From 2f33b5b56e9221f2613b34cd1a1a9d82d5ed4303 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 18 Nov 2007 17:08:06 +1100 Subject: nouveau: Very rough cut at gallium winsys + nv40 pipe driver. --- src/mesa/drivers/dri/nouveau_winsys/Makefile | 38 + src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 347 ++ .../drivers/dri/nouveau_winsys/nouveau_channel.c | 130 + .../drivers/dri/nouveau_winsys/nouveau_context.c | 228 + .../drivers/dri/nouveau_winsys/nouveau_context.h | 104 + .../drivers/dri/nouveau_winsys/nouveau_device.c | 144 + .../drivers/dri/nouveau_winsys/nouveau_device.h | 29 + src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 193 + src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h | 158 + src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h | 28 + .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 212 + .../drivers/dri/nouveau_winsys/nouveau_grobj.c | 107 + .../drivers/dri/nouveau_winsys/nouveau_local.h | 82 + src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c | 103 + .../drivers/dri/nouveau_winsys/nouveau_notifier.c | 140 + .../drivers/dri/nouveau_winsys/nouveau_screen.c | 253 + .../drivers/dri/nouveau_winsys/nouveau_screen.h | 17 + .../dri/nouveau_winsys/nouveau_swapbuffers.c | 98 + .../dri/nouveau_winsys/nouveau_swapbuffers.h | 10 + .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 90 + .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 251 + .../dri/nouveau_winsys/nouveau_winsys_pipe.h | 23 + .../dri/nouveau_winsys/nouveau_winsys_softpipe.c | 82 + src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 163 + src/mesa/drivers/dri/nouveau_winsys/nv50_region.c | 51 + src/mesa/pipe/Makefile | 3 +- src/mesa/pipe/nouveau/nouveau_bo.h | 50 + src/mesa/pipe/nouveau/nouveau_channel.h | 37 + src/mesa/pipe/nouveau/nouveau_class.h | 5261 ++++++++++++++++++++ src/mesa/pipe/nouveau/nouveau_grobj.h | 41 + src/mesa/pipe/nouveau/nouveau_notifier.h | 43 + src/mesa/pipe/nouveau/nouveau_winsys.h | 53 + src/mesa/pipe/nv40/Makefile | 30 + src/mesa/pipe/nv40/nv40_clear.c | 21 + src/mesa/pipe/nv40/nv40_context.c | 277 ++ src/mesa/pipe/nv40/nv40_context.h | 111 + src/mesa/pipe/nv40/nv40_dma.h | 62 + src/mesa/pipe/nv40/nv40_draw.c | 63 + src/mesa/pipe/nv40/nv40_fragprog.c | 642 +++ src/mesa/pipe/nv40/nv40_miptree.c | 60 + src/mesa/pipe/nv40/nv40_query.c | 98 + src/mesa/pipe/nv40/nv40_region.c | 85 + src/mesa/pipe/nv40/nv40_shader.h | 546 ++ src/mesa/pipe/nv40/nv40_state.c | 674 +++ src/mesa/pipe/nv40/nv40_state.h | 173 + src/mesa/pipe/nv40/nv40_state_emit.c | 112 + src/mesa/pipe/nv40/nv40_state_tex.c | 140 + src/mesa/pipe/nv40/nv40_surface.c | 229 + src/mesa/pipe/nv40/nv40_vbo.c | 222 + src/mesa/pipe/nv40/nv40_vertprog.c | 594 +++ src/mesa/pipe/nv40/nvgl_pipe.h | 198 + 51 files changed, 12905 insertions(+), 1 deletion(-) create mode 100644 src/mesa/drivers/dri/nouveau_winsys/Makefile create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv04_region.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv50_region.c create mode 100644 src/mesa/pipe/nouveau/nouveau_bo.h create mode 100644 src/mesa/pipe/nouveau/nouveau_channel.h create mode 100644 src/mesa/pipe/nouveau/nouveau_class.h create mode 100644 src/mesa/pipe/nouveau/nouveau_grobj.h create mode 100644 src/mesa/pipe/nouveau/nouveau_notifier.h create mode 100644 src/mesa/pipe/nouveau/nouveau_winsys.h create mode 100644 src/mesa/pipe/nv40/Makefile create mode 100644 src/mesa/pipe/nv40/nv40_clear.c create mode 100644 src/mesa/pipe/nv40/nv40_context.c create mode 100644 src/mesa/pipe/nv40/nv40_context.h create mode 100644 src/mesa/pipe/nv40/nv40_dma.h create mode 100644 src/mesa/pipe/nv40/nv40_draw.c create mode 100644 src/mesa/pipe/nv40/nv40_fragprog.c create mode 100644 src/mesa/pipe/nv40/nv40_miptree.c create mode 100644 src/mesa/pipe/nv40/nv40_query.c create mode 100644 src/mesa/pipe/nv40/nv40_region.c create mode 100644 src/mesa/pipe/nv40/nv40_shader.h create mode 100644 src/mesa/pipe/nv40/nv40_state.c create mode 100644 src/mesa/pipe/nv40/nv40_state.h create mode 100644 src/mesa/pipe/nv40/nv40_state_emit.c create mode 100644 src/mesa/pipe/nv40/nv40_state_tex.c create mode 100644 src/mesa/pipe/nv40/nv40_surface.c create mode 100644 src/mesa/pipe/nv40/nv40_vbo.c create mode 100644 src/mesa/pipe/nv40/nv40_vertprog.c create mode 100644 src/mesa/pipe/nv40/nvgl_pipe.h diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile new file mode 100644 index 00000000000..97b861526f4 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -0,0 +1,38 @@ + +TOP = ../../../../.. +include $(TOP)/configs/current + +LIBNAME = nouveau_dri.so + +MINIGLX_SOURCES = + +PIPE_DRIVERS = \ + $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a \ + $(TOP)/src/mesa/pipe/nv40/libnv40.a + +DRIVER_SOURCES = \ + nouveau_bo.c \ + nouveau_channel.c \ + nouveau_context.c \ + nouveau_device.c \ + nouveau_dma.c \ + nouveau_grobj.c \ + nouveau_lock.c \ + nouveau_notifier.c \ + nouveau_screen.c \ + nouveau_swapbuffers.c \ + nouveau_winsys.c \ + nouveau_winsys_pipe.c \ + nouveau_winsys_softpipe.c \ + nv04_region.c \ + nv50_region.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../Makefile.template + +symlinks: diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c new file mode 100644 index 00000000000..3b508cf92b3 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -0,0 +1,347 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +int +nouveau_bo_init(struct nouveau_device *userdev) +{ + struct nouveau_device_priv *nv = nouveau_device(userdev); + struct nouveau_channel *chan; + int ret; + + if ((ret = nouveau_channel_alloc(userdev, 0x80000001, 0x80000002, + &nv->bufmgr.channel))) + return ret; + chan = nv->bufmgr.channel; + + if ((ret = nouveau_notifier_alloc(nv->bufmgr.channel, 0x80000003, 1, + &nv->bufmgr.notify))) + return ret; + + if ((ret = nouveau_grobj_alloc(nv->bufmgr.channel, 0x80000004, 0x39, + &nv->bufmgr.m2mf))) + return ret; + + nouveau_notifier_reset(nv->bufmgr.notify, 0); + BEGIN_RING_CH(chan, nv->bufmgr.m2mf, 0x180, 1); + OUT_RING_CH (chan, nv->bufmgr.notify->handle); + BEGIN_RING_CH(chan, nv->bufmgr.m2mf, 0x104, 1); + OUT_RING_CH (chan, 0); + BEGIN_RING_CH(chan, nv->bufmgr.m2mf, 0x100, 1); + OUT_RING_CH (chan, 0); + FIRE_RING_CH (chan); + if ((ret = nouveau_notifier_wait_status(nv->bufmgr.notify, 0, 0, 2000))) + return ret; + + return 0; +} + +void +nouveau_bo_takedown(struct nouveau_device *userdev) +{ + struct nouveau_device_priv *nv = nouveau_device(userdev); + + nouveau_notifier_free(&nv->bufmgr.notify); + nouveau_grobj_free(&nv->bufmgr.m2mf); + nouveau_channel_free(&nv->bufmgr.channel); +} + +static int +nouveau_bo_realloc_gpu(struct nouveau_bo_priv *bo, uint32_t flags, int size) +{ + struct nouveau_device_priv *nv = nouveau_device(bo->base.device); + int ret; + + if (bo->drm.size && bo->drm.size != size) { + struct drm_nouveau_mem_free f; + + if (bo->map) { + drmUnmap(bo->map, bo->drm.size); + bo->map = NULL; + } + + f.flags = bo->drm.flags; + f.offset = bo->drm.offset; + drmCommandWrite(nv->fd, DRM_NOUVEAU_MEM_FREE, &f, sizeof(f)); + + bo->drm.size = 0; + } + + if (size && !bo->drm.size) { + if (flags) { + bo->drm.flags = 0; + if (flags & NOUVEAU_BO_VRAM) + bo->drm.flags |= NOUVEAU_MEM_FB; + if (flags & NOUVEAU_BO_GART) + bo->drm.flags |= (NOUVEAU_MEM_AGP | + NOUVEAU_MEM_PCI); + bo->drm.flags |= NOUVEAU_MEM_MAPPED; + } + + bo->drm.size = size; + + ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_MEM_ALLOC, + &bo->drm, sizeof(bo->drm)); + if (ret) { + free(bo); + return ret; + } + + ret = drmMap(nv->fd, bo->drm.map_handle, bo->drm.size, + &bo->map); + if (ret) { + bo->map = NULL; + nouveau_bo_del((void *)&bo); + return ret; + } + } + + return 0; +} + +int +nouveau_bo_new(struct nouveau_device *userdev, uint32_t flags, int align, + int size, struct nouveau_bo **userbo) +{ + struct nouveau_bo_priv *bo; + int ret; + + if (!userdev || !userbo || *userbo) + return -EINVAL; + + bo = calloc(1, sizeof(*bo)); + if (!bo) + return -ENOMEM; + bo->base.device = userdev; + bo->drm.alignment = align; + + if (flags & NOUVEAU_BO_PIN) { + ret = nouveau_bo_realloc_gpu(bo, flags, size); + if (ret) { + free(bo); + return ret; + } + } else { + bo->sysmem = malloc(size); + if (!bo->sysmem) { + free(bo); + return -ENOMEM; + } + } + + bo->base.size = size; + bo->base.offset = bo->drm.offset; + bo->base.handle = (unsigned long)bo; + bo->refcount = 1; + *userbo = &bo->base; + return 0; +} + +int +nouveau_bo_user(struct nouveau_device *userdev, void *ptr, int size, + struct nouveau_bo **userbo) +{ + struct nouveau_bo_priv *bo; + + if (!userdev || !userbo || *userbo) + return -EINVAL; + + bo = calloc(1, sizeof(*bo)); + if (!bo) + return -ENOMEM; + bo->base.device = userdev; + + bo->sysmem = ptr; + bo->user = 1; + + bo->base.size = size; + bo->base.offset = bo->drm.offset; + bo->base.handle = (unsigned long)bo; + bo->refcount = 1; + *userbo = &bo->base; + return 0; +} + +int +nouveau_bo_ref(struct nouveau_device *userdev, uint64_t handle, + struct nouveau_bo **userbo) +{ + struct nouveau_bo_priv *bo = (void *)(unsigned long)handle; + + if (!userdev || !userbo || *userbo) + return -EINVAL; + + bo->refcount++; + *userbo = &bo->base; + return 0; +} + +int +nouveau_bo_resize(struct nouveau_bo *userbo, int size) +{ + struct nouveau_bo_priv *bo = nouveau_bo(userbo); + int ret; + + if (!bo || bo->user) + return -EINVAL; + + if (bo->sysmem) { + bo->sysmem = realloc(bo->sysmem, size); + if (!bo->sysmem) + return -ENOMEM; + } else { + ret = nouveau_bo_realloc_gpu(bo, 0, size); + if (ret) + return ret; + } + + bo->base.size = size; + return 0; +} + +void +nouveau_bo_del(struct nouveau_bo **userbo) +{ + struct nouveau_bo_priv *bo; + + if (!userbo || !*userbo) + return; + bo = nouveau_bo(*userbo); + *userbo = NULL; + + if (--bo->refcount) + return; + + nouveau_bo_realloc_gpu(bo, 0, 0); + if (bo->sysmem && !bo->user) + free(bo->sysmem); + free(bo); +} + +int +nouveau_bo_map(struct nouveau_bo *userbo, uint32_t flags) +{ + struct nouveau_bo_priv *bo = nouveau_bo(userbo); + + if (!bo) + return -EINVAL; + + if (bo->sysmem) + userbo->map = bo->sysmem; + else + userbo->map = bo->map; + return 0; +} + +void +nouveau_bo_unmap(struct nouveau_bo *userbo) +{ + userbo->map = NULL; +} + +void +nouveau_bo_emit_reloc(struct nouveau_channel *userchan, void *ptr, + struct nouveau_bo *userbo, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_bo_priv *bo = nouveau_bo(userbo); + struct nouveau_bo_reloc *r; + int i, on_list = 0; + + for (i = 0; i < chan->nr_buffers; i++) { + if (chan->buffers[i].bo == bo) { + on_list = 1; + break; + } + } + + if (i >= 128) + return; + + if (on_list) { + chan->buffers[i].flags &= (flags | NOUVEAU_BO_RDWR); + chan->buffers[i].flags |= (flags & NOUVEAU_BO_RDWR); + } else { + chan->buffers[i].bo = bo; + chan->buffers[i].flags = flags; + chan->nr_buffers++; + } + + if (chan->num_relocs >= chan->max_relocs) + FIRE_RING_CH(userchan); + r = &chan->relocs[chan->num_relocs++]; + + r->ptr = ptr; + r->bo = bo; + r->data = data; + r->flags = flags; + r->vor = vor; + r->tor = tor; +} + +static int +nouveau_bo_upload(struct nouveau_bo_priv *bo) +{ + memcpy(bo->map, bo->sysmem, bo->drm.size); + return 0; +} + +void +nouveau_bo_validate(struct nouveau_channel *userchan) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + int i; + + for (i = 0; i < chan->nr_buffers; i++) { + struct nouveau_bo_priv *bo = chan->buffers[i].bo; + + if (!bo->drm.size) { + nouveau_bo_realloc_gpu(bo, chan->buffers[i].flags, + bo->base.size); + nouveau_bo_upload(bo); + } else + if (bo->user || bo->base.map) + nouveau_bo_upload(bo); + + if (!bo->user && !bo->base.map) { + free(bo->sysmem); + bo->sysmem = NULL; + } + + + bo->base.offset = bo->drm.offset; + if (bo->drm.flags & NOUVEAU_MEM_AGP) + bo->base.flags = NOUVEAU_BO_GART; + else + bo->base.flags = NOUVEAU_BO_VRAM; + } + chan->nr_buffers = 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c new file mode 100644 index 00000000000..ef41020b727 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c @@ -0,0 +1,130 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +int +nouveau_channel_alloc(struct nouveau_device *userdev, uint32_t fb_ctxdma, + uint32_t tt_ctxdma, struct nouveau_channel **userchan) +{ + struct nouveau_device_priv *nv = nouveau_device(userdev); + struct nouveau_channel_priv *chan; + int ret; + + if (!nv || !userchan || *userchan) + return -EINVAL; + + chan = calloc(1, sizeof(*chan)); + if (!chan) + return -ENOMEM; + chan->base.device = userdev; + + chan->drm.fb_ctxdma_handle = fb_ctxdma; + chan->drm.tt_ctxdma_handle = tt_ctxdma; + ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_CHANNEL_ALLOC, + &chan->drm, sizeof(chan->drm)); + if (ret) { + free(chan); + return ret; + } + + chan->base.id = chan->drm.channel; + if (nouveau_grobj_ref(&chan->base, chan->drm.fb_ctxdma_handle, + &chan->base.vram) || + nouveau_grobj_ref(&chan->base, chan->drm.tt_ctxdma_handle, + &chan->base.gart)) { + nouveau_channel_free((void *)&chan); + return -EINVAL; + } + + ret = drmMap(nv->fd, chan->drm.ctrl, chan->drm.ctrl_size, + (void*)&chan->user); + if (ret) { + nouveau_channel_free((void *)&chan); + return ret; + } + chan->put = &chan->user[0x40/4]; + chan->get = &chan->user[0x44/4]; + chan->ref_cnt = &chan->user[0x48/4]; + + ret = drmMap(nv->fd, chan->drm.notifier, chan->drm.notifier_size, + (drmAddressPtr)&chan->notifier_block); + if (ret) { + nouveau_channel_free((void *)&chan); + return ret; + } + + ret = drmMap(nv->fd, chan->drm.cmdbuf, chan->drm.cmdbuf_size, + (void*)&chan->pushbuf); + if (ret) { + nouveau_channel_free((void *)&chan); + return ret; + } + + chan->max_relocs = chan->drm.cmdbuf_size / 4; + chan->num_relocs = 0; + chan->relocs = + malloc(sizeof(struct nouveau_bo_reloc) * chan->max_relocs); + + nouveau_dma_channel_init(&chan->base); + + *userchan = &chan->base; + return 0; +} + +void +nouveau_channel_free(struct nouveau_channel **userchan) +{ + struct nouveau_channel_priv *chan; + + if (!userchan) + return; + chan = nouveau_channel(*userchan); + + if (chan) { + struct nouveau_device_priv *nv; + struct drm_nouveau_channel_free cf; + + nv = nouveau_device((*userchan)->device); + *userchan = NULL; + + FIRE_RING_CH(&chan->base); + + if (chan->relocs) + free(chan->relocs); + + nouveau_grobj_free(&chan->base.vram); + nouveau_grobj_free(&chan->base.gart); + + cf.channel = chan->drm.channel; + drmCommandWrite(nv->fd, DRM_NOUVEAU_CHANNEL_FREE, + &cf, sizeof(cf)); + free(chan); + } +} + + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c new file mode 100644 index 00000000000..584d236f349 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c @@ -0,0 +1,228 @@ +#include "glheader.h" +#include "context.h" +#include "extensions.h" + +#include "tnl/tnl.h" +#include "tnl/t_pipeline.h" +#include "tnl/t_vertex.h" + +#include "drivers/common/driverfuncs.h" + +#include "drirenderbuffer.h" + +#include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" + +#include "pipe/p_defines.h" +#include "pipe/p_context.h" + +#include "nouveau_context.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_winsys_pipe.h" + +#define need_GL_ARB_fragment_program +#define need_GL_ARB_multisample +#define need_GL_ARB_occlusion_query +#define need_GL_ARB_texture_compression +#define need_GL_ARB_vertex_program +#define need_GL_ARB_vertex_buffer_object +#define need_GL_EXT_compiled_vertex_array +#define need_GL_EXT_fog_coord +#define need_GL_EXT_secondary_color +#define need_GL_EXT_framebuffer_object +#include "extension_helper.h" + +const struct dri_extension common_extensions[] = +{ + { NULL, 0 } +}; + +const struct dri_extension nv40_extensions[] = +{ + { "GL_ARB_fragment_program", NULL }, + { "GL_ARB_multisample", GL_ARB_multisample_functions }, + { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, + { "GL_ARB_texture_border_clamp", NULL }, + { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, + { "GL_ARB_texture_cube_map", NULL }, + { "GL_ARB_texture_env_add", NULL }, + { "GL_ARB_texture_env_combine", NULL }, + { "GL_ARB_texture_env_crossbar", NULL }, + { "GL_ARB_texture_env_dot3", NULL }, + { "GL_ARB_texture_mirrored_repeat", NULL }, + { "GL_ARB_texture_non_power_of_two", NULL }, + { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, + { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, + { "GL_ATI_texture_env_combine3", NULL }, + { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, + { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, + { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, + { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, + { "GL_EXT_texture_edge_clamp", NULL }, + { "GL_EXT_texture_env_add", NULL }, + { "GL_EXT_texture_env_combine", NULL }, + { "GL_EXT_texture_env_dot3", NULL }, + { "GL_EXT_texture_mirror_clamp", NULL }, + { "GL_NV_texture_rectangle", NULL }, + { NULL, 0 } +}; + +#ifdef DEBUG +static const struct dri_debug_control debug_control[] = { + { "bo", DEBUG_BO }, + { NULL, 0 } +}; +int __nouveau_debug = 0; +#endif + +GLboolean +nouveau_context_create(const __GLcontextModes *glVis, + __DRIcontextPrivate *driContextPriv, + void *sharedContextPrivate) +{ + __DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv; + struct nouveau_screen *nv_screen = driScrnPriv->private; + struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); + struct nouveau_device_priv *nvdev; + struct pipe_context *pipe = NULL; + struct st_context *st_share = NULL; + int ret; + + if (sharedContextPrivate) { + st_share = ((struct nouveau_context *)sharedContextPrivate)->st; + } + + if ((ret = nouveau_device_get_param(nv_screen->device, + NOUVEAU_GETPARAM_CHIPSET_ID, + &nv->chipset))) { + NOUVEAU_ERR("Error determining chipset id: %d\n", ret); + return GL_FALSE; + } + + if ((ret = nouveau_channel_alloc(nv_screen->device, + 0x8003d001, 0x8003d002, + &nv->channel))) { + NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); + return GL_FALSE; + } + + driContextPriv->driverPrivate = (void *)nv; + nv->nv_screen = nv_screen; + nv->dri_screen = driScrnPriv; + + nvdev = nouveau_device(nv_screen->device); + nvdev->ctx = driContextPriv->hHWContext; + nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; + + driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, + nv->dri_screen->myNum, "nouveau"); +#ifdef DEBUG + __nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"), + debug_control); +#endif + + if ((ret = nouveau_grobj_alloc(nv->channel, 0x00000000, 0x30, + &nv->NvNull))) { + NOUVEAU_ERR("Error creating NULL object: %d\n", ret); + return GL_FALSE; + } + nv->next_handle = 0x80000000; + + if ((ret = nouveau_notifier_alloc(nv->channel, nv->next_handle++, 1, + &nv->sync_notifier))) { + NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); + return GL_FALSE; + } + + if (nv->chipset < 0x50) + ret = nouveau_region_init_nv04(nv); + else + ret = nouveau_region_init_nv50(nv); + if (ret) { + return GL_FALSE; + } + + if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { + pipe = nouveau_pipe_create(nv); + if (!pipe) + NOUVEAU_ERR("Couldn't create hw pipe\n"); + } + + if (!pipe) { + NOUVEAU_MSG("Using softpipe\n"); + pipe = nouveau_create_softpipe(nv); + if (!pipe) { + NOUVEAU_ERR("Error creating pipe, bailing\n"); + return GL_FALSE; + } + } + + pipe->priv = nv; + nv->st = st_create_context(pipe, glVis, st_share); + return GL_TRUE; +} + +void +nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) +{ + struct nouveau_context *nv = driContextPriv->driverPrivate; + + assert(nv); + + st_flush(nv->st, PIPE_FLUSH_WAIT); + st_destroy_context(nv->st); + + nouveau_grobj_free(&nv->NvCtxSurf2D); + nouveau_grobj_free(&nv->NvImageBlit); + nouveau_channel_free(&nv->channel); + + free(nv); +} + +GLboolean +nouveau_context_bind(__DRIcontextPrivate *driContextPriv, + __DRIdrawablePrivate *driDrawPriv, + __DRIdrawablePrivate *driReadPriv) +{ + struct nouveau_context *nv; + struct nouveau_framebuffer *draw, *read; + + if (!driContextPriv) { + st_make_current(NULL, NULL, NULL); + return GL_TRUE; + } + + nv = driContextPriv->driverPrivate; + draw = driDrawPriv->driverPrivate; + read = driReadPriv->driverPrivate; + + st_make_current(nv->st, draw->stfb, read->stfb); + + if ((nv->dri_drawable != driDrawPriv) || + (nv->last_stamp != driDrawPriv->lastStamp)) { + nv->dri_drawable = driDrawPriv; + st_resize_framebuffer(draw->stfb, driDrawPriv->w, + driDrawPriv->h); + nv->last_stamp = driDrawPriv->lastStamp; + } + + if (driDrawPriv != driReadPriv) { + st_resize_framebuffer(read->stfb, driReadPriv->w, + driReadPriv->h); + } + + return GL_TRUE; +} + +GLboolean +nouveau_context_unbind(__DRIcontextPrivate *driContextPriv) +{ + struct nouveau_context *nv = driContextPriv->driverPrivate; + (void)nv; + + st_flush(nv->st, 0); + return GL_TRUE; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h new file mode 100644 index 00000000000..cd04b08eca1 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -0,0 +1,104 @@ +#ifndef __NOUVEAU_CONTEXT_H__ +#define __NOUVEAU_CONTEXT_H__ + +#include "glheader.h" +#include "context.h" + +#include "dri_util.h" +#include "xmlconfig.h" + +#include "pipe/nouveau/nouveau_winsys.h" +#include "nouveau_device.h" +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +struct nouveau_framebuffer { + struct st_framebuffer *stfb; +}; + +struct nouveau_context { + struct st_context *st; + + /* Misc HW info */ + uint64_t chipset; + + /* DRI stuff */ + __DRIscreenPrivate *dri_screen; + __DRIdrawablePrivate *dri_drawable; + unsigned int last_stamp; + driOptionCache dri_option_cache; + drm_context_t drm_context; + drmLock drm_lock; + GLboolean locked; + struct nouveau_screen *nv_screen; + + /* Bufmgr */ + struct { + struct nouveau_channel *channel; + struct nouveau_notifier *notify; + struct nouveau_grobj *m2mf; + uint32_t m2mf_src_ctxdma; + uint32_t m2mf_dst_ctxdma; + uint32_t next_sequence; + } bo; + + /* Relocations */ + struct nouveau_bo *reloc_head; + + /* Hardware context */ + struct nouveau_channel *channel; + struct nouveau_notifier *sync_notifier; + struct nouveau_grobj *NvNull; + struct nouveau_grobj *NvCtxSurf2D; + struct nouveau_grobj *NvImageBlit; + struct nouveau_grobj *NvGdiRect; + uint32_t next_handle; + uint32_t next_sequence; + + /* pipe_region accel */ + int (*region_display)(void); + int (*region_copy)(struct nouveau_context *, struct pipe_region *, + unsigned, unsigned, unsigned, struct pipe_region *, + unsigned, unsigned, unsigned, unsigned, unsigned); + int (*region_fill)(struct nouveau_context *, struct pipe_region *, + unsigned, unsigned, unsigned, unsigned, unsigned, + unsigned); + int (*region_data)(struct nouveau_context *, struct pipe_region *, + unsigned, unsigned, unsigned, const void *, + unsigned, unsigned, unsigned, unsigned, unsigned); +}; + +static INLINE struct nouveau_context * +nouveau_context(GLcontext *ctx) +{ + return (struct nouveau_context *)ctx; +} + +extern GLboolean nouveau_context_create(const __GLcontextModes *, + __DRIcontextPrivate *, void *); +extern void nouveau_context_destroy(__DRIcontextPrivate *); +extern GLboolean nouveau_context_bind(__DRIcontextPrivate *, + __DRIdrawablePrivate *draw, + __DRIdrawablePrivate *read); +extern GLboolean nouveau_context_unbind(__DRIcontextPrivate *); + +#ifdef DEBUG +extern int __nouveau_debug; + +#define DEBUG_BO (1 << 0) + +#define DBG(flag, ...) do { \ + if (__nouveau_debug & (DEBUG_##flag)) \ + NOUVEAU_ERR(__VA_ARGS__); \ +} while(0) +#else +#define DBG(flag, ...) +#endif + +extern void LOCK_HARDWARE(struct nouveau_context *); +extern void UNLOCK_HARDWARE(struct nouveau_context *); + +extern int nouveau_region_init_nv04(struct nouveau_context *); +extern int nouveau_region_init_nv50(struct nouveau_context *); + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c new file mode 100644 index 00000000000..a3f55135504 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c @@ -0,0 +1,144 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" + +int +nouveau_device_open_existing(struct nouveau_device **userdev, int close, + int fd, drm_context_t ctx) +{ + struct nouveau_device_priv *nv; + int ret; + + if (!userdev || *userdev) + return -EINVAL; + + nv = calloc(1, sizeof(*nv)); + if (!nv) + return -ENOMEM; + nv->fd = fd; + nv->ctx = ctx; + nv->needs_close = close; + + drmCommandNone(nv->fd, DRM_NOUVEAU_CARD_INIT); + + if ((ret = nouveau_bo_init(&nv->base))) { + nouveau_device_close((void *)&nv); + return ret; + } + + *userdev = &nv->base; + return 0; +} + +int +nouveau_device_open(struct nouveau_device **userdev, const char *busid) +{ + drm_context_t ctx; + int fd, ret; + + if (!userdev || *userdev) + return -EINVAL; + + fd = drmOpen("nouveau", busid); + if (fd < 0) + return -EINVAL; + + ret = drmCreateContext(fd, &ctx); + if (ret) { + drmClose(fd); + return ret; + } + + ret = nouveau_device_open_existing(userdev, 1, fd, ctx); + if (ret) { + drmDestroyContext(fd, ctx); + drmClose(fd); + return ret; + } + + return 0; +} + +void +nouveau_device_close(struct nouveau_device **userdev) +{ + struct nouveau_device_priv *nv; + + if (userdev || !*userdev) + return; + nv = (struct nouveau_device_priv *)*userdev; + *userdev = NULL; + + nouveau_bo_takedown(&nv->base); + + if (nv->needs_close) { + drmDestroyContext(nv->fd, nv->ctx); + drmClose(nv->fd); + } + free(nv); +} + +int +nouveau_device_get_param(struct nouveau_device *userdev, + uint64_t param, uint64_t *value) +{ + struct nouveau_device_priv *nv = (struct nouveau_device_priv *)userdev; + struct drm_nouveau_getparam g; + int ret; + + if (!nv || !value) + return -EINVAL; + + g.param = param; + ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_GETPARAM, &g, sizeof(g)); + if (ret) + return ret; + + *value = g.value; + return 0; +} + +int +nouveau_device_set_param(struct nouveau_device *userdev, + uint64_t param, uint64_t value) +{ + struct nouveau_device_priv *nv = (struct nouveau_device_priv *)userdev; + struct drm_nouveau_setparam s; + int ret; + + if (!nv) + return -EINVAL; + + s.param = param; + s.value = value; + ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_SETPARAM, &s, sizeof(s)); + if (ret) + return ret; + + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h new file mode 100644 index 00000000000..744a89f74bf --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h @@ -0,0 +1,29 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DEVICE_H__ +#define __NOUVEAU_DEVICE_H__ + +struct nouveau_device { +}; + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c new file mode 100644 index 00000000000..6e123c44734 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -0,0 +1,193 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +#define READ_GET(ch) ((*(ch)->get - (ch)->dma.base) >> 2) +#define WRITE_PUT(ch, val) do { \ + volatile int dum; \ + NOUVEAU_DMA_BARRIER; \ + dum=READ_GET(ch); \ + *(ch)->put = (((val) << 2) + (ch)->dma.base); \ + NOUVEAU_DMA_BARRIER; \ +} while(0) + +void +nouveau_dma_channel_init(struct nouveau_channel *userchan) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + int i; + + chan->dma.base = chan->drm.put_base; + chan->dma.cur = chan->dma.put = RING_SKIPS; + chan->dma.max = (chan->drm.cmdbuf_size >> 2) - 2; + chan->dma.free = chan->dma.max - chan->dma.cur; + + for (i = 0; i < RING_SKIPS; i++) + chan->pushbuf[i] = 0x00000000; +} + +#define CHECK_TIMEOUT() do { \ + if ((NOUVEAU_TIME_MSEC() - t_start) > NOUVEAU_DMA_TIMEOUT) \ + return - EBUSY; \ +} while(0) + +int +nouveau_dma_wait(struct nouveau_channel *userchan, int size) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + uint32_t get, t_start; + + FIRE_RING_CH(userchan); + + t_start = NOUVEAU_TIME_MSEC(); + while (chan->dma.free < size) { + get = READ_GET(chan); + + if (chan->dma.put >= get) { + chan->dma.free = chan->dma.max - chan->dma.cur; + + if (chan->dma.free < size) { +#ifdef NOUVEAU_DMA_DEBUG + chan->dma.push_free = 1; +#endif + OUT_RING_CH(userchan, + 0x20000000 | chan->dma.base); + if (get <= RING_SKIPS) { + /*corner case - will be idle*/ + if (chan->dma.put <= RING_SKIPS) + WRITE_PUT(chan, RING_SKIPS + 1); + + do { + CHECK_TIMEOUT(); + get = READ_GET(chan); + } while (get <= RING_SKIPS); + } + + WRITE_PUT(chan, RING_SKIPS); + chan->dma.cur = chan->dma.put = RING_SKIPS; + chan->dma.free = get - (RING_SKIPS + 1); + } + } else { + chan->dma.free = get - chan->dma.cur - 1; + } + + CHECK_TIMEOUT(); + } + + return 0; +} + +#ifdef NOUVEAU_DMA_SUBCHAN_LRU +void +nouveau_dma_subc_bind(struct nouveau_grobj *grobj) +{ + struct nouveau_channel_priv *chan = nouveau_channel(grobj->channel); + int subc = -1, i; + + for (i = 0; i < 8; i++) { + if (chan->subchannel[i].grobj && + chan->subchannel[i].grobj->bound == + NOUVEAU_GROBJ_EXPLICIT_BIND) + continue; + if (chan->subchannel[i].seq < chan->subchannel[subc].seq) + subc = i; + } + assert(subc >= 0); + + if (chan->subchannel[subc].grobj) + chan->subchannel[subc].grobj->bound = 0; + chan->subchannel[subc].grobj = grobj; + grobj->subc = subc; + grobj->bound = NOUVEAU_GROBJ_BOUND; + + BEGIN_RING_CH(grobj->channel, grobj, 0, 1); + nouveau_dma_out (grobj->channel, grobj->handle); +} +#endif + +void +nouveau_dma_kickoff(struct nouveau_channel *userchan) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + uint32_t put_offset; + int i; + + if (chan->dma.cur == chan->dma.put) + return; + + if (chan->num_relocs) { + nouveau_bo_validate(userchan); + + for (i = 0; i < chan->num_relocs; i++) { + struct nouveau_bo_reloc *r = &chan->relocs[i]; + uint32_t push; + + if (r->flags & NOUVEAU_BO_LOW) { + push = r->bo->base.offset + r->data; + } else + if (r->flags & NOUVEAU_BO_HIGH) { + push = (r->bo->base.offset + r->data) >> 32; + } else { + push = r->data; + } + + if (r->flags & NOUVEAU_BO_OR) { + if (r->bo->base.flags & NOUVEAU_BO_VRAM) + push |= r->vor; + else + push |= r->tor; + } + + *r->ptr = push; + } + + chan->num_relocs = 0; + } + +#ifdef NOUVEAU_DMA_DEBUG + if (chan->dma.push_free) { + NOUVEAU_ERR("Packet incomplete: %d left\n", chan->dma.push_free); + return; + } +#endif + +#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF + for (i = chan->dma.put; i < chan->dma.cur; i++) + NOUVEAU_MSG("0x%08x\n", chan->pushbuf[i]); +#endif + + put_offset = (chan->dma.cur << 2) + chan->dma.base; +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("FIRE_RING %d/0x%08x\n", chan->drm.channel, put_offset); +#endif + chan->dma.put = chan->dma.cur; + NOUVEAU_DMA_BARRIER; + *chan->put = put_offset; + NOUVEAU_DMA_BARRIER; +} diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h new file mode 100644 index 00000000000..927841c4eb4 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h @@ -0,0 +1,158 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DMA_H__ +#define __NOUVEAU_DMA_H__ + +#include +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +#define RING_SKIPS 8 + +extern int nouveau_dma_wait(struct nouveau_channel *chan, int size); +extern void nouveau_dma_subc_bind(struct nouveau_grobj *); +extern void nouveau_dma_channel_init(struct nouveau_channel *); +extern void nouveau_dma_kickoff(struct nouveau_channel *); + +#ifdef NOUVEAU_DMA_DEBUG +static char faulty[1024]; +#endif + +static inline void +nouveau_dma_out(struct nouveau_channel *userchan, uint32_t data) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + +#ifdef NOUVEAU_DMA_DEBUG + if (chan->dma.push_free == 0) { + NOUVEAU_ERR("No space left in packet. Error at %s\n",faulty); + return; + } + chan->dma.push_free--; +#endif +#ifdef NOUVEAU_DMA_TRACE + { + uint32_t offset = (chan->dma.cur << 2) + chan->dma.base; + NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", + chan->drm.channel, offset, data); + } +#endif + chan->pushbuf[chan->dma.cur++] = data; +} + +static inline void +nouveau_dma_outp(struct nouveau_channel *userchan, uint32_t *ptr, int size) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + (void)chan; + +#ifdef NOUVEAU_DMA_DEBUG + if (chan->dma.push_free < size) { + NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", + chan->dma.push_free, size); + return; + } +#endif +#ifdef NOUVEAU_DMA_TRACE + while (size--) { + nouveau_dma_out(userchan, *ptr); + ptr++; + } +#else + memcpy(&chan->pushbuf[chan->dma.cur], ptr, size << 2); +#ifdef NOUVEAU_DMA_DEBUG + chan->dma.push_free -= size; +#endif + chan->dma.cur += size; +#endif +} + +static inline void +nouveau_dma_begin(struct nouveau_channel *userchan, struct nouveau_grobj *grobj, + int method, int size, const char* file, int line) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + int push_size = size + 1; + +#ifdef NOUVEAU_DMA_SUBCHAN_LRU + if (grobj->bound == NOUVEAU_GROBJ_UNBOUND) + nouveau_dma_subc_bind(grobj); + chan->subchannel[grobj->subc].seq = chan->subc_sequence++; +#endif + +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", chan->drm.channel, + grobj->handle, grobj->subc, method, size); +#endif + +#ifdef NOUVEAU_DMA_DEBUG + if (chan->dma.push_free) { + NOUVEAU_ERR("Previous packet incomplete: %d left. Error at %s\n", + chan->dma.push_free,faulty); + return; + } + sprintf(faulty,"%s:%d",file,line); +#endif + + if (chan->dma.free < push_size) { + if (nouveau_dma_wait(userchan, push_size) && + userchan->hang_notify) { + userchan->hang_notify(userchan); + } + } + chan->dma.free -= push_size; +#ifdef NOUVEAU_DMA_DEBUG + chan->dma.push_free = push_size; +#endif + + nouveau_dma_out(userchan, (size << 18) | (grobj->subc << 13) | method); +} + +static inline void +nouveau_dma_bind(struct nouveau_channel *userchan, struct nouveau_grobj *grobj, + int subc) +{ + struct nouveau_channel_priv *chan = nouveau_channel(userchan); + + if (chan->subchannel[subc].grobj == grobj) + return; + + if (chan->subchannel[subc].grobj) + chan->subchannel[subc].grobj->bound = NOUVEAU_GROBJ_UNBOUND; + chan->subchannel[subc].grobj = grobj; + grobj->subc = subc; + grobj->bound = NOUVEAU_GROBJ_EXPLICIT_BIND; + + nouveau_dma_begin(userchan, grobj, 0x0000, 1, __FUNCTION__, __LINE__); + nouveau_dma_out (userchan, grobj->handle); +} + +#define BIND_RING_CH(ch,gr,sc) nouveau_dma_bind((ch), (gr), (sc)) +#define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) +#define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) +#define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ + (dwords)) +#define FIRE_RING_CH(ch) nouveau_dma_kickoff((ch)) +#define WAIT_RING_CH(ch,sz) nouveau_dma_wait((ch), (sz)) + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h new file mode 100644 index 00000000000..1207c2d609c --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h @@ -0,0 +1,28 @@ +#ifndef _NOUVEAU_DRI_ +#define _NOUVEAU_DRI_ + +#include "xf86drm.h" +#include "drm.h" +#include "nouveau_drm.h" + +struct nouveau_dri { + uint32_t device_id; /**< \brief PCI device ID */ + uint32_t width; /**< \brief width in pixels of display */ + uint32_t height; /**< \brief height in scanlines of display */ + uint32_t depth; /**< \brief depth of display (8, 15, 16, 24) */ + uint32_t bpp; /**< \brief bit depth of display (8, 16, 24, 32) */ + + uint32_t bus_type; /**< \brief ths bus type */ + uint32_t bus_mode; /**< \brief bus mode (used for AGP, maybe also for PCI-E ?) */ + + uint32_t front_offset; /**< \brief front buffer offset */ + uint32_t front_pitch; /**< \brief front buffer pitch */ + uint32_t back_offset; /**< \brief private back buffer offset */ + uint32_t back_pitch; /**< \brief private back buffer pitch */ + uint32_t depth_offset; /**< \brief private depth buffer offset */ + uint32_t depth_pitch; /**< \brief private depth buffer pitch */ + +}; + +#endif + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h new file mode 100644 index 00000000000..1a21da69857 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -0,0 +1,212 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DRMIF_H__ +#define __NOUVEAU_DRMIF_H__ + +#include +#include +#include + +#include "nouveau_device.h" +#include "pipe/nouveau/nouveau_channel.h" +#include "pipe/nouveau/nouveau_grobj.h" +#include "pipe/nouveau/nouveau_notifier.h" +#include "pipe/nouveau/nouveau_bo.h" + +struct nouveau_device_priv { + struct nouveau_device base; + + int fd; + drm_context_t ctx; + drmLock *lock; + int needs_close; + + struct { + struct nouveau_channel *channel; + struct nouveau_notifier *notify; + struct nouveau_grobj *m2mf; + } bufmgr; +}; +#define nouveau_device(n) ((struct nouveau_device_priv *)(n)) + +extern int +nouveau_device_open_existing(struct nouveau_device **, int close, + int fd, drm_context_t ctx); + +extern int +nouveau_device_open(struct nouveau_device **, const char *busid); + +extern void +nouveau_device_close(struct nouveau_device **); + +extern int +nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v); + +extern int +nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val); + +struct nouveau_channel_priv { + struct nouveau_channel base; + + struct drm_nouveau_channel_alloc drm; + + struct { + struct nouveau_grobj *grobj; + uint32_t seq; + } subchannel[8]; + uint32_t subc_sequence; + + uint32_t *pushbuf; + void *notifier_block; + + volatile uint32_t *user; + volatile uint32_t *put; + volatile uint32_t *get; + volatile uint32_t *ref_cnt; + + struct { + uint32_t base, max; + uint32_t cur, put; + uint32_t free; + + int push_free; + } dma; + + struct { + struct nouveau_bo_priv *bo; + uint32_t flags; + } buffers[128]; + int nr_buffers; + + struct nouveau_bo_reloc *relocs; + int num_relocs; + int max_relocs; +}; +#define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) + +extern int +nouveau_channel_alloc(struct nouveau_device *, uint32_t fb, uint32_t tt, + struct nouveau_channel **); + +extern void +nouveau_channel_free(struct nouveau_channel **); + +struct nouveau_grobj_priv { + struct nouveau_grobj base; +}; +#define nouveau_grobj(n) ((struct nouveau_grobj_priv *)(n)) + +extern int nouveau_grobj_alloc(struct nouveau_channel *, uint32_t handle, + int class, struct nouveau_grobj **); +extern int nouveau_grobj_ref(struct nouveau_channel *, uint32_t handle, + struct nouveau_grobj **); +extern void nouveau_grobj_free(struct nouveau_grobj **); + + +struct nouveau_notifier_priv { + struct nouveau_notifier base; + + struct drm_nouveau_notifierobj_alloc drm; + volatile void *map; +}; +#define nouveau_notifier(n) ((struct nouveau_notifier_priv *)(n)) + +extern int +nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, int count, + struct nouveau_notifier **); + +extern void +nouveau_notifier_free(struct nouveau_notifier **); + +extern void +nouveau_notifier_reset(struct nouveau_notifier *, int id); + +extern uint32_t +nouveau_notifier_status(struct nouveau_notifier *, int id); + +extern uint32_t +nouveau_notifier_return_val(struct nouveau_notifier *, int id); + +extern int +nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, + int timeout); + +struct nouveau_bo_priv { + struct nouveau_bo base; + + struct drm_nouveau_mem_alloc drm; + void *map; + + void *sysmem; + int user; + + int refcount; +}; + +struct nouveau_bo_reloc { + struct nouveau_bo_priv *bo; + uint32_t *ptr; + uint32_t flags; + uint32_t data, vor, tor; +}; + +#define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) + +extern int +nouveau_bo_init(struct nouveau_device *); + +extern void +nouveau_bo_takedown(struct nouveau_device *); + +extern int +nouveau_bo_new(struct nouveau_device *, uint32_t flags, int align, int size, + struct nouveau_bo **); + +extern int +nouveau_bo_user(struct nouveau_device *, void *ptr, int size, + struct nouveau_bo **); + +extern int +nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); + +extern int +nouveau_bo_resize(struct nouveau_bo *, int size); + +extern void +nouveau_bo_del(struct nouveau_bo **); + +extern int +nouveau_bo_map(struct nouveau_bo *, uint32_t flags); + +extern void +nouveau_bo_unmap(struct nouveau_bo *); + +extern void +nouveau_bo_emit_reloc(struct nouveau_channel *chan, void *ptr, + struct nouveau_bo *, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor); + +extern void +nouveau_bo_validate(struct nouveau_channel *); + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c new file mode 100644 index 00000000000..8dab2023956 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c @@ -0,0 +1,107 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" + +int +nouveau_grobj_alloc(struct nouveau_channel *userchan, uint32_t handle, + int class, struct nouveau_grobj **usergrobj) +{ + struct nouveau_device_priv *nv = nouveau_device(userchan->device); + struct nouveau_grobj_priv *gr; + struct drm_nouveau_grobj_alloc g; + int ret; + + if (!nv || !usergrobj || *usergrobj) + return -EINVAL; + + gr = calloc(1, sizeof(*gr)); + if (!gr) + return -ENOMEM; + gr->base.channel = userchan; + gr->base.handle = handle; + gr->base.grclass = class; + + g.channel = userchan->id; + g.handle = handle; + g.class = class; + ret = drmCommandWrite(nv->fd, DRM_NOUVEAU_GROBJ_ALLOC, &g, sizeof(g)); + if (ret) { + nouveau_grobj_free((void *)&gr); + return ret; + } + + *usergrobj = &gr->base; + return 0; +} + +int +nouveau_grobj_ref(struct nouveau_channel *userchan, uint32_t handle, + struct nouveau_grobj **usergr) +{ + struct nouveau_grobj_priv *gr; + + if (!userchan || !usergr || *usergr) + return -EINVAL; + + gr = calloc(1, sizeof(*gr)); + if (!gr) + return -ENOMEM; + gr->base.channel = userchan; + gr->base.handle = handle; + gr->base.grclass = 0; + + *usergr = &gr->base; + return 0; +} + +void +nouveau_grobj_free(struct nouveau_grobj **usergrobj) +{ + struct nouveau_grobj_priv *gr; + + if (!usergrobj) + return; + gr = nouveau_grobj(*usergrobj); + *usergrobj = NULL; + + if (gr) { + struct nouveau_channel_priv *chan; + struct nouveau_device_priv *nv; + struct drm_nouveau_gpuobj_free f; + + chan = nouveau_channel(gr->base.channel); + nv = nouveau_device(chan->base.device); + + if (gr->base.grclass) { + f.channel = chan->drm.channel; + f.handle = gr->base.handle; + drmCommandWrite(nv->fd, DRM_NOUVEAU_GPUOBJ_FREE, + &f, sizeof(f)); + } + free(gr); + } +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h new file mode 100644 index 00000000000..4ada5d014bf --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -0,0 +1,82 @@ +#ifndef __NOUVEAU_LOCAL_H__ +#define __NOUVEAU_LOCAL_H__ + +#include + +/* Debug output */ +#define NOUVEAU_MSG(fmt, args...) do { \ + fprintf(stdout, "nouveau: "fmt, ##args); \ + fflush(stdout); \ +} while(0) + +#define NOUVEAU_ERR(fmt, args...) do { \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); \ + fflush(stderr); \ +} while(0) + +#define NOUVEAU_TIME_MSEC() 0 + +/* User FIFO control */ +//#define NOUVEAU_DMA_TRACE +//#define NOUVEAU_DMA_DEBUG +//#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF +#define NOUVEAU_DMA_SUBCHAN_LRU +#define NOUVEAU_DMA_BARRIER +#define NOUVEAU_DMA_TIMEOUT 2000 + +/* Push buffer access macros */ +#define BEGIN_RING(obj,mthd,size) do { \ + BEGIN_RING_CH(nv->channel, nv->obj, (mthd), (size)); \ +} while(0) + +#define OUT_RING(data) do { \ + OUT_RING_CH(nv->channel, (data)); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + OUT_RINGp_CH(nv->channel, (src), (size)); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define WAIT_RING(size) do { \ + WAIT_RING_CH(nv->channel, (size)); \ +} while(0) + +#define FIRE_RING() do { \ + FIRE_RING_CH(nv->channel); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + struct nouveau_channel_priv *chan = nouveau_channel(nv->channel); \ + nouveau_bo_emit_reloc(nv->channel, &chan->pushbuf[chan->dma.cur], \ + (void*)(bo), (data), (flags), (vor), (tor)); \ + OUT_RING(0); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + nv->channel->vram->handle, nv->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c new file mode 100644 index 00000000000..ffe7d01d667 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c @@ -0,0 +1,103 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include "glheader.h" +#include "context.h" +#include "extensions.h" + +#include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" + +#include "drirenderbuffer.h" +#include "vblank.h" +#include "utils.h" +#include "xmlpool.h" /* for symbolic values of enum-type options */ + +#include "nouveau_context.h" +#include "nouveau_screen.h" + +_glthread_DECLARE_STATIC_MUTEX( lockMutex ); + +static void +nouveau_contended_lock(struct nouveau_context *nv, GLuint flags) +{ + __DRIdrawablePrivate *dPriv = nv->dri_drawable; + __DRIscreenPrivate *sPriv = nv->dri_screen; + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + drmGetLock(nvdev->fd, nvdev->ctx, flags); + + /* If the window moved, may need to set a new cliprect now. + * + * NOTE: This releases and regains the hw lock, so all state + * checking must be done *after* this call: + */ + if (dPriv) + DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv); +} + +/* Lock the hardware and validate our state. + */ +void +LOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + char __ret=0; + + _glthread_LOCK_MUTEX(lockMutex); + assert(!nv->locked); + + DRM_CAS(nvdev->lock, nvdev->ctx, + (DRM_LOCK_HELD | nvdev->ctx), __ret); + + if (__ret) + nouveau_contended_lock(nv, 0); + nv->locked = GL_TRUE; +} + + + /* Unlock the hardware using the global current context + */ +void +UNLOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + assert(nv->locked); + nv->locked = GL_FALSE; + + DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx); + + _glthread_UNLOCK_MUTEX(lockMutex); +} diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c new file mode 100644 index 00000000000..e423d59b1df --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c @@ -0,0 +1,140 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +#define NOTIFIER(__v) \ + struct nouveau_notifier_priv *notifier = nouveau_notifier(user); \ + volatile uint32_t *n = (void*)notifier->map + (id * 32) + +int +nouveau_notifier_alloc(struct nouveau_channel *userchan, uint32_t handle, + int count, struct nouveau_notifier **usernotifier) +{ + struct nouveau_notifier_priv *notifier; + int ret; + + if (!userchan || !usernotifier || *usernotifier) + return -EINVAL; + + notifier = calloc(1, sizeof(*notifier)); + if (!notifier) + return -ENOMEM; + notifier->base.channel = userchan; + notifier->base.handle = handle; + + notifier->drm.channel = userchan->id; + notifier->drm.handle = handle; + notifier->drm.count = count; + if ((ret = drmCommandWriteRead(nouveau_device(userchan->device)->fd, + DRM_NOUVEAU_NOTIFIEROBJ_ALLOC, + ¬ifier->drm, + sizeof(notifier->drm)))) { + nouveau_notifier_free((void *)¬ifier); + return ret; + } + + notifier->map = (void *)nouveau_channel(userchan)->notifier_block + + notifier->drm.offset; + *usernotifier = ¬ifier->base; + return 0; +} + +void +nouveau_notifier_free(struct nouveau_notifier **usernotifier) +{ + + struct nouveau_notifier_priv *notifier; + + if (!usernotifier) + return; + notifier = nouveau_notifier(*usernotifier); + *usernotifier = NULL; + + if (notifier) { + struct nouveau_channel_priv *chan; + struct nouveau_device_priv *nv; + struct drm_nouveau_gpuobj_free f; + + chan = nouveau_channel(notifier->base.channel); + nv = nouveau_device(chan->base.device); + + f.channel = chan->drm.channel; + f.handle = notifier->base.handle; + drmCommandWrite(nv->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); + free(notifier); + } +} + +void +nouveau_notifier_reset(struct nouveau_notifier *user, int id) +{ + NOTIFIER(n); + + n[NV_NOTIFY_TIME_0 /4] = 0x00000000; + n[NV_NOTIFY_TIME_1 /4] = 0x00000000; + n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000; + n[NV_NOTIFY_STATE /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS << + NV_NOTIFY_STATE_STATUS_SHIFT); +} + +uint32_t +nouveau_notifier_status(struct nouveau_notifier *user, int id) +{ + NOTIFIER(n); + + return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; +} + +uint32_t +nouveau_notifier_return_val(struct nouveau_notifier *user, int id) +{ + NOTIFIER(n); + + return n[NV_NOTIFY_RETURN_VALUE/4]; +} + +int +nouveau_notifier_wait_status(struct nouveau_notifier *user, int id, + int status, int timeout) +{ + NOTIFIER(n); + uint32_t time = 0, t_start = NOUVEAU_TIME_MSEC(); + + while (time <= timeout) { + uint32_t v; + + v = n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; + if (v == status) + return 0; + + if (timeout) + time = NOUVEAU_TIME_MSEC() - t_start; + } + + return -EBUSY; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c new file mode 100644 index 00000000000..48f583cbfb4 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -0,0 +1,253 @@ +#include "glheader.h" +#include "context.h" +#include "framebuffer.h" +#include "matrix.h" +#include "renderbuffer.h" +#include "simple_list.h" +#include "utils.h" +#include "vblank.h" +#include "xmlpool.h" + +#include "pipe/p_context.h" +#include "state_tracker/st_public.h" +#include "state_tracker/st_cb_fbo.h" + +#include "nouveau_context.h" +#include "nouveau_device.h" +#include "nouveau_drm.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10 +#error nouveau_drm.h version does not match expected version +#endif + +PUBLIC const char __driConfigOptions[] = +DRI_CONF_BEGIN +DRI_CONF_END; +static const GLuint __driNConfigOptions = 0; + +extern const struct dri_extension common_extensions[]; +extern const struct dri_extension nv40_extensions[]; + +static GLboolean +nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_dri *nv_dri = driScrnPriv->pDevPriv; + struct nouveau_screen *nv_screen; + int ret; + + if (driScrnPriv->devPrivSize != sizeof(struct nouveau_dri)) { + NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); + return GL_FALSE; + } + + nv_screen = CALLOC_STRUCT(nouveau_screen); + if (!nv_screen) + return GL_FALSE; + nv_screen->driScrnPriv = driScrnPriv; + driScrnPriv->private = (void *)nv_screen; + + driParseOptionInfo(&nv_screen->option_cache, + __driConfigOptions, __driNConfigOptions); + + if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, + driScrnPriv->fd, 0))) { + NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); + return GL_FALSE; + } + + nv_screen->front_offset = nv_dri->front_offset; + nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); + + return GL_TRUE; +} + +static void +nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_screen *nv_screen = driScrnPriv->private; + + driScrnPriv->private = NULL; + FREE(nv_screen); +} + +static GLboolean +nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, + __DRIdrawablePrivate * driDrawPriv, + const __GLcontextModes *glVis, GLboolean pixmapBuffer) +{ + struct nouveau_framebuffer *nvfb; + + if (pixmapBuffer) + return GL_FALSE; + + nvfb = CALLOC_STRUCT(nouveau_framebuffer); + if (!nvfb) + return GL_FALSE; + + nvfb->stfb = st_create_framebuffer(glVis, GL_TRUE, (void*)nvfb); + if (!nvfb->stfb) { + free(nvfb); + return GL_FALSE; + } + + driDrawPriv->driverPrivate = (void *)nvfb; + return GL_TRUE; +} + +static void +nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) +{ + struct nouveau_framebuffer *nvfb; + + nvfb = (struct nouveau_framebuffer *)driDrawPriv; + st_unreference_framebuffer(&nvfb->stfb); + free(nvfb); +} + +static struct __DriverAPIRec +nouveau_api = { + .InitDriver = nouveau_screen_create, + .DestroyScreen = nouveau_screen_destroy, + .CreateContext = nouveau_context_create, + .DestroyContext = nouveau_context_destroy, + .CreateBuffer = nouveau_create_buffer, + .DestroyBuffer = nouveau_destroy_buffer, + .SwapBuffers = nouveau_swap_buffers, + .MakeCurrent = nouveau_context_bind, + .UnbindContext = nouveau_context_unbind, + .GetSwapInfo = NULL, + .GetMSC = NULL, + .WaitForMSC = NULL, + .WaitForSBC = NULL, + .SwapBuffersMSC = NULL, + .CopySubBuffer = nouveau_copy_sub_buffer, + .setTexOffset = NULL +}; + +static __GLcontextModes * +nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, + unsigned stencil_bits, GLboolean have_back_buffer) +{ + __GLcontextModes * modes; + __GLcontextModes * m; + unsigned num_modes; + unsigned depth_buffer_factor; + unsigned back_buffer_factor; + int i; + + static const struct { + GLenum format; + GLenum type; + } fb_format_array[] = { + { GL_RGB , GL_UNSIGNED_SHORT_5_6_5 }, + { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV }, + { GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV }, + }; + + /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't + * support pageflipping at all. + */ + static const GLenum back_buffer_modes[] = { + GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML + }; + + u_int8_t depth_bits_array[4] = { 0, 16, 24 }; + u_int8_t stencil_bits_array[4] = { 0, 0, 8 }; + + depth_buffer_factor = 4; + back_buffer_factor = (have_back_buffer) ? 3 : 1; + + num_modes = ((pixel_bits==16) ? 1 : 2) * + depth_buffer_factor * back_buffer_factor * 4; + modes = (*dri_interface->createContextModes)(num_modes, + sizeof(__GLcontextModes)); + m = modes; + + for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) { + if (!driFillInModes(&m, fb_format_array[i].format, + fb_format_array[i].type, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + GLX_TRUE_COLOR)) { + fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__ ); + return NULL; + } + + if (!driFillInModes(&m, fb_format_array[i].format, + fb_format_array[i].type, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + GLX_DIRECT_COLOR)) { + fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__ ); + return NULL; + } + } + + return modes; +} +PUBLIC void * +__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, + __DRIscreen *psc, const __GLcontextModes * modes, + const __DRIversion * ddx_version, + const __DRIversion * dri_version, + const __DRIversion * drm_version, + const __DRIframebuffer * frame_buffer, + void * pSAREA, int fd, int internal_api_version, + const __DRIinterfaceMethods * interface, + __GLcontextModes ** driver_modes) +{ + __DRIscreenPrivate *psp; + static const __DRIversion ddx_expected = { 1, 2, 0 }; + static const __DRIversion dri_expected = { 4, 0, 0 }; + static const __DRIversion drm_expected = + { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; + struct nouveau_dri *nv_dri = NULL; + + dri_interface = interface; + + if (!driCheckDriDdxDrmVersions2("nouveau", + dri_version, &dri_expected, + ddx_version, &ddx_expected, + drm_version, &drm_expected)) { + return NULL; + } + + if (drm_expected.patch != drm_version->patch) { + fprintf(stderr, "Incompatible DRM patch level.\n" + "Expected: %d\n" "Current : %d\n", + drm_expected.patch, drm_version->patch); + return NULL; + } + + psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, + ddx_version, dri_version, drm_version, + frame_buffer, pSAREA, fd, + internal_api_version, + &nouveau_api); + if (psp == NULL) + return NULL; + nv_dri = psp->pDevPriv; + + *driver_modes = nouveau_fill_in_modes(nv_dri->bpp, + (nv_dri->bpp == 16) ? 16 : 24, + (nv_dri->bpp == 16) ? 0 : 8, + 1); + + driInitExtensions(NULL, common_extensions, GL_FALSE); + driInitExtensions(NULL, nv40_extensions, GL_FALSE); + + return (void *)psp; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h new file mode 100644 index 00000000000..e53cd873a7e --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h @@ -0,0 +1,17 @@ +#ifndef __NOUVEAU_SCREEN_H__ +#define __NOUVEAU_SCREEN_H__ + +#include "xmlconfig.h" +#include "nouveau_device.h" + +struct nouveau_screen { + __DRIscreenPrivate *driScrnPriv; + driOptionCache option_cache; + + struct nouveau_device *device; + + uint32_t front_offset; + uint32_t front_pitch; +}; + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c new file mode 100644 index 00000000000..57598f9cab9 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c @@ -0,0 +1,98 @@ +#include "context.h" + +#include "pipe/p_context.h" +#include "state_tracker/st_public.h" +#include "state_tracker/st_cb_fbo.h" + +#include "nouveau_context.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +void +nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, + const drm_clip_rect_t *rect) +{ + struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; + struct nouveau_screen *nv_screen = nv->nv_screen; + struct pipe_region *p_region = surf->region; + drm_clip_rect_t *pbox; + int nbox, i; + + LOCK_HARDWARE(nv); + if (!dPriv->numClipRects) { + UNLOCK_HARDWARE(nv); + return; + } + pbox = dPriv->pClipRects; + nbox = dPriv->numClipRects; + + BEGIN_RING(NvCtxSurf2D, 0x184, 2); + OUT_RELOCo(p_region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RING (nv->channel->vram->handle); + + BEGIN_RING(NvCtxSurf2D, 0x300, 4); + OUT_RING ((p_region->cpp == 4) ? 6 : 4); + OUT_RING ((nv_screen->front_pitch << 16) | + (p_region->pitch * p_region->cpp)); + OUT_RELOCl(p_region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RING (nv_screen->front_offset); + + for (i = 0; i < nbox; i++, pbox++) { + int sx, sy, dx, dy, w, h; + + sx = pbox->x1 - dPriv->x; + sy = pbox->y1 - dPriv->y; + dx = pbox->x1; + dy = pbox->y1; + w = pbox->x2 - pbox->x1; + h = pbox->y2 - pbox->y1; + + BEGIN_RING(NvImageBlit, 0x300, 3); + OUT_RING ((sy << 16) | sx); + OUT_RING ((dy << 16) | dx); + OUT_RING (( h << 16) | w); + } + + FIRE_RING(); + UNLOCK_HARDWARE(nv); + + if (nv->last_stamp != dPriv->lastStamp) { + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + st_resize_framebuffer(nvfb->stfb, dPriv->w, dPriv->h); + nv->last_stamp = dPriv->lastStamp; + } +} + +void +nouveau_copy_sub_buffer(__DRIdrawablePrivate *dPriv, int x, int y, int w, int h) +{ + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + struct pipe_surface *surf; + + surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + drm_clip_rect_t rect; + rect.x1 = x; + rect.y1 = y; + rect.x2 = x + w; + rect.y2 = y + h; + + st_notify_swapbuffers(nvfb->stfb); + nouveau_copy_buffer(dPriv, surf, &rect); + } +} + +void +nouveau_swap_buffers(__DRIdrawablePrivate *dPriv) +{ + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + struct pipe_surface *surf; + + surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + st_notify_swapbuffers(nvfb->stfb); + nouveau_copy_buffer(dPriv, surf, NULL); + } +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h new file mode 100644 index 00000000000..825d3da6da5 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h @@ -0,0 +1,10 @@ +#ifndef __NOUVEAU_SWAPBUFFERS_H__ +#define __NOUVEAU_SWAPBUFFERS_H__ + +extern void nouveau_copy_buffer(__DRIdrawablePrivate *, struct pipe_surface *, + const drm_clip_rect_t *); +extern void nouveau_copy_sub_buffer(__DRIdrawablePrivate *, + int x, int y, int w, int h); +extern void nouveau_swap_buffers(__DRIdrawablePrivate *); + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c new file mode 100644 index 00000000000..a2e85a0aa7f --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -0,0 +1,90 @@ +#include "nouveau_context.h" +#include "nouveau_winsys_pipe.h" + +#include "pipe/nouveau/nouveau_winsys.h" + +static int +nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, + struct nouveau_notifier **notify) +{ + struct nouveau_context *nv = nvws->nv; + + return nouveau_notifier_alloc(nv->channel, nv->next_handle++, + count, notify); +} + +static int +nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, + struct nouveau_grobj **grobj) +{ + struct nouveau_context *nv = nvws->nv; + + return nouveau_grobj_alloc(nv->channel, nv->next_handle++, + grclass, grobj); +} + +static uint32_t * +nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) +{ + struct nouveau_channel_priv *chan = nouveau_channel(grobj->channel); + uint32_t *pushbuf; + + BEGIN_RING_CH(&chan->base, grobj, mthd, size); + pushbuf = &chan->pushbuf[chan->dma.cur]; + chan->dma.cur += size; +#ifdef NOUVEAU_DMA_DEBUG + chan->dma.push_free -= size; +#endif + return pushbuf; +} + +static void +nouveau_pipe_dma_kickoff(struct nouveau_channel *userchan) +{ + FIRE_RING_CH(userchan); +} + +struct pipe_context * +nouveau_pipe_create(struct nouveau_context *nv) +{ + struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); + struct pipe_context *(*hw_create)(struct pipe_winsys *, + struct nouveau_winsys *, + unsigned); + + if (!nvws) + return NULL; + + switch (nv->chipset & 0xf0) { + case 0x40: + hw_create = nv40_create; + break; + default: + NOUVEAU_ERR("Unknown chipset NV%02x\n", (int)nv->chipset); + return NULL; + } + + nvws->nv = nv; + nvws->channel = nv->channel; + + nvws->begin_ring = nouveau_pipe_dma_beginp; + nvws->out_reloc = nouveau_bo_emit_reloc; + nvws->fire_ring = nouveau_pipe_dma_kickoff; + + nvws->grobj_alloc = nouveau_pipe_grobj_alloc; + nvws->grobj_free = nouveau_grobj_free; + + nvws->notifier_alloc = nouveau_pipe_notifier_alloc; + nvws->notifier_free = nouveau_notifier_free; + nvws->notifier_reset = nouveau_notifier_reset; + nvws->notifier_status = nouveau_notifier_status; + nvws->notifier_retval = nouveau_notifier_return_val; + nvws->notifier_wait = nouveau_notifier_wait_status; + + nvws->region_copy = nv->region_copy; + nvws->region_fill = nv->region_fill; + nvws->region_data = nv->region_data; + + return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c new file mode 100644 index 00000000000..0dd2cec2ca1 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -0,0 +1,251 @@ +#include "pipe/p_winsys.h" +#include "pipe/p_defines.h" + +#include "nouveau_context.h" +#include "nouveau_device.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" +#include "nouveau_winsys_pipe.h" + +static void +nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, + void *context_private) +{ + struct nouveau_context *nv = context_private; + __DRIdrawablePrivate *dPriv = nv->dri_drawable; + + nouveau_copy_buffer(dPriv, surf, NULL); +} + +static void +nouveau_printf(struct pipe_winsys *pws, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + vfprintf(stderr, fmt, args); + va_end(args); +} + +static const char * +nouveau_get_name(struct pipe_winsys *pws) +{ + return "Nouveau/DRI"; +} + +static struct pipe_region * +nouveau_region_alloc(struct pipe_winsys *ws, unsigned cpp, + unsigned width, unsigned height, unsigned flags) +{ + struct pipe_region *region; + + region = calloc(sizeof(*region), 1); + region->cpp = cpp; + region->pitch = ((cpp * width + 63) & ~63) / cpp; + region->height = height; + region->refcount = 1; + region->buffer = ws->buffer_create(ws, 64); + + ws->buffer_data(ws, region->buffer, region->pitch * cpp * height, NULL, + PIPE_BUFFER_USAGE_PIXEL); + return region; +} + +static void +nouveau_region_release(struct pipe_winsys *pws, struct pipe_region **pregion) +{ + struct pipe_region *region; + + if (!pregion || !*pregion) + return; + region = *pregion; + *pregion = NULL; + + assert(region->refcount > 0); + region->refcount--; + + if (region->refcount == 0) { + assert(region->map_refcount == 0); + pws->buffer_reference(pws, ®ion->buffer, NULL); + free(region); + } +} + +static struct pipe_surface * +nouveau_surface_alloc(struct pipe_winsys *pws, unsigned format) +{ + struct pipe_surface *surf; + + surf = CALLOC_STRUCT(pipe_surface); + if (!surf) + return NULL; + + surf->format = format; + surf->refcount = 1; + surf->winsys = pws; + + return surf; +} + +static void +nouveau_surface_release(struct pipe_winsys *pws, struct pipe_surface **s) +{ + struct pipe_surface *surf = *s; *s = NULL; + + if (surf->refcount-- == 0) { + if (surf->region) + pws->region_release(pws, &surf->region); + free(surf); + } +} + +static struct pipe_buffer_handle * +nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_bo *nvbo = NULL; + + if (nouveau_bo_new(dev, NOUVEAU_BO_LOCAL, alignment, 0, &nvbo)) + return NULL; + return (struct pipe_buffer_handle *)nvbo; +} + +static struct pipe_buffer_handle * +nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_bo *nvbo = NULL; + + if (nouveau_bo_user(dev, ptr, bytes, &nvbo)) + return NULL; + return (struct pipe_buffer_handle *)nvbo; +} + +static void * +nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, + unsigned flags) +{ + struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + uint32_t map_flags = 0; + + if (flags & PIPE_BUFFER_FLAG_READ) + map_flags |= NOUVEAU_BO_RD; + if (flags & PIPE_BUFFER_FLAG_WRITE) + map_flags |= NOUVEAU_BO_WR; + + if (nouveau_bo_map(nvbo, map_flags)) + return NULL; + return nvbo->map; +} + +static void +nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer_handle *bo) +{ + struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + + nouveau_bo_unmap(nvbo); +} + +static void +nouveau_pipe_bo_reference(struct pipe_winsys *pws, + struct pipe_buffer_handle **ptr, + struct pipe_buffer_handle *bo) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_context *nv = nvpws->nv; + struct nouveau_device *dev = nv->nv_screen->device; + + if (*ptr) { + struct nouveau_bo *nvbo = (struct nouveau_bo *)*ptr; + FIRE_RING(); + nouveau_bo_del(&nvbo); + *ptr = NULL; + } + + if (bo) { + struct nouveau_bo *nvbo = (struct nouveau_bo *)bo, *new = NULL; + nouveau_bo_ref(dev, nvbo->handle, &new); + *ptr = bo; + } +} + +static void +nouveau_pipe_bo_data(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, + unsigned size, const void *data, unsigned usage) +{ + struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + + if (nvbo->size != size) + nouveau_bo_resize(nvbo, size); + + if (data) { + if (nouveau_bo_map(nvbo, NOUVEAU_BO_WR)) + return; + memcpy(nvbo->map, data, size); + nouveau_bo_unmap(nvbo); + } +} + +static void +nouveau_pipe_bo_subdata(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, + unsigned long offset, unsigned long size, + const void *data) +{ + struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + + if (nouveau_bo_map(nvbo, NOUVEAU_BO_WR)) + return; + memcpy(nvbo->map + offset, data, size); + nouveau_bo_unmap(nvbo); +} + +static void +nouveau_pipe_bo_get_subdata(struct pipe_winsys *pws, + struct pipe_buffer_handle *bo, unsigned long offset, + unsigned long size, void *data) +{ + struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + + if (nouveau_bo_map(nvbo, NOUVEAU_BO_RD)) + return; + memcpy(data, nvbo->map + offset, size); + nouveau_bo_unmap(nvbo); +} + +struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv) +{ + struct nouveau_pipe_winsys *nvpws; + struct pipe_winsys *pws; + + nvpws = CALLOC_STRUCT(nouveau_pipe_winsys); + if (!nvpws) + return NULL; + nvpws->nv = nv; + pws = &nvpws->pws; + + pws->flush_frontbuffer = nouveau_flush_frontbuffer; + pws->printf = nouveau_printf; + + pws->region_alloc = nouveau_region_alloc; + pws->region_release = nouveau_region_release; + + pws->surface_alloc = nouveau_surface_alloc; + pws->surface_release = nouveau_surface_release; + + pws->buffer_create = nouveau_pipe_bo_create; + pws->user_buffer_create = nouveau_pipe_bo_user_create; + pws->buffer_map = nouveau_pipe_bo_map; + pws->buffer_unmap = nouveau_pipe_bo_unmap; + pws->buffer_reference = nouveau_pipe_bo_reference; + pws->buffer_data = nouveau_pipe_bo_data; + pws->buffer_subdata = nouveau_pipe_bo_subdata; + pws->buffer_get_subdata= nouveau_pipe_bo_get_subdata; + + pws->get_name = nouveau_get_name; + + return &nvpws->pws; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h new file mode 100644 index 00000000000..bcac60851cb --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h @@ -0,0 +1,23 @@ +#ifndef NOUVEAU_PIPE_WINSYS_H +#define NOUVEAU_PIPE_WINSYS_H + +#include "pipe/p_context.h" +#include "pipe/p_winsys.h" +#include "nouveau_context.h" + +struct nouveau_pipe_winsys { + struct pipe_winsys pws; + + struct nouveau_context *nv; +}; + +extern struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv); + +struct pipe_context * +nouveau_create_softpipe(struct nouveau_context *nv); + +struct pipe_context * +nouveau_pipe_create(struct nouveau_context *nv); + +#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c new file mode 100644 index 00000000000..c86575f1875 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c @@ -0,0 +1,82 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * + **************************************************************************/ +/* + * Authors: Keith Whitwell + */ + +#include "imports.h" + +#include "pipe/softpipe/sp_winsys.h" +#include "pipe/p_defines.h" + +#include "nouveau_context.h" +#include "nouveau_winsys_pipe.h" + +struct nouveau_softpipe_winsys { + struct softpipe_winsys sws; + struct nouveau_context *nv; +}; + +/** + * Return list of surface formats supported by this driver. + */ +static boolean +nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) +{ + switch (format) { + case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_U_R5_G6_B5: + case PIPE_FORMAT_S8_Z24: + return TRUE; + default: + break; + }; + + return FALSE; +} + + + +struct pipe_context * +nouveau_create_softpipe(struct nouveau_context *nv) +{ + struct nouveau_softpipe_winsys *nvsws; + + nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); + + /* Fill in this struct with callbacks that softpipe will need to + * communicate with the window system, buffer manager, etc. + */ + nvsws->sws.is_format_supported = nouveau_is_format_supported; + nvsws->nv = nv; + + /* Create the softpipe context: + */ + return softpipe_create(nouveau_create_pipe_winsys(nv), &nvsws->sws); +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c new file mode 100644 index 00000000000..cbe70983f5c --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c @@ -0,0 +1,163 @@ +#include "pipe/p_context.h" + +#include "nouveau_context.h" + +static INLINE int +nv04_surface_format(int cpp) +{ + switch (cpp) { + case 1: return 0x1; + case 2: return 0x4; + case 4: return 0xb; + default: + return -1; + } +} + +static INLINE int +nv04_rect_format(int cpp) +{ + switch (cpp) { + case 1: return 0x3; + case 2: return 0x1; + case 4: return 0x3; + default: + return -1; + } +} + +static int +nv04_region_display(void) +{ + NOUVEAU_ERR("unimplemented\n"); + return 0; +} + +static int +nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + struct pipe_region *src, unsigned src_offset, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + int format; + + if (src->cpp != dst->cpp) + return 1; + + if ((format = nv04_surface_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + BEGIN_RING(NvCtxSurf2D, 0x0184, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(NvCtxSurf2D, 0x0300, 4); + OUT_RING (format); + OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); + OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(NvImageBlit, 0x0300, 3); + OUT_RING ((sy << 16) | sx); + OUT_RING ((dy << 16) | dx); + OUT_RING (( h << 16) | w); + + return 0; +} + +static int +nv04_region_fill(struct nouveau_context *nv, + struct pipe_region *dst, unsigned dst_offset, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + int cs2d_format, gdirect_format; + + if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + if ((gdirect_format = nv04_rect_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + BEGIN_RING(NvCtxSurf2D, 0x0184, 2); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(NvCtxSurf2D, 0x0300, 4); + OUT_RING (cs2d_format); + OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); + OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(NvGdiRect, 0x0300, 1); + OUT_RING (gdirect_format); + BEGIN_RING(NvGdiRect, 0x03fc, 1); + OUT_RING (value); + BEGIN_RING(NvGdiRect, 0x0400, 2); + OUT_RING ((dx << 16) | dy); + OUT_RING (( w << 16) | h); + + return 0; +} + +static int +nv04_region_data(struct nouveau_context *nv, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + const void *src, unsigned src_pitch, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + NOUVEAU_ERR("unimplemented!!\n"); + return 0; +} + +int +nouveau_region_init_nv04(struct nouveau_context *nv) +{ + int ret; + + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x62, + &nv->NvCtxSurf2D))) { + NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); + return 1; + } + BEGIN_RING(NvCtxSurf2D, 0x0184, 2); + OUT_RING (nv->channel->vram->handle); + OUT_RING (nv->channel->vram->handle); + + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x9f, + &nv->NvImageBlit))) { + NOUVEAU_ERR("Error creating blit object: %d\n", ret); + return 1; + } + BEGIN_RING(NvImageBlit, 0x0180, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(NvImageBlit, 0x019c, 1); + OUT_RING (nv->NvCtxSurf2D->handle); + BEGIN_RING(NvImageBlit, 0x02fc, 1); + OUT_RING (3); + + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x4a, + &nv->NvGdiRect))) { + NOUVEAU_ERR("Error creating rect object: %d\n", ret); + return 1; + } + BEGIN_RING(NvGdiRect, 0x0180, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(NvGdiRect, 0x0198, 1); + OUT_RING (nv->NvCtxSurf2D->handle); + BEGIN_RING(NvGdiRect, 0x02fc, 1); + OUT_RING (3); + BEGIN_RING(NvGdiRect, 0x0304, 1); + OUT_RING (2); + + nv->region_display = nv04_region_display; + nv->region_copy = nv04_region_copy; + nv->region_fill = nv04_region_fill; + nv->region_data = nv04_region_data; + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c new file mode 100644 index 00000000000..c7450c5c8d0 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c @@ -0,0 +1,51 @@ +#include "pipe/p_context.h" + +#include "nouveau_context.h" + +static int +nv50_region_display(void) +{ + NOUVEAU_ERR("unimplemented\n"); + return 0; +} + +static int +nv50_region_copy(struct nouveau_context *nv, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + struct pipe_region *src, unsigned src_offset, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + NOUVEAU_ERR("unimplemented!!\n"); + return 0; +} + +static int +nv50_region_fill(struct nouveau_context *nv, + struct pipe_region *dst, unsigned dst_offset, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + NOUVEAU_ERR("unimplemented!!\n"); + return 0; +} + +static int +nv50_region_data(struct nouveau_context *nv, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + const void *src, unsigned src_pitch, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + NOUVEAU_ERR("unimplemented!!\n"); + return 0; +} + +int +nouveau_region_init_nv50(struct nouveau_context *nv) +{ + nv->region_display = nv50_region_display; + nv->region_copy = nv50_region_copy; + nv->region_fill = nv50_region_fill; + nv->region_data = nv50_region_data; + return 0; +} + diff --git a/src/mesa/pipe/Makefile b/src/mesa/pipe/Makefile index d448f3f5a6a..135f1fd19d7 100644 --- a/src/mesa/pipe/Makefile +++ b/src/mesa/pipe/Makefile @@ -2,6 +2,7 @@ default: cd softpipe ; make cd i915simple ; make cd failover ; make + cd nv40; make clean: - rm -f `find . -name \*.[oa]` \ No newline at end of file + rm -f `find . -name \*.[oa]` diff --git a/src/mesa/pipe/nouveau/nouveau_bo.h b/src/mesa/pipe/nouveau/nouveau_bo.h new file mode 100644 index 00000000000..2b57ee92633 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_bo.h @@ -0,0 +1,50 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_BO_H__ +#define __NOUVEAU_BO_H__ + +/* Relocation/Buffer type flags */ +#define NOUVEAU_BO_VRAM (1 << 0) +#define NOUVEAU_BO_GART (1 << 1) +#define NOUVEAU_BO_RD (1 << 2) +#define NOUVEAU_BO_WR (1 << 3) +#define NOUVEAU_BO_RDWR (NOUVEAU_BO_RD | NOUVEAU_BO_WR) +#define NOUVEAU_BO_MAP (1 << 4) +#define NOUVEAU_BO_PIN (1 << 5) +#define NOUVEAU_BO_LOW (1 << 6) +#define NOUVEAU_BO_HIGH (1 << 7) +#define NOUVEAU_BO_OR (1 << 8) +#define NOUVEAU_BO_LOCAL (1 << 9) + +struct nouveau_bo { + struct nouveau_device *device; + uint64_t handle; + + uint64_t size; + void *map; + + uint32_t flags; + uint64_t offset; +}; + +#endif diff --git a/src/mesa/pipe/nouveau/nouveau_channel.h b/src/mesa/pipe/nouveau/nouveau_channel.h new file mode 100644 index 00000000000..eccc7b3bd4d --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_channel.h @@ -0,0 +1,37 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_CHANNEL_H__ +#define __NOUVEAU_CHANNEL_H__ + +struct nouveau_channel { + struct nouveau_device *device; + int id; + + struct nouveau_grobj *vram; + struct nouveau_grobj *gart; + + void *user_private; + void (*hang_notify)(struct nouveau_channel *); +}; + +#endif diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h new file mode 100644 index 00000000000..94b667e44d2 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -0,0 +1,5261 @@ +#ifndef NOUVEAU_REG_H +#define NOUVEAU_REG_H 1 + + +#define NV01_ROOT 0x00000001 + + + +#define NV01_CONTEXT_DMA 0x00000002 + + + +#define NV01_DEVICE 0x00000003 + + + +#define NV01_TIMER 0x00000004 + +#define NV01_TIMER_SYNCHRONIZE 0x00000100 +#define NV01_TIMER_STOP_ALARM 0x00000104 +#define NV01_TIMER_DMA_NOTIFY 0x00000180 +#define NV01_TIMER_TIME(x) (0x00000300+((x)*4)) +#define NV01_TIMER_TIME__SIZE 0x00000002 +#define NV01_TIMER_ALARM_NOTIFY 0x00000308 + + +#define NV_IMAGE_STENCIL 0x00000010 + +#define NV_IMAGE_STENCIL_NOTIFY 0x00000104 +#define NV_IMAGE_STENCIL_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_STENCIL_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_STENCIL_IMAGE_INPUT(x) (0x00000204+((x)*4)) +#define NV_IMAGE_STENCIL_IMAGE_INPUT__SIZE 0x00000002 + + +#define NV_IMAGE_BLEND_AND 0x00000011 + +#define NV_IMAGE_BLEND_AND_NOP 0x00000100 +#define NV_IMAGE_BLEND_AND_NOTIFY 0x00000104 +#define NV_IMAGE_BLEND_AND_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_BLEND_AND_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_BLEND_AND_BETA_INPUT 0x00000204 +#define NV_IMAGE_BLEND_AND_IMAGE_INPUT 0x00000208 + + +#define NV01_CONTEXT_BETA1 0x00000012 + +#define NV01_CONTEXT_BETA1_NOP 0x00000100 +#define NV01_CONTEXT_BETA1_NOTIFY 0x00000104 +#define NV01_CONTEXT_BETA1_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_BETA1_BETA_1D31 0x00000300 + + +#define NV_IMAGE_ROP_AND 0x00000013 + +#define NV_IMAGE_ROP_AND_NOTIFY 0x00000104 +#define NV_IMAGE_ROP_AND_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_ROP_AND_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_ROP_AND_ROP_INPUT 0x00000204 +#define NV_IMAGE_ROP_AND_IMAGE_INPUT(x) (0x00000208+((x)*4)) +#define NV_IMAGE_ROP_AND_IMAGE_INPUT__SIZE 0x00000002 + + +#define NV_IMAGE_COLOR_KEY 0x00000015 + + + +#define NV01_CONTEXT_COLOR_KEY 0x00000017 + +#define NV01_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 +#define NV01_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A8Y8 0x00000001 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X24Y8 0x00000002 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A1R5G5B5 0x00000003 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X17R5G5B5 0x00000004 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A8R8G8B8 0x00000005 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X8R8G8B8 0x00000006 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A16Y16 0x00000007 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16Y16 0x00000008 +#define NV01_CONTEXT_COLOR_KEY_COLOR 0x00000304 + + +#define NV01_CONTEXT_PATTERN 0x00000018 + +#define NV01_CONTEXT_PATTERN_NOTIFY 0x00000104 +#define NV01_CONTEXT_PATTERN_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_PATTERN_COLOR_FORMAT 0x00000300 +#define NV01_CONTEXT_PATTERN_MONOCHROME_FORMAT 0x00000304 +#define NV01_CONTEXT_PATTERN_SHAPE 0x00000308 +#define NV01_CONTEXT_PATTERN_COLOR(x) (0x00000310+((x)*4)) +#define NV01_CONTEXT_PATTERN_COLOR__SIZE 0x00000002 +#define NV01_CONTEXT_PATTERN_PATTERN(x) (0x00000318+((x)*4)) +#define NV01_CONTEXT_PATTERN_PATTERN__SIZE 0x00000002 + + +#define NV01_CONTEXT_CLIP_RECTANGLE 0x00000019 + +#define NV01_CONTEXT_CLIP_RECTANGLE_NOP 0x00000104 +#define NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY 0x00000104 +#define NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT 0x00000300 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_SHIFT 0 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_MASK 0x0000ffff +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_SHIFT 16 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_MASK 0xffff0000 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE 0x00000304 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_SHIFT 0 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_MASK 0x0000ffff +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_SHIFT 16 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_LINE 0x0000001c + +#define NV01_RENDER_SOLID_LINE_NOP 0x00000104 +#define NV01_RENDER_SOLID_LINE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_LINE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_LINE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_LINE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_LINE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_LINE_ROP 0x0000018c +#define NV01_RENDER_SOLID_LINE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_LINE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_LINE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_LINE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_PREMULT 0x00000003 +#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A8Y8 0x00000001 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X24Y8 0x00000002 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A1R5G5B5 0x00000003 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X17R5G5B5 0x00000004 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A8R8G8B8 0x00000005 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X8R8G8B8 0x00000006 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A16Y16 0x00000007 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16Y16 0x00000008 +#define NV01_RENDER_SOLID_LINE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0(x) (0x00000400+((x)*8)) +#define NV01_RENDER_SOLID_LINE_LINE_POINT0__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1(x) (0x00000404+((x)*8)) +#define NV01_RENDER_SOLID_LINE_LINE_POINT1__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X(x) (0x00000480+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y(x) (0x00000484+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X(x) (0x00000488+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y(x) (0x0000048c+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_POLYLINE(x) (0x00000500+((x)*4)) +#define NV01_RENDER_SOLID_LINE_POLYLINE__SIZE 0x00000020 +#define NV01_RENDER_SOLID_LINE_POLYLINE_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_POLYLINE_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X(x) (0x00000580+((x)*8)) +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y(x) (0x00000584+((x)*8)) +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR(x) (0x00000600+((x)*8)) +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT(x) (0x00000604+((x)*8)) +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_TRIANGLE 0x0000001d + +#define NV01_RENDER_SOLID_TRIANGLE_NOP 0x00000100 +#define NV01_RENDER_SOLID_TRIANGLE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_TRIANGLE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_TRIANGLE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_TRIANGLE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_TRIANGLE_ROP 0x0000018c +#define NV01_RENDER_SOLID_TRIANGLE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_TRIANGLE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_PREMULT 0x00000003 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_TRIANGLE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_TRIANGLE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0 0x00000310 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1 0x00000314 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2 0x00000318 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_X 0x00000320 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_Y 0x00000324 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_X 0x00000328 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_Y 0x0000032c +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_X 0x00000330 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_Y 0x00000334 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH(x) (0x00000400+((x)*4)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH__SIZE 0x00000020 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X(x) (0x00000480+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y(x) (0x00000484+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR(x) (0x00000500+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0(x) (0x00000504+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1(x) (0x00000508+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2(x) (0x0000050c+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR(x) (0x00000580+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT(x) (0x00000584+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_RECTANGLE 0x0000001e + +#define NV01_RENDER_SOLID_RECTANGLE_NOP 0x00000100 +#define NV01_RENDER_SOLID_RECTANGLE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_RECTANGLE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_RECTANGLE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_RECTANGLE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_RECTANGLE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_RECTANGLE_ROP 0x0000018c +#define NV01_RENDER_SOLID_RECTANGLE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_RECTANGLE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_PREMULT 0x00000003 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_RECTANGLE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_RECTANGLE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT(x) (0x00000400+((x)*8)) +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE__SIZE 0x00000010 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_SHIFT 0 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_MASK 0x0000ffff +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_SHIFT 16 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_MASK 0xffff0000 + + +#define NV01_IMAGE_BLIT 0x0000001f + +#define NV01_IMAGE_BLIT_NOP 0x00000100 +#define NV01_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV01_IMAGE_BLIT_PATCH 0x0000010c +#define NV01_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV01_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV01_IMAGE_BLIT_PATTERN 0x0000018c +#define NV01_IMAGE_BLIT_ROP 0x00000190 +#define NV01_IMAGE_BLIT_BETA1 0x00000194 +#define NV01_IMAGE_BLIT_SURFACE 0x0000019c +#define NV01_IMAGE_BLIT_OPERATION 0x000002fc +#define NV01_IMAGE_BLIT_IMAGE_INPUT 0x00000204 +#define NV01_IMAGE_BLIT_POINT_IN 0x00000300 +#define NV01_IMAGE_BLIT_POINT_IN_X_SHIFT 0 +#define NV01_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 +#define NV01_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 +#define NV01_IMAGE_BLIT_POINT_OUT 0x00000304 +#define NV01_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 +#define NV01_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 +#define NV01_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 +#define NV01_IMAGE_BLIT_SIZE 0x00000308 +#define NV01_IMAGE_BLIT_SIZE_W_SHIFT 0 +#define NV01_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_SIZE_H_SHIFT 16 +#define NV01_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 + + +#define NV01_IMAGE_FROM_CPU 0x00000021 + +#define NV01_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV01_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV01_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV01_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV01_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 +#define NV01_IMAGE_FROM_CPU_PATTERN 0x0000018c +#define NV01_IMAGE_FROM_CPU_ROP 0x00000190 +#define NV01_IMAGE_FROM_CPU_BETA1 0x00000194 +#define NV01_IMAGE_FROM_CPU_SURFACE 0x00000198 +#define NV01_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_IMAGE_FROM_CPU_OPERATION_ROP_AND 0x00000001 +#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_AND 0x00000002 +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_PREMULT 0x00000003 +#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_PREMULT 0x00000004 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_Y8 0x00000001 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A1R5G5B5 0x00000002 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X1R5G5B5 0x00000003 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A8R8G8B8 0x00000004 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X8R8G8B8 0x00000005 +#define NV01_IMAGE_FROM_CPU_POINT 0x00000304 +#define NV01_IMAGE_FROM_CPU_POINT_X_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_POINT_X_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_POINT_Y_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_POINT_Y_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_SIZE_IN 0x0000030c +#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV01_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 + + +#define NV01_NULL 0x00000030 + + + +#define NV03_STRETCHED_IMAGE_FROM_CPU 0x00000036 + +#define NV03_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV03_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV03_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV03_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV03_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 +#define NV03_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c +#define NV03_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000194 +#define NV03_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_DX_DU 0x00000308 +#define NV03_STRETCHED_IMAGE_FROM_CPU_DY_DV 0x0000030c +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT 0x00000310 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE 0x00000314 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 + + +#define NV03_SCALED_IMAGE_FROM_MEMORY 0x00000037 + +#define NV03_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 +#define NV03_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 +#define NV03_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 +#define NV03_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c +#define NV03_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 +#define NV03_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000194 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000003 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000004 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT 0x00000310 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE 0x00000314 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DU_DX 0x00000318 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DV_DY 0x0000031c +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE 0x00000400 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT 0x00000404 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_MASK 0x00ff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CENTER 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CORNER 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_SHIFT 24 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_MASK 0xff000000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_OFFSET 0x00000408 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT 0x0000040c +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_MASK 0xffff0000 + + +#define NV04_DVD_SUBPICTURE 0x00000038 + +#define NV04_DVD_SUBPICTURE_NOP 0x00000100 +#define NV04_DVD_SUBPICTURE_NOTIFY 0x00000104 +#define NV04_DVD_SUBPICTURE_WAIT_FOR_IDLE 0x00000108 +#define NV04_DVD_SUBPICTURE_DMA_NOTIFY 0x00000180 +#define NV04_DVD_SUBPICTURE_DMA_OVERLAY 0x00000184 +#define NV04_DVD_SUBPICTURE_DMA_IMAGEIN 0x00000188 +#define NV04_DVD_SUBPICTURE_DMA_IMAGEOUT 0x0000018c +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT 0x00000300 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE 0x00000304 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT 0x00000308 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_OFFSET 0x0000030c +#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DU_DX 0x00000310 +#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DV_DY 0x00000314 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE 0x00000318 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT 0x0000031c +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEIN_OFFSET 0x00000320 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT 0x00000324 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DU_DX 0x00000328 +#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DV_DY 0x0000032c +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE 0x00000330 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT 0x00000334 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_OFFSET 0x00000338 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT 0x0000033c +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK 0xffff0000 + + +#define NV_MEMORY_TO_MEMORY_FORMAT 0x00000039 + +#define NV_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 +#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 +#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c +#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 +#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 +#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 +#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c +#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 +#define NV_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 + + +#define NV01_MEMORY_LOCAL_BANKED 0x0000003d + + + +#define NV01_MAPPING_SYSTEM 0x0000003e + + + +#define NV03_MEMORY_LOCAL_CURSOR 0x0000003f + + + +#define NV01_MEMORY_LOCAL_LINEAR 0x00000040 + + + +#define NV01_MAPPING_LOCAL 0x00000041 + + + +#define NV04_CONTEXT_SURFACES_2D 0x00000042 + +#define NV04_CONTEXT_SURFACES_2D_NOP 0x00000100 +#define NV04_CONTEXT_SURFACES_2D_NOTIFY 0x00000104 +#define NV04_CONTEXT_SURFACES_2D_PM_TRIGGER 0x00000140 +#define NV04_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE 0x00000184 +#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_DESTIN 0x00000188 +#define NV04_CONTEXT_SURFACES_2D_FORMAT 0x00000300 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y8 0x00000001 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_Z1R5G5B5 0x00000002 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_X1R5G5B5 0x00000003 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5 0x00000004 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y16 0x00000005 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_Z8R8G8B8 0x00000006 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_X8R8G8B8 0x00000007 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_X1A7R8G8B8 0x00000009 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8 0x0000000a +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y32 0x0000000b +#define NV04_CONTEXT_SURFACES_2D_PITCH 0x00000304 +#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 +#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 +#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 +#define NV04_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c + + +#define NV03_CONTEXT_ROP 0x00000043 + +#define NV03_CONTEXT_ROP_NOP 0x00000100 +#define NV03_CONTEXT_ROP_NOTIFY 0x00000104 +#define NV03_CONTEXT_ROP_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_ROP_ROP 0x00000300 +#define NV03_CONTEXT_ROP_ROP_VALUE0_SHIFT 0 +#define NV03_CONTEXT_ROP_ROP_VALUE0_MASK 0x0000000f +#define NV03_CONTEXT_ROP_ROP_VALUE0_CLEAR 0x00000000 +#define NV03_CONTEXT_ROP_ROP_VALUE0_NOR 0x00000001 +#define NV03_CONTEXT_ROP_ROP_VALUE0_AND_INVERTED 0x00000002 +#define NV03_CONTEXT_ROP_ROP_VALUE0_COPY_INVERTED 0x00000003 +#define NV03_CONTEXT_ROP_ROP_VALUE0_AND_REVERSE 0x00000004 +#define NV03_CONTEXT_ROP_ROP_VALUE0_INVERT 0x00000005 +#define NV03_CONTEXT_ROP_ROP_VALUE0_XOR 0x00000006 +#define NV03_CONTEXT_ROP_ROP_VALUE0_NAND 0x00000007 +#define NV03_CONTEXT_ROP_ROP_VALUE0_AND 0x00000008 +#define NV03_CONTEXT_ROP_ROP_VALUE0_EQUI 0x00000009 +#define NV03_CONTEXT_ROP_ROP_VALUE0_NOOP 0x0000000a +#define NV03_CONTEXT_ROP_ROP_VALUE0_OR_INVERTED 0x0000000b +#define NV03_CONTEXT_ROP_ROP_VALUE0_COPY 0x0000000c +#define NV03_CONTEXT_ROP_ROP_VALUE0_OR_REVERSE 0x0000000d +#define NV03_CONTEXT_ROP_ROP_VALUE0_OR 0x0000000e +#define NV03_CONTEXT_ROP_ROP_VALUE0_SET 0x0000000f +#define NV03_CONTEXT_ROP_ROP_VALUE1_SHIFT 4 +#define NV03_CONTEXT_ROP_ROP_VALUE1_MASK 0x000000f0 +#define NV03_CONTEXT_ROP_ROP_VALUE1_CLEAR 0x00000000 +#define NV03_CONTEXT_ROP_ROP_VALUE1_NOR 0x00000001 +#define NV03_CONTEXT_ROP_ROP_VALUE1_AND_INVERTED 0x00000002 +#define NV03_CONTEXT_ROP_ROP_VALUE1_COPY_INVERTED 0x00000003 +#define NV03_CONTEXT_ROP_ROP_VALUE1_AND_REVERSE 0x00000004 +#define NV03_CONTEXT_ROP_ROP_VALUE1_INVERT 0x00000005 +#define NV03_CONTEXT_ROP_ROP_VALUE1_XOR 0x00000006 +#define NV03_CONTEXT_ROP_ROP_VALUE1_NAND 0x00000007 +#define NV03_CONTEXT_ROP_ROP_VALUE1_AND 0x00000008 +#define NV03_CONTEXT_ROP_ROP_VALUE1_EQUI 0x00000009 +#define NV03_CONTEXT_ROP_ROP_VALUE1_NOOP 0x0000000a +#define NV03_CONTEXT_ROP_ROP_VALUE1_OR_INVERTED 0x0000000b +#define NV03_CONTEXT_ROP_ROP_VALUE1_COPY 0x0000000c +#define NV03_CONTEXT_ROP_ROP_VALUE1_OR_REVERSE 0x0000000d +#define NV03_CONTEXT_ROP_ROP_VALUE1_OR 0x0000000e +#define NV03_CONTEXT_ROP_ROP_VALUE1_SET 0x0000000f + + +#define NV04_IMAGE_PATTERN 0x00000044 + +#define NV04_IMAGE_PATTERN_NOP 0x00000100 +#define NV04_IMAGE_PATTERN_NOTIFY 0x00000104 +#define NV04_IMAGE_PATTERN_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT 0x00000300 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A16R5G6B5 0x00000001 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_X16A1R5G5B5 0x00000002 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT 0x00000304 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_CGA6 0x00000001 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_LE 0x00000002 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE 0x00000308 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_8X8 0x00000000 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_64X1 0x00000001 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_1X64 0x00000002 +#define NV04_IMAGE_PATTERN_PATTERN_SELECT 0x0000030c +#define NV04_IMAGE_PATTERN_PATTERN_SELECT_MONO 0x00000001 +#define NV04_IMAGE_PATTERN_PATTERN_SELECT_COLOR 0x00000002 +#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR0 0x00000310 +#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR1 0x00000314 +#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0 0x00000318 +#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1 0x0000031c +#define NV04_IMAGE_PATTERN_PATTERN_Y8(x) (0x00000400+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_Y8__SIZE 0x00000010 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_MASK 0x000000ff +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_SHIFT 8 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_MASK 0x0000ff00 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_MASK 0x00ff0000 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_SHIFT 24 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_MASK 0xff000000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5(x) (0x00000500+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5__SIZE 0x00000020 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_MASK 0x0000001f +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_SHIFT 5 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_MASK 0x000007e0 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_SHIFT 11 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_MASK 0x0000f800 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_MASK 0x001f0000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_SHIFT 21 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_MASK 0x07e00000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_SHIFT 27 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_MASK 0xf8000000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5(x) (0x00000600+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5__SIZE 0x00000020 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_MASK 0x0000001f +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_SHIFT 5 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_MASK 0x000003e0 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_SHIFT 10 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_MASK 0x00007c00 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_MASK 0x001f0000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_SHIFT 21 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_MASK 0x03e00000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_SHIFT 26 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_MASK 0x7c000000 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8(x) (0x00000700+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8__SIZE 0x00000040 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_MASK 0x000000ff +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_SHIFT 8 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_MASK 0x0000ff00 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_MASK 0x00ff0000 + + +#define NV03_VIDEO_LUT_CURSOR_DAC 0x00000046 + +#define NV03_VIDEO_LUT_CURSOR_DAC_SYNCHRONIZE 0x00000100 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_IMAGE 0x00000104 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_CURSOR 0x00000108 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_DAC 0x0000010c +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_NOTIFY 0x00000180 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE(x) (0x00000184+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT(x) (0x0000018c+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR(x) (0x00000194+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_GET 0x000002fc +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET(x) (0x00000300+((x)*8)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT(x) (0x00000304+((x)*8)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET(x) (0x00000340+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT(x) (0x00000344+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT(x) (0x00000348+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A 0x00000358 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE(x) (0x00000380+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC(x) (0x00000384+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC(x) (0x00000388+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE(x) (0x0000038c+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_PIXEL_CLOCK 0x000003a0 + + +#define NV03_DX3_TEXTURED_TRIANGLE 0x00000048 + +#define NV03_DX3_TEXTURED_TRIANGLE_NOP 0x00000100 +#define NV03_DX3_TEXTURED_TRIANGLE_NOTIFY 0x00000104 +#define NV03_DX3_TEXTURED_TRIANGLE_PATCH 0x0000010c +#define NV03_DX3_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV03_DX3_TEXTURED_TRIANGLE_DMA_TEXTURE 0x00000184 +#define NV03_DX3_TEXTURED_TRIANGLE_CLIP_RECTANGLE 0x00000188 +#define NV03_DX3_TEXTURED_TRIANGLE_SURFACE 0x0000018c +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_OFFSET 0x00000304 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT 0x00000308 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_MASK 0x0000ffff +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_MASK 0x0f000000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_SHIFT 28 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_MASK 0xf0000000 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER 0x0000030c +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_MASK 0x0000001f +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_MASK 0x00001f00 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_MASK 0x00ff0000 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR 0x00000310 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_MASK 0x000000ff +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_MASK 0x0000ff00 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_MASK 0x00ff0000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT 0x00000314 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_MASK 0x0000000f +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_SHIFT 4 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_MASK 0x00000030 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_SHIFT 6 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_MASK 0x000000c0 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_MASK 0x00000f00 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_SHIFT 12 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_MASK 0x00007000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_PERSPECTIVE_ENABLE (1 << 15) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_MASK 0x07000000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_SHIFT 27 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_MASK 0x18000000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_BETA (1 << 29) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_DST_BLEND (1 << 30) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SRC_BLEND (1 << 31) +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL 0x00000318 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_MASK 0x000000ff +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_MASK 0xffffff00 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR(x) (0x00001000+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_MASK 0x0000000f +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_SHIFT 4 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_MASK 0x000000f0 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_MASK 0x00000f00 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_SHIFT 12 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_MASK 0x0000f000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_MASK 0xff000000 +#define NV03_DX3_TEXTURED_TRIANGLE_COLOR(x) (0x00001004+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_COLOR__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_X(x) (0x00001008+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_X__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_Y(x) (0x0000100c+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_Y__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_Z(x) (0x00001010+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_Z__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_M(x) (0x00001014+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_M__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_U(x) (0x00001018+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_U__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_V(x) (0x0000101c+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_V__SIZE 0x00000040 + + +#define NV04_GDI_RECTANGLE_TEXT 0x0000004a + +#define NV04_GDI_RECTANGLE_TEXT_NOP 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 +#define NV04_GDI_RECTANGLE_TEXT_PATCH 0x0000010c +#define NV04_GDI_RECTANGLE_TEXT_PM_TRIGGER 0x00000140 +#define NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 +#define NV04_GDI_RECTANGLE_TEXT_DMA_FONTS 0x00000184 +#define NV04_GDI_RECTANGLE_TEXT_PATTERN 0x00000188 +#define NV04_GDI_RECTANGLE_TEXT_ROP 0x0000018c +#define NV04_GDI_RECTANGLE_TEXT_BETA1 0x00000190 +#define NV04_GDI_RECTANGLE_TEXT_BETA4 0x00000194 +#define NV04_GDI_RECTANGLE_TEXT_SURFACE 0x00000198 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_X16A1R5G5B5 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_CGA6 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(x) (0x00000400+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0 0x000005f4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1 0x000005f8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_B 0x000005fc +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0(x) (0x00000600+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1(x) (0x00000604+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x000007ec +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x000007f0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_C 0x000007f4 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C 0x000007f8 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C 0x000007fc +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000800+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000080 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x00000be4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x00000be8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR0_E 0x00000bec +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_E 0x00000bf0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x00000bf4 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x00000bf8 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E 0x00000bfc +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00000c00+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000080 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F 0x00000ff0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_MASK 0x0fffffff +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_SHIFT 28 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_MASK 0xf0000000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0 0x00000ff4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1 0x00000ff8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_F 0x00000ffc +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F(x) (0x00001000+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F__SIZE 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_MASK 0x000000ff +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_SHIFT 8 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_MASK 0x000fff00 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_SHIFT 20 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_MASK 0xfff00000 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G 0x000017f0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_MASK 0x0fffffff +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_SHIFT 28 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_MASK 0xf0000000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0 0x000017f4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1 0x000017f8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_G 0x000017fc +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT(x) (0x00001800+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT__SIZE 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX(x) (0x00001804+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX__SIZE 0x00000100 + + +#define NV03_GDI_RECTANGLE_TEXT 0x0000004b + +#define NV03_GDI_RECTANGLE_TEXT_NOP 0x00000100 +#define NV03_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 +#define NV03_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 +#define NV03_GDI_RECTANGLE_TEXT_PATTERN 0x00000184 +#define NV03_GDI_RECTANGLE_TEXT_ROP 0x00000188 +#define NV03_GDI_RECTANGLE_TEXT_BETA1 0x0000018c +#define NV03_GDI_RECTANGLE_TEXT_SURFACE 0x00000190 +#define NV03_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc +#define NV03_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT 0x00000400 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE 0x00000404 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B 0x000007f4 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B 0x000007f8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_B 0x000007fc +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0 0x00000800 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1 0x00000804 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x00000bec +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x00000bf0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_C 0x00000bf4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C 0x00000bf8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C 0x00000bfc +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000c00+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000020 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0 0x00000fe8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1 0x00000fec +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_D 0x00000ff0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D 0x00000ff4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D 0x00000ff8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D 0x00000ffc +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D(x) (0x00001000+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D__SIZE 0x00000020 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x000013e4 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x000013e8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR0_E 0x000013ec +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_E 0x000013f0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x000013f4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x000013f8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E 0x000013fc +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00001400+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000020 + + +#define NV04_SWIZZLED_SURFACE 0x00000052 + +#define NV04_SWIZZLED_SURFACE_NOP 0x00000100 +#define NV04_SWIZZLED_SURFACE_NOTIFY 0x00000104 +#define NV04_SWIZZLED_SURFACE_DMA_NOTIFY 0x00000180 +#define NV04_SWIZZLED_SURFACE_DMA_IMAGE 0x00000184 +#define NV04_SWIZZLED_SURFACE_FORMAT 0x00000300 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_SHIFT 0 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK 0x000000ff +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y8 0x00000001 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000002 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000003 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_R5G6B5 0x00000004 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y16 0x00000005 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000006 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000007 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000009 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8 0x0000000a +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y32 0x0000000b +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_MASK 0xff000000 +#define NV04_SWIZZLED_SURFACE_OFFSET 0x00000304 + + +#define NV04_CONTEXT_SURFACES_3D 0x00000053 + +#define NV04_CONTEXT_SURFACES_3D_NOP 0x00000100 +#define NV04_CONTEXT_SURFACES_3D_NOTIFY 0x00000104 +#define NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_SURFACES_3D_DMA_COLOR 0x00000184 +#define NV04_CONTEXT_SURFACES_3D_DMA_ZETA 0x00000188 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL 0x000002f8 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL 0x000002fc +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_FORMAT 0x00000300 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK 0x000000ff +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000001 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000002 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000004 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000005 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000006 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000007 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT 8 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK 0x0000ff00 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_PITCH 0x00000001 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SWIZZLE 0x00000002 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_MASK 0xff000000 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE 0x00000304 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_PITCH 0x00000308 +#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x0000030c +#define NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000310 + + +#define NV04_DX5_TEXTURED_TRIANGLE 0x00000054 + +#define NV04_DX5_TEXTURED_TRIANGLE_NOP 0x00000100 +#define NV04_DX5_TEXTURED_TRIANGLE_NOTIFY 0x00000104 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_A 0x00000184 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_B 0x00000188 +#define NV04_DX5_TEXTURED_TRIANGLE_SURFACE 0x0000018c +#define NV04_DX5_TEXTURED_TRIANGLE_COLORKEY 0x00000300 +#define NV04_DX5_TEXTURED_TRIANGLE_OFFSET 0x00000304 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT 0x00000308 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_MASK 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_SHIFT 2 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_MASK 0x0000000c +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CENTER 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CENTER 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_Y8 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A1R5G5B5 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X1R5G5B5 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A4R4G4B4 0x00000004 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_R5G6B5 0x00000005 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A8R8G8B8 0x00000006 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X8R8G8B8 0x00000007 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER 0x00000004 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP 0x00000005 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPU (1 << 27) +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_REPEAT 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MIRRORED_REPEAT 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_BORDER 0x00000004 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP 0x00000005 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPV (1 << 31) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER 0x0000030c +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND 0x00000310 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_MASK 0x0000000f +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT 0x00000001 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD 0x00000002 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_PHONG 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK 0x0f000000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK 0xf0000000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL 0x00000314 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE (1 << 12) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN (1 << 13) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT 14 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_MASK 0x0000c000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_MASK 0x00300000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE (1 << 22) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE (1 << 23) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_MASK 0x3f000000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT 30 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_MASK 0xc0000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR 0x00000318 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(x) (0x00000400+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY(x) (0x00000404+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ(x) (0x00000408+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW(x) (0x0000040c+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR(x) (0x00000410+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR(x) (0x00000414+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU(x) (0x00000418+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV(x) (0x0000041c+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(x) (0x00000600+((x)*4)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE__SIZE 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 + + +#define NV04_DX6_MULTITEX_TRIANGLE 0x00000055 + +#define NV04_DX6_MULTITEX_TRIANGLE_NOP 0x00000100 +#define NV04_DX6_MULTITEX_TRIANGLE_NOTIFY 0x00000104 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_A 0x00000184 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_B 0x00000188 +#define NV04_DX6_MULTITEX_TRIANGLE_SURFACE 0x0000018c +#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET(x) (0x00000308+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT(x) (0x00000310+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPU (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPV (1 << 31) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER(x) (0x00000318+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA 0x00000320 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR 0x00000324 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA 0x0000032c +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR 0x00000330 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR 0x00000334 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND 0x00000338 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK 0x0f000000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK 0xf0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0 0x0000033c +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_TEST_ENABLE (1 << 12) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ORIGIN (1 << 13) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_SHIFT 14 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_MASK 0x0000c000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_MASK 0x00300000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_DITHER_ENABLE (1 << 22) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_PERSPECTIVE_ENABLE (1 << 23) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_WRITE_ENABLE (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_ENABLE (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_ENABLE (1 << 26) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_ENABLE (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_ENABLE (1 << 28) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_ENABLE (1 << 29) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_SHIFT 30 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_MASK 0xc0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1 0x00000340 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2 0x00000344 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR 0x00000348 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX(x) (0x00000400+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY(x) (0x00000404+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ(x) (0x00000408+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW(x) (0x0000040c+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR(x) (0x00000410+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR(x) (0x00000414+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0(x) (0x00000418+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0(x) (0x0000041c+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1(x) (0x00000420+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1(x) (0x00000424+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE(x) (0x00000540+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE__SIZE 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 + + +#define NV10_TCL_PRIMITIVE_3D 0x00000056 + +#define NV10_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV10_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 +#define NV10_TCL_PRIMITIVE_3D_DISPLAY_LIST 0x0000018c +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 +#define NV10_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 +#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00000218+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00000220+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT 7 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK 0x00000780 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8 0x0000000b +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3 0x0000000e +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT 0x00000012 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT 0x00000013 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8 0x0000001a +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT_2 0x0000001b +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT 0x0000001d +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT 0x00000020 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT (1 << 11) +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_REPEAT 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_SHIFT 28 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MASK 0xf0000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_REPEAT 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00000228+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_MASK 0x00000030 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00000248+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 28 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0xf0000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SHIFT 5 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000268+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SHIFT 5 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR(x) (0x00000270+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM (1 << 14) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS (1 << 15) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT 17 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000280+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM (1 << 14) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS (1 << 15) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT 17 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_SHIFT 27 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_MASK 0x38000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SHIFT 5 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_MASK 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL 0x00000294 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL (1 << 1) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE 0x00000298 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP_2 0x00000802 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP2 0x00000803 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR 0x00000804 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR_2 0x00002601 +#define NV10_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 +#define NV10_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c +#define NV10_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 +#define NV10_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 +#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV10_TCL_PRIMITIVE_3D_POINT_SMOOTH_ENABLE 0x0000031c +#define NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_WEIGHT_ENABLE 0x00000328 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_B (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G (1 << 8) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R (1 << 16) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A (1 << 24) +#define NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c +#define NV10_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c +#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV10_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 +#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_R 0x000003a8 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_G 0x000003ac +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_B 0x000003b0 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_A 0x000003b4 +#define NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL 0x000003b8 +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT0 (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT1 (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT2 (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT3 (1 << 6) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT4 (1 << 8) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT5 (1 << 10) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT6 (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT7 (1 << 14) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_FALSE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 +#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE 0x000003e8 +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_POINT_SIZE 0x000003ec +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x00000540+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000580+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000600+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000604+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000608+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x00000680 +#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x00000684 +#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x00000688 +#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R0x000006c4 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G0x000006c8 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B0x000006cc +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X 0x000006e8 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Y 0x000006ec +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Z 0x000006f0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_W 0x000006f4 +#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x000006f8+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00000c04 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00000c08 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00000c18 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x00000c1c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00000c20 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W 0x00000c24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00000c30 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00000c34 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00000c38 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000c40 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000c44 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R 0x00000c50 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G 0x00000c54 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B 0x00000c58 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A 0x00000c5c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R 0x00000c60 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G 0x00000c64 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B 0x00000c68 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x00000c6c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R 0x00000c80 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G 0x00000c84 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B 0x00000c88 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I 0x00000c8c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00000c90 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00000c94 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00000c98 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x00000ca0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x00000ca4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x00000ca8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x00000cac +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x00000cb0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x00000cb4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x00000cb8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x00000cbc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x00000cc0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x00000cc8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x00000ccc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x00000cd0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x00000cd4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x00000cd8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x00000cdc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00000ce0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_WGH_1F 0x00000ce4 +#define NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x00000cec +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE 0x00000cf0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_POS 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS 0x00000d04 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL 0x00000d08 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL 0x00000d0c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00000dfc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END 0x000013fc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001800 + + +#define NV04_CONTEXT_COLOR_KEY 0x00000057 + +#define NV04_CONTEXT_COLOR_KEY_NOP 0x00000100 +#define NV04_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 +#define NV04_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 +#define NV04_CONTEXT_COLOR_KEY_COLOR 0x00000304 + + +#define NV03_CONTEXT_SURFACES_2D 0x00000058 + +#define NV03_CONTEXT_SURFACES_2D_SYNCHRONIZE 0x00000100 +#define NV03_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_SURFACES_2D_DMA_SOURCE 0x00000184 +#define NV03_CONTEXT_SURFACES_2D_DMA_DESTIN 0x00000188 +#define NV03_CONTEXT_SURFACES_2D_COLOR_FORMAT 0x00000300 +#define NV03_CONTEXT_SURFACES_2D_PITCH 0x00000304 +#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 +#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff +#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 +#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 +#define NV03_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 +#define NV03_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c + + +#define NV03_CONTEXT_SURFACES_3D 0x0000005a + +#define NV03_CONTEXT_SURFACES_3D_SYNCHRONIZE 0x00000100 +#define NV03_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_SURFACES_3D_DMA_SURfACE 0x00000184 +#define NV03_CONTEXT_SURFACES_3D_PITCH 0x00000300 +#define NV03_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x00000304 +#define NV03_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000308 + + +#define NV04_RENDER_SOLID_LINE 0x0000005c + +#define NV04_RENDER_SOLID_LINE_SURFACE 0x00000198 + + +#define NV_RENDER_SOLID_TRIANGLE 0x0000005d + + + +#define NV_RENDER_SOLID_RECTANGLE 0x0000005e + +#define NV_RENDER_SOLID_RECTANGLE_SURFACE 0x00000198 + + +#define NV_IMAGE_BLIT 0x0000005f + +#define NV_IMAGE_BLIT_NOP 0x00000100 +#define NV_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV_IMAGE_BLIT_PATTERN 0x0000018c +#define NV_IMAGE_BLIT_ROP 0x00000190 +#define NV_IMAGE_BLIT_SURFACE 0x0000019c +#define NV_IMAGE_BLIT_OPERATION 0x000002fc +#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 +#define NV_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 +#define NV_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 +#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000003 +#define NV_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000004 +#define NV_IMAGE_BLIT_POINT_IN 0x00000300 +#define NV_IMAGE_BLIT_POINT_IN_X_SHIFT 0 +#define NV_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff +#define NV_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 +#define NV_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 +#define NV_IMAGE_BLIT_POINT_OUT 0x00000304 +#define NV_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 +#define NV_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff +#define NV_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 +#define NV_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 +#define NV_IMAGE_BLIT_SIZE 0x00000308 +#define NV_IMAGE_BLIT_SIZE_W_SHIFT 0 +#define NV_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff +#define NV_IMAGE_BLIT_SIZE_H_SHIFT 16 +#define NV_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 + + +#define NV04_INDEXED_IMAGE_FROM_CPU 0x00000060 + +#define NV04_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV04_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV04_INDEXED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 +#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 +#define NV04_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec +#define NV04_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 +#define NV04_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 +#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 +#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc +#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR 0x00000400 + + +#define NV_IMAGE_FROM_CPU 0x00000061 + + + +#define NV10_CONTEXT_SURFACES_2D 0x00000062 + + + +#define NV05_SCALED_IMAGE_FROM_MEMORY 0x00000063 + +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 + + +#define NV01_IMAGE_SRCCOPY_AND 0x00000064 + +#define NV01_IMAGE_SRCCOPY_AND_NOTIFY 0x00000104 +#define NV01_IMAGE_SRCCOPY_AND_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_SRCCOPY_AND_IMAGE_OUTPUT 0x00000200 +#define NV01_IMAGE_SRCCOPY_AND_IMAGE_INPUT 0x00000204 + + +#define NV05_INDEXED_IMAGE_FROM_CPU 0x00000064 + +#define NV05_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV05_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV05_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV05_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_KEY 0x00000188 +#define NV05_INDEXED_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x0000018c +#define NV05_INDEXED_IMAGE_FROM_CPU_PATTERN 0x00000190 +#define NV05_INDEXED_IMAGE_FROM_CPU_ROP 0x00000194 +#define NV05_INDEXED_IMAGE_FROM_CPU_BETA1 0x00000198 +#define NV05_INDEXED_IMAGE_FROM_CPU_BETA4 0x0000019c +#define NV05_INDEXED_IMAGE_FROM_CPU_SURFACE 0x000001a0 +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000003e0 +#define NV05_INDEXED_IMAGE_FROM_CPU_OPERATION 0x000003e4 +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 +#define NV05_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec +#define NV05_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 +#define NV05_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 +#define NV05_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 +#define NV05_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc +#define NV05_INDEXED_IMAGE_FROM_CPU_INDICES 0x00000400 + + +#define NV05_IMAGE_FROM_CPU 0x00000065 + +#define NV05_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV05_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV05_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV05_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV05_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 +#define NV05_IMAGE_FROM_CPU_PATTERN 0x0000018c +#define NV05_IMAGE_FROM_CPU_ROP 0x00000190 +#define NV05_IMAGE_FROM_CPU_BETA1 0x00000194 +#define NV05_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV05_IMAGE_FROM_CPU_SURFACE 0x0000019c +#define NV05_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 +#define NV05_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV05_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV05_IMAGE_FROM_CPU_POINT 0x00000304 +#define NV05_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 +#define NV05_IMAGE_FROM_CPU_SIZE_IN 0x0000030c +#define NV05_IMAGE_FROM_CPU_COLOR 0x00000400 + + +#define NV05_STRETCHED_IMAGE_FROM_CPU 0x00000066 + +#define NV05_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV05_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV05_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV05_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 +#define NV05_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c +#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 +#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA4 0x00000194 +#define NV05_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000198 +#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 +#define NV05_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV05_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 +#define NV05_STRETCHED_IMAGE_FROM_CPU_DELTA_DX_DU 0x00000308 +#define NV05_STRETCHED_IMAGE_FROM_CPU_DELTA_DY_DV 0x0000030c +#define NV05_STRETCHED_IMAGE_FROM_CPU_CLIP_0 0x00000310 +#define NV05_STRETCHED_IMAGE_FROM_CPU_CLIP_1 0x00000314 +#define NV05_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 +#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR 0x00000400 + + +#define NV04_IMAGE_BLEND_PREMULT 0x00000067 + +#define NV04_IMAGE_BLEND_PREMULT_NOP 0x00000100 +#define NV04_IMAGE_BLEND_PREMULT_NOTIFY 0x00000104 +#define NV04_IMAGE_BLEND_PREMULT_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_BLEND_PREMULT_IMAGE_OUTPUT 0x00000200 +#define NV04_IMAGE_BLEND_PREMULT_BETA_INPUT 0x00000204 +#define NV04_IMAGE_BLEND_PREMULT_IMAGE_INPUT 0x00000208 + + +#define NV03_CHANNEL_PIO 0x0000006a + + + +#define NV03_CHANNEL_DMA 0x0000006b + + + +#define NV04_BETA_SOLID 0x00000072 + +#define NV04_BETA_SOLID_NOP 0x00000100 +#define NV04_BETA_SOLID_NOTIFY 0x00000104 +#define NV04_BETA_SOLID_DMA_NOTIFY 0x00000180 +#define NV04_BETA_SOLID_BETA_OUTPUT 0x00000200 +#define NV04_BETA_SOLID_BETA_FACTOR 0x00000300 + + +#define NV_STRETCHED_IMAGE_FROM_CPU 0x00000076 + +#define NV_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 +#define NV_STRETCHED_IMAGE_FROM_CPU_DELTA_DU_DX 0x00000308 +#define NV_STRETCHED_IMAGE_FROM_CPU_DELTA_DV_DY 0x0000030c +#define NV_STRETCHED_IMAGE_FROM_CPU_CLIP_0 0x00000310 +#define NV_STRETCHED_IMAGE_FROM_CPU_CLIP_1 0x00000314 +#define NV_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 +#define NV_STRETCHED_IMAGE_FROM_CPU_COLOR 0x00000400 + + +#define NV04_SCALED_IMAGE_FROM_MEMORY 0x00000077 + +#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA4 0x00000194 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000198 + + +#define NV10_TEXTURE_FROM_CPU 0x0000007b + +#define NV10_TEXTURE_FROM_CPU_NOP 0x00000100 +#define NV10_TEXTURE_FROM_CPU_NOTIFY 0x00000104 +#define NV10_TEXTURE_FROM_CPU_WAIT_FOR_IDLE 0x00000108 +#define NV10_TEXTURE_FROM_CPU_PM_TRIGGER 0x00000140 +#define NV10_TEXTURE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV10_TEXTURE_FROM_CPU_SURFACE 0x00000184 +#define NV10_TEXTURE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV10_TEXTURE_FROM_CPU_POINT 0x00000304 +#define NV10_TEXTURE_FROM_CPU_POINT_X_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_POINT_X_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_POINT_Y_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_POINT_Y_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_SIZE 0x00000308 +#define NV10_TEXTURE_FROM_CPU_SIZE_W_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_SIZE_W_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_SIZE_H_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_SIZE_H_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL 0x0000030c +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL 0x00000310 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV10_TEXTURE_FROM_CPU_COLOR__SIZE 0x00000100 + + +#define NV10_VIDEO_DISPLAY 0x0000007c + + + +#define NV10_DVD_SUBPICTURE 0x00000088 + + + +#define NV10_SCALED_IMAGE_FROM_MEMORY 0x00000089 + +#define NV10_SCALED_IMAGE_FROM_MEMORY_WAIT_FOR_IDLE 0x00000108 + + +#define NV10_IMAGE_FROM_CPU 0x0000008a + +#define NV10_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV10_IMAGE_FROM_CPU_SURFACE 0x0000019c +#define NV10_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 + + +#define NV10_CONTEXT_SURFACES_3D 0x00000093 + + + +#define NV10_DX5_TEXTURE_TRIANGLE 0x00000094 + + + +#define NV10_DX6_MULTI_TEXTURE_TRIANGLE 0x00000095 + + + +#define NV11_TCL_PRIMITIVE_3D 0x00000096 + +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000d40 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000d44 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f + + +#define NV20_TCL_PRIMITIVE_3D 0x00000097 + +#define NV20_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV20_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV20_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a4 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001a8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 +#define NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c +#define NV20_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 +#define NV20_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 +#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 +#define NV20_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c +#define NV20_TCL_PRIMITIVE_3D_LIGHT_CONTROL 0x00000294 +#define NV20_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c +#define NV20_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 +#define NV20_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 +#define NV20_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c +#define NV20_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 +#define NV20_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 +#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV20_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV20_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c +#define NV20_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c +#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV20_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 +#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 +#define NV20_TCL_PRIMITIVE_3D_SEPARATE_SPECULAR_ENABLE 0x000003b8 +#define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc +#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_POINT_SIZE 0x0000043c +#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000009c0 +#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000009c4 +#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000009c8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c +#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x00000a30+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000007 +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000ac0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000af0 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000af4 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000af8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000afc +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00001500 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00001504 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00001508 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00001518 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x0000151c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00001520 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY 0x00001528 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z 0x0000152c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00001530 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00001534 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00001538 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00001540 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00001544 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_X 0x00001550 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Y 0x00001554 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Z 0x00001558 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_W 0x0000155c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_X 0x00001560 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Y 0x00001564 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Z 0x00001568 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x0000156c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_X 0x00001580 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Y 0x00001584 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Z 0x00001588 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I 0x0000158c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_MASK 0x000000ff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_SHIFT 8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_MASK 0x0000ff00 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_MASK 0x00ff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_SHIFT 24 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_MASK 0xff000000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00001590 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00001594 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00001598 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x000015a0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x000015a4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x000015a8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x000015ac +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x000015b0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x000015b4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x000015b8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x000015bc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x000015c0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x000015c8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x000015cc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x000015d0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x000015d4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x000015d8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x000015dc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S 0x000015e0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T 0x000015e4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I 0x000015e8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S 0x000015f0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T 0x000015f4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R 0x000015f8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q 0x000015fc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST 0x00001600 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ 0x00001604 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S 0x00001608 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T 0x0000160c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I 0x00001610 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S 0x00001620 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T 0x00001624 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R 0x00001628 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q 0x0000162c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST 0x00001630 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ 0x00001634 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x000016bc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x000017bc +#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x000017c0 +#define NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 +#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE 0x000017f8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x000017fc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001818 +#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001b00+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001b04+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001b08+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001b0c+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001b14+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x00001c30 +#define NV20_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x00001c50 +#define NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE 0x00001d90 +#define NV20_TCL_PRIMITIVE_3D_CLEAR_BUFFERS 0x00001d94 +#define NV20_TCL_PRIMITIVE_3D_RC_COLOR0 0x00001e20 +#define NV20_TCL_PRIMITIVE_3D_RC_COLOR1 0x00001e24 +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00001e40+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_ENABLE 0x00001e60 +#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP 0x00001e70 +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID 0x00001ea4 + + +#define NV17_TCL_PRIMITIVE_3D 0x00000099 + +#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x000001ac +#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001b0 +#define NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE 0x000002bc +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH 0x00000d5c +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE 0x00000d68 +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE 0x00001658 + + +#define NV20_SWIZZLED_SURFACE 0x0000009e + + + +#define NV12_IMAGE_BLIT 0x0000009f + +#define NV12_IMAGE_BLIT_BETA1 0x00000194 +#define NV12_IMAGE_BLIT_BETA4 0x00000198 + + +#define NV30_CONTEXT_SURFACES_2D 0x00000362 + + + +#define NV30_STRETCHED_IMAGE_FROM_CPU 0x00000366 + + + +#define NV30_TEXTURE_FROM_CPU 0x0000037b + + + +#define NV30_SCALED_IMAGE_FROM_MEMORY 0x00000389 + + + +#define NV30_IMAGE_FROM_CPU 0x0000038a + + + +#define NV30_TCL_PRIMITIVE_3D 0x00000397 + + + +#define NV30_SWIZZLED_SURFACE 0x0000039e + + + +#define NV36_TCL_PRIMITIVE_3D 0x00000497 + + + +#define NV25_TCL_PRIMITIVE_3D 0x00000597 + +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x0000019c +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a0 +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001ac +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY9 0x000001b0 + + +#define NV34_TCL_PRIMITIVE_3D 0x00000697 + +#define NV34_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV34_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV34_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x0000018c +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000194 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x00000198 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a4 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a8 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001ac +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001b0 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_FORMAT 0x00000208 +#define NV34_TCL_PRIMITIVE_3D_PITCH 0x0000020c +#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 +#define NV34_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 +#define NV34_TCL_PRIMITIVE_3D_COLOR1_OFFSET 0x00000218 +#define NV34_TCL_PRIMITIVE_3D_COLOR1_PITCH 0x0000021c +#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_PITCH 0x0000022c +#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_OFFSET 0x00000230 +#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_UNK0250(x) (0x00000250+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_UNK0250__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x0000030c +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000310 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000314 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000318 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_COLOR 0x0000031c +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION 0x00000320 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MIN 0x00008007 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MAX 0x00008008 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000324 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK 0xff000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE 0x00000328 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK 0x0000032c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC 0x00000330 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF 0x00000334 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK 0x00000338 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL 0x0000033c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL 0x00000340 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS 0x00000344 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE 0x00000348 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK 0x0000034c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF 0x00000354 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK 0x00000358 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL 0x00000368 +#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV34_TCL_PRIMITIVE_3D_FOG_ENABLE 0x0000036c +#define NV34_TCL_PRIMITIVE_3D_FOG_COLOR 0x00000370 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000378 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f +#define NV34_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x0000037c +#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R 0x000003a0 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G 0x000003a4 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B 0x000003a8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A 0x000003b4 +#define NV34_TCL_PRIMITIVE_3D_LINE_WIDTH 0x000003b8 +#define NV34_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x000003bc +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x000008c0 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x000008c4 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000008c8 +#define NV34_TCL_PRIMITIVE_3D_FOG_MODE 0x000008cc +#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000008d4 +#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV34_TCL_PRIMITIVE_3D_RC_COLOR0 0x000008ec +#define NV34_TCL_PRIMITIVE_3D_RC_COLOR1 0x000008f0 +#define NV34_TCL_PRIMITIVE_3D_RC_FINAL0 0x000008f4 +#define NV34_TCL_PRIMITIVE_3D_RC_FINAL1 0x000008f8 +#define NV34_TCL_PRIMITIVE_3D_RC_ENABLE 0x000008fc +#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000900+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000904+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000914+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R0x00000a10 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G0x00000a14 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B0x00000a18 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000a30 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000a34 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000a38 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000a3c +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000a6c +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x00000a74 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000a90 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000a94 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x0000000c +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R0x000017a0 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G0x000017a4 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B0x000017a8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R 0x000017b0 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G 0x000017b4 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B 0x000017b8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A 0x000017c0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00001808 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV34_TCL_PRIMITIVE_3D_VERTEX_DATA 0x00001818 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x00001828 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x0000182c +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE 0x00001830 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE 0x00001834 +#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x0000183c +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST(x) (0x00001920+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_S_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_S_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_T_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_T_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I__SIZE 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY 0x00001980 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW 0x00001984 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_SHIFT 4 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_MASK 0x000000f0 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 20 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_SHIFT 28 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001a08+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_REPEAT 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_REPEAT 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MIRRORED_REPEAT 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_EDGE 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_BORDER 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_REPEAT 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MIRRORED_REPEAT 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_EDGE 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_BORDER 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001a14+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x000f0000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0x0f000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_MASK 0xff000000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_CLEAR_DEPTH_VALUE 0x00001d8c +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00001e54 +#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_POINT_SIZE 0x00001ee0 +#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00001ee4 +#define NV34_TCL_PRIMITIVE_3D_POINT_SPRITE 0x00001ee8 +#define NV34_TCL_PRIMITIVE_3D_UNK1f80(x) (0x00001f80+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_UNK1f80__SIZE 0x00000010 + + +#define NV40_CONTEXT_SURFACES_2D 0x00003062 + + + +#define NV40_STRETCHED_IMAGE_FROM_CPU 0x00003066 + + + +#define NV40_TEXTURE_FROM_CPU 0x0000307b + + + +#define NV40_SCALED_IMAGE_FROM_MEMORY 0x00003089 + + + +#define NV40_IMAGE_FROM_CPU 0x0000308a + + + +#define NV40_SWIZZLED_SURFACE 0x0000309e + + + +#define NV40TCL 0x00004097 + +#define NV40TCL_REF_CNT 0x00000050 +#define NV40TCL_NOP 0x00000100 +#define NV40TCL_NOTIFY 0x00000104 +#define NV40TCL_DMA_NOTIFY 0x00000180 +#define NV40TCL_DMA_TEXTURE0 0x00000184 +#define NV40TCL_DMA_TEXTURE1 0x00000188 +#define NV40TCL_DMA_COLOR1 0x0000018c +#define NV40TCL_DMA_COLOR0 0x00000194 +#define NV40TCL_DMA_ZETA 0x00000198 +#define NV40TCL_DMA_VTXBUF0 0x0000019c +#define NV40TCL_DMA_VTXBUF1 0x000001a0 +#define NV40TCL_DMA_FENCE 0x000001a4 +#define NV40TCL_DMA_QUERY 0x000001a8 +#define NV40TCL_DMA_UNK01AC 0x000001ac +#define NV40TCL_DMA_UNK01B0 0x000001b0 +#define NV40TCL_DMA_COLOR2 0x000001b4 +#define NV40TCL_DMA_COLOR3 0x000001b8 +#define NV40TCL_RT_HORIZ 0x00000200 +#define NV40TCL_RT_HORIZ_W_SHIFT 16 +#define NV40TCL_RT_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_RT_HORIZ_X_SHIFT 0 +#define NV40TCL_RT_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_RT_VERT 0x00000204 +#define NV40TCL_RT_VERT_H_SHIFT 16 +#define NV40TCL_RT_VERT_H_MASK 0xffff0000 +#define NV40TCL_RT_VERT_Y_SHIFT 0 +#define NV40TCL_RT_VERT_Y_MASK 0x0000ffff +#define NV40TCL_RT_FORMAT 0x00000208 +#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT 24 +#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_MASK 0xff000000 +#define NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT 16 +#define NV40TCL_RT_FORMAT_LOG2_WIDTH_MASK 0x00ff0000 +#define NV40TCL_RT_FORMAT_TYPE_SHIFT 8 +#define NV40TCL_RT_FORMAT_TYPE_MASK 0x00000f00 +#define NV40TCL_RT_FORMAT_TYPE_LINEAR 0x00000001 +#define NV40TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000002 +#define NV40TCL_RT_FORMAT_DEPTH_SHIFT 5 +#define NV40TCL_RT_FORMAT_DEPTH_MASK 0x000000e0 +#define NV40TCL_RT_FORMAT_DEPTH_Z16 0x00000001 +#define NV40TCL_RT_FORMAT_DEPTH_Z24S8 0x00000002 +#define NV40TCL_RT_FORMAT_COLOR_SHIFT 0 +#define NV40TCL_RT_FORMAT_COLOR_MASK 0x0000001f +#define NV40TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV40TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 +#define NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV40TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV40TCL_COLOR0_PITCH 0x0000020c +#define NV40TCL_COLOR0_OFFSET 0x00000210 +#define NV40TCL_ZETA_OFFSET 0x00000214 +#define NV40TCL_COLOR1_OFFSET 0x00000218 +#define NV40TCL_COLOR1_PITCH 0x0000021c +#define NV40TCL_RT_ENABLE 0x00000220 +#define NV40TCL_RT_ENABLE_MRT (1 << 4) +#define NV40TCL_RT_ENABLE_COLOR3 (1 << 3) +#define NV40TCL_RT_ENABLE_COLOR2 (1 << 2) +#define NV40TCL_RT_ENABLE_COLOR1 (1 << 1) +#define NV40TCL_RT_ENABLE_COLOR0 (1 << 0) +#define NV40TCL_ZETA_PITCH 0x0000022c +#define NV40TCL_COLOR2_PITCH 0x00000280 +#define NV40TCL_COLOR3_PITCH 0x00000284 +#define NV40TCL_COLOR2_OFFSET 0x00000288 +#define NV40TCL_COLOR3_OFFSET 0x0000028c +#define NV40TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV40TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV40TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV40TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV40TCL_DITHER_ENABLE 0x00000300 +#define NV40TCL_ALPHA_TEST_ENABLE 0x00000304 +#define NV40TCL_ALPHA_TEST_FUNC 0x00000308 +#define NV40TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 +#define NV40TCL_ALPHA_TEST_FUNC_LESS 0x00000201 +#define NV40TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 +#define NV40TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 +#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV40TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 +#define NV40TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 +#define NV40TCL_ALPHA_TEST_REF 0x0000030c +#define NV40TCL_BLEND_ENABLE 0x00000310 +#define NV40TCL_BLEND_FUNC_SRC 0x00000314 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV40TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_DST 0x00000318 +#define NV40TCL_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV40TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_COLOR 0x0000031c +#define NV40TCL_BLEND_EQUATION 0x00000320 +#define NV40TCL_BLEND_EQUATION_RGB_SHIFT 0 +#define NV40TCL_BLEND_EQUATION_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 +#define NV40TCL_BLEND_EQUATION_RGB_MIN 0x00008007 +#define NV40TCL_BLEND_EQUATION_RGB_MAX 0x00008008 +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV40TCL_BLEND_EQUATION_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_EQUATION_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 +#define NV40TCL_BLEND_EQUATION_ALPHA_MIN 0x00008007 +#define NV40TCL_BLEND_EQUATION_ALPHA_MAX 0x00008008 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV40TCL_COLOR_MASK 0x00000324 +#define NV40TCL_COLOR_MASK_BUFFER0_B_SHIFT 0 +#define NV40TCL_COLOR_MASK_BUFFER0_B_MASK 0x000000ff +#define NV40TCL_COLOR_MASK_BUFFER0_G_SHIFT 8 +#define NV40TCL_COLOR_MASK_BUFFER0_G_MASK 0x0000ff00 +#define NV40TCL_COLOR_MASK_BUFFER0_R_SHIFT 16 +#define NV40TCL_COLOR_MASK_BUFFER0_R_MASK 0x00ff0000 +#define NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT 24 +#define NV40TCL_COLOR_MASK_BUFFER0_A_MASK 0xff000000 +#define NV40TCL_STENCIL_BACK_ENABLE 0x00000328 +#define NV40TCL_STENCIL_BACK_MASK 0x0000032c +#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000334 +#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000338 +#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000033c +#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000344 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000348 +#define NV40TCL_STENCIL_FRONT_MASK 0x0000034c +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000354 +#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 +#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_SHADE_MODEL 0x00000368 +#define NV40TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV40TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV40TCL_MRT_COLOR_MASK 0x00000370 +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_A (1 << 4) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_R (1 << 5) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_G (1 << 6) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_B (1 << 7) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_A (1 << 8) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_R (1 << 9) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_G (1 << 10) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_B (1 << 11) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_A (1 << 12) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_R (1 << 13) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_G (1 << 14) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_B (1 << 15) +#define NV40TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV40TCL_COLOR_LOGIC_OP 0x00000378 +#define NV40TCL_COLOR_LOGIC_OP_CLEAR 0x00001500 +#define NV40TCL_COLOR_LOGIC_OP_AND 0x00001501 +#define NV40TCL_COLOR_LOGIC_OP_AND_REVERSE 0x00001502 +#define NV40TCL_COLOR_LOGIC_OP_COPY 0x00001503 +#define NV40TCL_COLOR_LOGIC_OP_AND_INVERTED 0x00001504 +#define NV40TCL_COLOR_LOGIC_OP_NOOP 0x00001505 +#define NV40TCL_COLOR_LOGIC_OP_XOR 0x00001506 +#define NV40TCL_COLOR_LOGIC_OP_OR 0x00001507 +#define NV40TCL_COLOR_LOGIC_OP_NOR 0x00001508 +#define NV40TCL_COLOR_LOGIC_OP_EQUIV 0x00001509 +#define NV40TCL_COLOR_LOGIC_OP_INVERT 0x0000150a +#define NV40TCL_COLOR_LOGIC_OP_OR_REVERSE 0x0000150b +#define NV40TCL_COLOR_LOGIC_OP_COPY_INVERTED 0x0000150c +#define NV40TCL_COLOR_LOGIC_OP_OR_INVERTED 0x0000150d +#define NV40TCL_COLOR_LOGIC_OP_NAND 0x0000150e +#define NV40TCL_COLOR_LOGIC_OP_SET 0x0000150f +#define NV40TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV40TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV40TCL_LINE_WIDTH 0x000003b8 +#define NV40TCL_LINE_SMOOTH_ENABLE 0x000003bc +#define NV40TCL_UNK03C0(x) (0x000003c0+((x)*4)) +#define NV40TCL_UNK03C0__SIZE 0x00000010 +#define NV40TCL_UNK0400(x) (0x00000400+((x)*4)) +#define NV40TCL_UNK0400__SIZE 0x00000010 +#define NV40TCL_UNK0440(x) (0x00000440+((x)*4)) +#define NV40TCL_UNK0440__SIZE 0x00000020 +#define NV40TCL_SCISSOR_HORIZ 0x000008c0 +#define NV40TCL_SCISSOR_HORIZ_X_SHIFT 0 +#define NV40TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_SCISSOR_HORIZ_W_SHIFT 16 +#define NV40TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_SCISSOR_VERT 0x000008c4 +#define NV40TCL_SCISSOR_VERT_Y_SHIFT 0 +#define NV40TCL_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV40TCL_SCISSOR_VERT_H_SHIFT 16 +#define NV40TCL_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV40TCL_FOG_MODE 0x000008cc +#define NV40TCL_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV40TCL_FOG_EQUATION_LINEAR 0x000008d4 +#define NV40TCL_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV40TCL_FP_ADDRESS 0x000008e4 +#define NV40TCL_FP_ADDRESS_OFFSET_SHIFT 8 +#define NV40TCL_FP_ADDRESS_OFFSET_MASK 0xffffff00 +#define NV40TCL_FP_ADDRESS_DMA1 (1 << 1) +#define NV40TCL_FP_ADDRESS_DMA0 (1 << 0) +#define NV40TCL_VIEWPORT_HORIZ 0x00000a00 +#define NV40TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV40TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV40TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_VIEWPORT_VERT 0x00000a04 +#define NV40TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV40TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV40TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV40TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV40TCL_VIEWPORT_TRANSLATE_X 0x00000a20 +#define NV40TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 +#define NV40TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 +#define NV40TCL_VIEWPORT_TRANSLATE_W 0x00000a2c +#define NV40TCL_VIEWPORT_SCALE_X 0x00000a30 +#define NV40TCL_VIEWPORT_SCALE_Y 0x00000a34 +#define NV40TCL_VIEWPORT_SCALE_Z 0x00000a38 +#define NV40TCL_VIEWPORT_SCALE_W 0x00000a3c +#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV40TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV40TCL_DEPTH_FUNC 0x00000a6c +#define NV40TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV40TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV40TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV40TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV40TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV40TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV40TCL_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV40TCL_DEPTH_TEST_ENABLE 0x00000a74 +#define NV40TCL_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV40TCL_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV40TCL_UNK0B40(x) (0x00000b40+((x)*4)) +#define NV40TCL_UNK0B40__SIZE 0x00000008 +#define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) +#define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV40TCL_VTX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 +#define NV40TCL_VTXBUF_ADDRESS(x) (0x00001680+((x)*4)) +#define NV40TCL_VTXBUF_ADDRESS__SIZE 0x00000010 +#define NV40TCL_VTXBUF_ADDRESS_DMA1 (1 << 31) +#define NV40TCL_VTXBUF_ADDRESS_OFFSET_SHIFT 0 +#define NV40TCL_VTXBUF_ADDRESS_OFFSET_MASK 0x0fffffff +#define NV40TCL_VTX_CACHE_INVALIDATE 0x00001714 +#define NV40TCL_VTXFMT(x) (0x00001740+((x)*4)) +#define NV40TCL_VTXFMT__SIZE 0x00000010 +#define NV40TCL_VTXFMT_TYPE_SHIFT 0 +#define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f +#define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 +#define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 +#define NV40TCL_VTXFMT_SIZE_SHIFT 4 +#define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 +#define NV40TCL_VTXFMT_STRIDE_SHIFT 8 +#define NV40TCL_VTXFMT_STRIDE_MASK 0x0000ff00 +#define NV40TCL_QUERY_RESET 0x000017c8 +#define NV40TCL_QUERY_UNK17CC 0x000017cc +#define NV40TCL_QUERY_GET 0x00001800 +#define NV40TCL_QUERY_GET_UNK24_SHIFT 24 +#define NV40TCL_QUERY_GET_UNK24_MASK 0xff000000 +#define NV40TCL_QUERY_GET_OFFSET_SHIFT 0 +#define NV40TCL_QUERY_GET_OFFSET_MASK 0x00ffffff +#define NV40TCL_BEGIN_END 0x00001808 +#define NV40TCL_BEGIN_END_STOP 0x00000000 +#define NV40TCL_BEGIN_END_POINTS 0x00000001 +#define NV40TCL_BEGIN_END_LINES 0x00000002 +#define NV40TCL_BEGIN_END_LINE_LOOP 0x00000003 +#define NV40TCL_BEGIN_END_LINE_STRIP 0x00000004 +#define NV40TCL_BEGIN_END_TRIANGLES 0x00000005 +#define NV40TCL_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV40TCL_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV40TCL_BEGIN_END_QUADS 0x00000008 +#define NV40TCL_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV40TCL_BEGIN_END_POLYGON 0x0000000a +#define NV40TCL_VB_ELEMENT_U16 0x0000180c +#define NV40TCL_VB_ELEMENT_U16_1_SHIFT 16 +#define NV40TCL_VB_ELEMENT_U16_1_MASK 0xffff0000 +#define NV40TCL_VB_ELEMENT_U16_0_SHIFT 0 +#define NV40TCL_VB_ELEMENT_U16_0_MASK 0x0000ffff +#define NV40TCL_VB_ELEMENT_U32 0x00001810 +#define NV40TCL_VB_VERTEX_BATCH 0x00001814 +#define NV40TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 +#define NV40TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 +#define NV40TCL_VB_VERTEX_BATCH_START_SHIFT 0 +#define NV40TCL_VB_VERTEX_BATCH_START_MASK 0x00ffffff +#define NV40TCL_VERTEX_DATA 0x00001818 +#define NV40TCL_POLYGON_MODE_FRONT 0x00001828 +#define NV40TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV40TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV40TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV40TCL_POLYGON_MODE_BACK 0x0000182c +#define NV40TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV40TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV40TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV40TCL_CULL_FACE 0x00001830 +#define NV40TCL_CULL_FACE_FRONT 0x00000404 +#define NV40TCL_CULL_FACE_BACK 0x00000405 +#define NV40TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV40TCL_FRONT_FACE 0x00001834 +#define NV40TCL_FRONT_FACE_CW 0x00000900 +#define NV40TCL_FRONT_FACE_CCW 0x00000901 +#define NV40TCL_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV40TCL_CULL_FACE_ENABLE 0x0000183c +#define NV40TCL_TEX_SIZE1(x) (0x00001840+((x)*4)) +#define NV40TCL_TEX_SIZE1__SIZE 0x00000008 +#define NV40TCL_TEX_SIZE1_DEPTH_SHIFT 20 +#define NV40TCL_TEX_SIZE1_DEPTH_MASK 0xfff00000 +#define NV40TCL_TEX_SIZE1_PITCH_SHIFT 0 +#define NV40TCL_TEX_SIZE1_PITCH_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV40TCL_VTX_ATTR_2F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV40TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV40TCL_VTX_ATTR_2I__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_2I_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_2I_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_0(x) (0x00001900+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_0__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_0_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_0_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_1(x) (0x00001904+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_1__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_1_W_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_1_Z_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff +#define NV40TCL_TEX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV40TCL_TEX_OFFSET__SIZE 0x00000010 +#define NV40TCL_TEX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV40TCL_TEX_FORMAT__SIZE 0x00000010 +#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT 16 +#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_MASK 0x000f0000 +#define NV40TCL_TEX_FORMAT_RECT (1 << 14) +#define NV40TCL_TEX_FORMAT_LINEAR (1 << 13) +#define NV40TCL_TEX_FORMAT_FORMAT_SHIFT 8 +#define NV40TCL_TEX_FORMAT_FORMAT_MASK 0x00001f00 +#define NV40TCL_TEX_FORMAT_FORMAT_L8 0x00000001 +#define NV40TCL_TEX_FORMAT_FORMAT_A1R5G5B5 0x00000002 +#define NV40TCL_TEX_FORMAT_FORMAT_A4R4G4B4 0x00000003 +#define NV40TCL_TEX_FORMAT_FORMAT_R5G6B5 0x00000004 +#define NV40TCL_TEX_FORMAT_FORMAT_A8R8G8B8 0x00000005 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT1 0x00000006 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT3 0x00000007 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT5 0x00000008 +#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00000018 +#define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 +#define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 +#define NV40TCL_TEX_FORMAT_NO_BORDER (1 << 3) +#define NV40TCL_TEX_FORMAT_CUBIC (1 << 2) +#define NV40TCL_TEX_FORMAT_DMA1 (1 << 1) +#define NV40TCL_TEX_FORMAT_DMA0 (1 << 0) +#define NV40TCL_TEX_WRAP(x) (0x00001a08+((x)*32)) +#define NV40TCL_TEX_WRAP__SIZE 0x00000010 +#define NV40TCL_TEX_WRAP_S_SHIFT 0 +#define NV40TCL_TEX_WRAP_S_MASK 0x000000ff +#define NV40TCL_TEX_WRAP_S_REPEAT 0x00000001 +#define NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV40TCL_TEX_WRAP_S_CLAMP 0x00000005 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE 0x00000006 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER 0x00000007 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP 0x00000008 +#define NV40TCL_TEX_WRAP_T_SHIFT 8 +#define NV40TCL_TEX_WRAP_T_MASK 0x0000ff00 +#define NV40TCL_TEX_WRAP_T_REPEAT 0x00000001 +#define NV40TCL_TEX_WRAP_T_MIRRORED_REPEAT 0x00000002 +#define NV40TCL_TEX_WRAP_T_CLAMP_TO_EDGE 0x00000003 +#define NV40TCL_TEX_WRAP_T_CLAMP_TO_BORDER 0x00000004 +#define NV40TCL_TEX_WRAP_T_CLAMP 0x00000005 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_EDGE 0x00000006 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_BORDER 0x00000007 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP 0x00000008 +#define NV40TCL_TEX_WRAP_R_SHIFT 16 +#define NV40TCL_TEX_WRAP_R_MASK 0x00ff0000 +#define NV40TCL_TEX_WRAP_R_REPEAT 0x00000001 +#define NV40TCL_TEX_WRAP_R_MIRRORED_REPEAT 0x00000002 +#define NV40TCL_TEX_WRAP_R_CLAMP_TO_EDGE 0x00000003 +#define NV40TCL_TEX_WRAP_R_CLAMP_TO_BORDER 0x00000004 +#define NV40TCL_TEX_WRAP_R_CLAMP 0x00000005 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE 0x00000006 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER 0x00000007 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP 0x00000008 +#define NV40TCL_TEX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV40TCL_TEX_ENABLE__SIZE 0x00000010 +#define NV40TCL_TEX_ENABLE_ENABLE (1 << 31) +#define NV40TCL_TEX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV40TCL_TEX_SWIZZLE__SIZE 0x00000010 +#define NV40TCL_TEX_SWIZZLE_S0_X_SHIFT 14 +#define NV40TCL_TEX_SWIZZLE_S0_X_MASK 0x0000c000 +#define NV40TCL_TEX_SWIZZLE_S0_X_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_X_ONE 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S0_X_S1 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT 12 +#define NV40TCL_TEX_SWIZZLE_S0_Y_MASK 0x00003000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_ONE 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S0_Y_S1 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT 10 +#define NV40TCL_TEX_SWIZZLE_S0_Z_MASK 0x00000c00 +#define NV40TCL_TEX_SWIZZLE_S0_Z_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_Z_ONE 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S0_Z_S1 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S0_W_SHIFT 8 +#define NV40TCL_TEX_SWIZZLE_S0_W_MASK 0x00000300 +#define NV40TCL_TEX_SWIZZLE_S0_W_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_W_ONE 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S0_W_S1 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_X_SHIFT 6 +#define NV40TCL_TEX_SWIZZLE_S1_X_MASK 0x000000c0 +#define NV40TCL_TEX_SWIZZLE_S1_X_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_X_Z 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S1_X_Y 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_X_X 0x00000003 +#define NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT 4 +#define NV40TCL_TEX_SWIZZLE_S1_Y_MASK 0x00000030 +#define NV40TCL_TEX_SWIZZLE_S1_Y_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_Y_Z 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S1_Y_Y 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_Y_X 0x00000003 +#define NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT 2 +#define NV40TCL_TEX_SWIZZLE_S1_Z_MASK 0x0000000c +#define NV40TCL_TEX_SWIZZLE_S1_Z_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_Z_Z 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S1_Z_Y 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_Z_X 0x00000003 +#define NV40TCL_TEX_SWIZZLE_S1_W_SHIFT 0 +#define NV40TCL_TEX_SWIZZLE_S1_W_MASK 0x00000003 +#define NV40TCL_TEX_SWIZZLE_S1_W_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_W_Z 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S1_W_Y 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_W_X 0x00000003 +#define NV40TCL_TEX_FILTER(x) (0x00001a14+((x)*32)) +#define NV40TCL_TEX_FILTER__SIZE 0x00000010 +#define NV40TCL_TEX_FILTER_MIN_SHIFT 16 +#define NV40TCL_TEX_FILTER_MIN_MASK 0x000f0000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST 0x00000001 +#define NV40TCL_TEX_FILTER_MIN_LINEAR 0x00000002 +#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV40TCL_TEX_FILTER_MAG_SHIFT 24 +#define NV40TCL_TEX_FILTER_MAG_MASK 0x0f000000 +#define NV40TCL_TEX_FILTER_MAG_NEAREST 0x00000001 +#define NV40TCL_TEX_FILTER_MAG_LINEAR 0x00000002 +#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_NEAREST 0x00000003 +#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_NEAREST 0x00000004 +#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_LINEAR 0x00000005 +#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_LINEAR 0x00000006 +#define NV40TCL_TEX_SIZE0(x) (0x00001a18+((x)*32)) +#define NV40TCL_TEX_SIZE0__SIZE 0x00000010 +#define NV40TCL_TEX_SIZE0_H_SHIFT 0 +#define NV40TCL_TEX_SIZE0_H_MASK 0x0000ffff +#define NV40TCL_TEX_SIZE0_W_SHIFT 16 +#define NV40TCL_TEX_SIZE0_W_MASK 0xffff0000 +#define NV40TCL_TEX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV40TCL_TEX_BORDER_COLOR__SIZE 0x00000010 +#define NV40TCL_TEX_BORDER_COLOR_B_SHIFT 0 +#define NV40TCL_TEX_BORDER_COLOR_B_MASK 0x000000ff +#define NV40TCL_TEX_BORDER_COLOR_G_SHIFT 8 +#define NV40TCL_TEX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV40TCL_TEX_BORDER_COLOR_R_SHIFT 16 +#define NV40TCL_TEX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV40TCL_TEX_BORDER_COLOR_A_SHIFT 24 +#define NV40TCL_TEX_BORDER_COLOR_A_MASK 0xff000000 +#define NV40TCL_VTX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_W__SIZE 0x00000010 +#define NV40TCL_FP_CONTROL 0x00001d60 +#define NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT 24 +#define NV40TCL_FP_CONTROL_TEMP_COUNT_MASK 0xff000000 +#define NV40TCL_FP_CONTROL_KIL (1 << 7) +#define NV40TCL_CLEAR_VALUE_DEPTH 0x00001d8c +#define NV40TCL_CLEAR_VALUE_COLOR 0x00001d90 +#define NV40TCL_CLEAR_BUFFERS 0x00001d94 +#define NV40TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) +#define NV40TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) +#define NV40TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) +#define NV40TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) +#define NV40TCL_CLEAR_BUFFERS_STENCIL (1 << 1) +#define NV40TCL_CLEAR_BUFFERS_DEPTH (1 << 0) +#define NV40TCL_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV40TCL_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV40TCL_VP_UPLOAD_FROM_ID 0x00001e9c +#define NV40TCL_VP_START_FROM_ID 0x00001ea0 +#define NV40TCL_POINT_SIZE 0x00001ee0 +#define NV40TCL_POINT_SPRITE 0x00001ee8 +#define NV40TCL_VP_UPLOAD_CONST_ID 0x00001efc +#define NV40TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 +#define NV40TCL_TEX_CACHE_CTL 0x00001fd8 +#define NV40TCL_VP_ATTRIB_EN 0x00001ff0 +#define NV40TCL_VP_RESULT_EN 0x00001ff4 + + +#define NV44_TCL_PRIMITIVE_3D 0x00004497 + + + +#define NV50_PRIMITIVE_2D 0x0000502d + +#define NV50_PRIMITIVE_2D_DMA_NOTIFY 0x00000180 +#define NV50_PRIMITIVE_2D_DMA_IN_MEMORY0 0x00000184 +#define NV50_PRIMITIVE_2D_DMA_IN_MEMORY1 0x00000188 +#define NV50_PRIMITIVE_2D_DMA_IN_MEMORY2 0x0000018c + + +#define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 + + + +#define NV50_TCL_PRIMITIVE_3D 0x00005097 + +#define NV50_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV50_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV50_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0(x) (0x00000184+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0__SIZE 0x0000000b +#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1(x) (0x000001c0+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1__SIZE 0x00000008 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00000314 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F(x) (0x00000318+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F__SIZE 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F(x) (0x000003c0+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F__SIZE 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F(x) (0x000003c8+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F__SIZE 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F(x) (0x000003d0+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F__SIZE 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F(x) (0x000003d8+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F__SIZE 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F(x) (0x00000400+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F__SIZE 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F(x) (0x00000420+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F__SIZE 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F(x) (0x00000430+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F__SIZE 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F(x) (0x00000440+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F__SIZE 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X(x) (0x00000500+((x)*16)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X__SIZE 0x00000010 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y(x) (0x00000504+((x)*16)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y__SIZE 0x00000010 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z(x) (0x00000508+((x)*16)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z__SIZE 0x00000010 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W(x) (0x0000050c+((x)*16)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W__SIZE 0x00000010 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x000006a0 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x000006a4 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I 0x000006a8 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I 0x000006ac +#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY 0x00000700 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW 0x00000704 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x00000740 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x00000744 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x00000748 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x0000074c +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST 0x00000750 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ 0x00000754 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST 0x00000758 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ 0x0000075c +#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000790 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000794 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x0000088c +#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I 0x00000890 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00000900+((x)*16)) +#define NV50_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0(x) (0x00000a00+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0__SIZE 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1(x) (0x00000a0c+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1__SIZE 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000c00 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000c04 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_SHIFT 0 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_SHIFT 16 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000c08 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000c0c +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x00000d00+((x)*8)) +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x00000d04+((x)*8)) +#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_FIRST 0x00000d74 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_COUNT 0x00000d78 +#define NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR(x) (0x00000d80+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR__SIZE 0x00000004 +#define NV50_TCL_PRIMITIVE_3D_CLEAR_DEPTH 0x00000d90 +#define NV50_TCL_PRIMITIVE_3D_CLEAR_STENCIL 0x00000da0 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x00000dac +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000db0 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000db4 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000dc0 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000dc4 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000dc8 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x00000e04 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_L_SHIFT 0 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_L_MASK 0x0000ffff +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_R_SHIFT 16 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_R_MASK 0xffff0000 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x00000e08 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_T_SHIFT 0 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_T_MASK 0x0000ffff +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_B_SHIFT 16 +#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_B_MASK 0xffff0000 +#define NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID 0x00000f00 +#define NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST(x) (0x00000f04+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST__SIZE 0x00000010 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF 0x00000f54 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK 0x00000f58 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK 0x00000f5c +#define NV50_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x000012cc +#define NV50_TCL_PRIMITIVE_3D_SHADE_MODEL 0x000012d4 +#define NV50_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV50_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x000012e8 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x000012ec +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x0000130c +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00001310 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x00001314 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50_TCL_PRIMITIVE_3D_BLEND_COLOR(x) (0x0000131c+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_BLEND_COLOR__SIZE 0x00000004 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB 0x00001340 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_MIN 0x00008007 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_MAX 0x00008008 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB 0x00001344 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB 0x00001348 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA 0x0000134c +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_MIN 0x00008007 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_MAX 0x00008008 +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a +#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA 0x00001350 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA 0x00001358 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50_TCL_PRIMITIVE_3D_UNK1360(x) (0x00001360+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_UNK1360__SIZE 0x00000008 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE 0x00001380 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL 0x00001384 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL 0x00001388 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS 0x0000138c +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC 0x00001390 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF 0x00001394 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK 0x00001398 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK 0x0000139c +#define NV50_TCL_PRIMITIVE_3D_LINE_WIDTH 0x000013b0 +#define NV50_TCL_PRIMITIVE_3D_VP_INST 0x0000140c +#define NV50_TCL_PRIMITIVE_3D_POINT_SIZE 0x00001518 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x0000156c +#define NV50_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00001570 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE 0x00001594 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL 0x00001598 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL 0x0000159c +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS 0x000015a0 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC 0x000015a4 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x000015bc +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN 0x000015dc +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_POINTS 0x00000000 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_LINES 0x00000001 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_LINE_LOOP 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_LINE_STRIP 0x00000003 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_TRIANGLES 0x00000004 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_TRIANGLE_STRIP 0x00000005 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_TRIANGLE_FAN 0x00000006 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_QUADS 0x00000007 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_QUAD_STRIP 0x00000008 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_POLYGON 0x00000009 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_END 0x000015e0 +#define NV50_TCL_PRIMITIVE_3D_VERTEX_DATA 0x00001640 +#define NV50_TCL_PRIMITIVE_3D_VP_INST_PARAMS(x) (0x00001650+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_VP_INST_PARAMS__SIZE 0x00000002 +#define NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE 0x0000166c +#define NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN 0x00001680 +#define NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000168c +#define NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001700+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00001918 +#define NV50_TCL_PRIMITIVE_3D_FRONT_FACE 0x0000191c +#define NV50_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV50_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV50_TCL_PRIMITIVE_3D_CULL_FACE 0x00001920 +#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_ENABLE 0x000019c4 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP 0x000019c8 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_AND 0x00001501 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_COPY 0x00001503 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_NOOP 0x00001505 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_XOR 0x00001506 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_OR 0x00001507 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_NOR 0x00001508 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_INVERT 0x0000150a +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_NAND 0x0000150e +#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_SET 0x0000150f +#define NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS 0x000019d0 +#define NV50_TCL_PRIMITIVE_3D_COLOR_MASK(x) (0x00001a00+((x)*4)) +#define NV50_TCL_PRIMITIVE_3D_COLOR_MASK__SIZE 0x00000008 + + +#define NV52_TCL_PRIMITIVE_3D 0x00008297 + + + +#endif /* NOUVEAU_REG_H */ diff --git a/src/mesa/pipe/nouveau/nouveau_grobj.h b/src/mesa/pipe/nouveau/nouveau_grobj.h new file mode 100644 index 00000000000..a36db7f3bc4 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_grobj.h @@ -0,0 +1,41 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_GROBJ_H__ +#define __NOUVEAU_GROBJ_H__ + +#include "nouveau_channel.h" + +struct nouveau_grobj { + struct nouveau_channel *channel; + int grclass; + uint32_t handle; + + enum { + NOUVEAU_GROBJ_UNBOUND = 0, + NOUVEAU_GROBJ_BOUND = 1, + NOUVEAU_GROBJ_EXPLICIT_BIND = 2, + } bound; + int subc; +}; + +#endif diff --git a/src/mesa/pipe/nouveau/nouveau_notifier.h b/src/mesa/pipe/nouveau/nouveau_notifier.h new file mode 100644 index 00000000000..35adde1e324 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_notifier.h @@ -0,0 +1,43 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_NOTIFIER_H__ +#define __NOUVEAU_NOTIFIER_H__ + +#define NV_NOTIFIER_SIZE 32 +#define NV_NOTIFY_TIME_0 0x00000000 +#define NV_NOTIFY_TIME_1 0x00000004 +#define NV_NOTIFY_RETURN_VALUE 0x00000008 +#define NV_NOTIFY_STATE 0x0000000C +#define NV_NOTIFY_STATE_STATUS_MASK 0xFF000000 +#define NV_NOTIFY_STATE_STATUS_SHIFT 24 +#define NV_NOTIFY_STATE_STATUS_COMPLETED 0x00 +#define NV_NOTIFY_STATE_STATUS_IN_PROCESS 0x01 +#define NV_NOTIFY_STATE_ERROR_CODE_MASK 0x0000FFFF +#define NV_NOTIFY_STATE_ERROR_CODE_SHIFT 0 + +struct nouveau_notifier { + struct nouveau_channel *channel; + uint32_t handle; +}; + +#endif diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h new file mode 100644 index 00000000000..beee49a134f --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -0,0 +1,53 @@ +#ifndef NOUVEAU_WINSYS_H +#define NOUVEAU_WINSYS_H + +#include +#include "pipe/p_winsys.h" + +#include "pipe/nouveau/nouveau_bo.h" +#include "pipe/nouveau/nouveau_channel.h" +#include "pipe/nouveau/nouveau_class.h" +#include "pipe/nouveau/nouveau_grobj.h" +#include "pipe/nouveau/nouveau_notifier.h" + +struct nouveau_winsys { + struct nouveau_context *nv; + + struct nouveau_channel *channel; + + /*XXX: this is crappy, and bound to be slow.. however, it's nice and + * simple, it'll do for the moment*/ + uint32_t *(*begin_ring)(struct nouveau_grobj *, int mthd, int size); + void (*out_reloc)(struct nouveau_channel *, void *ptr, + struct nouveau_bo *, uint32_t data, + uint32_t flags, uint32_t vor, uint32_t tor); + void (*fire_ring)(struct nouveau_channel *); + + int (*grobj_alloc)(struct nouveau_winsys *, int grclass, + struct nouveau_grobj **); + void (*grobj_free)(struct nouveau_grobj **); + + int (*notifier_alloc)(struct nouveau_winsys *, int count, + struct nouveau_notifier **); + void (*notifier_free)(struct nouveau_notifier **); + void (*notifier_reset)(struct nouveau_notifier *, int id); + uint32_t (*notifier_status)(struct nouveau_notifier *, int id); + uint32_t (*notifier_retval)(struct nouveau_notifier *, int id); + int (*notifier_wait)(struct nouveau_notifier *, int id, + int status, int timeout); + + int (*region_copy)(struct nouveau_context *, struct pipe_region *, + unsigned, unsigned, unsigned, struct pipe_region *, + unsigned, unsigned, unsigned, unsigned, unsigned); + int (*region_fill)(struct nouveau_context *, struct pipe_region *, + unsigned, unsigned, unsigned, unsigned, unsigned, + unsigned); + int (*region_data)(struct nouveau_context *, struct pipe_region *, + unsigned, unsigned, unsigned, const void *, + unsigned, unsigned, unsigned, unsigned, unsigned); +}; + +extern struct pipe_context * +nv40_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); + +#endif diff --git a/src/mesa/pipe/nv40/Makefile b/src/mesa/pipe/nv40/Makefile new file mode 100644 index 00000000000..90c8542da45 --- /dev/null +++ b/src/mesa/pipe/nv40/Makefile @@ -0,0 +1,30 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv40 + +DRIVER_SOURCES = \ + nv40_clear.c \ + nv40_context.c \ + nv40_draw.c \ + nv40_fragprog.c \ + nv40_miptree.c \ + nv40_query.c \ + nv40_region.c \ + nv40_state.c \ + nv40_state_emit.c \ + nv40_state_tex.c \ + nv40_surface.c \ + nv40_vbo.c \ + nv40_vertprog.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../Makefile.template + +symlinks: + diff --git a/src/mesa/pipe/nv40/nv40_clear.c b/src/mesa/pipe/nv40/nv40_clear.c new file mode 100644 index 00000000000..f3b7a23689f --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_clear.c @@ -0,0 +1,21 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv40_context.h" +#include "nv40_dma.h" + + +void +nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + /*XXX: We're actually Z24_S8... */ + if (ps->format == PIPE_FORMAT_S8_Z24) { + clearValue = (((clearValue & 0xff000000) >> 24) | + ((clearValue & 0x00ffffff) << 8)); + } + + pipe->region_fill(pipe, ps->region, 0, 0, 0, ps->width, ps->height, + clearValue); +} diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c new file mode 100644 index 00000000000..ff66095c5fc --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -0,0 +1,277 @@ +#include "pipe/draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" +#include "nv40_dma.h" + +static boolean +nv40_is_format_supported(struct pipe_context *pipe, uint format) +{ + switch (format) { + case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_U_R5_G6_B5: + case PIPE_FORMAT_S8_Z24: + return TRUE; + default: + break; + }; + + return FALSE; +} + +static const char * +nv40_get_name(struct pipe_context *pipe) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv40->chipset); + return buffer; +} + +static const char * +nv40_get_vendor(struct pipe_context *pipe) +{ + return "nouveau"; +} + +static int +nv40_get_param(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 16; + case PIPE_CAP_NPOT_TEXTURES: + return 1; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 0; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 4; + case PIPE_CAP_OCCLUSION_QUERY: + return 1; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv40_get_paramf(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv40_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + if (flags & PIPE_FLUSH_TEXTURE_CACHE) { + BEGIN_RING(curie, 0x1fd8, 1); + OUT_RING (2); + BEGIN_RING(curie, 0x1fd8, 1); + OUT_RING (1); + } + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv40->sync, 0); + BEGIN_RING(curie, 0x104, 1); + OUT_RING (0); + BEGIN_RING(curie, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv40->sync, 0, 0, 2000); +} + +static void +nv40_destroy(struct pipe_context *pipe) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + draw_destroy(nv40->draw); + free(nv40); +} + +static boolean +nv40_init_hwctx(struct nv40_context *nv40, int curie_class) +{ + struct nouveau_winsys *nvws = nv40->nvws; + int ret; + + if ((ret = nvws->notifier_alloc(nvws, nv40->num_query_objects, + &nv40->query))) { + NOUVEAU_ERR("Error creating query notifier objects: %d\n", ret); + return FALSE; + } + + if ((ret = nvws->grobj_alloc(nvws, curie_class, + &nv40->curie))) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + BEGIN_RING(curie, NV40TCL_DMA_NOTIFY, 1); + OUT_RING (nv40->sync->handle); + BEGIN_RING(curie, NV40TCL_DMA_TEXTURE0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(curie, NV40TCL_DMA_VTXBUF0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(curie, NV40TCL_DMA_FENCE, 2); + OUT_RING (0); + OUT_RING (nv40->query->handle); + BEGIN_RING(curie, NV40TCL_DMA_UNK01AC, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + + BEGIN_RING(curie, 0x1ea4, 3); + OUT_RING (0x00000010); + OUT_RING (0x01000100); + OUT_RING (0xff800006); + + /* vtxprog output routing */ + BEGIN_RING(curie, 0x1fc4, 1); + OUT_RING (0x06144321); + BEGIN_RING(curie, 0x1fc8, 2); + OUT_RING (0xedcba987); + OUT_RING (0x00000021); + BEGIN_RING(curie, 0x1fd0, 1); + OUT_RING (0x00171615); + BEGIN_RING(curie, 0x1fd4, 1); + OUT_RING (0x001b1a19); + + BEGIN_RING(curie, 0x1ef8, 1); + OUT_RING (0x0020ffff); + BEGIN_RING(curie, 0x1d64, 1); + OUT_RING (0x00d30000); + BEGIN_RING(curie, 0x1e94, 1); + OUT_RING (0x00000001); + + FIRE_RING (); + return TRUE; +} + +#define GRCLASS4097_CHIPSETS 0x00000baf +#define GRCLASS4497_CHIPSETS 0x00005450 +struct pipe_context * +nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv40_context *nv40; + int curie_class, ret; + + if ((chipset & 0xf0) != 0x40) { + NOUVEAU_ERR("Not a NV4X chipset\n"); + return NULL; + } + + if (GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) { + curie_class = 0x4097; + } else + if (GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) { + curie_class = 0x4497; + } else { + NOUVEAU_ERR("Unknown NV4X chipset: NV%02x\n", chipset); + return NULL; + } + + nv40 = CALLOC_STRUCT(nv40_context); + if (!nv40) + return NULL; + nv40->chipset = chipset; + nv40->nvws = nvws; + + if ((ret = nvws->notifier_alloc(nvws, 1, &nv40->sync))) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + free(nv40); + return NULL; + } + + nv40->num_query_objects = 32; + nv40->query_objects = calloc(nv40->num_query_objects, + sizeof(struct pipe_query_object *)); + if (!nv40->query_objects) { + free(nv40); + return NULL; + } + + if (!nv40_init_hwctx(nv40, curie_class)) { + free(nv40); + return NULL; + } + + nv40->pipe.winsys = pipe_winsys; + + nv40->pipe.destroy = nv40_destroy; + nv40->pipe.is_format_supported = nv40_is_format_supported; + nv40->pipe.get_name = nv40_get_name; + nv40->pipe.get_vendor = nv40_get_vendor; + nv40->pipe.get_param = nv40_get_param; + nv40->pipe.get_paramf = nv40_get_paramf; + + nv40->pipe.draw_arrays = nv40_draw_arrays; + nv40->pipe.draw_elements = nv40_draw_elements; + nv40->pipe.clear = nv40_clear; + + nv40->pipe.begin_query = nv40_query_begin; + nv40->pipe.end_query = nv40_query_end; + nv40->pipe.wait_query = nv40_query_wait; + + nv40->pipe.mipmap_tree_layout = nv40_miptree_layout; + + nv40->pipe.flush = nv40_flush; + + nv40_init_region_functions(nv40); + nv40_init_surface_functions(nv40); + nv40_init_state_functions(nv40); + + nv40->draw = draw_create(); + assert(nv40->draw); + draw_set_rasterize_stage(nv40->draw, nv40_draw_render_stage(nv40)); + + return &nv40->pipe; +} + + diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h new file mode 100644 index 00000000000..63be38299fb --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -0,0 +1,111 @@ +#ifndef __NV40_CONTEXT_H__ +#define __NV40_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/draw/draw_vertex.h" + +#include "pipe/nouveau/nouveau_winsys.h" + +#include "nv40_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV40_NEW_TEXTURE (1 << 0) +#define NV40_NEW_VERTPROG (1 << 1) +#define NV40_NEW_FRAGPROG (1 << 2) +#define NV40_NEW_ARRAYS (1 << 3) + +struct nv40_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct draw_context *draw; + + int chipset; + struct nouveau_grobj *curie; + struct nouveau_notifier *sync; + uint32_t *pushbuf; + + /* query objects */ + struct nouveau_notifier *query; + struct pipe_query_object **query_objects; + uint num_query_objects; + + uint32_t dirty; + + struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct pipe_mipmap_tree *tex_miptree[PIPE_MAX_SAMPLERS]; + uint32_t tex_dirty; + + struct { + struct nv40_vertex_program *vp; + struct nv40_vertex_program *active_vp; + + struct pipe_buffer_handle *constant_buf; + } vertprog; + + struct { + struct nv40_fragment_program *fp; + struct nv40_fragment_program *active_fp; + + struct pipe_buffer_handle *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; +}; + + +extern void nv40_init_region_functions(struct nv40_context *nv40); +extern void nv40_init_surface_functions(struct nv40_context *nv40); +extern void nv40_init_state_functions(struct nv40_context *nv40); + +/* nv40_draw.c */ +extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); + +/* nv40_miptree.c */ +extern boolean nv40_miptree_layout(struct pipe_context *, + struct pipe_mipmap_tree *); + +/* nv40_vertprog.c */ +extern void nv40_vertprog_translate(struct nv40_context *, + struct nv40_vertex_program *); +extern void nv40_vertprog_bind(struct nv40_context *, + struct nv40_vertex_program *); + +/* nv40_fragprog.c */ +extern void nv40_fragprog_translate(struct nv40_context *, + struct nv40_fragment_program *); +extern void nv40_fragprog_bind(struct nv40_context *, + struct nv40_fragment_program *); + +/* nv40_state.c and friends */ +extern void nv40_emit_hw_state(struct nv40_context *nv40); +extern void nv40_state_tex_update(struct nv40_context *nv40); + +/* nv40_vbo.c */ +extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv40_draw_elements(struct pipe_context *pipe, + struct pipe_buffer_handle *indexBuffer, + unsigned indexSize, + unsigned mode, unsigned start, + unsigned count); +extern void nv40_vbo_arrays_update(struct nv40_context *nv40); + +/* nv40_clear.c */ +extern void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +/* nv40_query.c */ +extern void nv40_query_begin(struct pipe_context *, struct pipe_query_object *); +extern void nv40_query_end(struct pipe_context *, struct pipe_query_object *); +extern void nv40_query_wait(struct pipe_context *, struct pipe_query_object *); + +#endif diff --git a/src/mesa/pipe/nv40/nv40_dma.h b/src/mesa/pipe/nv40/nv40_dma.h new file mode 100644 index 00000000000..3775ce6e72d --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_dma.h @@ -0,0 +1,62 @@ +#ifndef __NV40_DMA_H__ +#define __NV40_DMA_H__ + +#include "pipe/nouveau/nouveau_winsys.h" + +#define BEGIN_RING(obj,mthd,size) do { \ + nv40->pushbuf = nv40->nvws->begin_ring(nv40->obj, (mthd), (size)); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + +#define OUT_RING(data) do { \ + (*nv40->pushbuf++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + memcpy(nv40->pushbuf, (src), (size) * 4); \ + nv40->pushbuf += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define FIRE_RING() do { \ + nv40->nvws->fire_ring(nv40->nvws->channel); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + nv40->nvws->out_reloc(nv40->nvws->channel, nv40->pushbuf, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ + OUT_RING(0); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + nv40->nvws->channel->vram->handle, \ + nv40->nvws->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/mesa/pipe/nv40/nv40_draw.c b/src/mesa/pipe/nv40/nv40_draw.c new file mode 100644 index 00000000000..52ce493ea2d --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_draw.c @@ -0,0 +1,63 @@ +#include "pipe/draw/draw_private.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" + +struct nv40_draw_stage { + struct draw_stage draw; + struct nv40_context *nv40; +}; + +static void +nv40_draw_begin(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_end(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_point(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_line(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_tri(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_reset_stipple_counter(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +struct draw_stage * +nv40_draw_render_stage(struct nv40_context *nv40) +{ + struct nv40_draw_stage *nv40draw = CALLOC_STRUCT(nv40_draw_stage); + + nv40draw->nv40 = nv40; + nv40draw->draw.draw = nv40->draw; + nv40draw->draw.begin = nv40_draw_begin; + nv40draw->draw.point = nv40_draw_point; + nv40draw->draw.line = nv40_draw_line; + nv40draw->draw.tri = nv40_draw_tri; + nv40draw->draw.end = nv40_draw_end; + nv40draw->draw.reset_stipple_counter = nv40_draw_reset_stipple_counter; + + return &nv40draw->draw; +} + diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c new file mode 100644 index 00000000000..48b783eebef --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -0,0 +1,642 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/tgsi/exec/tgsi_token.h" +#include "pipe/tgsi/exec/tgsi_parse.h" + +#include "nv40_context.h" +#include "nv40_dma.h" + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 1 +#define MASK_Y 2 +#define MASK_Z 4 +#define MASK_W 8 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE NV40_FP_OP_DST_SCALE_1X +#define DEF_CTEST NV40_FP_OP_COND_TR +#include "nv40_shader.h" + +#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv40_sr_neg((s)) +#define abs(s) nv40_sr_abs((s)) +#define scale(s,v) nv40_sr_scale((s), NV40_FP_OP_DST_SCALE_##v) + +static uint32_t +passthrough_fp_data[] = { + 0x01403e81, 0x1c9dc801, 0x0001c800, 0x3fe1c800 +}; + +static struct nv40_fragment_program +passthrough_fp = { + .pipe = NULL, + .translated = TRUE, + .insn = passthrough_fp_data, + .insn_len = sizeof(passthrough_fp_data) / sizeof(uint32_t), + .buffer = NULL, + .uses_kil = 0, + .num_regs = 2, +}; + +struct nv40_fpc { + struct nv40_fragment_program *fp; + + uint attrib_map[PIPE_MAX_SHADER_INPUTS]; + + int high_temp; + int temp_temp_count; + + uint depth_id; + uint colour_id; + + boolean inst_has_const; + int inst_const_id; +}; + +static INLINE struct nv40_sreg +nv40_sr_temp(struct nv40_fpc *fpc) +{ + int idx; + + idx = fpc->temp_temp_count++; + idx += fpc->high_temp + 1; + return nv40_sr(0, NV40_FP_REG_TYPE_TEMP, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv40_fp_arith((cc), (s), NV40_FP_OP_OPCODE_##o, \ + (d), (m), (s0), (s1), (s2)) +#define tex(cc,s,o,u,d,m,s0,s1,s2) \ + nv40_fp_tex((cc), (s), NV40_FP_OP_OPCODE_##o, (u), \ + (d), (m), (s0), none, none) +#define temp(fpc) nv40_sr_temp((fpc)) + +static void +emit_src(struct nv40_fpc *fpc, uint32_t *hw, int pos, struct nv40_sreg src) +{ + uint32_t sr = 0; + + sr |= (src.type << NV40_FP_REG_TYPE_SHIFT); + if (src.type == NV40_FP_REG_TYPE_INPUT) { + hw[0] |= (src.index << NV40_FP_OP_INPUT_SRC_SHIFT); + } else + if (src.type == NV40_FP_REG_TYPE_CONST) { + fpc->inst_has_const = TRUE; + } else + if (src.type == NV40_FP_REG_TYPE_TEMP) { + sr |= (src.index << NV40_FP_REG_SRC_SHIFT); + } + + if (src.negate) + sr |= NV40_FP_REG_NEGATE; + + if (src.abs) + hw[1] |= (1 << (29 + pos)); + + sr |= ((src.swz[0] << NV40_FP_REG_SWZ_X_SHIFT) | + (src.swz[1] << NV40_FP_REG_SWZ_Y_SHIFT) | + (src.swz[2] << NV40_FP_REG_SWZ_Z_SHIFT) | + (src.swz[3] << NV40_FP_REG_SWZ_W_SHIFT)); + + hw[pos + 1] |= sr; +} + +static void +emit_dst(struct nv40_fpc *fpc, uint32_t *hw, struct nv40_sreg dst) +{ + struct nv40_fragment_program *fp = fpc->fp; + + if (dst.output) { + if (dst.index == 1) { + fp->writes_depth = 1; + } else { + hw[0] |= NV40_FP_OP_UNK0_7; + } + } else { + if (fp->num_regs < (dst.index + 1)) + fp->num_regs = dst.index + 1; + } + hw[0] |= (dst.index << NV40_FP_OP_OUT_REG_SHIFT); +} + +static void +nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, + struct nv40_sreg dst, int mask, + struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) +{ + struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fp->insn_len]; + + fpc->inst_has_const = FALSE; + + if (op == NV40_FP_OP_OPCODE_KIL) + fp->uses_kil = TRUE; + hw[0] |= (op << NV40_FP_OP_OPCODE_SHIFT); + hw[0] |= (mask << NV40_FP_OP_OUTMASK_SHIFT); + hw[2] |= (dst.dst_scale << NV40_FP_OP_DST_SCALE_SHIFT); + + if (sat) + hw[0] |= NV40_FP_OP_OUT_SAT; + + if (dst.cc_update) + hw[0] |= NV40_FP_OP_COND_WRITE_ENABLE; + hw[1] |= (dst.cc_test << NV40_FP_OP_COND_SHIFT); + hw[1] |= ((dst.cc_swz[0] << NV40_FP_OP_COND_SWZ_X_SHIFT) | + (dst.cc_swz[1] << NV40_FP_OP_COND_SWZ_Y_SHIFT) | + (dst.cc_swz[2] << NV40_FP_OP_COND_SWZ_Z_SHIFT) | + (dst.cc_swz[3] << NV40_FP_OP_COND_SWZ_W_SHIFT)); + + emit_dst(fpc, hw, dst); + emit_src(fpc, hw, 0, s0); + emit_src(fpc, hw, 1, s1); + emit_src(fpc, hw, 2, s2); + + fp->insn_len += 4; + if (fpc->inst_has_const) { + fp->consts[fp->num_consts].pipe_id = fpc->inst_const_id; + fp->consts[fp->num_consts].hw_id = fp->insn_len; + fp->num_consts++; + fp->insn_len += 4; + } +} + +static void +nv40_fp_tex(struct nv40_fpc *fpc, int sat, int op, int unit, + struct nv40_sreg dst, int mask, + struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) +{ + struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fp->insn_len]; + + nv40_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); + hw[0] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); +} + +static INLINE struct nv40_sreg +tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) +{ + struct nv40_sreg src; + uint type, index; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + type = NV40_FP_REG_TYPE_INPUT; + index = fpc->attrib_map[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_CONSTANT: + type = NV40_FP_REG_TYPE_CONST; + index = fsrc->SrcRegister.Index; + break; + case TGSI_FILE_TEMPORARY: + type = NV40_FP_REG_TYPE_TEMP; + index = fsrc->SrcRegister.Index + 1; + if (fpc->high_temp < index) + fpc->high_temp = index; + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src = nv40_sr(0, type, index); + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv40_sreg +tgsi_dst(struct nv40_fpc *fpc, const struct tgsi_full_dst_register *fdst) { + int out, idx; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + out = 1; + if (fdst->DstRegister.Index == fpc->colour_id) + idx = 0; + else + idx = 1; + break; + case TGSI_FILE_TEMPORARY: + out = 0; + idx = fdst->DstRegister.Index + 1; + if (fpc->high_temp < idx) + fpc->high_temp = idx; + break; + case TGSI_FILE_NULL: + break; + default: + NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); + break; + } + + return nv40_sr(out, NV40_FP_REG_TYPE_TEMP, idx); +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, + const struct tgsi_full_instruction *finst) +{ + struct nv40_sreg src[3], dst, tmp; + struct nv40_sreg none = nv40_sr(0, NV40_FP_REG_TYPE_INPUT, 0); + int mask, sat, unit; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_RET) + return TRUE; + + fpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(fpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + NOUVEAU_MSG("extra src attr %d\n", + fsrc->SrcRegister.Index); + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_CONSTANT: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + case TGSI_FILE_SAMPLER: + unit = fsrc->SrcRegister.Index; + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_CMP: + tmp = temp(fpc); + arith(fpc, sat, MOV, dst, mask, src[2], none, none); + tmp.cc_update = 1; + arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); + dst.cc_test = NV40_VP_INST_COND_LT; + arith(fpc, sat, MOV, dst, mask, src[1], none, none); + break; + case TGSI_OPCODE_COS: + arith(fpc, sat, COS, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(fpc, sat, DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none); + arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X), + swz(src[1], W, W, W, W), none); + break; + case TGSI_OPCODE_DST: + arith(fpc, sat, DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(fpc, sat, EX2, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FLR: + arith(fpc, sat, FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(fpc, sat, FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_KIL: + case TGSI_OPCODE_KILP: + /*XXX: Which is NV, which is ARB kil? ARB implemented here. + *XXX: Don't need temp, can update CC0 without writing dst + */ + tmp = temp(fpc); + tmp.cc_update = 1; + arith(fpc, 0, MOV, tmp, MASK_ALL, src[0], none, none); + dst.cc_test = NV40_FP_OP_COND_LT; + arith(fpc, 0, KIL, dst, 0, none, none, none); + break; + case TGSI_OPCODE_LG2: + arith(fpc, sat, LG2, dst, mask, src[0], none, none); + break; +// case TGSI_OPCODE_LIT: + case TGSI_OPCODE_LRP: + tmp = temp(fpc); + arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); + arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); + break; + case TGSI_OPCODE_MAD: + arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(fpc, sat, MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(fpc, sat, MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(fpc, sat, MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(fpc); + arith(fpc, 0, LG2, tmp, MASK_X, + swz(src[0], X, X, X, X), none, none); + arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(fpc, sat, EX2, dst, mask, + swz(tmp, X, X, X, X), none, none); + break; + case TGSI_OPCODE_RCP: + arith(fpc, sat, RCP, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_RFL: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); + arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); + arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, + swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); + arith(fpc, sat, MAD, dst, mask, + swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); + break; + case TGSI_OPCODE_RSQ: + tmp = temp(fpc); + arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, + abs(swz(src[0], X, X, X, X)), none, none); + arith(fpc, sat, EX2, dst, mask, + neg(swz(tmp, X, X, X, X)), none, none); + break; + case TGSI_OPCODE_SCS: + if (mask & MASK_X) { + arith(fpc, sat, COS, dst, MASK_X, + swz(src[0], X, X, X, X), none, none); + } + if (mask & MASK_Y) { + arith(fpc, sat, SIN, dst, MASK_Y, + swz(src[0], X, X, X, X), none, none); + } + break; + case TGSI_OPCODE_SIN: + arith(fpc, sat, SIN, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_SGE: + arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); + break; + case TGSI_OPCODE_TEX: + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_TXB: + tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); + break; +#if 0 /* XXX: reimplement on top of TEX */ + case TGSI_OPCODE_TXP: + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + break; +#endif + case TGSI_OPCODE_XPD: + tmp = temp(fpc); + arith(fpc, 0, MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(fpc, sat, MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv40_fragprog_parse_decl_attrib(struct nv40_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV40_FP_OP_INPUT_SRC_POSITION; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV40_FP_OP_INPUT_SRC_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV40_FP_OP_INPUT_SRC_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV40_FP_OP_INPUT_SRC_FOGC; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV40_FP_OP_INPUT_SRC_TC(fdec->Semantic. + SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad input semantic\n"); + return FALSE; + } + + fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + fpc->depth_id = fdec->u.DeclarationRange.First; + break; + case TGSI_SEMANTIC_COLOR: + fpc->colour_id = fdec->u.DeclarationRange.First; + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + return TRUE; +} + +void +nv40_fragprog_translate(struct nv40_context *nv40, + struct nv40_fragment_program *fp) +{ + struct tgsi_parse_context parse; + struct nv40_fpc *fpc = NULL; + int i; + + fpc = calloc(1, sizeof(struct nv40_fpc)); + if (!fpc) + return; + fp->insn = calloc(1, 128*4*sizeof(uint32_t)); + fpc->fp = fp; + fpc->high_temp = -1; + fp->num_regs = 2; + + tgsi_parse_init(&parse, fp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_INPUT: + if (!nv40_fragprog_parse_decl_attrib(fpc, fdec)) + goto out_err; + break; + case TGSI_FILE_OUTPUT: + if (!nv40_fragprog_parse_decl_output(fpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + + finst = &parse.FullToken.FullInstruction; + if (!nv40_fragprog_parse_instruction(fpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + if (fpc->inst_has_const == FALSE) + fp->insn[fp->insn_len - 4] |= 0x00000001; + else + fp->insn[fp->insn_len - 8] |= 0x00000001; + fp->insn[fp->insn_len++] = 0x00000001; + + fp->translated = TRUE; + fp->on_hw = FALSE; +out_err: + tgsi_parse_free(&parse); + free(fpc); +} + +void +nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) +{ + struct pipe_winsys *ws = nv40->pipe.winsys; + uint32_t fp_control; + + if (!fp->translated) { + NOUVEAU_ERR("fragprog invalid, using passthrough shader\n"); + fp = &passthrough_fp; + } + + if (!fp->on_hw) { + if (!fp->buffer) + fp->buffer = ws->buffer_create(ws, 0x100); + + nv40->pipe.winsys->buffer_data(nv40->pipe.winsys, fp->buffer, + fp->insn_len * sizeof(uint32_t), + fp->insn, + PIPE_BUFFER_USAGE_PIXEL); + fp->on_hw = TRUE; + } + + fp_control = fp->num_regs << NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT; + if (fp->uses_kil) + fp_control |= NV40TCL_FP_CONTROL_KIL; + if (fp->writes_depth) + fp_control |= 0xe; + + BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); + OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); + BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); + OUT_RING (fp_control); + + nv40->fragprog.active_fp = fp; +} + diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c new file mode 100644 index 00000000000..6b85823d8cf --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -0,0 +1,60 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" + +boolean +nv40_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + uint width, height, depth, offset; + boolean swizzled = FALSE; + int l; + + mt->pitch = mt->width0; + mt->total_height = 0; + + width = mt->width0; + height = mt->height0; + depth = mt->depth0; + offset = 0; + for (l = mt->first_level; l <= mt->last_level; l++) { + uint pitch, f; + + mt->level[l].width = width; + mt->level[l].height = height; + mt->level[l].depth = depth; + mt->level[l].level_offset = offset; + + if (!swizzled) + pitch = mt->width0; + else + pitch = width; + + if (mt->target == PIPE_TEXTURE_CUBE) + mt->level[l].nr_images = 6; + else + if (mt->target == PIPE_TEXTURE_3D) + mt->level[l].nr_images = 3; + else + mt->level[l].nr_images = 1; + mt->level[l].image_offset = + malloc(mt->level[l].nr_images * sizeof(unsigned)); + + for (f = 0; f < mt->level[l].nr_images; f++) { + mt->level[l].image_offset[f] = + (offset - mt->level[l].level_offset) / mt->cpp; + mt->total_height += height; + + offset += (pitch * mt->cpp * height); + } + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + depth = MAX2(1, depth >> 1); + } + + return TRUE; +} + diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c new file mode 100644 index 00000000000..efd81e66408 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -0,0 +1,98 @@ +#include "pipe/p_context.h" + +#include "nv40_context.h" +#include "nv40_dma.h" + +static uint +nv40_query_object_find(struct nv40_context *nv40, struct pipe_query_object *q) +{ + int id; + + for (id = 0; id < nv40->num_query_objects; id++) { + if (nv40->query_objects[id] == q) + return id; + } + + return -1; +} + +void +nv40_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + int id; + + assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + id = nv40_query_object_find(nv40, NULL); + assert(id >= 0); + nv40->query_objects[id] = q; + + nv40->nvws->notifier_reset(nv40->query, id); + q->ready = 0; + + BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); + OUT_RING (1); + BEGIN_RING(curie, NV40TCL_QUERY_UNK17CC, 1); + OUT_RING (1); +} + +static void +nv40_query_update(struct pipe_context *pipe, struct pipe_query_object *q) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + int id; + + id = nv40_query_object_find(nv40, q); + assert(id >= 0); + + if (nv40->nvws->notifier_status(nv40->query, id) == 0) { + q->ready = 1; + q->count = nv40->nvws->notifier_retval(nv40->query, id); + nv40->query_objects[id] = NULL; + } +} + +void +nv40_query_end(struct pipe_context *pipe, struct pipe_query_object *q) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + int id; + + id = nv40_query_object_find(nv40, q); + assert(id >= 0); + + BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); + OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | + ((id * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); + FIRE_RING (); + + /*XXX: Some apps spin waiting for GL_QUERY_RESULT_AVAILABLE_ARB. + * Core mesa won't ask the driver to update the query object's + * status in this case, so the app waits forever.. fix this some + * day. + */ +#if 0 + nv40_query_update(pipe, q); +#else + nv40_query_wait(pipe, q); +#endif +} + +void +nv40_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) +{ + nv40_query_update(pipe, q); + if (!q->ready) { + struct nv40_context *nv40 = (struct nv40_context *)pipe; + int id; + + id = nv40_query_object_find(nv40, q); + assert(id >= 0); + + nv40->nvws->notifier_wait(nv40->query, id, 0, 0); + nv40_query_update(pipe, q); + assert(q->ready); + } +} + diff --git a/src/mesa/pipe/nv40/nv40_region.c b/src/mesa/pipe/nv40/nv40_region.c new file mode 100644 index 00000000000..f62bf89d183 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_region.c @@ -0,0 +1,85 @@ +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" + +#include "nv40_context.h" +#include "nv40_dma.h" + +static ubyte * +nv40_region_map(struct pipe_context *pipe, struct pipe_region *region) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct pipe_winsys *ws = nv40->pipe.winsys; + + if (!region->map_refcount++) { + region->map = ws->buffer_map(ws, region->buffer, + PIPE_BUFFER_FLAG_WRITE | + PIPE_BUFFER_FLAG_READ); + } + + return region->map; +} + +static void +nv40_region_unmap(struct pipe_context *pipe, struct pipe_region *region) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct pipe_winsys *ws = nv40->pipe.winsys; + + if (!--region->map_refcount) { + ws->buffer_unmap(ws, region->buffer); + region->map = NULL; + } +} + +static void +nv40_region_data(struct pipe_context *pipe, + struct pipe_region *dst, + unsigned dst_offset, + unsigned dstx, unsigned dsty, + const void *src, unsigned src_pitch, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->region_data(nvws->nv, dst, dst_offset, dstx, dsty, + src, src_pitch, srcx, srcy, width, height); +} + + +static void +nv40_region_copy(struct pipe_context *pipe, struct pipe_region *dst, + unsigned dst_offset, unsigned dstx, unsigned dsty, + struct pipe_region *src, unsigned src_offset, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->region_copy(nvws->nv, dst, dst_offset, dstx, dsty, + src, src_offset, srcx, srcy, width, height); +} + +static void +nv40_region_fill(struct pipe_context *pipe, + struct pipe_region *dst, unsigned dst_offset, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, unsigned value) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->region_fill(nvws->nv, dst, dst_offset, dstx, dsty, + width, height, value); +} + +void +nv40_init_region_functions(struct nv40_context *nv40) +{ + nv40->pipe.region_map = nv40_region_map; + nv40->pipe.region_unmap = nv40_region_unmap; + nv40->pipe.region_data = nv40_region_data; + nv40->pipe.region_copy = nv40_region_copy; + nv40->pipe.region_fill = nv40_region_fill; +} + diff --git a/src/mesa/pipe/nv40/nv40_shader.h b/src/mesa/pipe/nv40/nv40_shader.h new file mode 100644 index 00000000000..5b2cf3e2935 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_shader.h @@ -0,0 +1,546 @@ +#ifndef __NV40_SHADER_H__ +#define __NV40_SHADER_H__ + +/* Vertex programs instruction set + * + * The NV40 instruction set is very similar to NV30. Most fields are in + * a slightly different position in the instruction however. + * + * Merged instructions + * In some cases it is possible to put two instructions into one opcode + * slot. The rules for when this is OK is not entirely clear to me yet. + * + * There are separate writemasks and dest temp register fields for each + * grouping of instructions. There is however only one field with the + * ID of a result register. Writing to temp/result regs is selected by + * setting VEC_RESULT/SCA_RESULT. + * + * Temporary registers + * The source/dest temp register fields have been extended by 1 bit, to + * give a total of 32 temporary registers. + * + * Relative Addressing + * NV40 can use an address register to index into vertex attribute regs. + * This is done by putting the offset value into INPUT_SRC and setting + * the INDEX_INPUT flag. + * + * Conditional execution (see NV_vertex_program{2,3} for details) + * There is a second condition code register on NV40, it's use is enabled + * by setting the COND_REG_SELECT_1 flag. + * + * Texture lookup + * TODO + */ + +/* ---- OPCODE BITS 127:96 / data DWORD 0 --- */ +#define NV40_VP_INST_VEC_RESULT (1 << 30) +/* uncertain.. */ +#define NV40_VP_INST_COND_UPDATE_ENABLE ((1 << 14)|1<<29) +/* use address reg as index into attribs */ +#define NV40_VP_INST_INDEX_INPUT (1 << 27) +#define NV40_VP_INST_COND_REG_SELECT_1 (1 << 25) +#define NV40_VP_INST_ADDR_REG_SELECT_1 (1 << 24) +#define NV40_VP_INST_SRC2_ABS (1 << 23) +#define NV40_VP_INST_SRC1_ABS (1 << 22) +#define NV40_VP_INST_SRC0_ABS (1 << 21) +#define NV40_VP_INST_VEC_DEST_TEMP_SHIFT 15 +#define NV40_VP_INST_VEC_DEST_TEMP_MASK (0x1F << 15) +#define NV40_VP_INST_COND_TEST_ENABLE (1 << 13) +#define NV40_VP_INST_COND_SHIFT 10 +#define NV40_VP_INST_COND_MASK (0x7 << 10) +# define NV40_VP_INST_COND_FL 0 +# define NV40_VP_INST_COND_LT 1 +# define NV40_VP_INST_COND_EQ 2 +# define NV40_VP_INST_COND_LE 3 +# define NV40_VP_INST_COND_GT 4 +# define NV40_VP_INST_COND_NE 5 +# define NV40_VP_INST_COND_GE 6 +# define NV40_VP_INST_COND_TR 7 +#define NV40_VP_INST_COND_SWZ_X_SHIFT 8 +#define NV40_VP_INST_COND_SWZ_X_MASK (3 << 8) +#define NV40_VP_INST_COND_SWZ_Y_SHIFT 6 +#define NV40_VP_INST_COND_SWZ_Y_MASK (3 << 6) +#define NV40_VP_INST_COND_SWZ_Z_SHIFT 4 +#define NV40_VP_INST_COND_SWZ_Z_MASK (3 << 4) +#define NV40_VP_INST_COND_SWZ_W_SHIFT 2 +#define NV40_VP_INST_COND_SWZ_W_MASK (3 << 2) +#define NV40_VP_INST_COND_SWZ_ALL_SHIFT 2 +#define NV40_VP_INST_COND_SWZ_ALL_MASK (0xFF << 2) +#define NV40_VP_INST_ADDR_SWZ_SHIFT 0 +#define NV40_VP_INST_ADDR_SWZ_MASK (0x03 << 0) +#define NV40_VP_INST0_KNOWN ( \ + NV40_VP_INST_INDEX_INPUT | \ + NV40_VP_INST_COND_REG_SELECT_1 | \ + NV40_VP_INST_ADDR_REG_SELECT_1 | \ + NV40_VP_INST_SRC2_ABS | \ + NV40_VP_INST_SRC1_ABS | \ + NV40_VP_INST_SRC0_ABS | \ + NV40_VP_INST_VEC_DEST_TEMP_MASK | \ + NV40_VP_INST_COND_TEST_ENABLE | \ + NV40_VP_INST_COND_MASK | \ + NV40_VP_INST_COND_SWZ_ALL_MASK | \ + NV40_VP_INST_ADDR_SWZ_MASK) + +/* ---- OPCODE BITS 95:64 / data DWORD 1 --- */ +#define NV40_VP_INST_VEC_OPCODE_SHIFT 22 +#define NV40_VP_INST_VEC_OPCODE_MASK (0x1F << 22) +# define NV40_VP_INST_OP_NOP 0x00 +# define NV40_VP_INST_OP_MOV 0x01 +# define NV40_VP_INST_OP_MUL 0x02 +# define NV40_VP_INST_OP_ADD 0x03 +# define NV40_VP_INST_OP_MAD 0x04 +# define NV40_VP_INST_OP_DP3 0x05 +# define NV40_VP_INST_OP_DP4 0x07 +# define NV40_VP_INST_OP_DPH 0x06 +# define NV40_VP_INST_OP_DST 0x08 +# define NV40_VP_INST_OP_MIN 0x09 +# define NV40_VP_INST_OP_MAX 0x0A +# define NV40_VP_INST_OP_SLT 0x0B +# define NV40_VP_INST_OP_SGE 0x0C +# define NV40_VP_INST_OP_ARL 0x0D +# define NV40_VP_INST_OP_FRC 0x0E +# define NV40_VP_INST_OP_FLR 0x0F +# define NV40_VP_INST_OP_SEQ 0x10 +# define NV40_VP_INST_OP_SFL 0x11 +# define NV40_VP_INST_OP_SGT 0x12 +# define NV40_VP_INST_OP_SLE 0x13 +# define NV40_VP_INST_OP_SNE 0x14 +# define NV40_VP_INST_OP_STR 0x15 +# define NV40_VP_INST_OP_SSG 0x16 +# define NV40_VP_INST_OP_ARR 0x17 +# define NV40_VP_INST_OP_ARA 0x18 +# define NV40_VP_INST_OP_TXWHAT 0x19 +#define NV40_VP_INST_SCA_OPCODE_SHIFT 27 +#define NV40_VP_INST_SCA_OPCODE_MASK (0x1F << 27) +# define NV40_VP_INST_OP_RCP 0x02 +# define NV40_VP_INST_OP_RCC 0x03 +# define NV40_VP_INST_OP_RSQ 0x04 +# define NV40_VP_INST_OP_EXP 0x05 +# define NV40_VP_INST_OP_LOG 0x06 +# define NV40_VP_INST_OP_LIT 0x07 +# define NV40_VP_INST_OP_BRA 0x09 +# define NV40_VP_INST_OP_CAL 0x0B +# define NV40_VP_INST_OP_RET 0x0C +# define NV40_VP_INST_OP_LG2 0x0D +# define NV40_VP_INST_OP_EX2 0x0E +# define NV40_VP_INST_OP_SIN 0x0F +# define NV40_VP_INST_OP_COS 0x10 +# define NV40_VP_INST_OP_PUSHA 0x13 +# define NV40_VP_INST_OP_POPA 0x14 +#define NV40_VP_INST_CONST_SRC_SHIFT 12 +#define NV40_VP_INST_CONST_SRC_MASK (0xFF << 12) +#define NV40_VP_INST_INPUT_SRC_SHIFT 8 +#define NV40_VP_INST_INPUT_SRC_MASK (0x0F << 8) +# define NV40_VP_INST_IN_POS 0 +# define NV40_VP_INST_IN_WEIGHT 1 +# define NV40_VP_INST_IN_NORMAL 2 +# define NV40_VP_INST_IN_COL0 3 +# define NV40_VP_INST_IN_COL1 4 +# define NV40_VP_INST_IN_FOGC 5 +# define NV40_VP_INST_IN_TC0 8 +# define NV40_VP_INST_IN_TC(n) (8+n) +#define NV40_VP_INST_SRC0H_SHIFT 0 +#define NV40_VP_INST_SRC0H_MASK (0xFF << 0) +#define NV40_VP_INST1_KNOWN ( \ + NV40_VP_INST_VEC_OPCODE_MASK | \ + NV40_VP_INST_SCA_OPCODE_MASK | \ + NV40_VP_INST_CONST_SRC_MASK | \ + NV40_VP_INST_INPUT_SRC_MASK | \ + NV40_VP_INST_SRC0H_MASK \ + ) + +/* ---- OPCODE BITS 63:32 / data DWORD 2 --- */ +#define NV40_VP_INST_SRC0L_SHIFT 23 +#define NV40_VP_INST_SRC0L_MASK (0x1FF << 23) +#define NV40_VP_INST_SRC1_SHIFT 6 +#define NV40_VP_INST_SRC1_MASK (0x1FFFF << 6) +#define NV40_VP_INST_SRC2H_SHIFT 0 +#define NV40_VP_INST_SRC2H_MASK (0x3F << 0) +#define NV40_VP_INST_IADDRH_SHIFT 0 +#define NV40_VP_INST_IADDRH_MASK (0x1F << 0) + +/* ---- OPCODE BITS 31:0 / data DWORD 3 --- */ +#define NV40_VP_INST_IADDRL_SHIFT 29 +#define NV40_VP_INST_IADDRL_MASK (7 << 29) +#define NV40_VP_INST_SRC2L_SHIFT 21 +#define NV40_VP_INST_SRC2L_MASK (0x7FF << 21) +#define NV40_VP_INST_SCA_WRITEMASK_SHIFT 17 +#define NV40_VP_INST_SCA_WRITEMASK_MASK (0xF << 17) +# define NV40_VP_INST_SCA_WRITEMASK_X (1 << 20) +# define NV40_VP_INST_SCA_WRITEMASK_Y (1 << 19) +# define NV40_VP_INST_SCA_WRITEMASK_Z (1 << 18) +# define NV40_VP_INST_SCA_WRITEMASK_W (1 << 17) +#define NV40_VP_INST_VEC_WRITEMASK_SHIFT 13 +#define NV40_VP_INST_VEC_WRITEMASK_MASK (0xF << 13) +# define NV40_VP_INST_VEC_WRITEMASK_X (1 << 16) +# define NV40_VP_INST_VEC_WRITEMASK_Y (1 << 15) +# define NV40_VP_INST_VEC_WRITEMASK_Z (1 << 14) +# define NV40_VP_INST_VEC_WRITEMASK_W (1 << 13) +#define NV40_VP_INST_SCA_RESULT (1 << 12) +#define NV40_VP_INST_SCA_DEST_TEMP_SHIFT 7 +#define NV40_VP_INST_SCA_DEST_TEMP_MASK (0x1F << 7) +#define NV40_VP_INST_DEST_SHIFT 2 +#define NV40_VP_INST_DEST_MASK (31 << 2) +# define NV40_VP_INST_DEST_POS 0 +# define NV40_VP_INST_DEST_COL0 1 +# define NV40_VP_INST_DEST_COL1 2 +# define NV40_VP_INST_DEST_BFC0 3 +# define NV40_VP_INST_DEST_BFC1 4 +# define NV40_VP_INST_DEST_FOGC 5 +# define NV40_VP_INST_DEST_PSZ 6 +# define NV40_VP_INST_DEST_TC0 7 +# define NV40_VP_INST_DEST_TC(n) (7+n) +# define NV40_VP_INST_DEST_TEMP 0x1F +#define NV40_VP_INST_INDEX_CONST (1 << 1) +#define NV40_VP_INST_LAST (1 << 0) +#define NV40_VP_INST3_KNOWN ( \ + NV40_VP_INST_SRC2L_MASK |\ + NV40_VP_INST_SCA_WRITEMASK_MASK |\ + NV40_VP_INST_VEC_WRITEMASK_MASK |\ + NV40_VP_INST_SCA_DEST_TEMP_MASK |\ + NV40_VP_INST_DEST_MASK |\ + NV40_VP_INST_INDEX_CONST) + +/* Useful to split the source selection regs into their pieces */ +#define NV40_VP_SRC0_HIGH_SHIFT 9 +#define NV40_VP_SRC0_HIGH_MASK 0x0001FE00 +#define NV40_VP_SRC0_LOW_MASK 0x000001FF +#define NV40_VP_SRC2_HIGH_SHIFT 11 +#define NV40_VP_SRC2_HIGH_MASK 0x0001F800 +#define NV40_VP_SRC2_LOW_MASK 0x000007FF + +/* Source selection - these are the bits you fill NV40_VP_INST_SRCn with */ +#define NV40_VP_SRC_NEGATE (1 << 16) +#define NV40_VP_SRC_SWZ_X_SHIFT 14 +#define NV40_VP_SRC_SWZ_X_MASK (3 << 14) +#define NV40_VP_SRC_SWZ_Y_SHIFT 12 +#define NV40_VP_SRC_SWZ_Y_MASK (3 << 12) +#define NV40_VP_SRC_SWZ_Z_SHIFT 10 +#define NV40_VP_SRC_SWZ_Z_MASK (3 << 10) +#define NV40_VP_SRC_SWZ_W_SHIFT 8 +#define NV40_VP_SRC_SWZ_W_MASK (3 << 8) +#define NV40_VP_SRC_SWZ_ALL_SHIFT 8 +#define NV40_VP_SRC_SWZ_ALL_MASK (0xFF << 8) +#define NV40_VP_SRC_TEMP_SRC_SHIFT 2 +#define NV40_VP_SRC_TEMP_SRC_MASK (0x1F << 2) +#define NV40_VP_SRC_REG_TYPE_SHIFT 0 +#define NV40_VP_SRC_REG_TYPE_MASK (3 << 0) +# define NV40_VP_SRC_REG_TYPE_UNK0 0 +# define NV40_VP_SRC_REG_TYPE_TEMP 1 +# define NV40_VP_SRC_REG_TYPE_INPUT 2 +# define NV40_VP_SRC_REG_TYPE_CONST 3 + + +/* + * Each fragment program opcode appears to be comprised of 4 32-bit values. + * + * 0 - Opcode, output reg/mask, ATTRIB source + * 1 - Source 0 + * 2 - Source 1 + * 3 - Source 2 + * + * There appears to be no special difference between result regs and temp regs. + * result.color == R0.xyzw + * result.depth == R1.z + * When the fragprog contains instructions to write depth, + * NV30_TCL_PRIMITIVE_3D_UNK1D78=0 otherwise it is set to 1. + * + * Constants are inserted directly after the instruction that uses them. + * + * It appears that it's not possible to use two input registers in one + * instruction as the input sourcing is done in the instruction dword + * and not the source selection dwords. As such instructions such as: + * + * ADD result.color, fragment.color, fragment.texcoord[0]; + * + * must be split into two MOV's and then an ADD (nvidia does this) but + * I'm not sure why it's not just one MOV and then source the second input + * in the ADD instruction.. + * + * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary + * negation requires multiplication with a const. + * + * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO and + * SWIZZLE_ONE. + * + * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as + * SWIZZLE_ZERO is implemented simply by not writing to the relevant components + * of the destination. + * + * Looping + * Loops appear to be fairly expensive on NV40 at least, the proprietary + * driver goes to a lot of effort to avoid using the native looping + * instructions. If the total number of *executed* instructions between + * REP/ENDREP or LOOP/ENDLOOP is <=500, the driver will unroll the loop. + * The maximum loop count is 255. + * + * Conditional execution + * TODO + * + * Non-native instructions: + * LIT + * LRP - MAD+MAD + * SUB - ADD, negate second source + * RSQ - LG2 + EX2 + * POW - LG2 + MUL + EX2 + * SCS - COS + SIN + * XPD + * DP2 - MUL + ADD + * NRM + */ + +//== Opcode / Destination selection == +#define NV40_FP_OP_PROGRAM_END (1 << 0) +#define NV40_FP_OP_OUT_REG_SHIFT 1 +#define NV40_FP_OP_OUT_REG_MASK (31 << 1) +/* Needs to be set when writing outputs to get expected result.. */ +#define NV40_FP_OP_UNK0_7 (1 << 7) +#define NV40_FP_OP_COND_WRITE_ENABLE (1 << 8) +#define NV40_FP_OP_OUTMASK_SHIFT 9 +#define NV40_FP_OP_OUTMASK_MASK (0xF << 9) +# define NV40_FP_OP_OUT_X (1 << 9) +# define NV40_FP_OP_OUT_Y (1 <<10) +# define NV40_FP_OP_OUT_Z (1 <<11) +# define NV40_FP_OP_OUT_W (1 <<12) +/* Uncertain about these, especially the input_src values.. it's possible that + * they can be dynamically changed. + */ +#define NV40_FP_OP_INPUT_SRC_SHIFT 13 +#define NV40_FP_OP_INPUT_SRC_MASK (15 << 13) +# define NV40_FP_OP_INPUT_SRC_POSITION 0x0 +# define NV40_FP_OP_INPUT_SRC_COL0 0x1 +# define NV40_FP_OP_INPUT_SRC_COL1 0x2 +# define NV40_FP_OP_INPUT_SRC_FOGC 0x3 +# define NV40_FP_OP_INPUT_SRC_TC0 0x4 +# define NV40_FP_OP_INPUT_SRC_TC(n) (0x4 + n) +# define NV40_FP_OP_INPUT_SRC_FACING 0xE +#define NV40_FP_OP_TEX_UNIT_SHIFT 17 +#define NV40_FP_OP_TEX_UNIT_MASK (0xF << 17) +#define NV40_FP_OP_PRECISION_SHIFT 22 +#define NV40_FP_OP_PRECISION_MASK (3 << 22) +# define NV40_FP_PRECISION_FP32 0 +# define NV40_FP_PRECISION_FP16 1 +# define NV40_FP_PRECISION_FX12 2 +#define NV40_FP_OP_OPCODE_SHIFT 24 +#define NV40_FP_OP_OPCODE_MASK (0x3F << 24) +# define NV40_FP_OP_OPCODE_NOP 0x00 +# define NV40_FP_OP_OPCODE_MOV 0x01 +# define NV40_FP_OP_OPCODE_MUL 0x02 +# define NV40_FP_OP_OPCODE_ADD 0x03 +# define NV40_FP_OP_OPCODE_MAD 0x04 +# define NV40_FP_OP_OPCODE_DP3 0x05 +# define NV40_FP_OP_OPCODE_DP4 0x06 +# define NV40_FP_OP_OPCODE_DST 0x07 +# define NV40_FP_OP_OPCODE_MIN 0x08 +# define NV40_FP_OP_OPCODE_MAX 0x09 +# define NV40_FP_OP_OPCODE_SLT 0x0A +# define NV40_FP_OP_OPCODE_SGE 0x0B +# define NV40_FP_OP_OPCODE_SLE 0x0C +# define NV40_FP_OP_OPCODE_SGT 0x0D +# define NV40_FP_OP_OPCODE_SNE 0x0E +# define NV40_FP_OP_OPCODE_SEQ 0x0F +# define NV40_FP_OP_OPCODE_FRC 0x10 +# define NV40_FP_OP_OPCODE_FLR 0x11 +# define NV40_FP_OP_OPCODE_KIL 0x12 +# define NV40_FP_OP_OPCODE_PK4B 0x13 +# define NV40_FP_OP_OPCODE_UP4B 0x14 +/* DDX/DDY can only write to XY */ +# define NV40_FP_OP_OPCODE_DDX 0x15 +# define NV40_FP_OP_OPCODE_DDY 0x16 +# define NV40_FP_OP_OPCODE_TEX 0x17 +# define NV40_FP_OP_OPCODE_TXP 0x18 +# define NV40_FP_OP_OPCODE_TXD 0x19 +# define NV40_FP_OP_OPCODE_RCP 0x1A +# define NV40_FP_OP_OPCODE_EX2 0x1C +# define NV40_FP_OP_OPCODE_LG2 0x1D +# define NV40_FP_OP_OPCODE_COS 0x22 +# define NV40_FP_OP_OPCODE_SIN 0x23 +# define NV40_FP_OP_OPCODE_PK2H 0x24 +# define NV40_FP_OP_OPCODE_UP2H 0x25 +# define NV40_FP_OP_OPCODE_PK4UB 0x27 +# define NV40_FP_OP_OPCODE_UP4UB 0x28 +# define NV40_FP_OP_OPCODE_PK2US 0x29 +# define NV40_FP_OP_OPCODE_UP2US 0x2A +# define NV40_FP_OP_OPCODE_DP2A 0x2E +# define NV40_FP_OP_OPCODE_TXL 0x2F +# define NV40_FP_OP_OPCODE_TXB 0x31 +# define NV40_FP_OP_OPCODE_DIV 0x3A +# define NV40_FP_OP_OPCODE_UNK_LIT 0x3C +/* The use of these instructions appears to be indicated by bit 31 of DWORD 2.*/ +# define NV40_FP_OP_BRA_OPCODE_BRK 0x0 +# define NV40_FP_OP_BRA_OPCODE_CAL 0x1 +# define NV40_FP_OP_BRA_OPCODE_IF 0x2 +# define NV40_FP_OP_BRA_OPCODE_LOOP 0x3 +# define NV40_FP_OP_BRA_OPCODE_REP 0x4 +# define NV40_FP_OP_BRA_OPCODE_RET 0x5 +#define NV40_FP_OP_OUT_SAT (1 << 31) + +/* high order bits of SRC0 */ +#define NV40_FP_OP_OUT_ABS (1 << 29) +#define NV40_FP_OP_COND_SWZ_W_SHIFT 27 +#define NV40_FP_OP_COND_SWZ_W_MASK (3 << 27) +#define NV40_FP_OP_COND_SWZ_Z_SHIFT 25 +#define NV40_FP_OP_COND_SWZ_Z_MASK (3 << 25) +#define NV40_FP_OP_COND_SWZ_Y_SHIFT 23 +#define NV40_FP_OP_COND_SWZ_Y_MASK (3 << 23) +#define NV40_FP_OP_COND_SWZ_X_SHIFT 21 +#define NV40_FP_OP_COND_SWZ_X_MASK (3 << 21) +#define NV40_FP_OP_COND_SWZ_ALL_SHIFT 21 +#define NV40_FP_OP_COND_SWZ_ALL_MASK (0xFF << 21) +#define NV40_FP_OP_COND_SHIFT 18 +#define NV40_FP_OP_COND_MASK (0x07 << 18) +# define NV40_FP_OP_COND_FL 0 +# define NV40_FP_OP_COND_LT 1 +# define NV40_FP_OP_COND_EQ 2 +# define NV40_FP_OP_COND_LE 3 +# define NV40_FP_OP_COND_GT 4 +# define NV40_FP_OP_COND_NE 5 +# define NV40_FP_OP_COND_GE 6 +# define NV40_FP_OP_COND_TR 7 + +/* high order bits of SRC1 */ +#define NV40_FP_OP_OPCODE_IS_BRANCH (1<<31) +#define NV40_FP_OP_DST_SCALE_SHIFT 28 +#define NV40_FP_OP_DST_SCALE_MASK (3 << 28) +#define NV40_FP_OP_DST_SCALE_1X 0 +#define NV40_FP_OP_DST_SCALE_2X 1 +#define NV40_FP_OP_DST_SCALE_4X 2 +#define NV40_FP_OP_DST_SCALE_8X 3 +#define NV40_FP_OP_DST_SCALE_INV_2X 5 +#define NV40_FP_OP_DST_SCALE_INV_4X 6 +#define NV40_FP_OP_DST_SCALE_INV_8X 7 + +/* SRC1 LOOP */ +#define NV40_FP_OP_LOOP_INCR_SHIFT 19 +#define NV40_FP_OP_LOOP_INCR_MASK (0xFF << 19) +#define NV40_FP_OP_LOOP_INDEX_SHIFT 10 +#define NV40_FP_OP_LOOP_INDEX_MASK (0xFF << 10) +#define NV40_FP_OP_LOOP_COUNT_SHIFT 2 +#define NV40_FP_OP_LOOP_COUNT_MASK (0xFF << 2) + +/* SRC1 IF */ +#define NV40_FP_OP_ELSE_ID_SHIFT 2 +#define NV40_FP_OP_ELSE_ID_MASK (0xFF << 2) + +/* SRC1 CAL */ +#define NV40_FP_OP_IADDR_SHIFT 2 +#define NV40_FP_OP_IADDR_MASK (0xFF << 2) + +/* SRC1 REP + * I have no idea why there are 3 count values here.. but they + * have always been filled with the same value in my tests so + * far.. + */ +#define NV40_FP_OP_REP_COUNT1_SHIFT 2 +#define NV40_FP_OP_REP_COUNT1_MASK (0xFF << 2) +#define NV40_FP_OP_REP_COUNT2_SHIFT 10 +#define NV40_FP_OP_REP_COUNT2_MASK (0xFF << 10) +#define NV40_FP_OP_REP_COUNT3_SHIFT 19 +#define NV40_FP_OP_REP_COUNT3_MASK (0xFF << 19) + +/* SRC2 REP/IF */ +#define NV40_FP_OP_END_ID_SHIFT 2 +#define NV40_FP_OP_END_ID_MASK (0xFF << 2) + +// SRC2 high-order +#define NV40_FP_OP_INDEX_INPUT (1 << 30) +#define NV40_FP_OP_ADDR_INDEX_SHIFT 19 +#define NV40_FP_OP_ADDR_INDEX_MASK (0xF << 19) + +//== Register selection == +#define NV40_FP_REG_TYPE_SHIFT 0 +#define NV40_FP_REG_TYPE_MASK (3 << 0) +# define NV40_FP_REG_TYPE_TEMP 0 +# define NV40_FP_REG_TYPE_INPUT 1 +# define NV40_FP_REG_TYPE_CONST 2 +#define NV40_FP_REG_SRC_SHIFT 2 +#define NV40_FP_REG_SRC_MASK (31 << 2) +#define NV40_FP_REG_UNK_0 (1 << 8) +#define NV40_FP_REG_SWZ_ALL_SHIFT 9 +#define NV40_FP_REG_SWZ_ALL_MASK (255 << 9) +#define NV40_FP_REG_SWZ_X_SHIFT 9 +#define NV40_FP_REG_SWZ_X_MASK (3 << 9) +#define NV40_FP_REG_SWZ_Y_SHIFT 11 +#define NV40_FP_REG_SWZ_Y_MASK (3 << 11) +#define NV40_FP_REG_SWZ_Z_SHIFT 13 +#define NV40_FP_REG_SWZ_Z_MASK (3 << 13) +#define NV40_FP_REG_SWZ_W_SHIFT 15 +#define NV40_FP_REG_SWZ_W_MASK (3 << 15) +# define NV40_FP_SWIZZLE_X 0 +# define NV40_FP_SWIZZLE_Y 1 +# define NV40_FP_SWIZZLE_Z 2 +# define NV40_FP_SWIZZLE_W 3 +#define NV40_FP_REG_NEGATE (1 << 17) + +struct nv40_sreg { + int output; + int type; + int index; + + int dst_scale; + + int negate; + int abs; + int swz[4]; + + int cc_update; + int cc_update_reg; + int cc_test; + int cc_test_reg; + int cc_swz[4]; +}; + +static INLINE struct nv40_sreg +nv40_sr(int out, int type, int index) +{ + struct nv40_sreg temp = { + .output = out, + .type = type, + .index = index, + .dst_scale = DEF_SCALE, + .abs = 0, + .negate = 0, + .swz = { 0, 1, 2, 3 }, + .cc_update = 0, + .cc_update_reg = 0, + .cc_test = DEF_CTEST, + .cc_test_reg = 0, + .cc_swz = { 0, 1, 2, 3 }, + }; + return temp; +} + +static INLINE struct nv40_sreg +nv40_sr_swz(struct nv40_sreg src, int x, int y, int z, int w) +{ + struct nv40_sreg dst = src; + + dst.swz[SWZ_X] = src.swz[x]; + dst.swz[SWZ_Y] = src.swz[y]; + dst.swz[SWZ_Z] = src.swz[z]; + dst.swz[SWZ_W] = src.swz[w]; + return dst; +} + +static INLINE struct nv40_sreg +nv40_sr_neg(struct nv40_sreg src) +{ + src.negate = !src.negate; + return src; +} + +static INLINE struct nv40_sreg +nv40_sr_abs(struct nv40_sreg src) +{ + src.abs = 1; + return src; +} + +static INLINE struct nv40_sreg +nv40_sr_scale(struct nv40_sreg src, int scale) +{ + src.dst_scale = scale; + return src; +} + +#endif diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c new file mode 100644 index 00000000000..e38a5ea534a --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -0,0 +1,674 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" +#include "nv40_dma.h" +#include "nv40_state.h" + +#include "nvgl_pipe.h" + +static void * +nv40_alpha_test_state_create(struct pipe_context *pipe, + const struct pipe_alpha_test_state *cso) +{ + struct nv40_alpha_test_state *at; + + at = malloc(sizeof(struct nv40_alpha_test_state)); + + at->enabled = cso->enabled ? 1 : 0; + if (at->enabled) { + at->func = nvgl_comparison_op(cso->func); + at->ref = float_to_ubyte(cso->ref); + } + + return (void *)at; +} + +static void +nv40_alpha_test_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_alpha_test_state *at = hwcso; + + if (at->enabled) { + BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + OUT_RING (at->enabled); + OUT_RING (at->func); + OUT_RING (at->ref); + } else { + BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 1); + OUT_RING (0); + } +} + +static void +nv40_alpha_test_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv40_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv40_blend_state *cb; + + cb = malloc(sizeof(struct nv40_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + if (cb->b_enable) { + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | + (nvgl_blend_eqn(cso->rgb_func))); + } + + cb->l_enable = cso->logicop_enable ? 1 : 0; + if (cb->l_enable) { + cb->l_op = nvgl_logicop_func(cso->logicop_func); + } + + cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + + cb->d_enable = cso->dither ? 1 : 0; + + return (void *)cb; +} + +static void +nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_blend_state *cb = hwcso; + + BEGIN_RING(curie, NV40TCL_DITHER_ENABLE, 1); + OUT_RING (cb->d_enable); + + if (cb->b_enable) { + BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 3); + OUT_RING (cb->b_enable); + OUT_RING (cb->b_srcfunc); + OUT_RING (cb->b_dstfunc); + BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 2); + OUT_RING (cb->b_eqn); + OUT_RING (cb->c_mask); + } else { + BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 1); + OUT_RING (0); + } + + if (cb->l_enable) { + BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + OUT_RING (cb->l_enable); + OUT_RING (cb->l_op); + } else { + BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); + OUT_RING (0); + } +} + +static void +nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv40_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv40_sampler_state *ps; + + ps = malloc(sizeof(struct nv40_sampler_state)); + + ps->wrap = ((nv40_tex_wrap_mode(cso->wrap_r) << 16) | + (nv40_tex_wrap_mode(cso->wrap_t) << 8) | + (nv40_tex_wrap_mode(cso->wrap_s) << 0)); + ps->filt = ((nv40_tex_filter(cso->min_img_filter, + cso->min_mip_filter) << 16) | + (nv40_tex_filter(cso->mag_img_filter, + PIPE_TEX_MIPFILTER_NONE) << 24)); + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, + void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_sampler_state *ps = hwcso; + + nv40->tex_sampler[unit] = ps; + nv40->tex_dirty |= (1 << unit); + + nv40->dirty |= NV40_NEW_TEXTURE; +} + +static void +nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv40_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv40_rasterizer_state *rs; + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + rs = malloc(sizeof(struct nv40_rasterizer_state)); + + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; + + rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; + rs->line_smooth_en = cso->line_smooth ? 1 : 0; + rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; + rs->line_stipple = (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor; + + rs->point_size = *(uint32_t*)&cso->point_size; + + rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; + rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; + + if (cso->front_winding == PIPE_WINDING_CCW) { + rs->front_face = 0x0901; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + } else { + rs->front_face = 0x0900; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + } + + rs->cull_face_en = 0; + rs->cull_face = 0x0900; + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + rs->cull_face = 0x0901; + /* fall-through */ + case PIPE_WINDING_CW: + rs->cull_face_en = 1; + break; + case PIPE_WINDING_NONE: + default: + break; + } + + return (void *)rs; +} + +static void +nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_rasterizer_state *rs = hwcso; + + BEGIN_RING(curie, NV40TCL_SHADE_MODEL, 1); + OUT_RING (rs->shade_model); + + BEGIN_RING(curie, NV40TCL_LINE_WIDTH, 2); + OUT_RING (rs->line_width); + OUT_RING (rs->line_smooth_en); + BEGIN_RING(curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + OUT_RING (rs->line_stipple_en); + OUT_RING (rs->line_stipple); + + BEGIN_RING(curie, NV40TCL_POINT_SIZE, 1); + OUT_RING (rs->point_size); + + BEGIN_RING(curie, NV40TCL_POLYGON_MODE_FRONT, 6); + OUT_RING (rs->poly_mode_front); + OUT_RING (rs->poly_mode_back); + OUT_RING (rs->cull_face); + OUT_RING (rs->front_face); + OUT_RING (rs->poly_smooth_en); + OUT_RING (rs->cull_face_en); + + BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + OUT_RING (rs->poly_stipple_en); +} + +static void +nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv40_depth_stencil_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_state *cso) +{ + struct nv40_depth_stencil_state *zs; + + /*XXX: ignored: + * depth.occlusion_count + * depth.clear + * stencil.clear_value + */ + zs = malloc(sizeof(struct nv40_depth_stencil_state)); + + zs->depth.func = nvgl_comparison_op(cso->depth.func); + zs->depth.write_enable = cso->depth.writemask ? 1 : 0; + zs->depth.test_enable = cso->depth.enabled ? 1 : 0; + + zs->stencil.back.enable = cso->stencil.back_enabled ? 1 : 0; + zs->stencil.back.wmask = cso->stencil.write_mask[1]; + zs->stencil.back.func = + nvgl_comparison_op(cso->stencil.back_func); + zs->stencil.back.ref = cso->stencil.ref_value[1]; + zs->stencil.back.vmask = cso->stencil.value_mask[1]; + zs->stencil.back.fail = nvgl_stencil_op(cso->stencil.back_fail_op); + zs->stencil.back.zfail = nvgl_stencil_op(cso->stencil.back_zfail_op); + zs->stencil.back.zpass = nvgl_stencil_op(cso->stencil.back_zpass_op); + + zs->stencil.front.enable= cso->stencil.front_enabled ? 1 : 0; + zs->stencil.front.wmask = cso->stencil.write_mask[0]; + zs->stencil.front.func = + nvgl_comparison_op(cso->stencil.front_func); + zs->stencil.front.ref = cso->stencil.ref_value[0]; + zs->stencil.front.vmask = cso->stencil.value_mask[0]; + zs->stencil.front.fail = nvgl_stencil_op(cso->stencil.front_fail_op); + zs->stencil.front.zfail = nvgl_stencil_op(cso->stencil.front_zfail_op); + zs->stencil.front.zpass = nvgl_stencil_op(cso->stencil.front_zpass_op); + + return (void *)zs; +} + +static void +nv40_depth_stencil_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_depth_stencil_state *zs = hwcso; + + BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); + OUT_RINGp ((uint32_t *)&zs->depth, 3); + BEGIN_RING(curie, NV40TCL_STENCIL_BACK_ENABLE, 16); + OUT_RINGp ((uint32_t *)&zs->stencil.back, 8); + OUT_RINGp ((uint32_t *)&zs->stencil.front, 8); +} + +static void +nv40_depth_stencil_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv40_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv40_vertex_program *vp; + + vp = calloc(1, sizeof(struct nv40_vertex_program)); + vp->pipe = cso; + + return (void *)vp; +} + +static void +nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_vertex_program *vp = hwcso; + + nv40->vertprog.vp = vp; + nv40->dirty |= NV40_NEW_VERTPROG; +} + +static void +nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv40_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv40_fragment_program *fp; + + fp = calloc(1, sizeof(struct nv40_fragment_program)); + fp->pipe = cso; + + return (void *)fp; +} + +static void +nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_fragment_program *fp = hwcso; + + nv40->fragprog.fp = fp; + nv40->dirty |= NV40_NEW_FRAGPROG; +} + +static void +nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv40_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + BEGIN_RING(curie, NV40TCL_BLEND_COLOR, 1); + OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0)); +} + +static void +nv40_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + nv40->dirty |= NV40_NEW_VERTPROG; +} + +static void +nv40_set_clear_color_state(struct pipe_context *pipe, + const struct pipe_clear_color_state *ccol) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + BEGIN_RING(curie, NV40TCL_CLEAR_VALUE_COLOR, 1); + OUT_RING ((float_to_ubyte(ccol->color[3]) << 24) | + (float_to_ubyte(ccol->color[0]) << 16) | + (float_to_ubyte(ccol->color[1]) << 8) | + (float_to_ubyte(ccol->color[2]) << 0)); +} + +static void +nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + if (shader == PIPE_SHADER_VERTEX) { + nv40->vertprog.constant_buf = buf->buffer; + nv40->dirty |= NV40_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv40->fragprog.constant_buf = buf->buffer; + nv40->dirty |= NV40_NEW_FRAGPROG; + } +} + +static void +nv40_set_feedback_state(struct pipe_context *pipe, + const struct pipe_feedback_state *feedback) +{ + NOUVEAU_ERR("\n"); +} + +#define get_region(surf) ((surf) ? surf->region : NULL) +static void +nv40_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + struct pipe_region *region; + uint32_t rt_enable = 0, rt_format = 0; + + if ((region = get_region(fb->cbufs[0]))) { + rt_enable |= NV40TCL_RT_ENABLE_COLOR0; + + BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); + OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); + OUT_RING (region->pitch * region->cpp); + OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if ((region = get_region(fb->cbufs[1]))) { + rt_enable |= NV40TCL_RT_ENABLE_COLOR1; + + BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); + OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); + OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (region->pitch * region->cpp); + } + + if ((region = get_region(fb->cbufs[2]))) { + rt_enable |= NV40TCL_RT_ENABLE_COLOR2; + + BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); + OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); + OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); + OUT_RING (region->pitch * region->cpp); + } + + if ((region = get_region(fb->cbufs[3]))) { + rt_enable |= NV40TCL_RT_ENABLE_COLOR3; + + BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); + OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); + OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); + OUT_RING (region->pitch * region->cpp); + } + + if ((region = get_region(fb->zbuf))) { + BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); + OUT_RELOCo(region->buffer, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); + BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); + OUT_RELOCl(region->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); + BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); + OUT_RING (region->pitch * region->cpp); + } + + if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | + NV40TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV40TCL_RT_ENABLE_MRT; + BEGIN_RING(curie, NV40TCL_RT_ENABLE, 1); + OUT_RING (rt_enable); + + if (0) { +#if 0 + rt_format |= (log2width << + NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT); + rt_format |= (log2height << + NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT); +#endif + rt_format |= (NV40TCL_RT_FORMAT_TYPE_SWIZZLED << + NV40TCL_RT_FORMAT_TYPE_SHIFT); + } else { + rt_format |= (NV40TCL_RT_FORMAT_TYPE_LINEAR << + NV40TCL_RT_FORMAT_TYPE_SHIFT); + } + + if (fb->cbufs[0]->format == PIPE_FORMAT_U_R5_G6_B5) { + rt_format |= (NV40TCL_RT_FORMAT_COLOR_R5G6B5 << + NV40TCL_RT_FORMAT_COLOR_SHIFT); + } else { + rt_format |= (NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 << + NV40TCL_RT_FORMAT_COLOR_SHIFT); + } + + if (fb->zbuf && fb->zbuf->format == PIPE_FORMAT_U_Z16) { + rt_format |= (NV40TCL_RT_FORMAT_DEPTH_Z16 << + NV40TCL_RT_FORMAT_DEPTH_SHIFT); + } else { + rt_format |= (NV40TCL_RT_FORMAT_DEPTH_Z24S8 << + NV40TCL_RT_FORMAT_DEPTH_SHIFT); + } + + BEGIN_RING(curie, NV40TCL_RT_HORIZ, 3); + OUT_RING ((fb->cbufs[0]->width << 16) | 0); + OUT_RING ((fb->cbufs[0]->height << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(curie, NV40TCL_VIEWPORT_HORIZ, 2); + OUT_RING ((fb->cbufs[0]->width << 16) | 0); + OUT_RING ((fb->cbufs[0]->height << 16) | 0); + BEGIN_RING(curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((fb->cbufs[0]->width - 1) << 16) | 0); + OUT_RING (((fb->cbufs[0]->height - 1) << 16) | 0); +} + +static void +nv40_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + OUT_RINGp ((uint32_t *)stipple->stipple, 32); +} + +static void +nv40_set_sampler_units(struct pipe_context *pipe, + uint num_samplers, const uint *units) +{ +} + +static void +nv40_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + BEGIN_RING(curie, NV40TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny); +} + +static void +nv40_set_texture_state(struct pipe_context *pipe, unsigned unit, + struct pipe_mipmap_tree *miptree) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + nv40->tex_miptree[unit] = miptree; + nv40->tex_dirty |= unit; + + nv40->dirty |= NV40_NEW_TEXTURE; +} + +static void +nv40_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + BEGIN_RING(curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + OUT_RINGf (vpt->translate[0]); + OUT_RINGf (vpt->translate[1]); + OUT_RINGf (vpt->translate[2]); + OUT_RINGf (vpt->translate[3]); + OUT_RINGf (vpt->scale[0]); + OUT_RINGf (vpt->scale[1]); + OUT_RINGf (vpt->scale[2]); + OUT_RINGf (vpt->scale[3]); +} + +static void +nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + nv40->vtxbuf[index] = *vb; + + nv40->dirty |= NV40_NEW_ARRAYS; +} + +static void +nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + nv40->vtxelt[index] = *ve; + + nv40->dirty |= NV40_NEW_ARRAYS; +} + +static void +nv40_set_feedback_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_feedback_buffer *fbb) +{ + NOUVEAU_ERR("\n"); +} + +void +nv40_init_state_functions(struct nv40_context *nv40) +{ + nv40->pipe.create_alpha_test_state = nv40_alpha_test_state_create; + nv40->pipe.bind_alpha_test_state = nv40_alpha_test_state_bind; + nv40->pipe.delete_alpha_test_state = nv40_alpha_test_state_delete; + + nv40->pipe.create_blend_state = nv40_blend_state_create; + nv40->pipe.bind_blend_state = nv40_blend_state_bind; + nv40->pipe.delete_blend_state = nv40_blend_state_delete; + + nv40->pipe.create_sampler_state = nv40_sampler_state_create; + nv40->pipe.bind_sampler_state = nv40_sampler_state_bind; + nv40->pipe.delete_sampler_state = nv40_sampler_state_delete; + + nv40->pipe.create_rasterizer_state = nv40_rasterizer_state_create; + nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind; + nv40->pipe.delete_rasterizer_state = nv40_rasterizer_state_delete; + + nv40->pipe.create_depth_stencil_state = nv40_depth_stencil_state_create; + nv40->pipe.bind_depth_stencil_state = nv40_depth_stencil_state_bind; + nv40->pipe.delete_depth_stencil_state = nv40_depth_stencil_state_delete; + + nv40->pipe.create_vs_state = nv40_vp_state_create; + nv40->pipe.bind_vs_state = nv40_vp_state_bind; + nv40->pipe.delete_vs_state = nv40_vp_state_delete; + + nv40->pipe.create_fs_state = nv40_fp_state_create; + nv40->pipe.bind_fs_state = nv40_fp_state_bind; + nv40->pipe.delete_fs_state = nv40_fp_state_delete; + + nv40->pipe.set_blend_color = nv40_set_blend_color; + nv40->pipe.set_clip_state = nv40_set_clip_state; + nv40->pipe.set_clear_color_state = nv40_set_clear_color_state; + nv40->pipe.set_constant_buffer = nv40_set_constant_buffer; +// nv40->pipe.set_feedback_state = nv40_set_feedback_state; + nv40->pipe.set_framebuffer_state = nv40_set_framebuffer_state; + nv40->pipe.set_polygon_stipple = nv40_set_polygon_stipple; + nv40->pipe.set_sampler_units = nv40_set_sampler_units; + nv40->pipe.set_scissor_state = nv40_set_scissor_state; + nv40->pipe.set_texture_state = nv40_set_texture_state; + nv40->pipe.set_viewport_state = nv40_set_viewport_state; + + nv40->pipe.set_vertex_buffer = nv40_set_vertex_buffer; + nv40->pipe.set_vertex_element = nv40_set_vertex_element; + +// nv40->pipe.set_feedback_buffer = nv40_set_feedback_buffer; +} + diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h new file mode 100644 index 00000000000..1535037f638 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -0,0 +1,173 @@ +#ifndef __NV40_STATE_H__ +#define __NV40_STATE_H__ + +#include "pipe/p_state.h" + +struct nv40_alpha_test_state { + uint32_t enabled; + uint32_t func; + uint32_t ref; +}; + +struct nv40_blend_state { + uint32_t b_enable; + uint32_t b_srcfunc; + uint32_t b_dstfunc; + uint32_t b_eqn; + + uint32_t l_enable; + uint32_t l_op; + + uint32_t c_mask; + + uint32_t d_enable; +}; + +struct nv40_sampler_state { + uint32_t wrap; + uint32_t filt; + uint32_t bcol; +}; + +struct nv40_rasterizer_state { + uint32_t shade_model; + + uint32_t line_width; + uint32_t line_smooth_en; + uint32_t line_stipple_en; + uint32_t line_stipple; + + uint32_t point_size; + + uint32_t poly_smooth_en; + uint32_t poly_stipple_en; + + uint32_t poly_mode_front; + uint32_t poly_mode_back; + + uint32_t front_face; + uint32_t cull_face; + uint32_t cull_face_en; + +}; + +struct nv40_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + boolean on_hw; + int start_ip; + + uint32_t *insn; + int insn_len; + + struct { + int pipe_id; + int hw_id; + float value[4]; + } consts[256]; + int num_consts; + + uint32_t ir; + uint32_t or; +}; + +struct nv40_fragment_program { + const struct pipe_shader_state *pipe; + + boolean translated; + boolean on_hw; + + uint32_t *insn; + int insn_len; + + struct { + int pipe_id; + int hw_id; + } consts[256]; + int num_consts; + + struct pipe_buffer_handle *buffer; + + boolean uses_kil; + boolean writes_depth; + int num_regs; +}; + +struct nv40_depth_push { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; +}; + +struct nv40_stencil_push { + uint32_t enable; + uint32_t wmask; + uint32_t func; + uint32_t ref; + uint32_t vmask; + uint32_t fail; + uint32_t zfail; + uint32_t zpass; +}; + +struct nv40_depth_stencil_state { + struct nv40_depth_push depth; + union { + struct nv40_stencil_push back; + struct nv40_stencil_push front; + } stencil; +}; + +static INLINE unsigned +nv40_tex_wrap_mode(unsigned wrap) { + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + return NV40TCL_TEX_WRAP_S_REPEAT; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + return NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + return NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER; + case PIPE_TEX_WRAP_CLAMP: + return NV40TCL_TEX_WRAP_S_CLAMP; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + return NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + return NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + return NV40TCL_TEX_WRAP_S_MIRROR_CLAMP; + default: + return NV40TCL_TEX_WRAP_S_REPEAT; + } +} + +static INLINE unsigned +nv40_tex_filter(unsigned f0, unsigned f1) { + switch (f0) { + case PIPE_TEX_FILTER_NEAREST: + switch (f1) { + case PIPE_TEX_MIPFILTER_NEAREST: + return NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST; + case PIPE_TEX_MIPFILTER_LINEAR: + return NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR; + case PIPE_TEX_MIPFILTER_NONE: + default: + return NV40TCL_TEX_FILTER_MIN_NEAREST; + } + case PIPE_TEX_FILTER_LINEAR: + default: + switch (f1) { + case PIPE_TEX_MIPFILTER_NEAREST: + return NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST; + case PIPE_TEX_MIPFILTER_LINEAR: + return NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR; + case PIPE_TEX_MIPFILTER_NONE: + default: + return NV40TCL_TEX_FILTER_MIN_LINEAR; + } + } +} + +#endif diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c new file mode 100644 index 00000000000..a29c70538f5 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -0,0 +1,112 @@ +#include "nv40_context.h" +#include "nv40_dma.h" +#include "nv40_state.h" + +static INLINE void +nv40_state_update_fragprog(struct nv40_context *nv40) +{ + struct pipe_context *pipe = (struct pipe_context *)nv40; + struct nv40_fragment_program *fp = nv40->fragprog.fp; + float *map; + int i; + + if (!fp->translated) + nv40_fragprog_translate(nv40, fp); + + if (fp->num_consts) { + map = pipe->winsys->buffer_map(pipe->winsys, + nv40->fragprog.constant_buf, + PIPE_BUFFER_FLAG_READ); + for (i = 0; i < fp->num_consts; i++) { + uint pid = fp->consts[i].pipe_id; + + if (pid == -1) + continue; + + if (!memcmp(&fp->insn[fp->consts[i].hw_id], &map[pid*4], + 4 * sizeof(float))) + continue; + + memcpy(&fp->insn[fp->consts[i].hw_id], &map[pid*4], + 4 * sizeof(float)); + fp->on_hw = 0; + } + pipe->winsys->buffer_unmap(pipe->winsys, + nv40->fragprog.constant_buf); + } +} + +static INLINE void +nv40_state_update_vertprog(struct nv40_context *nv40) +{ + struct pipe_context *pipe = (struct pipe_context *)nv40; + struct nv40_vertex_program *vp = nv40->vertprog.vp; + float *map; + int i; + + if (!nv40->vertprog.vp->translated) + nv40_vertprog_translate(nv40, nv40->vertprog.vp); + + if (vp->num_consts) { + map = pipe->winsys->buffer_map(pipe->winsys, + nv40->vertprog.constant_buf, + PIPE_BUFFER_FLAG_READ); + for (i = 0; i < vp->num_consts; i++) { + uint pid = vp->consts[i].pipe_id; + + if (pid >= 0) { + if (!memcmp(vp->consts[i].value, &map[pid*4], + 4 * sizeof(float))) + continue; + memcpy(vp->consts[i].value, &map[pid*4], + 4 * sizeof(float)); + } + + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (vp->consts[i].hw_id); + OUT_RINGp ((uint32_t *)vp->consts[i].value, 4); + } + pipe->winsys->buffer_unmap(pipe->winsys, + nv40->vertprog.constant_buf); + } +} + +void +nv40_emit_hw_state(struct nv40_context *nv40) +{ + if (nv40->dirty & NV40_NEW_FRAGPROG) { + struct nv40_fragment_program *cur = nv40->fragprog.fp; + + nv40_state_update_fragprog(nv40); + + if (cur->on_hw) + nv40->dirty &= ~NV40_NEW_FRAGPROG; + + if (!cur->on_hw || cur != nv40->fragprog.active_fp) + nv40_fragprog_bind(nv40, cur); + } + + if (nv40->dirty & NV40_NEW_TEXTURE) + nv40_state_tex_update(nv40); + + if (nv40->dirty & (NV40_NEW_TEXTURE | NV40_NEW_FRAGPROG)) { + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (1); + nv40->dirty &= ~(NV40_NEW_TEXTURE | NV40_NEW_FRAGPROG); + } + + if (nv40->dirty & NV40_NEW_VERTPROG) { + nv40_state_update_vertprog(nv40); + if (nv40->vertprog.vp != nv40->vertprog.active_vp) + nv40_vertprog_bind(nv40, nv40->vertprog.vp); + nv40->dirty &= ~NV40_NEW_VERTPROG; + } + + if (nv40->dirty & NV40_NEW_ARRAYS) { + nv40_vbo_arrays_update(nv40); + nv40->dirty &= ~NV40_NEW_ARRAYS; + } +} + diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c new file mode 100644 index 00000000000..a92d6250a20 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -0,0 +1,140 @@ +#include "nv40_context.h" +#include "nv40_dma.h" + +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV40TCL_TEX_FORMAT_FORMAT_##tf, \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x << NV40TCL_TEX_SWIZZLE_S0_X_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0y << NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0z << NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0w << NV40TCL_TEX_SWIZZLE_S0_W_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S1_X_##ts1x << NV40TCL_TEX_SWIZZLE_S1_X_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S1_X_##ts1y << NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S1_X_##ts1z << NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT) | \ + (NV40TCL_TEX_SWIZZLE_S1_X_##ts1w << NV40TCL_TEX_SWIZZLE_S1_W_SHIFT), \ +} + +struct nv40_texture_format { + boolean defined; + uint pipe; + int format; + int swizzle; +}; + +static struct nv40_texture_format +nv40_texture_formats[] = { + _(U_A8_R8_G8_B8, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), + _(U_A1_R5_G5_B5, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), + _(U_A4_R4_G4_B4, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), + _(U_R5_G6_B5 , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), + _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, Z, W, X, Y), +// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +}; + +static struct nv40_texture_format * +nv40_tex_format(uint pipe_format) +{ + struct nv40_texture_format *tf = nv40_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + +static INLINE int +nv40_tex_dims(uint pipe_target) +{ + switch (pipe_target) { + case PIPE_TEXTURE_1D: return 1; + case PIPE_TEXTURE_2D: return 2; + case PIPE_TEXTURE_3D: return 3; + case PIPE_TEXTURE_CUBE: return 2; + default: + NOUVEAU_ERR("AII unknown pipe target: %d\n", pipe_target); + return 2; + } +} + +static void +nv40_tex_unit_enable(struct nv40_context *nv40, int unit) +{ + struct nouveau_winsys *nvws = nv40->nvws; + struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; + struct pipe_mipmap_tree *mt = nv40->tex_miptree[unit]; + struct nv40_texture_format *tf; + uint32_t txf, txs, txp; + int swizzled = 0; /*XXX: implement in region code? */ + + tf = nv40_tex_format(mt->format); + if (!tf->defined) { + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", mt->format); + return; + } + + txf = (tf->format | 0x80) << NV40TCL_TEX_FORMAT_FORMAT_SHIFT; + txf |= ((mt->last_level - mt->first_level + 1) << + NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); + if (1) /* XXX */ + txf |= NV40TCL_TEX_FORMAT_NO_BORDER; + + txf |= (nv40_tex_dims(mt->target) << NV40TCL_TEX_FORMAT_DIMS_SHIFT); + if (0) /*XXX*/ + txf |= NV40TCL_TEX_FORMAT_RECT; + + if (swizzled) { + txp = 0; + } else { + txp = mt->pitch * mt->cpp; + txf |= NV40TCL_TEX_FORMAT_LINEAR; + } + + txs = tf->swizzle; + if (mt->format == PIPE_FORMAT_U_A8_L8) + txs |= (1<<16); /*nfi*/ + + BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); + OUT_RELOCl(mt->region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD); + OUT_RELOCd(mt->region->buffer, txf, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_OR | NOUVEAU_BO_RD, NV40TCL_TEX_FORMAT_DMA0, + NV40TCL_TEX_FORMAT_DMA1); + OUT_RING (ps->wrap); + OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | + (0x00078000) /* mipmap related? */); + OUT_RING (txs); + OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); + OUT_RING ((mt->width0 << NV40TCL_TEX_SIZE0_W_SHIFT) | mt->height0); + OUT_RING (ps->bcol); + BEGIN_RING(curie, NV40TCL_TEX_SIZE1(unit), 1); + OUT_RING ((mt->depth0 << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); +} + +void +nv40_state_tex_update(struct nv40_context *nv40) +{ + while (nv40->tex_dirty) { + int unit = ffs(nv40->tex_dirty) - 1; + + if (nv40->tex_miptree[unit]) { + nv40_tex_unit_enable(nv40, unit); + } else { + BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); + OUT_RING (0); + } + + nv40->tex_dirty &= ~(1 << unit); + } +} + diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c new file mode 100644 index 00000000000..84e0d792681 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -0,0 +1,229 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv40_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" + + +#define CLIP_TILE \ + do { \ + if (x >= ps->width) \ + return; \ + if (y >= ps->height) \ + return; \ + if (x + w > ps->width) \ + w = ps->width - x; \ + if (y + h > ps->height) \ + h = ps->height -y; \ + } while(0) + + +/** + * Note: this is exactly like a8r8g8b8_get_tile() in sp_surface.c + * Share it someday. + */ +static void +nv40_get_tile_rgba(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, float *p) +{ + const unsigned *src + = ((const unsigned *) (ps->region->map + ps->offset)) + + y * ps->region->pitch + x; + unsigned i, j; + unsigned w0 = w; + + CLIP_TILE; + + switch (ps->format) { + case PIPE_FORMAT_U_A8_R8_G8_B8: + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++) { + const unsigned pixel = src[j]; + pRow[0] = UBYTE_TO_FLOAT((pixel >> 16) & 0xff); + pRow[1] = UBYTE_TO_FLOAT((pixel >> 8) & 0xff); + pRow[2] = UBYTE_TO_FLOAT((pixel >> 0) & 0xff); + pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); + pRow += 4; + } + src += ps->region->pitch; + p += w0 * 4; + } + break; + case PIPE_FORMAT_S8_Z24: + { + const float scale = 1.0 / (float) 0xffffff; + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++) { + const unsigned pixel = src[j]; + pRow[0] = + pRow[1] = + pRow[2] = + pRow[3] = (pixel & 0xffffff) * scale; + pRow += 4; + } + src += ps->region->pitch; + p += w0 * 4; + } + } + break; + default: + assert(0); + } +} + + +static void +nv40_put_tile_rgba(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, const float *p) +{ + /* TODO */ + assert(0); +} + + +/* + * XXX note: same as code in sp_surface.c + */ +static void +nv40_get_tile(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, + void *p, int dst_stride) +{ + const uint cpp = ps->region->cpp; + const uint w0 = w; + const ubyte *pSrc; + ubyte *pDest; + uint i; + + assert(ps->region->map); + + CLIP_TILE; + + if (dst_stride == 0) { + dst_stride = w0 * cpp; + } + + pSrc = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pDest = (ubyte *) p; + + for (i = 0; i < h; i++) { + memcpy(pDest, pSrc, w0 * cpp); + pDest += dst_stride; + pSrc += ps->region->pitch * cpp; + } +} + + +/* + * XXX note: same as code in sp_surface.c + */ +static void +nv40_put_tile(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, + const void *p, int src_stride) +{ + const uint cpp = ps->region->cpp; + const uint w0 = w; + const ubyte *pSrc; + ubyte *pDest; + uint i; + + assert(ps->region->map); + + CLIP_TILE; + + if (src_stride == 0) { + src_stride = w0 * cpp; + } + + pSrc = (const ubyte *) p; + pDest = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + + for (i = 0; i < h; i++) { + memcpy(pDest, pSrc, w0 * cpp); + pDest += ps->region->pitch * cpp; + pSrc += src_stride; + } +} + + +/* + * XXX note: same as code in sp_surface.c + */ +static struct pipe_surface * +nv40_get_tex_surface(struct pipe_context *pipe, + struct pipe_mipmap_tree *mt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_surface *ps; + unsigned offset; /* in bytes */ + + offset = mt->level[level].level_offset; + + if (mt->target == PIPE_TEXTURE_CUBE) { + offset += mt->level[level].image_offset[face] * mt->cpp; + } + else if (mt->target == PIPE_TEXTURE_3D) { + offset += mt->level[level].image_offset[zslice] * mt->cpp; + } + else { + assert(face == 0); + assert(zslice == 0); + } + + ps = pipe->winsys->surface_alloc(pipe->winsys, mt->format); + if (ps) { + assert(ps->format); + assert(ps->refcount); + pipe_region_reference(&ps->region, mt->region); + ps->width = mt->level[level].width; + ps->height = mt->level[level].height; + ps->offset = offset; + } + return ps; +} + + +void +nv40_init_surface_functions(struct nv40_context *nv40) +{ + nv40->pipe.get_tex_surface = nv40_get_tex_surface; + nv40->pipe.get_tile = nv40_get_tile; + nv40->pipe.put_tile = nv40_put_tile; + nv40->pipe.get_tile_rgba = nv40_get_tile_rgba; + nv40->pipe.put_tile_rgba = nv40_put_tile_rgba; +} diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c new file mode 100644 index 00000000000..aa930476b6f --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -0,0 +1,222 @@ +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" +#include "nv40_dma.h" +#include "nv40_state.h" +#include "nvgl_pipe.h" + +boolean +nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, + unsigned count) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + unsigned nr; + + if (nv40->dirty) + nv40_emit_hw_state(nv40); + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(curie, NV40TCL_VB_VERTEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + /*XXX: large arrays (nr>2047) will blow up */ + nr = count >> 8; + if (nr) { + assert (nr <= 2047); + + BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, nr); + while (nr--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + pipe->flush(pipe, PIPE_FLUSH_WAIT); + return TRUE; +} + +static INLINE void +nv40_draw_elements_u08(struct nv40_context *nv40, void *ib, + unsigned start, unsigned count) +{ + uint8_t *elts = (uint8_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2046); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv40_draw_elements_u16(struct nv40_context *nv40, void *ib, + unsigned start, unsigned count) +{ + uint16_t *elts = (uint16_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2046); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, + unsigned start, unsigned count) +{ + uint32_t *elts = (uint32_t *)ib + start; + int push; + + while (count) { + push = MIN2(count, 2047); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U32, push); + OUT_RINGp (elts, push); + + count -= push; + elts += push; + } +} + +boolean +nv40_draw_elements(struct pipe_context *pipe, + struct pipe_buffer_handle *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + void *ib; + + if (nv40->dirty) + nv40_emit_hw_state(nv40); + + ib = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, + PIPE_BUFFER_FLAG_READ); + if (!ib) { + NOUVEAU_ERR("Couldn't map index buffer!!\n"); + return FALSE; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + switch (indexSize) { + case 1: + nv40_draw_elements_u08(nv40, ib, start, count); + break; + case 2: + nv40_draw_elements_u16(nv40, ib, start, count); + break; + case 4: + nv40_draw_elements_u32(nv40, ib, start, count); + break; + default: + NOUVEAU_ERR("unsupported elt size %d\n", indexSize); + break; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + pipe->winsys->buffer_unmap(pipe->winsys, ib); + pipe->flush(pipe, PIPE_FLUSH_WAIT); + return TRUE; +} + +static INLINE int +nv40_vbo_format_to_ncomp(uint format) +{ + switch (format) { + case PIPE_FORMAT_R32G32B32A32_FLOAT: return 4; + case PIPE_FORMAT_R32G32B32_FLOAT: return 3; + case PIPE_FORMAT_R32G32_FLOAT: return 2; + case PIPE_FORMAT_R32_FLOAT: return 1; + default: + NOUVEAU_ERR("AII, unknown vbo format %d\n", format); + return 1; + } +} + +void +nv40_vbo_arrays_update(struct nv40_context *nv40) +{ + struct nouveau_winsys *nvws = nv40->nvws; + struct nv40_vertex_program *vp = nv40->vertprog.vp; + uint32_t inputs, vtxfmt[16]; + int hw, num_hw; + + inputs = vp->ir; + for (hw = 0; hw < 16 && inputs; hw++) { + if (inputs & (1 << hw)) { + num_hw = hw; + inputs &= ~(1 << hw); + } + } + num_hw++; + + inputs = vp->ir; + BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + for (hw = 0; hw < num_hw; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + + if (!(inputs & (1 << hw))) { + OUT_RING(0); + vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; + continue; + } + + ve = &nv40->vtxelt[hw]; + vb = &nv40->vtxbuf[ve->vertex_buffer_index]; + + OUT_RELOC(vb->buffer, vb->buffer_offset + ve->src_offset, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, + NV40TCL_VTXBUF_ADDRESS_DMA1); + vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | + (nv40_vbo_format_to_ncomp(ve->src_format) << + NV40TCL_VTXFMT_SIZE_SHIFT) | + NV40TCL_VTXFMT_TYPE_FLOAT); + } + + BEGIN_RING(curie, 0x1710, 1); + OUT_RING (0); /* vtx cache flush */ + BEGIN_RING(curie, NV40TCL_VTXFMT(0), num_hw); + OUT_RINGp (vtxfmt, num_hw); +} + diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c new file mode 100644 index 00000000000..be550e47430 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -0,0 +1,594 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/tgsi/exec/tgsi_token.h" +#include "pipe/tgsi/exec/tgsi_parse.h" + +#include "nv40_context.h" +#include "nv40_dma.h" +#include "nv40_state.h" + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 8 +#define MASK_Y 4 +#define MASK_Z 2 +#define MASK_W 1 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE 0 +#define DEF_CTEST 0 +#include "nv40_shader.h" + +#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv40_sr_neg((s)) +#define abs(s) nv40_sr_abs((s)) + +static uint32_t +passthrough_vp_data[] = { + 0x40041c6c, 0x0040010d, 0x8106c083, 0x6041ff84, + 0x40041c6c, 0x0040000d, 0x8106c083, 0x6041ff81, +}; + +static struct nv40_vertex_program +passthrough_vp = { + .pipe = NULL, + .translated = TRUE, + + .insn = passthrough_vp_data, + .insn_len = sizeof(passthrough_vp_data) / sizeof(uint32_t), + + .ir = 0x00000003, + .or = 0x00000001, +}; + +struct nv40_vpc { + struct nv40_vertex_program *vp; + + uint output_map[PIPE_MAX_SHADER_OUTPUTS]; + + int high_temp; + int temp_temp_count; +}; + +static INLINE struct nv40_sreg +nv40_sr_temp(struct nv40_vpc *vpc) +{ + int idx; + + idx = vpc->temp_temp_count++; + idx += vpc->high_temp; + return nv40_sr(0, NV40_VP_SRC_REG_TYPE_TEMP, idx); +} + +static INLINE struct nv40_sreg +nv40_sr_const(struct nv40_vpc *vpc, int pipe, + float x, float y, float z, float w) +{ + struct nv40_vertex_program *vp = vpc->vp; + int idx = vp->num_consts; + + vp->consts[idx].pipe_id = pipe; + vp->consts[idx].hw_id = idx; + vp->consts[idx].value[0] = x; + vp->consts[idx].value[1] = y; + vp->consts[idx].value[2] = z; + vp->consts[idx].value[3] = w; + vp->num_consts++; + + return nv40_sr(0, NV40_VP_SRC_REG_TYPE_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv40_vp_arith((cc), (s), NV40_VP_INST_##o, (d), (m), (s0), (s1), (s2)) +#define temp(vpc) nv40_sr_temp((vpc)) +#define constant(v,p,x,y,z,w) nv40_sr_const((v), (p), (x), (y), (z), (w)) + +static void +emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src) +{ + struct nv40_vertex_program *vp = vpc->vp; + uint32_t sr = 0; + + sr |= (src.type << NV40_VP_SRC_REG_TYPE_SHIFT); + if (src.type == NV40_VP_SRC_REG_TYPE_INPUT) { + vp->ir |= (1 << src.index); + hw[1] |= (src.index << NV40_VP_INST_INPUT_SRC_SHIFT); + } else + if (src.type == NV40_VP_SRC_REG_TYPE_CONST) { + hw[1] |= (src.index << NV40_VP_INST_CONST_SRC_SHIFT); + } else { + sr |= (src.index << NV40_VP_SRC_TEMP_SRC_SHIFT); + } + + if (src.negate) + sr |= NV40_VP_SRC_NEGATE; + + if (src.abs) + hw[0] |= (1 << (21 + pos)); + + sr |= ((src.swz[0] << NV40_VP_SRC_SWZ_X_SHIFT) | + (src.swz[1] << NV40_VP_SRC_SWZ_Y_SHIFT) | + (src.swz[2] << NV40_VP_SRC_SWZ_Z_SHIFT) | + (src.swz[3] << NV40_VP_SRC_SWZ_W_SHIFT)); + + switch (pos) { + case 0: + hw[1] |= ((sr & NV40_VP_SRC0_HIGH_MASK) >> + NV40_VP_SRC0_HIGH_SHIFT) << NV40_VP_INST_SRC0H_SHIFT; + hw[2] |= (sr & NV40_VP_SRC0_LOW_MASK) << + NV40_VP_INST_SRC0L_SHIFT; + break; + case 1: + hw[2] |= sr << NV40_VP_INST_SRC1_SHIFT; + break; + case 2: + hw[2] |= ((sr & NV40_VP_SRC2_HIGH_MASK) >> + NV40_VP_SRC2_HIGH_SHIFT) << NV40_VP_INST_SRC2H_SHIFT; + hw[3] |= (sr & NV40_VP_SRC2_LOW_MASK) << + NV40_VP_INST_SRC2L_SHIFT; + break; + default: + assert(0); + } +} + +static void +emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) +{ + struct nv40_vertex_program *vp = vpc->vp; + + if (dst.output == 0) { + hw[3] |= NV40_VP_INST_DEST_MASK; + if (slot == 0) { + hw[0] |= (dst.index << + NV40_VP_INST_VEC_DEST_TEMP_SHIFT); + } else { + hw[3] |= (dst.index << + NV40_VP_INST_SCA_DEST_TEMP_SHIFT); + } + } else { + switch (dst.index) { + case NV40_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; + case NV40_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; + case NV40_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; + case NV40_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; + case NV40_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; + case NV40_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; + case NV40_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; + case NV40_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; + case NV40_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; + case NV40_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; + case NV40_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; + case NV40_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; + case NV40_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; + case NV40_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + default: + break; + } + + hw[3] |= (dst.index << NV40_VP_INST_DEST_SHIFT); + if (slot == 0) { + hw[0] |= NV40_VP_INST_VEC_RESULT; + hw[0] |= NV40_VP_INST_VEC_DEST_TEMP_MASK; + } else { + hw[3] |= NV40_VP_INST_SCA_RESULT; + hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; + } + } +} + +static void +nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op, + struct nv40_sreg dst, int mask, + struct nv40_sreg s0, struct nv40_sreg s1, + struct nv40_sreg s2) +{ + struct nv40_vertex_program *vp = vpc->vp; + uint32_t *hw = &vp->insn[vp->insn_len]; + + hw[0] |= (NV40_VP_INST_COND_TR << NV40_VP_INST_COND_SHIFT); + hw[0] |= ((0 << NV40_VP_INST_COND_SWZ_X_SHIFT) | + (1 << NV40_VP_INST_COND_SWZ_Y_SHIFT) | + (2 << NV40_VP_INST_COND_SWZ_Z_SHIFT) | + (3 << NV40_VP_INST_COND_SWZ_W_SHIFT)); + + if (slot == 0) { + hw[1] |= (op << NV40_VP_INST_VEC_OPCODE_SHIFT); + hw[3] |= (NV40_VP_INST_SCA_RESULT | + NV40_VP_INST_SCA_DEST_TEMP_MASK); + hw[3] |= (mask << NV40_VP_INST_VEC_WRITEMASK_SHIFT); + } else { + hw[1] |= (op << NV40_VP_INST_SCA_OPCODE_SHIFT); + hw[1] |= (NV40_VP_INST_VEC_DEST_TEMP_MASK | (1 << 20)); + hw[3] |= (mask << NV40_VP_INST_SCA_WRITEMASK_SHIFT); + } + + emit_dst(vpc, hw, slot, dst); + emit_src(vpc, hw, 0, s0); + emit_src(vpc, hw, 1, s1); + emit_src(vpc, hw, 2, s2); + + vp->insn_len += 4; +} + +static INLINE struct nv40_sreg +tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv40_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv40_sr(0, NV40_VP_SRC_REG_TYPE_INPUT, + fsrc->SrcRegister.Index); + break; + case TGSI_FILE_CONSTANT: + src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + break; + case TGSI_FILE_TEMPORARY: + if (vpc->high_temp < fsrc->SrcRegister.Index) + vpc->high_temp = fsrc->SrcRegister.Index; + src = nv40_sr(0, NV40_VP_SRC_REG_TYPE_TEMP, + fsrc->SrcRegister.Index); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv40_sreg +tgsi_dst(struct nv40_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + uint out, idx; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + out = 1; + idx = vpc->output_map[fdst->DstRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + out = 0; + idx = fdst->DstRegister.Index; + if (vpc->high_temp < idx) + vpc->high_temp = idx; + break; + default: + NOUVEAU_ERR("bad dst file\n"); + break; + } + + return nv40_sr(out, NV40_VP_SRC_REG_TYPE_TEMP, idx); +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, + const struct tgsi_full_instruction *finst) +{ + struct nv40_sreg src[3], dst, tmp; + struct nv40_sreg none = nv40_sr(0, NV40_VP_SRC_REG_TYPE_INPUT, 0); + int mask; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_RET) + return TRUE; + + vpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(vpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + NOUVEAU_MSG("extra src attr %d\n", + fsrc->SrcRegister.Index); + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_CONSTANT: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); + break; + case TGSI_OPCODE_ARL: + arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DST: + arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_EXP: + arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_FLR: + arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_LG2: + arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LIT: + arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LOG: + arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_MAD: + arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(vpc); + arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, + swz(src[0], X, X, X, X)); + arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(vpc, 1, OP_EX2, dst, mask, none, none, + swz(tmp, X, X, X, X)); + break; + case TGSI_OPCODE_RCP: + arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_RSQ: + arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_SGE: + arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); + break; + case TGSI_OPCODE_XPD: + tmp = temp(vpc); + arith(vpc, 0, OP_MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV40_VP_INST_DEST_POS; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV40_VP_INST_DEST_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV40_VP_INST_DEST_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_BCOLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV40_VP_INST_DEST_BFC0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV40_VP_INST_DEST_BFC1; + } else { + NOUVEAU_ERR("bad bcolour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV40_VP_INST_DEST_FOGC; + break; + case TGSI_SEMANTIC_PSIZE: + hw = NV40_VP_INST_DEST_PSZ; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV40_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + vpc->output_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +void +nv40_vertprog_translate(struct nv40_context *nv40, + struct nv40_vertex_program *vp) +{ + struct tgsi_parse_context parse; + struct nv40_vpc *vpc = NULL; + + vpc = calloc(1, sizeof(struct nv40_vpc)); + if (!vpc) + return; + vp->insn = calloc(1, 128*4*sizeof(uint32_t)); + vpc->vp = vp; + vpc->high_temp = -1; + + tgsi_parse_init(&parse, vp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_OUTPUT: + if (!nv40_vertprog_parse_decl_output(vpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + + finst = &parse.FullToken.FullInstruction; + if (!nv40_vertprog_parse_instruction(vpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + vp->insn[vp->insn_len - 1] |= NV40_VP_INST_LAST; +#if 0 + { + int i; + for (i = 0; i < vp->insn_len; i++) + NOUVEAU_ERR("inst[%d] = 0x%08x\n", i, vp->insn[i]); + } +#endif + vp->translated = TRUE; +out_err: + tgsi_parse_free(&parse); + free(vpc); +} + +void +nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) +{ + int i; + + if (!vp->translated) { + NOUVEAU_ERR("vtxprog invalid, using passthrough shader\n"); + vp = &passthrough_vp; + } + + if (!vp->on_hw) { + if (nv40->vertprog.active_vp) + nv40->vertprog.active_vp->on_hw = FALSE; + vp->on_hw = TRUE; + vp->start_ip = 0; + + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (vp->start_ip); + for (i = 0; i < vp->insn_len; i += 4) { + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (&vp->insn[i], 4); + } + } + + BEGIN_RING(curie, NV40TCL_VP_START_FROM_ID, 1); + OUT_RING (vp->start_ip); + BEGIN_RING(curie, NV40TCL_VP_ATTRIB_EN, 2); + OUT_RING (vp->ir); + OUT_RING (vp->or); + + nv40->vertprog.active_vp = vp; +} diff --git a/src/mesa/pipe/nv40/nvgl_pipe.h b/src/mesa/pipe/nv40/nvgl_pipe.h new file mode 100644 index 00000000000..15ff318023d --- /dev/null +++ b/src/mesa/pipe/nv40/nvgl_pipe.h @@ -0,0 +1,198 @@ +#ifndef __NVGL_PIPE_H__ +#define __NVGL_PIPE_H__ + +#include + +static INLINE unsigned +nvgl_blend_func(unsigned factor) +{ + switch (factor) { + case PIPE_BLENDFACTOR_ONE: + return GL_ONE; + case PIPE_BLENDFACTOR_SRC_COLOR: + return GL_SRC_COLOR; + case PIPE_BLENDFACTOR_SRC_ALPHA: + return GL_SRC_ALPHA; + case PIPE_BLENDFACTOR_DST_ALPHA: + return GL_DST_ALPHA; + case PIPE_BLENDFACTOR_DST_COLOR: + return GL_DST_COLOR; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + return GL_SRC_ALPHA_SATURATE; + case PIPE_BLENDFACTOR_CONST_COLOR: + return GL_CONSTANT_COLOR; + case PIPE_BLENDFACTOR_CONST_ALPHA: + return GL_CONSTANT_ALPHA; + case PIPE_BLENDFACTOR_ZERO: + return GL_ZERO; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + return GL_ONE_MINUS_SRC_COLOR; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + return GL_ONE_MINUS_SRC_ALPHA; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + return GL_ONE_MINUS_DST_ALPHA; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + return GL_ONE_MINUS_DST_COLOR; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + return GL_ONE_MINUS_CONSTANT_COLOR; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + return GL_ONE_MINUS_CONSTANT_ALPHA; + default: + return GL_ONE; + } +} + +static INLINE unsigned +nvgl_blend_eqn(unsigned func) +{ + switch (func) { + case PIPE_BLEND_ADD: + return GL_FUNC_ADD; + case PIPE_BLEND_SUBTRACT: + return GL_FUNC_SUBTRACT; + case PIPE_BLEND_REVERSE_SUBTRACT: + return GL_FUNC_REVERSE_SUBTRACT; + case PIPE_BLEND_MIN: + return GL_MIN; + case PIPE_BLEND_MAX: + return GL_MAX; + default: + return GL_FUNC_ADD; + } +} + +static INLINE unsigned +nvgl_logicop_func(unsigned func) +{ + switch (func) { + case PIPE_LOGICOP_CLEAR: + return GL_CLEAR; + case PIPE_LOGICOP_NOR: + return GL_NOR; + case PIPE_LOGICOP_AND_INVERTED: + return GL_AND_INVERTED; + case PIPE_LOGICOP_COPY_INVERTED: + return GL_COPY_INVERTED; + case PIPE_LOGICOP_AND_REVERSE: + return GL_AND_REVERSE; + case PIPE_LOGICOP_INVERT: + return GL_INVERT; + case PIPE_LOGICOP_XOR: + return GL_XOR; + case PIPE_LOGICOP_NAND: + return GL_NAND; + case PIPE_LOGICOP_AND: + return GL_AND; + case PIPE_LOGICOP_EQUIV: + return GL_EQUIV; + case PIPE_LOGICOP_NOOP: + return GL_NOOP; + case PIPE_LOGICOP_OR_INVERTED: + return GL_OR_INVERTED; + case PIPE_LOGICOP_COPY: + return GL_COPY; + case PIPE_LOGICOP_OR_REVERSE: + return GL_OR_REVERSE; + case PIPE_LOGICOP_OR: + return GL_OR; + case PIPE_LOGICOP_SET: + return GL_SET; + default: + return GL_CLEAR; + } +} + +static INLINE unsigned +nvgl_comparison_op(unsigned op) +{ + switch (op) { + case PIPE_FUNC_NEVER: + return GL_NEVER; + case PIPE_FUNC_LESS: + return GL_LESS; + case PIPE_FUNC_EQUAL: + return GL_EQUAL; + case PIPE_FUNC_LEQUAL: + return GL_LEQUAL; + case PIPE_FUNC_GREATER: + return GL_GREATER; + case PIPE_FUNC_NOTEQUAL: + return GL_NOTEQUAL; + case PIPE_FUNC_GEQUAL: + return GL_GEQUAL; + case PIPE_FUNC_ALWAYS: + return GL_ALWAYS; + default: + return GL_NEVER; + } +} + +static INLINE unsigned +nvgl_polygon_mode(unsigned mode) +{ + switch (mode) { + case PIPE_POLYGON_MODE_FILL: + return GL_FILL; + case PIPE_POLYGON_MODE_LINE: + return GL_LINE; + case PIPE_POLYGON_MODE_POINT: + return GL_POINT; + default: + return GL_FILL; + } +} + +static INLINE unsigned +nvgl_stencil_op(unsigned op) +{ + switch (op) { + case PIPE_STENCIL_OP_KEEP: + return GL_KEEP; + case PIPE_STENCIL_OP_ZERO: + return GL_ZERO; + case PIPE_STENCIL_OP_REPLACE: + return GL_REPLACE; + case PIPE_STENCIL_OP_INCR: + return GL_INCR; + case PIPE_STENCIL_OP_DECR: + return GL_DECR; + case PIPE_STENCIL_OP_INCR_WRAP: + return GL_INCR_WRAP; + case PIPE_STENCIL_OP_DECR_WRAP: + return GL_DECR_WRAP; + case PIPE_STENCIL_OP_INVERT: + return GL_INVERT; + default: + return GL_KEEP; + } +} + +static INLINE unsigned +nvgl_primitive(unsigned prim) { + switch (prim) { + case PIPE_PRIM_POINTS: + return GL_POINTS + 1; + case PIPE_PRIM_LINES: + return GL_LINES + 1; + case PIPE_PRIM_LINE_LOOP: + return GL_LINE_LOOP + 1; + case PIPE_PRIM_LINE_STRIP: + return GL_LINE_STRIP + 1; + case PIPE_PRIM_TRIANGLES: + return GL_TRIANGLES + 1; + case PIPE_PRIM_TRIANGLE_STRIP: + return GL_TRIANGLE_STRIP + 1; + case PIPE_PRIM_TRIANGLE_FAN: + return GL_TRIANGLE_FAN + 1; + case PIPE_PRIM_QUADS: + return GL_QUADS + 1; + case PIPE_PRIM_QUAD_STRIP: + return GL_QUAD_STRIP + 1; + case PIPE_PRIM_POLYGON: + return GL_POLYGON + 1; + default: + return GL_POINTS + 1; + } +} + +#endif -- cgit v1.2.3 From bc449c28c6455276fb7b833491468b144efee236 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 18 Nov 2007 17:49:30 +1100 Subject: nouveau: oops, when'd that disappear.. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index 48f583cbfb4..16c4324164b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -155,8 +155,8 @@ nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML }; - u_int8_t depth_bits_array[4] = { 0, 16, 24 }; - u_int8_t stencil_bits_array[4] = { 0, 0, 8 }; + u_int8_t depth_bits_array[4] = { 0, 16, 24, 24 }; + u_int8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; depth_buffer_factor = 4; back_buffer_factor = (have_back_buffer) ? 3 : 1; -- cgit v1.2.3 From 3ab26c864cb8401e919de01772c419b0299811fb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 18 Nov 2007 17:59:45 +1100 Subject: nv40: some very rough guesses for get_paramf() --- src/mesa/pipe/nv40/nv40_context.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index ff66095c5fc..ef5f21cb9c3 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -79,10 +79,14 @@ nv40_get_paramf(struct pipe_context *pipe, int param) switch (param) { case PIPE_CAP_MAX_LINE_WIDTH: case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; case PIPE_CAP_MAX_POINT_WIDTH: case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; -- cgit v1.2.3 From a1d622190f2e7dba578d693845277d6f32504b09 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 18 Nov 2007 21:47:18 +1100 Subject: nv40: support TXP again --- src/mesa/pipe/nv40/nv40_fragprog.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 48b783eebef..3be13143476 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -446,16 +446,15 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); break; case TGSI_OPCODE_TEX: - tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == + TGSI_EXTSWIZZLE_W) { + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + } else + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); break; case TGSI_OPCODE_TXB: tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); break; -#if 0 /* XXX: reimplement on top of TEX */ - case TGSI_OPCODE_TXP: - tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); - break; -#endif case TGSI_OPCODE_XPD: tmp = temp(fpc); arith(fpc, 0, MUL, tmp, mask, -- cgit v1.2.3 From f940603037844b91ffed84c390bef0ee57ffe8cc Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 18 Nov 2007 22:12:34 +1100 Subject: nv40: fix thinko == fix 3d textures --- src/mesa/pipe/nv40/nv40_miptree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 6b85823d8cf..2d8ef8ced77 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -36,7 +36,7 @@ nv40_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) mt->level[l].nr_images = 6; else if (mt->target == PIPE_TEXTURE_3D) - mt->level[l].nr_images = 3; + mt->level[l].nr_images = mt->level[l].depth; else mt->level[l].nr_images = 1; mt->level[l].image_offset = -- cgit v1.2.3 From c7c6253169798658547ec6fc6eb9cdefc68b58d3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 18 Nov 2007 23:08:33 +1100 Subject: nouveau: m2mf fallback path for region copies. --- .../drivers/dri/nouveau_winsys/nouveau_context.h | 1 + src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 55 ++++++++++++++++++++++ 2 files changed, 56 insertions(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index cd04b08eca1..f2557af935d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -52,6 +52,7 @@ struct nouveau_context { struct nouveau_grobj *NvCtxSurf2D; struct nouveau_grobj *NvImageBlit; struct nouveau_grobj *NvGdiRect; + struct nouveau_grobj *NvM2MF; uint32_t next_handle; uint32_t next_sequence; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c index cbe70983f5c..f41e45c73a6 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c @@ -33,6 +33,40 @@ nv04_region_display(void) return 0; } +static int +nv04_region_copy_m2mf(struct nouveau_context *nv, struct pipe_region *dst, + unsigned dst_offset, struct pipe_region *src, + unsigned src_offset, unsigned line_len, unsigned height) +{ + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_WR); + + while (height) { + int count = (height > 2047) ? 2047 : height; + + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_WR); + OUT_RING (src->pitch * src->cpp); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (line_len); + OUT_RING (count); + OUT_RING (0x0101); + OUT_RING (0); + + height -= count; + src_offset += src->pitch * count; + dst_offset += dst->pitch * count; + } + + return 0; +} + static int nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, unsigned dst_offset, unsigned dx, unsigned dy, @@ -44,6 +78,19 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, if (src->cpp != dst->cpp) return 1; + NOUVEAU_ERR("preg\n"); + + /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback + * to NV_MEMORY_TO_MEMORY_FORMAT in this case. + */ + if ((src_offset & 63) || (dst_offset & 63)) { + dst_offset += (dy * dst->pitch + dx) * dst->cpp; + src_offset += (sy * src->pitch + sx) * src->cpp; + return nv04_region_copy_m2mf(nv, dst, dst_offset, src, + src_offset, w * src->cpp, h); + + } + if ((format = nv04_surface_format(dst->cpp)) < 0) { NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); return 1; @@ -119,6 +166,14 @@ nouveau_region_init_nv04(struct nouveau_context *nv) { int ret; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, + &nv->NvM2MF))) { + NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); + return 1; + } + BEGIN_RING(NvM2MF, 0x0180, 1); + OUT_RING (nv->sync_notifier->handle); + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x62, &nv->NvCtxSurf2D))) { NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); -- cgit v1.2.3 From 060127af38bd3a732e64cf23f06cd4cef26ac67e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 13:22:11 +1100 Subject: nouveau: update to latest header --- src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 65 +- src/mesa/pipe/nouveau/nouveau_class.h | 10925 ++++++++++---------- src/mesa/pipe/nv40/nv40_state.c | 122 +- src/mesa/pipe/nv40/nv40_state.h | 51 - src/mesa/pipe/nv40/nv40_state_tex.c | 49 +- 5 files changed, 5877 insertions(+), 5335 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c index f41e45c73a6..98f0d554058 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c @@ -6,9 +6,9 @@ static INLINE int nv04_surface_format(int cpp) { switch (cpp) { - case 1: return 0x1; - case 2: return 0x4; - case 4: return 0xb; + case 1: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; + case 2: return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; + case 4: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; default: return -1; } @@ -18,9 +18,9 @@ static INLINE int nv04_rect_format(int cpp) { switch (cpp) { - case 1: return 0x3; - case 2: return 0x1; - case 4: return 0x3; + case 1: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + case 2: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; + case 4: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; default: return -1; } @@ -78,8 +78,6 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, if (src->cpp != dst->cpp) return 1; - NOUVEAU_ERR("preg\n"); - /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback * to NV_MEMORY_TO_MEMORY_FORMAT in this case. */ @@ -96,10 +94,10 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, return 1; } - BEGIN_RING(NvCtxSurf2D, 0x0184, 2); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvCtxSurf2D, 0x0300, 4); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (format); OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); @@ -131,20 +129,21 @@ nv04_region_fill(struct nouveau_context *nv, return 1; } - BEGIN_RING(NvCtxSurf2D, 0x0184, 2); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvCtxSurf2D, 0x0300, 4); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (cs2d_format); OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvGdiRect, 0x0300, 1); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); OUT_RING (gdirect_format); - BEGIN_RING(NvGdiRect, 0x03fc, 1); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); OUT_RING (value); - BEGIN_RING(NvGdiRect, 0x0400, 2); + BEGIN_RING(NvGdiRect, + NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); OUT_RING ((dx << 16) | dy); OUT_RING (( w << 16) | h); @@ -164,6 +163,7 @@ nv04_region_data(struct nouveau_context *nv, struct pipe_region *dst, int nouveau_region_init_nv04(struct nouveau_context *nv) { + unsigned class; int ret; if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, @@ -171,43 +171,48 @@ nouveau_region_init_nv04(struct nouveau_context *nv) NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); return 1; } - BEGIN_RING(NvM2MF, 0x0180, 1); + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x62, + class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : + NV10_CONTEXT_SURFACES_2D; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, &nv->NvCtxSurf2D))) { NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); return 1; } - BEGIN_RING(NvCtxSurf2D, 0x0184, 2); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RING (nv->channel->vram->handle); OUT_RING (nv->channel->vram->handle); - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x9f, + class = nv->chipset < 0x10 ? NV_IMAGE_BLIT : + NV12_IMAGE_BLIT; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, &nv->NvImageBlit))) { NOUVEAU_ERR("Error creating blit object: %d\n", ret); return 1; } - BEGIN_RING(NvImageBlit, 0x0180, 1); + BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvImageBlit, 0x019c, 1); + BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_SURFACE, 1); OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvImageBlit, 0x02fc, 1); - OUT_RING (3); + BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_OPERATION, 1); + OUT_RING (NV_IMAGE_BLIT_OPERATION_SRCCOPY); - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x4a, + class = NV04_GDI_RECTANGLE_TEXT; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, &nv->NvGdiRect))) { NOUVEAU_ERR("Error creating rect object: %d\n", ret); return 1; } - BEGIN_RING(NvGdiRect, 0x0180, 1); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvGdiRect, 0x0198, 1); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvGdiRect, 0x02fc, 1); - OUT_RING (3); - BEGIN_RING(NvGdiRect, 0x0304, 1); - OUT_RING (2); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); + OUT_RING (NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); + OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); nv->region_display = nv04_region_display; nv->region_copy = nv04_region_copy; diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h index 94b667e44d2..22b526a54ea 100644 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -1,5260 +1,5783 @@ +/************************************************************************* + + Autogenerated file, do not edit ! + +************************************************************************** + + Copyright (C) 2006-2007 : + Dmitry Baryshkov, + Laurent Carlier, + Matthieu Castet, + Dawid Gajownik, + Jeremy Kolb, + Stephane Loeuillet, + Patrice Mandin, + Stephane Marchesin, + Serge Martin, + Sylvain Munaut, + Ben Skeggs, + Erik Waling, + koala_br, + sturmflut. + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +*************************************************************************/ + + #ifndef NOUVEAU_REG_H #define NOUVEAU_REG_H 1 -#define NV01_ROOT 0x00000001 - +#define NV01_ROOT 0x00000001 -#define NV01_CONTEXT_DMA 0x00000002 +#define NV01_CONTEXT_DMA 0x00000002 -#define NV01_DEVICE 0x00000003 +#define NV01_DEVICE 0x00000003 -#define NV01_TIMER 0x00000004 -#define NV01_TIMER_SYNCHRONIZE 0x00000100 -#define NV01_TIMER_STOP_ALARM 0x00000104 -#define NV01_TIMER_DMA_NOTIFY 0x00000180 -#define NV01_TIMER_TIME(x) (0x00000300+((x)*4)) -#define NV01_TIMER_TIME__SIZE 0x00000002 -#define NV01_TIMER_ALARM_NOTIFY 0x00000308 +#define NV01_TIMER 0x00000004 +#define NV01_TIMER_SYNCHRONIZE 0x00000100 +#define NV01_TIMER_STOP_ALARM 0x00000104 +#define NV01_TIMER_DMA_NOTIFY 0x00000180 +#define NV01_TIMER_TIME(x) (0x00000300+((x)*4)) +#define NV01_TIMER_TIME__SIZE 0x00000002 +#define NV01_TIMER_ALARM_NOTIFY 0x00000308 -#define NV_IMAGE_STENCIL 0x00000010 -#define NV_IMAGE_STENCIL_NOTIFY 0x00000104 -#define NV_IMAGE_STENCIL_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_STENCIL_IMAGE_OUTPUT 0x00000200 -#define NV_IMAGE_STENCIL_IMAGE_INPUT(x) (0x00000204+((x)*4)) -#define NV_IMAGE_STENCIL_IMAGE_INPUT__SIZE 0x00000002 +#define NV_IMAGE_STENCIL 0x00000010 +#define NV_IMAGE_STENCIL_NOTIFY 0x00000104 +#define NV_IMAGE_STENCIL_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_STENCIL_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_STENCIL_IMAGE_INPUT(x) (0x00000204+((x)*4)) +#define NV_IMAGE_STENCIL_IMAGE_INPUT__SIZE 0x00000002 -#define NV_IMAGE_BLEND_AND 0x00000011 -#define NV_IMAGE_BLEND_AND_NOP 0x00000100 -#define NV_IMAGE_BLEND_AND_NOTIFY 0x00000104 -#define NV_IMAGE_BLEND_AND_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_BLEND_AND_IMAGE_OUTPUT 0x00000200 -#define NV_IMAGE_BLEND_AND_BETA_INPUT 0x00000204 -#define NV_IMAGE_BLEND_AND_IMAGE_INPUT 0x00000208 +#define NV_IMAGE_BLEND_AND 0x00000011 +#define NV_IMAGE_BLEND_AND_NOP 0x00000100 +#define NV_IMAGE_BLEND_AND_NOTIFY 0x00000104 +#define NV_IMAGE_BLEND_AND_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_BLEND_AND_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_BLEND_AND_BETA_INPUT 0x00000204 +#define NV_IMAGE_BLEND_AND_IMAGE_INPUT 0x00000208 -#define NV01_CONTEXT_BETA1 0x00000012 -#define NV01_CONTEXT_BETA1_NOP 0x00000100 -#define NV01_CONTEXT_BETA1_NOTIFY 0x00000104 -#define NV01_CONTEXT_BETA1_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_BETA1_BETA_1D31 0x00000300 +#define NV01_CONTEXT_BETA1 0x00000012 +#define NV01_CONTEXT_BETA1_NOP 0x00000100 +#define NV01_CONTEXT_BETA1_NOTIFY 0x00000104 +#define NV01_CONTEXT_BETA1_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_BETA1_BETA_1D31 0x00000300 -#define NV_IMAGE_ROP_AND 0x00000013 -#define NV_IMAGE_ROP_AND_NOTIFY 0x00000104 -#define NV_IMAGE_ROP_AND_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_ROP_AND_IMAGE_OUTPUT 0x00000200 -#define NV_IMAGE_ROP_AND_ROP_INPUT 0x00000204 -#define NV_IMAGE_ROP_AND_IMAGE_INPUT(x) (0x00000208+((x)*4)) -#define NV_IMAGE_ROP_AND_IMAGE_INPUT__SIZE 0x00000002 - - -#define NV_IMAGE_COLOR_KEY 0x00000015 - - - -#define NV01_CONTEXT_COLOR_KEY 0x00000017 - -#define NV01_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 -#define NV01_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A8Y8 0x00000001 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X24Y8 0x00000002 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A1R5G5B5 0x00000003 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X17R5G5B5 0x00000004 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A8R8G8B8 0x00000005 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X8R8G8B8 0x00000006 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A16Y16 0x00000007 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16Y16 0x00000008 -#define NV01_CONTEXT_COLOR_KEY_COLOR 0x00000304 - - -#define NV01_CONTEXT_PATTERN 0x00000018 - -#define NV01_CONTEXT_PATTERN_NOTIFY 0x00000104 -#define NV01_CONTEXT_PATTERN_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_PATTERN_COLOR_FORMAT 0x00000300 -#define NV01_CONTEXT_PATTERN_MONOCHROME_FORMAT 0x00000304 -#define NV01_CONTEXT_PATTERN_SHAPE 0x00000308 -#define NV01_CONTEXT_PATTERN_COLOR(x) (0x00000310+((x)*4)) -#define NV01_CONTEXT_PATTERN_COLOR__SIZE 0x00000002 -#define NV01_CONTEXT_PATTERN_PATTERN(x) (0x00000318+((x)*4)) -#define NV01_CONTEXT_PATTERN_PATTERN__SIZE 0x00000002 - - -#define NV01_CONTEXT_CLIP_RECTANGLE 0x00000019 - -#define NV01_CONTEXT_CLIP_RECTANGLE_NOP 0x00000104 -#define NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY 0x00000104 -#define NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT 0x00000300 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_SHIFT 0 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_MASK 0x0000ffff -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_SHIFT 16 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_MASK 0xffff0000 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE 0x00000304 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_SHIFT 0 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_MASK 0x0000ffff -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_SHIFT 16 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_MASK 0xffff0000 - - -#define NV01_RENDER_SOLID_LINE 0x0000001c - -#define NV01_RENDER_SOLID_LINE_NOP 0x00000104 -#define NV01_RENDER_SOLID_LINE_NOTIFY 0x00000104 -#define NV01_RENDER_SOLID_LINE_PATCH 0x0000010c -#define NV01_RENDER_SOLID_LINE_DMA_NOTIFY 0x00000180 -#define NV01_RENDER_SOLID_LINE_CLIP_RECTANGLE 0x00000184 -#define NV01_RENDER_SOLID_LINE_PATTERN 0x00000188 -#define NV01_RENDER_SOLID_LINE_ROP 0x0000018c -#define NV01_RENDER_SOLID_LINE_BETA1 0x00000190 -#define NV01_RENDER_SOLID_LINE_SURFACE 0x00000194 -#define NV01_RENDER_SOLID_LINE_OPERATION 0x000002fc -#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_RENDER_SOLID_LINE_OPERATION_ROP_AND 0x00000001 -#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_AND 0x00000002 -#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_PREMULT 0x00000003 -#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_PREMULT 0x00000004 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT 0x00000300 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A8Y8 0x00000001 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X24Y8 0x00000002 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A1R5G5B5 0x00000003 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X17R5G5B5 0x00000004 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A8R8G8B8 0x00000005 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X8R8G8B8 0x00000006 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A16Y16 0x00000007 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16Y16 0x00000008 -#define NV01_RENDER_SOLID_LINE_COLOR 0x00000304 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0(x) (0x00000400+((x)*8)) -#define NV01_RENDER_SOLID_LINE_LINE_POINT0__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1(x) (0x00000404+((x)*8)) -#define NV01_RENDER_SOLID_LINE_LINE_POINT1__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X(x) (0x00000480+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y(x) (0x00000484+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X(x) (0x00000488+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y(x) (0x0000048c+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_POLYLINE(x) (0x00000500+((x)*4)) -#define NV01_RENDER_SOLID_LINE_POLYLINE__SIZE 0x00000020 -#define NV01_RENDER_SOLID_LINE_POLYLINE_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_POLYLINE_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X(x) (0x00000580+((x)*8)) -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y(x) (0x00000584+((x)*8)) -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR(x) (0x00000600+((x)*8)) -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT(x) (0x00000604+((x)*8)) -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_MASK 0xffff0000 - - -#define NV01_RENDER_SOLID_TRIANGLE 0x0000001d - -#define NV01_RENDER_SOLID_TRIANGLE_NOP 0x00000100 -#define NV01_RENDER_SOLID_TRIANGLE_NOTIFY 0x00000104 -#define NV01_RENDER_SOLID_TRIANGLE_PATCH 0x0000010c -#define NV01_RENDER_SOLID_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV01_RENDER_SOLID_TRIANGLE_CLIP_RECTANGLE 0x00000184 -#define NV01_RENDER_SOLID_TRIANGLE_PATTERN 0x00000188 -#define NV01_RENDER_SOLID_TRIANGLE_ROP 0x0000018c -#define NV01_RENDER_SOLID_TRIANGLE_BETA1 0x00000190 -#define NV01_RENDER_SOLID_TRIANGLE_SURFACE 0x00000194 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION 0x000002fc -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_ROP_AND 0x00000001 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_AND 0x00000002 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_PREMULT 0x00000003 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_PREMULT 0x00000004 -#define NV01_RENDER_SOLID_TRIANGLE_COLOR_FORMAT 0x00000300 -#define NV01_RENDER_SOLID_TRIANGLE_COLOR 0x00000304 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0 0x00000310 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1 0x00000314 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2 0x00000318 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_X 0x00000320 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_Y 0x00000324 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_X 0x00000328 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_Y 0x0000032c -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_X 0x00000330 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_Y 0x00000334 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH(x) (0x00000400+((x)*4)) -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH__SIZE 0x00000020 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X(x) (0x00000480+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y(x) (0x00000484+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR(x) (0x00000500+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0(x) (0x00000504+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1(x) (0x00000508+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2(x) (0x0000050c+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR(x) (0x00000580+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT(x) (0x00000584+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_MASK 0xffff0000 - - -#define NV01_RENDER_SOLID_RECTANGLE 0x0000001e - -#define NV01_RENDER_SOLID_RECTANGLE_NOP 0x00000100 -#define NV01_RENDER_SOLID_RECTANGLE_NOTIFY 0x00000104 -#define NV01_RENDER_SOLID_RECTANGLE_PATCH 0x0000010c -#define NV01_RENDER_SOLID_RECTANGLE_DMA_NOTIFY 0x00000180 -#define NV01_RENDER_SOLID_RECTANGLE_CLIP_RECTANGLE 0x00000184 -#define NV01_RENDER_SOLID_RECTANGLE_PATTERN 0x00000188 -#define NV01_RENDER_SOLID_RECTANGLE_ROP 0x0000018c -#define NV01_RENDER_SOLID_RECTANGLE_BETA1 0x00000190 -#define NV01_RENDER_SOLID_RECTANGLE_SURFACE 0x00000194 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION 0x000002fc -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_ROP_AND 0x00000001 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_AND 0x00000002 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_PREMULT 0x00000003 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_PREMULT 0x00000004 -#define NV01_RENDER_SOLID_RECTANGLE_COLOR_FORMAT 0x00000300 -#define NV01_RENDER_SOLID_RECTANGLE_COLOR 0x00000304 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT(x) (0x00000400+((x)*8)) -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT__SIZE 0x00000010 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_SHIFT 0 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_SHIFT 16 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE__SIZE 0x00000010 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_SHIFT 0 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_MASK 0x0000ffff -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_SHIFT 16 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_MASK 0xffff0000 - - -#define NV01_IMAGE_BLIT 0x0000001f - -#define NV01_IMAGE_BLIT_NOP 0x00000100 -#define NV01_IMAGE_BLIT_NOTIFY 0x00000104 -#define NV01_IMAGE_BLIT_PATCH 0x0000010c -#define NV01_IMAGE_BLIT_DMA_NOTIFY 0x00000180 -#define NV01_IMAGE_BLIT_COLOR_KEY 0x00000184 -#define NV01_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 -#define NV01_IMAGE_BLIT_PATTERN 0x0000018c -#define NV01_IMAGE_BLIT_ROP 0x00000190 -#define NV01_IMAGE_BLIT_BETA1 0x00000194 -#define NV01_IMAGE_BLIT_SURFACE 0x0000019c -#define NV01_IMAGE_BLIT_OPERATION 0x000002fc -#define NV01_IMAGE_BLIT_IMAGE_INPUT 0x00000204 -#define NV01_IMAGE_BLIT_POINT_IN 0x00000300 -#define NV01_IMAGE_BLIT_POINT_IN_X_SHIFT 0 -#define NV01_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff -#define NV01_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 -#define NV01_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 -#define NV01_IMAGE_BLIT_POINT_OUT 0x00000304 -#define NV01_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 -#define NV01_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff -#define NV01_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 -#define NV01_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 -#define NV01_IMAGE_BLIT_SIZE 0x00000308 -#define NV01_IMAGE_BLIT_SIZE_W_SHIFT 0 -#define NV01_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff -#define NV01_IMAGE_BLIT_SIZE_H_SHIFT 16 -#define NV01_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 - - -#define NV01_IMAGE_FROM_CPU 0x00000021 - -#define NV01_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV01_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV01_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV01_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV01_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 -#define NV01_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 -#define NV01_IMAGE_FROM_CPU_PATTERN 0x0000018c -#define NV01_IMAGE_FROM_CPU_ROP 0x00000190 -#define NV01_IMAGE_FROM_CPU_BETA1 0x00000194 -#define NV01_IMAGE_FROM_CPU_SURFACE 0x00000198 -#define NV01_IMAGE_FROM_CPU_OPERATION 0x000002fc -#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_IMAGE_FROM_CPU_OPERATION_ROP_AND 0x00000001 -#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_AND 0x00000002 -#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_PREMULT 0x00000003 -#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_PREMULT 0x00000004 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_Y8 0x00000001 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A1R5G5B5 0x00000002 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X1R5G5B5 0x00000003 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A8R8G8B8 0x00000004 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X8R8G8B8 0x00000005 -#define NV01_IMAGE_FROM_CPU_POINT 0x00000304 -#define NV01_IMAGE_FROM_CPU_POINT_X_SHIFT 0 -#define NV01_IMAGE_FROM_CPU_POINT_X_MASK 0x0000ffff -#define NV01_IMAGE_FROM_CPU_POINT_Y_SHIFT 16 -#define NV01_IMAGE_FROM_CPU_POINT_Y_MASK 0xffff0000 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_SHIFT 0 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_MASK 0x0000ffff -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_SHIFT 16 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_MASK 0xffff0000 -#define NV01_IMAGE_FROM_CPU_SIZE_IN 0x0000030c -#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 -#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff -#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 -#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 -#define NV01_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) -#define NV01_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 - - -#define NV01_NULL 0x00000030 - - - -#define NV03_STRETCHED_IMAGE_FROM_CPU 0x00000036 - -#define NV03_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV03_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV03_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV03_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 -#define NV03_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 -#define NV03_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c -#define NV03_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000194 -#define NV03_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_DX_DU 0x00000308 -#define NV03_STRETCHED_IMAGE_FROM_CPU_DY_DV 0x0000030c -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT 0x00000310 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE 0x00000314 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 - - -#define NV03_SCALED_IMAGE_FROM_MEMORY 0x00000037 - -#define NV03_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 -#define NV03_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 -#define NV03_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 -#define NV03_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c -#define NV03_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 -#define NV03_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000194 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000003 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000004 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT 0x00000310 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE 0x00000314 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DU_DX 0x00000318 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DV_DY 0x0000031c -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE 0x00000400 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT 0x00000404 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_MASK 0x00ff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CENTER 0x00000001 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CORNER 0x00000002 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_SHIFT 24 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_MASK 0xff000000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_OFFSET 0x00000408 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT 0x0000040c -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_MASK 0xffff0000 - - -#define NV04_DVD_SUBPICTURE 0x00000038 - -#define NV04_DVD_SUBPICTURE_NOP 0x00000100 -#define NV04_DVD_SUBPICTURE_NOTIFY 0x00000104 -#define NV04_DVD_SUBPICTURE_WAIT_FOR_IDLE 0x00000108 -#define NV04_DVD_SUBPICTURE_DMA_NOTIFY 0x00000180 -#define NV04_DVD_SUBPICTURE_DMA_OVERLAY 0x00000184 -#define NV04_DVD_SUBPICTURE_DMA_IMAGEIN 0x00000188 -#define NV04_DVD_SUBPICTURE_DMA_IMAGEOUT 0x0000018c -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT 0x00000300 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE 0x00000304 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT 0x00000308 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_OFFSET 0x0000030c -#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DU_DX 0x00000310 -#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DV_DY 0x00000314 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE 0x00000318 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT 0x0000031c -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEIN_OFFSET 0x00000320 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT 0x00000324 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DU_DX 0x00000328 -#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DV_DY 0x0000032c -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE 0x00000330 -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_SHIFT 0 -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_SHIFT 16 -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT 0x00000334 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_SHIFT 0 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_SHIFT 16 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_OVERLAY_OFFSET 0x00000338 -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT 0x0000033c -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_SHIFT 0 -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_SHIFT 16 -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK 0xffff0000 - - -#define NV_MEMORY_TO_MEMORY_FORMAT 0x00000039 - -#define NV_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 -#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 -#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 -#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 -#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 -#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c -#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 -#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 -#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 -#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c -#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 -#define NV_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 - - -#define NV01_MEMORY_LOCAL_BANKED 0x0000003d - - - -#define NV01_MAPPING_SYSTEM 0x0000003e - - - -#define NV03_MEMORY_LOCAL_CURSOR 0x0000003f - - - -#define NV01_MEMORY_LOCAL_LINEAR 0x00000040 - - - -#define NV01_MAPPING_LOCAL 0x00000041 - - - -#define NV04_CONTEXT_SURFACES_2D 0x00000042 - -#define NV04_CONTEXT_SURFACES_2D_NOP 0x00000100 -#define NV04_CONTEXT_SURFACES_2D_NOTIFY 0x00000104 -#define NV04_CONTEXT_SURFACES_2D_PM_TRIGGER 0x00000140 -#define NV04_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 -#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE 0x00000184 -#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_DESTIN 0x00000188 -#define NV04_CONTEXT_SURFACES_2D_FORMAT 0x00000300 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y8 0x00000001 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_Z1R5G5B5 0x00000002 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_X1R5G5B5 0x00000003 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5 0x00000004 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y16 0x00000005 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_Z8R8G8B8 0x00000006 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_X8R8G8B8 0x00000007 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_X1A7R8G8B8 0x00000009 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8 0x0000000a -#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y32 0x0000000b -#define NV04_CONTEXT_SURFACES_2D_PITCH 0x00000304 -#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 -#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 -#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 -#define NV04_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c - - -#define NV03_CONTEXT_ROP 0x00000043 - -#define NV03_CONTEXT_ROP_NOP 0x00000100 -#define NV03_CONTEXT_ROP_NOTIFY 0x00000104 -#define NV03_CONTEXT_ROP_DMA_NOTIFY 0x00000180 -#define NV03_CONTEXT_ROP_ROP 0x00000300 -#define NV03_CONTEXT_ROP_ROP_VALUE0_SHIFT 0 -#define NV03_CONTEXT_ROP_ROP_VALUE0_MASK 0x0000000f -#define NV03_CONTEXT_ROP_ROP_VALUE0_CLEAR 0x00000000 -#define NV03_CONTEXT_ROP_ROP_VALUE0_NOR 0x00000001 -#define NV03_CONTEXT_ROP_ROP_VALUE0_AND_INVERTED 0x00000002 -#define NV03_CONTEXT_ROP_ROP_VALUE0_COPY_INVERTED 0x00000003 -#define NV03_CONTEXT_ROP_ROP_VALUE0_AND_REVERSE 0x00000004 -#define NV03_CONTEXT_ROP_ROP_VALUE0_INVERT 0x00000005 -#define NV03_CONTEXT_ROP_ROP_VALUE0_XOR 0x00000006 -#define NV03_CONTEXT_ROP_ROP_VALUE0_NAND 0x00000007 -#define NV03_CONTEXT_ROP_ROP_VALUE0_AND 0x00000008 -#define NV03_CONTEXT_ROP_ROP_VALUE0_EQUI 0x00000009 -#define NV03_CONTEXT_ROP_ROP_VALUE0_NOOP 0x0000000a -#define NV03_CONTEXT_ROP_ROP_VALUE0_OR_INVERTED 0x0000000b -#define NV03_CONTEXT_ROP_ROP_VALUE0_COPY 0x0000000c -#define NV03_CONTEXT_ROP_ROP_VALUE0_OR_REVERSE 0x0000000d -#define NV03_CONTEXT_ROP_ROP_VALUE0_OR 0x0000000e -#define NV03_CONTEXT_ROP_ROP_VALUE0_SET 0x0000000f -#define NV03_CONTEXT_ROP_ROP_VALUE1_SHIFT 4 -#define NV03_CONTEXT_ROP_ROP_VALUE1_MASK 0x000000f0 -#define NV03_CONTEXT_ROP_ROP_VALUE1_CLEAR 0x00000000 -#define NV03_CONTEXT_ROP_ROP_VALUE1_NOR 0x00000001 -#define NV03_CONTEXT_ROP_ROP_VALUE1_AND_INVERTED 0x00000002 -#define NV03_CONTEXT_ROP_ROP_VALUE1_COPY_INVERTED 0x00000003 -#define NV03_CONTEXT_ROP_ROP_VALUE1_AND_REVERSE 0x00000004 -#define NV03_CONTEXT_ROP_ROP_VALUE1_INVERT 0x00000005 -#define NV03_CONTEXT_ROP_ROP_VALUE1_XOR 0x00000006 -#define NV03_CONTEXT_ROP_ROP_VALUE1_NAND 0x00000007 -#define NV03_CONTEXT_ROP_ROP_VALUE1_AND 0x00000008 -#define NV03_CONTEXT_ROP_ROP_VALUE1_EQUI 0x00000009 -#define NV03_CONTEXT_ROP_ROP_VALUE1_NOOP 0x0000000a -#define NV03_CONTEXT_ROP_ROP_VALUE1_OR_INVERTED 0x0000000b -#define NV03_CONTEXT_ROP_ROP_VALUE1_COPY 0x0000000c -#define NV03_CONTEXT_ROP_ROP_VALUE1_OR_REVERSE 0x0000000d -#define NV03_CONTEXT_ROP_ROP_VALUE1_OR 0x0000000e -#define NV03_CONTEXT_ROP_ROP_VALUE1_SET 0x0000000f - - -#define NV04_IMAGE_PATTERN 0x00000044 - -#define NV04_IMAGE_PATTERN_NOP 0x00000100 -#define NV04_IMAGE_PATTERN_NOTIFY 0x00000104 -#define NV04_IMAGE_PATTERN_DMA_NOTIFY 0x00000180 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT 0x00000300 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A16R5G6B5 0x00000001 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT_X16A1R5G5B5 0x00000002 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT 0x00000304 -#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_CGA6 0x00000001 -#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_LE 0x00000002 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE 0x00000308 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_8X8 0x00000000 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_64X1 0x00000001 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_1X64 0x00000002 -#define NV04_IMAGE_PATTERN_PATTERN_SELECT 0x0000030c -#define NV04_IMAGE_PATTERN_PATTERN_SELECT_MONO 0x00000001 -#define NV04_IMAGE_PATTERN_PATTERN_SELECT_COLOR 0x00000002 -#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR0 0x00000310 -#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR1 0x00000314 -#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0 0x00000318 -#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1 0x0000031c -#define NV04_IMAGE_PATTERN_PATTERN_Y8(x) (0x00000400+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_Y8__SIZE 0x00000010 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_MASK 0x000000ff -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_SHIFT 8 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_MASK 0x0000ff00 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_MASK 0x00ff0000 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_SHIFT 24 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_MASK 0xff000000 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5(x) (0x00000500+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5__SIZE 0x00000020 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_MASK 0x0000001f -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_SHIFT 5 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_MASK 0x000007e0 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_SHIFT 11 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_MASK 0x0000f800 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_MASK 0x001f0000 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_SHIFT 21 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_MASK 0x07e00000 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_SHIFT 27 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_MASK 0xf8000000 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5(x) (0x00000600+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5__SIZE 0x00000020 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_MASK 0x0000001f -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_SHIFT 5 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_MASK 0x000003e0 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_SHIFT 10 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_MASK 0x00007c00 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_MASK 0x001f0000 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_SHIFT 21 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_MASK 0x03e00000 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_SHIFT 26 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_MASK 0x7c000000 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8(x) (0x00000700+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8__SIZE 0x00000040 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_MASK 0x000000ff -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_SHIFT 8 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_MASK 0x0000ff00 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_MASK 0x00ff0000 - - -#define NV03_VIDEO_LUT_CURSOR_DAC 0x00000046 - -#define NV03_VIDEO_LUT_CURSOR_DAC_SYNCHRONIZE 0x00000100 -#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_IMAGE 0x00000104 -#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_CURSOR 0x00000108 -#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_DAC 0x0000010c -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_NOTIFY 0x00000180 -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE(x) (0x00000184+((x)*4)) -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT(x) (0x0000018c+((x)*4)) -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR(x) (0x00000194+((x)*4)) -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_GET 0x000002fc -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET(x) (0x00000300+((x)*8)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT(x) (0x00000304+((x)*8)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET(x) (0x00000340+((x)*12)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT(x) (0x00000344+((x)*12)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_MASK 0xffff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT(x) (0x00000348+((x)*12)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A 0x00000358 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_MASK 0xffff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE(x) (0x00000380+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_MASK 0xffff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC(x) (0x00000384+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC(x) (0x00000388+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE(x) (0x0000038c+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_PIXEL_CLOCK 0x000003a0 - - -#define NV03_DX3_TEXTURED_TRIANGLE 0x00000048 - -#define NV03_DX3_TEXTURED_TRIANGLE_NOP 0x00000100 -#define NV03_DX3_TEXTURED_TRIANGLE_NOTIFY 0x00000104 -#define NV03_DX3_TEXTURED_TRIANGLE_PATCH 0x0000010c -#define NV03_DX3_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV03_DX3_TEXTURED_TRIANGLE_DMA_TEXTURE 0x00000184 -#define NV03_DX3_TEXTURED_TRIANGLE_CLIP_RECTANGLE 0x00000188 -#define NV03_DX3_TEXTURED_TRIANGLE_SURFACE 0x0000018c -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_OFFSET 0x00000304 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT 0x00000308 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_MASK 0x0000ffff -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_MASK 0x000f0000 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT 20 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK 0x00f00000 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_SHIFT 24 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_MASK 0x0f000000 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_SHIFT 28 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_MASK 0xf0000000 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER 0x0000030c -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_MASK 0x0000001f -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_MASK 0x00001f00 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_MASK 0x00ff0000 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR 0x00000310 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_MASK 0x000000ff -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_MASK 0x0000ff00 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_MASK 0x00ff0000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT 0x00000314 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_MASK 0x0000000f -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_SHIFT 4 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_MASK 0x00000030 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_SHIFT 6 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_MASK 0x000000c0 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_MASK 0x00000f00 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_SHIFT 12 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_MASK 0x00007000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_PERSPECTIVE_ENABLE (1 << 15) -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_MASK 0x000f0000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_SHIFT 20 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_MASK 0x00f00000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_SHIFT 24 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_MASK 0x07000000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_SHIFT 27 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_MASK 0x18000000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_BETA (1 << 29) -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_DST_BLEND (1 << 30) -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SRC_BLEND (1 << 31) -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL 0x00000318 -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_MASK 0x000000ff -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_MASK 0xffffff00 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR(x) (0x00001000+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_MASK 0x0000000f -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_SHIFT 4 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_MASK 0x000000f0 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_MASK 0x00000f00 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_SHIFT 12 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_MASK 0x0000f000 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_MASK 0x000f0000 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_SHIFT 20 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_MASK 0x00f00000 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_SHIFT 24 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_MASK 0xff000000 -#define NV03_DX3_TEXTURED_TRIANGLE_COLOR(x) (0x00001004+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_COLOR__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_X(x) (0x00001008+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_X__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_Y(x) (0x0000100c+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_Y__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_Z(x) (0x00001010+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_Z__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_M(x) (0x00001014+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_M__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_U(x) (0x00001018+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_U__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_V(x) (0x0000101c+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_V__SIZE 0x00000040 - - -#define NV04_GDI_RECTANGLE_TEXT 0x0000004a - -#define NV04_GDI_RECTANGLE_TEXT_NOP 0x00000100 -#define NV04_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 -#define NV04_GDI_RECTANGLE_TEXT_PATCH 0x0000010c -#define NV04_GDI_RECTANGLE_TEXT_PM_TRIGGER 0x00000140 -#define NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 -#define NV04_GDI_RECTANGLE_TEXT_DMA_FONTS 0x00000184 -#define NV04_GDI_RECTANGLE_TEXT_PATTERN 0x00000188 -#define NV04_GDI_RECTANGLE_TEXT_ROP 0x0000018c -#define NV04_GDI_RECTANGLE_TEXT_BETA1 0x00000190 -#define NV04_GDI_RECTANGLE_TEXT_BETA4 0x00000194 -#define NV04_GDI_RECTANGLE_TEXT_SURFACE 0x00000198 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5 0x00000001 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_X16A1R5G5B5 0x00000002 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_CGA6 0x00000001 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE 0x00000002 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(x) (0x00000400+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0 0x000005f4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1 0x000005f8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_B 0x000005fc -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0(x) (0x00000600+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1(x) (0x00000604+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x000007ec -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x000007f0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_C 0x000007f4 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C 0x000007f8 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_POINT_C 0x000007fc -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000800+((x)*4)) -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000080 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x00000be4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x00000be8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR0_E 0x00000bec -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_E 0x00000bf0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x00000bf4 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x00000bf8 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_POINT_E 0x00000bfc -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00000c00+((x)*4)) -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000080 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F 0x00000ff0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_MASK 0x0fffffff -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_SHIFT 28 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_MASK 0xf0000000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0 0x00000ff4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1 0x00000ff8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_F 0x00000ffc -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F(x) (0x00001000+((x)*4)) -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F__SIZE 0x00000100 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_MASK 0x000000ff -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_SHIFT 8 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_MASK 0x000fff00 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_SHIFT 20 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_MASK 0xfff00000 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G 0x000017f0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_MASK 0x0fffffff -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_SHIFT 28 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_MASK 0xf0000000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0 0x000017f4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1 0x000017f8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_G 0x000017fc -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT(x) (0x00001800+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT__SIZE 0x00000100 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX(x) (0x00001804+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX__SIZE 0x00000100 - - -#define NV03_GDI_RECTANGLE_TEXT 0x0000004b - -#define NV03_GDI_RECTANGLE_TEXT_NOP 0x00000100 -#define NV03_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 -#define NV03_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 -#define NV03_GDI_RECTANGLE_TEXT_PATTERN 0x00000184 -#define NV03_GDI_RECTANGLE_TEXT_ROP 0x00000188 -#define NV03_GDI_RECTANGLE_TEXT_BETA1 0x0000018c -#define NV03_GDI_RECTANGLE_TEXT_SURFACE 0x00000190 -#define NV03_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc -#define NV03_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT 0x00000400 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE 0x00000404 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B 0x000007f4 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B 0x000007f8 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_B 0x000007fc -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0 0x00000800 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1 0x00000804 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x00000bec -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x00000bf0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_C 0x00000bf4 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C 0x00000bf8 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_POINT_C 0x00000bfc -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000c00+((x)*4)) -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000020 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0 0x00000fe8 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1 0x00000fec -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_D 0x00000ff0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D 0x00000ff4 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D 0x00000ff8 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_POINT_D 0x00000ffc -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D(x) (0x00001000+((x)*4)) -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D__SIZE 0x00000020 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x000013e4 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x000013e8 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR0_E 0x000013ec -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_E 0x000013f0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x000013f4 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x000013f8 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_POINT_E 0x000013fc -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00001400+((x)*4)) -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000020 - - -#define NV04_SWIZZLED_SURFACE 0x00000052 - -#define NV04_SWIZZLED_SURFACE_NOP 0x00000100 -#define NV04_SWIZZLED_SURFACE_NOTIFY 0x00000104 -#define NV04_SWIZZLED_SURFACE_DMA_NOTIFY 0x00000180 -#define NV04_SWIZZLED_SURFACE_DMA_IMAGE 0x00000184 -#define NV04_SWIZZLED_SURFACE_FORMAT 0x00000300 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_SHIFT 0 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK 0x000000ff -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y8 0x00000001 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000002 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000003 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_R5G6B5 0x00000004 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y16 0x00000005 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000006 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000007 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000009 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8 0x0000000a -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y32 0x0000000b -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_MASK 0xff000000 -#define NV04_SWIZZLED_SURFACE_OFFSET 0x00000304 - - -#define NV04_CONTEXT_SURFACES_3D 0x00000053 - -#define NV04_CONTEXT_SURFACES_3D_NOP 0x00000100 -#define NV04_CONTEXT_SURFACES_3D_NOTIFY 0x00000104 -#define NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 -#define NV04_CONTEXT_SURFACES_3D_DMA_COLOR 0x00000184 -#define NV04_CONTEXT_SURFACES_3D_DMA_ZETA 0x00000188 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL 0x000002f8 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL 0x000002fc -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_FORMAT 0x00000300 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK 0x000000ff -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000001 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000002 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_R5G6B5 0x00000003 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000004 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000005 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000006 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000007 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_A8R8G8B8 0x00000008 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT 8 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK 0x0000ff00 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_PITCH 0x00000001 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SWIZZLE 0x00000002 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_MASK 0xff000000 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE 0x00000304 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_PITCH 0x00000308 -#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x0000030c -#define NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000310 - - -#define NV04_DX5_TEXTURED_TRIANGLE 0x00000054 - -#define NV04_DX5_TEXTURED_TRIANGLE_NOP 0x00000100 -#define NV04_DX5_TEXTURED_TRIANGLE_NOTIFY 0x00000104 -#define NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV04_DX5_TEXTURED_TRIANGLE_DMA_A 0x00000184 -#define NV04_DX5_TEXTURED_TRIANGLE_DMA_B 0x00000188 -#define NV04_DX5_TEXTURED_TRIANGLE_SURFACE 0x0000018c -#define NV04_DX5_TEXTURED_TRIANGLE_COLORKEY 0x00000300 -#define NV04_DX5_TEXTURED_TRIANGLE_OFFSET 0x00000304 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT 0x00000308 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_MASK 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_SHIFT 2 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_MASK 0x0000000c -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CENTER 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CENTER 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_Y8 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A1R5G5B5 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X1R5G5B5 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A4R4G4B4 0x00000004 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_R5G6B5 0x00000005 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A8R8G8B8 0x00000006 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X8R8G8B8 0x00000007 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER 0x00000004 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP 0x00000005 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPU (1 << 27) -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_REPEAT 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MIRRORED_REPEAT 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_BORDER 0x00000004 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP 0x00000005 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPV (1 << 31) -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER 0x0000030c -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND 0x00000310 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_MASK 0x0000000f -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT 0x00000001 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD 0x00000002 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_PHONG 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK 0x0f000000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT 28 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK 0xf0000000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL 0x00000314 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE (1 << 12) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN (1 << 13) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT 14 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_MASK 0x0000c000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_MASK 0x00300000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE (1 << 22) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE (1 << 23) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_MASK 0x3f000000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT 30 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_MASK 0xc0000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR 0x00000318 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(x) (0x00000400+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY(x) (0x00000404+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ(x) (0x00000408+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW(x) (0x0000040c+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR(x) (0x00000410+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_MASK 0x0000ff00 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_MASK 0xff000000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR(x) (0x00000414+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_MASK 0x0000ff00 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_MASK 0xff000000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU(x) (0x00000418+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV(x) (0x0000041c+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(x) (0x00000600+((x)*4)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE__SIZE 0x00000040 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 - - -#define NV04_DX6_MULTITEX_TRIANGLE 0x00000055 - -#define NV04_DX6_MULTITEX_TRIANGLE_NOP 0x00000100 -#define NV04_DX6_MULTITEX_TRIANGLE_NOTIFY 0x00000104 -#define NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV04_DX6_MULTITEX_TRIANGLE_DMA_A 0x00000184 -#define NV04_DX6_MULTITEX_TRIANGLE_DMA_B 0x00000188 -#define NV04_DX6_MULTITEX_TRIANGLE_SURFACE 0x0000018c -#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET(x) (0x00000308+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET__SIZE 0x00000002 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT(x) (0x00000310+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT__SIZE 0x00000002 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPU (1 << 27) -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPV (1 << 31) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER(x) (0x00000318+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER__SIZE 0x00000002 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA 0x00000320 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR 0x00000324 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA 0x0000032c -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR 0x00000330 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR 0x00000334 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND 0x00000338 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK 0x0f000000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT 28 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK 0xf0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0 0x0000033c -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_TEST_ENABLE (1 << 12) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ORIGIN (1 << 13) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_SHIFT 14 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_MASK 0x0000c000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_MASK 0x00300000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_DITHER_ENABLE (1 << 22) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_PERSPECTIVE_ENABLE (1 << 23) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_WRITE_ENABLE (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_ENABLE (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_ENABLE (1 << 26) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_ENABLE (1 << 27) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_ENABLE (1 << 28) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_ENABLE (1 << 29) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_SHIFT 30 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_MASK 0xc0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1 0x00000340 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_MASK 0x000000f0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2 0x00000344 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_MASK 0x000000f0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR 0x00000348 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX(x) (0x00000400+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY(x) (0x00000404+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ(x) (0x00000408+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW(x) (0x0000040c+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR(x) (0x00000410+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR(x) (0x00000414+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0(x) (0x00000418+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0(x) (0x0000041c+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1(x) (0x00000420+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1(x) (0x00000424+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE(x) (0x00000540+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE__SIZE 0x00000030 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 - - -#define NV10_TCL_PRIMITIVE_3D 0x00000056 - -#define NV10_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV10_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 -#define NV10_TCL_PRIMITIVE_3D_DISPLAY_LIST 0x0000018c -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 -#define NV10_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 -#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00000218+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00000220+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT 7 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK 0x00000780 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8 0x0000000b -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3 0x0000000e -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT 0x00000012 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT 0x00000013 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8 0x0000001a -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT_2 0x0000001b -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT 0x0000001d -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT 0x00000020 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT (1 << 11) -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_REPEAT 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_SHIFT 28 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MASK 0xf0000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_REPEAT 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00000228+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_MASK 0x00000030 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00000248+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 28 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0xf0000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SHIFT 5 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000268+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SHIFT 5 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR(x) (0x00000270+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM (1 << 14) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS (1 << 15) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT 17 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000280+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM (1 << 14) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS (1 << 15) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT 17 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_SHIFT 27 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_MASK 0x38000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SHIFT 5 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_MASK 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL 0x00000294 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL (1 << 1) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE 0x00000298 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP_2 0x00000802 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP2 0x00000803 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR 0x00000804 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR_2 0x00002601 -#define NV10_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 -#define NV10_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c -#define NV10_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 -#define NV10_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 -#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 -#define NV10_TCL_PRIMITIVE_3D_POINT_SMOOTH_ENABLE 0x0000031c -#define NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_WEIGHT_ENABLE 0x00000328 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_B (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G (1 << 8) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R (1 << 16) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A (1 << 24) -#define NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c -#define NV10_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c -#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV10_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 -#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_R 0x000003a8 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_G 0x000003ac -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_B 0x000003b0 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_A 0x000003b4 -#define NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL 0x000003b8 -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT0 (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT1 (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT2 (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT3 (1 << 6) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT4 (1 << 8) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT5 (1 << 10) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT6 (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT7 (1 << 14) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_FALSE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 -#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE 0x000003e8 -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_POINT_SIZE 0x000003ec -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x00000540+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000580+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000600+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000604+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000608+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x00000680 -#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x00000684 -#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x00000688 -#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R0x000006c4 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G0x000006c8 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B0x000006cc -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X 0x000006e8 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Y 0x000006ec -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Z 0x000006f0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_W 0x000006f4 -#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x000006f8+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00000c04 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00000c08 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00000c18 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x00000c1c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00000c20 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W 0x00000c24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00000c30 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00000c34 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00000c38 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000c40 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000c44 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R 0x00000c50 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G 0x00000c54 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B 0x00000c58 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A 0x00000c5c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R 0x00000c60 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G 0x00000c64 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B 0x00000c68 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x00000c6c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R 0x00000c80 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G 0x00000c84 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B 0x00000c88 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I 0x00000c8c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00000c90 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00000c94 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00000c98 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x00000ca0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x00000ca4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x00000ca8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x00000cac -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x00000cb0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x00000cb4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x00000cb8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x00000cbc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x00000cc0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x00000cc8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x00000ccc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x00000cd0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x00000cd4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x00000cd8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x00000cdc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00000ce0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_WGH_1F 0x00000ce4 -#define NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x00000cec -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE 0x00000cf0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_POS 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS 0x00000d04 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL 0x00000d08 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL 0x00000d0c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00000dfc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END 0x000013fc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001800 - - -#define NV04_CONTEXT_COLOR_KEY 0x00000057 - -#define NV04_CONTEXT_COLOR_KEY_NOP 0x00000100 -#define NV04_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 -#define NV04_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 -#define NV04_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 -#define NV04_CONTEXT_COLOR_KEY_COLOR 0x00000304 - - -#define NV03_CONTEXT_SURFACES_2D 0x00000058 - -#define NV03_CONTEXT_SURFACES_2D_SYNCHRONIZE 0x00000100 -#define NV03_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 -#define NV03_CONTEXT_SURFACES_2D_DMA_SOURCE 0x00000184 -#define NV03_CONTEXT_SURFACES_2D_DMA_DESTIN 0x00000188 -#define NV03_CONTEXT_SURFACES_2D_COLOR_FORMAT 0x00000300 -#define NV03_CONTEXT_SURFACES_2D_PITCH 0x00000304 -#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 -#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff -#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 -#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 -#define NV03_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 -#define NV03_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c - - -#define NV03_CONTEXT_SURFACES_3D 0x0000005a - -#define NV03_CONTEXT_SURFACES_3D_SYNCHRONIZE 0x00000100 -#define NV03_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 -#define NV03_CONTEXT_SURFACES_3D_DMA_SURfACE 0x00000184 -#define NV03_CONTEXT_SURFACES_3D_PITCH 0x00000300 -#define NV03_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x00000304 -#define NV03_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000308 - - -#define NV04_RENDER_SOLID_LINE 0x0000005c - -#define NV04_RENDER_SOLID_LINE_SURFACE 0x00000198 - - -#define NV_RENDER_SOLID_TRIANGLE 0x0000005d - - - -#define NV_RENDER_SOLID_RECTANGLE 0x0000005e - -#define NV_RENDER_SOLID_RECTANGLE_SURFACE 0x00000198 - - -#define NV_IMAGE_BLIT 0x0000005f - -#define NV_IMAGE_BLIT_NOP 0x00000100 -#define NV_IMAGE_BLIT_NOTIFY 0x00000104 -#define NV_IMAGE_BLIT_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_BLIT_COLOR_KEY 0x00000184 -#define NV_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 -#define NV_IMAGE_BLIT_PATTERN 0x0000018c -#define NV_IMAGE_BLIT_ROP 0x00000190 -#define NV_IMAGE_BLIT_SURFACE 0x0000019c -#define NV_IMAGE_BLIT_OPERATION 0x000002fc -#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 -#define NV_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 -#define NV_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 -#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000003 -#define NV_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000004 -#define NV_IMAGE_BLIT_POINT_IN 0x00000300 -#define NV_IMAGE_BLIT_POINT_IN_X_SHIFT 0 -#define NV_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff -#define NV_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 -#define NV_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 -#define NV_IMAGE_BLIT_POINT_OUT 0x00000304 -#define NV_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 -#define NV_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff -#define NV_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 -#define NV_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 -#define NV_IMAGE_BLIT_SIZE 0x00000308 -#define NV_IMAGE_BLIT_SIZE_W_SHIFT 0 -#define NV_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff -#define NV_IMAGE_BLIT_SIZE_H_SHIFT 16 -#define NV_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 - - -#define NV04_INDEXED_IMAGE_FROM_CPU 0x00000060 - -#define NV04_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV04_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV04_INDEXED_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 -#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 -#define NV04_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec -#define NV04_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 -#define NV04_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 -#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 -#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc -#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR 0x00000400 - - -#define NV_IMAGE_FROM_CPU 0x00000061 - - - -#define NV10_CONTEXT_SURFACES_2D 0x00000062 +#define NV_IMAGE_ROP_AND 0x00000013 +#define NV_IMAGE_ROP_AND_NOTIFY 0x00000104 +#define NV_IMAGE_ROP_AND_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_ROP_AND_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_ROP_AND_ROP_INPUT 0x00000204 +#define NV_IMAGE_ROP_AND_IMAGE_INPUT(x) (0x00000208+((x)*4)) +#define NV_IMAGE_ROP_AND_IMAGE_INPUT__SIZE 0x00000002 + + +#define NV_IMAGE_COLOR_KEY 0x00000015 + + + +#define NV01_CONTEXT_COLOR_KEY 0x00000017 + +#define NV01_CONTEXT_COLOR_KEY_NOP 0x00000100 +#define NV01_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 +#define NV01_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A8Y8 0x00000001 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X24Y8 0x00000002 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A1R5G5B5 0x00000003 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X17R5G5B5 0x00000004 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A8R8G8B8 0x00000005 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X8R8G8B8 0x00000006 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A16Y16 0x00000007 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16Y16 0x00000008 +#define NV01_CONTEXT_COLOR_KEY_COLOR 0x00000304 + + +#define NV01_CONTEXT_PATTERN 0x00000018 + +#define NV01_CONTEXT_PATTERN_NOP 0x00000100 +#define NV01_CONTEXT_PATTERN_NOTIFY 0x00000104 +#define NV01_CONTEXT_PATTERN_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_PATTERN_COLOR_FORMAT 0x00000300 +#define NV01_CONTEXT_PATTERN_MONOCHROME_FORMAT 0x00000304 +#define NV01_CONTEXT_PATTERN_SHAPE 0x00000308 +#define NV01_CONTEXT_PATTERN_COLOR(x) (0x00000310+((x)*4)) +#define NV01_CONTEXT_PATTERN_COLOR__SIZE 0x00000002 +#define NV01_CONTEXT_PATTERN_PATTERN(x) (0x00000318+((x)*4)) +#define NV01_CONTEXT_PATTERN_PATTERN__SIZE 0x00000002 + + +#define NV01_CONTEXT_CLIP_RECTANGLE 0x00000019 + +#define NV01_CONTEXT_CLIP_RECTANGLE_NOP 0x00000100 +#define NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY 0x00000104 +#define NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT 0x00000300 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_SHIFT 0 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_MASK 0x0000ffff +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_SHIFT 16 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_MASK 0xffff0000 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE 0x00000304 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_SHIFT 0 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_MASK 0x0000ffff +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_SHIFT 16 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_LINE 0x0000001c + +#define NV01_RENDER_SOLID_LINE_NOP 0x00000100 +#define NV01_RENDER_SOLID_LINE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_LINE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_LINE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_LINE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_LINE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_LINE_ROP 0x0000018c +#define NV01_RENDER_SOLID_LINE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_LINE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_LINE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_LINE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY 0x00000003 +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A8Y8 0x00000001 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X24Y8 0x00000002 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A1R5G5B5 0x00000003 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X17R5G5B5 0x00000004 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A8R8G8B8 0x00000005 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X8R8G8B8 0x00000006 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A16Y16 0x00000007 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16Y16 0x00000008 +#define NV01_RENDER_SOLID_LINE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0(x) (0x00000400+((x)*8)) +#define NV01_RENDER_SOLID_LINE_LINE_POINT0__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1(x) (0x00000404+((x)*8)) +#define NV01_RENDER_SOLID_LINE_LINE_POINT1__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X(x) (0x00000480+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y(x) (0x00000484+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X(x) (0x00000488+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y(x) (0x0000048c+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_POLYLINE(x) (0x00000500+((x)*4)) +#define NV01_RENDER_SOLID_LINE_POLYLINE__SIZE 0x00000020 +#define NV01_RENDER_SOLID_LINE_POLYLINE_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_POLYLINE_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X(x) (0x00000580+((x)*8)) +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y(x) (0x00000584+((x)*8)) +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR(x) (0x00000600+((x)*8)) +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT(x) (0x00000604+((x)*8)) +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_TRIANGLE 0x0000001d + +#define NV01_RENDER_SOLID_TRIANGLE_NOP 0x00000100 +#define NV01_RENDER_SOLID_TRIANGLE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_TRIANGLE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_TRIANGLE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_TRIANGLE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_TRIANGLE_ROP 0x0000018c +#define NV01_RENDER_SOLID_TRIANGLE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_TRIANGLE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY 0x00000003 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_RENDER_SOLID_TRIANGLE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_TRIANGLE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0 0x00000310 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1 0x00000314 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2 0x00000318 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_X 0x00000320 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_Y 0x00000324 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_X 0x00000328 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_Y 0x0000032c +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_X 0x00000330 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_Y 0x00000334 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH(x) (0x00000400+((x)*4)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH__SIZE 0x00000020 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X(x) (0x00000480+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y(x) (0x00000484+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR(x) (0x00000500+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0(x) (0x00000504+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1(x) (0x00000508+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2(x) (0x0000050c+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR(x) (0x00000580+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT(x) (0x00000584+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_RECTANGLE 0x0000001e + +#define NV01_RENDER_SOLID_RECTANGLE_NOP 0x00000100 +#define NV01_RENDER_SOLID_RECTANGLE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_RECTANGLE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_RECTANGLE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_RECTANGLE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_RECTANGLE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_RECTANGLE_ROP 0x0000018c +#define NV01_RENDER_SOLID_RECTANGLE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_RECTANGLE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY 0x00000003 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_RENDER_SOLID_RECTANGLE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_RECTANGLE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT(x) (0x00000400+((x)*8)) +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE__SIZE 0x00000010 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_SHIFT 0 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_MASK 0x0000ffff +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_SHIFT 16 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_MASK 0xffff0000 + + +#define NV01_IMAGE_BLIT 0x0000001f + +#define NV01_IMAGE_BLIT_NOP 0x00000100 +#define NV01_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV01_IMAGE_BLIT_PATCH 0x0000010c +#define NV01_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV01_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV01_IMAGE_BLIT_PATTERN 0x0000018c +#define NV01_IMAGE_BLIT_ROP 0x00000190 +#define NV01_IMAGE_BLIT_BETA1 0x00000194 +#define NV01_IMAGE_BLIT_SURFACE 0x0000019c +#define NV01_IMAGE_BLIT_OPERATION 0x000002fc +#define NV01_IMAGE_BLIT_IMAGE_INPUT 0x00000204 +#define NV01_IMAGE_BLIT_POINT_IN 0x00000300 +#define NV01_IMAGE_BLIT_POINT_IN_X_SHIFT 0 +#define NV01_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 +#define NV01_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 +#define NV01_IMAGE_BLIT_POINT_OUT 0x00000304 +#define NV01_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 +#define NV01_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 +#define NV01_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 +#define NV01_IMAGE_BLIT_SIZE 0x00000308 +#define NV01_IMAGE_BLIT_SIZE_W_SHIFT 0 +#define NV01_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_SIZE_H_SHIFT 16 +#define NV01_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 + + +#define NV01_IMAGE_FROM_CPU 0x00000021 + +#define NV01_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV01_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV01_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV01_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV01_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 +#define NV01_IMAGE_FROM_CPU_PATTERN 0x0000018c +#define NV01_IMAGE_FROM_CPU_ROP 0x00000190 +#define NV01_IMAGE_FROM_CPU_BETA1 0x00000194 +#define NV01_IMAGE_FROM_CPU_SURFACE 0x00000198 +#define NV01_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_IMAGE_FROM_CPU_OPERATION_ROP_AND 0x00000001 +#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_AND 0x00000002 +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY 0x00000003 +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_Y8 0x00000001 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A1R5G5B5 0x00000002 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X1R5G5B5 0x00000003 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A8R8G8B8 0x00000004 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X8R8G8B8 0x00000005 +#define NV01_IMAGE_FROM_CPU_POINT 0x00000304 +#define NV01_IMAGE_FROM_CPU_POINT_X_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_POINT_X_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_POINT_Y_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_POINT_Y_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_SIZE_IN 0x0000030c +#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV01_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 + + +#define NV01_NULL 0x00000030 + + + +#define NV03_STRETCHED_IMAGE_FROM_CPU 0x00000036 + +#define NV03_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV03_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV03_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV03_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV03_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 +#define NV03_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c +#define NV03_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000194 +#define NV03_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_DX_DU 0x00000308 +#define NV03_STRETCHED_IMAGE_FROM_CPU_DY_DV 0x0000030c +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT 0x00000310 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE 0x00000314 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 + + +#define NV03_SCALED_IMAGE_FROM_MEMORY 0x00000037 + +#define NV03_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 +#define NV03_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 +#define NV03_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 +#define NV03_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c +#define NV03_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 +#define NV03_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000194 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY 0x00000003 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000005 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT 0x00000310 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE 0x00000314 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DU_DX 0x00000318 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DV_DY 0x0000031c +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE 0x00000400 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT 0x00000404 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_MASK 0x00ff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CENTER 0x00010000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CORNER 0x00020000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_SHIFT 24 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_MASK 0xff000000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_OFFSET 0x00000408 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT 0x0000040c +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_MASK 0xffff0000 + + +#define NV04_DVD_SUBPICTURE 0x00000038 + +#define NV04_DVD_SUBPICTURE_NOP 0x00000100 +#define NV04_DVD_SUBPICTURE_NOTIFY 0x00000104 +#define NV04_DVD_SUBPICTURE_WAIT_FOR_IDLE 0x00000108 +#define NV04_DVD_SUBPICTURE_DMA_NOTIFY 0x00000180 +#define NV04_DVD_SUBPICTURE_DMA_OVERLAY 0x00000184 +#define NV04_DVD_SUBPICTURE_DMA_IMAGEIN 0x00000188 +#define NV04_DVD_SUBPICTURE_DMA_IMAGEOUT 0x0000018c +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT 0x00000300 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE 0x00000304 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT 0x00000308 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_OFFSET 0x0000030c +#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DU_DX 0x00000310 +#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DV_DY 0x00000314 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE 0x00000318 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT 0x0000031c +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEIN_OFFSET 0x00000320 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT 0x00000324 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DU_DX 0x00000328 +#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DV_DY 0x0000032c +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE 0x00000330 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT 0x00000334 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_OFFSET 0x00000338 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT 0x0000033c +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK 0xffff0000 + + +#define NV_MEMORY_TO_MEMORY_FORMAT 0x00000039 + +#define NV_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 +#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 +#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c +#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 +#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 +#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 +#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c +#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 +#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 +#define NV_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 + + +#define NV01_MEMORY_LOCAL_BANKED 0x0000003d + + + +#define NV01_MAPPING_SYSTEM 0x0000003e + + + +#define NV03_MEMORY_LOCAL_CURSOR 0x0000003f + + + +#define NV01_MEMORY_LOCAL_LINEAR 0x00000040 + + + +#define NV01_MAPPING_LOCAL 0x00000041 + + + +#define NV04_CONTEXT_SURFACES_2D 0x00000042 + +#define NV04_CONTEXT_SURFACES_2D_NOP 0x00000100 +#define NV04_CONTEXT_SURFACES_2D_NOTIFY 0x00000104 +#define NV04_CONTEXT_SURFACES_2D_PM_TRIGGER 0x00000140 +#define NV04_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE 0x00000184 +#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_DESTIN 0x00000188 +#define NV04_CONTEXT_SURFACES_2D_FORMAT 0x00000300 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y8 0x00000001 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_Z1R5G5B5 0x00000002 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_X1R5G5B5 0x00000003 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5 0x00000004 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y16 0x00000005 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_Z8R8G8B8 0x00000006 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_X8R8G8B8 0x00000007 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_X1A7R8G8B8 0x00000009 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8 0x0000000a +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y32 0x0000000b +#define NV04_CONTEXT_SURFACES_2D_PITCH 0x00000304 +#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 +#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 +#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 +#define NV04_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c + + +#define NV03_CONTEXT_ROP 0x00000043 + +#define NV03_CONTEXT_ROP_NOP 0x00000100 +#define NV03_CONTEXT_ROP_NOTIFY 0x00000104 +#define NV03_CONTEXT_ROP_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_ROP_ROP 0x00000300 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SHIFT 0 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_MASK 0x0000000f +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_CLEAR 0x00000000 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOR 0x00000001 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_INVERTED 0x00000002 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY_INVERTED 0x00000003 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_REVERSE 0x00000004 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_INVERT 0x00000005 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_XOR 0x00000006 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NAND 0x00000007 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND 0x00000008 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_EQUI 0x00000009 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOOP 0x0000000a +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_INVERTED 0x0000000b +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY 0x0000000c +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_REVERSE 0x0000000d +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR 0x0000000e +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SET 0x0000000f +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SHIFT 4 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_MASK 0x000000f0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_CLEAR 0x00000000 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOR 0x00000010 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_INVERTED 0x00000020 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY_INVERTED 0x00000030 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_REVERSE 0x00000040 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_INVERT 0x00000050 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_XOR 0x00000060 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NAND 0x00000070 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND 0x00000080 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_EQUI 0x00000090 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOOP 0x000000a0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_INVERTED 0x000000b0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY 0x000000c0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_REVERSE 0x000000d0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR 0x000000e0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SET 0x000000f0 + + +#define NV04_IMAGE_PATTERN 0x00000044 + +#define NV04_IMAGE_PATTERN_NOP 0x00000100 +#define NV04_IMAGE_PATTERN_NOTIFY 0x00000104 +#define NV04_IMAGE_PATTERN_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT 0x00000300 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A16R5G6B5 0x00000001 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_X16A1R5G5B5 0x00000002 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT 0x00000304 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_CGA6 0x00000001 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_LE 0x00000002 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE 0x00000308 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_8X8 0x00000000 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_64X1 0x00000001 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_1X64 0x00000002 +#define NV04_IMAGE_PATTERN_PATTERN_SELECT 0x0000030c +#define NV04_IMAGE_PATTERN_PATTERN_SELECT_MONO 0x00000001 +#define NV04_IMAGE_PATTERN_PATTERN_SELECT_COLOR 0x00000002 +#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR0 0x00000310 +#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR1 0x00000314 +#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0 0x00000318 +#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1 0x0000031c +#define NV04_IMAGE_PATTERN_PATTERN_Y8(x) (0x00000400+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_Y8__SIZE 0x00000010 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_MASK 0x000000ff +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_SHIFT 8 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_MASK 0x0000ff00 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_MASK 0x00ff0000 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_SHIFT 24 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_MASK 0xff000000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5(x) (0x00000500+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5__SIZE 0x00000020 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_MASK 0x0000001f +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_SHIFT 5 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_MASK 0x000007e0 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_SHIFT 11 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_MASK 0x0000f800 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_MASK 0x001f0000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_SHIFT 21 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_MASK 0x07e00000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_SHIFT 27 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_MASK 0xf8000000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5(x) (0x00000600+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5__SIZE 0x00000020 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_MASK 0x0000001f +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_SHIFT 5 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_MASK 0x000003e0 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_SHIFT 10 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_MASK 0x00007c00 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_MASK 0x001f0000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_SHIFT 21 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_MASK 0x03e00000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_SHIFT 26 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_MASK 0x7c000000 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8(x) (0x00000700+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8__SIZE 0x00000040 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_MASK 0x000000ff +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_SHIFT 8 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_MASK 0x0000ff00 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_MASK 0x00ff0000 + + +#define NV03_VIDEO_LUT_CURSOR_DAC 0x00000046 + +#define NV03_VIDEO_LUT_CURSOR_DAC_SYNCHRONIZE 0x00000100 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_IMAGE 0x00000104 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_CURSOR 0x00000108 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_DAC 0x0000010c +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_NOTIFY 0x00000180 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE(x) (0x00000184+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT(x) (0x0000018c+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR(x) (0x00000194+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_GET 0x000002fc +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET(x) (0x00000300+((x)*8)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT(x) (0x00000304+((x)*8)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET(x) (0x00000340+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT(x) (0x00000344+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT(x) (0x00000348+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A 0x00000358 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE(x) (0x00000380+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC(x) (0x00000384+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC(x) (0x00000388+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE(x) (0x0000038c+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_PIXEL_CLOCK 0x000003a0 + + +#define NV03_DX3_TEXTURED_TRIANGLE 0x00000048 + +#define NV03_DX3_TEXTURED_TRIANGLE_NOP 0x00000100 +#define NV03_DX3_TEXTURED_TRIANGLE_NOTIFY 0x00000104 +#define NV03_DX3_TEXTURED_TRIANGLE_PATCH 0x0000010c +#define NV03_DX3_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV03_DX3_TEXTURED_TRIANGLE_DMA_TEXTURE 0x00000184 +#define NV03_DX3_TEXTURED_TRIANGLE_CLIP_RECTANGLE 0x00000188 +#define NV03_DX3_TEXTURED_TRIANGLE_SURFACE 0x0000018c +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_OFFSET 0x00000304 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT 0x00000308 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_MASK 0x0000ffff +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_MASK 0x0f000000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_SHIFT 28 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_MASK 0xf0000000 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER 0x0000030c +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_MASK 0x0000001f +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_MASK 0x00001f00 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_MASK 0x00ff0000 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR 0x00000310 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_MASK 0x000000ff +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_MASK 0x0000ff00 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_MASK 0x00ff0000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT 0x00000314 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_MASK 0x0000000f +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_SHIFT 4 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_MASK 0x00000030 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_SHIFT 6 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_MASK 0x000000c0 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_MASK 0x00000f00 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_SHIFT 12 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_MASK 0x00007000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_PERSPECTIVE_ENABLE (1 << 15) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_MASK 0x07000000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_SHIFT 27 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_MASK 0x18000000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_BETA (1 << 29) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_DST_BLEND (1 << 30) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SRC_BLEND (1 << 31) +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL 0x00000318 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_MASK 0x000000ff +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_MASK 0xffffff00 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR(x) (0x00001000+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_MASK 0x0000000f +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_SHIFT 4 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_MASK 0x000000f0 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_MASK 0x00000f00 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_SHIFT 12 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_MASK 0x0000f000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_MASK 0xff000000 +#define NV03_DX3_TEXTURED_TRIANGLE_COLOR(x) (0x00001004+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_COLOR__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_X(x) (0x00001008+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_X__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_Y(x) (0x0000100c+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_Y__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_Z(x) (0x00001010+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_Z__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_M(x) (0x00001014+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_M__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_U(x) (0x00001018+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_U__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_V(x) (0x0000101c+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_V__SIZE 0x00000040 + + +#define NV04_GDI_RECTANGLE_TEXT 0x0000004a + +#define NV04_GDI_RECTANGLE_TEXT_NOP 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 +#define NV04_GDI_RECTANGLE_TEXT_PATCH 0x0000010c +#define NV04_GDI_RECTANGLE_TEXT_PM_TRIGGER 0x00000140 +#define NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 +#define NV04_GDI_RECTANGLE_TEXT_DMA_FONTS 0x00000184 +#define NV04_GDI_RECTANGLE_TEXT_PATTERN 0x00000188 +#define NV04_GDI_RECTANGLE_TEXT_ROP 0x0000018c +#define NV04_GDI_RECTANGLE_TEXT_BETA1 0x00000190 +#define NV04_GDI_RECTANGLE_TEXT_BETA4 0x00000194 +#define NV04_GDI_RECTANGLE_TEXT_SURFACE 0x00000198 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_AND 0x00000000 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_ROP_AND 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_AND 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY 0x00000003 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_PREMULT 0x00000005 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_X16A1R5G5B5 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_CGA6 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(x) (0x00000400+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0 0x000005f4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1 0x000005f8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_B 0x000005fc +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0(x) (0x00000600+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1(x) (0x00000604+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x000007ec +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x000007f0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_C 0x000007f4 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C 0x000007f8 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C 0x000007fc +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000800+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000080 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x00000be4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x00000be8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR0_E 0x00000bec +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_E 0x00000bf0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x00000bf4 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x00000bf8 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E 0x00000bfc +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00000c00+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000080 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F 0x00000ff0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_MASK 0x0fffffff +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_SHIFT 28 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_MASK 0xf0000000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0 0x00000ff4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1 0x00000ff8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_F 0x00000ffc +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F(x) (0x00001000+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F__SIZE 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_MASK 0x000000ff +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_SHIFT 8 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_MASK 0x000fff00 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_SHIFT 20 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_MASK 0xfff00000 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G 0x000017f0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_MASK 0x0fffffff +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_SHIFT 28 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_MASK 0xf0000000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0 0x000017f4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1 0x000017f8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_G 0x000017fc +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT(x) (0x00001800+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT__SIZE 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX(x) (0x00001804+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX__SIZE 0x00000100 + + +#define NV03_GDI_RECTANGLE_TEXT 0x0000004b + +#define NV03_GDI_RECTANGLE_TEXT_NOP 0x00000100 +#define NV03_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 +#define NV03_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 +#define NV03_GDI_RECTANGLE_TEXT_PATTERN 0x00000184 +#define NV03_GDI_RECTANGLE_TEXT_ROP 0x00000188 +#define NV03_GDI_RECTANGLE_TEXT_BETA1 0x0000018c +#define NV03_GDI_RECTANGLE_TEXT_SURFACE 0x00000190 +#define NV03_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc +#define NV03_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT 0x00000400 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE 0x00000404 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B 0x000007f4 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B 0x000007f8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_B 0x000007fc +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0 0x00000800 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1 0x00000804 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x00000bec +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x00000bf0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_C 0x00000bf4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C 0x00000bf8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C 0x00000bfc +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000c00+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000020 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0 0x00000fe8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1 0x00000fec +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_D 0x00000ff0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D 0x00000ff4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D 0x00000ff8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D 0x00000ffc +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D(x) (0x00001000+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D__SIZE 0x00000020 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x000013e4 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x000013e8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR0_E 0x000013ec +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_E 0x000013f0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x000013f4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x000013f8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E 0x000013fc +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00001400+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000020 + + +#define NV04_SWIZZLED_SURFACE 0x00000052 + +#define NV04_SWIZZLED_SURFACE_NOP 0x00000100 +#define NV04_SWIZZLED_SURFACE_NOTIFY 0x00000104 +#define NV04_SWIZZLED_SURFACE_DMA_NOTIFY 0x00000180 +#define NV04_SWIZZLED_SURFACE_DMA_IMAGE 0x00000184 +#define NV04_SWIZZLED_SURFACE_FORMAT 0x00000300 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_SHIFT 0 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK 0x000000ff +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y8 0x00000001 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000002 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000003 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_R5G6B5 0x00000004 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y16 0x00000005 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000006 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000007 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000009 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8 0x0000000a +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y32 0x0000000b +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_MASK 0xff000000 +#define NV04_SWIZZLED_SURFACE_OFFSET 0x00000304 + + +#define NV04_CONTEXT_SURFACES_3D 0x00000053 + +#define NV04_CONTEXT_SURFACES_3D_NOP 0x00000100 +#define NV04_CONTEXT_SURFACES_3D_NOTIFY 0x00000104 +#define NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_SURFACES_3D_DMA_COLOR 0x00000184 +#define NV04_CONTEXT_SURFACES_3D_DMA_ZETA 0x00000188 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL 0x000002f8 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL 0x000002fc +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_FORMAT 0x00000300 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK 0x000000ff +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000001 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000002 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000004 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000005 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000006 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000007 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT 8 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK 0x0000ff00 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_PITCH 0x00000100 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SWIZZLE 0x00000200 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_MASK 0xff000000 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE 0x00000304 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_PITCH 0x00000308 +#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x0000030c +#define NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000310 + + +#define NV04_DX5_TEXTURED_TRIANGLE 0x00000054 + +#define NV04_DX5_TEXTURED_TRIANGLE_NOP 0x00000100 +#define NV04_DX5_TEXTURED_TRIANGLE_NOTIFY 0x00000104 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_A 0x00000184 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_B 0x00000188 +#define NV04_DX5_TEXTURED_TRIANGLE_SURFACE 0x0000018c +#define NV04_DX5_TEXTURED_TRIANGLE_COLORKEY 0x00000300 +#define NV04_DX5_TEXTURED_TRIANGLE_OFFSET 0x00000304 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT 0x00000308 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_MASK 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_SHIFT 2 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_MASK 0x0000000c +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CENTER 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER 0x00000020 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CENTER 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER 0x00000080 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_Y8 0x00000100 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A1R5G5B5 0x00000200 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X1R5G5B5 0x00000300 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A4R4G4B4 0x00000400 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_R5G6B5 0x00000500 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A8R8G8B8 0x00000600 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X8R8G8B8 0x00000700 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT 0x01000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT 0x02000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE 0x03000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER 0x04000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP 0x05000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPU (1 << 27) +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_REPEAT 0x10000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MIRRORED_REPEAT 0x20000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE 0x30000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_BORDER 0x40000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP 0x50000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPV (1 << 31) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER 0x0000030c +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST 0x01000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR 0x02000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST 0x10000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR 0x20000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x30000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x40000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x50000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x60000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND 0x00000310 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_MASK 0x0000000f +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD 0x00000080 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_PHONG 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK 0x0f000000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK 0xf0000000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL 0x00000314 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE (1 << 12) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN (1 << 13) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT 14 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_MASK 0x0000c000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_MASK 0x00300000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE (1 << 22) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE (1 << 23) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_MASK 0x3f000000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT 30 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_MASK 0xc0000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR 0x00000318 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(x) (0x00000400+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY(x) (0x00000404+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ(x) (0x00000408+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW(x) (0x0000040c+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR(x) (0x00000410+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR(x) (0x00000414+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU(x) (0x00000418+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV(x) (0x0000041c+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(x) (0x00000600+((x)*4)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE__SIZE 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 + + +#define NV04_DX6_MULTITEX_TRIANGLE 0x00000055 + +#define NV04_DX6_MULTITEX_TRIANGLE_NOP 0x00000100 +#define NV04_DX6_MULTITEX_TRIANGLE_NOTIFY 0x00000104 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_A 0x00000184 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_B 0x00000188 +#define NV04_DX6_MULTITEX_TRIANGLE_SURFACE 0x0000018c +#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET(x) (0x00000308+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT(x) (0x00000310+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPU (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPV (1 << 31) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER(x) (0x00000318+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA 0x00000320 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR 0x00000324 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA 0x0000032c +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR 0x00000330 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR 0x00000334 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND 0x00000338 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK 0x0f000000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK 0xf0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0 0x0000033c +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_TEST_ENABLE (1 << 12) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ORIGIN (1 << 13) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_SHIFT 14 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_MASK 0x0000c000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_MASK 0x00300000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_DITHER_ENABLE (1 << 22) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_PERSPECTIVE_ENABLE (1 << 23) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_WRITE_ENABLE (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_ENABLE (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_ENABLE (1 << 26) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_ENABLE (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_ENABLE (1 << 28) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_ENABLE (1 << 29) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_SHIFT 30 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_MASK 0xc0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1 0x00000340 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2 0x00000344 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR 0x00000348 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX(x) (0x00000400+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY(x) (0x00000404+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ(x) (0x00000408+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW(x) (0x0000040c+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR(x) (0x00000410+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR(x) (0x00000414+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0(x) (0x00000418+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0(x) (0x0000041c+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1(x) (0x00000420+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1(x) (0x00000424+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE(x) (0x00000540+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE__SIZE 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 + + +#define NV10_TCL_PRIMITIVE_3D 0x00000056 + +#define NV10_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV10_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 +#define NV10_TCL_PRIMITIVE_3D_DISPLAY_LIST 0x0000018c +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 +#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 +#define NV10_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 +#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00000218+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00000220+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT 7 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK 0x00000780 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8 0x00000080 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT 0x00000180 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_X8R8G8B8 0x00000380 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8 0x00000580 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1 0x00000600 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3 0x00000700 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5 0x00000780 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT 0x00000980 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT2 0x00000d80 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT 0x00000e80 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT 0x00001000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16 0x00001900 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16_RECT 0x00001a80 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00002500 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00002580 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00002600 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT (1 << 11) +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_REPEAT 0x01000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x02000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x03000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x04000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP 0x05000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_SHIFT 28 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MASK 0xf0000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_REPEAT 0x10000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x20000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x30000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x40000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP 0x50000000 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00000228+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_MASK 0x00000030 +#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00000248+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x01000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x02000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 28 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0xf0000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x10000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x20000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x30000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x40000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x50000000 +#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x60000000 +#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SHIFT 5 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_FOG 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_FOG 0x00030000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_FOG 0x03000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000268+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SHIFT 5 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_FOG 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_FOG 0x00030000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_FOG 0x03000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR(x) (0x00000270+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000030 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000080 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000090 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x000000c0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x000000d0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM (1 << 14) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS (1 << 15) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT 17 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00020000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00040000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000280+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000030 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000080 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000090 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x000000c0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x000000d0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM (1 << 14) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS (1 << 15) +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT 17 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00020000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00040000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_SHIFT 27 +#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_MASK 0x38000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_FOG 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SHIFT 5 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_MASK 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_FOG 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_FOG 0x00030000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_FOG 0x03000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_MASK 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_FOG 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_NV 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE1_NV 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SHIFT 13 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_MASK 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_MASK 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_FOG 0x00030000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_NV 0x000c0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE1_NV 0x000d0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SHIFT 21 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_MASK 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_MASK 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_FOG 0x03000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_NV 0x0c000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE1_NV 0x0d000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SHIFT 29 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_MASK 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL 0x00000294 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL (1 << 1) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE 0x00000298 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP 0x00000800 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP_2 0x00000802 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP2 0x00000803 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR 0x00000804 +#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR_2 0x00002601 +#define NV10_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 +#define NV10_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c +#define NV10_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 +#define NV10_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 +#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV10_TCL_PRIMITIVE_3D_POINT_SMOOTH_ENABLE 0x0000031c +#define NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_WEIGHT_ENABLE 0x00000328 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_B (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G (1 << 8) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R (1 << 16) +#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A (1 << 24) +#define NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c +#define NV10_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c +#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV10_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 +#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 +#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_R 0x000003a8 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_G 0x000003ac +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_B 0x000003b0 +#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_A 0x000003b4 +#define NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL 0x000003b8 +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT0 (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT1 (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT2 (1 << 4) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT3 (1 << 6) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT4 (1 << 8) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT5 (1 << 10) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT6 (1 << 12) +#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT7 (1 << 14) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_FALSE 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 +#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE 0x000003e8 +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) +#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) +#define NV10_TCL_PRIMITIVE_3D_POINT_SIZE 0x000003ec +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x00000540+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000580+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000600+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000604+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000608+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) +#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x00000680 +#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x00000684 +#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x00000688 +#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000006c4 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000006c8 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000006cc +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X 0x000006e8 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Y 0x000006ec +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Z 0x000006f0 +#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_W 0x000006f4 +#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x000006f8+((x)*4)) +#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) +#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00000c00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00000c04 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00000c08 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00000c18 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x00000c1c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00000c20 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W 0x00000c24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00000c30 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00000c34 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00000c38 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000c40 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000c44 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R 0x00000c50 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G 0x00000c54 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B 0x00000c58 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A 0x00000c5c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R 0x00000c60 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G 0x00000c64 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B 0x00000c68 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x00000c6c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R 0x00000c80 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G 0x00000c84 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B 0x00000c88 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I 0x00000c8c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK 0x000000ff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK 0x00ff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00000c90 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00000c94 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00000c98 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x00000ca0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x00000ca4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x00000ca8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x00000cac +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x00000cb0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x00000cb4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x00000cb8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x00000cbc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x00000cc0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x00000cc8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x00000ccc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x00000cd0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x00000cd4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x00000cd8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x00000cdc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00000ce0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_WGH_1F 0x00000ce4 +#define NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x00000cec +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE 0x00000cf0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_POS 0x00000d00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS 0x00000d04 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL 0x00000d08 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL 0x00000d0c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00000dfc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX 0x00000e00 +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END 0x000013fc +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 +#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001800 + + +#define NV04_CONTEXT_COLOR_KEY 0x00000057 + + + +#define NV03_CONTEXT_SURFACES_2D 0x00000058 + +#define NV03_CONTEXT_SURFACES_2D_SYNCHRONIZE 0x00000100 +#define NV03_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_SURFACES_2D_DMA_SOURCE 0x00000184 +#define NV03_CONTEXT_SURFACES_2D_DMA_DESTIN 0x00000188 +#define NV03_CONTEXT_SURFACES_2D_COLOR_FORMAT 0x00000300 +#define NV03_CONTEXT_SURFACES_2D_PITCH 0x00000304 +#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 +#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff +#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 +#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 +#define NV03_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 +#define NV03_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c + + +#define NV03_CONTEXT_SURFACES_3D 0x0000005a + +#define NV03_CONTEXT_SURFACES_3D_SYNCHRONIZE 0x00000100 +#define NV03_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_SURFACES_3D_DMA_SURFACE 0x00000184 +#define NV03_CONTEXT_SURFACES_3D_PITCH 0x00000300 +#define NV03_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x00000304 +#define NV03_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000308 + + +#define NV04_RENDER_SOLID_LINE 0x0000005c + +#define NV04_RENDER_SOLID_LINE_SURFACE 0x00000198 + + +#define NV04_RENDER_SOLID_TRIANGLE 0x0000005d + + + +#define NV04_RENDER_SOLID_RECTANGLE 0x0000005e + +#define NV04_RENDER_SOLID_RECTANGLE_SURFACE 0x00000198 -#define NV05_SCALED_IMAGE_FROM_MEMORY 0x00000063 +#define NV_IMAGE_BLIT 0x0000005f -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 +#define NV_IMAGE_BLIT_NOP 0x00000100 +#define NV_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV_IMAGE_BLIT_PATTERN 0x0000018c +#define NV_IMAGE_BLIT_ROP 0x00000190 +#define NV_IMAGE_BLIT_BETA4 0x00000198 +#define NV_IMAGE_BLIT_SURFACE 0x0000019c +#define NV_IMAGE_BLIT_OPERATION 0x000002fc +#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 +#define NV_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 +#define NV_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 +#define NV_IMAGE_BLIT_OPERATION_SRCCOPY 0x00000003 +#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000005 -#define NV01_IMAGE_SRCCOPY_AND 0x00000064 +#define NV04_INDEXED_IMAGE_FROM_CPU 0x00000060 -#define NV01_IMAGE_SRCCOPY_AND_NOTIFY 0x00000104 -#define NV01_IMAGE_SRCCOPY_AND_DMA_NOTIFY 0x00000180 -#define NV01_IMAGE_SRCCOPY_AND_IMAGE_OUTPUT 0x00000200 -#define NV01_IMAGE_SRCCOPY_AND_IMAGE_INPUT 0x00000204 +#define NV04_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV04_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV04_INDEXED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 +#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 +#define NV04_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec +#define NV04_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 +#define NV04_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 +#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 +#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc +#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR 0x00000400 -#define NV05_INDEXED_IMAGE_FROM_CPU 0x00000064 +#define NV_IMAGE_FROM_CPU 0x00000061 -#define NV05_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV05_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV05_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV05_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 -#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_KEY 0x00000188 -#define NV05_INDEXED_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x0000018c -#define NV05_INDEXED_IMAGE_FROM_CPU_PATTERN 0x00000190 -#define NV05_INDEXED_IMAGE_FROM_CPU_ROP 0x00000194 -#define NV05_INDEXED_IMAGE_FROM_CPU_BETA1 0x00000198 -#define NV05_INDEXED_IMAGE_FROM_CPU_BETA4 0x0000019c -#define NV05_INDEXED_IMAGE_FROM_CPU_SURFACE 0x000001a0 -#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000003e0 -#define NV05_INDEXED_IMAGE_FROM_CPU_OPERATION 0x000003e4 -#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 -#define NV05_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec -#define NV05_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 -#define NV05_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 -#define NV05_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 -#define NV05_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc -#define NV05_INDEXED_IMAGE_FROM_CPU_INDICES 0x00000400 - - -#define NV05_IMAGE_FROM_CPU 0x00000065 - -#define NV05_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV05_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV05_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV05_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 -#define NV05_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 -#define NV05_IMAGE_FROM_CPU_PATTERN 0x0000018c -#define NV05_IMAGE_FROM_CPU_ROP 0x00000190 -#define NV05_IMAGE_FROM_CPU_BETA1 0x00000194 -#define NV05_IMAGE_FROM_CPU_BETA4 0x00000198 -#define NV05_IMAGE_FROM_CPU_SURFACE 0x0000019c -#define NV05_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 -#define NV05_IMAGE_FROM_CPU_OPERATION 0x000002fc -#define NV05_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV05_IMAGE_FROM_CPU_POINT 0x00000304 -#define NV05_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 -#define NV05_IMAGE_FROM_CPU_SIZE_IN 0x0000030c -#define NV05_IMAGE_FROM_CPU_COLOR 0x00000400 - - -#define NV05_STRETCHED_IMAGE_FROM_CPU 0x00000066 - -#define NV05_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV05_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV05_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 -#define NV05_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 -#define NV05_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c -#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 -#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA4 0x00000194 -#define NV05_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000198 -#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 -#define NV05_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc -#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV05_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 -#define NV05_STRETCHED_IMAGE_FROM_CPU_DELTA_DX_DU 0x00000308 -#define NV05_STRETCHED_IMAGE_FROM_CPU_DELTA_DY_DV 0x0000030c -#define NV05_STRETCHED_IMAGE_FROM_CPU_CLIP_0 0x00000310 -#define NV05_STRETCHED_IMAGE_FROM_CPU_CLIP_1 0x00000314 -#define NV05_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 -#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR 0x00000400 +#define NV_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV_IMAGE_FROM_CPU_SURFACE 0x0000019c -#define NV04_IMAGE_BLEND_PREMULT 0x00000067 +#define NV10_CONTEXT_SURFACES_2D 0x00000062 -#define NV04_IMAGE_BLEND_PREMULT_NOP 0x00000100 -#define NV04_IMAGE_BLEND_PREMULT_NOTIFY 0x00000104 -#define NV04_IMAGE_BLEND_PREMULT_DMA_NOTIFY 0x00000180 -#define NV04_IMAGE_BLEND_PREMULT_IMAGE_OUTPUT 0x00000200 -#define NV04_IMAGE_BLEND_PREMULT_BETA_INPUT 0x00000204 -#define NV04_IMAGE_BLEND_PREMULT_IMAGE_INPUT 0x00000208 -#define NV03_CHANNEL_PIO 0x0000006a +#define NV05_SCALED_IMAGE_FROM_MEMORY 0x00000063 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 -#define NV03_CHANNEL_DMA 0x0000006b +#define NV01_IMAGE_SRCCOPY_AND 0x00000064 +#define NV01_IMAGE_SRCCOPY_AND_NOTIFY 0x00000104 +#define NV01_IMAGE_SRCCOPY_AND_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_SRCCOPY_AND_IMAGE_OUTPUT 0x00000200 +#define NV01_IMAGE_SRCCOPY_AND_IMAGE_INPUT 0x00000204 -#define NV04_BETA_SOLID 0x00000072 +#define NV05_INDEXED_IMAGE_FROM_CPU 0x00000064 -#define NV04_BETA_SOLID_NOP 0x00000100 -#define NV04_BETA_SOLID_NOTIFY 0x00000104 -#define NV04_BETA_SOLID_DMA_NOTIFY 0x00000180 -#define NV04_BETA_SOLID_BETA_OUTPUT 0x00000200 -#define NV04_BETA_SOLID_BETA_FACTOR 0x00000300 +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_KEY 0x00000188 +#define NV05_INDEXED_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x0000018c +#define NV05_INDEXED_IMAGE_FROM_CPU_PATTERN 0x00000190 +#define NV05_INDEXED_IMAGE_FROM_CPU_ROP 0x00000194 +#define NV05_INDEXED_IMAGE_FROM_CPU_BETA1 0x00000198 +#define NV05_INDEXED_IMAGE_FROM_CPU_BETA4 0x0000019c +#define NV05_INDEXED_IMAGE_FROM_CPU_SURFACE 0x000001a0 +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000003e0 +#define NV05_INDEXED_IMAGE_FROM_CPU_OPERATION 0x000003e4 +#define NV05_INDEXED_IMAGE_FROM_CPU_INDICES 0x00000400 -#define NV_STRETCHED_IMAGE_FROM_CPU 0x00000076 +#define NV05_IMAGE_FROM_CPU 0x00000065 -#define NV_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 -#define NV_STRETCHED_IMAGE_FROM_CPU_DELTA_DU_DX 0x00000308 -#define NV_STRETCHED_IMAGE_FROM_CPU_DELTA_DV_DY 0x0000030c -#define NV_STRETCHED_IMAGE_FROM_CPU_CLIP_0 0x00000310 -#define NV_STRETCHED_IMAGE_FROM_CPU_CLIP_1 0x00000314 -#define NV_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 -#define NV_STRETCHED_IMAGE_FROM_CPU_COLOR 0x00000400 +#define NV05_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV05_IMAGE_FROM_CPU_SURFACE 0x0000019c -#define NV04_SCALED_IMAGE_FROM_MEMORY 0x00000077 +#define NV05_STRETCHED_IMAGE_FROM_CPU 0x00000066 -#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA4 0x00000194 -#define NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000198 +#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA4 0x00000194 +#define NV05_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000198 +#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 -#define NV10_TEXTURE_FROM_CPU 0x0000007b +#define NV04_IMAGE_BLEND_PREMULT 0x00000067 -#define NV10_TEXTURE_FROM_CPU_NOP 0x00000100 -#define NV10_TEXTURE_FROM_CPU_NOTIFY 0x00000104 -#define NV10_TEXTURE_FROM_CPU_WAIT_FOR_IDLE 0x00000108 -#define NV10_TEXTURE_FROM_CPU_PM_TRIGGER 0x00000140 -#define NV10_TEXTURE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV10_TEXTURE_FROM_CPU_SURFACE 0x00000184 -#define NV10_TEXTURE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV10_TEXTURE_FROM_CPU_POINT 0x00000304 -#define NV10_TEXTURE_FROM_CPU_POINT_X_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_POINT_X_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_POINT_Y_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_POINT_Y_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_SIZE 0x00000308 -#define NV10_TEXTURE_FROM_CPU_SIZE_W_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_SIZE_W_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_SIZE_H_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_SIZE_H_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL 0x0000030c -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL 0x00000310 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) -#define NV10_TEXTURE_FROM_CPU_COLOR__SIZE 0x00000100 - - -#define NV10_VIDEO_DISPLAY 0x0000007c - - - -#define NV10_DVD_SUBPICTURE 0x00000088 - - - -#define NV10_SCALED_IMAGE_FROM_MEMORY 0x00000089 - -#define NV10_SCALED_IMAGE_FROM_MEMORY_WAIT_FOR_IDLE 0x00000108 - - -#define NV10_IMAGE_FROM_CPU 0x0000008a - -#define NV10_IMAGE_FROM_CPU_BETA4 0x00000198 -#define NV10_IMAGE_FROM_CPU_SURFACE 0x0000019c -#define NV10_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 - - -#define NV10_CONTEXT_SURFACES_3D 0x00000093 - - - -#define NV10_DX5_TEXTURE_TRIANGLE 0x00000094 - - - -#define NV10_DX6_MULTI_TEXTURE_TRIANGLE 0x00000095 - - - -#define NV11_TCL_PRIMITIVE_3D 0x00000096 - -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000d40 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000d44 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f - - -#define NV20_TCL_PRIMITIVE_3D 0x00000097 - -#define NV20_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV20_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV20_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a4 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001a8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 -#define NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c -#define NV20_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 -#define NV20_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 -#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 -#define NV20_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c -#define NV20_TCL_PRIMITIVE_3D_LIGHT_CONTROL 0x00000294 -#define NV20_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c -#define NV20_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 -#define NV20_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 -#define NV20_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c -#define NV20_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 -#define NV20_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 -#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 -#define NV20_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV20_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV20_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c -#define NV20_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c -#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV20_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 -#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 -#define NV20_TCL_PRIMITIVE_3D_SEPARATE_SPECULAR_ENABLE 0x000003b8 -#define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc -#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_POINT_SIZE 0x0000043c -#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000009c0 -#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000009c4 -#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000009c8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c -#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x00000a30+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000007 -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000ac0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000af0 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000af4 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000af8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000afc -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00001500 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00001504 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00001508 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00001518 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x0000151c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00001520 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY 0x00001528 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z 0x0000152c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00001530 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00001534 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00001538 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00001540 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00001544 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_X 0x00001550 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Y 0x00001554 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Z 0x00001558 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_W 0x0000155c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_X 0x00001560 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Y 0x00001564 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Z 0x00001568 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x0000156c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_X 0x00001580 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Y 0x00001584 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Z 0x00001588 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I 0x0000158c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_MASK 0x000000ff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_SHIFT 8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_MASK 0x0000ff00 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_MASK 0x00ff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_SHIFT 24 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_MASK 0xff000000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00001590 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00001594 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00001598 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x000015a0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x000015a4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x000015a8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x000015ac -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x000015b0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x000015b4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x000015b8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x000015bc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x000015c0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x000015c8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x000015cc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x000015d0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x000015d4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x000015d8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x000015dc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S 0x000015e0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T 0x000015e4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I 0x000015e8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S 0x000015f0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T 0x000015f4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R 0x000015f8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q 0x000015fc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST 0x00001600 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ 0x00001604 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S 0x00001608 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T 0x0000160c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I 0x00001610 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S 0x00001620 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T 0x00001624 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R 0x00001628 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q 0x0000162c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST 0x00001630 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ 0x00001634 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x000016bc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 -#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x000017bc -#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x000017c0 -#define NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 -#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE 0x000017f8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x000017fc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001818 -#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001b00+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001b04+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001b08+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001b0c+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001b14+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x00001c30 -#define NV20_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x00001c50 -#define NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE 0x00001d90 -#define NV20_TCL_PRIMITIVE_3D_CLEAR_BUFFERS 0x00001d94 -#define NV20_TCL_PRIMITIVE_3D_RC_COLOR0 0x00001e20 -#define NV20_TCL_PRIMITIVE_3D_RC_COLOR1 0x00001e24 -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00001e40+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_ENABLE 0x00001e60 -#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP 0x00001e70 -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID 0x00001ea4 - - -#define NV17_TCL_PRIMITIVE_3D 0x00000099 - -#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x000001ac -#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001b0 -#define NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE 0x000002bc -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH 0x00000d5c -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE 0x00000d68 -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE 0x00001658 - - -#define NV20_SWIZZLED_SURFACE 0x0000009e - - - -#define NV12_IMAGE_BLIT 0x0000009f - -#define NV12_IMAGE_BLIT_BETA1 0x00000194 -#define NV12_IMAGE_BLIT_BETA4 0x00000198 - - -#define NV30_CONTEXT_SURFACES_2D 0x00000362 - - - -#define NV30_STRETCHED_IMAGE_FROM_CPU 0x00000366 - - - -#define NV30_TEXTURE_FROM_CPU 0x0000037b - - - -#define NV30_SCALED_IMAGE_FROM_MEMORY 0x00000389 - - - -#define NV30_IMAGE_FROM_CPU 0x0000038a - - - -#define NV30_TCL_PRIMITIVE_3D 0x00000397 - - - -#define NV30_SWIZZLED_SURFACE 0x0000039e - - - -#define NV36_TCL_PRIMITIVE_3D 0x00000497 - - - -#define NV25_TCL_PRIMITIVE_3D 0x00000597 - -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x0000019c -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a0 -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001ac -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY9 0x000001b0 - - -#define NV34_TCL_PRIMITIVE_3D 0x00000697 - -#define NV34_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV34_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV34_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x0000018c -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000194 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x00000198 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a4 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a8 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001ac -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001b0 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_FORMAT 0x00000208 -#define NV34_TCL_PRIMITIVE_3D_PITCH 0x0000020c -#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 -#define NV34_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 -#define NV34_TCL_PRIMITIVE_3D_COLOR1_OFFSET 0x00000218 -#define NV34_TCL_PRIMITIVE_3D_COLOR1_PITCH 0x0000021c -#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_PITCH 0x0000022c -#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_OFFSET 0x00000230 -#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_UNK0250(x) (0x00000250+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_UNK0250__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x0000030c -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000310 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000314 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000318 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_COLOR 0x0000031c -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION 0x00000320 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MIN 0x00008007 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MAX 0x00008008 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000324 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK 0xff000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE 0x00000328 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK 0x0000032c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC 0x00000330 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF 0x00000334 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK 0x00000338 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL 0x0000033c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL 0x00000340 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS 0x00000344 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE 0x00000348 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK 0x0000034c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC 0x00000350 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF 0x00000354 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK 0x00000358 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL 0x0000035c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL 0x00000360 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS 0x00000364 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL 0x00000368 -#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV34_TCL_PRIMITIVE_3D_FOG_ENABLE 0x0000036c -#define NV34_TCL_PRIMITIVE_3D_FOG_COLOR 0x00000370 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000374 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000378 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f -#define NV34_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x0000037c -#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R 0x000003a0 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G 0x000003a4 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B 0x000003a8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A 0x000003b4 -#define NV34_TCL_PRIMITIVE_3D_LINE_WIDTH 0x000003b8 -#define NV34_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x000003bc -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x000008c0 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x000008c4 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000008c8 -#define NV34_TCL_PRIMITIVE_3D_FOG_MODE 0x000008cc -#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000008d0 -#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000008d4 -#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000008d8 -#define NV34_TCL_PRIMITIVE_3D_RC_COLOR0 0x000008ec -#define NV34_TCL_PRIMITIVE_3D_RC_COLOR1 0x000008f0 -#define NV34_TCL_PRIMITIVE_3D_RC_FINAL0 0x000008f4 -#define NV34_TCL_PRIMITIVE_3D_RC_FINAL1 0x000008f8 -#define NV34_TCL_PRIMITIVE_3D_RC_ENABLE 0x000008fc -#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000900+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000904+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000914+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R0x00000a10 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G0x00000a14 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B0x00000a18 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000a30 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000a34 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000a38 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000a3c -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000a6c -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x00000a70 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x00000a74 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000a78 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000a7c -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000a90 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000a94 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x0000000c -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R0x000017a0 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G0x000017a4 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B0x000017a8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R 0x000017b0 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G 0x000017b4 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B 0x000017b8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A 0x000017c0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00001808 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV34_TCL_PRIMITIVE_3D_VERTEX_DATA 0x00001818 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x00001828 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x0000182c -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE 0x00001830 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE 0x00001834 -#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00001838 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x0000183c -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST(x) (0x00001920+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_S_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_S_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_T_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_2I_ST_T_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I__SIZE 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY 0x00001980 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW 0x00001984 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001a00+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001a04+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_SHIFT 4 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_MASK 0x000000f0 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 20 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_SHIFT 28 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001a08+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_REPEAT 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_REPEAT 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MIRRORED_REPEAT 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_EDGE 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_BORDER 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_REPEAT 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MIRRORED_REPEAT 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_EDGE 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_BORDER 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001a0c+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE(x) (0x00001a10+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001a14+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x000f0000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0x0f000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_MASK 0xff000000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_CLEAR_DEPTH_VALUE 0x00001d8c -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE 0x00001db4 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN 0x00001db8 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00001e54 -#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_POINT_SIZE 0x00001ee0 -#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00001ee4 -#define NV34_TCL_PRIMITIVE_3D_POINT_SPRITE 0x00001ee8 -#define NV34_TCL_PRIMITIVE_3D_UNK1f80(x) (0x00001f80+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_UNK1f80__SIZE 0x00000010 - - -#define NV40_CONTEXT_SURFACES_2D 0x00003062 - - - -#define NV40_STRETCHED_IMAGE_FROM_CPU 0x00003066 - - - -#define NV40_TEXTURE_FROM_CPU 0x0000307b - - - -#define NV40_SCALED_IMAGE_FROM_MEMORY 0x00003089 - - - -#define NV40_IMAGE_FROM_CPU 0x0000308a - - - -#define NV40_SWIZZLED_SURFACE 0x0000309e - - - -#define NV40TCL 0x00004097 - -#define NV40TCL_REF_CNT 0x00000050 -#define NV40TCL_NOP 0x00000100 -#define NV40TCL_NOTIFY 0x00000104 -#define NV40TCL_DMA_NOTIFY 0x00000180 -#define NV40TCL_DMA_TEXTURE0 0x00000184 -#define NV40TCL_DMA_TEXTURE1 0x00000188 -#define NV40TCL_DMA_COLOR1 0x0000018c -#define NV40TCL_DMA_COLOR0 0x00000194 -#define NV40TCL_DMA_ZETA 0x00000198 -#define NV40TCL_DMA_VTXBUF0 0x0000019c -#define NV40TCL_DMA_VTXBUF1 0x000001a0 -#define NV40TCL_DMA_FENCE 0x000001a4 -#define NV40TCL_DMA_QUERY 0x000001a8 -#define NV40TCL_DMA_UNK01AC 0x000001ac -#define NV40TCL_DMA_UNK01B0 0x000001b0 -#define NV40TCL_DMA_COLOR2 0x000001b4 -#define NV40TCL_DMA_COLOR3 0x000001b8 -#define NV40TCL_RT_HORIZ 0x00000200 -#define NV40TCL_RT_HORIZ_W_SHIFT 16 -#define NV40TCL_RT_HORIZ_W_MASK 0xffff0000 -#define NV40TCL_RT_HORIZ_X_SHIFT 0 -#define NV40TCL_RT_HORIZ_X_MASK 0x0000ffff -#define NV40TCL_RT_VERT 0x00000204 -#define NV40TCL_RT_VERT_H_SHIFT 16 -#define NV40TCL_RT_VERT_H_MASK 0xffff0000 -#define NV40TCL_RT_VERT_Y_SHIFT 0 -#define NV40TCL_RT_VERT_Y_MASK 0x0000ffff -#define NV40TCL_RT_FORMAT 0x00000208 -#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT 24 -#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_MASK 0xff000000 -#define NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT 16 -#define NV40TCL_RT_FORMAT_LOG2_WIDTH_MASK 0x00ff0000 -#define NV40TCL_RT_FORMAT_TYPE_SHIFT 8 -#define NV40TCL_RT_FORMAT_TYPE_MASK 0x00000f00 -#define NV40TCL_RT_FORMAT_TYPE_LINEAR 0x00000001 -#define NV40TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000002 -#define NV40TCL_RT_FORMAT_DEPTH_SHIFT 5 -#define NV40TCL_RT_FORMAT_DEPTH_MASK 0x000000e0 -#define NV40TCL_RT_FORMAT_DEPTH_Z16 0x00000001 -#define NV40TCL_RT_FORMAT_DEPTH_Z24S8 0x00000002 -#define NV40TCL_RT_FORMAT_COLOR_SHIFT 0 -#define NV40TCL_RT_FORMAT_COLOR_MASK 0x0000001f -#define NV40TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 -#define NV40TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 -#define NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 -#define NV40TCL_RT_FORMAT_COLOR_B8 0x00000009 -#define NV40TCL_COLOR0_PITCH 0x0000020c -#define NV40TCL_COLOR0_OFFSET 0x00000210 -#define NV40TCL_ZETA_OFFSET 0x00000214 -#define NV40TCL_COLOR1_OFFSET 0x00000218 -#define NV40TCL_COLOR1_PITCH 0x0000021c -#define NV40TCL_RT_ENABLE 0x00000220 -#define NV40TCL_RT_ENABLE_MRT (1 << 4) -#define NV40TCL_RT_ENABLE_COLOR3 (1 << 3) -#define NV40TCL_RT_ENABLE_COLOR2 (1 << 2) -#define NV40TCL_RT_ENABLE_COLOR1 (1 << 1) -#define NV40TCL_RT_ENABLE_COLOR0 (1 << 0) -#define NV40TCL_ZETA_PITCH 0x0000022c -#define NV40TCL_COLOR2_PITCH 0x00000280 -#define NV40TCL_COLOR3_PITCH 0x00000284 -#define NV40TCL_COLOR2_OFFSET 0x00000288 -#define NV40TCL_COLOR3_OFFSET 0x0000028c -#define NV40TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) -#define NV40TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV40TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) -#define NV40TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV40TCL_DITHER_ENABLE 0x00000300 -#define NV40TCL_ALPHA_TEST_ENABLE 0x00000304 -#define NV40TCL_ALPHA_TEST_FUNC 0x00000308 -#define NV40TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 -#define NV40TCL_ALPHA_TEST_FUNC_LESS 0x00000201 -#define NV40TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 -#define NV40TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 -#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 -#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 -#define NV40TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 -#define NV40TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 -#define NV40TCL_ALPHA_TEST_REF 0x0000030c -#define NV40TCL_BLEND_ENABLE 0x00000310 -#define NV40TCL_BLEND_FUNC_SRC 0x00000314 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 -#define NV40TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff -#define NV40TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV40TCL_BLEND_FUNC_DST 0x00000318 -#define NV40TCL_BLEND_FUNC_DST_RGB_SHIFT 0 -#define NV40TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff -#define NV40TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV40TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV40TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV40TCL_BLEND_COLOR 0x0000031c -#define NV40TCL_BLEND_EQUATION 0x00000320 -#define NV40TCL_BLEND_EQUATION_RGB_SHIFT 0 -#define NV40TCL_BLEND_EQUATION_RGB_MASK 0x0000ffff -#define NV40TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 -#define NV40TCL_BLEND_EQUATION_RGB_MIN 0x00008007 -#define NV40TCL_BLEND_EQUATION_RGB_MAX 0x00008008 -#define NV40TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a -#define NV40TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV40TCL_BLEND_EQUATION_ALPHA_SHIFT 16 -#define NV40TCL_BLEND_EQUATION_ALPHA_MASK 0xffff0000 -#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 -#define NV40TCL_BLEND_EQUATION_ALPHA_MIN 0x00008007 -#define NV40TCL_BLEND_EQUATION_ALPHA_MAX 0x00008008 -#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a -#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV40TCL_COLOR_MASK 0x00000324 -#define NV40TCL_COLOR_MASK_BUFFER0_B_SHIFT 0 -#define NV40TCL_COLOR_MASK_BUFFER0_B_MASK 0x000000ff -#define NV40TCL_COLOR_MASK_BUFFER0_G_SHIFT 8 -#define NV40TCL_COLOR_MASK_BUFFER0_G_MASK 0x0000ff00 -#define NV40TCL_COLOR_MASK_BUFFER0_R_SHIFT 16 -#define NV40TCL_COLOR_MASK_BUFFER0_R_MASK 0x00ff0000 -#define NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT 24 -#define NV40TCL_COLOR_MASK_BUFFER0_A_MASK 0xff000000 -#define NV40TCL_STENCIL_BACK_ENABLE 0x00000328 -#define NV40TCL_STENCIL_BACK_MASK 0x0000032c -#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000334 -#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000338 -#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000033c -#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000344 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000348 -#define NV40TCL_STENCIL_FRONT_MASK 0x0000034c -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000354 -#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 -#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000035c -#define NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV40TCL_SHADE_MODEL 0x00000368 -#define NV40TCL_SHADE_MODEL_FLAT 0x00001d00 -#define NV40TCL_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV40TCL_MRT_COLOR_MASK 0x00000370 -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_A (1 << 4) -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_R (1 << 5) -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_G (1 << 6) -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_B (1 << 7) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_A (1 << 8) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_R (1 << 9) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_G (1 << 10) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_B (1 << 11) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_A (1 << 12) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_R (1 << 13) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_G (1 << 14) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_B (1 << 15) -#define NV40TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 -#define NV40TCL_COLOR_LOGIC_OP 0x00000378 -#define NV40TCL_COLOR_LOGIC_OP_CLEAR 0x00001500 -#define NV40TCL_COLOR_LOGIC_OP_AND 0x00001501 -#define NV40TCL_COLOR_LOGIC_OP_AND_REVERSE 0x00001502 -#define NV40TCL_COLOR_LOGIC_OP_COPY 0x00001503 -#define NV40TCL_COLOR_LOGIC_OP_AND_INVERTED 0x00001504 -#define NV40TCL_COLOR_LOGIC_OP_NOOP 0x00001505 -#define NV40TCL_COLOR_LOGIC_OP_XOR 0x00001506 -#define NV40TCL_COLOR_LOGIC_OP_OR 0x00001507 -#define NV40TCL_COLOR_LOGIC_OP_NOR 0x00001508 -#define NV40TCL_COLOR_LOGIC_OP_EQUIV 0x00001509 -#define NV40TCL_COLOR_LOGIC_OP_INVERT 0x0000150a -#define NV40TCL_COLOR_LOGIC_OP_OR_REVERSE 0x0000150b -#define NV40TCL_COLOR_LOGIC_OP_COPY_INVERTED 0x0000150c -#define NV40TCL_COLOR_LOGIC_OP_OR_INVERTED 0x0000150d -#define NV40TCL_COLOR_LOGIC_OP_NAND 0x0000150e -#define NV40TCL_COLOR_LOGIC_OP_SET 0x0000150f -#define NV40TCL_DEPTH_RANGE_NEAR 0x00000394 -#define NV40TCL_DEPTH_RANGE_FAR 0x00000398 -#define NV40TCL_LINE_WIDTH 0x000003b8 -#define NV40TCL_LINE_SMOOTH_ENABLE 0x000003bc -#define NV40TCL_UNK03C0(x) (0x000003c0+((x)*4)) -#define NV40TCL_UNK03C0__SIZE 0x00000010 -#define NV40TCL_UNK0400(x) (0x00000400+((x)*4)) -#define NV40TCL_UNK0400__SIZE 0x00000010 -#define NV40TCL_UNK0440(x) (0x00000440+((x)*4)) -#define NV40TCL_UNK0440__SIZE 0x00000020 -#define NV40TCL_SCISSOR_HORIZ 0x000008c0 -#define NV40TCL_SCISSOR_HORIZ_X_SHIFT 0 -#define NV40TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff -#define NV40TCL_SCISSOR_HORIZ_W_SHIFT 16 -#define NV40TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 -#define NV40TCL_SCISSOR_VERT 0x000008c4 -#define NV40TCL_SCISSOR_VERT_Y_SHIFT 0 -#define NV40TCL_SCISSOR_VERT_Y_MASK 0x0000ffff -#define NV40TCL_SCISSOR_VERT_H_SHIFT 16 -#define NV40TCL_SCISSOR_VERT_H_MASK 0xffff0000 -#define NV40TCL_FOG_MODE 0x000008cc -#define NV40TCL_FOG_EQUATION_CONSTANT 0x000008d0 -#define NV40TCL_FOG_EQUATION_LINEAR 0x000008d4 -#define NV40TCL_FOG_EQUATION_QUADRATIC 0x000008d8 -#define NV40TCL_FP_ADDRESS 0x000008e4 -#define NV40TCL_FP_ADDRESS_OFFSET_SHIFT 8 -#define NV40TCL_FP_ADDRESS_OFFSET_MASK 0xffffff00 -#define NV40TCL_FP_ADDRESS_DMA1 (1 << 1) -#define NV40TCL_FP_ADDRESS_DMA0 (1 << 0) -#define NV40TCL_VIEWPORT_HORIZ 0x00000a00 -#define NV40TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV40TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV40TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV40TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV40TCL_VIEWPORT_VERT 0x00000a04 -#define NV40TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV40TCL_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV40TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV40TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV40TCL_VIEWPORT_TRANSLATE_X 0x00000a20 -#define NV40TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 -#define NV40TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 -#define NV40TCL_VIEWPORT_TRANSLATE_W 0x00000a2c -#define NV40TCL_VIEWPORT_SCALE_X 0x00000a30 -#define NV40TCL_VIEWPORT_SCALE_Y 0x00000a34 -#define NV40TCL_VIEWPORT_SCALE_Z 0x00000a38 -#define NV40TCL_VIEWPORT_SCALE_W 0x00000a3c -#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 -#define NV40TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 -#define NV40TCL_DEPTH_FUNC 0x00000a6c -#define NV40TCL_DEPTH_FUNC_NEVER 0x00000200 -#define NV40TCL_DEPTH_FUNC_LESS 0x00000201 -#define NV40TCL_DEPTH_FUNC_EQUAL 0x00000202 -#define NV40TCL_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV40TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV40TCL_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV40TCL_DEPTH_WRITE_ENABLE 0x00000a70 -#define NV40TCL_DEPTH_TEST_ENABLE 0x00000a74 -#define NV40TCL_POLYGON_OFFSET_FACTOR 0x00000a78 -#define NV40TCL_POLYGON_OFFSET_UNITS 0x00000a7c -#define NV40TCL_UNK0B40(x) (0x00000b40+((x)*4)) -#define NV40TCL_UNK0B40__SIZE 0x00000008 -#define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) -#define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) -#define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV40TCL_VTX_ATTR_3F_X(x) (0x00001500+((x)*16)) -#define NV40TCL_VTX_ATTR_3F_X__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_3F_Y(x) (0x00001504+((x)*16)) -#define NV40TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_3F_Z(x) (0x00001508+((x)*16)) -#define NV40TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 -#define NV40TCL_VTXBUF_ADDRESS(x) (0x00001680+((x)*4)) -#define NV40TCL_VTXBUF_ADDRESS__SIZE 0x00000010 -#define NV40TCL_VTXBUF_ADDRESS_DMA1 (1 << 31) -#define NV40TCL_VTXBUF_ADDRESS_OFFSET_SHIFT 0 -#define NV40TCL_VTXBUF_ADDRESS_OFFSET_MASK 0x0fffffff -#define NV40TCL_VTX_CACHE_INVALIDATE 0x00001714 -#define NV40TCL_VTXFMT(x) (0x00001740+((x)*4)) -#define NV40TCL_VTXFMT__SIZE 0x00000010 -#define NV40TCL_VTXFMT_TYPE_SHIFT 0 -#define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f -#define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 -#define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 -#define NV40TCL_VTXFMT_SIZE_SHIFT 4 -#define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 -#define NV40TCL_VTXFMT_STRIDE_SHIFT 8 -#define NV40TCL_VTXFMT_STRIDE_MASK 0x0000ff00 -#define NV40TCL_QUERY_RESET 0x000017c8 -#define NV40TCL_QUERY_UNK17CC 0x000017cc -#define NV40TCL_QUERY_GET 0x00001800 -#define NV40TCL_QUERY_GET_UNK24_SHIFT 24 -#define NV40TCL_QUERY_GET_UNK24_MASK 0xff000000 -#define NV40TCL_QUERY_GET_OFFSET_SHIFT 0 -#define NV40TCL_QUERY_GET_OFFSET_MASK 0x00ffffff -#define NV40TCL_BEGIN_END 0x00001808 -#define NV40TCL_BEGIN_END_STOP 0x00000000 -#define NV40TCL_BEGIN_END_POINTS 0x00000001 -#define NV40TCL_BEGIN_END_LINES 0x00000002 -#define NV40TCL_BEGIN_END_LINE_LOOP 0x00000003 -#define NV40TCL_BEGIN_END_LINE_STRIP 0x00000004 -#define NV40TCL_BEGIN_END_TRIANGLES 0x00000005 -#define NV40TCL_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV40TCL_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV40TCL_BEGIN_END_QUADS 0x00000008 -#define NV40TCL_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV40TCL_BEGIN_END_POLYGON 0x0000000a -#define NV40TCL_VB_ELEMENT_U16 0x0000180c -#define NV40TCL_VB_ELEMENT_U16_1_SHIFT 16 -#define NV40TCL_VB_ELEMENT_U16_1_MASK 0xffff0000 -#define NV40TCL_VB_ELEMENT_U16_0_SHIFT 0 -#define NV40TCL_VB_ELEMENT_U16_0_MASK 0x0000ffff -#define NV40TCL_VB_ELEMENT_U32 0x00001810 -#define NV40TCL_VB_VERTEX_BATCH 0x00001814 -#define NV40TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 -#define NV40TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 -#define NV40TCL_VB_VERTEX_BATCH_START_SHIFT 0 -#define NV40TCL_VB_VERTEX_BATCH_START_MASK 0x00ffffff -#define NV40TCL_VERTEX_DATA 0x00001818 -#define NV40TCL_POLYGON_MODE_FRONT 0x00001828 -#define NV40TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV40TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV40TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV40TCL_POLYGON_MODE_BACK 0x0000182c -#define NV40TCL_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV40TCL_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV40TCL_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV40TCL_CULL_FACE 0x00001830 -#define NV40TCL_CULL_FACE_FRONT 0x00000404 -#define NV40TCL_CULL_FACE_BACK 0x00000405 -#define NV40TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV40TCL_FRONT_FACE 0x00001834 -#define NV40TCL_FRONT_FACE_CW 0x00000900 -#define NV40TCL_FRONT_FACE_CCW 0x00000901 -#define NV40TCL_POLYGON_SMOOTH_ENABLE 0x00001838 -#define NV40TCL_CULL_FACE_ENABLE 0x0000183c -#define NV40TCL_TEX_SIZE1(x) (0x00001840+((x)*4)) -#define NV40TCL_TEX_SIZE1__SIZE 0x00000008 -#define NV40TCL_TEX_SIZE1_DEPTH_SHIFT 20 -#define NV40TCL_TEX_SIZE1_DEPTH_MASK 0xfff00000 -#define NV40TCL_TEX_SIZE1_PITCH_SHIFT 0 -#define NV40TCL_TEX_SIZE1_PITCH_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) -#define NV40TCL_VTX_ATTR_2F_X__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) -#define NV40TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) -#define NV40TCL_VTX_ATTR_2I__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 -#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_2I_X_SHIFT 0 -#define NV40TCL_VTX_ATTR_2I_X_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_4I_0(x) (0x00001900+((x)*8)) -#define NV40TCL_VTX_ATTR_4I_0__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4I_0_Y_SHIFT 16 -#define NV40TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_4I_0_X_SHIFT 0 -#define NV40TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_4I_1(x) (0x00001904+((x)*8)) -#define NV40TCL_VTX_ATTR_4I_1__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4I_1_W_SHIFT 16 -#define NV40TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_4I_1_Z_SHIFT 0 -#define NV40TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff -#define NV40TCL_TEX_OFFSET(x) (0x00001a00+((x)*32)) -#define NV40TCL_TEX_OFFSET__SIZE 0x00000010 -#define NV40TCL_TEX_FORMAT(x) (0x00001a04+((x)*32)) -#define NV40TCL_TEX_FORMAT__SIZE 0x00000010 -#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT 16 -#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_MASK 0x000f0000 -#define NV40TCL_TEX_FORMAT_RECT (1 << 14) -#define NV40TCL_TEX_FORMAT_LINEAR (1 << 13) -#define NV40TCL_TEX_FORMAT_FORMAT_SHIFT 8 -#define NV40TCL_TEX_FORMAT_FORMAT_MASK 0x00001f00 -#define NV40TCL_TEX_FORMAT_FORMAT_L8 0x00000001 -#define NV40TCL_TEX_FORMAT_FORMAT_A1R5G5B5 0x00000002 -#define NV40TCL_TEX_FORMAT_FORMAT_A4R4G4B4 0x00000003 -#define NV40TCL_TEX_FORMAT_FORMAT_R5G6B5 0x00000004 -#define NV40TCL_TEX_FORMAT_FORMAT_A8R8G8B8 0x00000005 -#define NV40TCL_TEX_FORMAT_FORMAT_DXT1 0x00000006 -#define NV40TCL_TEX_FORMAT_FORMAT_DXT3 0x00000007 -#define NV40TCL_TEX_FORMAT_FORMAT_DXT5 0x00000008 -#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00000018 -#define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 -#define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 -#define NV40TCL_TEX_FORMAT_NO_BORDER (1 << 3) -#define NV40TCL_TEX_FORMAT_CUBIC (1 << 2) -#define NV40TCL_TEX_FORMAT_DMA1 (1 << 1) -#define NV40TCL_TEX_FORMAT_DMA0 (1 << 0) -#define NV40TCL_TEX_WRAP(x) (0x00001a08+((x)*32)) -#define NV40TCL_TEX_WRAP__SIZE 0x00000010 -#define NV40TCL_TEX_WRAP_S_SHIFT 0 -#define NV40TCL_TEX_WRAP_S_MASK 0x000000ff -#define NV40TCL_TEX_WRAP_S_REPEAT 0x00000001 -#define NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT 0x00000002 -#define NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE 0x00000003 -#define NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER 0x00000004 -#define NV40TCL_TEX_WRAP_S_CLAMP 0x00000005 -#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE 0x00000006 -#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER 0x00000007 -#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP 0x00000008 -#define NV40TCL_TEX_WRAP_T_SHIFT 8 -#define NV40TCL_TEX_WRAP_T_MASK 0x0000ff00 -#define NV40TCL_TEX_WRAP_T_REPEAT 0x00000001 -#define NV40TCL_TEX_WRAP_T_MIRRORED_REPEAT 0x00000002 -#define NV40TCL_TEX_WRAP_T_CLAMP_TO_EDGE 0x00000003 -#define NV40TCL_TEX_WRAP_T_CLAMP_TO_BORDER 0x00000004 -#define NV40TCL_TEX_WRAP_T_CLAMP 0x00000005 -#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_EDGE 0x00000006 -#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_BORDER 0x00000007 -#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP 0x00000008 -#define NV40TCL_TEX_WRAP_R_SHIFT 16 -#define NV40TCL_TEX_WRAP_R_MASK 0x00ff0000 -#define NV40TCL_TEX_WRAP_R_REPEAT 0x00000001 -#define NV40TCL_TEX_WRAP_R_MIRRORED_REPEAT 0x00000002 -#define NV40TCL_TEX_WRAP_R_CLAMP_TO_EDGE 0x00000003 -#define NV40TCL_TEX_WRAP_R_CLAMP_TO_BORDER 0x00000004 -#define NV40TCL_TEX_WRAP_R_CLAMP 0x00000005 -#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE 0x00000006 -#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER 0x00000007 -#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP 0x00000008 -#define NV40TCL_TEX_ENABLE(x) (0x00001a0c+((x)*32)) -#define NV40TCL_TEX_ENABLE__SIZE 0x00000010 -#define NV40TCL_TEX_ENABLE_ENABLE (1 << 31) -#define NV40TCL_TEX_SWIZZLE(x) (0x00001a10+((x)*32)) -#define NV40TCL_TEX_SWIZZLE__SIZE 0x00000010 -#define NV40TCL_TEX_SWIZZLE_S0_X_SHIFT 14 -#define NV40TCL_TEX_SWIZZLE_S0_X_MASK 0x0000c000 -#define NV40TCL_TEX_SWIZZLE_S0_X_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_X_ONE 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S0_X_S1 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT 12 -#define NV40TCL_TEX_SWIZZLE_S0_Y_MASK 0x00003000 -#define NV40TCL_TEX_SWIZZLE_S0_Y_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_Y_ONE 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S0_Y_S1 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT 10 -#define NV40TCL_TEX_SWIZZLE_S0_Z_MASK 0x00000c00 -#define NV40TCL_TEX_SWIZZLE_S0_Z_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_Z_ONE 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S0_Z_S1 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S0_W_SHIFT 8 -#define NV40TCL_TEX_SWIZZLE_S0_W_MASK 0x00000300 -#define NV40TCL_TEX_SWIZZLE_S0_W_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_W_ONE 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S0_W_S1 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S1_X_SHIFT 6 -#define NV40TCL_TEX_SWIZZLE_S1_X_MASK 0x000000c0 -#define NV40TCL_TEX_SWIZZLE_S1_X_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_X_Z 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S1_X_Y 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S1_X_X 0x00000003 -#define NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT 4 -#define NV40TCL_TEX_SWIZZLE_S1_Y_MASK 0x00000030 -#define NV40TCL_TEX_SWIZZLE_S1_Y_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_Y_Z 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S1_Y_Y 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S1_Y_X 0x00000003 -#define NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT 2 -#define NV40TCL_TEX_SWIZZLE_S1_Z_MASK 0x0000000c -#define NV40TCL_TEX_SWIZZLE_S1_Z_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_Z_Z 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S1_Z_Y 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S1_Z_X 0x00000003 -#define NV40TCL_TEX_SWIZZLE_S1_W_SHIFT 0 -#define NV40TCL_TEX_SWIZZLE_S1_W_MASK 0x00000003 -#define NV40TCL_TEX_SWIZZLE_S1_W_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_W_Z 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S1_W_Y 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S1_W_X 0x00000003 -#define NV40TCL_TEX_FILTER(x) (0x00001a14+((x)*32)) -#define NV40TCL_TEX_FILTER__SIZE 0x00000010 -#define NV40TCL_TEX_FILTER_MIN_SHIFT 16 -#define NV40TCL_TEX_FILTER_MIN_MASK 0x000f0000 -#define NV40TCL_TEX_FILTER_MIN_NEAREST 0x00000001 -#define NV40TCL_TEX_FILTER_MIN_LINEAR 0x00000002 -#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV40TCL_TEX_FILTER_MAG_SHIFT 24 -#define NV40TCL_TEX_FILTER_MAG_MASK 0x0f000000 -#define NV40TCL_TEX_FILTER_MAG_NEAREST 0x00000001 -#define NV40TCL_TEX_FILTER_MAG_LINEAR 0x00000002 -#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_NEAREST 0x00000003 -#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_NEAREST 0x00000004 -#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_LINEAR 0x00000005 -#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_LINEAR 0x00000006 -#define NV40TCL_TEX_SIZE0(x) (0x00001a18+((x)*32)) -#define NV40TCL_TEX_SIZE0__SIZE 0x00000010 -#define NV40TCL_TEX_SIZE0_H_SHIFT 0 -#define NV40TCL_TEX_SIZE0_H_MASK 0x0000ffff -#define NV40TCL_TEX_SIZE0_W_SHIFT 16 -#define NV40TCL_TEX_SIZE0_W_MASK 0xffff0000 -#define NV40TCL_TEX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) -#define NV40TCL_TEX_BORDER_COLOR__SIZE 0x00000010 -#define NV40TCL_TEX_BORDER_COLOR_B_SHIFT 0 -#define NV40TCL_TEX_BORDER_COLOR_B_MASK 0x000000ff -#define NV40TCL_TEX_BORDER_COLOR_G_SHIFT 8 -#define NV40TCL_TEX_BORDER_COLOR_G_MASK 0x0000ff00 -#define NV40TCL_TEX_BORDER_COLOR_R_SHIFT 16 -#define NV40TCL_TEX_BORDER_COLOR_R_MASK 0x00ff0000 -#define NV40TCL_TEX_BORDER_COLOR_A_SHIFT 24 -#define NV40TCL_TEX_BORDER_COLOR_A_MASK 0xff000000 -#define NV40TCL_VTX_ATTR_4F_X(x) (0x00001c00+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_X__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_W__SIZE 0x00000010 -#define NV40TCL_FP_CONTROL 0x00001d60 -#define NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT 24 -#define NV40TCL_FP_CONTROL_TEMP_COUNT_MASK 0xff000000 -#define NV40TCL_FP_CONTROL_KIL (1 << 7) -#define NV40TCL_CLEAR_VALUE_DEPTH 0x00001d8c -#define NV40TCL_CLEAR_VALUE_COLOR 0x00001d90 -#define NV40TCL_CLEAR_BUFFERS 0x00001d94 -#define NV40TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) -#define NV40TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) -#define NV40TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) -#define NV40TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) -#define NV40TCL_CLEAR_BUFFERS_STENCIL (1 << 1) -#define NV40TCL_CLEAR_BUFFERS_DEPTH (1 << 0) -#define NV40TCL_LINE_STIPPLE_ENABLE 0x00001db4 -#define NV40TCL_LINE_STIPPLE_PATTERN 0x00001db8 -#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 -#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff -#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 -#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 -#define NV40TCL_VP_UPLOAD_FROM_ID 0x00001e9c -#define NV40TCL_VP_START_FROM_ID 0x00001ea0 -#define NV40TCL_POINT_SIZE 0x00001ee0 -#define NV40TCL_POINT_SPRITE 0x00001ee8 -#define NV40TCL_VP_UPLOAD_CONST_ID 0x00001efc -#define NV40TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 -#define NV40TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 -#define NV40TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 -#define NV40TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 -#define NV40TCL_TEX_CACHE_CTL 0x00001fd8 -#define NV40TCL_VP_ATTRIB_EN 0x00001ff0 -#define NV40TCL_VP_RESULT_EN 0x00001ff4 - - -#define NV44_TCL_PRIMITIVE_3D 0x00004497 - - - -#define NV50_PRIMITIVE_2D 0x0000502d - -#define NV50_PRIMITIVE_2D_DMA_NOTIFY 0x00000180 -#define NV50_PRIMITIVE_2D_DMA_IN_MEMORY0 0x00000184 -#define NV50_PRIMITIVE_2D_DMA_IN_MEMORY1 0x00000188 -#define NV50_PRIMITIVE_2D_DMA_IN_MEMORY2 0x0000018c - - -#define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 - - - -#define NV50_TCL_PRIMITIVE_3D 0x00005097 - -#define NV50_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV50_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV50_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0(x) (0x00000184+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0__SIZE 0x0000000b -#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1(x) (0x000001c0+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1__SIZE 0x00000008 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00000314 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F(x) (0x00000318+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F__SIZE 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F(x) (0x000003c0+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F__SIZE 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F(x) (0x000003c8+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F__SIZE 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F(x) (0x000003d0+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F__SIZE 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F(x) (0x000003d8+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F__SIZE 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F(x) (0x00000400+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F__SIZE 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F(x) (0x00000420+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F__SIZE 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F(x) (0x00000430+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F__SIZE 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F(x) (0x00000440+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F__SIZE 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X(x) (0x00000500+((x)*16)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X__SIZE 0x00000010 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y(x) (0x00000504+((x)*16)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y__SIZE 0x00000010 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z(x) (0x00000508+((x)*16)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z__SIZE 0x00000010 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W(x) (0x0000050c+((x)*16)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W__SIZE 0x00000010 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x000006a0 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x000006a4 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I 0x000006a8 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I 0x000006ac -#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY 0x00000700 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW 0x00000704 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x00000740 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x00000744 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x00000748 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x0000074c -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST 0x00000750 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ 0x00000754 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST 0x00000758 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ 0x0000075c -#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000790 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000794 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x0000088c -#define NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I 0x00000890 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00000900+((x)*16)) -#define NV50_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0(x) (0x00000a00+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0__SIZE 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1(x) (0x00000a0c+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1__SIZE 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000c00 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000c04 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_SHIFT 0 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_SHIFT 16 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000c08 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000c0c -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x00000d00+((x)*8)) -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x00000d04+((x)*8)) -#define NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_FIRST 0x00000d74 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_COUNT 0x00000d78 -#define NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR(x) (0x00000d80+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR__SIZE 0x00000004 -#define NV50_TCL_PRIMITIVE_3D_CLEAR_DEPTH 0x00000d90 -#define NV50_TCL_PRIMITIVE_3D_CLEAR_STENCIL 0x00000da0 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x00000dac -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000db0 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000db4 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000dc0 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000dc4 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000dc8 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x00000e04 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_L_SHIFT 0 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_L_MASK 0x0000ffff -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_R_SHIFT 16 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_R_MASK 0xffff0000 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x00000e08 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_T_SHIFT 0 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_T_MASK 0x0000ffff -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_B_SHIFT 16 -#define NV50_TCL_PRIMITIVE_3D_SCISSOR_VERT_B_MASK 0xffff0000 -#define NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID 0x00000f00 -#define NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST(x) (0x00000f04+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST__SIZE 0x00000010 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF 0x00000f54 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK 0x00000f58 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK 0x00000f5c -#define NV50_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x000012cc -#define NV50_TCL_PRIMITIVE_3D_SHADE_MODEL 0x000012d4 -#define NV50_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV50_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x000012e8 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x000012ec -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x0000130c -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00001310 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x00001314 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV50_TCL_PRIMITIVE_3D_BLEND_COLOR(x) (0x0000131c+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_BLEND_COLOR__SIZE 0x00000004 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB 0x00001340 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_MIN 0x00008007 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_MAX 0x00008008 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB 0x00001344 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB 0x00001348 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA 0x0000134c -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_MIN 0x00008007 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_MAX 0x00008008 -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a -#define NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA 0x00001350 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA 0x00001358 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50_TCL_PRIMITIVE_3D_UNK1360(x) (0x00001360+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_UNK1360__SIZE 0x00000008 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE 0x00001380 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL 0x00001384 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL 0x00001388 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS 0x0000138c -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC 0x00001390 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF 0x00001394 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK 0x00001398 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK 0x0000139c -#define NV50_TCL_PRIMITIVE_3D_LINE_WIDTH 0x000013b0 -#define NV50_TCL_PRIMITIVE_3D_VP_INST 0x0000140c -#define NV50_TCL_PRIMITIVE_3D_POINT_SIZE 0x00001518 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x0000156c -#define NV50_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00001570 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE 0x00001594 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL 0x00001598 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL 0x0000159c -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS 0x000015a0 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC 0x000015a4 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x000015bc -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN 0x000015dc -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_POINTS 0x00000000 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_LINES 0x00000001 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_LINE_LOOP 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_LINE_STRIP 0x00000003 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_TRIANGLES 0x00000004 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_TRIANGLE_STRIP 0x00000005 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_TRIANGLE_FAN 0x00000006 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_QUADS 0x00000007 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_QUAD_STRIP 0x00000008 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN_POLYGON 0x00000009 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_END 0x000015e0 -#define NV50_TCL_PRIMITIVE_3D_VERTEX_DATA 0x00001640 -#define NV50_TCL_PRIMITIVE_3D_VP_INST_PARAMS(x) (0x00001650+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_VP_INST_PARAMS__SIZE 0x00000002 -#define NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE 0x0000166c -#define NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN 0x00001680 -#define NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000168c -#define NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001700+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00001918 -#define NV50_TCL_PRIMITIVE_3D_FRONT_FACE 0x0000191c -#define NV50_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV50_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV50_TCL_PRIMITIVE_3D_CULL_FACE 0x00001920 -#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV50_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_ENABLE 0x000019c4 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP 0x000019c8 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_AND 0x00001501 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_COPY 0x00001503 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_NOOP 0x00001505 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_XOR 0x00001506 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_OR 0x00001507 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_NOR 0x00001508 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_INVERT 0x0000150a -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_NAND 0x0000150e -#define NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP_SET 0x0000150f -#define NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS 0x000019d0 -#define NV50_TCL_PRIMITIVE_3D_COLOR_MASK(x) (0x00001a00+((x)*4)) -#define NV50_TCL_PRIMITIVE_3D_COLOR_MASK__SIZE 0x00000008 - - -#define NV52_TCL_PRIMITIVE_3D 0x00008297 +#define NV04_IMAGE_BLEND_PREMULT_NOP 0x00000100 +#define NV04_IMAGE_BLEND_PREMULT_NOTIFY 0x00000104 +#define NV04_IMAGE_BLEND_PREMULT_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_BLEND_PREMULT_IMAGE_OUTPUT 0x00000200 +#define NV04_IMAGE_BLEND_PREMULT_BETA_INPUT 0x00000204 +#define NV04_IMAGE_BLEND_PREMULT_IMAGE_INPUT 0x00000208 + + +#define NV03_CHANNEL_PIO 0x0000006a + + + +#define NV03_CHANNEL_DMA 0x0000006b + + + +#define NV04_BETA_SOLID 0x00000072 + +#define NV04_BETA_SOLID_NOP 0x00000100 +#define NV04_BETA_SOLID_NOTIFY 0x00000104 +#define NV04_BETA_SOLID_DMA_NOTIFY 0x00000180 +#define NV04_BETA_SOLID_BETA_OUTPUT 0x00000200 +#define NV04_BETA_SOLID_BETA_FACTOR 0x00000300 + + +#define NV04_STRETCHED_IMAGE_FROM_CPU 0x00000076 + + + +#define NV04_SCALED_IMAGE_FROM_MEMORY 0x00000077 + +#define NV04_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 +#define NV04_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 +#define NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 +#define NV04_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c +#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 +#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA4 0x00000194 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000198 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY 0x00000003 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000005 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT 0x00000310 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE 0x00000314 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX 0x00000318 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY 0x0000031c +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE 0x00000400 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT 0x00000404 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_MASK 0x00ff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER 0x00010000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CORNER 0x00020000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_SHIFT 24 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_MASK 0xff000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE 0x00000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_BILINEAR 0x01000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_ADDRESS 0x00000408 +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT 0x0000040c +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_MASK 0xffff0000 + + +#define NV10_TEXTURE_FROM_CPU 0x0000007b + +#define NV10_TEXTURE_FROM_CPU_NOP 0x00000100 +#define NV10_TEXTURE_FROM_CPU_NOTIFY 0x00000104 +#define NV10_TEXTURE_FROM_CPU_WAIT_FOR_IDLE 0x00000108 +#define NV10_TEXTURE_FROM_CPU_PM_TRIGGER 0x00000140 +#define NV10_TEXTURE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV10_TEXTURE_FROM_CPU_SURFACE 0x00000184 +#define NV10_TEXTURE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV10_TEXTURE_FROM_CPU_POINT 0x00000304 +#define NV10_TEXTURE_FROM_CPU_POINT_X_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_POINT_X_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_POINT_Y_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_POINT_Y_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_SIZE 0x00000308 +#define NV10_TEXTURE_FROM_CPU_SIZE_W_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_SIZE_W_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_SIZE_H_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_SIZE_H_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL 0x0000030c +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL 0x00000310 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV10_TEXTURE_FROM_CPU_COLOR__SIZE 0x00000700 + + +#define NV10_VIDEO_DISPLAY 0x0000007c + + + +#define NV10_DVD_SUBPICTURE 0x00000088 + + + +#define NV10_SCALED_IMAGE_FROM_MEMORY 0x00000089 + +#define NV10_SCALED_IMAGE_FROM_MEMORY_WAIT_FOR_IDLE 0x00000108 + + +#define NV10_IMAGE_FROM_CPU 0x0000008a + +#define NV10_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 + + +#define NV10_CONTEXT_SURFACES_3D 0x00000093 + + + +#define NV10_DX5_TEXTURE_TRIANGLE 0x00000094 + + + +#define NV10_DX6_MULTI_TEXTURE_TRIANGLE 0x00000095 + + + +#define NV11_TCL_PRIMITIVE_3D 0x00000096 + +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000d40 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000d44 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f + + +#define NV20_TCL_PRIMITIVE_3D 0x00000097 + +#define NV20_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV20_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV20_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a4 +#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001a8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 +#define NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c +#define NV20_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 +#define NV20_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 +#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 +#define NV20_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c +#define NV20_TCL_PRIMITIVE_3D_LIGHT_CONTROL 0x00000294 +#define NV20_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c +#define NV20_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 +#define NV20_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 +#define NV20_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c +#define NV20_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 +#define NV20_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 +#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV20_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV20_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c +#define NV20_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c +#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV20_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 +#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 +#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 +#define NV20_TCL_PRIMITIVE_3D_SEPARATE_SPECULAR_ENABLE 0x000003b8 +#define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc +#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_POINT_SIZE 0x0000043c +#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000009c0 +#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000009c4 +#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000009c8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c +#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x00000a30+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000007 +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000ac0+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000af0 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000af4 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000af8 +#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000afc +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00001500 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00001504 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00001508 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00001518 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x0000151c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00001520 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY 0x00001528 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z 0x0000152c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00001530 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00001534 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00001538 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00001540 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00001544 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_X 0x00001550 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Y 0x00001554 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Z 0x00001558 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_W 0x0000155c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_X 0x00001560 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Y 0x00001564 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Z 0x00001568 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x0000156c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_X 0x00001580 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Y 0x00001584 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Z 0x00001588 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I 0x0000158c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_MASK 0x000000ff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_SHIFT 8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_MASK 0x0000ff00 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_MASK 0x00ff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_SHIFT 24 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_MASK 0xff000000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00001590 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00001594 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00001598 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x000015a0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x000015a4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x000015a8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x000015ac +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x000015b0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x000015b4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x000015b8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x000015bc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x000015c0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x000015c8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x000015cc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x000015d0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x000015d4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x000015d8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x000015dc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S 0x000015e0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T 0x000015e4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I 0x000015e8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S 0x000015f0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T 0x000015f4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R 0x000015f8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q 0x000015fc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST 0x00001600 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ 0x00001604 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S 0x00001608 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T 0x0000160c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I 0x00001610 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S 0x00001620 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T 0x00001624 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R 0x00001628 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q 0x0000162c +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST 0x00001630 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ 0x00001634 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT 16 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 +#define NV20_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x000016bc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x000017bc +#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x000017c0 +#define NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 +#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE 0x000017f8 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x000017fc +#define NV20_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 +#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001818 +#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001b00+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001b04+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001b08+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001b0c+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001b14+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) +#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV20_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x00001c30 +#define NV20_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x00001c50 +#define NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE 0x00001d90 +#define NV20_TCL_PRIMITIVE_3D_CLEAR_BUFFERS 0x00001d94 +#define NV20_TCL_PRIMITIVE_3D_RC_COLOR0 0x00001e20 +#define NV20_TCL_PRIMITIVE_3D_RC_COLOR1 0x00001e24 +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00001e40+((x)*4)) +#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 +#define NV20_TCL_PRIMITIVE_3D_RC_ENABLE 0x00001e60 +#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP 0x00001e70 +#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID 0x00001ea4 + + +#define NV17_TCL_PRIMITIVE_3D 0x00000099 + +#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x000001ac +#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001b0 +#define NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE 0x000002bc +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH 0x00000d5c +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE 0x00000d68 +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c +#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE 0x00001658 + + +#define NV20_SWIZZLED_SURFACE 0x0000009e + + + +#define NV12_IMAGE_BLIT 0x0000009f + + + +#define NV30_CONTEXT_SURFACES_2D 0x00000362 + + + +#define NV30_STRETCHED_IMAGE_FROM_CPU 0x00000366 + + + +#define NV30_TEXTURE_FROM_CPU 0x0000037b + + + +#define NV30_SCALED_IMAGE_FROM_MEMORY 0x00000389 + + + +#define NV30_IMAGE_FROM_CPU 0x0000038a + + + +#define NV30_TCL_PRIMITIVE_3D 0x00000397 + + + +#define NV30_SWIZZLED_SURFACE 0x0000039e + + + +#define NV36_TCL_PRIMITIVE_3D 0x00000497 + + + +#define NV25_TCL_PRIMITIVE_3D 0x00000597 + +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x0000019c +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a0 +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001ac +#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY9 0x000001b0 + + +#define NV34_TCL_PRIMITIVE_3D 0x00000697 + +#define NV34_TCL_PRIMITIVE_3D_NOP 0x00000100 +#define NV34_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 +#define NV34_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x0000018c +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000194 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x00000198 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a4 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a8 +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001ac +#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001b0 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_FORMAT 0x00000208 +#define NV34_TCL_PRIMITIVE_3D_PITCH 0x0000020c +#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 +#define NV34_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 +#define NV34_TCL_PRIMITIVE_3D_COLOR1_OFFSET 0x00000218 +#define NV34_TCL_PRIMITIVE_3D_COLOR1_PITCH 0x0000021c +#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_PITCH 0x0000022c +#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_OFFSET 0x00000230 +#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_UNK0250(x) (0x00000250+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_UNK0250__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x0000030c +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000310 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000314 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000318 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_COLOR 0x0000031c +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION 0x00000320 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MIN 0x00008007 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MAX 0x00008008 +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000324 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK 0xff000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE 0x00000328 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK 0x0000032c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC 0x00000330 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF 0x00000334 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK 0x00000338 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL 0x0000033c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL 0x00000340 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS 0x00000344 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE 0x00000348 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK 0x0000034c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF 0x00000354 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK 0x00000358 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL 0x00000368 +#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 +#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV34_TCL_PRIMITIVE_3D_FOG_ENABLE 0x0000036c +#define NV34_TCL_PRIMITIVE_3D_FOG_COLOR 0x00000370 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000378 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f +#define NV34_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x0000037c +#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R 0x000003a0 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G 0x000003a4 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B 0x000003a8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A 0x000003b4 +#define NV34_TCL_PRIMITIVE_3D_LINE_WIDTH 0x000003b8 +#define NV34_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x000003bc +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x000008c0 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x000008c4 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000008c8 +#define NV34_TCL_PRIMITIVE_3D_FOG_MODE 0x000008cc +#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000008d4 +#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV34_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM 0x000008e4 +#define NV34_TCL_PRIMITIVE_3D_RC_COLOR0 0x000008ec +#define NV34_TCL_PRIMITIVE_3D_RC_COLOR1 0x000008f0 +#define NV34_TCL_PRIMITIVE_3D_RC_FINAL0 0x000008f4 +#define NV34_TCL_PRIMITIVE_3D_RC_FINAL1 0x000008f8 +#define NV34_TCL_PRIMITIVE_3D_RC_ENABLE 0x000008fc +#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000900+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000904+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000914+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x00000a10 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x00000a14 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x00000a18 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000a30 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000a34 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000a38 +#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000a3c +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000a6c +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV34_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x00000a74 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000a90 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000a94 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) +#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x0000000c +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000017a0 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000017a4 +#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000017a8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R 0x000017b0 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G 0x000017b4 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B 0x000017b8 +#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A 0x000017c0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00001808 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV34_TCL_PRIMITIVE_3D_VERTEX_DATA 0x00001818 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x00001828 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x0000182c +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE 0x00001830 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE 0x00001834 +#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 +#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 +#define NV34_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x0000183c +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_Y_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_Y_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I__SIZE 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY 0x00001980 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW 0x00001984 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_SHIFT 4 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_MASK 0x000000f0 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8 0x00000100 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4 0x00000400 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8 0x00000600 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_X8R8G8B8 0x00000700 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8 0x00000b00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1 0x00000c00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3 0x00000e00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5 0x00000f00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT 0x00001000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT 0x00001200 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT 0x00001300 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8 0x00001a00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT2 0x00001b00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT 0x00001d00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT 0x00002000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16 0x00003200 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16_RECT 0x00003500 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00004a00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00004b00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00004c00 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT (1 << 12) +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_RECT (1 << 14) +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 20 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_SHIFT 28 +#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001a08+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_REPEAT 0x00000001 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP 0x00000005 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_REPEAT 0x00000100 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MIRRORED_REPEAT 0x00000200 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_EDGE 0x00000300 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_BORDER 0x00000400 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP 0x00000500 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_REPEAT 0x00010000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MIRRORED_REPEAT 0x00020000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_EDGE 0x00030000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_BORDER 0x00040000 +#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP 0x00050000 +#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_SWIZZLE_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_SWIZZLE_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_RECT_PITCH_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_RECT_PITCH_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001a14+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x000f0000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x00010000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x00020000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00030000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00040000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00050000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00060000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0x0f000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x01000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x02000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_MASK 0x000000ff +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_SHIFT 8 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_SHIFT 24 +#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_MASK 0xff000000 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W__SIZE 0x00000010 +#define NV34_TCL_PRIMITIVE_3D_CLEAR_DEPTH_VALUE 0x00001d8c +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00001e54 +#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS__SIZE 0x00000008 +#define NV34_TCL_PRIMITIVE_3D_POINT_SIZE 0x00001ee0 +#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00001ee4 +#define NV34_TCL_PRIMITIVE_3D_POINT_SPRITE 0x00001ee8 +#define NV34_TCL_PRIMITIVE_3D_UNK1f80(x) (0x00001f80+((x)*4)) +#define NV34_TCL_PRIMITIVE_3D_UNK1f80__SIZE 0x00000010 + + +#define NV40_CONTEXT_SURFACES_2D 0x00003062 + + + +#define NV40_STRETCHED_IMAGE_FROM_CPU 0x00003066 + + + +#define NV40_TEXTURE_FROM_CPU 0x0000307b + + + +#define NV40_SCALED_IMAGE_FROM_MEMORY 0x00003089 + + + +#define NV40_IMAGE_FROM_CPU 0x0000308a + + + +#define NV40_SWIZZLED_SURFACE 0x0000309e + + + +#define NV40TCL 0x00004097 + +#define NV40TCL_REF_CNT 0x00000050 +#define NV40TCL_NOP 0x00000100 +#define NV40TCL_NOTIFY 0x00000104 +#define NV40TCL_DMA_NOTIFY 0x00000180 +#define NV40TCL_DMA_TEXTURE0 0x00000184 +#define NV40TCL_DMA_TEXTURE1 0x00000188 +#define NV40TCL_DMA_COLOR1 0x0000018c +#define NV40TCL_DMA_COLOR0 0x00000194 +#define NV40TCL_DMA_ZETA 0x00000198 +#define NV40TCL_DMA_VTXBUF0 0x0000019c +#define NV40TCL_DMA_VTXBUF1 0x000001a0 +#define NV40TCL_DMA_FENCE 0x000001a4 +#define NV40TCL_DMA_QUERY 0x000001a8 +#define NV40TCL_DMA_UNK01AC 0x000001ac +#define NV40TCL_DMA_UNK01B0 0x000001b0 +#define NV40TCL_DMA_COLOR2 0x000001b4 +#define NV40TCL_DMA_COLOR3 0x000001b8 +#define NV40TCL_RT_HORIZ 0x00000200 +#define NV40TCL_RT_HORIZ_W_SHIFT 16 +#define NV40TCL_RT_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_RT_HORIZ_X_SHIFT 0 +#define NV40TCL_RT_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_RT_VERT 0x00000204 +#define NV40TCL_RT_VERT_H_SHIFT 16 +#define NV40TCL_RT_VERT_H_MASK 0xffff0000 +#define NV40TCL_RT_VERT_Y_SHIFT 0 +#define NV40TCL_RT_VERT_Y_MASK 0x0000ffff +#define NV40TCL_RT_FORMAT 0x00000208 +#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT 24 +#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_MASK 0xff000000 +#define NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT 16 +#define NV40TCL_RT_FORMAT_LOG2_WIDTH_MASK 0x00ff0000 +#define NV40TCL_RT_FORMAT_TYPE_SHIFT 8 +#define NV40TCL_RT_FORMAT_TYPE_MASK 0x00000f00 +#define NV40TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 +#define NV40TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 +#define NV40TCL_RT_FORMAT_ZETA_SHIFT 5 +#define NV40TCL_RT_FORMAT_ZETA_MASK 0x000000e0 +#define NV40TCL_RT_FORMAT_ZETA_Z16 0x00000020 +#define NV40TCL_RT_FORMAT_ZETA_Z24S8 0x00000040 +#define NV40TCL_RT_FORMAT_COLOR_SHIFT 0 +#define NV40TCL_RT_FORMAT_COLOR_MASK 0x0000001f +#define NV40TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV40TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 +#define NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV40TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV40TCL_COLOR0_PITCH 0x0000020c +#define NV40TCL_COLOR0_OFFSET 0x00000210 +#define NV40TCL_ZETA_OFFSET 0x00000214 +#define NV40TCL_COLOR1_OFFSET 0x00000218 +#define NV40TCL_COLOR1_PITCH 0x0000021c +#define NV40TCL_RT_ENABLE 0x00000220 +#define NV40TCL_RT_ENABLE_MRT (1 << 4) +#define NV40TCL_RT_ENABLE_COLOR3 (1 << 3) +#define NV40TCL_RT_ENABLE_COLOR2 (1 << 2) +#define NV40TCL_RT_ENABLE_COLOR1 (1 << 1) +#define NV40TCL_RT_ENABLE_COLOR0 (1 << 0) +#define NV40TCL_ZETA_PITCH 0x0000022c +#define NV40TCL_COLOR2_PITCH 0x00000280 +#define NV40TCL_COLOR3_PITCH 0x00000284 +#define NV40TCL_COLOR2_OFFSET 0x00000288 +#define NV40TCL_COLOR3_OFFSET 0x0000028c +#define NV40TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV40TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV40TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV40TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV40TCL_DITHER_ENABLE 0x00000300 +#define NV40TCL_ALPHA_TEST_ENABLE 0x00000304 +#define NV40TCL_ALPHA_TEST_FUNC 0x00000308 +#define NV40TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 +#define NV40TCL_ALPHA_TEST_FUNC_LESS 0x00000201 +#define NV40TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 +#define NV40TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 +#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV40TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 +#define NV40TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 +#define NV40TCL_ALPHA_TEST_REF 0x0000030c +#define NV40TCL_BLEND_ENABLE 0x00000310 +#define NV40TCL_BLEND_FUNC_SRC 0x00000314 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV40TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV40TCL_BLEND_FUNC_DST 0x00000318 +#define NV40TCL_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV40TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV40TCL_BLEND_COLOR 0x0000031c +#define NV40TCL_BLEND_EQUATION 0x00000320 +#define NV40TCL_BLEND_EQUATION_RGB_SHIFT 0 +#define NV40TCL_BLEND_EQUATION_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 +#define NV40TCL_BLEND_EQUATION_RGB_MIN 0x00008007 +#define NV40TCL_BLEND_EQUATION_RGB_MAX 0x00008008 +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV40TCL_BLEND_EQUATION_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_EQUATION_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x80060000 +#define NV40TCL_BLEND_EQUATION_ALPHA_MIN 0x80070000 +#define NV40TCL_BLEND_EQUATION_ALPHA_MAX 0x80080000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x800a0000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x800b0000 +#define NV40TCL_COLOR_MASK 0x00000324 +#define NV40TCL_COLOR_MASK_BUFFER0_B_SHIFT 0 +#define NV40TCL_COLOR_MASK_BUFFER0_B_MASK 0x000000ff +#define NV40TCL_COLOR_MASK_BUFFER0_G_SHIFT 8 +#define NV40TCL_COLOR_MASK_BUFFER0_G_MASK 0x0000ff00 +#define NV40TCL_COLOR_MASK_BUFFER0_R_SHIFT 16 +#define NV40TCL_COLOR_MASK_BUFFER0_R_MASK 0x00ff0000 +#define NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT 24 +#define NV40TCL_COLOR_MASK_BUFFER0_A_MASK 0xff000000 +#define NV40TCL_STENCIL_BACK_ENABLE 0x00000328 +#define NV40TCL_STENCIL_BACK_MASK 0x0000032c +#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000334 +#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000338 +#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000033c +#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000344 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000348 +#define NV40TCL_STENCIL_FRONT_MASK 0x0000034c +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000354 +#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 +#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_SHADE_MODEL 0x00000368 +#define NV40TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV40TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV40TCL_MRT_COLOR_MASK 0x00000370 +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_A (1 << 4) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_R (1 << 5) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_G (1 << 6) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_B (1 << 7) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_A (1 << 8) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_R (1 << 9) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_G (1 << 10) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_B (1 << 11) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_A (1 << 12) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_R (1 << 13) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_G (1 << 14) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_B (1 << 15) +#define NV40TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV40TCL_COLOR_LOGIC_OP 0x00000378 +#define NV40TCL_COLOR_LOGIC_OP_CLEAR 0x00001500 +#define NV40TCL_COLOR_LOGIC_OP_AND 0x00001501 +#define NV40TCL_COLOR_LOGIC_OP_AND_REVERSE 0x00001502 +#define NV40TCL_COLOR_LOGIC_OP_COPY 0x00001503 +#define NV40TCL_COLOR_LOGIC_OP_AND_INVERTED 0x00001504 +#define NV40TCL_COLOR_LOGIC_OP_NOOP 0x00001505 +#define NV40TCL_COLOR_LOGIC_OP_XOR 0x00001506 +#define NV40TCL_COLOR_LOGIC_OP_OR 0x00001507 +#define NV40TCL_COLOR_LOGIC_OP_NOR 0x00001508 +#define NV40TCL_COLOR_LOGIC_OP_EQUIV 0x00001509 +#define NV40TCL_COLOR_LOGIC_OP_INVERT 0x0000150a +#define NV40TCL_COLOR_LOGIC_OP_OR_REVERSE 0x0000150b +#define NV40TCL_COLOR_LOGIC_OP_COPY_INVERTED 0x0000150c +#define NV40TCL_COLOR_LOGIC_OP_OR_INVERTED 0x0000150d +#define NV40TCL_COLOR_LOGIC_OP_NAND 0x0000150e +#define NV40TCL_COLOR_LOGIC_OP_SET 0x0000150f +#define NV40TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV40TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV40TCL_LINE_WIDTH 0x000003b8 +#define NV40TCL_LINE_SMOOTH_ENABLE 0x000003bc +#define NV40TCL_UNK03C0(x) (0x000003c0+((x)*4)) +#define NV40TCL_UNK03C0__SIZE 0x00000010 +#define NV40TCL_UNK0400(x) (0x00000400+((x)*4)) +#define NV40TCL_UNK0400__SIZE 0x00000010 +#define NV40TCL_UNK0440(x) (0x00000440+((x)*4)) +#define NV40TCL_UNK0440__SIZE 0x00000020 +#define NV40TCL_SCISSOR_HORIZ 0x000008c0 +#define NV40TCL_SCISSOR_HORIZ_X_SHIFT 0 +#define NV40TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_SCISSOR_HORIZ_W_SHIFT 16 +#define NV40TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_SCISSOR_VERT 0x000008c4 +#define NV40TCL_SCISSOR_VERT_Y_SHIFT 0 +#define NV40TCL_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV40TCL_SCISSOR_VERT_H_SHIFT 16 +#define NV40TCL_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV40TCL_FOG_MODE 0x000008cc +#define NV40TCL_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV40TCL_FOG_EQUATION_LINEAR 0x000008d4 +#define NV40TCL_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV40TCL_FP_ADDRESS 0x000008e4 +#define NV40TCL_FP_ADDRESS_OFFSET_SHIFT 8 +#define NV40TCL_FP_ADDRESS_OFFSET_MASK 0xffffff00 +#define NV40TCL_FP_ADDRESS_DMA1 (1 << 1) +#define NV40TCL_FP_ADDRESS_DMA0 (1 << 0) +#define NV40TCL_VIEWPORT_HORIZ 0x00000a00 +#define NV40TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV40TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV40TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_VIEWPORT_VERT 0x00000a04 +#define NV40TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV40TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV40TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV40TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV40TCL_VIEWPORT_TRANSLATE_X 0x00000a20 +#define NV40TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 +#define NV40TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 +#define NV40TCL_VIEWPORT_TRANSLATE_W 0x00000a2c +#define NV40TCL_VIEWPORT_SCALE_X 0x00000a30 +#define NV40TCL_VIEWPORT_SCALE_Y 0x00000a34 +#define NV40TCL_VIEWPORT_SCALE_Z 0x00000a38 +#define NV40TCL_VIEWPORT_SCALE_W 0x00000a3c +#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV40TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV40TCL_DEPTH_FUNC 0x00000a6c +#define NV40TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV40TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV40TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV40TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV40TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV40TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV40TCL_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV40TCL_DEPTH_TEST_ENABLE 0x00000a74 +#define NV40TCL_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV40TCL_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV40TCL_UNK0B40(x) (0x00000b40+((x)*4)) +#define NV40TCL_UNK0B40__SIZE 0x00000008 +#define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) +#define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV40TCL_VTX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 +#define NV40TCL_VTXBUF_ADDRESS(x) (0x00001680+((x)*4)) +#define NV40TCL_VTXBUF_ADDRESS__SIZE 0x00000010 +#define NV40TCL_VTXBUF_ADDRESS_DMA1 (1 << 31) +#define NV40TCL_VTXBUF_ADDRESS_OFFSET_SHIFT 0 +#define NV40TCL_VTXBUF_ADDRESS_OFFSET_MASK 0x0fffffff +#define NV40TCL_VTX_CACHE_INVALIDATE 0x00001714 +#define NV40TCL_VTXFMT(x) (0x00001740+((x)*4)) +#define NV40TCL_VTXFMT__SIZE 0x00000010 +#define NV40TCL_VTXFMT_TYPE_SHIFT 0 +#define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f +#define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 +#define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 +#define NV40TCL_VTXFMT_SIZE_SHIFT 4 +#define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 +#define NV40TCL_VTXFMT_STRIDE_SHIFT 8 +#define NV40TCL_VTXFMT_STRIDE_MASK 0x0000ff00 +#define NV40TCL_QUERY_RESET 0x000017c8 +#define NV40TCL_QUERY_UNK17CC 0x000017cc +#define NV40TCL_QUERY_GET 0x00001800 +#define NV40TCL_QUERY_GET_UNK24_SHIFT 24 +#define NV40TCL_QUERY_GET_UNK24_MASK 0xff000000 +#define NV40TCL_QUERY_GET_OFFSET_SHIFT 0 +#define NV40TCL_QUERY_GET_OFFSET_MASK 0x00ffffff +#define NV40TCL_BEGIN_END 0x00001808 +#define NV40TCL_BEGIN_END_STOP 0x00000000 +#define NV40TCL_BEGIN_END_POINTS 0x00000001 +#define NV40TCL_BEGIN_END_LINES 0x00000002 +#define NV40TCL_BEGIN_END_LINE_LOOP 0x00000003 +#define NV40TCL_BEGIN_END_LINE_STRIP 0x00000004 +#define NV40TCL_BEGIN_END_TRIANGLES 0x00000005 +#define NV40TCL_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV40TCL_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV40TCL_BEGIN_END_QUADS 0x00000008 +#define NV40TCL_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV40TCL_BEGIN_END_POLYGON 0x0000000a +#define NV40TCL_VB_ELEMENT_U16 0x0000180c +#define NV40TCL_VB_ELEMENT_U16_1_SHIFT 16 +#define NV40TCL_VB_ELEMENT_U16_1_MASK 0xffff0000 +#define NV40TCL_VB_ELEMENT_U16_0_SHIFT 0 +#define NV40TCL_VB_ELEMENT_U16_0_MASK 0x0000ffff +#define NV40TCL_VB_ELEMENT_U32 0x00001810 +#define NV40TCL_VB_VERTEX_BATCH 0x00001814 +#define NV40TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 +#define NV40TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 +#define NV40TCL_VB_VERTEX_BATCH_START_SHIFT 0 +#define NV40TCL_VB_VERTEX_BATCH_START_MASK 0x00ffffff +#define NV40TCL_VERTEX_DATA 0x00001818 +#define NV40TCL_POLYGON_MODE_FRONT 0x00001828 +#define NV40TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV40TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV40TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV40TCL_POLYGON_MODE_BACK 0x0000182c +#define NV40TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV40TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV40TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV40TCL_CULL_FACE 0x00001830 +#define NV40TCL_CULL_FACE_FRONT 0x00000404 +#define NV40TCL_CULL_FACE_BACK 0x00000405 +#define NV40TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV40TCL_FRONT_FACE 0x00001834 +#define NV40TCL_FRONT_FACE_CW 0x00000900 +#define NV40TCL_FRONT_FACE_CCW 0x00000901 +#define NV40TCL_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV40TCL_CULL_FACE_ENABLE 0x0000183c +#define NV40TCL_TEX_SIZE1(x) (0x00001840+((x)*4)) +#define NV40TCL_TEX_SIZE1__SIZE 0x00000008 +#define NV40TCL_TEX_SIZE1_DEPTH_SHIFT 20 +#define NV40TCL_TEX_SIZE1_DEPTH_MASK 0xfff00000 +#define NV40TCL_TEX_SIZE1_PITCH_SHIFT 0 +#define NV40TCL_TEX_SIZE1_PITCH_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV40TCL_VTX_ATTR_2F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV40TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV40TCL_VTX_ATTR_2I__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_2I_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_2I_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_0(x) (0x00001900+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_0__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_0_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_0_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_1(x) (0x00001904+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_1__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_1_W_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_1_Z_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff +#define NV40TCL_TEX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV40TCL_TEX_OFFSET__SIZE 0x00000010 +#define NV40TCL_TEX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV40TCL_TEX_FORMAT__SIZE 0x00000010 +#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT 16 +#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_MASK 0x000f0000 +#define NV40TCL_TEX_FORMAT_RECT (1 << 14) +#define NV40TCL_TEX_FORMAT_LINEAR (1 << 13) +#define NV40TCL_TEX_FORMAT_FORMAT_SHIFT 8 +#define NV40TCL_TEX_FORMAT_FORMAT_MASK 0x00001f00 +#define NV40TCL_TEX_FORMAT_FORMAT_L8 0x00000100 +#define NV40TCL_TEX_FORMAT_FORMAT_A1R5G5B5 0x00000200 +#define NV40TCL_TEX_FORMAT_FORMAT_A4R4G4B4 0x00000300 +#define NV40TCL_TEX_FORMAT_FORMAT_R5G6B5 0x00000400 +#define NV40TCL_TEX_FORMAT_FORMAT_A8R8G8B8 0x00000500 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT1 0x00000600 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT3 0x00000700 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT5 0x00000800 +#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00001800 +#define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 +#define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 +#define NV40TCL_TEX_FORMAT_DIMS_1D 0x00000010 +#define NV40TCL_TEX_FORMAT_DIMS_2D 0x00000020 +#define NV40TCL_TEX_FORMAT_DIMS_3D 0x00000030 +#define NV40TCL_TEX_FORMAT_NO_BORDER (1 << 3) +#define NV40TCL_TEX_FORMAT_CUBIC (1 << 2) +#define NV40TCL_TEX_FORMAT_DMA1 (1 << 1) +#define NV40TCL_TEX_FORMAT_DMA0 (1 << 0) +#define NV40TCL_TEX_WRAP(x) (0x00001a08+((x)*32)) +#define NV40TCL_TEX_WRAP__SIZE 0x00000010 +#define NV40TCL_TEX_WRAP_S_SHIFT 0 +#define NV40TCL_TEX_WRAP_S_MASK 0x000000ff +#define NV40TCL_TEX_WRAP_S_REPEAT 0x00000001 +#define NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV40TCL_TEX_WRAP_S_CLAMP 0x00000005 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE 0x00000006 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER 0x00000007 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP 0x00000008 +#define NV40TCL_TEX_WRAP_T_SHIFT 8 +#define NV40TCL_TEX_WRAP_T_MASK 0x0000ff00 +#define NV40TCL_TEX_WRAP_T_REPEAT 0x00000100 +#define NV40TCL_TEX_WRAP_T_MIRRORED_REPEAT 0x00000200 +#define NV40TCL_TEX_WRAP_T_CLAMP_TO_EDGE 0x00000300 +#define NV40TCL_TEX_WRAP_T_CLAMP_TO_BORDER 0x00000400 +#define NV40TCL_TEX_WRAP_T_CLAMP 0x00000500 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_EDGE 0x00000600 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_BORDER 0x00000700 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP 0x00000800 +#define NV40TCL_TEX_WRAP_R_SHIFT 16 +#define NV40TCL_TEX_WRAP_R_MASK 0x00ff0000 +#define NV40TCL_TEX_WRAP_R_REPEAT 0x00010000 +#define NV40TCL_TEX_WRAP_R_MIRRORED_REPEAT 0x00020000 +#define NV40TCL_TEX_WRAP_R_CLAMP_TO_EDGE 0x00030000 +#define NV40TCL_TEX_WRAP_R_CLAMP_TO_BORDER 0x00040000 +#define NV40TCL_TEX_WRAP_R_CLAMP 0x00050000 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE 0x00060000 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER 0x00070000 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP 0x00080000 +#define NV40TCL_TEX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV40TCL_TEX_ENABLE__SIZE 0x00000010 +#define NV40TCL_TEX_ENABLE_ENABLE (1 << 31) +#define NV40TCL_TEX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV40TCL_TEX_SWIZZLE__SIZE 0x00000010 +#define NV40TCL_TEX_SWIZZLE_S0_X_SHIFT 14 +#define NV40TCL_TEX_SWIZZLE_S0_X_MASK 0x0000c000 +#define NV40TCL_TEX_SWIZZLE_S0_X_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_X_ONE 0x00004000 +#define NV40TCL_TEX_SWIZZLE_S0_X_S1 0x00008000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT 12 +#define NV40TCL_TEX_SWIZZLE_S0_Y_MASK 0x00003000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_ONE 0x00001000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_S1 0x00002000 +#define NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT 10 +#define NV40TCL_TEX_SWIZZLE_S0_Z_MASK 0x00000c00 +#define NV40TCL_TEX_SWIZZLE_S0_Z_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_Z_ONE 0x00000400 +#define NV40TCL_TEX_SWIZZLE_S0_Z_S1 0x00000800 +#define NV40TCL_TEX_SWIZZLE_S0_W_SHIFT 8 +#define NV40TCL_TEX_SWIZZLE_S0_W_MASK 0x00000300 +#define NV40TCL_TEX_SWIZZLE_S0_W_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_W_ONE 0x00000100 +#define NV40TCL_TEX_SWIZZLE_S0_W_S1 0x00000200 +#define NV40TCL_TEX_SWIZZLE_S1_X_SHIFT 6 +#define NV40TCL_TEX_SWIZZLE_S1_X_MASK 0x000000c0 +#define NV40TCL_TEX_SWIZZLE_S1_X_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_X_Z 0x00000040 +#define NV40TCL_TEX_SWIZZLE_S1_X_Y 0x00000080 +#define NV40TCL_TEX_SWIZZLE_S1_X_X 0x000000c0 +#define NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT 4 +#define NV40TCL_TEX_SWIZZLE_S1_Y_MASK 0x00000030 +#define NV40TCL_TEX_SWIZZLE_S1_Y_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_Y_Z 0x00000010 +#define NV40TCL_TEX_SWIZZLE_S1_Y_Y 0x00000020 +#define NV40TCL_TEX_SWIZZLE_S1_Y_X 0x00000030 +#define NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT 2 +#define NV40TCL_TEX_SWIZZLE_S1_Z_MASK 0x0000000c +#define NV40TCL_TEX_SWIZZLE_S1_Z_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_Z_Z 0x00000004 +#define NV40TCL_TEX_SWIZZLE_S1_Z_Y 0x00000008 +#define NV40TCL_TEX_SWIZZLE_S1_Z_X 0x0000000c +#define NV40TCL_TEX_SWIZZLE_S1_W_SHIFT 0 +#define NV40TCL_TEX_SWIZZLE_S1_W_MASK 0x00000003 +#define NV40TCL_TEX_SWIZZLE_S1_W_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_W_Z 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S1_W_Y 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_W_X 0x00000003 +#define NV40TCL_TEX_FILTER(x) (0x00001a14+((x)*32)) +#define NV40TCL_TEX_FILTER__SIZE 0x00000010 +#define NV40TCL_TEX_FILTER_MIN_SHIFT 16 +#define NV40TCL_TEX_FILTER_MIN_MASK 0x000f0000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST 0x00010000 +#define NV40TCL_TEX_FILTER_MIN_LINEAR 0x00020000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST 0x00030000 +#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST 0x00040000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR 0x00050000 +#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR 0x00060000 +#define NV40TCL_TEX_FILTER_MAG_SHIFT 24 +#define NV40TCL_TEX_FILTER_MAG_MASK 0x0f000000 +#define NV40TCL_TEX_FILTER_MAG_NEAREST 0x01000000 +#define NV40TCL_TEX_FILTER_MAG_LINEAR 0x02000000 +#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV40TCL_TEX_SIZE0(x) (0x00001a18+((x)*32)) +#define NV40TCL_TEX_SIZE0__SIZE 0x00000010 +#define NV40TCL_TEX_SIZE0_H_SHIFT 0 +#define NV40TCL_TEX_SIZE0_H_MASK 0x0000ffff +#define NV40TCL_TEX_SIZE0_W_SHIFT 16 +#define NV40TCL_TEX_SIZE0_W_MASK 0xffff0000 +#define NV40TCL_TEX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV40TCL_TEX_BORDER_COLOR__SIZE 0x00000010 +#define NV40TCL_TEX_BORDER_COLOR_B_SHIFT 0 +#define NV40TCL_TEX_BORDER_COLOR_B_MASK 0x000000ff +#define NV40TCL_TEX_BORDER_COLOR_G_SHIFT 8 +#define NV40TCL_TEX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV40TCL_TEX_BORDER_COLOR_R_SHIFT 16 +#define NV40TCL_TEX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV40TCL_TEX_BORDER_COLOR_A_SHIFT 24 +#define NV40TCL_TEX_BORDER_COLOR_A_MASK 0xff000000 +#define NV40TCL_VTX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_W__SIZE 0x00000010 +#define NV40TCL_FP_CONTROL 0x00001d60 +#define NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT 24 +#define NV40TCL_FP_CONTROL_TEMP_COUNT_MASK 0xff000000 +#define NV40TCL_FP_CONTROL_KIL (1 << 7) +#define NV40TCL_FENCE_OFFSET 0x00001d6c +#define NV40TCL_FENCE_VALUE 0x00001d70 +#define NV40TCL_CLEAR_VALUE_DEPTH 0x00001d8c +#define NV40TCL_CLEAR_VALUE_COLOR 0x00001d90 +#define NV40TCL_CLEAR_BUFFERS 0x00001d94 +#define NV40TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) +#define NV40TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) +#define NV40TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) +#define NV40TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) +#define NV40TCL_CLEAR_BUFFERS_STENCIL (1 << 1) +#define NV40TCL_CLEAR_BUFFERS_DEPTH (1 << 0) +#define NV40TCL_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV40TCL_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV40TCL_VP_UPLOAD_FROM_ID 0x00001e9c +#define NV40TCL_VP_START_FROM_ID 0x00001ea0 +#define NV40TCL_POINT_SIZE 0x00001ee0 +#define NV40TCL_POINT_SPRITE 0x00001ee8 +#define NV40TCL_VP_UPLOAD_CONST_ID 0x00001efc +#define NV40TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 +#define NV40TCL_TEX_CACHE_CTL 0x00001fd8 +#define NV40TCL_VP_ATTRIB_EN 0x00001ff0 +#define NV40TCL_VP_RESULT_EN 0x00001ff4 + + +#define NV44TCL 0x00004497 + + + +#define NV50_2D 0x0000502d + +#define NV50_2D_NOP 0x00000100 +#define NV50_2D_NOTIFY 0x00000104 +#define NV50_2D_DMA_NOTIFY 0x00000180 +#define NV50_2D_DMA_IN_MEMORY0 0x00000184 +#define NV50_2D_DMA_IN_MEMORY1 0x00000188 +#define NV50_2D_DMA_IN_MEMORY2 0x0000018c +#define NV50_2D_DST_FORMAT 0x00000200 +#define NV50_2D_DST_FORMAT_32BPP 0x000000cf +#define NV50_2D_DST_FORMAT_24BPP 0x000000e6 +#define NV50_2D_DST_FORMAT_16BPP 0x000000e8 +#define NV50_2D_DST_FORMAT_8BPP 0x000000f3 +#define NV50_2D_DST_FORMAT_15BPP 0x000000f8 +#define NV50_2D_DST_PITCH 0x00000214 +#define NV50_2D_DST_WIDTH 0x00000218 +#define NV50_2D_DST_HEIGHT 0x0000021c +#define NV50_2D_DST_ADDRESS_HIGH 0x00000220 +#define NV50_2D_DST_ADDRESS_LOW 0x00000224 +#define NV50_2D_SRC_FORMAT 0x00000230 +#define NV50_2D_SRC_FORMAT_32BPP 0x000000cf +#define NV50_2D_SRC_FORMAT_24BPP 0x000000e6 +#define NV50_2D_SRC_FORMAT_16BPP 0x000000e8 +#define NV50_2D_SRC_FORMAT_8BPP 0x000000f3 +#define NV50_2D_SRC_FORMAT_15BPP 0x000000f8 +#define NV50_2D_SRC_PITCH 0x00000244 +#define NV50_2D_SRC_WIDTH 0x00000248 +#define NV50_2D_SRC_HEIGHT 0x0000024c +#define NV50_2D_SRC_ADDRESS_HIGH 0x00000250 +#define NV50_2D_SRC_ADDRESS_LOW 0x00000254 +#define NV50_2D_CLIP_X 0x00000280 +#define NV50_2D_CLIP_Y 0x00000284 +#define NV50_2D_CLIP_Z 0x00000288 +#define NV50_2D_CLIP_W 0x0000028c +#define NV50_2D_ROP 0x000002a0 +#define NV50_2D_OPERATION 0x000002ac +#define NV50_2D_OPERATION_SRCCOPY_AND 0x00000000 +#define NV50_2D_OPERATION_ROP_AND 0x00000001 +#define NV50_2D_OPERATION_BLEND_AND 0x00000002 +#define NV50_2D_OPERATION_SRCCOPY 0x00000003 +#define NV50_2D_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV50_2D_OPERATION_BLEND_PREMULT 0x00000005 +#define NV50_2D_PATTERN_FORMAT 0x000002e8 +#define NV50_2D_PATTERN_FORMAT_16BPP 0x00000000 +#define NV50_2D_PATTERN_FORMAT_15BPP 0x00000001 +#define NV50_2D_PATTERN_FORMAT_32BPP 0x00000002 +#define NV50_2D_PATTERN_FORMAT_8BPP 0x00000003 +#define NV50_2D_PATTERN_COLOR(x) (0x000002f0+((x)*4)) +#define NV50_2D_PATTERN_COLOR__SIZE 0x00000002 +#define NV50_2D_PATTERN_BITMAP(x) (0x000002f8+((x)*4)) +#define NV50_2D_PATTERN_BITMAP__SIZE 0x00000002 +#define NV50_2D_RECT_FORMAT 0x00000584 +#define NV50_2D_RECT_FORMAT_32BPP 0x000000cf +#define NV50_2D_RECT_FORMAT_24BPP 0x000000e6 +#define NV50_2D_RECT_FORMAT_16BPP 0x000000e8 +#define NV50_2D_RECT_FORMAT_8BPP 0x000000f3 +#define NV50_2D_RECT_FORMAT_15BPP 0x000000f8 +#define NV50_2D_RECT_COLOR 0x00000588 +#define NV50_2D_RECT_X1 0x00000600 +#define NV50_2D_RECT_Y1 0x00000604 +#define NV50_2D_RECT_X2 0x00000608 +#define NV50_2D_RECT_Y2 0x0000060c +#define NV50_2D_BLIT_DST_X 0x000008b0 +#define NV50_2D_BLIT_DST_Y 0x000008b4 +#define NV50_2D_BLIT_DST_W 0x000008b8 +#define NV50_2D_BLIT_DST_H 0x000008bc +#define NV50_2D_BLIT_SRC_X 0x000008d4 +#define NV50_2D_BLIT_SRC_Y 0x000008dc + + +#define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 + + + +#define NV50TCL 0x00005097 + +#define NV50TCL_NOP 0x00000100 +#define NV50TCL_NOTIFY 0x00000104 +#define NV50TCL_DMA_NOTIFY 0x00000180 +#define NV50TCL_DMA_IN_MEMORY0(x) (0x00000184+((x)*4)) +#define NV50TCL_DMA_IN_MEMORY0__SIZE 0x0000000b +#define NV50TCL_DMA_IN_MEMORY1(x) (0x000001c0+((x)*4)) +#define NV50TCL_DMA_IN_MEMORY1__SIZE 0x00000008 +#define NV50TCL_RT_ADDRESS_HIGH(x) (0x00000200+((x)*32)) +#define NV50TCL_RT_ADDRESS_HIGH__SIZE 0x00000008 +#define NV50TCL_RT_ADDRESS_LOW(x) (0x00000204+((x)*32)) +#define NV50TCL_RT_ADDRESS_LOW__SIZE 0x00000008 +#define NV50TCL_RT_FORMAT(x) (0x00000208+((x)*32)) +#define NV50TCL_RT_FORMAT__SIZE 0x00000008 +#define NV50TCL_RT_UNK3(x) (0x0000020c+((x)*32)) +#define NV50TCL_RT_UNK3__SIZE 0x00000008 +#define NV50TCL_RT_UNK4(x) (0x00000210+((x)*32)) +#define NV50TCL_RT_UNK4__SIZE 0x00000008 +#define NV50TCL_VTX_ATTR_1F(x) (0x00000300+((x)*4)) +#define NV50TCL_VTX_ATTR_1F__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2F_X(x) (0x00000380+((x)*8)) +#define NV50TCL_VTX_ATTR_2F_X__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2F_Y(x) (0x00000384+((x)*8)) +#define NV50TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_X(x) (0x00000400+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_X__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_Y(x) (0x00000404+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_Z(x) (0x00000408+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_W(x) (0x0000040c+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_W__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_X(x) (0x00000500+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_X__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_Y(x) (0x00000504+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_Z(x) (0x00000508+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_W(x) (0x0000050c+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_W__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2I(x) (0x00000680+((x)*4)) +#define NV50TCL_VTX_ATTR_2I__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2I_X_SHIFT 0 +#define NV50TCL_VTX_ATTR_2I_X_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV50TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4I_0(x) (0x00000700+((x)*8)) +#define NV50TCL_VTX_ATTR_4I_0__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4I_0_X_SHIFT 0 +#define NV50TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4I_0_Y_SHIFT 16 +#define NV50TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4I_1(x) (0x00000704+((x)*8)) +#define NV50TCL_VTX_ATTR_4I_1__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4I_1_Z_SHIFT 0 +#define NV50TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4I_1_W_SHIFT 16 +#define NV50TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4NI_0(x) (0x00000780+((x)*8)) +#define NV50TCL_VTX_ATTR_4NI_0__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4NI_0_X_SHIFT 0 +#define NV50TCL_VTX_ATTR_4NI_0_X_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4NI_0_Y_SHIFT 16 +#define NV50TCL_VTX_ATTR_4NI_0_Y_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4NI_1(x) (0x00000784+((x)*8)) +#define NV50TCL_VTX_ATTR_4NI_1__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4NI_1_Z_SHIFT 0 +#define NV50TCL_VTX_ATTR_4NI_1_Z_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4NI_1_W_SHIFT 16 +#define NV50TCL_VTX_ATTR_4NI_1_W_MASK 0xffff0000 +#define NV50TCL_VERTEX_ARRAY_FORMAT(x) (0x00000900+((x)*16)) +#define NV50TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV50TCL_VIEWPORT_UNK0(x) (0x00000a00+((x)*4)) +#define NV50TCL_VIEWPORT_UNK0__SIZE 0x00000003 +#define NV50TCL_VIEWPORT_UNK1(x) (0x00000a0c+((x)*4)) +#define NV50TCL_VIEWPORT_UNK1__SIZE 0x00000003 +#define NV50TCL_VIEWPORT_HORIZ 0x00000c00 +#define NV50TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV50TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV50TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV50TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV50TCL_VIEWPORT_VERT 0x00000c04 +#define NV50TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV50TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV50TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV50TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV50TCL_DEPTH_RANGE_NEAR 0x00000c08 +#define NV50TCL_DEPTH_RANGE_FAR 0x00000c0c +#define NV50TCL_VIEWPORT_CLIP_HORIZ(x) (0x00000d00+((x)*8)) +#define NV50TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV50TCL_VIEWPORT_CLIP_VERT(x) (0x00000d04+((x)*8)) +#define NV50TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV50TCL_VERTEX_BUFFER_FIRST 0x00000d74 +#define NV50TCL_VERTEX_BUFFER_COUNT 0x00000d78 +#define NV50TCL_CLEAR_COLOR(x) (0x00000d80+((x)*4)) +#define NV50TCL_CLEAR_COLOR__SIZE 0x00000004 +#define NV50TCL_CLEAR_DEPTH 0x00000d90 +#define NV50TCL_CLEAR_STENCIL 0x00000da0 +#define NV50TCL_POLYGON_MODE_FRONT 0x00000dac +#define NV50TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV50TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV50TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV50TCL_POLYGON_MODE_BACK 0x00000db0 +#define NV50TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV50TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV50TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV50TCL_POLYGON_SMOOTH_ENABLE 0x00000db4 +#define NV50TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000dc0 +#define NV50TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000dc4 +#define NV50TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000dc8 +#define NV50TCL_SCISSOR_HORIZ 0x00000e04 +#define NV50TCL_SCISSOR_HORIZ_L_SHIFT 0 +#define NV50TCL_SCISSOR_HORIZ_L_MASK 0x0000ffff +#define NV50TCL_SCISSOR_HORIZ_R_SHIFT 16 +#define NV50TCL_SCISSOR_HORIZ_R_MASK 0xffff0000 +#define NV50TCL_SCISSOR_VERT 0x00000e08 +#define NV50TCL_SCISSOR_VERT_T_SHIFT 0 +#define NV50TCL_SCISSOR_VERT_T_MASK 0x0000ffff +#define NV50TCL_SCISSOR_VERT_B_SHIFT 16 +#define NV50TCL_SCISSOR_VERT_B_MASK 0xffff0000 +#define NV50TCL_VP_UPLOAD_CONST_ID 0x00000f00 +#define NV50TCL_VP_UPLOAD_CONST(x) (0x00000f04+((x)*4)) +#define NV50TCL_VP_UPLOAD_CONST__SIZE 0x00000010 +#define NV50TCL_STENCIL_FRONT_FUNC_REF 0x00000f54 +#define NV50TCL_STENCIL_FRONT_MASK 0x00000f58 +#define NV50TCL_STENCIL_FRONT_FUNC_MASK 0x00000f5c +#define NV50TCL_GP_ADDRESS_HIGH 0x00000f70 +#define NV50TCL_GP_ADDRESS_LOW 0x00000f74 +#define NV50TCL_VP_ADDRESS_HIGH 0x00000f7c +#define NV50TCL_VP_ADDRESS_LOW 0x00000f80 +#define NV50TCL_FP_ADDRESS_HIGH 0x00000fa4 +#define NV50TCL_FP_ADDRESS_LOW 0x00000fa8 +#define NV50TCL_ZETA_ADDRESS_HIGH 0x00000fe0 +#define NV50TCL_ZETA_ADDRESS_LOW 0x00000fe4 +#define NV50TCL_UNKFF4 0x00000ff4 +#define NV50TCL_UNKFF4_W_SHIFT 16 +#define NV50TCL_UNKFF4_W_MASK 0xffff0000 +#define NV50TCL_UNKFF8 0x00000ff8 +#define NV50TCL_UNKFF8_H_SHIFT 16 +#define NV50TCL_UNKFF8_H_MASK 0xffff0000 +#define NV50TCL_RT_HORIZ(x) (0x00001240+((x)*8)) +#define NV50TCL_RT_HORIZ__SIZE 0x00000008 +#define NV50TCL_RT_VERT(x) (0x00001244+((x)*8)) +#define NV50TCL_RT_VERT__SIZE 0x00000008 +#define NV50TCL_DEPTH_TEST_ENABLE 0x000012cc +#define NV50TCL_SHADE_MODEL 0x000012d4 +#define NV50TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV50TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV50TCL_DEPTH_WRITE_ENABLE 0x000012e8 +#define NV50TCL_ALPHA_TEST_ENABLE 0x000012ec +#define NV50TCL_DEPTH_TEST_FUNC 0x0000130c +#define NV50TCL_DEPTH_TEST_FUNC_NEVER 0x00000200 +#define NV50TCL_DEPTH_TEST_FUNC_LESS 0x00000201 +#define NV50TCL_DEPTH_TEST_FUNC_EQUAL 0x00000202 +#define NV50TCL_DEPTH_TEST_FUNC_LEQUAL 0x00000203 +#define NV50TCL_DEPTH_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_DEPTH_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_DEPTH_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_DEPTH_TEST_FUNC_GEQUAL 0x00000206 +#define NV50TCL_DEPTH_TEST_FUNC_ALWAYS 0x00000207 +#define NV50TCL_ALPHA_TEST_REF 0x00001310 +#define NV50TCL_ALPHA_TEST_FUNC 0x00001314 +#define NV50TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 +#define NV50TCL_ALPHA_TEST_FUNC_LESS 0x00000201 +#define NV50TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 +#define NV50TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 +#define NV50TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 +#define NV50TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 +#define NV50TCL_BLEND_COLOR(x) (0x0000131c+((x)*4)) +#define NV50TCL_BLEND_COLOR__SIZE 0x00000004 +#define NV50TCL_BLEND_EQUATION_RGB 0x00001340 +#define NV50TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 +#define NV50TCL_BLEND_EQUATION_RGB_MIN 0x00008007 +#define NV50TCL_BLEND_EQUATION_RGB_MAX 0x00008008 +#define NV50TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a +#define NV50TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV50TCL_BLEND_FUNC_SRC_RGB 0x00001344 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_FUNC_DST_RGB 0x00001348 +#define NV50TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_EQUATION_ALPHA 0x0000134c +#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 +#define NV50TCL_BLEND_EQUATION_ALPHA_MIN 0x00008007 +#define NV50TCL_BLEND_EQUATION_ALPHA_MAX 0x00008008 +#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a +#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV50TCL_BLEND_FUNC_SRC_ALPHA 0x00001350 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_FUNC_DST_ALPHA 0x00001358 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_ENABLE(x) (0x00001360+((x)*4)) +#define NV50TCL_BLEND_ENABLE__SIZE 0x00000008 +#define NV50TCL_STENCIL_BACK_ENABLE 0x00001380 +#define NV50TCL_STENCIL_BACK_OP_FAIL 0x00001384 +#define NV50TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL 0x00001388 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_BACK_OP_ZPASS 0x0000138c +#define NV50TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV50TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC 0x00001390 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50TCL_STENCIL_BACK_FUNC_REF 0x00001394 +#define NV50TCL_STENCIL_BACK_MASK 0x00001398 +#define NV50TCL_STENCIL_BACK_FUNC_MASK 0x0000139c +#define NV50TCL_LINE_WIDTH 0x000013b0 +#define NV50TCL_VP_START_ID 0x0000140c +#define NV50TCL_GP_START_ID 0x00001410 +#define NV50TCL_FP_START_ID 0x00001414 +#define NV50TCL_POINT_SIZE 0x00001518 +#define NV50TCL_TEX_CB0_ADDRESS_HIGH 0x0000155c +#define NV50TCL_TEX_CB0_ADDRESS_LOW 0x00001560 +#define NV50TCL_POLYGON_OFFSET_FACTOR 0x0000156c +#define NV50TCL_LINE_SMOOTH_ENABLE 0x00001570 +#define NV50TCL_TEX_CB1_ADDRESS_HIGH 0x00001574 +#define NV50TCL_TEX_CB1_ADDRESS_LOW 0x00001578 +#define NV50TCL_STENCIL_FRONT_ENABLE 0x00001594 +#define NV50TCL_STENCIL_FRONT_OP_FAIL 0x00001598 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL 0x0000159c +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS 0x000015a0 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC 0x000015a4 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50TCL_POLYGON_OFFSET_UNITS 0x000015bc +#define NV50TCL_VERTEX_BEGIN 0x000015dc +#define NV50TCL_VERTEX_BEGIN_POINTS 0x00000000 +#define NV50TCL_VERTEX_BEGIN_LINES 0x00000001 +#define NV50TCL_VERTEX_BEGIN_LINE_LOOP 0x00000002 +#define NV50TCL_VERTEX_BEGIN_LINE_STRIP 0x00000003 +#define NV50TCL_VERTEX_BEGIN_TRIANGLES 0x00000004 +#define NV50TCL_VERTEX_BEGIN_TRIANGLE_STRIP 0x00000005 +#define NV50TCL_VERTEX_BEGIN_TRIANGLE_FAN 0x00000006 +#define NV50TCL_VERTEX_BEGIN_QUADS 0x00000007 +#define NV50TCL_VERTEX_BEGIN_QUAD_STRIP 0x00000008 +#define NV50TCL_VERTEX_BEGIN_POLYGON 0x00000009 +#define NV50TCL_VERTEX_END 0x000015e0 +#define NV50TCL_VERTEX_DATA 0x00001640 +#define NV50TCL_VP_ATTR_EN_0 0x00001650 +#define NV50TCL_VP_ATTR_EN_0_7_SHIFT 28 +#define NV50TCL_VP_ATTR_EN_0_7_MASK 0xf0000000 +#define NV50TCL_VP_ATTR_EN_0_7_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNNN 0x10000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYNN 0x20000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYNN 0x30000000 +#define NV50TCL_VP_ATTR_EN_0_7_NNZN 0x40000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNZN 0x50000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYZN 0x60000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYZN 0x70000000 +#define NV50TCL_VP_ATTR_EN_0_7_NNNW 0x80000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNNW 0x90000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYNW 0xa0000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYNW 0xb0000000 +#define NV50TCL_VP_ATTR_EN_0_7_NNZW 0xc0000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNZW 0xd0000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYZW 0xe0000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYZW 0xf0000000 +#define NV50TCL_VP_ATTR_EN_0_6_SHIFT 24 +#define NV50TCL_VP_ATTR_EN_0_6_MASK 0x0f000000 +#define NV50TCL_VP_ATTR_EN_0_6_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNNN 0x01000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYNN 0x02000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYNN 0x03000000 +#define NV50TCL_VP_ATTR_EN_0_6_NNZN 0x04000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNZN 0x05000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYZN 0x06000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYZN 0x07000000 +#define NV50TCL_VP_ATTR_EN_0_6_NNNW 0x08000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNNW 0x09000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYNW 0x0a000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYNW 0x0b000000 +#define NV50TCL_VP_ATTR_EN_0_6_NNZW 0x0c000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNZW 0x0d000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYZW 0x0e000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYZW 0x0f000000 +#define NV50TCL_VP_ATTR_EN_0_5_SHIFT 20 +#define NV50TCL_VP_ATTR_EN_0_5_MASK 0x00f00000 +#define NV50TCL_VP_ATTR_EN_0_5_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_5_XNNN 0x00100000 +#define NV50TCL_VP_ATTR_EN_0_5_NYNN 0x00200000 +#define NV50TCL_VP_ATTR_EN_0_5_XYNN 0x00300000 +#define NV50TCL_VP_ATTR_EN_0_5_NNZN 0x00400000 +#define NV50TCL_VP_ATTR_EN_0_5_XNZN 0x00500000 +#define NV50TCL_VP_ATTR_EN_0_5_NYZN 0x00600000 +#define NV50TCL_VP_ATTR_EN_0_5_XYZN 0x00700000 +#define NV50TCL_VP_ATTR_EN_0_5_NNNW 0x00800000 +#define NV50TCL_VP_ATTR_EN_0_5_XNNW 0x00900000 +#define NV50TCL_VP_ATTR_EN_0_5_NYNW 0x00a00000 +#define NV50TCL_VP_ATTR_EN_0_5_XYNW 0x00b00000 +#define NV50TCL_VP_ATTR_EN_0_5_NNZW 0x00c00000 +#define NV50TCL_VP_ATTR_EN_0_5_XNZW 0x00d00000 +#define NV50TCL_VP_ATTR_EN_0_5_NYZW 0x00e00000 +#define NV50TCL_VP_ATTR_EN_0_5_XYZW 0x00f00000 +#define NV50TCL_VP_ATTR_EN_0_4_SHIFT 16 +#define NV50TCL_VP_ATTR_EN_0_4_MASK 0x000f0000 +#define NV50TCL_VP_ATTR_EN_0_4_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_4_XNNN 0x00010000 +#define NV50TCL_VP_ATTR_EN_0_4_NYNN 0x00020000 +#define NV50TCL_VP_ATTR_EN_0_4_XYNN 0x00030000 +#define NV50TCL_VP_ATTR_EN_0_4_NNZN 0x00040000 +#define NV50TCL_VP_ATTR_EN_0_4_XNZN 0x00050000 +#define NV50TCL_VP_ATTR_EN_0_4_NYZN 0x00060000 +#define NV50TCL_VP_ATTR_EN_0_4_XYZN 0x00070000 +#define NV50TCL_VP_ATTR_EN_0_4_NNNW 0x00080000 +#define NV50TCL_VP_ATTR_EN_0_4_XNNW 0x00090000 +#define NV50TCL_VP_ATTR_EN_0_4_NYNW 0x000a0000 +#define NV50TCL_VP_ATTR_EN_0_4_XYNW 0x000b0000 +#define NV50TCL_VP_ATTR_EN_0_4_NNZW 0x000c0000 +#define NV50TCL_VP_ATTR_EN_0_4_XNZW 0x000d0000 +#define NV50TCL_VP_ATTR_EN_0_4_NYZW 0x000e0000 +#define NV50TCL_VP_ATTR_EN_0_4_XYZW 0x000f0000 +#define NV50TCL_VP_ATTR_EN_0_3_SHIFT 12 +#define NV50TCL_VP_ATTR_EN_0_3_MASK 0x0000f000 +#define NV50TCL_VP_ATTR_EN_0_3_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_3_XNNN 0x00001000 +#define NV50TCL_VP_ATTR_EN_0_3_NYNN 0x00002000 +#define NV50TCL_VP_ATTR_EN_0_3_XYNN 0x00003000 +#define NV50TCL_VP_ATTR_EN_0_3_NNZN 0x00004000 +#define NV50TCL_VP_ATTR_EN_0_3_XNZN 0x00005000 +#define NV50TCL_VP_ATTR_EN_0_3_NYZN 0x00006000 +#define NV50TCL_VP_ATTR_EN_0_3_XYZN 0x00007000 +#define NV50TCL_VP_ATTR_EN_0_3_NNNW 0x00008000 +#define NV50TCL_VP_ATTR_EN_0_3_XNNW 0x00009000 +#define NV50TCL_VP_ATTR_EN_0_3_NYNW 0x0000a000 +#define NV50TCL_VP_ATTR_EN_0_3_XYNW 0x0000b000 +#define NV50TCL_VP_ATTR_EN_0_3_NNZW 0x0000c000 +#define NV50TCL_VP_ATTR_EN_0_3_XNZW 0x0000d000 +#define NV50TCL_VP_ATTR_EN_0_3_NYZW 0x0000e000 +#define NV50TCL_VP_ATTR_EN_0_3_XYZW 0x0000f000 +#define NV50TCL_VP_ATTR_EN_0_2_SHIFT 8 +#define NV50TCL_VP_ATTR_EN_0_2_MASK 0x00000f00 +#define NV50TCL_VP_ATTR_EN_0_2_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_2_XNNN 0x00000100 +#define NV50TCL_VP_ATTR_EN_0_2_NYNN 0x00000200 +#define NV50TCL_VP_ATTR_EN_0_2_XYNN 0x00000300 +#define NV50TCL_VP_ATTR_EN_0_2_NNZN 0x00000400 +#define NV50TCL_VP_ATTR_EN_0_2_XNZN 0x00000500 +#define NV50TCL_VP_ATTR_EN_0_2_NYZN 0x00000600 +#define NV50TCL_VP_ATTR_EN_0_2_XYZN 0x00000700 +#define NV50TCL_VP_ATTR_EN_0_2_NNNW 0x00000800 +#define NV50TCL_VP_ATTR_EN_0_2_XNNW 0x00000900 +#define NV50TCL_VP_ATTR_EN_0_2_NYNW 0x00000a00 +#define NV50TCL_VP_ATTR_EN_0_2_XYNW 0x00000b00 +#define NV50TCL_VP_ATTR_EN_0_2_NNZW 0x00000c00 +#define NV50TCL_VP_ATTR_EN_0_2_XNZW 0x00000d00 +#define NV50TCL_VP_ATTR_EN_0_2_NYZW 0x00000e00 +#define NV50TCL_VP_ATTR_EN_0_2_XYZW 0x00000f00 +#define NV50TCL_VP_ATTR_EN_0_1_SHIFT 4 +#define NV50TCL_VP_ATTR_EN_0_1_MASK 0x000000f0 +#define NV50TCL_VP_ATTR_EN_0_1_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_1_XNNN 0x00000010 +#define NV50TCL_VP_ATTR_EN_0_1_NYNN 0x00000020 +#define NV50TCL_VP_ATTR_EN_0_1_XYNN 0x00000030 +#define NV50TCL_VP_ATTR_EN_0_1_NNZN 0x00000040 +#define NV50TCL_VP_ATTR_EN_0_1_XNZN 0x00000050 +#define NV50TCL_VP_ATTR_EN_0_1_NYZN 0x00000060 +#define NV50TCL_VP_ATTR_EN_0_1_XYZN 0x00000070 +#define NV50TCL_VP_ATTR_EN_0_1_NNNW 0x00000080 +#define NV50TCL_VP_ATTR_EN_0_1_XNNW 0x00000090 +#define NV50TCL_VP_ATTR_EN_0_1_NYNW 0x000000a0 +#define NV50TCL_VP_ATTR_EN_0_1_XYNW 0x000000b0 +#define NV50TCL_VP_ATTR_EN_0_1_NNZW 0x000000c0 +#define NV50TCL_VP_ATTR_EN_0_1_XNZW 0x000000d0 +#define NV50TCL_VP_ATTR_EN_0_1_NYZW 0x000000e0 +#define NV50TCL_VP_ATTR_EN_0_1_XYZW 0x000000f0 +#define NV50TCL_VP_ATTR_EN_0_0_SHIFT 0 +#define NV50TCL_VP_ATTR_EN_0_0_MASK 0x0000000f +#define NV50TCL_VP_ATTR_EN_0_0_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_0_XNNN 0x00000001 +#define NV50TCL_VP_ATTR_EN_0_0_NYNN 0x00000002 +#define NV50TCL_VP_ATTR_EN_0_0_XYNN 0x00000003 +#define NV50TCL_VP_ATTR_EN_0_0_NNZN 0x00000004 +#define NV50TCL_VP_ATTR_EN_0_0_XNZN 0x00000005 +#define NV50TCL_VP_ATTR_EN_0_0_NYZN 0x00000006 +#define NV50TCL_VP_ATTR_EN_0_0_XYZN 0x00000007 +#define NV50TCL_VP_ATTR_EN_0_0_NNNW 0x00000008 +#define NV50TCL_VP_ATTR_EN_0_0_XNNW 0x00000009 +#define NV50TCL_VP_ATTR_EN_0_0_NYNW 0x0000000a +#define NV50TCL_VP_ATTR_EN_0_0_XYNW 0x0000000b +#define NV50TCL_VP_ATTR_EN_0_0_NNZW 0x0000000c +#define NV50TCL_VP_ATTR_EN_0_0_XNZW 0x0000000d +#define NV50TCL_VP_ATTR_EN_0_0_NYZW 0x0000000e +#define NV50TCL_VP_ATTR_EN_0_0_XYZW 0x0000000f +#define NV50TCL_VP_ATTR_EN_1 0x00001654 +#define NV50TCL_VP_ATTR_EN_1_15_SHIFT 28 +#define NV50TCL_VP_ATTR_EN_1_15_MASK 0xf0000000 +#define NV50TCL_VP_ATTR_EN_1_15_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNNN 0x10000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYNN 0x20000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYNN 0x30000000 +#define NV50TCL_VP_ATTR_EN_1_15_NNZN 0x40000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNZN 0x50000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYZN 0x60000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYZN 0x70000000 +#define NV50TCL_VP_ATTR_EN_1_15_NNNW 0x80000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNNW 0x90000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYNW 0xa0000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYNW 0xb0000000 +#define NV50TCL_VP_ATTR_EN_1_15_NNZW 0xc0000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNZW 0xd0000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYZW 0xe0000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYZW 0xf0000000 +#define NV50TCL_VP_ATTR_EN_1_14_SHIFT 24 +#define NV50TCL_VP_ATTR_EN_1_14_MASK 0x0f000000 +#define NV50TCL_VP_ATTR_EN_1_14_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNNN 0x01000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYNN 0x02000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYNN 0x03000000 +#define NV50TCL_VP_ATTR_EN_1_14_NNZN 0x04000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNZN 0x05000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYZN 0x06000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYZN 0x07000000 +#define NV50TCL_VP_ATTR_EN_1_14_NNNW 0x08000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNNW 0x09000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYNW 0x0a000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYNW 0x0b000000 +#define NV50TCL_VP_ATTR_EN_1_14_NNZW 0x0c000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNZW 0x0d000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYZW 0x0e000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYZW 0x0f000000 +#define NV50TCL_VP_ATTR_EN_1_13_SHIFT 20 +#define NV50TCL_VP_ATTR_EN_1_13_MASK 0x00f00000 +#define NV50TCL_VP_ATTR_EN_1_13_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_13_XNNN 0x00100000 +#define NV50TCL_VP_ATTR_EN_1_13_NYNN 0x00200000 +#define NV50TCL_VP_ATTR_EN_1_13_XYNN 0x00300000 +#define NV50TCL_VP_ATTR_EN_1_13_NNZN 0x00400000 +#define NV50TCL_VP_ATTR_EN_1_13_XNZN 0x00500000 +#define NV50TCL_VP_ATTR_EN_1_13_NYZN 0x00600000 +#define NV50TCL_VP_ATTR_EN_1_13_XYZN 0x00700000 +#define NV50TCL_VP_ATTR_EN_1_13_NNNW 0x00800000 +#define NV50TCL_VP_ATTR_EN_1_13_XNNW 0x00900000 +#define NV50TCL_VP_ATTR_EN_1_13_NYNW 0x00a00000 +#define NV50TCL_VP_ATTR_EN_1_13_XYNW 0x00b00000 +#define NV50TCL_VP_ATTR_EN_1_13_NNZW 0x00c00000 +#define NV50TCL_VP_ATTR_EN_1_13_XNZW 0x00d00000 +#define NV50TCL_VP_ATTR_EN_1_13_NYZW 0x00e00000 +#define NV50TCL_VP_ATTR_EN_1_13_XYZW 0x00f00000 +#define NV50TCL_VP_ATTR_EN_1_12_SHIFT 16 +#define NV50TCL_VP_ATTR_EN_1_12_MASK 0x000f0000 +#define NV50TCL_VP_ATTR_EN_1_12_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_12_XNNN 0x00010000 +#define NV50TCL_VP_ATTR_EN_1_12_NYNN 0x00020000 +#define NV50TCL_VP_ATTR_EN_1_12_XYNN 0x00030000 +#define NV50TCL_VP_ATTR_EN_1_12_NNZN 0x00040000 +#define NV50TCL_VP_ATTR_EN_1_12_XNZN 0x00050000 +#define NV50TCL_VP_ATTR_EN_1_12_NYZN 0x00060000 +#define NV50TCL_VP_ATTR_EN_1_12_XYZN 0x00070000 +#define NV50TCL_VP_ATTR_EN_1_12_NNNW 0x00080000 +#define NV50TCL_VP_ATTR_EN_1_12_XNNW 0x00090000 +#define NV50TCL_VP_ATTR_EN_1_12_NYNW 0x000a0000 +#define NV50TCL_VP_ATTR_EN_1_12_XYNW 0x000b0000 +#define NV50TCL_VP_ATTR_EN_1_12_NNZW 0x000c0000 +#define NV50TCL_VP_ATTR_EN_1_12_XNZW 0x000d0000 +#define NV50TCL_VP_ATTR_EN_1_12_NYZW 0x000e0000 +#define NV50TCL_VP_ATTR_EN_1_12_XYZW 0x000f0000 +#define NV50TCL_VP_ATTR_EN_1_11_SHIFT 12 +#define NV50TCL_VP_ATTR_EN_1_11_MASK 0x0000f000 +#define NV50TCL_VP_ATTR_EN_1_11_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_11_XNNN 0x00001000 +#define NV50TCL_VP_ATTR_EN_1_11_NYNN 0x00002000 +#define NV50TCL_VP_ATTR_EN_1_11_XYNN 0x00003000 +#define NV50TCL_VP_ATTR_EN_1_11_NNZN 0x00004000 +#define NV50TCL_VP_ATTR_EN_1_11_XNZN 0x00005000 +#define NV50TCL_VP_ATTR_EN_1_11_NYZN 0x00006000 +#define NV50TCL_VP_ATTR_EN_1_11_XYZN 0x00007000 +#define NV50TCL_VP_ATTR_EN_1_11_NNNW 0x00008000 +#define NV50TCL_VP_ATTR_EN_1_11_XNNW 0x00009000 +#define NV50TCL_VP_ATTR_EN_1_11_NYNW 0x0000a000 +#define NV50TCL_VP_ATTR_EN_1_11_XYNW 0x0000b000 +#define NV50TCL_VP_ATTR_EN_1_11_NNZW 0x0000c000 +#define NV50TCL_VP_ATTR_EN_1_11_XNZW 0x0000d000 +#define NV50TCL_VP_ATTR_EN_1_11_NYZW 0x0000e000 +#define NV50TCL_VP_ATTR_EN_1_11_XYZW 0x0000f000 +#define NV50TCL_VP_ATTR_EN_1_10_SHIFT 8 +#define NV50TCL_VP_ATTR_EN_1_10_MASK 0x00000f00 +#define NV50TCL_VP_ATTR_EN_1_10_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_10_XNNN 0x00000100 +#define NV50TCL_VP_ATTR_EN_1_10_NYNN 0x00000200 +#define NV50TCL_VP_ATTR_EN_1_10_XYNN 0x00000300 +#define NV50TCL_VP_ATTR_EN_1_10_NNZN 0x00000400 +#define NV50TCL_VP_ATTR_EN_1_10_XNZN 0x00000500 +#define NV50TCL_VP_ATTR_EN_1_10_NYZN 0x00000600 +#define NV50TCL_VP_ATTR_EN_1_10_XYZN 0x00000700 +#define NV50TCL_VP_ATTR_EN_1_10_NNNW 0x00000800 +#define NV50TCL_VP_ATTR_EN_1_10_XNNW 0x00000900 +#define NV50TCL_VP_ATTR_EN_1_10_NYNW 0x00000a00 +#define NV50TCL_VP_ATTR_EN_1_10_XYNW 0x00000b00 +#define NV50TCL_VP_ATTR_EN_1_10_NNZW 0x00000c00 +#define NV50TCL_VP_ATTR_EN_1_10_XNZW 0x00000d00 +#define NV50TCL_VP_ATTR_EN_1_10_NYZW 0x00000e00 +#define NV50TCL_VP_ATTR_EN_1_10_XYZW 0x00000f00 +#define NV50TCL_VP_ATTR_EN_1_9_SHIFT 4 +#define NV50TCL_VP_ATTR_EN_1_9_MASK 0x000000f0 +#define NV50TCL_VP_ATTR_EN_1_9_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_9_XNNN 0x00000010 +#define NV50TCL_VP_ATTR_EN_1_9_NYNN 0x00000020 +#define NV50TCL_VP_ATTR_EN_1_9_XYNN 0x00000030 +#define NV50TCL_VP_ATTR_EN_1_9_NNZN 0x00000040 +#define NV50TCL_VP_ATTR_EN_1_9_XNZN 0x00000050 +#define NV50TCL_VP_ATTR_EN_1_9_NYZN 0x00000060 +#define NV50TCL_VP_ATTR_EN_1_9_XYZN 0x00000070 +#define NV50TCL_VP_ATTR_EN_1_9_NNNW 0x00000080 +#define NV50TCL_VP_ATTR_EN_1_9_XNNW 0x00000090 +#define NV50TCL_VP_ATTR_EN_1_9_NYNW 0x000000a0 +#define NV50TCL_VP_ATTR_EN_1_9_XYNW 0x000000b0 +#define NV50TCL_VP_ATTR_EN_1_9_NNZW 0x000000c0 +#define NV50TCL_VP_ATTR_EN_1_9_XNZW 0x000000d0 +#define NV50TCL_VP_ATTR_EN_1_9_NYZW 0x000000e0 +#define NV50TCL_VP_ATTR_EN_1_9_XYZW 0x000000f0 +#define NV50TCL_VP_ATTR_EN_1_8_SHIFT 0 +#define NV50TCL_VP_ATTR_EN_1_8_MASK 0x0000000f +#define NV50TCL_VP_ATTR_EN_1_8_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_8_XNNN 0x00000001 +#define NV50TCL_VP_ATTR_EN_1_8_NYNN 0x00000002 +#define NV50TCL_VP_ATTR_EN_1_8_XYNN 0x00000003 +#define NV50TCL_VP_ATTR_EN_1_8_NNZN 0x00000004 +#define NV50TCL_VP_ATTR_EN_1_8_XNZN 0x00000005 +#define NV50TCL_VP_ATTR_EN_1_8_NYZN 0x00000006 +#define NV50TCL_VP_ATTR_EN_1_8_XYZN 0x00000007 +#define NV50TCL_VP_ATTR_EN_1_8_NNNW 0x00000008 +#define NV50TCL_VP_ATTR_EN_1_8_XNNW 0x00000009 +#define NV50TCL_VP_ATTR_EN_1_8_NYNW 0x0000000a +#define NV50TCL_VP_ATTR_EN_1_8_XYNW 0x0000000b +#define NV50TCL_VP_ATTR_EN_1_8_NNZW 0x0000000c +#define NV50TCL_VP_ATTR_EN_1_8_XNZW 0x0000000d +#define NV50TCL_VP_ATTR_EN_1_8_NYZW 0x0000000e +#define NV50TCL_VP_ATTR_EN_1_8_XYZW 0x0000000f +#define NV50TCL_LINE_STIPPLE_ENABLE 0x0000166c +#define NV50TCL_LINE_STIPPLE_PATTERN 0x00001680 +#define NV50TCL_POLYGON_STIPPLE_ENABLE 0x0000168c +#define NV50TCL_VP_REG_HPOS 0x000016bc +#define NV50TCL_VP_REG_HPOS_X_SHIFT 0 +#define NV50TCL_VP_REG_HPOS_X_MASK 0x000000ff +#define NV50TCL_VP_REG_HPOS_Y_SHIFT 8 +#define NV50TCL_VP_REG_HPOS_Y_MASK 0x0000ff00 +#define NV50TCL_VP_REG_HPOS_Z_SHIFT 16 +#define NV50TCL_VP_REG_HPOS_Z_MASK 0x00ff0000 +#define NV50TCL_VP_REG_HPOS_W_SHIFT 24 +#define NV50TCL_VP_REG_HPOS_W_MASK 0xff000000 +#define NV50TCL_VP_REG_COL0 0x000016c0 +#define NV50TCL_VP_REG_COL0_X_SHIFT 0 +#define NV50TCL_VP_REG_COL0_X_MASK 0x000000ff +#define NV50TCL_VP_REG_COL0_Y_SHIFT 8 +#define NV50TCL_VP_REG_COL0_Y_MASK 0x0000ff00 +#define NV50TCL_VP_REG_COL0_Z_SHIFT 16 +#define NV50TCL_VP_REG_COL0_Z_MASK 0x00ff0000 +#define NV50TCL_VP_REG_COL0_W_SHIFT 24 +#define NV50TCL_VP_REG_COL0_W_MASK 0xff000000 +#define NV50TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001700+((x)*4)) +#define NV50TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV50TCL_CULL_FACE_ENABLE 0x00001918 +#define NV50TCL_FRONT_FACE 0x0000191c +#define NV50TCL_FRONT_FACE_CW 0x00000900 +#define NV50TCL_FRONT_FACE_CCW 0x00000901 +#define NV50TCL_CULL_FACE 0x00001920 +#define NV50TCL_CULL_FACE_FRONT 0x00000404 +#define NV50TCL_CULL_FACE_BACK 0x00000405 +#define NV50TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV50TCL_LOGIC_OP_ENABLE 0x000019c4 +#define NV50TCL_LOGIC_OP 0x000019c8 +#define NV50TCL_LOGIC_OP_CLEAR 0x00001500 +#define NV50TCL_LOGIC_OP_AND 0x00001501 +#define NV50TCL_LOGIC_OP_AND_REVERSE 0x00001502 +#define NV50TCL_LOGIC_OP_COPY 0x00001503 +#define NV50TCL_LOGIC_OP_AND_INVERTED 0x00001504 +#define NV50TCL_LOGIC_OP_NOOP 0x00001505 +#define NV50TCL_LOGIC_OP_XOR 0x00001506 +#define NV50TCL_LOGIC_OP_OR 0x00001507 +#define NV50TCL_LOGIC_OP_NOR 0x00001508 +#define NV50TCL_LOGIC_OP_EQUIV 0x00001509 +#define NV50TCL_LOGIC_OP_INVERT 0x0000150a +#define NV50TCL_LOGIC_OP_OR_REVERSE 0x0000150b +#define NV50TCL_LOGIC_OP_COPY_INVERTED 0x0000150c +#define NV50TCL_LOGIC_OP_OR_INVERTED 0x0000150d +#define NV50TCL_LOGIC_OP_NAND 0x0000150e +#define NV50TCL_LOGIC_OP_SET 0x0000150f +#define NV50TCL_CLEAR_BUFFERS 0x000019d0 +#define NV50TCL_COLOR_MASK(x) (0x00001a00+((x)*4)) +#define NV50TCL_COLOR_MASK__SIZE 0x00000008 + + +#define NV54TCL 0x00008297 diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index e38a5ea534a..a9790484b9b 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -119,21 +119,101 @@ nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) free(hwcso); } + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV40TCL_TEX_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV40TCL_TEX_WRAP_S_CLAMP; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP; + break; + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV40TCL_TEX_WRAP_S_REPEAT; + break; + } + + return ret >> NV40TCL_TEX_WRAP_S_SHIFT; +} + static void * nv40_sampler_state_create(struct pipe_context *pipe, const struct pipe_sampler_state *cso) { struct nv40_sampler_state *ps; - + uint32_t filter = 0; + ps = malloc(sizeof(struct nv40_sampler_state)); - ps->wrap = ((nv40_tex_wrap_mode(cso->wrap_r) << 16) | - (nv40_tex_wrap_mode(cso->wrap_t) << 8) | - (nv40_tex_wrap_mode(cso->wrap_s) << 0)); - ps->filt = ((nv40_tex_filter(cso->min_img_filter, - cso->min_mip_filter) << 16) | - (nv40_tex_filter(cso->mag_img_filter, - PIPE_TEX_MIPFILTER_NONE) << 24)); + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV40TCL_TEX_FILTER_MAG_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV40TCL_TEX_FILTER_MAG_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV40TCL_TEX_FILTER_MIN_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV40TCL_TEX_FILTER_MIN_NEAREST; + break; + } + break; + } + + + ps->wrap = ((wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) | + (wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_R_SHIFT)); + ps->filt = filter; ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | (float_to_ubyte(cso->border_color[0]) << 16) | (float_to_ubyte(cso->border_color[1]) << 8) | @@ -497,33 +577,23 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, OUT_RING (rt_enable); if (0) { -#if 0 - rt_format |= (log2width << - NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT); - rt_format |= (log2height << - NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT); -#endif - rt_format |= (NV40TCL_RT_FORMAT_TYPE_SWIZZLED << - NV40TCL_RT_FORMAT_TYPE_SHIFT); + rt_format |= (0 << NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT); + rt_format |= (0 << NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT); + rt_format |= NV40TCL_RT_FORMAT_TYPE_SWIZZLED; } else { - rt_format |= (NV40TCL_RT_FORMAT_TYPE_LINEAR << - NV40TCL_RT_FORMAT_TYPE_SHIFT); + rt_format |= NV40TCL_RT_FORMAT_TYPE_LINEAR; } if (fb->cbufs[0]->format == PIPE_FORMAT_U_R5_G6_B5) { - rt_format |= (NV40TCL_RT_FORMAT_COLOR_R5G6B5 << - NV40TCL_RT_FORMAT_COLOR_SHIFT); + rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; } else { - rt_format |= (NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 << - NV40TCL_RT_FORMAT_COLOR_SHIFT); + rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; } if (fb->zbuf && fb->zbuf->format == PIPE_FORMAT_U_Z16) { - rt_format |= (NV40TCL_RT_FORMAT_DEPTH_Z16 << - NV40TCL_RT_FORMAT_DEPTH_SHIFT); + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; } else { - rt_format |= (NV40TCL_RT_FORMAT_DEPTH_Z24S8 << - NV40TCL_RT_FORMAT_DEPTH_SHIFT); + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; } BEGIN_RING(curie, NV40TCL_RT_HORIZ, 3); diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 1535037f638..ae162b95cef 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -119,55 +119,4 @@ struct nv40_depth_stencil_state { } stencil; }; -static INLINE unsigned -nv40_tex_wrap_mode(unsigned wrap) { - switch (wrap) { - case PIPE_TEX_WRAP_REPEAT: - return NV40TCL_TEX_WRAP_S_REPEAT; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - return NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - return NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - return NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER; - case PIPE_TEX_WRAP_CLAMP: - return NV40TCL_TEX_WRAP_S_CLAMP; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - return NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - return NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - return NV40TCL_TEX_WRAP_S_MIRROR_CLAMP; - default: - return NV40TCL_TEX_WRAP_S_REPEAT; - } -} - -static INLINE unsigned -nv40_tex_filter(unsigned f0, unsigned f1) { - switch (f0) { - case PIPE_TEX_FILTER_NEAREST: - switch (f1) { - case PIPE_TEX_MIPFILTER_NEAREST: - return NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST; - case PIPE_TEX_MIPFILTER_LINEAR: - return NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR; - case PIPE_TEX_MIPFILTER_NONE: - default: - return NV40TCL_TEX_FILTER_MIN_NEAREST; - } - case PIPE_TEX_FILTER_LINEAR: - default: - switch (f1) { - case PIPE_TEX_MIPFILTER_NEAREST: - return NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST; - case PIPE_TEX_MIPFILTER_LINEAR: - return NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR; - case PIPE_TEX_MIPFILTER_NONE: - default: - return NV40TCL_TEX_FILTER_MIN_LINEAR; - } - } -} - #endif diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index a92d6250a20..4acf80c9cee 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -6,14 +6,10 @@ TRUE, \ PIPE_FORMAT_##m, \ NV40TCL_TEX_FORMAT_FORMAT_##tf, \ - (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x << NV40TCL_TEX_SWIZZLE_S0_X_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S0_X_##ts0y << NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S0_X_##ts0z << NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S0_X_##ts0w << NV40TCL_TEX_SWIZZLE_S0_W_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S1_X_##ts1x << NV40TCL_TEX_SWIZZLE_S1_X_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S1_X_##ts1y << NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S1_X_##ts1z << NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT) | \ - (NV40TCL_TEX_SWIZZLE_S1_X_##ts1w << NV40TCL_TEX_SWIZZLE_S1_W_SHIFT), \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x | NV40TCL_TEX_SWIZZLE_S0_Y_##ts0y | \ + NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w | \ + NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y | \ + NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ } struct nv40_texture_format { @@ -53,20 +49,6 @@ nv40_tex_format(uint pipe_format) return NULL; } -static INLINE int -nv40_tex_dims(uint pipe_target) -{ - switch (pipe_target) { - case PIPE_TEXTURE_1D: return 1; - case PIPE_TEXTURE_2D: return 2; - case PIPE_TEXTURE_3D: return 3; - case PIPE_TEXTURE_CUBE: return 2; - default: - NOUVEAU_ERR("AII unknown pipe target: %d\n", pipe_target); - return 2; - } -} - static void nv40_tex_unit_enable(struct nv40_context *nv40, int unit) { @@ -78,20 +60,33 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) int swizzled = 0; /*XXX: implement in region code? */ tf = nv40_tex_format(mt->format); - if (!tf->defined) { + if (!tf || !tf->defined) { NOUVEAU_ERR("Unsupported texture format: 0x%x\n", mt->format); return; } - txf = (tf->format | 0x80) << NV40TCL_TEX_FORMAT_FORMAT_SHIFT; + txf = tf->format | 0x8000; txf |= ((mt->last_level - mt->first_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); + if (1) /* XXX */ txf |= NV40TCL_TEX_FORMAT_NO_BORDER; - txf |= (nv40_tex_dims(mt->target) << NV40TCL_TEX_FORMAT_DIMS_SHIFT); - if (0) /*XXX*/ - txf |= NV40TCL_TEX_FORMAT_RECT; + switch (mt->target) { + case PIPE_TEXTURE_2D: + case PIPE_TEXTURE_CUBE: + txf |= NV40TCL_TEX_FORMAT_DIMS_2D; + break; + case PIPE_TEXTURE_3D: + txf |= NV40TCL_TEX_FORMAT_DIMS_3D; + break; + case PIPE_TEXTURE_1D: + txf |= NV40TCL_TEX_FORMAT_DIMS_1D; + break; + default: + NOUVEAU_ERR("Unknown target %d\n", mt->target); + return; + } if (swizzled) { txp = 0; -- cgit v1.2.3 From b4c813313a9f299ee79145d2b58377db1ecf8cd8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 15:34:26 +1100 Subject: nv40: remove use of temps for KILP, implement KIL --- src/mesa/pipe/nv40/nv40_fragprog.c | 92 +++++++++++++++++++++----------------- src/mesa/pipe/nv40/nv40_shader.h | 10 +++-- src/mesa/pipe/nv40/nv40_vertprog.c | 73 +++++++++++++++++------------- 3 files changed, 100 insertions(+), 75 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 3be13143476..9b89cf5939a 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -58,13 +58,13 @@ struct nv40_fpc { }; static INLINE struct nv40_sreg -nv40_sr_temp(struct nv40_fpc *fpc) +temp(struct nv40_fpc *fpc) { int idx; idx = fpc->temp_temp_count++; idx += fpc->high_temp + 1; - return nv40_sr(0, NV40_FP_REG_TYPE_TEMP, idx); + return nv40_sr(NV40SR_TEMP, idx); } #define arith(cc,s,o,d,m,s0,s1,s2) \ @@ -73,22 +73,30 @@ nv40_sr_temp(struct nv40_fpc *fpc) #define tex(cc,s,o,u,d,m,s0,s1,s2) \ nv40_fp_tex((cc), (s), NV40_FP_OP_OPCODE_##o, (u), \ (d), (m), (s0), none, none) -#define temp(fpc) nv40_sr_temp((fpc)) static void emit_src(struct nv40_fpc *fpc, uint32_t *hw, int pos, struct nv40_sreg src) { uint32_t sr = 0; - sr |= (src.type << NV40_FP_REG_TYPE_SHIFT); - if (src.type == NV40_FP_REG_TYPE_INPUT) { + switch (src.type) { + case NV40SR_INPUT: + sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); hw[0] |= (src.index << NV40_FP_OP_INPUT_SRC_SHIFT); - } else - if (src.type == NV40_FP_REG_TYPE_CONST) { - fpc->inst_has_const = TRUE; - } else - if (src.type == NV40_FP_REG_TYPE_TEMP) { + break; + case NV40SR_TEMP: + sr |= (NV40_FP_REG_TYPE_TEMP << NV40_FP_REG_TYPE_SHIFT); sr |= (src.index << NV40_FP_REG_SRC_SHIFT); + break; + case NV40SR_CONST: + sr |= (NV40_FP_REG_TYPE_CONST << NV40_FP_REG_TYPE_SHIFT); + fpc->inst_has_const = TRUE; + break; + case NV40SR_NONE: + sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); + break; + default: + assert(0); } if (src.negate) @@ -110,16 +118,25 @@ emit_dst(struct nv40_fpc *fpc, uint32_t *hw, struct nv40_sreg dst) { struct nv40_fragment_program *fp = fpc->fp; - if (dst.output) { + switch (dst.type) { + case NV40SR_TEMP: + if (fp->num_regs < (dst.index + 1)) + fp->num_regs = dst.index + 1; + break; + case NV40SR_OUTPUT: if (dst.index == 1) { fp->writes_depth = 1; } else { hw[0] |= NV40_FP_OP_UNK0_7; } - } else { - if (fp->num_regs < (dst.index + 1)) - fp->num_regs = dst.index + 1; + break; + case NV40SR_NONE: + hw[0] |= (1 << 30); + break; + default: + assert(0); } + hw[0] |= (dst.index << NV40_FP_OP_OUT_REG_SHIFT); } @@ -180,29 +197,25 @@ static INLINE struct nv40_sreg tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) { struct nv40_sreg src; - uint type, index; switch (fsrc->SrcRegister.File) { case TGSI_FILE_INPUT: - type = NV40_FP_REG_TYPE_INPUT; - index = fpc->attrib_map[fsrc->SrcRegister.Index]; + src = nv40_sr(NV40SR_INPUT, + fpc->attrib_map[fsrc->SrcRegister.Index]); break; case TGSI_FILE_CONSTANT: - type = NV40_FP_REG_TYPE_CONST; - index = fsrc->SrcRegister.Index; + src = nv40_sr(NV40SR_CONST, fsrc->SrcRegister.Index); break; case TGSI_FILE_TEMPORARY: - type = NV40_FP_REG_TYPE_TEMP; - index = fsrc->SrcRegister.Index + 1; - if (fpc->high_temp < index) - fpc->high_temp = index; + src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index + 1); + if (fpc->high_temp < src.index) + fpc->high_temp = src.index; break; default: NOUVEAU_ERR("bad src file\n"); break; } - src = nv40_sr(0, type, index); src.abs = fsrc->SrcRegisterExtMod.Absolute; src.negate = fsrc->SrcRegister.Negate; src.swz[0] = fsrc->SrcRegister.SwizzleX; @@ -214,30 +227,26 @@ tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) static INLINE struct nv40_sreg tgsi_dst(struct nv40_fpc *fpc, const struct tgsi_full_dst_register *fdst) { - int out, idx; + int idx; switch (fdst->DstRegister.File) { case TGSI_FILE_OUTPUT: - out = 1; if (fdst->DstRegister.Index == fpc->colour_id) - idx = 0; + return nv40_sr(NV40SR_OUTPUT, 0); else - idx = 1; + return nv40_sr(NV40SR_OUTPUT, 1); break; case TGSI_FILE_TEMPORARY: - out = 0; idx = fdst->DstRegister.Index + 1; if (fpc->high_temp < idx) fpc->high_temp = idx; - break; + return nv40_sr(NV40SR_TEMP, idx); case TGSI_FILE_NULL: - break; + return nv40_sr(NV40SR_NONE, 0); default: NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); - break; + return nv40_sr(NV40SR_NONE, 0); } - - return nv40_sr(out, NV40_FP_REG_TYPE_TEMP, idx); } static INLINE int @@ -256,8 +265,8 @@ static boolean nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, const struct tgsi_full_instruction *finst) { + const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); struct nv40_sreg src[3], dst, tmp; - struct nv40_sreg none = nv40_sr(0, NV40_FP_REG_TYPE_INPUT, 0); int mask, sat, unit; int ai = -1, ci = -1; int i; @@ -361,14 +370,13 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, arith(fpc, sat, FRC, dst, mask, src[0], none, none); break; case TGSI_OPCODE_KIL: + arith(fpc, 0, KIL, none, 0, none, none, none); + break; case TGSI_OPCODE_KILP: - /*XXX: Which is NV, which is ARB kil? ARB implemented here. - *XXX: Don't need temp, can update CC0 without writing dst - */ - tmp = temp(fpc); - tmp.cc_update = 1; - arith(fpc, 0, MOV, tmp, MASK_ALL, src[0], none, none); - dst.cc_test = NV40_FP_OP_COND_LT; + dst = nv40_sr(NV40SR_NONE, 0); + dst.cc_update = 1; + arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); + dst.cc_update = 0; dst.cc_test = NV40_FP_OP_COND_LT; arith(fpc, 0, KIL, dst, 0, none, none, none); break; case TGSI_OPCODE_LG2: diff --git a/src/mesa/pipe/nv40/nv40_shader.h b/src/mesa/pipe/nv40/nv40_shader.h index 5b2cf3e2935..855562eb0d1 100644 --- a/src/mesa/pipe/nv40/nv40_shader.h +++ b/src/mesa/pipe/nv40/nv40_shader.h @@ -472,8 +472,13 @@ # define NV40_FP_SWIZZLE_W 3 #define NV40_FP_REG_NEGATE (1 << 17) +#define NV40SR_NONE 0 +#define NV40SR_OUTPUT 1 +#define NV40SR_INPUT 2 +#define NV40SR_TEMP 3 +#define NV40SR_CONST 4 + struct nv40_sreg { - int output; int type; int index; @@ -491,10 +496,9 @@ struct nv40_sreg { }; static INLINE struct nv40_sreg -nv40_sr(int out, int type, int index) +nv40_sr(int type, int index) { struct nv40_sreg temp = { - .output = out, .type = type, .index = index, .dst_scale = DEF_SCALE, diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index be550e47430..68334b7420e 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -54,18 +54,17 @@ struct nv40_vpc { }; static INLINE struct nv40_sreg -nv40_sr_temp(struct nv40_vpc *vpc) +temp(struct nv40_vpc *vpc) { int idx; idx = vpc->temp_temp_count++; idx += vpc->high_temp; - return nv40_sr(0, NV40_VP_SRC_REG_TYPE_TEMP, idx); + return nv40_sr(NV40SR_TEMP, idx); } static INLINE struct nv40_sreg -nv40_sr_const(struct nv40_vpc *vpc, int pipe, - float x, float y, float z, float w) +constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w) { struct nv40_vertex_program *vp = vpc->vp; int idx = vp->num_consts; @@ -78,13 +77,11 @@ nv40_sr_const(struct nv40_vpc *vpc, int pipe, vp->consts[idx].value[3] = w; vp->num_consts++; - return nv40_sr(0, NV40_VP_SRC_REG_TYPE_CONST, idx); + return nv40_sr(NV40SR_CONST, idx); } #define arith(cc,s,o,d,m,s0,s1,s2) \ nv40_vp_arith((cc), (s), NV40_VP_INST_##o, (d), (m), (s0), (s1), (s2)) -#define temp(vpc) nv40_sr_temp((vpc)) -#define constant(v,p,x,y,z,w) nv40_sr_const((v), (p), (x), (y), (z), (w)) static void emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src) @@ -92,15 +89,28 @@ emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src) struct nv40_vertex_program *vp = vpc->vp; uint32_t sr = 0; - sr |= (src.type << NV40_VP_SRC_REG_TYPE_SHIFT); - if (src.type == NV40_VP_SRC_REG_TYPE_INPUT) { + switch (src.type) { + case NV40SR_TEMP: + sr |= (NV40_VP_SRC_REG_TYPE_TEMP << NV40_VP_SRC_REG_TYPE_SHIFT); + sr |= (src.index << NV40_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV40SR_INPUT: + sr |= (NV40_VP_SRC_REG_TYPE_INPUT << + NV40_VP_SRC_REG_TYPE_SHIFT); vp->ir |= (1 << src.index); hw[1] |= (src.index << NV40_VP_INST_INPUT_SRC_SHIFT); - } else - if (src.type == NV40_VP_SRC_REG_TYPE_CONST) { + break; + case NV40SR_CONST: + sr |= (NV40_VP_SRC_REG_TYPE_CONST << + NV40_VP_SRC_REG_TYPE_SHIFT); hw[1] |= (src.index << NV40_VP_INST_CONST_SRC_SHIFT); - } else { - sr |= (src.index << NV40_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV40SR_NONE: + sr |= (NV40_VP_SRC_REG_TYPE_INPUT << + NV40_VP_SRC_REG_TYPE_SHIFT); + break; + default: + assert(0); } if (src.negate) @@ -140,16 +150,18 @@ emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) { struct nv40_vertex_program *vp = vpc->vp; - if (dst.output == 0) { + switch (dst.type) { + case NV40SR_TEMP: hw[3] |= NV40_VP_INST_DEST_MASK; if (slot == 0) { hw[0] |= (dst.index << NV40_VP_INST_VEC_DEST_TEMP_SHIFT); } else { - hw[3] |= (dst.index << + hw[3] |= (dst.index << NV40_VP_INST_SCA_DEST_TEMP_SHIFT); } - } else { + break; + case NV40SR_OUTPUT: switch (dst.index) { case NV40_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; case NV40_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; @@ -166,7 +178,7 @@ emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) case NV40_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; case NV40_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; default: - break; + break; } hw[3] |= (dst.index << NV40_VP_INST_DEST_SHIFT); @@ -177,6 +189,9 @@ emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) hw[3] |= NV40_VP_INST_SCA_RESULT; hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; } + break; + default: + assert(0); } } @@ -220,8 +235,7 @@ tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { switch (fsrc->SrcRegister.File) { case TGSI_FILE_INPUT: - src = nv40_sr(0, NV40_VP_SRC_REG_TYPE_INPUT, - fsrc->SrcRegister.Index); + src = nv40_sr(NV40SR_INPUT, fsrc->SrcRegister.Index); break; case TGSI_FILE_CONSTANT: src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); @@ -229,8 +243,7 @@ tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { case TGSI_FILE_TEMPORARY: if (vpc->high_temp < fsrc->SrcRegister.Index) vpc->high_temp = fsrc->SrcRegister.Index; - src = nv40_sr(0, NV40_VP_SRC_REG_TYPE_TEMP, - fsrc->SrcRegister.Index); + src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index); break; default: NOUVEAU_ERR("bad src file\n"); @@ -248,25 +261,25 @@ tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { static INLINE struct nv40_sreg tgsi_dst(struct nv40_vpc *vpc, const struct tgsi_full_dst_register *fdst) { - uint out, idx; + struct nv40_sreg dst; switch (fdst->DstRegister.File) { case TGSI_FILE_OUTPUT: - out = 1; - idx = vpc->output_map[fdst->DstRegister.Index]; + dst = nv40_sr(NV40SR_OUTPUT, + vpc->output_map[fdst->DstRegister.Index]); + break; case TGSI_FILE_TEMPORARY: - out = 0; - idx = fdst->DstRegister.Index; - if (vpc->high_temp < idx) - vpc->high_temp = idx; + dst = nv40_sr(NV40SR_TEMP, fdst->DstRegister.Index); + if (vpc->high_temp < dst.index) + vpc->high_temp = dst.index; break; default: NOUVEAU_ERR("bad dst file\n"); break; } - return nv40_sr(out, NV40_VP_SRC_REG_TYPE_TEMP, idx); + return dst; } static INLINE int @@ -286,7 +299,7 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, const struct tgsi_full_instruction *finst) { struct nv40_sreg src[3], dst, tmp; - struct nv40_sreg none = nv40_sr(0, NV40_VP_SRC_REG_TYPE_INPUT, 0); + struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); int mask; int ai = -1, ci = -1; int i; -- cgit v1.2.3 From 04fcee96b1ad9caf5b903a0ca72a1929ba2aae50 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 19:08:33 +1100 Subject: nv40: track pipe const id --- src/mesa/pipe/nv40/nv40_fragprog.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 9b89cf5939a..a5c562ef716 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -91,6 +91,7 @@ emit_src(struct nv40_fpc *fpc, uint32_t *hw, int pos, struct nv40_sreg src) case NV40SR_CONST: sr |= (NV40_FP_REG_TYPE_CONST << NV40_FP_REG_TYPE_SHIFT); fpc->inst_has_const = TRUE; + fpc->inst_const_id = src.index; break; case NV40SR_NONE: sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); -- cgit v1.2.3 From 85d626508ad9f8deef251058c12b03d46468a962 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 19:09:47 +1100 Subject: Add support for Z24_S8 to GL state tracker. --- src/mesa/pipe/p_format.h | 2 ++ src/mesa/state_tracker/st_cb_clear.c | 8 ++++---- src/mesa/state_tracker/st_cb_readpixels.c | 9 +++++++++ src/mesa/state_tracker/st_format.c | 9 +++++++++ 4 files changed, 24 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/p_format.h b/src/mesa/pipe/p_format.h index d478a6c58df..f42f987c4e3 100644 --- a/src/mesa/pipe/p_format.h +++ b/src/mesa/pipe/p_format.h @@ -171,6 +171,7 @@ static INLINE uint pf_get(pipe_format_rgbazs_t f, uint shift, uint mask) #define _PIPE_FORMAT_RRRG _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G ) #define _PIPE_FORMAT_Z000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_SZ00 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_S, PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) +#define _PIPE_FORMAT_ZS00 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_Z, PIPE_FORMAT_COMP_S, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_S000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_S, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) /** @@ -222,6 +223,7 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f) #define PIPE_FORMAT_U_Z32 _PIPE_FORMAT_RGBAZS_8N( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) /**< uint Z/depth */ #define PIPE_FORMAT_F_Z32 _PIPE_FORMAT_RGBAZS_8N( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) /**< float Z/depth */ #define PIPE_FORMAT_S8_Z24 _PIPE_FORMAT_RGBAZS_8N( _PIPE_FORMAT_SZ00, 1, 3, 0, 0, PIPE_FORMAT_TYPE_UNORM ) /**< 8-bit stencil + 24-bit Z */ +#define PIPE_FORMAT_Z24_S8 _PIPE_FORMAT_RGBAZS_8N( _PIPE_FORMAT_ZS00, 3, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ) /**< 24-bit Z + 8-bit stencil */ #define PIPE_FORMAT_U_S8 _PIPE_FORMAT_RGBAZS_8N( _PIPE_FORMAT_S000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) /**< 8-bit stencil */ #define PIPE_FORMAT_R64_FLOAT _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) #define PIPE_FORMAT_R64G64_FLOAT _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 3e591170da7..219a5afcbdd 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -91,6 +91,8 @@ depth_value(GLuint pipeFormat, GLfloat value) return (uint) (value * 0xffffffff); case PIPE_FORMAT_S8_Z24: return (uint) (value * 0xffffff); + case PIPE_FORMAT_Z24_S8: + return ((uint) (value * 0xffffff)) << 8; default: assert(0); return 0; @@ -103,7 +105,7 @@ is_depth_stencil_format(GLuint pipeFormat) { switch (pipeFormat) { case PIPE_FORMAT_S8_Z24: - /*case PIPE_FORMAT_Z24_S8:*/ + case PIPE_FORMAT_Z24_S8: return GL_TRUE; default: return GL_FALSE; @@ -521,11 +523,9 @@ clear_depth_stencil_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) case PIPE_FORMAT_S8_Z24: clearValue |= ctx->Stencil.Clear << 24; break; -#if 0 case PIPE_FORMAT_Z24_S8: - clearValue = (clearValue << 8) | clearVal; + clearValue |= clearValue | ctx->Stencil.Clear; break; -#endif default: assert(0); } diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index c347a0b6888..2e7c01672a1 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -100,6 +100,15 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, } } break; + case PIPE_FORMAT_Z24_S8: + { + const uint *src = (uint *) stmap + srcY * ps->region->pitch + x; + GLint k; + for (k = 0; k < width; k++) { + values[k] = src[k] & 0xff; + } + } + break; default: assert(0); } diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index ddba6509415..c6b5bc968f0 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -320,6 +320,9 @@ default_depth_format( if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) { return PIPE_FORMAT_S8_Z24; } + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) { + return PIPE_FORMAT_Z24_S8; + } return PIPE_FORMAT_NONE; } @@ -486,6 +489,8 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, case GL_DEPTH_COMPONENT24: if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) return PIPE_FORMAT_S8_Z24; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) + return PIPE_FORMAT_Z24_S8; /* fall-through */ case GL_DEPTH_COMPONENT32: if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_Z32 )) @@ -503,12 +508,16 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, return PIPE_FORMAT_U_S8; if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) return PIPE_FORMAT_S8_Z24; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) + return PIPE_FORMAT_Z24_S8; return PIPE_FORMAT_NONE; case GL_DEPTH_STENCIL_EXT: case GL_DEPTH24_STENCIL8_EXT: if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) return PIPE_FORMAT_S8_Z24; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) + return PIPE_FORMAT_Z24_S8; return PIPE_FORMAT_NONE; default: -- cgit v1.2.3 From f01e305e989189fce146d26793e795aa83ff980d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 19:10:58 +1100 Subject: nv40: use native Z24_S8 format for zeta buffer --- src/mesa/pipe/nv40/nv40_clear.c | 6 ------ src/mesa/pipe/nv40/nv40_context.c | 2 +- src/mesa/pipe/nv40/nv40_surface.c | 4 ++-- 3 files changed, 3 insertions(+), 9 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_clear.c b/src/mesa/pipe/nv40/nv40_clear.c index f3b7a23689f..7890107c957 100644 --- a/src/mesa/pipe/nv40/nv40_clear.c +++ b/src/mesa/pipe/nv40/nv40_clear.c @@ -10,12 +10,6 @@ void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { - /*XXX: We're actually Z24_S8... */ - if (ps->format == PIPE_FORMAT_S8_Z24) { - clearValue = (((clearValue & 0xff000000) >> 24) | - ((clearValue & 0x00ffffff) << 8)); - } - pipe->region_fill(pipe, ps->region, 0, 0, 0, ps->width, ps->height, clearValue); } diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index ef5f21cb9c3..8462d08f8d9 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -12,7 +12,7 @@ nv40_is_format_supported(struct pipe_context *pipe, uint format) switch (format) { case PIPE_FORMAT_U_A8_R8_G8_B8: case PIPE_FORMAT_U_R5_G6_B5: - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_Z24_S8: return TRUE; default: break; diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index 84e0d792681..5292b5171ff 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -79,7 +79,7 @@ nv40_get_tile_rgba(struct pipe_context *pipe, p += w0 * 4; } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_Z24_S8: { const float scale = 1.0 / (float) 0xffffff; for (i = 0; i < h; i++) { @@ -89,7 +89,7 @@ nv40_get_tile_rgba(struct pipe_context *pipe, pRow[0] = pRow[1] = pRow[2] = - pRow[3] = (pixel & 0xffffff) * scale; + pRow[3] = ((pixel & 0xffffff) >> 8) * scale; pRow += 4; } src += ps->region->pitch; -- cgit v1.2.3 From 30837fd24f76131ef5ea77a1396c304d9fc87f4e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 20:56:47 +1100 Subject: nv40: fix a couple of typos --- src/mesa/pipe/nv40/nv40_state.c | 4 ++-- src/mesa/pipe/nv40/nv40_state_tex.c | 1 + 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index a9790484b9b..37b07352bec 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -210,9 +210,9 @@ nv40_sampler_state_create(struct pipe_context *pipe, } - ps->wrap = ((wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_S_SHIFT) | + ps->wrap = ((wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_S_SHIFT) | (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) | - (wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_R_SHIFT)); + (wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_R_SHIFT)); ps->filt = filter; ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | (float_to_ubyte(cso->border_color[0]) << 16) | diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 4acf80c9cee..832abbf1f39 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -33,6 +33,7 @@ nv40_texture_formats[] = { // _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), // _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), // _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + {}, }; static struct nv40_texture_format * -- cgit v1.2.3 From b9b5f4b3c16f0b7bc8ae2d7cca03597e0029cb02 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 21:13:33 +1100 Subject: nv40: "rect" textures, anisotropic filtering --- src/mesa/pipe/nv40/nv40_context.c | 2 +- src/mesa/pipe/nv40/nv40_state.c | 42 ++++++++++++++++++++++++++++++++----- src/mesa/pipe/nv40/nv40_state.h | 2 ++ src/mesa/pipe/nv40/nv40_state_tex.c | 5 +++-- 4 files changed, 43 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 8462d08f8d9..6f270916432 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -52,7 +52,7 @@ nv40_get_param(struct pipe_context *pipe, int param) case PIPE_CAP_S3TC: return 0; case PIPE_CAP_ANISOTROPIC_FILTER: - return 0; + return 1; case PIPE_CAP_POINT_SPRITE: return 0; case PIPE_CAP_MAX_RENDER_TARGETS: diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 37b07352bec..9b58409e51e 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -166,7 +166,42 @@ nv40_sampler_state_create(struct pipe_context *pipe, uint32_t filter = 0; ps = malloc(sizeof(struct nv40_sampler_state)); - + + ps->fmt = 0; + if (!cso->normalized_coords) + ps->fmt |= NV40TCL_TEX_FORMAT_RECT; + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) | + (wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_R_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy >= 2.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + + if (cso->max_anisotropy >= 16.0) { + ps->en |= (7 << 4); + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= (6 << 4); + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= (5 << 4); + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= (4 << 4); + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= (3 << 4); + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= (2 << 4); + } else { + ps->en |= (1 << 4); /* 2.0 */ + } + } + switch (cso->mag_img_filter) { case PIPE_TEX_FILTER_LINEAR: filter |= NV40TCL_TEX_FILTER_MAG_LINEAR; @@ -209,11 +244,8 @@ nv40_sampler_state_create(struct pipe_context *pipe, break; } - - ps->wrap = ((wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_S_SHIFT) | - (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) | - (wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_R_SHIFT)); ps->filt = filter; + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | (float_to_ubyte(cso->border_color[0]) << 16) | (float_to_ubyte(cso->border_color[1]) << 8) | diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index ae162b95cef..14bf5d83e3e 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -24,7 +24,9 @@ struct nv40_blend_state { }; struct nv40_sampler_state { + uint32_t fmt; uint32_t wrap; + uint32_t en; uint32_t filt; uint32_t bcol; }; diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 832abbf1f39..cf2a7a9fe0f 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -66,7 +66,8 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) return; } - txf = tf->format | 0x8000; + txf = ps->fmt; + txf |= tf->format | 0x8000; txf |= ((mt->last_level - mt->first_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); @@ -107,7 +108,7 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) NOUVEAU_BO_OR | NOUVEAU_BO_RD, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); OUT_RING (ps->wrap); - OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | + OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | (0x00078000) /* mipmap related? */); OUT_RING (txs); OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); -- cgit v1.2.3 From 0d0349faff07df663eb49fe1a7209e1c6a28505b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 22:09:56 +1100 Subject: nouveau: temporary workaround for dodgy buffer code --- src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c index 98f0d554058..1160a8f340d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c @@ -108,6 +108,14 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, OUT_RING ((dy << 16) | dx); OUT_RING (( h << 16) | w); + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(NvGdiRect, 0x104, 1); + OUT_RING (0); + BEGIN_RING(NvGdiRect, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); + return 0; } @@ -147,6 +155,14 @@ nv04_region_fill(struct nouveau_context *nv, OUT_RING ((dx << 16) | dy); OUT_RING (( w << 16) | h); + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(NvGdiRect, 0x104, 1); + OUT_RING (0); + BEGIN_RING(NvGdiRect, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); + return 0; } -- cgit v1.2.3 From 0655cdcf48daecbe10ff8c16c443686a64848d1b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 20 Nov 2007 22:34:55 +1100 Subject: nv40: force reupload of all consts on vtxprog change --- src/mesa/pipe/nv40/nv40_state_emit.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index a29c70538f5..3224c829994 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -42,11 +42,14 @@ nv40_state_update_vertprog(struct nv40_context *nv40) struct pipe_context *pipe = (struct pipe_context *)nv40; struct nv40_vertex_program *vp = nv40->vertprog.vp; float *map; - int i; + int i, force_consts = 0; if (!nv40->vertprog.vp->translated) nv40_vertprog_translate(nv40, nv40->vertprog.vp); + if (nv40->vertprog.vp != nv40->vertprog.active_vp) + force_consts = 1; + if (vp->num_consts) { map = pipe->winsys->buffer_map(pipe->winsys, nv40->vertprog.constant_buf, @@ -55,7 +58,8 @@ nv40_state_update_vertprog(struct nv40_context *nv40) uint pid = vp->consts[i].pipe_id; if (pid >= 0) { - if (!memcmp(vp->consts[i].value, &map[pid*4], + if (!force_consts && + !memcmp(vp->consts[i].value, &map[pid*4], 4 * sizeof(float))) continue; memcpy(vp->consts[i].value, &map[pid*4], -- cgit v1.2.3 From ff7edad0776603aa62b8f684db20c84bde75ea0b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 21 Nov 2007 19:20:38 +1100 Subject: nv40: Better teximage layout, probably not 100% correct still. Gallium represents image layout by saying that each mipmap level has a number of "face" images within it. However, nv40 represents them as "faces" that have a number of mipmap levels. I'm not sure if the gallium representation allows this, but I've made an attempt to match it as closely as possible. CUBE/3D textures with mipmaps are probably broken, but untested currently. --- src/mesa/pipe/nv40/nv40_miptree.c | 69 ++++++++++++++++++++----------------- src/mesa/pipe/nv40/nv40_state_tex.c | 4 ++- 2 files changed, 41 insertions(+), 32 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 2d8ef8ced77..15639f01e90 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -8,53 +8,60 @@ boolean nv40_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - uint width, height, depth, offset; boolean swizzled = FALSE; - int l; + uint width = mt->width0, height = mt->height0, depth = mt->depth0; + uint offset; + int nr_faces, l, f; mt->pitch = mt->width0; mt->total_height = 0; - width = mt->width0; - height = mt->height0; - depth = mt->depth0; - offset = 0; - for (l = mt->first_level; l <= mt->last_level; l++) { - uint pitch, f; + if (mt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else + if (mt->target == PIPE_TEXTURE_3D) { + nr_faces = mt->depth0; + } else { + nr_faces = 1; + } + for (l = mt->first_level; l <= mt->last_level; l++) { mt->level[l].width = width; mt->level[l].height = height; mt->level[l].depth = depth; - mt->level[l].level_offset = offset; - - if (!swizzled) - pitch = mt->width0; - else - pitch = width; - - if (mt->target == PIPE_TEXTURE_CUBE) - mt->level[l].nr_images = 6; - else - if (mt->target == PIPE_TEXTURE_3D) - mt->level[l].nr_images = mt->level[l].depth; - else - mt->level[l].nr_images = 1; - mt->level[l].image_offset = - malloc(mt->level[l].nr_images * sizeof(unsigned)); - - for (f = 0; f < mt->level[l].nr_images; f++) { - mt->level[l].image_offset[f] = - (offset - mt->level[l].level_offset) / mt->cpp; - mt->total_height += height; + mt->level[l].level_offset = 0; - offset += (pitch * mt->cpp * height); - } + mt->level[l].nr_images = nr_faces; + mt->level[l].image_offset = malloc(nr_faces * sizeof(unsigned)); + for (f = 0; f < nr_faces; f++) + mt->total_height += height; width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); depth = MAX2(1, depth >> 1); } + offset = 0; + for (f = 0; f < nr_faces; f++) { + for (l = mt->first_level; l <= mt->last_level; l++) { + if (f == 0) { + mt->level[l].level_offset = offset; + } + + uint pitch; + + if (swizzled) + pitch = mt->level[l].width * mt->cpp; + else + pitch = mt->width0 * mt->cpp; + pitch = (pitch + 63) & ~63; + + mt->level[l].image_offset[f] = + (offset - mt->level[l].level_offset) / mt->cpp; + offset += pitch * mt->level[l].height; + } + } + return TRUE; } diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index cf2a7a9fe0f..bd779df9b36 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -75,8 +75,10 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) txf |= NV40TCL_TEX_FORMAT_NO_BORDER; switch (mt->target) { - case PIPE_TEXTURE_2D: case PIPE_TEXTURE_CUBE: + txf |= NV40TCL_TEX_FORMAT_CUBIC; + /* fall-through */ + case PIPE_TEXTURE_2D: txf |= NV40TCL_TEX_FORMAT_DIMS_2D; break; case PIPE_TEXTURE_3D: -- cgit v1.2.3 From dd1500b8b436dc4a009455489ade7dcf65f03d07 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 21 Nov 2007 19:21:34 +1100 Subject: nouveau: wait for sync after m2mf also, to be thourough. These will all go away at some point I promise, want to rule out sync issues while I bash nv40 into shape. --- src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c index 1160a8f340d..0fdb38e105c 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c @@ -64,6 +64,14 @@ nv04_region_copy_m2mf(struct nouveau_context *nv, struct pipe_region *dst, dst_offset += dst->pitch * count; } + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(NvM2MF, 0x104, 1); + OUT_RING (0); + BEGIN_RING(NvM2MF, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); + return 0; } -- cgit v1.2.3 From 558f3abb49475ba0aad585e7fb0c07121aecde6f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 23 Nov 2007 18:54:11 +1100 Subject: nv40: support for keeping multiple vtxprogs on the hw at the same time. --- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 87 +++++++++++++++++ src/mesa/pipe/nouveau/nouveau_winsys.h | 17 ++++ src/mesa/pipe/nv40/nv40_context.c | 8 ++ src/mesa/pipe/nv40/nv40_context.h | 11 ++- src/mesa/pipe/nv40/nv40_fragprog.c | 2 +- src/mesa/pipe/nv40/nv40_state.c | 4 +- src/mesa/pipe/nv40/nv40_state.h | 8 +- src/mesa/pipe/nv40/nv40_state_emit.c | 53 +--------- src/mesa/pipe/nv40/nv40_vbo.c | 2 +- src/mesa/pipe/nv40/nv40_vertprog.c | 108 +++++++++++++++++++-- 10 files changed, 231 insertions(+), 69 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index a2e85a0aa7f..bdc89b95318 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -3,6 +3,89 @@ #include "pipe/nouveau/nouveau_winsys.h" +static int +nouveau_resource_init(struct nouveau_resource **heap, int size) +{ + struct nouveau_resource *r; + + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = 0; + r->size = size; + *heap = r; + return 0; +} + +static int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!heap || !size || !res || *res) + return 1; + + while (heap) { + if (!heap->in_use && heap->size >= size) { + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = (heap->start + heap->size) - size; + r->size = size; + r->in_use = TRUE; + r->priv = priv; + + heap->size -= size; + + r->next = heap->next; + if (heap->next) + heap->next->prev = r; + r->prev = heap; + heap->next = r; + + *res = r; + return 0; + } + + heap = heap->next; + } + + return 1; +} + +static void +nouveau_resource_free(struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!res || !*res) + return; + r = *res; + + if (r->prev && !r->prev->in_use) { + r->prev->next = r->next; + if (r->next) + r->next->prev = r->prev; + r->prev->size += r->size; + free(r); + } else + if (r->next && !r->next->in_use) { + r->next->prev = r->prev; + if (r->prev) + r->prev->next = r->next; + r->next->size += r->size; + r->next->start = r->start; + free(r); + } else { + r->in_use = FALSE; + } + + *res = NULL; +} + static int nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, struct nouveau_notifier **notify) @@ -67,6 +150,10 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->nv = nv; nvws->channel = nv->channel; + nvws->res_init = nouveau_resource_init; + nvws->res_alloc = nouveau_resource_alloc; + nvws->res_free = nouveau_resource_free; + nvws->begin_ring = nouveau_pipe_dma_beginp; nvws->out_reloc = nouveau_bo_emit_reloc; nvws->fire_ring = nouveau_pipe_dma_kickoff; diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index beee49a134f..28f17cdc8fa 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -3,6 +3,7 @@ #include #include "pipe/p_winsys.h" +#include "pipe/p_defines.h" #include "pipe/nouveau/nouveau_bo.h" #include "pipe/nouveau/nouveau_channel.h" @@ -10,11 +11,27 @@ #include "pipe/nouveau/nouveau_grobj.h" #include "pipe/nouveau/nouveau_notifier.h" +struct nouveau_resource { + struct nouveau_resource *prev; + struct nouveau_resource *next; + + boolean in_use; + void *priv; + + uint start; + uint size; +}; + struct nouveau_winsys { struct nouveau_context *nv; struct nouveau_channel *channel; + int (*res_init)(struct nouveau_resource **heap, int size); + int (*res_alloc)(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **); + void (*res_free)(struct nouveau_resource **); + /*XXX: this is crappy, and bound to be slow.. however, it's nice and * simple, it'll do for the moment*/ uint32_t *(*begin_ring)(struct nouveau_grobj *, int mthd, int size); diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 6f270916432..ba9d875e0ed 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -241,6 +241,14 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, return NULL; } + if (nvws->res_init(&nv40->vertprog.exec_heap, 512) || + nvws->res_init(&nv40->vertprog.data_heap, 256)) { + nvws->res_free(&nv40->vertprog.exec_heap); + nvws->res_free(&nv40->vertprog.data_heap); + free(nv40); + return NULL; + } + if (!nv40_init_hwctx(nv40, curie_class)) { free(nv40); return NULL; diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 63be38299fb..ec07a883706 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -44,16 +44,19 @@ struct nv40_context { uint32_t tex_dirty; struct { - struct nv40_vertex_program *vp; - struct nv40_vertex_program *active_vp; + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + struct nv40_vertex_program *active; + + struct nv40_vertex_program *current; struct pipe_buffer_handle *constant_buf; } vertprog; struct { - struct nv40_fragment_program *fp; - struct nv40_fragment_program *active_fp; + struct nv40_fragment_program *active; + struct nv40_fragment_program *current; struct pipe_buffer_handle *constant_buf; } fragprog; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index a5c562ef716..09b68513bac 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -645,6 +645,6 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); OUT_RING (fp_control); - nv40->fragprog.active_fp = fp; + nv40->fragprog.active = fp; } diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 9b58409e51e..b1928a5f91a 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -443,7 +443,7 @@ nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nv40_vertex_program *vp = hwcso; - nv40->vertprog.vp = vp; + nv40->vertprog.current = vp; nv40->dirty |= NV40_NEW_VERTPROG; } @@ -471,7 +471,7 @@ nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nv40_fragment_program *fp = hwcso; - nv40->fragprog.fp = fp; + nv40->fragprog.current = fp; nv40->dirty |= NV40_NEW_FRAGPROG; } diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 14bf5d83e3e..d0114972ef2 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -57,11 +57,13 @@ struct nv40_vertex_program { const struct pipe_shader_state *pipe; boolean translated; - boolean on_hw; - int start_ip; + struct nouveau_resource *exec; uint32_t *insn; - int insn_len; + uint insn_len; + + struct nouveau_resource *data; + uint data_start; struct { int pipe_id; diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 3224c829994..55c7d36b129 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -6,7 +6,7 @@ static INLINE void nv40_state_update_fragprog(struct nv40_context *nv40) { struct pipe_context *pipe = (struct pipe_context *)nv40; - struct nv40_fragment_program *fp = nv40->fragprog.fp; + struct nv40_fragment_program *fp = nv40->fragprog.current; float *map; int i; @@ -36,57 +36,18 @@ nv40_state_update_fragprog(struct nv40_context *nv40) } } -static INLINE void -nv40_state_update_vertprog(struct nv40_context *nv40) -{ - struct pipe_context *pipe = (struct pipe_context *)nv40; - struct nv40_vertex_program *vp = nv40->vertprog.vp; - float *map; - int i, force_consts = 0; - - if (!nv40->vertprog.vp->translated) - nv40_vertprog_translate(nv40, nv40->vertprog.vp); - - if (nv40->vertprog.vp != nv40->vertprog.active_vp) - force_consts = 1; - - if (vp->num_consts) { - map = pipe->winsys->buffer_map(pipe->winsys, - nv40->vertprog.constant_buf, - PIPE_BUFFER_FLAG_READ); - for (i = 0; i < vp->num_consts; i++) { - uint pid = vp->consts[i].pipe_id; - - if (pid >= 0) { - if (!force_consts && - !memcmp(vp->consts[i].value, &map[pid*4], - 4 * sizeof(float))) - continue; - memcpy(vp->consts[i].value, &map[pid*4], - 4 * sizeof(float)); - } - - BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5); - OUT_RING (vp->consts[i].hw_id); - OUT_RINGp ((uint32_t *)vp->consts[i].value, 4); - } - pipe->winsys->buffer_unmap(pipe->winsys, - nv40->vertprog.constant_buf); - } -} - void nv40_emit_hw_state(struct nv40_context *nv40) { if (nv40->dirty & NV40_NEW_FRAGPROG) { - struct nv40_fragment_program *cur = nv40->fragprog.fp; + struct nv40_fragment_program *cur = nv40->fragprog.current; nv40_state_update_fragprog(nv40); if (cur->on_hw) nv40->dirty &= ~NV40_NEW_FRAGPROG; - if (!cur->on_hw || cur != nv40->fragprog.active_fp) + if (!cur->on_hw || cur != nv40->fragprog.active) nv40_fragprog_bind(nv40, cur); } @@ -101,12 +62,8 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->dirty &= ~(NV40_NEW_TEXTURE | NV40_NEW_FRAGPROG); } - if (nv40->dirty & NV40_NEW_VERTPROG) { - nv40_state_update_vertprog(nv40); - if (nv40->vertprog.vp != nv40->vertprog.active_vp) - nv40_vertprog_bind(nv40, nv40->vertprog.vp); - nv40->dirty &= ~NV40_NEW_VERTPROG; - } + if (nv40->dirty & NV40_NEW_VERTPROG) + nv40_vertprog_bind(nv40, nv40->vertprog.current); if (nv40->dirty & NV40_NEW_ARRAYS) { nv40_vbo_arrays_update(nv40); diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index aa930476b6f..2da0001b288 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -176,7 +176,7 @@ void nv40_vbo_arrays_update(struct nv40_context *nv40) { struct nouveau_winsys *nvws = nv40->nvws; - struct nv40_vertex_program *vp = nv40->vertprog.vp; + struct nv40_vertex_program *vp = nv40->vertprog.active; uint32_t inputs, vtxfmt[16]; int hw, num_hw; diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index 68334b7420e..04bdaacc98a 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -576,21 +576,108 @@ out_err: void nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) { + struct nouveau_winsys *nvws = nv40->nvws; + struct pipe_context *pipe = &nv40->pipe; + boolean upload_code = FALSE, upload_data = FALSE; + float *map; int i; + /* Translate TGSI shader into hw bytecode */ if (!vp->translated) { - NOUVEAU_ERR("vtxprog invalid, using passthrough shader\n"); - vp = &passthrough_vp; + nv40_vertprog_translate(nv40, vp); + if (!vp->translated) + assert(0); } - if (!vp->on_hw) { - if (nv40->vertprog.active_vp) - nv40->vertprog.active_vp->on_hw = FALSE; - vp->on_hw = TRUE; - vp->start_ip = 0; + /* Allocate hw vtxprog exec slots */ + /*XXX: when we do branching, need to patch targets if program moves. + */ + if (!vp->exec) { + struct nouveau_resource *heap = nv40->vertprog.exec_heap; + uint vplen = vp->insn_len / 4; + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { + while (heap->next && heap->size < vplen) { + struct nv40_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->exec); + } + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) + assert(0); + } + + upload_code = TRUE; + } + + /* Allocate hw vtxprog const slots */ + if (vp->num_consts && !vp->data) { + struct nouveau_resource *heap = nv40->vertprog.data_heap; + int count = vp->num_consts; + + if (nvws->res_alloc(heap, count, vp, &vp->data)) { + while (heap->next && heap->size < count) { + struct nv40_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, count, vp, &vp->data)) + assert(0); + } + + upload_data = TRUE; + } + + /* If constants moved, patch the vtxprog to fix the offsets */ + if (vp->num_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->insn_len; i += 4) { + int id; + + id = (vp->insn[i + 1] & NV40_VP_INST_CONST_SRC_MASK) >> + NV40_VP_INST_CONST_SRC_SHIFT; + id -= vp->data_start; + id += vp->data->start; + + vp->insn[i + 1] &= ~NV40_VP_INST_CONST_SRC_MASK; + vp->insn[i + 1] |= (id << NV40_VP_INST_CONST_SRC_SHIFT); + } + + vp->data_start = vp->data->start; + upload_code = TRUE; + } + /* Update + Upload constant values */ + if (vp->num_consts) { + map = pipe->winsys->buffer_map(pipe->winsys, + nv40->vertprog.constant_buf, + PIPE_BUFFER_FLAG_READ); + for (i = 0; i < vp->num_consts; i++) { + uint pid = vp->consts[i].pipe_id; + + if (pid >= 0) { + if (!upload_data && + !memcmp(vp->consts[i].value, &map[pid*4], + 4 * sizeof(float))) + continue; + memcpy(vp->consts[i].value, &map[pid*4], + 4 * sizeof(float)); + } + + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (vp->consts[i].hw_id + vp->data->start); + OUT_RINGp ((uint32_t *)vp->consts[i].value, 4); + } + pipe->winsys->buffer_unmap(pipe->winsys, + nv40->vertprog.constant_buf); + } + + /* Upload vtxprog */ + if (upload_code) { BEGIN_RING(curie, NV40TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (vp->start_ip); + OUT_RING (vp->exec->start); for (i = 0; i < vp->insn_len; i += 4) { BEGIN_RING(curie, NV40TCL_VP_UPLOAD_INST(0), 4); OUT_RINGp (&vp->insn[i], 4); @@ -598,10 +685,11 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) } BEGIN_RING(curie, NV40TCL_VP_START_FROM_ID, 1); - OUT_RING (vp->start_ip); + OUT_RING (vp->exec->start); BEGIN_RING(curie, NV40TCL_VP_ATTRIB_EN, 2); OUT_RING (vp->ir); OUT_RING (vp->or); - nv40->vertprog.active_vp = vp; + nv40->vertprog.active = vp; } + -- cgit v1.2.3 From 8b1c8f5e758e161bacee3f5eca2fd531f2bf56aa Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 Nov 2007 11:25:32 +1100 Subject: nv40: fp arbitrary swizzling/negation Pretty bad implementation, goal was to get glBitmap working. --- src/mesa/pipe/nv40/nv40_fragprog.c | 79 ++++++++++++++++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_shader.h | 2 + 2 files changed, 81 insertions(+) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 09b68513bac..e6fee2bf5ae 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -4,6 +4,7 @@ #include "pipe/tgsi/exec/tgsi_token.h" #include "pipe/tgsi/exec/tgsi_parse.h" +#include "pipe/tgsi/exec/tgsi_util.h" #include "nv40_context.h" #include "nv40_dma.h" @@ -262,6 +263,66 @@ tgsi_mask(uint tgsi) return mask; } +static boolean +src_native_swz(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc, + struct nv40_sreg *src) +{ + const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); + struct nv40_sreg tgsi = tgsi_src(fpc, fsrc); + uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; + uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, + fsrc->SrcRegisterExtSwz.NegateY, + fsrc->SrcRegisterExtSwz.NegateZ, + fsrc->SrcRegisterExtSwz.NegateW }; + uint c; + + for (c = 0; c < 4; c++) { + switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + mask |= (1 << c); + break; + case TGSI_EXTSWIZZLE_ZERO: + zero_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + case TGSI_EXTSWIZZLE_ONE: + one_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + default: + assert(0); + } + + if (!tgsi.negate && neg[c]) + neg_mask |= (1 << c); + } + + if (mask == MASK_ALL) + return TRUE; + + *src = temp(fpc); + + if (mask) + arith(fpc, 0, MOV, *src, mask, tgsi, none, none); + + if (zero_mask) + arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); + + if (one_mask) + arith(fpc, 0, STR, *src, one_mask, *src, none, none); + + if (neg_mask) { + struct nv40_sreg one = temp(fpc); + arith(fpc, 0, STR, one, neg_mask, one, none, none); + arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none); + } + + return FALSE; +} + static boolean nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, const struct tgsi_full_instruction *finst) @@ -289,6 +350,18 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc; fsrc = &finst->FullSrcRegisters[i]; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + case TGSI_FILE_CONSTANT: + case TGSI_FILE_TEMPORARY: + if (!src_native_swz(fpc, fsrc, &src[i])) + continue; + break; + default: + break; + } + switch (fsrc->SrcRegister.File) { case TGSI_FILE_INPUT: if (ai == -1 || ai == fsrc->SrcRegister.Index) { @@ -625,6 +698,12 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) if (!fp->buffer) fp->buffer = ws->buffer_create(ws, 0x100); +#if 0 + int i; + for (i = 0; i < fp->insn_len; i++) + NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); +#endif + nv40->pipe.winsys->buffer_data(nv40->pipe.winsys, fp->buffer, fp->insn_len * sizeof(uint32_t), fp->insn, diff --git a/src/mesa/pipe/nv40/nv40_shader.h b/src/mesa/pipe/nv40/nv40_shader.h index 855562eb0d1..207dafd7483 100644 --- a/src/mesa/pipe/nv40/nv40_shader.h +++ b/src/mesa/pipe/nv40/nv40_shader.h @@ -353,6 +353,8 @@ # define NV40_FP_OP_OPCODE_RCP 0x1A # define NV40_FP_OP_OPCODE_EX2 0x1C # define NV40_FP_OP_OPCODE_LG2 0x1D +# define NV40_FP_OP_OPCODE_STR 0x20 +# define NV40_FP_OP_OPCODE_SFL 0x21 # define NV40_FP_OP_OPCODE_COS 0x22 # define NV40_FP_OP_OPCODE_SIN 0x23 # define NV40_FP_OP_OPCODE_PK2H 0x24 -- cgit v1.2.3 From 25acd4eba0b6863bc34d4aeb7ba93fb8f95f5d13 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 Nov 2007 12:02:12 +1100 Subject: nv40: point sprites --- src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c | 2 ++ src/mesa/pipe/nv40/nv40_context.c | 2 +- src/mesa/pipe/nv40/nv40_state.c | 14 ++++++++++++++ src/mesa/pipe/nv40/nv40_state.h | 1 + 4 files changed, 18 insertions(+), 1 deletion(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c index 584d236f349..67df43647aa 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c @@ -25,6 +25,7 @@ #define need_GL_ARB_fragment_program #define need_GL_ARB_multisample #define need_GL_ARB_occlusion_query +#define need_GL_ARB_point_parameters #define need_GL_ARB_texture_compression #define need_GL_ARB_vertex_program #define need_GL_ARB_vertex_buffer_object @@ -44,6 +45,7 @@ const struct dri_extension nv40_extensions[] = { "GL_ARB_fragment_program", NULL }, { "GL_ARB_multisample", GL_ARB_multisample_functions }, { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, + { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, { "GL_ARB_texture_border_clamp", NULL }, { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, { "GL_ARB_texture_cube_map", NULL }, diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index ba9d875e0ed..8706ed1c07c 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -54,7 +54,7 @@ nv40_get_param(struct pipe_context *pipe, int param) case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: - return 0; + return 1; case PIPE_CAP_MAX_RENDER_TARGETS: return 4; case PIPE_CAP_OCCLUSION_QUERY: diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index b1928a5f91a..1b8989b8443 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -278,6 +278,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { struct nv40_rasterizer_state *rs; + int i; /*XXX: ignored: * light_twoside @@ -326,6 +327,16 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, break; } + if (cso->point_sprite) { + rs->point_sprite = (1 << 0); + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + rs->point_sprite |= (1 << (8 + i)); + } + } else { + rs->point_sprite = 0; + } + return (void *)rs; } @@ -358,6 +369,9 @@ nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); OUT_RING (rs->poly_stipple_en); + + BEGIN_RING(curie, NV40TCL_POINT_SPRITE, 1); + OUT_RING (rs->point_sprite); } static void diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index d0114972ef2..c6d22ceaa86 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -51,6 +51,7 @@ struct nv40_rasterizer_state { uint32_t cull_face; uint32_t cull_face_en; + uint32_t point_sprite; }; struct nv40_vertex_program { -- cgit v1.2.3 From b8c5419faf0db76f38aa1dae8f8cfdf9660bc8ae Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 Nov 2007 12:06:42 +1100 Subject: nv40: remove unused stuff --- src/mesa/pipe/nv40/nv40_state.c | 17 +---------------- 1 file changed, 1 insertion(+), 16 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 1b8989b8443..3f23f6cb609 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -546,20 +546,12 @@ nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, } } -static void -nv40_set_feedback_state(struct pipe_context *pipe, - const struct pipe_feedback_state *feedback) -{ - NOUVEAU_ERR("\n"); -} - #define get_region(surf) ((surf) ? surf->region : NULL) static void nv40_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nouveau_winsys *nvws = nv40->nvws; struct pipe_region *region; uint32_t rt_enable = 0, rt_format = 0; @@ -732,13 +724,6 @@ nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, nv40->dirty |= NV40_NEW_ARRAYS; } -static void -nv40_set_feedback_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_feedback_buffer *fbb) -{ - NOUVEAU_ERR("\n"); -} - void nv40_init_state_functions(struct nv40_context *nv40) { @@ -774,7 +759,6 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.set_clip_state = nv40_set_clip_state; nv40->pipe.set_clear_color_state = nv40_set_clear_color_state; nv40->pipe.set_constant_buffer = nv40_set_constant_buffer; -// nv40->pipe.set_feedback_state = nv40_set_feedback_state; nv40->pipe.set_framebuffer_state = nv40_set_framebuffer_state; nv40->pipe.set_polygon_stipple = nv40_set_polygon_stipple; nv40->pipe.set_sampler_units = nv40_set_sampler_units; @@ -785,6 +769,7 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.set_vertex_buffer = nv40_set_vertex_buffer; nv40->pipe.set_vertex_element = nv40_set_vertex_element; +// nv40->pipe.set_feedback_state = nv40_set_feedback_state; // nv40->pipe.set_feedback_buffer = nv40_set_feedback_buffer; } -- cgit v1.2.3 From 2a2756a019ecdd5406ef84019610a0016868a39b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 Nov 2007 12:39:32 +1100 Subject: nv40: be more flexible with render target setup --- src/mesa/pipe/nv40/nv40_state.c | 172 ++++++++++++++++++++++++++-------------- 1 file changed, 112 insertions(+), 60 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 3f23f6cb609..1e7513c0bbc 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -546,104 +546,156 @@ nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, } } -#define get_region(surf) ((surf) ? surf->region : NULL) static void nv40_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct pipe_region *region; - uint32_t rt_enable = 0, rt_format = 0; + struct pipe_region *region[4], *zregion; + uint32_t rt_enable, rt_format, w, h; + int i, colour_format = 0, zeta_format = 0; + + rt_enable = 0; + for (i = 0; i < 4; i++) { + if (!fb->cbufs[i]) + continue; + + if (colour_format) { + assert(w == fb->cbufs[i]->width); + assert(h == fb->cbufs[i]->height); + assert(colour_format == fb->cbufs[i]->format); + } else { + w = fb->cbufs[i]->width; + h = fb->cbufs[i]->height; + colour_format = fb->cbufs[i]->format; + rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); + region[i] = fb->cbufs[i]->region; + } + } + + if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | + NV40TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV40TCL_RT_ENABLE_MRT; - if ((region = get_region(fb->cbufs[0]))) { - rt_enable |= NV40TCL_RT_ENABLE_COLOR0; + if (fb->zbuf) { + if (colour_format) { + assert(w == fb->zbuf->width); + assert(h == fb->zbuf->height); + } else { + w = fb->zbuf->width; + h = fb->zbuf->height; + } + + zeta_format = fb->zbuf->format; + zregion = fb->zbuf->region; + } + if (fb->sbuf) { + if (colour_format) { + assert(w == fb->sbuf->width); + assert(h == fb->sbuf->height); + } else { + w = fb->zbuf->width; + h = fb->zbuf->height; + } + + if (zeta_format) { + assert(fb->sbuf->format == zeta_format); + assert(fb->sbuf->region == zregion); + } else { + zeta_format = fb->sbuf->format; + zregion = fb->sbuf->region; + } + } + + rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_U_A8_R8_G8_B8: + case 0: + rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_U_R5_G6_B5: + rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + switch (zeta_format) { + case PIPE_FORMAT_U_Z16: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; + break; + case PIPE_FORMAT_Z24_S8: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; + break; + case 0: + break; + default: + assert(0); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(region[0]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); - OUT_RING (region->pitch * region->cpp); - OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (region[0]->pitch * region[0]->cpp); + OUT_RELOCl(region[0]->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); } - if ((region = get_region(fb->cbufs[1]))) { - rt_enable |= NV40TCL_RT_ENABLE_COLOR1; - + if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(region[1]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); - OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RING (region->pitch * region->cpp); + OUT_RELOCl(region[1]->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (region[1]->pitch * region[1]->cpp); } - if ((region = get_region(fb->cbufs[2]))) { - rt_enable |= NV40TCL_RT_ENABLE_COLOR2; - + if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(region[2]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(region[2]->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); - OUT_RING (region->pitch * region->cpp); + OUT_RING (region[2]->pitch * region[2]->cpp); } - if ((region = get_region(fb->cbufs[3]))) { - rt_enable |= NV40TCL_RT_ENABLE_COLOR3; - + if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(region[3]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(region[3]->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); - OUT_RING (region->pitch * region->cpp); + OUT_RING (region[3]->pitch * region[3]->cpp); } - if ((region = get_region(fb->zbuf))) { + if (zeta_format) { BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(region->buffer, + OUT_RELOCo(zregion->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(region->buffer, 0, + OUT_RELOCl(zregion->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); - OUT_RING (region->pitch * region->cpp); + OUT_RING (zregion->pitch * zregion->cpp); } - if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | - NV40TCL_RT_ENABLE_COLOR3)) - rt_enable |= NV40TCL_RT_ENABLE_MRT; BEGIN_RING(curie, NV40TCL_RT_ENABLE, 1); OUT_RING (rt_enable); - - if (0) { - rt_format |= (0 << NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT); - rt_format |= (0 << NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT); - rt_format |= NV40TCL_RT_FORMAT_TYPE_SWIZZLED; - } else { - rt_format |= NV40TCL_RT_FORMAT_TYPE_LINEAR; - } - - if (fb->cbufs[0]->format == PIPE_FORMAT_U_R5_G6_B5) { - rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; - } else { - rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; - } - - if (fb->zbuf && fb->zbuf->format == PIPE_FORMAT_U_Z16) { - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; - } else { - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; - } - BEGIN_RING(curie, NV40TCL_RT_HORIZ, 3); - OUT_RING ((fb->cbufs[0]->width << 16) | 0); - OUT_RING ((fb->cbufs[0]->height << 16) | 0); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); OUT_RING (rt_format); BEGIN_RING(curie, NV40TCL_VIEWPORT_HORIZ, 2); - OUT_RING ((fb->cbufs[0]->width << 16) | 0); - OUT_RING ((fb->cbufs[0]->height << 16) | 0); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); BEGIN_RING(curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((fb->cbufs[0]->width - 1) << 16) | 0); - OUT_RING (((fb->cbufs[0]->height - 1) << 16) | 0); + OUT_RING (((w - 1) << 16) | 0); + OUT_RING (((h - 1) << 16) | 0); } static void -- cgit v1.2.3 From 042b7dfd0e15ccd0fcad0c141477091c006e0815 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 Nov 2007 13:05:56 +1100 Subject: nv40: fix build after gallium changes --- src/mesa/pipe/nv40/nv40_fragprog.c | 6 +++--- src/mesa/pipe/nv40/nv40_vertprog.c | 4 ++-- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index e6fee2bf5ae..ec8cac676e9 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -2,9 +2,9 @@ #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "pipe/tgsi/exec/tgsi_token.h" -#include "pipe/tgsi/exec/tgsi_parse.h" -#include "pipe/tgsi/exec/tgsi_util.h" +#include "pipe/p_shader_tokens.h" +#include "pipe/tgsi/util/tgsi_parse.h" +#include "pipe/tgsi/util/tgsi_util.h" #include "nv40_context.h" #include "nv40_dma.h" diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index 04bdaacc98a..66722bafa58 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -2,8 +2,8 @@ #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "pipe/tgsi/exec/tgsi_token.h" -#include "pipe/tgsi/exec/tgsi_parse.h" +#include "pipe/p_shader_tokens.h" +#include "pipe/tgsi/util/tgsi_parse.h" #include "nv40_context.h" #include "nv40_dma.h" -- cgit v1.2.3 From 97f8c39d916e80c0dbb0eae9a6c5109555a1f1b5 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 5 Dec 2007 11:57:50 +1100 Subject: nouveau: g8x winsys support, yay softpipe! --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 33 ----- .../drivers/dri/nouveau_winsys/nouveau_context.c | 30 ++++ .../drivers/dri/nouveau_winsys/nouveau_context.h | 14 +- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 6 - .../drivers/dri/nouveau_winsys/nouveau_screen.c | 2 + .../drivers/dri/nouveau_winsys/nouveau_screen.h | 2 + .../dri/nouveau_winsys/nouveau_swapbuffers.c | 18 +-- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 45 +++++- src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 92 ++++++------ src/mesa/drivers/dri/nouveau_winsys/nv50_region.c | 154 +++++++++++++++++++-- src/mesa/pipe/nouveau/nouveau_winsys.h | 6 +- src/mesa/pipe/nv40/nv40_fragprog.c | 1 - src/mesa/pipe/nv40/nv40_miptree.c | 1 - src/mesa/pipe/nv40/nv40_region.c | 6 +- src/mesa/pipe/nv40/nv40_state_tex.c | 1 - src/mesa/pipe/nv40/nv40_vbo.c | 1 - 16 files changed, 285 insertions(+), 127 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 3b508cf92b3..f1981b97770 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -31,45 +31,12 @@ int nouveau_bo_init(struct nouveau_device *userdev) { - struct nouveau_device_priv *nv = nouveau_device(userdev); - struct nouveau_channel *chan; - int ret; - - if ((ret = nouveau_channel_alloc(userdev, 0x80000001, 0x80000002, - &nv->bufmgr.channel))) - return ret; - chan = nv->bufmgr.channel; - - if ((ret = nouveau_notifier_alloc(nv->bufmgr.channel, 0x80000003, 1, - &nv->bufmgr.notify))) - return ret; - - if ((ret = nouveau_grobj_alloc(nv->bufmgr.channel, 0x80000004, 0x39, - &nv->bufmgr.m2mf))) - return ret; - - nouveau_notifier_reset(nv->bufmgr.notify, 0); - BEGIN_RING_CH(chan, nv->bufmgr.m2mf, 0x180, 1); - OUT_RING_CH (chan, nv->bufmgr.notify->handle); - BEGIN_RING_CH(chan, nv->bufmgr.m2mf, 0x104, 1); - OUT_RING_CH (chan, 0); - BEGIN_RING_CH(chan, nv->bufmgr.m2mf, 0x100, 1); - OUT_RING_CH (chan, 0); - FIRE_RING_CH (chan); - if ((ret = nouveau_notifier_wait_status(nv->bufmgr.notify, 0, 0, 2000))) - return ret; - return 0; } void nouveau_bo_takedown(struct nouveau_device *userdev) { - struct nouveau_device_priv *nv = nouveau_device(userdev); - - nouveau_notifier_free(&nv->bufmgr.notify); - nouveau_grobj_free(&nv->bufmgr.m2mf); - nouveau_channel_free(&nv->bufmgr.channel); } static int diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c index 67df43647aa..70655a1a7d9 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c @@ -125,6 +125,36 @@ nouveau_context_create(const __GLcontextModes *glVis, debug_control); #endif + /*XXX: Hack up a fake region and buffer object for front buffer. + * This will go away with TTM, replaced with a simple reference + * of the front buffer handle passed to us by the DDX. + */ + { + struct pipe_region *fb_region; + struct nouveau_bo_priv *fb_bo; + + fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); + fb_bo->drm.offset = nv_screen->front_offset; + fb_bo->drm.flags = NOUVEAU_MEM_FB; + fb_bo->drm.size = nv_screen->front_pitch * + nv_screen->front_height; + fb_bo->refcount = 1; + fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; + fb_bo->base.offset = fb_bo->drm.offset; + fb_bo->base.handle = (unsigned long)fb_bo; + fb_bo->base.size = fb_bo->drm.size; + fb_bo->base.device = nv_screen->device; + + fb_region = calloc(1, sizeof(struct pipe_region)); + fb_region->cpp = nv_screen->front_cpp; + fb_region->pitch = nv_screen->front_pitch / fb_region->cpp; + fb_region->height = nv_screen->front_height; + fb_region->refcount = 1; + fb_region->buffer = (void *)fb_bo; + + nv->frontbuffer = fb_region; + } + if ((ret = nouveau_grobj_alloc(nv->channel, 0x00000000, 0x30, &nv->NvNull))) { NOUVEAU_ERR("Error creating NULL object: %d\n", ret); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index f2557af935d..70110f97683 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -31,6 +31,7 @@ struct nouveau_context { drmLock drm_lock; GLboolean locked; struct nouveau_screen *nv_screen; + struct pipe_region *frontbuffer; /* Bufmgr */ struct { @@ -53,14 +54,19 @@ struct nouveau_context { struct nouveau_grobj *NvImageBlit; struct nouveau_grobj *NvGdiRect; struct nouveau_grobj *NvM2MF; + struct nouveau_grobj *Nv2D; uint32_t next_handle; uint32_t next_sequence; /* pipe_region accel */ - int (*region_display)(void); - int (*region_copy)(struct nouveau_context *, struct pipe_region *, - unsigned, unsigned, unsigned, struct pipe_region *, - unsigned, unsigned, unsigned, unsigned, unsigned); + struct pipe_region *region_src, *region_dst; + unsigned region_src_offset, region_dst_offset; + int (*region_copy_prep)(struct nouveau_context *, + struct pipe_region *dst, uint32_t dst_offset, + struct pipe_region *src, uint32_t src_offset); + void (*region_copy)(struct nouveau_context *, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h); + void (*region_copy_done)(struct nouveau_context *); int (*region_fill)(struct nouveau_context *, struct pipe_region *, unsigned, unsigned, unsigned, unsigned, unsigned, unsigned); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 1a21da69857..df98890f8e2 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -40,12 +40,6 @@ struct nouveau_device_priv { drm_context_t ctx; drmLock *lock; int needs_close; - - struct { - struct nouveau_channel *channel; - struct nouveau_notifier *notify; - struct nouveau_grobj *m2mf; - } bufmgr; }; #define nouveau_device(n) ((struct nouveau_device_priv *)(n)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index 16c4324164b..6554082e622 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -61,6 +61,8 @@ nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) nv_screen->front_offset = nv_dri->front_offset; nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); + nv_screen->front_cpp = nv_dri->bpp / 8; + nv_screen->front_height = nv_dri->height; return GL_TRUE; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h index e53cd873a7e..019823bd44d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h @@ -12,6 +12,8 @@ struct nouveau_screen { uint32_t front_offset; uint32_t front_pitch; + uint32_t front_cpp; + uint32_t front_height; }; #endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c index 57598f9cab9..5b2e76b942f 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c @@ -14,7 +14,6 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, const drm_clip_rect_t *rect) { struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; - struct nouveau_screen *nv_screen = nv->nv_screen; struct pipe_region *p_region = surf->region; drm_clip_rect_t *pbox; int nbox, i; @@ -27,17 +26,7 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, pbox = dPriv->pClipRects; nbox = dPriv->numClipRects; - BEGIN_RING(NvCtxSurf2D, 0x184, 2); - OUT_RELOCo(p_region->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RING (nv->channel->vram->handle); - - BEGIN_RING(NvCtxSurf2D, 0x300, 4); - OUT_RING ((p_region->cpp == 4) ? 6 : 4); - OUT_RING ((nv_screen->front_pitch << 16) | - (p_region->pitch * p_region->cpp)); - OUT_RELOCl(p_region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RING (nv_screen->front_offset); - + nv->region_copy_prep(nv, nv->frontbuffer, 0, p_region, 0); for (i = 0; i < nbox; i++, pbox++) { int sx, sy, dx, dy, w, h; @@ -48,10 +37,7 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, w = pbox->x2 - pbox->x1; h = pbox->y2 - pbox->y1; - BEGIN_RING(NvImageBlit, 0x300, 3); - OUT_RING ((sy << 16) | sx); - OUT_RING ((dy << 16) | dx); - OUT_RING (( h << 16) | w); + nv->region_copy(nv, dx, dy, sx, sy, w, h); } FIRE_RING(); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index bdc89b95318..6840ec5f754 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -127,6 +127,45 @@ nouveau_pipe_dma_kickoff(struct nouveau_channel *userchan) FIRE_RING_CH(userchan); } +static int +nouveau_pipe_region_copy(struct nouveau_winsys *nvws, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + struct pipe_region *src, unsigned src_offset, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct nouveau_context *nv = nvws->nv; + + if (nv->region_copy_prep(nv, dst, dst_offset, src, src_offset)) + return 1; + nv->region_copy(nv, dx, dy, sx, sy, w, h); + nv->region_copy_done(nv); + + return 0; +} + +static int +nouveau_pipe_region_fill(struct nouveau_winsys *nvws, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + unsigned w, unsigned h, unsigned value) +{ + if (nvws->nv->region_fill(nvws->nv, dst, dst_offset, dx, dy, + w, h, value)) + return 1; + return 0; +} + +static int +nouveau_pipe_region_data(struct nouveau_winsys *nvws, struct pipe_region *dst, + unsigned dst_offset, unsigned dx, unsigned dy, + const void *src, unsigned src_pitch, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + if (nvws->nv->region_data(nvws->nv, dst, dst_offset, dx, dy, src, + src_pitch, sx, sy, w, h)) + return 1; + return 0; +} + struct pipe_context * nouveau_pipe_create(struct nouveau_context *nv) { @@ -168,9 +207,9 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->notifier_retval = nouveau_notifier_return_val; nvws->notifier_wait = nouveau_notifier_wait_status; - nvws->region_copy = nv->region_copy; - nvws->region_fill = nv->region_fill; - nvws->region_data = nv->region_data; + nvws->region_copy = nouveau_pipe_region_copy; + nvws->region_fill = nouveau_pipe_region_fill; + nvws->region_data = nouveau_pipe_region_data; return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c index 0fdb38e105c..d41588d8907 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c @@ -26,26 +26,19 @@ nv04_rect_format(int cpp) } } -static int -nv04_region_display(void) -{ - NOUVEAU_ERR("unimplemented\n"); - return 0; -} - -static int -nv04_region_copy_m2mf(struct nouveau_context *nv, struct pipe_region *dst, - unsigned dst_offset, struct pipe_region *src, - unsigned src_offset, unsigned line_len, unsigned height) +static void +nv04_region_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) { - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_WR); + struct pipe_region *dst = nv->region_dst; + struct pipe_region *src = nv->region_dst; + unsigned dst_offset, src_offset; + + dst_offset = nv->region_dst_offset + (dy * dst->pitch + dx) * dst->cpp; + src_offset = nv->region_src_offset + (sy * src->pitch + sx) * src->cpp; - while (height) { - int count = (height > 2047) ? 2047 : height; + while (h) { + int count = (h > 2047) ? 2047 : h; BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | @@ -54,32 +47,31 @@ nv04_region_copy_m2mf(struct nouveau_context *nv, struct pipe_region *dst, NOUVEAU_BO_GART | NOUVEAU_BO_WR); OUT_RING (src->pitch * src->cpp); OUT_RING (dst->pitch * dst->cpp); - OUT_RING (line_len); + OUT_RING (w * src->cpp); OUT_RING (count); OUT_RING (0x0101); OUT_RING (0); - height -= count; + h -= count; src_offset += src->pitch * count; dst_offset += dst->pitch * count; } +} - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(NvM2MF, 0x104, 1); - OUT_RING (0); - BEGIN_RING(NvM2MF, 0x100, 1); - OUT_RING (0); - FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); - - return 0; +static void +nv04_region_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + BEGIN_RING(NvImageBlit, 0x0300, 3); + OUT_RING ((sy << 16) | sx); + OUT_RING ((dy << 16) | dx); + OUT_RING (( h << 16) | w); } static int -nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - struct pipe_region *src, unsigned src_offset, - unsigned sx, unsigned sy, unsigned w, unsigned h) +nv04_region_copy_prep(struct nouveau_context *nv, + struct pipe_region *dst, unsigned dst_offset, + struct pipe_region *src, unsigned src_offset) { int format; @@ -90,10 +82,18 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, * to NV_MEMORY_TO_MEMORY_FORMAT in this case. */ if ((src_offset & 63) || (dst_offset & 63)) { - dst_offset += (dy * dst->pitch + dx) * dst->cpp; - src_offset += (sy * src->pitch + sx) * src->cpp; - return nv04_region_copy_m2mf(nv, dst, dst_offset, src, - src_offset, w * src->cpp, h); + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_WR); + + nv->region_copy = nv04_region_copy_m2mf; + nv->region_dst = dst; + nv->region_dst_offset = dst_offset; + nv->region_src = src; + nv->region_src_offset = src_offset; + return 0; } @@ -101,21 +101,24 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); return 1; } + nv->region_copy = nv04_region_copy_blit; BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (format); OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvImageBlit, 0x0300, 3); - OUT_RING ((sy << 16) | sx); - OUT_RING ((dy << 16) | dx); - OUT_RING (( h << 16) | w); + return 0; +} +static void +nv04_region_copy_done(struct nouveau_context *nv) +{ nouveau_notifier_reset(nv->sync_notifier, 0); BEGIN_RING(NvGdiRect, 0x104, 1); OUT_RING (0); @@ -123,8 +126,6 @@ nv04_region_copy(struct nouveau_context *nv, struct pipe_region *dst, OUT_RING (0); FIRE_RING(); nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); - - return 0; } static int @@ -238,8 +239,9 @@ nouveau_region_init_nv04(struct nouveau_context *nv) BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); - nv->region_display = nv04_region_display; - nv->region_copy = nv04_region_copy; + nv->region_copy_prep = nv04_region_copy_prep; + nv->region_copy = nv04_region_copy_blit; + nv->region_copy_done = nv04_region_copy_done; nv->region_fill = nv04_region_fill; nv->region_data = nv04_region_data; return 0; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c index c7450c5c8d0..75bfcaf6c8a 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c @@ -2,21 +2,94 @@ #include "nouveau_context.h" +static INLINE int +nv50_format(int cpp) +{ + switch (cpp) { + case 4: return NV50_2D_DST_FORMAT_32BPP; + case 3: return NV50_2D_DST_FORMAT_24BPP; + case 2: return NV50_2D_DST_FORMAT_16BPP; + case 1: return NV50_2D_DST_FORMAT_8BPP; + default: + return -1; + } +} + static int -nv50_region_display(void) +nv50_region_copy_prep(struct nouveau_context *nv, + struct pipe_region *dst, unsigned dst_offset, + struct pipe_region *src, unsigned src_offset) { - NOUVEAU_ERR("unimplemented\n"); + int surf_format; + + assert(src->cpp == dst->cpp); + + surf_format = nv50_format(dst->cpp); + assert(surf_format >= 0); + + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + OUT_RELOCh(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); + OUT_RING (0); + OUT_RING (0); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + + BEGIN_RING(Nv2D, NV50_2D_SRC_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_SRC_PITCH, 5); + OUT_RING (src->pitch * src->cpp); + OUT_RING (src->pitch); + OUT_RING (src->height); + OUT_RELOCh(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + return 0; } -static int -nv50_region_copy(struct nouveau_context *nv, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - struct pipe_region *src, unsigned src_offset, +static void +nv50_region_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, unsigned sx, unsigned sy, unsigned w, unsigned h) { - NOUVEAU_ERR("unimplemented!!\n"); - return 0; + BEGIN_RING(Nv2D, 0x0110, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, NV50_2D_BLIT_DST_X, 12); + OUT_RING (dx); + OUT_RING (dy); + OUT_RING (w); + OUT_RING (h); + OUT_RING (0); + OUT_RING (1); + OUT_RING (0); + OUT_RING (1); + OUT_RING (0); + OUT_RING (sx); + OUT_RING (0); + OUT_RING (sy); +} + +static void +nv50_region_copy_done(struct nouveau_context *nv) +{ + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(Nv2D, 0x104, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); } static int @@ -25,7 +98,52 @@ nv50_region_fill(struct nouveau_context *nv, unsigned dx, unsigned dy, unsigned w, unsigned h, unsigned value) { - NOUVEAU_ERR("unimplemented!!\n"); + int surf_format, rect_format; + + surf_format = nv50_format(dst->cpp); + if (surf_format < 0) + return 1; + + rect_format = nv50_format(dst->cpp); + if (rect_format < 0) + return 1; + + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY1, 1); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + OUT_RELOCh(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); + OUT_RING (0); + OUT_RING (0); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + + BEGIN_RING(Nv2D, 0x0580, 4); + OUT_RING (4); + OUT_RING (rect_format); + OUT_RING (value); + + BEGIN_RING(Nv2D, NV50_2D_RECT_X1, 4); + OUT_RING (dx); + OUT_RING (dy); + OUT_RING (dx + w); + OUT_RING (dy + h); + + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(Nv2D, 0x104, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); + return 0; } @@ -42,8 +160,24 @@ nv50_region_data(struct nouveau_context *nv, struct pipe_region *dst, int nouveau_region_init_nv50(struct nouveau_context *nv) { - nv->region_display = nv50_region_display; + int ret; + + ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, NV50_2D, + &nv->Nv2D); + if (ret) + return ret; + + BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RING (nv->channel->vram->handle); + OUT_RING (nv->channel->vram->handle); + BEGIN_RING(Nv2D, NV50_2D_OPERATION, 1); + OUT_RING (NV50_2D_OPERATION_SRCCOPY); + + nv->region_copy_prep = nv50_region_copy_prep; nv->region_copy = nv50_region_copy; + nv->region_copy_done = nv50_region_copy_done; nv->region_fill = nv50_region_fill; nv->region_data = nv50_region_data; return 0; diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 28f17cdc8fa..1dc15f93ecd 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -53,13 +53,13 @@ struct nouveau_winsys { int (*notifier_wait)(struct nouveau_notifier *, int id, int status, int timeout); - int (*region_copy)(struct nouveau_context *, struct pipe_region *, + int (*region_copy)(struct nouveau_winsys *, struct pipe_region *, unsigned, unsigned, unsigned, struct pipe_region *, unsigned, unsigned, unsigned, unsigned, unsigned); - int (*region_fill)(struct nouveau_context *, struct pipe_region *, + int (*region_fill)(struct nouveau_winsys *, struct pipe_region *, unsigned, unsigned, unsigned, unsigned, unsigned, unsigned); - int (*region_data)(struct nouveau_context *, struct pipe_region *, + int (*region_data)(struct nouveau_winsys *, struct pipe_region *, unsigned, unsigned, unsigned, const void *, unsigned, unsigned, unsigned, unsigned, unsigned); }; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index ec8cac676e9..8f3eeb1f481 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -620,7 +620,6 @@ nv40_fragprog_translate(struct nv40_context *nv40, { struct tgsi_parse_context parse; struct nv40_fpc *fpc = NULL; - int i; fpc = calloc(1, sizeof(struct nv40_fpc)); if (!fpc) diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 15639f01e90..f5c4206f402 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -7,7 +7,6 @@ boolean nv40_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; boolean swizzled = FALSE; uint width = mt->width0, height = mt->height0, depth = mt->depth0; uint offset; diff --git a/src/mesa/pipe/nv40/nv40_region.c b/src/mesa/pipe/nv40/nv40_region.c index f62bf89d183..572b82bca3a 100644 --- a/src/mesa/pipe/nv40/nv40_region.c +++ b/src/mesa/pipe/nv40/nv40_region.c @@ -42,7 +42,7 @@ nv40_region_data(struct pipe_context *pipe, struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nouveau_winsys *nvws = nv40->nvws; - nvws->region_data(nvws->nv, dst, dst_offset, dstx, dsty, + nvws->region_data(nvws, dst, dst_offset, dstx, dsty, src, src_pitch, srcx, srcy, width, height); } @@ -56,7 +56,7 @@ nv40_region_copy(struct pipe_context *pipe, struct pipe_region *dst, struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nouveau_winsys *nvws = nv40->nvws; - nvws->region_copy(nvws->nv, dst, dst_offset, dstx, dsty, + nvws->region_copy(nvws, dst, dst_offset, dstx, dsty, src, src_offset, srcx, srcy, width, height); } @@ -69,7 +69,7 @@ nv40_region_fill(struct pipe_context *pipe, struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nouveau_winsys *nvws = nv40->nvws; - nvws->region_fill(nvws->nv, dst, dst_offset, dstx, dsty, + nvws->region_fill(nvws, dst, dst_offset, dstx, dsty, width, height, value); } diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index bd779df9b36..6902b41ba71 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -53,7 +53,6 @@ nv40_tex_format(uint pipe_format) static void nv40_tex_unit_enable(struct nv40_context *nv40, int unit) { - struct nouveau_winsys *nvws = nv40->nvws; struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; struct pipe_mipmap_tree *mt = nv40->tex_miptree[unit]; struct nv40_texture_format *tf; diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 2da0001b288..cc71fb99de2 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -175,7 +175,6 @@ nv40_vbo_format_to_ncomp(uint format) void nv40_vbo_arrays_update(struct nv40_context *nv40) { - struct nouveau_winsys *nvws = nv40->nvws; struct nv40_vertex_program *vp = nv40->vertprog.active; uint32_t inputs, vtxfmt[16]; int hw, num_hw; -- cgit v1.2.3 From 47b418b8fa5fd242e9021503d6ec329ac3d56fb0 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 5 Dec 2007 14:01:01 +1100 Subject: nv50: 3d driver skeleton --- src/mesa/drivers/dri/nouveau_winsys/Makefile | 3 +- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 4 + src/mesa/drivers/dri/nouveau_winsys/nv50_region.c | 2 +- src/mesa/pipe/Makefile | 1 + src/mesa/pipe/nouveau/nouveau_gldefs.h | 198 +++++++++++++++++ src/mesa/pipe/nouveau/nouveau_winsys.h | 3 + src/mesa/pipe/nv40/nv40_context.h | 1 + src/mesa/pipe/nv40/nv40_state.c | 2 - src/mesa/pipe/nv40/nv40_vbo.c | 1 - src/mesa/pipe/nv40/nvgl_pipe.h | 198 ----------------- src/mesa/pipe/nv50/Makefile | 26 +++ src/mesa/pipe/nv50/nv50_clear.c | 15 ++ src/mesa/pipe/nv50/nv50_context.c | 215 ++++++++++++++++++ src/mesa/pipe/nv50/nv50_context.h | 62 ++++++ src/mesa/pipe/nv50/nv50_dma.h | 62 ++++++ src/mesa/pipe/nv50/nv50_draw.c | 63 ++++++ src/mesa/pipe/nv50/nv50_miptree.c | 13 ++ src/mesa/pipe/nv50/nv50_query.c | 23 ++ src/mesa/pipe/nv50/nv50_region.c | 85 ++++++++ src/mesa/pipe/nv50/nv50_state.c | 242 +++++++++++++++++++++ src/mesa/pipe/nv50/nv50_state.h | 7 + src/mesa/pipe/nv50/nv50_surface.c | 229 +++++++++++++++++++ src/mesa/pipe/nv50/nv50_vbo.c | 25 +++ 23 files changed, 1277 insertions(+), 203 deletions(-) create mode 100644 src/mesa/pipe/nouveau/nouveau_gldefs.h delete mode 100644 src/mesa/pipe/nv40/nvgl_pipe.h create mode 100644 src/mesa/pipe/nv50/Makefile create mode 100644 src/mesa/pipe/nv50/nv50_clear.c create mode 100644 src/mesa/pipe/nv50/nv50_context.c create mode 100644 src/mesa/pipe/nv50/nv50_context.h create mode 100644 src/mesa/pipe/nv50/nv50_dma.h create mode 100644 src/mesa/pipe/nv50/nv50_draw.c create mode 100644 src/mesa/pipe/nv50/nv50_miptree.c create mode 100644 src/mesa/pipe/nv50/nv50_query.c create mode 100644 src/mesa/pipe/nv50/nv50_region.c create mode 100644 src/mesa/pipe/nv50/nv50_state.c create mode 100644 src/mesa/pipe/nv50/nv50_state.h create mode 100644 src/mesa/pipe/nv50/nv50_surface.c create mode 100644 src/mesa/pipe/nv50/nv50_vbo.c diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index 97b861526f4..292ec5e0a43 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -8,7 +8,8 @@ MINIGLX_SOURCES = PIPE_DRIVERS = \ $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a \ - $(TOP)/src/mesa/pipe/nv40/libnv40.a + $(TOP)/src/mesa/pipe/nv40/libnv40.a \ + $(TOP)/src/mesa/pipe/nv50/libnv50.a DRIVER_SOURCES = \ nouveau_bo.c \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index 6840ec5f754..e35b6009e34 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -181,6 +181,10 @@ nouveau_pipe_create(struct nouveau_context *nv) case 0x40: hw_create = nv40_create; break; + case 0x50: + case 0x80: + hw_create = nv50_create; + break; default: NOUVEAU_ERR("Unknown chipset NV%02x\n", (int)nv->chipset); return NULL; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c index 75bfcaf6c8a..168cbdfb7c2 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c @@ -125,7 +125,7 @@ nv50_region_fill(struct nouveau_context *nv, OUT_RING (dst->pitch); OUT_RING (dst->height); - BEGIN_RING(Nv2D, 0x0580, 4); + BEGIN_RING(Nv2D, 0x0580, 3); OUT_RING (4); OUT_RING (rect_format); OUT_RING (value); diff --git a/src/mesa/pipe/Makefile b/src/mesa/pipe/Makefile index 135f1fd19d7..7394ad7f611 100644 --- a/src/mesa/pipe/Makefile +++ b/src/mesa/pipe/Makefile @@ -3,6 +3,7 @@ default: cd i915simple ; make cd failover ; make cd nv40; make + cd nv50; make clean: rm -f `find . -name \*.[oa]` diff --git a/src/mesa/pipe/nouveau/nouveau_gldefs.h b/src/mesa/pipe/nouveau/nouveau_gldefs.h new file mode 100644 index 00000000000..15ff318023d --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_gldefs.h @@ -0,0 +1,198 @@ +#ifndef __NVGL_PIPE_H__ +#define __NVGL_PIPE_H__ + +#include + +static INLINE unsigned +nvgl_blend_func(unsigned factor) +{ + switch (factor) { + case PIPE_BLENDFACTOR_ONE: + return GL_ONE; + case PIPE_BLENDFACTOR_SRC_COLOR: + return GL_SRC_COLOR; + case PIPE_BLENDFACTOR_SRC_ALPHA: + return GL_SRC_ALPHA; + case PIPE_BLENDFACTOR_DST_ALPHA: + return GL_DST_ALPHA; + case PIPE_BLENDFACTOR_DST_COLOR: + return GL_DST_COLOR; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + return GL_SRC_ALPHA_SATURATE; + case PIPE_BLENDFACTOR_CONST_COLOR: + return GL_CONSTANT_COLOR; + case PIPE_BLENDFACTOR_CONST_ALPHA: + return GL_CONSTANT_ALPHA; + case PIPE_BLENDFACTOR_ZERO: + return GL_ZERO; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + return GL_ONE_MINUS_SRC_COLOR; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + return GL_ONE_MINUS_SRC_ALPHA; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + return GL_ONE_MINUS_DST_ALPHA; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + return GL_ONE_MINUS_DST_COLOR; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + return GL_ONE_MINUS_CONSTANT_COLOR; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + return GL_ONE_MINUS_CONSTANT_ALPHA; + default: + return GL_ONE; + } +} + +static INLINE unsigned +nvgl_blend_eqn(unsigned func) +{ + switch (func) { + case PIPE_BLEND_ADD: + return GL_FUNC_ADD; + case PIPE_BLEND_SUBTRACT: + return GL_FUNC_SUBTRACT; + case PIPE_BLEND_REVERSE_SUBTRACT: + return GL_FUNC_REVERSE_SUBTRACT; + case PIPE_BLEND_MIN: + return GL_MIN; + case PIPE_BLEND_MAX: + return GL_MAX; + default: + return GL_FUNC_ADD; + } +} + +static INLINE unsigned +nvgl_logicop_func(unsigned func) +{ + switch (func) { + case PIPE_LOGICOP_CLEAR: + return GL_CLEAR; + case PIPE_LOGICOP_NOR: + return GL_NOR; + case PIPE_LOGICOP_AND_INVERTED: + return GL_AND_INVERTED; + case PIPE_LOGICOP_COPY_INVERTED: + return GL_COPY_INVERTED; + case PIPE_LOGICOP_AND_REVERSE: + return GL_AND_REVERSE; + case PIPE_LOGICOP_INVERT: + return GL_INVERT; + case PIPE_LOGICOP_XOR: + return GL_XOR; + case PIPE_LOGICOP_NAND: + return GL_NAND; + case PIPE_LOGICOP_AND: + return GL_AND; + case PIPE_LOGICOP_EQUIV: + return GL_EQUIV; + case PIPE_LOGICOP_NOOP: + return GL_NOOP; + case PIPE_LOGICOP_OR_INVERTED: + return GL_OR_INVERTED; + case PIPE_LOGICOP_COPY: + return GL_COPY; + case PIPE_LOGICOP_OR_REVERSE: + return GL_OR_REVERSE; + case PIPE_LOGICOP_OR: + return GL_OR; + case PIPE_LOGICOP_SET: + return GL_SET; + default: + return GL_CLEAR; + } +} + +static INLINE unsigned +nvgl_comparison_op(unsigned op) +{ + switch (op) { + case PIPE_FUNC_NEVER: + return GL_NEVER; + case PIPE_FUNC_LESS: + return GL_LESS; + case PIPE_FUNC_EQUAL: + return GL_EQUAL; + case PIPE_FUNC_LEQUAL: + return GL_LEQUAL; + case PIPE_FUNC_GREATER: + return GL_GREATER; + case PIPE_FUNC_NOTEQUAL: + return GL_NOTEQUAL; + case PIPE_FUNC_GEQUAL: + return GL_GEQUAL; + case PIPE_FUNC_ALWAYS: + return GL_ALWAYS; + default: + return GL_NEVER; + } +} + +static INLINE unsigned +nvgl_polygon_mode(unsigned mode) +{ + switch (mode) { + case PIPE_POLYGON_MODE_FILL: + return GL_FILL; + case PIPE_POLYGON_MODE_LINE: + return GL_LINE; + case PIPE_POLYGON_MODE_POINT: + return GL_POINT; + default: + return GL_FILL; + } +} + +static INLINE unsigned +nvgl_stencil_op(unsigned op) +{ + switch (op) { + case PIPE_STENCIL_OP_KEEP: + return GL_KEEP; + case PIPE_STENCIL_OP_ZERO: + return GL_ZERO; + case PIPE_STENCIL_OP_REPLACE: + return GL_REPLACE; + case PIPE_STENCIL_OP_INCR: + return GL_INCR; + case PIPE_STENCIL_OP_DECR: + return GL_DECR; + case PIPE_STENCIL_OP_INCR_WRAP: + return GL_INCR_WRAP; + case PIPE_STENCIL_OP_DECR_WRAP: + return GL_DECR_WRAP; + case PIPE_STENCIL_OP_INVERT: + return GL_INVERT; + default: + return GL_KEEP; + } +} + +static INLINE unsigned +nvgl_primitive(unsigned prim) { + switch (prim) { + case PIPE_PRIM_POINTS: + return GL_POINTS + 1; + case PIPE_PRIM_LINES: + return GL_LINES + 1; + case PIPE_PRIM_LINE_LOOP: + return GL_LINE_LOOP + 1; + case PIPE_PRIM_LINE_STRIP: + return GL_LINE_STRIP + 1; + case PIPE_PRIM_TRIANGLES: + return GL_TRIANGLES + 1; + case PIPE_PRIM_TRIANGLE_STRIP: + return GL_TRIANGLE_STRIP + 1; + case PIPE_PRIM_TRIANGLE_FAN: + return GL_TRIANGLE_FAN + 1; + case PIPE_PRIM_QUADS: + return GL_QUADS + 1; + case PIPE_PRIM_QUAD_STRIP: + return GL_QUAD_STRIP + 1; + case PIPE_PRIM_POLYGON: + return GL_POLYGON + 1; + default: + return GL_POINTS + 1; + } +} + +#endif diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 1dc15f93ecd..40e4bfae54e 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -67,4 +67,7 @@ struct nouveau_winsys { extern struct pipe_context * nv40_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); +extern struct pipe_context * +nv50_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); + #endif diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index ec07a883706..65a197ae2f1 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -8,6 +8,7 @@ #include "pipe/draw/draw_vertex.h" #include "pipe/nouveau/nouveau_winsys.h" +#include "pipe/nouveau/nouveau_gldefs.h" #include "nv40_state.h" diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 1e7513c0bbc..937de7face1 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -6,8 +6,6 @@ #include "nv40_dma.h" #include "nv40_state.h" -#include "nvgl_pipe.h" - static void * nv40_alpha_test_state_create(struct pipe_context *pipe, const struct pipe_alpha_test_state *cso) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index cc71fb99de2..5e919c5ca97 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -5,7 +5,6 @@ #include "nv40_context.h" #include "nv40_dma.h" #include "nv40_state.h" -#include "nvgl_pipe.h" boolean nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, diff --git a/src/mesa/pipe/nv40/nvgl_pipe.h b/src/mesa/pipe/nv40/nvgl_pipe.h deleted file mode 100644 index 15ff318023d..00000000000 --- a/src/mesa/pipe/nv40/nvgl_pipe.h +++ /dev/null @@ -1,198 +0,0 @@ -#ifndef __NVGL_PIPE_H__ -#define __NVGL_PIPE_H__ - -#include - -static INLINE unsigned -nvgl_blend_func(unsigned factor) -{ - switch (factor) { - case PIPE_BLENDFACTOR_ONE: - return GL_ONE; - case PIPE_BLENDFACTOR_SRC_COLOR: - return GL_SRC_COLOR; - case PIPE_BLENDFACTOR_SRC_ALPHA: - return GL_SRC_ALPHA; - case PIPE_BLENDFACTOR_DST_ALPHA: - return GL_DST_ALPHA; - case PIPE_BLENDFACTOR_DST_COLOR: - return GL_DST_COLOR; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - return GL_SRC_ALPHA_SATURATE; - case PIPE_BLENDFACTOR_CONST_COLOR: - return GL_CONSTANT_COLOR; - case PIPE_BLENDFACTOR_CONST_ALPHA: - return GL_CONSTANT_ALPHA; - case PIPE_BLENDFACTOR_ZERO: - return GL_ZERO; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - return GL_ONE_MINUS_SRC_COLOR; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - return GL_ONE_MINUS_SRC_ALPHA; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - return GL_ONE_MINUS_DST_ALPHA; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - return GL_ONE_MINUS_DST_COLOR; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - return GL_ONE_MINUS_CONSTANT_COLOR; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - return GL_ONE_MINUS_CONSTANT_ALPHA; - default: - return GL_ONE; - } -} - -static INLINE unsigned -nvgl_blend_eqn(unsigned func) -{ - switch (func) { - case PIPE_BLEND_ADD: - return GL_FUNC_ADD; - case PIPE_BLEND_SUBTRACT: - return GL_FUNC_SUBTRACT; - case PIPE_BLEND_REVERSE_SUBTRACT: - return GL_FUNC_REVERSE_SUBTRACT; - case PIPE_BLEND_MIN: - return GL_MIN; - case PIPE_BLEND_MAX: - return GL_MAX; - default: - return GL_FUNC_ADD; - } -} - -static INLINE unsigned -nvgl_logicop_func(unsigned func) -{ - switch (func) { - case PIPE_LOGICOP_CLEAR: - return GL_CLEAR; - case PIPE_LOGICOP_NOR: - return GL_NOR; - case PIPE_LOGICOP_AND_INVERTED: - return GL_AND_INVERTED; - case PIPE_LOGICOP_COPY_INVERTED: - return GL_COPY_INVERTED; - case PIPE_LOGICOP_AND_REVERSE: - return GL_AND_REVERSE; - case PIPE_LOGICOP_INVERT: - return GL_INVERT; - case PIPE_LOGICOP_XOR: - return GL_XOR; - case PIPE_LOGICOP_NAND: - return GL_NAND; - case PIPE_LOGICOP_AND: - return GL_AND; - case PIPE_LOGICOP_EQUIV: - return GL_EQUIV; - case PIPE_LOGICOP_NOOP: - return GL_NOOP; - case PIPE_LOGICOP_OR_INVERTED: - return GL_OR_INVERTED; - case PIPE_LOGICOP_COPY: - return GL_COPY; - case PIPE_LOGICOP_OR_REVERSE: - return GL_OR_REVERSE; - case PIPE_LOGICOP_OR: - return GL_OR; - case PIPE_LOGICOP_SET: - return GL_SET; - default: - return GL_CLEAR; - } -} - -static INLINE unsigned -nvgl_comparison_op(unsigned op) -{ - switch (op) { - case PIPE_FUNC_NEVER: - return GL_NEVER; - case PIPE_FUNC_LESS: - return GL_LESS; - case PIPE_FUNC_EQUAL: - return GL_EQUAL; - case PIPE_FUNC_LEQUAL: - return GL_LEQUAL; - case PIPE_FUNC_GREATER: - return GL_GREATER; - case PIPE_FUNC_NOTEQUAL: - return GL_NOTEQUAL; - case PIPE_FUNC_GEQUAL: - return GL_GEQUAL; - case PIPE_FUNC_ALWAYS: - return GL_ALWAYS; - default: - return GL_NEVER; - } -} - -static INLINE unsigned -nvgl_polygon_mode(unsigned mode) -{ - switch (mode) { - case PIPE_POLYGON_MODE_FILL: - return GL_FILL; - case PIPE_POLYGON_MODE_LINE: - return GL_LINE; - case PIPE_POLYGON_MODE_POINT: - return GL_POINT; - default: - return GL_FILL; - } -} - -static INLINE unsigned -nvgl_stencil_op(unsigned op) -{ - switch (op) { - case PIPE_STENCIL_OP_KEEP: - return GL_KEEP; - case PIPE_STENCIL_OP_ZERO: - return GL_ZERO; - case PIPE_STENCIL_OP_REPLACE: - return GL_REPLACE; - case PIPE_STENCIL_OP_INCR: - return GL_INCR; - case PIPE_STENCIL_OP_DECR: - return GL_DECR; - case PIPE_STENCIL_OP_INCR_WRAP: - return GL_INCR_WRAP; - case PIPE_STENCIL_OP_DECR_WRAP: - return GL_DECR_WRAP; - case PIPE_STENCIL_OP_INVERT: - return GL_INVERT; - default: - return GL_KEEP; - } -} - -static INLINE unsigned -nvgl_primitive(unsigned prim) { - switch (prim) { - case PIPE_PRIM_POINTS: - return GL_POINTS + 1; - case PIPE_PRIM_LINES: - return GL_LINES + 1; - case PIPE_PRIM_LINE_LOOP: - return GL_LINE_LOOP + 1; - case PIPE_PRIM_LINE_STRIP: - return GL_LINE_STRIP + 1; - case PIPE_PRIM_TRIANGLES: - return GL_TRIANGLES + 1; - case PIPE_PRIM_TRIANGLE_STRIP: - return GL_TRIANGLE_STRIP + 1; - case PIPE_PRIM_TRIANGLE_FAN: - return GL_TRIANGLE_FAN + 1; - case PIPE_PRIM_QUADS: - return GL_QUADS + 1; - case PIPE_PRIM_QUAD_STRIP: - return GL_QUAD_STRIP + 1; - case PIPE_PRIM_POLYGON: - return GL_POLYGON + 1; - default: - return GL_POINTS + 1; - } -} - -#endif diff --git a/src/mesa/pipe/nv50/Makefile b/src/mesa/pipe/nv50/Makefile new file mode 100644 index 00000000000..af3fa0e7ecd --- /dev/null +++ b/src/mesa/pipe/nv50/Makefile @@ -0,0 +1,26 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv50 + +DRIVER_SOURCES = \ + nv50_clear.c \ + nv50_context.c \ + nv50_draw.c \ + nv50_miptree.c \ + nv50_query.c \ + nv50_region.c \ + nv50_state.c \ + nv50_surface.c \ + nv50_vbo.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../Makefile.template + +symlinks: + diff --git a/src/mesa/pipe/nv50/nv50_clear.c b/src/mesa/pipe/nv50/nv50_clear.c new file mode 100644 index 00000000000..85af1af78dd --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_clear.c @@ -0,0 +1,15 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv50_context.h" +#include "nv50_dma.h" + + +void +nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->region_fill(pipe, ps->region, 0, 0, 0, ps->width, ps->height, + clearValue); +} diff --git a/src/mesa/pipe/nv50/nv50_context.c b/src/mesa/pipe/nv50/nv50_context.c new file mode 100644 index 00000000000..6fceb13174a --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_context.c @@ -0,0 +1,215 @@ +#include "pipe/draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" +#include "nv50_dma.h" + +static boolean +nv50_is_format_supported(struct pipe_context *pipe, uint format) +{ + switch (format) { + case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_Z24_S8: + return TRUE; + default: + break; + }; + + return FALSE; +} + +static const char * +nv50_get_name(struct pipe_context *pipe) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv50->chipset); + return buffer; +} + +static const char * +nv50_get_vendor(struct pipe_context *pipe) +{ + return "nouveau"; +} + +static int +nv50_get_param(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 32; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 0; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 8; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv50_get_paramf(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv50_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv50->sync, 0); + BEGIN_RING(tesla, 0x104, 1); + OUT_RING (0); + BEGIN_RING(tesla, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv50->sync, 0, 0, 2000); +} + +static void +nv50_destroy(struct pipe_context *pipe) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + + draw_destroy(nv50->draw); + free(nv50); +} + +static boolean +nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) +{ + struct nouveau_winsys *nvws = nv50->nvws; + int ret; + + if ((ret = nvws->grobj_alloc(nvws, tesla_class, &nv50->tesla))) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + BEGIN_RING(tesla, NV50TCL_DMA_NOTIFY, 1); + OUT_RING (nv50->sync->handle); + + FIRE_RING (); + return TRUE; +} + +#define GRCLASS5097_CHIPSETS 0x00000000 +#define GRCLASS8297_CHIPSETS 0x00000010 +struct pipe_context * +nv50_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv50_context *nv50; + int tesla_class, ret; + + if ((chipset & 0xf0) != 0x50 && (chipset & 0xf0) != 0x80) { + NOUVEAU_ERR("Not a G8x chipset\n"); + return NULL; + } + + if (GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) { + tesla_class = 0x5097; + } else + if (GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) { + tesla_class = 0x8297; + } else { + NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset); + return NULL; + } + + nv50 = CALLOC_STRUCT(nv50_context); + if (!nv50) + return NULL; + nv50->chipset = chipset; + nv50->nvws = nvws; + + if ((ret = nvws->notifier_alloc(nvws, 1, &nv50->sync))) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + free(nv50); + return NULL; + } + + if (!nv50_init_hwctx(nv50, tesla_class)) { + free(nv50); + return NULL; + } + + nv50->pipe.winsys = pipe_winsys; + + nv50->pipe.destroy = nv50_destroy; + nv50->pipe.is_format_supported = nv50_is_format_supported; + nv50->pipe.get_name = nv50_get_name; + nv50->pipe.get_vendor = nv50_get_vendor; + nv50->pipe.get_param = nv50_get_param; + nv50->pipe.get_paramf = nv50_get_paramf; + + nv50->pipe.draw_arrays = nv50_draw_arrays; + nv50->pipe.draw_elements = nv50_draw_elements; + nv50->pipe.clear = nv50_clear; + + nv50->pipe.begin_query = nv50_query_begin; + nv50->pipe.end_query = nv50_query_end; + nv50->pipe.wait_query = nv50_query_wait; + + nv50->pipe.mipmap_tree_layout = nv50_miptree_layout; + + nv50->pipe.flush = nv50_flush; + + nv50_init_region_functions(nv50); + nv50_init_surface_functions(nv50); + nv50_init_state_functions(nv50); + + nv50->draw = draw_create(); + assert(nv50->draw); + draw_set_rasterize_stage(nv50->draw, nv50_draw_render_stage(nv50)); + + return &nv50->pipe; +} + + diff --git a/src/mesa/pipe/nv50/nv50_context.h b/src/mesa/pipe/nv50/nv50_context.h new file mode 100644 index 00000000000..f7fa48553d0 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_context.h @@ -0,0 +1,62 @@ +#ifndef __NV50_CONTEXT_H__ +#define __NV50_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/draw/draw_vertex.h" + +#include "pipe/nouveau/nouveau_winsys.h" +#include "pipe/nouveau/nouveau_gldefs.h" + +#include "nv50_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +struct nv50_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct draw_context *draw; + + int chipset; + struct nouveau_grobj *tesla; + struct nouveau_notifier *sync; + uint32_t *pushbuf; +}; + + +extern void nv50_init_region_functions(struct nv50_context *nv50); +extern void nv50_init_surface_functions(struct nv50_context *nv50); +extern void nv50_init_state_functions(struct nv50_context *nv50); + +/* nv50_draw.c */ +extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50); + +/* nv50_miptree.c */ +extern boolean nv50_miptree_layout(struct pipe_context *, + struct pipe_mipmap_tree *); +/* nv50_vbo.c */ +extern boolean nv50_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv50_draw_elements(struct pipe_context *pipe, + struct pipe_buffer_handle *indexBuffer, + unsigned indexSize, + unsigned mode, unsigned start, + unsigned count); + +/* nv50_clear.c */ +extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +/* nv50_query.c */ +extern void nv50_query_begin(struct pipe_context *, struct pipe_query_object *); +extern void nv50_query_end(struct pipe_context *, struct pipe_query_object *); +extern void nv50_query_wait(struct pipe_context *, struct pipe_query_object *); + + +#endif diff --git a/src/mesa/pipe/nv50/nv50_dma.h b/src/mesa/pipe/nv50/nv50_dma.h new file mode 100644 index 00000000000..f8121b61cc6 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_dma.h @@ -0,0 +1,62 @@ +#ifndef __NV50_DMA_H__ +#define __NV50_DMA_H__ + +#include "pipe/nouveau/nouveau_winsys.h" + +#define BEGIN_RING(obj,mthd,size) do { \ + nv50->pushbuf = nv50->nvws->begin_ring(nv50->obj, (mthd), (size)); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + +#define OUT_RING(data) do { \ + (*nv50->pushbuf++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + memcpy(nv50->pushbuf, (src), (size) * 4); \ + nv50->pushbuf += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define FIRE_RING() do { \ + nv50->nvws->fire_ring(nv50->nvws->channel); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + nv50->nvws->out_reloc(nv50->nvws->channel, nv50->pushbuf, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ + OUT_RING(0); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + nv50->nvws->channel->vram->handle, \ + nv50->nvws->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/mesa/pipe/nv50/nv50_draw.c b/src/mesa/pipe/nv50/nv50_draw.c new file mode 100644 index 00000000000..97088920284 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_draw.c @@ -0,0 +1,63 @@ +#include "pipe/draw/draw_private.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" + +struct nv50_draw_stage { + struct draw_stage draw; + struct nv50_context *nv50; +}; + +static void +nv50_draw_begin(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_end(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_point(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_line(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_tri(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_reset_stipple_counter(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +struct draw_stage * +nv50_draw_render_stage(struct nv50_context *nv50) +{ + struct nv50_draw_stage *nv50draw = CALLOC_STRUCT(nv50_draw_stage); + + nv50draw->nv50 = nv50; + nv50draw->draw.draw = nv50->draw; + nv50draw->draw.begin = nv50_draw_begin; + nv50draw->draw.point = nv50_draw_point; + nv50draw->draw.line = nv50_draw_line; + nv50draw->draw.tri = nv50_draw_tri; + nv50draw->draw.end = nv50_draw_end; + nv50draw->draw.reset_stipple_counter = nv50_draw_reset_stipple_counter; + + return &nv50draw->draw; +} + diff --git a/src/mesa/pipe/nv50/nv50_miptree.c b/src/mesa/pipe/nv50/nv50_miptree.c new file mode 100644 index 00000000000..cf0e530f317 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_miptree.c @@ -0,0 +1,13 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" + +boolean +nv50_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) +{ + NOUVEAU_ERR("unimplemented\n"); + return TRUE; +} + diff --git a/src/mesa/pipe/nv50/nv50_query.c b/src/mesa/pipe/nv50/nv50_query.c new file mode 100644 index 00000000000..1ce48943b99 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_query.c @@ -0,0 +1,23 @@ +#include "pipe/p_context.h" + +#include "nv50_context.h" +#include "nv50_dma.h" + +void +nv50_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +void +nv50_query_end(struct pipe_context *pipe, struct pipe_query_object *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +void +nv50_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + diff --git a/src/mesa/pipe/nv50/nv50_region.c b/src/mesa/pipe/nv50/nv50_region.c new file mode 100644 index 00000000000..d5a071c8b26 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_region.c @@ -0,0 +1,85 @@ +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" + +#include "nv50_context.h" +#include "nv50_dma.h" + +static ubyte * +nv50_region_map(struct pipe_context *pipe, struct pipe_region *region) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct pipe_winsys *ws = nv50->pipe.winsys; + + if (!region->map_refcount++) { + region->map = ws->buffer_map(ws, region->buffer, + PIPE_BUFFER_FLAG_WRITE | + PIPE_BUFFER_FLAG_READ); + } + + return region->map; +} + +static void +nv50_region_unmap(struct pipe_context *pipe, struct pipe_region *region) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct pipe_winsys *ws = nv50->pipe.winsys; + + if (!--region->map_refcount) { + ws->buffer_unmap(ws, region->buffer); + region->map = NULL; + } +} + +static void +nv50_region_data(struct pipe_context *pipe, + struct pipe_region *dst, + unsigned dst_offset, + unsigned dstx, unsigned dsty, + const void *src, unsigned src_pitch, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->region_data(nvws, dst, dst_offset, dstx, dsty, + src, src_pitch, srcx, srcy, width, height); +} + + +static void +nv50_region_copy(struct pipe_context *pipe, struct pipe_region *dst, + unsigned dst_offset, unsigned dstx, unsigned dsty, + struct pipe_region *src, unsigned src_offset, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->region_copy(nvws, dst, dst_offset, dstx, dsty, + src, src_offset, srcx, srcy, width, height); +} + +static void +nv50_region_fill(struct pipe_context *pipe, + struct pipe_region *dst, unsigned dst_offset, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, unsigned value) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->region_fill(nvws, dst, dst_offset, dstx, dsty, + width, height, value); +} + +void +nv50_init_region_functions(struct nv50_context *nv50) +{ + nv50->pipe.region_map = nv50_region_map; + nv50->pipe.region_unmap = nv50_region_unmap; + nv50->pipe.region_data = nv50_region_data; + nv50->pipe.region_copy = nv50_region_copy; + nv50->pipe.region_fill = nv50_region_fill; +} + diff --git a/src/mesa/pipe/nv50/nv50_state.c b/src/mesa/pipe/nv50/nv50_state.c new file mode 100644 index 00000000000..0674c30dc80 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_state.c @@ -0,0 +1,242 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" +#include "nv50_dma.h" +#include "nv50_state.h" + +static void * +nv50_alpha_test_state_create(struct pipe_context *pipe, + const struct pipe_alpha_test_state *cso) +{ +} + +static void +nv50_alpha_test_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_alpha_test_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ +} + +static void +nv50_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ +} + +static void +nv50_sampler_state_bind(struct pipe_context *pipe, unsigned unit, + void *hwcso) +{ +} + +static void +nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ +} + +static void +nv50_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_depth_stencil_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_state *cso) +{ +} + +static void +nv50_depth_stencil_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_depth_stencil_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ +} + +static void +nv50_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ +} + +static void +nv50_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ +} + +static void +nv50_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv50_set_clear_color_state(struct pipe_context *pipe, + const struct pipe_clear_color_state *ccol) +{ +} + +static void +nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ +} + +static void +nv50_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ +} + +static void +nv50_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ +} + +static void +nv50_set_sampler_units(struct pipe_context *pipe, + uint num_samplers, const uint *units) +{ +} + +static void +nv50_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ +} + +static void +nv50_set_texture_state(struct pipe_context *pipe, unsigned unit, + struct pipe_mipmap_tree *miptree) +{ +} + +static void +nv50_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ +} + +static void +nv50_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ +} + +static void +nv50_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ +} + +void +nv50_init_state_functions(struct nv50_context *nv50) +{ + nv50->pipe.create_alpha_test_state = nv50_alpha_test_state_create; + nv50->pipe.bind_alpha_test_state = nv50_alpha_test_state_bind; + nv50->pipe.delete_alpha_test_state = nv50_alpha_test_state_delete; + + nv50->pipe.create_blend_state = nv50_blend_state_create; + nv50->pipe.bind_blend_state = nv50_blend_state_bind; + nv50->pipe.delete_blend_state = nv50_blend_state_delete; + + nv50->pipe.create_sampler_state = nv50_sampler_state_create; + nv50->pipe.bind_sampler_state = nv50_sampler_state_bind; + nv50->pipe.delete_sampler_state = nv50_sampler_state_delete; + + nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create; + nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; + nv50->pipe.delete_rasterizer_state = nv50_rasterizer_state_delete; + + nv50->pipe.create_depth_stencil_state = nv50_depth_stencil_state_create; + nv50->pipe.bind_depth_stencil_state = nv50_depth_stencil_state_bind; + nv50->pipe.delete_depth_stencil_state = nv50_depth_stencil_state_delete; + + nv50->pipe.create_vs_state = nv50_vp_state_create; + nv50->pipe.bind_vs_state = nv50_vp_state_bind; + nv50->pipe.delete_vs_state = nv50_vp_state_delete; + + nv50->pipe.create_fs_state = nv50_fp_state_create; + nv50->pipe.bind_fs_state = nv50_fp_state_bind; + nv50->pipe.delete_fs_state = nv50_fp_state_delete; + + nv50->pipe.set_blend_color = nv50_set_blend_color; + nv50->pipe.set_clip_state = nv50_set_clip_state; + nv50->pipe.set_clear_color_state = nv50_set_clear_color_state; + nv50->pipe.set_constant_buffer = nv50_set_constant_buffer; + nv50->pipe.set_framebuffer_state = nv50_set_framebuffer_state; + nv50->pipe.set_polygon_stipple = nv50_set_polygon_stipple; + nv50->pipe.set_sampler_units = nv50_set_sampler_units; + nv50->pipe.set_scissor_state = nv50_set_scissor_state; + nv50->pipe.set_texture_state = nv50_set_texture_state; + nv50->pipe.set_viewport_state = nv50_set_viewport_state; + + nv50->pipe.set_vertex_buffer = nv50_set_vertex_buffer; + nv50->pipe.set_vertex_element = nv50_set_vertex_element; + +// nv50->pipe.set_feedback_state = nv50_set_feedback_state; +// nv50->pipe.set_feedback_buffer = nv50_set_feedback_buffer; +} + diff --git a/src/mesa/pipe/nv50/nv50_state.h b/src/mesa/pipe/nv50/nv50_state.h new file mode 100644 index 00000000000..a3b781d4c61 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_state.h @@ -0,0 +1,7 @@ +#ifndef __NV50_STATE_H__ +#define __NV50_STATE_H__ + +#include "pipe/p_state.h" + + +#endif diff --git a/src/mesa/pipe/nv50/nv50_surface.c b/src/mesa/pipe/nv50/nv50_surface.c new file mode 100644 index 00000000000..68013c0c1da --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_surface.c @@ -0,0 +1,229 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv50_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" + + +#define CLIP_TILE \ + do { \ + if (x >= ps->width) \ + return; \ + if (y >= ps->height) \ + return; \ + if (x + w > ps->width) \ + w = ps->width - x; \ + if (y + h > ps->height) \ + h = ps->height -y; \ + } while(0) + + +/** + * Note: this is exactly like a8r8g8b8_get_tile() in sp_surface.c + * Share it someday. + */ +static void +nv50_get_tile_rgba(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, float *p) +{ + const unsigned *src + = ((const unsigned *) (ps->region->map + ps->offset)) + + y * ps->region->pitch + x; + unsigned i, j; + unsigned w0 = w; + + CLIP_TILE; + + switch (ps->format) { + case PIPE_FORMAT_U_A8_R8_G8_B8: + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++) { + const unsigned pixel = src[j]; + pRow[0] = UBYTE_TO_FLOAT((pixel >> 16) & 0xff); + pRow[1] = UBYTE_TO_FLOAT((pixel >> 8) & 0xff); + pRow[2] = UBYTE_TO_FLOAT((pixel >> 0) & 0xff); + pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); + pRow += 4; + } + src += ps->region->pitch; + p += w0 * 4; + } + break; + case PIPE_FORMAT_Z24_S8: + { + const float scale = 1.0 / (float) 0xffffff; + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++) { + const unsigned pixel = src[j]; + pRow[0] = + pRow[1] = + pRow[2] = + pRow[3] = ((pixel & 0xffffff) >> 8) * scale; + pRow += 4; + } + src += ps->region->pitch; + p += w0 * 4; + } + } + break; + default: + assert(0); + } +} + + +static void +nv50_put_tile_rgba(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, const float *p) +{ + /* TODO */ + assert(0); +} + + +/* + * XXX note: same as code in sp_surface.c + */ +static void +nv50_get_tile(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, + void *p, int dst_stride) +{ + const uint cpp = ps->region->cpp; + const uint w0 = w; + const ubyte *pSrc; + ubyte *pDest; + uint i; + + assert(ps->region->map); + + CLIP_TILE; + + if (dst_stride == 0) { + dst_stride = w0 * cpp; + } + + pSrc = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pDest = (ubyte *) p; + + for (i = 0; i < h; i++) { + memcpy(pDest, pSrc, w0 * cpp); + pDest += dst_stride; + pSrc += ps->region->pitch * cpp; + } +} + + +/* + * XXX note: same as code in sp_surface.c + */ +static void +nv50_put_tile(struct pipe_context *pipe, + struct pipe_surface *ps, + uint x, uint y, uint w, uint h, + const void *p, int src_stride) +{ + const uint cpp = ps->region->cpp; + const uint w0 = w; + const ubyte *pSrc; + ubyte *pDest; + uint i; + + assert(ps->region->map); + + CLIP_TILE; + + if (src_stride == 0) { + src_stride = w0 * cpp; + } + + pSrc = (const ubyte *) p; + pDest = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + + for (i = 0; i < h; i++) { + memcpy(pDest, pSrc, w0 * cpp); + pDest += ps->region->pitch * cpp; + pSrc += src_stride; + } +} + + +/* + * XXX note: same as code in sp_surface.c + */ +static struct pipe_surface * +nv50_get_tex_surface(struct pipe_context *pipe, + struct pipe_mipmap_tree *mt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_surface *ps; + unsigned offset; /* in bytes */ + + offset = mt->level[level].level_offset; + + if (mt->target == PIPE_TEXTURE_CUBE) { + offset += mt->level[level].image_offset[face] * mt->cpp; + } + else if (mt->target == PIPE_TEXTURE_3D) { + offset += mt->level[level].image_offset[zslice] * mt->cpp; + } + else { + assert(face == 0); + assert(zslice == 0); + } + + ps = pipe->winsys->surface_alloc(pipe->winsys, mt->format); + if (ps) { + assert(ps->format); + assert(ps->refcount); + pipe_region_reference(&ps->region, mt->region); + ps->width = mt->level[level].width; + ps->height = mt->level[level].height; + ps->offset = offset; + } + return ps; +} + + +void +nv50_init_surface_functions(struct nv50_context *nv50) +{ + nv50->pipe.get_tex_surface = nv50_get_tex_surface; + nv50->pipe.get_tile = nv50_get_tile; + nv50->pipe.put_tile = nv50_put_tile; + nv50->pipe.get_tile_rgba = nv50_get_tile_rgba; + nv50->pipe.put_tile_rgba = nv50_put_tile_rgba; +} diff --git a/src/mesa/pipe/nv50/nv50_vbo.c b/src/mesa/pipe/nv50/nv50_vbo.c new file mode 100644 index 00000000000..faae08f6014 --- /dev/null +++ b/src/mesa/pipe/nv50/nv50_vbo.c @@ -0,0 +1,25 @@ +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" +#include "nv50_dma.h" +#include "nv50_state.h" + +boolean +nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, + unsigned count) +{ + NOUVEAU_ERR("unimplemented\n"); + return TRUE; +} + +boolean +nv50_draw_elements(struct pipe_context *pipe, + struct pipe_buffer_handle *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ + NOUVEAU_ERR("unimplemented\n"); + return TRUE; +} + -- cgit v1.2.3 From fecb3ce5860caa498f531c28db043c3c2cb845a8 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 7 Dec 2007 02:31:40 +0100 Subject: Use write posting in the kickoff function too. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 6e123c44734..d3239959034 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -137,6 +137,7 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) struct nouveau_channel_priv *chan = nouveau_channel(userchan); uint32_t put_offset; int i; + volatile int dum; if (chan->dma.cur == chan->dma.put) return; @@ -188,6 +189,7 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) #endif chan->dma.put = chan->dma.cur; NOUVEAU_DMA_BARRIER; + dum = READ_GET(chan); *chan->put = put_offset; NOUVEAU_DMA_BARRIER; } -- cgit v1.2.3 From 72513ae3a7b4f2a0b90c46e4ef0b9c10e4c74ed2 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Wed, 28 Nov 2007 19:04:54 +0100 Subject: Move dimensions from struct pipe_region to struct pipe_surface. --- .../drivers/dri/intel_winsys/intel_swapbuffers.c | 6 +- .../drivers/dri/intel_winsys/intel_winsys_pipe.c | 30 +-- src/mesa/pipe/failover/fo_context.c | 6 +- src/mesa/pipe/i915simple/i915_clear.c | 9 +- src/mesa/pipe/i915simple/i915_regions.c | 172 --------------- src/mesa/pipe/i915simple/i915_state_emit.c | 25 +-- src/mesa/pipe/i915simple/i915_surface.c | 185 +++++++++++++++- src/mesa/pipe/p_context.h | 48 ++-- src/mesa/pipe/p_state.h | 5 +- src/mesa/pipe/p_winsys.h | 6 +- src/mesa/pipe/softpipe/sp_clear.c | 16 +- src/mesa/pipe/softpipe/sp_region.c | 186 ---------------- src/mesa/pipe/softpipe/sp_surface.c | 244 ++++++++++++++++++--- src/mesa/pipe/xlib/xm_buffer.c | 5 +- src/mesa/pipe/xlib/xm_surface.c | 2 +- src/mesa/pipe/xlib/xm_winsys.c | 22 +- src/mesa/state_tracker/st_cb_drawpixels.c | 50 ++--- src/mesa/state_tracker/st_cb_fbo.c | 8 +- src/mesa/state_tracker/st_cb_readpixels.c | 6 +- src/mesa/state_tracker/st_cb_texture.c | 35 +-- src/mesa/state_tracker/st_mipmap_tree.c | 54 +++-- 21 files changed, 550 insertions(+), 570 deletions(-) diff --git a/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c b/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c index 9fcb2dac273..f96209d1b06 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c @@ -89,15 +89,15 @@ intelDisplaySurface(__DRIdrawablePrivate *dPriv, const int pitch = intelScreen->front.pitch / intelScreen->front.cpp; const int cpp = intelScreen->front.cpp; const struct pipe_region *srcRegion = surf->region; - const int srcpitch = srcRegion->pitch; + const int srcpitch = surf->pitch; int BR13, CMD; int i; ASSERT(srcRegion); - ASSERT(srcRegion->cpp == cpp); + ASSERT(surf->cpp == cpp); DBG(SWAP, "screen pitch %d src surface pitch %d\n", - pitch, srcRegion->pitch); + pitch, surf->pitch); if (cpp == 2) { BR13 = (pitch * cpp) | (0xCC << 16) | (1 << 24); diff --git a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c index ae02f98a783..7b3aa994824 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c @@ -173,14 +173,10 @@ intel_flush_frontbuffer( struct pipe_winsys *winsys, } -static struct pipe_region * -intel_i915_region_alloc(struct pipe_winsys *winsys, - unsigned cpp, unsigned width, - unsigned height, unsigned flags) +static unsigned +intel_i915_surface_pitch(struct pipe_winsys *winsys, + unsigned cpp, unsigned width, unsigned flags) { - struct pipe_region *region = calloc(sizeof(*region), 1); - const unsigned alignment = 64; - /* Choose a pitch to match hardware requirements - requires 64 byte * alignment of render targets. * @@ -188,24 +184,29 @@ intel_i915_region_alloc(struct pipe_winsys *winsys, * clearly want to be able to render to textures under some * circumstances, but maybe not always a requirement. */ - unsigned pitch; /* XXX is the pitch different for textures vs. drawables? */ if (flags & PIPE_SURFACE_FLAG_TEXTURE) /* or PIPE_SURFACE_FLAG_RENDER? */ - pitch = ((cpp * width + 63) & ~63) / cpp; + return ((cpp * width + 63) & ~63) / cpp; else - pitch = ((cpp * width + 63) & ~63) / cpp; + return ((cpp * width + 63) & ~63) / cpp; +} + + +static struct pipe_region * +intel_i915_region_alloc(struct pipe_winsys *winsys, + unsigned size, unsigned flags) +{ + struct pipe_region *region = calloc(sizeof(*region), 1); + const unsigned alignment = 64; - region->cpp = cpp; - region->pitch = pitch; - region->height = height; /* needed? */ region->refcount = 1; region->buffer = winsys->buffer_create( winsys, alignment ); winsys->buffer_data( winsys, region->buffer, - pitch * cpp * height, + size, NULL, PIPE_BUFFER_USAGE_PIXEL ); @@ -301,6 +302,7 @@ intel_create_pipe_winsys( int fd ) iws->winsys.get_name = intel_get_name; iws->winsys.region_alloc = intel_i915_region_alloc; iws->winsys.region_release = intel_i915_region_release; + iws->winsys.surface_pitch = intel_i915_surface_pitch; iws->winsys.surface_alloc = intel_i915_surface_alloc; iws->winsys.surface_release = intel_i915_surface_release; diff --git a/src/mesa/pipe/failover/fo_context.c b/src/mesa/pipe/failover/fo_context.c index c5fab73fb13..0cc9cab408c 100644 --- a/src/mesa/pipe/failover/fo_context.c +++ b/src/mesa/pipe/failover/fo_context.c @@ -142,9 +142,9 @@ struct pipe_context *failover_create( struct pipe_context *hw, failover->pipe.region_map = hw->region_map; failover->pipe.region_unmap = hw->region_unmap; - failover->pipe.region_data = hw->region_data; - failover->pipe.region_copy = hw->region_copy; - failover->pipe.region_fill = hw->region_fill; + failover->pipe.surface_data = hw->surface_data; + failover->pipe.surface_copy = hw->surface_copy; + failover->pipe.surface_fill = hw->surface_fill; failover->pipe.mipmap_tree_layout = hw->mipmap_tree_layout; failover->pipe.flush = hw->flush; diff --git a/src/mesa/pipe/i915simple/i915_clear.c b/src/mesa/pipe/i915simple/i915_clear.c index e8087df5a47..cde69daacc0 100644 --- a/src/mesa/pipe/i915simple/i915_clear.c +++ b/src/mesa/pipe/i915simple/i915_clear.c @@ -43,12 +43,5 @@ void i915_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { - int x, y, w, h; - - x = 0; - y = 0; - w = ps->width; - h = ps->height; - - pipe->region_fill(pipe, ps->region, 0, x, y, w, h, clearValue); + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); } diff --git a/src/mesa/pipe/i915simple/i915_regions.c b/src/mesa/pipe/i915simple/i915_regions.c index e8c4c92bc88..82fdec83d09 100644 --- a/src/mesa/pipe/i915simple/i915_regions.c +++ b/src/mesa/pipe/i915simple/i915_regions.c @@ -34,7 +34,6 @@ #include "pipe/p_defines.h" #include "pipe/p_winsys.h" #include "i915_context.h" -#include "i915_blit.h" @@ -70,181 +69,10 @@ i915_region_unmap(struct pipe_context *pipe, struct pipe_region *region) } -/* - * XXX Move this into core Mesa? - */ -static void -_mesa_copy_rect(ubyte * dst, - unsigned cpp, - unsigned dst_pitch, - unsigned dst_x, - unsigned dst_y, - unsigned width, - unsigned height, - const ubyte * src, - unsigned src_pitch, - unsigned src_x, - unsigned src_y) -{ - unsigned i; - - dst_pitch *= cpp; - src_pitch *= cpp; - dst += dst_x * cpp; - src += src_x * cpp; - dst += dst_y * dst_pitch; - src += src_y * dst_pitch; - width *= cpp; - - if (width == dst_pitch && width == src_pitch) - memcpy(dst, src, height * width); - else { - for (i = 0; i < height; i++) { - memcpy(dst, src, width); - dst += dst_pitch; - src += src_pitch; - } - } -} - - -/* Upload data to a rectangular sub-region. Lots of choices how to do this: - * - * - memcpy by span to current destination - * - upload data as new buffer and blit - * - * Currently always memcpy. - */ -static void -i915_region_data(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - const void *src, unsigned src_pitch, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - _mesa_copy_rect(pipe->region_map(pipe, dst) + dst_offset, - dst->cpp, - dst->pitch, - dstx, dsty, width, height, src, src_pitch, srcx, srcy); - - pipe->region_unmap(pipe, dst); -} - - -/* Assumes all values are within bounds -- no checking at this level - - * do it higher up if required. - */ -static void -i915_region_copy(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - struct pipe_region *src, - unsigned src_offset, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - assert( dst != src ); - assert( dst->cpp == src->cpp ); - - if (0) { - _mesa_copy_rect(pipe->region_map(pipe, dst) + dst_offset, - dst->cpp, - dst->pitch, - dstx, dsty, - width, height, - pipe->region_map(pipe, src) + src_offset, - src->pitch, - srcx, srcy); - - pipe->region_unmap(pipe, src); - pipe->region_unmap(pipe, dst); - } - else { - i915_copy_blit( i915_context(pipe), - dst->cpp, - (short) src->pitch, src->buffer, src_offset, - (short) dst->pitch, dst->buffer, dst_offset, - (short) srcx, (short) srcy, (short) dstx, (short) dsty, (short) width, (short) height ); - } -} - -/* Fill a rectangular sub-region. Need better logic about when to - * push buffers into AGP - will currently do so whenever possible. - */ -static ubyte * -get_pointer(struct pipe_region *dst, unsigned x, unsigned y) -{ - return dst->map + (y * dst->pitch + x) * dst->cpp; -} - - -static void -i915_region_fill(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - unsigned width, unsigned height, unsigned value) -{ - if (0) { - unsigned i, j; - - (void)pipe->region_map(pipe, dst); - - switch (dst->cpp) { - case 1: { - ubyte *row = get_pointer(dst, dstx, dsty); - for (i = 0; i < height; i++) { - memset(row, value, width); - row += dst->pitch; - } - } - break; - case 2: { - ushort *row = (ushort *) get_pointer(dst, dstx, dsty); - for (i = 0; i < height; i++) { - for (j = 0; j < width; j++) - row[j] = (ushort) value; - row += dst->pitch; - } - } - break; - case 4: { - unsigned *row = (unsigned *) get_pointer(dst, dstx, dsty); - for (i = 0; i < height; i++) { - for (j = 0; j < width; j++) - row[j] = value; - row += dst->pitch; - } - } - break; - default: - assert(0); - break; - } - } - else { - i915_fill_blit( i915_context(pipe), - dst->cpp, - (short) dst->pitch, - dst->buffer, dst_offset, - (short) dstx, (short) dsty, - (short) width, (short) height, - value ); - } -} - - - - - void i915_init_region_functions(struct i915_context *i915) { i915->pipe.region_map = i915_region_map; i915->pipe.region_unmap = i915_region_unmap; - i915->pipe.region_data = i915_region_data; - i915->pipe.region_copy = i915_region_copy; - i915->pipe.region_fill = i915_region_fill; } diff --git a/src/mesa/pipe/i915simple/i915_state_emit.c b/src/mesa/pipe/i915simple/i915_state_emit.c index d793e92a14d..900a91d8963 100644 --- a/src/mesa/pipe/i915simple/i915_state_emit.c +++ b/src/mesa/pipe/i915simple/i915_state_emit.c @@ -208,10 +208,11 @@ i915_emit_hardware_state(struct i915_context *i915 ) /* 8 dwords, 2 relocs */ if (i915->hardware_dirty & I915_HW_STATIC) { - if (i915->framebuffer.cbufs[0]) { - struct pipe_region *cbuf_region = i915->framebuffer.cbufs[0]->region; - unsigned pitch = (cbuf_region->pitch * - cbuf_region->cpp); + struct pipe_surface *cbuf_surface = i915->framebuffer.cbufs[0]; + struct pipe_surface *depth_surface = i915->framebuffer.zbuf; + + if (cbuf_surface) { + unsigned pitch = (cbuf_surface->pitch * cbuf_surface->cpp); OUT_BATCH(_3DSTATE_BUF_INFO_CMD); @@ -219,17 +220,15 @@ i915_emit_hardware_state(struct i915_context *i915 ) BUF_3D_PITCH(pitch) | /* pitch in bytes */ BUF_3D_USE_FENCE); - OUT_RELOC(cbuf_region->buffer, + OUT_RELOC(cbuf_surface->region->buffer, I915_BUFFER_ACCESS_WRITE, 0); } /* What happens if no zbuf?? */ - if (i915->framebuffer.zbuf) { - struct pipe_region *depth_region = i915->framebuffer.zbuf->region; - unsigned zpitch = (depth_region->pitch * - depth_region->cpp); + if (depth_surface) { + unsigned zpitch = (depth_surface->pitch * depth_surface->cpp); OUT_BATCH(_3DSTATE_BUF_INFO_CMD); @@ -237,7 +236,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) BUF_3D_PITCH(zpitch) | /* pitch in bytes */ BUF_3D_USE_FENCE); - OUT_RELOC(depth_region->buffer, + OUT_RELOC(depth_surface->region->buffer, I915_BUFFER_ACCESS_WRITE, 0); } @@ -245,13 +244,13 @@ i915_emit_hardware_state(struct i915_context *i915 ) { unsigned cformat, zformat = 0; - if (i915->framebuffer.cbufs[0]) - cformat = i915->framebuffer.cbufs[0]->format; + if (cbuf_surface) + cformat = cbuf_surface->format; else cformat = PIPE_FORMAT_U_A8_R8_G8_B8; /* arbitrary */ cformat = translate_format(cformat); - if (i915->framebuffer.zbuf) + if (depth_surface) zformat = translate_depth_format( i915->framebuffer.zbuf->format ); OUT_BATCH(_3DSTATE_DST_BUF_VARS_CMD); diff --git a/src/mesa/pipe/i915simple/i915_surface.c b/src/mesa/pipe/i915simple/i915_surface.c index bec7ddb7319..e4a5de00d77 100644 --- a/src/mesa/pipe/i915simple/i915_surface.c +++ b/src/mesa/pipe/i915simple/i915_surface.c @@ -26,6 +26,7 @@ **************************************************************************/ #include "i915_context.h" +#include "i915_blit.h" #include "i915_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" @@ -57,7 +58,7 @@ i915_get_tile_rgba(struct pipe_context *pipe, { const unsigned *src = ((const unsigned *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -75,7 +76,7 @@ i915_get_tile_rgba(struct pipe_context *pipe, pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } break; @@ -92,7 +93,7 @@ i915_get_tile_rgba(struct pipe_context *pipe, pRow[3] = (pixel & 0xffffff) * scale; pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -122,7 +123,7 @@ i915_get_tile(struct pipe_context *pipe, uint x, uint y, uint w, uint h, void *p, int dst_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const uint w0 = w; const ubyte *pSrc; ubyte *pDest; @@ -136,13 +137,13 @@ i915_get_tile(struct pipe_context *pipe, dst_stride = w0 * cpp; } - pSrc = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pSrc = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; pDest = (ubyte *) p; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); pDest += dst_stride; - pSrc += ps->region->pitch * cpp; + pSrc += ps->pitch * cpp; } } @@ -156,7 +157,7 @@ i915_put_tile(struct pipe_context *pipe, uint x, uint y, uint w, uint h, const void *p, int src_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const uint w0 = w; const ubyte *pSrc; ubyte *pDest; @@ -171,11 +172,11 @@ i915_put_tile(struct pipe_context *pipe, } pSrc = (const ubyte *) p; - pDest = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pDest = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); - pDest += ps->region->pitch * cpp; + pDest += ps->pitch * cpp; pSrc += src_stride; } } @@ -210,14 +211,177 @@ i915_get_tex_surface(struct pipe_context *pipe, assert(ps->format); assert(ps->refcount); pipe_region_reference(&ps->region, mt->region); + ps->cpp = mt->cpp; ps->width = mt->level[level].width; ps->height = mt->level[level].height; + ps->pitch = mt->pitch; ps->offset = offset; } return ps; } +/* + * XXX Move this into core Mesa? + */ +static void +_mesa_copy_rect(ubyte * dst, + unsigned cpp, + unsigned dst_pitch, + unsigned dst_x, + unsigned dst_y, + unsigned width, + unsigned height, + const ubyte * src, + unsigned src_pitch, + unsigned src_x, + unsigned src_y) +{ + unsigned i; + + dst_pitch *= cpp; + src_pitch *= cpp; + dst += dst_x * cpp; + src += src_x * cpp; + dst += dst_y * dst_pitch; + src += src_y * dst_pitch; + width *= cpp; + + if (width == dst_pitch && width == src_pitch) + memcpy(dst, src, height * width); + else { + for (i = 0; i < height; i++) { + memcpy(dst, src, width); + dst += dst_pitch; + src += src_pitch; + } + } +} + + +/* Upload data to a rectangular sub-region. Lots of choices how to do this: + * + * - memcpy by span to current destination + * - upload data as new buffer and blit + * + * Currently always memcpy. + */ +static void +i915_surface_data(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + const void *src, unsigned src_pitch, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + _mesa_copy_rect(pipe->region_map(pipe, dst->region) + dst->offset, + dst->cpp, + dst->pitch, + dstx, dsty, width, height, src, src_pitch, srcx, srcy); + + pipe->region_unmap(pipe, dst->region); +} + + +/* Assumes all values are within bounds -- no checking at this level - + * do it higher up if required. + */ +static void +i915_surface_copy(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + struct pipe_surface *src, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + assert( dst != src ); + assert( dst->cpp == src->cpp ); + + if (0) { + _mesa_copy_rect(pipe->region_map(pipe, dst->region) + dst->offset, + dst->cpp, + dst->pitch, + dstx, dsty, + width, height, + pipe->region_map(pipe, src->region) + src->offset, + src->pitch, + srcx, srcy); + + pipe->region_unmap(pipe, src->region); + pipe->region_unmap(pipe, dst->region); + } + else { + i915_copy_blit( i915_context(pipe), + dst->cpp, + (short) src->pitch, src->region->buffer, src->offset, + (short) dst->pitch, dst->region->buffer, dst->offset, + (short) srcx, (short) srcy, (short) dstx, (short) dsty, (short) width, (short) height ); + } +} + +/* Fill a rectangular sub-region. Need better logic about when to + * push buffers into AGP - will currently do so whenever possible. + */ +static ubyte * +get_pointer(struct pipe_surface *dst, unsigned x, unsigned y) +{ + return dst->region->map + (y * dst->pitch + x) * dst->cpp; +} + + +static void +i915_surface_fill(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, unsigned value) +{ + if (0) { + unsigned i, j; + + (void)pipe->region_map(pipe, dst->region); + + switch (dst->cpp) { + case 1: { + ubyte *row = get_pointer(dst, dstx, dsty); + for (i = 0; i < height; i++) { + memset(row, value, width); + row += dst->pitch; + } + } + break; + case 2: { + ushort *row = (ushort *) get_pointer(dst, dstx, dsty); + for (i = 0; i < height; i++) { + for (j = 0; j < width; j++) + row[j] = (ushort) value; + row += dst->pitch; + } + } + break; + case 4: { + unsigned *row = (unsigned *) get_pointer(dst, dstx, dsty); + for (i = 0; i < height; i++) { + for (j = 0; j < width; j++) + row[j] = value; + row += dst->pitch; + } + } + break; + default: + assert(0); + break; + } + } + else { + i915_fill_blit( i915_context(pipe), + dst->cpp, + (short) dst->pitch, + dst->region->buffer, dst->offset, + (short) dstx, (short) dsty, + (short) width, (short) height, + value ); + } +} + + void i915_init_surface_functions(struct i915_context *i915) { @@ -226,4 +390,7 @@ i915_init_surface_functions(struct i915_context *i915) i915->pipe.put_tile = i915_put_tile; i915->pipe.get_tile_rgba = i915_get_tile_rgba; i915->pipe.put_tile_rgba = i915_put_tile_rgba; + i915->pipe.surface_data = i915_surface_data; + i915->pipe.surface_copy = i915_surface_copy; + i915->pipe.surface_fill = i915_surface_fill; } diff --git a/src/mesa/pipe/p_context.h b/src/mesa/pipe/p_context.h index 8bed958feb3..e145b22f2ff 100644 --- a/src/mesa/pipe/p_context.h +++ b/src/mesa/pipe/p_context.h @@ -208,30 +208,30 @@ struct pipe_context { void (*region_unmap)(struct pipe_context *pipe, struct pipe_region *r); - void (*region_data)(struct pipe_context *pipe, - struct pipe_region *dest, - unsigned dest_offset, - unsigned destx, unsigned desty, - const void *src, unsigned src_stride, - unsigned srcx, unsigned srcy, - unsigned width, unsigned height); - - void (*region_copy)(struct pipe_context *pipe, - struct pipe_region *dest, - unsigned dest_offset, - unsigned destx, unsigned desty, - struct pipe_region *src, /* don't make this const - - need to map/unmap */ - unsigned src_offset, - unsigned srcx, unsigned srcy, - unsigned width, unsigned height); - - void (*region_fill)(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - unsigned width, unsigned height, - unsigned value); + + /* + * Surface functions + */ + void (*surface_data)(struct pipe_context *pipe, + struct pipe_surface *dest, + unsigned destx, unsigned desty, + const void *src, unsigned src_stride, + unsigned srcx, unsigned srcy, + unsigned width, unsigned height); + + void (*surface_copy)(struct pipe_context *pipe, + struct pipe_surface *dest, + unsigned destx, unsigned desty, + struct pipe_surface *src, /* don't make this const - + need to map/unmap */ + unsigned srcx, unsigned srcy, + unsigned width, unsigned height); + + void (*surface_fill)(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, + unsigned value); /* diff --git a/src/mesa/pipe/p_state.h b/src/mesa/pipe/p_state.h index 570f44e24ed..642734aeb80 100644 --- a/src/mesa/pipe/p_state.h +++ b/src/mesa/pipe/p_state.h @@ -265,9 +265,6 @@ struct pipe_region struct pipe_buffer_handle *buffer; /**< driver private buffer handle */ unsigned refcount; /**< Reference count for region */ - unsigned cpp; /**< bytes per pixel */ - unsigned pitch; /**< in pixels */ - unsigned height; /**< in pixels */ ubyte *map; /**< only non-NULL when region is actually mapped */ unsigned map_refcount; /**< Reference count for mapping */ }; @@ -281,7 +278,9 @@ struct pipe_surface { struct pipe_region *region; unsigned format; /**< PIPE_FORMAT_x */ + unsigned cpp; /**< bytes per pixel */ unsigned width, height; + unsigned pitch; /**< in pixels */ unsigned offset; /**< offset from start of region, in bytes */ unsigned refcount; struct pipe_winsys *winsys; /**< winsys which owns/created the surface */ diff --git a/src/mesa/pipe/p_winsys.h b/src/mesa/pipe/p_winsys.h index 41d522e11ea..5adca1d6fd6 100644 --- a/src/mesa/pipe/p_winsys.h +++ b/src/mesa/pipe/p_winsys.h @@ -80,12 +80,14 @@ struct pipe_winsys * flags is bitmask of PIPE_SURFACE_FLAG_RENDER, PIPE_SURFACE_FLAG_TEXTURE */ struct pipe_region *(*region_alloc)(struct pipe_winsys *ws, - unsigned cpp, unsigned width, - unsigned height, unsigned flags); + unsigned size, unsigned flags); void (*region_release)(struct pipe_winsys *ws, struct pipe_region **r); + unsigned (*surface_pitch)(struct pipe_winsys *ws, unsigned cpp, + unsigned with, unsigned flags); + /** allocate a new surface (no context dependency) */ struct pipe_surface *(*surface_alloc)(struct pipe_winsys *ws, unsigned format); diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index 87f850b6fd9..a4276362b99 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -47,23 +47,9 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { struct softpipe_context *softpipe = softpipe_context(pipe); - unsigned x, y, w, h; softpipe_update_derived(softpipe); /* not needed?? */ - /* Use the X coord to trick region_fill() into filling at an offset - * from the start of the region. Perhaps pipe_region should have the - * 'offset' field, not pipe_surface??? - */ - assert(ps->offset % ps->region->cpp == 0); - x = ps->offset / ps->region->cpp; - y = 0; - w = ps->width; - h = ps->height; - - assert(w <= ps->region->pitch); - assert(h <= ps->region->height); - if (ps == sp_tile_cache_get_surface(softpipe->zbuf_cache)) { float clear[4]; clear[0] = 1.0; /* XXX hack */ @@ -78,7 +64,7 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, sp_tile_cache_clear(softpipe->cbuf_cache[0], clear); } - pipe->region_fill(pipe, ps->region, 0, x, y, w, h, clearValue); + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); #if 0 diff --git a/src/mesa/pipe/softpipe/sp_region.c b/src/mesa/pipe/softpipe/sp_region.c index fef63ef2f6d..58dc6bb96e4 100644 --- a/src/mesa/pipe/softpipe/sp_region.c +++ b/src/mesa/pipe/softpipe/sp_region.c @@ -70,196 +70,10 @@ sp_region_unmap(struct pipe_context *pipe, struct pipe_region *region) } - -/** - * Copy 2D rect from one place to another. - * Position and sizes are in pixels. - */ -static void -copy_rect(ubyte * dst, - unsigned cpp, - unsigned dst_pitch, - unsigned dst_x, - unsigned dst_y, - unsigned width, - unsigned height, - const ubyte * src, - unsigned src_pitch, - unsigned src_x, - unsigned src_y) -{ - unsigned i; - - dst_pitch *= cpp; - src_pitch *= cpp; - dst += dst_x * cpp; - src += src_x * cpp; - dst += dst_y * dst_pitch; - src += src_y * src_pitch; - width *= cpp; - - if (width == dst_pitch && width == src_pitch) - memcpy(dst, src, height * width); - else { - for (i = 0; i < height; i++) { - memcpy(dst, src, width); - dst += dst_pitch; - src += src_pitch; - } - } -} - - -/* Upload data to a rectangular sub-region. Lots of choices how to do this: - * - * - memcpy by span to current destination - * - upload data as new buffer and blit - * - * Currently always memcpy. - */ -static void -sp_region_data(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - const void *src, unsigned src_pitch, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - copy_rect(pipe->region_map(pipe, dst) + dst_offset, - dst->cpp, - dst->pitch, - dstx, dsty, width, height, src, src_pitch, srcx, srcy); - - pipe->region_unmap(pipe, dst); -} - -/* Assumes all values are within bounds -- no checking at this level - - * do it higher up if required. - */ -static void -sp_region_copy(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - struct pipe_region *src, - unsigned src_offset, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - ubyte *src_map, *dst_map; - assert( dst->cpp == src->cpp ); - - dst_map = pipe->region_map(pipe, dst); - src_map = pipe->region_map(pipe, src); - copy_rect(dst_map + dst_offset, - dst->cpp, - dst->pitch, - dstx, dsty, - width, height, - src_map + src_offset, - src->pitch, - srcx, srcy); - - pipe->region_unmap(pipe, src); - pipe->region_unmap(pipe, dst); -} - - -static ubyte * -get_pointer(struct pipe_region *dst, unsigned x, unsigned y) -{ - return dst->map + (y * dst->pitch + x) * dst->cpp; -} - - -#define UBYTE_TO_USHORT(B) ((B) | ((B) << 8)) - - -/** - * Fill a rectangular sub-region. Need better logic about when to - * push buffers into AGP - will currently do so whenever possible. - */ -static void -sp_region_fill(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - unsigned width, unsigned height, unsigned value) -{ - unsigned i, j; - - assert(dst->pitch > 0); - assert(width <= dst->pitch); - - (void)pipe->region_map(pipe, dst); - - switch (dst->cpp) { - case 1: - { - ubyte *row = get_pointer(dst, dstx, dsty); - for (i = 0; i < height; i++) { - memset(row, value, width); - row += dst->pitch; - } - } - break; - case 2: - { - ushort *row = (ushort *) get_pointer(dst, dstx, dsty); - for (i = 0; i < height; i++) { - for (j = 0; j < width; j++) - row[j] = (ushort) value; - row += dst->pitch; - } - } - break; - case 4: - { - unsigned *row = (unsigned *) get_pointer(dst, dstx, dsty); - for (i = 0; i < height; i++) { - for (j = 0; j < width; j++) - row[j] = value; - row += dst->pitch; - } - } - break; - case 8: - { - /* expand the 4-byte clear value to an 8-byte value */ - ushort *row = (ushort *) get_pointer(dst, dstx, dsty); - ushort val0 = UBYTE_TO_USHORT((value >> 0) & 0xff); - ushort val1 = UBYTE_TO_USHORT((value >> 8) & 0xff); - ushort val2 = UBYTE_TO_USHORT((value >> 16) & 0xff); - ushort val3 = UBYTE_TO_USHORT((value >> 24) & 0xff); - for (i = 0; i < height; i++) { - for (j = 0; j < width; j++) { - row[j*4+0] = val0; - row[j*4+1] = val1; - row[j*4+2] = val2; - row[j*4+3] = val3; - } - row += dst->pitch * 4; - } - } - break; - default: - assert(0); - break; - } - - pipe->region_unmap( pipe, dst ); -} - - - - - void sp_init_region_functions(struct softpipe_context *sp) { sp->pipe.region_map = sp_region_map; sp->pipe.region_unmap = sp_region_unmap; - sp->pipe.region_data = sp_region_data; - sp->pipe.region_copy = sp_region_copy; - sp->pipe.region_fill = sp_region_fill; } diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index b7c9d4f0045..2ddf3ab99c2 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -77,7 +77,7 @@ a8r8g8b8_get_tile(struct pipe_surface *ps, { const unsigned *src = ((const unsigned *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -95,7 +95,7 @@ a8r8g8b8_get_tile(struct pipe_surface *ps, pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -108,7 +108,7 @@ a8r8g8b8_put_tile(struct pipe_surface *ps, { unsigned *dst = ((unsigned *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -127,7 +127,7 @@ a8r8g8b8_put_tile(struct pipe_surface *ps, dst[j] = (a << 24) | (r << 16) | (g << 8) | b; pRow += 4; } - dst += ps->region->pitch; + dst += ps->pitch; p += w0 * 4; } } @@ -141,7 +141,7 @@ a1r5g5b5_get_tile(struct pipe_surface *ps, { const ushort *src = ((const ushort *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; assert(ps->format == PIPE_FORMAT_U_A1_R5_G5_B5); @@ -155,7 +155,7 @@ a1r5g5b5_get_tile(struct pipe_surface *ps, p[3] = ((pixel >> 15) ) * 1.0f; p += 4; } - src += ps->region->pitch; + src += ps->pitch; } } @@ -172,7 +172,7 @@ z16_get_tile(struct pipe_surface *ps, { const ushort *src = ((const ushort *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; const float scale = 1.0f / 65535.0f; unsigned i, j; unsigned w0 = w; @@ -189,7 +189,7 @@ z16_get_tile(struct pipe_surface *ps, pRow[j * 4 + 2] = pRow[j * 4 + 3] = src[j] * scale; } - src += ps->region->pitch; + src += ps->pitch; p += 4 * w0; } } @@ -205,7 +205,7 @@ l8_get_tile(struct pipe_surface *ps, { const ubyte *src = ((const ubyte *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -222,7 +222,7 @@ l8_get_tile(struct pipe_surface *ps, pRow[3] = 1.0; pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -236,7 +236,7 @@ a8_get_tile(struct pipe_surface *ps, { const ubyte *src = ((const ubyte *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -253,7 +253,7 @@ a8_get_tile(struct pipe_surface *ps, pRow[3] = UBYTE_TO_FLOAT(src[j]); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -267,7 +267,7 @@ r16g16b16a16_get_tile(struct pipe_surface *ps, { const short *src = ((const short *) (ps->region->map + ps->offset)) - + (y * ps->region->pitch + x) * 4; + + (y * ps->pitch + x) * 4; unsigned i, j; unsigned w0 = w; @@ -286,7 +286,7 @@ r16g16b16a16_get_tile(struct pipe_surface *ps, pRow += 4; pixel += 4; } - src += ps->region->pitch * 4; + src += ps->pitch * 4; p += w0 * 4; } } @@ -299,7 +299,7 @@ r16g16b16a16_put_tile(struct pipe_surface *ps, { short *dst = ((short *) (ps->region->map + ps->offset)) - + (y * ps->region->pitch + x) * 4; + + (y * ps->pitch + x) * 4; unsigned i, j; unsigned w0 = w; @@ -321,7 +321,7 @@ r16g16b16a16_put_tile(struct pipe_surface *ps, dst[j*4+3] = a; pRow += 4; } - dst += ps->region->pitch * 4; + dst += ps->pitch * 4; p += w0 * 4; } } @@ -336,7 +336,7 @@ i8_get_tile(struct pipe_surface *ps, { const ubyte *src = ((const ubyte *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -353,7 +353,7 @@ i8_get_tile(struct pipe_surface *ps, pRow[3] = UBYTE_TO_FLOAT(src[j]); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -367,7 +367,7 @@ a8_l8_get_tile(struct pipe_surface *ps, { const ushort *src = ((const ushort *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -385,7 +385,7 @@ a8_l8_get_tile(struct pipe_surface *ps, pRow[3] = UBYTE_TO_FLOAT(p >> 8); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -404,7 +404,7 @@ z32_get_tile(struct pipe_surface *ps, { const uint *src = ((const uint *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; const double scale = 1.0 / (double) 0xffffffff; unsigned i, j; unsigned w0 = w; @@ -421,7 +421,7 @@ z32_get_tile(struct pipe_surface *ps, pRow[j * 4 + 2] = pRow[j * 4 + 3] = (float) (scale * src[j]); } - src += ps->region->pitch; + src += ps->pitch; p += 4 * w0; } } @@ -438,7 +438,7 @@ s8z24_get_tile(struct pipe_surface *ps, { const uint *src = ((const uint *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; const double scale = 1.0 / ((1 << 24) - 1); unsigned i, j; unsigned w0 = w; @@ -455,7 +455,7 @@ s8z24_get_tile(struct pipe_surface *ps, pRow[j * 4 + 2] = pRow[j * 4 + 3] = (float) (scale * (src[j] & 0xffffff)); } - src += ps->region->pitch; + src += ps->pitch; p += 4 * w0; } } @@ -472,7 +472,7 @@ z24s8_get_tile(struct pipe_surface *ps, { const uint *src = ((const uint *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + + y * ps->pitch + x; const double scale = 1.0 / ((1 << 24) - 1); unsigned i, j; unsigned w0 = w; @@ -489,7 +489,7 @@ z24s8_get_tile(struct pipe_surface *ps, pRow[j * 4 + 2] = pRow[j * 4 + 3] = (float) (scale * (src[j] >> 8)); } - src += ps->region->pitch; + src += ps->pitch; p += 4 * w0; } } @@ -525,8 +525,10 @@ softpipe_get_tex_surface(struct pipe_context *pipe, assert(ps->format); assert(ps->refcount); pipe_region_reference(&ps->region, mt->region); + ps->cpp = mt->cpp; ps->width = mt->level[level].width; ps->height = mt->level[level].height; + ps->pitch = mt->pitch; ps->offset = offset; } return ps; @@ -541,7 +543,7 @@ softpipe_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, uint x, uint y, uint w, uint h, void *p, int dst_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const ubyte *pSrc; ubyte *pDest; uint i; @@ -554,13 +556,13 @@ softpipe_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, CLIP_TILE; - pSrc = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pSrc = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; pDest = (ubyte *) p; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w * cpp); pDest += dst_stride; - pSrc += ps->region->pitch * cpp; + pSrc += ps->pitch * cpp; } } @@ -573,7 +575,7 @@ softpipe_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, uint x, uint y, uint w, uint h, const void *p, int src_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const ubyte *pSrc; ubyte *pDest; uint i; @@ -587,11 +589,11 @@ softpipe_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, CLIP_TILE; pSrc = (const ubyte *) p; - pDest = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pDest = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w * cpp); - pDest += ps->region->pitch * cpp; + pDest += ps->pitch * cpp; pSrc += src_stride; } } @@ -691,6 +693,180 @@ softpipe_put_tile_rgba(struct pipe_context *pipe, } +/** + * Copy 2D rect from one place to another. + * Position and sizes are in pixels. + */ +static void +copy_rect(ubyte * dst, + unsigned cpp, + unsigned dst_pitch, + unsigned dst_x, + unsigned dst_y, + unsigned width, + unsigned height, + const ubyte * src, + unsigned src_pitch, + unsigned src_x, + unsigned src_y) +{ + unsigned i; + + dst_pitch *= cpp; + src_pitch *= cpp; + dst += dst_x * cpp; + src += src_x * cpp; + dst += dst_y * dst_pitch; + src += src_y * src_pitch; + width *= cpp; + + if (width == dst_pitch && width == src_pitch) + memcpy(dst, src, height * width); + else { + for (i = 0; i < height; i++) { + memcpy(dst, src, width); + dst += dst_pitch; + src += src_pitch; + } + } +} + + +/* Upload data to a rectangular sub-region. Lots of choices how to do this: + * + * - memcpy by span to current destination + * - upload data as new buffer and blit + * + * Currently always memcpy. + */ +static void +sp_surface_data(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + const void *src, unsigned src_pitch, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + copy_rect(pipe->region_map(pipe, dst->region) + dst->offset, + dst->cpp, + dst->pitch, + dstx, dsty, width, height, src, src_pitch, srcx, srcy); + + pipe->region_unmap(pipe, dst->region); +} + +/* Assumes all values are within bounds -- no checking at this level - + * do it higher up if required. + */ +static void +sp_surface_copy(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + struct pipe_surface *src, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + ubyte *src_map, *dst_map; + assert( dst->cpp == src->cpp ); + + dst_map = pipe->region_map(pipe, dst->region); + src_map = pipe->region_map(pipe, src->region); + copy_rect(dst_map + dst->offset, + dst->cpp, + dst->pitch, + dstx, dsty, + width, height, + src_map + src->offset, + src->pitch, + srcx, srcy); + + pipe->region_unmap(pipe, src->region); + pipe->region_unmap(pipe, dst->region); +} + + +static ubyte * +get_pointer(struct pipe_surface *dst, unsigned x, unsigned y) +{ + return dst->region->map + (y * dst->pitch + x) * dst->cpp; +} + + +#define UBYTE_TO_USHORT(B) ((B) | ((B) << 8)) + + +/** + * Fill a rectangular sub-region. Need better logic about when to + * push buffers into AGP - will currently do so whenever possible. + */ +static void +sp_surface_fill(struct pipe_context *pipe, + struct pipe_surface *dst, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, unsigned value) +{ + unsigned i, j; + + assert(dst->pitch > 0); + assert(width <= dst->pitch); + + (void)pipe->region_map(pipe, dst->region); + + switch (dst->cpp) { + case 1: + { + ubyte *row = get_pointer(dst, dstx, dsty); + for (i = 0; i < height; i++) { + memset(row, value, width); + row += dst->pitch; + } + } + break; + case 2: + { + ushort *row = (ushort *) get_pointer(dst, dstx, dsty); + for (i = 0; i < height; i++) { + for (j = 0; j < width; j++) + row[j] = (ushort) value; + row += dst->pitch; + } + } + break; + case 4: + { + unsigned *row = (unsigned *) get_pointer(dst, dstx, dsty); + for (i = 0; i < height; i++) { + for (j = 0; j < width; j++) + row[j] = value; + row += dst->pitch; + } + } + break; + case 8: + { + /* expand the 4-byte clear value to an 8-byte value */ + ushort *row = (ushort *) get_pointer(dst, dstx, dsty); + ushort val0 = UBYTE_TO_USHORT((value >> 0) & 0xff); + ushort val1 = UBYTE_TO_USHORT((value >> 8) & 0xff); + ushort val2 = UBYTE_TO_USHORT((value >> 16) & 0xff); + ushort val3 = UBYTE_TO_USHORT((value >> 24) & 0xff); + for (i = 0; i < height; i++) { + for (j = 0; j < width; j++) { + row[j*4+0] = val0; + row[j*4+1] = val1; + row[j*4+2] = val2; + row[j*4+3] = val3; + } + row += dst->pitch * 4; + } + } + break; + default: + assert(0); + break; + } + + pipe->region_unmap( pipe, dst->region ); +} + void sp_init_surface_functions(struct softpipe_context *sp) @@ -700,4 +876,8 @@ sp_init_surface_functions(struct softpipe_context *sp) sp->pipe.get_tile_rgba = softpipe_get_tile_rgba; sp->pipe.put_tile_rgba = softpipe_put_tile_rgba; + + sp->pipe.surface_data = sp_surface_data; + sp->pipe.surface_copy = sp_surface_copy; + sp->pipe.surface_fill = sp_surface_fill; } diff --git a/src/mesa/pipe/xlib/xm_buffer.c b/src/mesa/pipe/xlib/xm_buffer.c index 7dc85bf2ebe..71a2003e6df 100644 --- a/src/mesa/pipe/xlib/xm_buffer.c +++ b/src/mesa/pipe/xlib/xm_buffer.c @@ -260,9 +260,8 @@ finish_surface_init(GLcontext *ctx, struct xmesa_renderbuffer *xrb) { struct pipe_context *pipe = ctx->st->pipe; if (!xrb->St.surface->region) { - int w = 1, h = 1; - xrb->St.surface->region = pipe->winsys->region_alloc(pipe->winsys, - 1, w, h, 0x0); + xrb->St.surface->region = pipe->winsys->region_alloc(pipe->winsys, 1, + 0x0); } } diff --git a/src/mesa/pipe/xlib/xm_surface.c b/src/mesa/pipe/xlib/xm_surface.c index 43e50507476..23945630959 100644 --- a/src/mesa/pipe/xlib/xm_surface.c +++ b/src/mesa/pipe/xlib/xm_surface.c @@ -624,7 +624,7 @@ xmesa_new_color_surface(struct pipe_winsys *winsys, GLuint pipeFormat) * The region's size will get set in the xmesa_alloc_front/back_storage() * functions. */ - xms->surface.region = winsys->region_alloc(winsys, 1, 1, 1, 0x0); + xms->surface.region = winsys->region_alloc(winsys, 1, 0x0); return &xms->surface; } diff --git a/src/mesa/pipe/xlib/xm_winsys.c b/src/mesa/pipe/xlib/xm_winsys.c index b842cf766e3..99816a811d6 100644 --- a/src/mesa/pipe/xlib/xm_winsys.c +++ b/src/mesa/pipe/xlib/xm_winsys.c @@ -220,27 +220,30 @@ round_up(unsigned n, unsigned multiple) } +static unsigned +xm_surface_pitch(struct pipe_winsys *winsys, unsigned cpp, unsigned width, + unsigned flags) +{ + return round_up(width, 64 / cpp); +} + + static struct pipe_region * -xm_region_alloc(struct pipe_winsys *winsys, - unsigned cpp, unsigned width, unsigned height, unsigned flags) +xm_region_alloc(struct pipe_winsys *winsys, unsigned size, unsigned flags) { struct pipe_region *region = CALLOC_STRUCT(pipe_region); const unsigned alignment = 64; - region->cpp = cpp; - region->pitch = round_up(width, alignment / cpp); - region->height = height; region->refcount = 1; - assert(region->pitch > 0); + assert(size > 0); - region->buffer = winsys->buffer_create( winsys, alignment ) -; + region->buffer = winsys->buffer_create( winsys, alignment ); /* NULL data --> just allocate the space */ winsys->buffer_data( winsys, region->buffer, - region->pitch * cpp * height, + size, NULL, PIPE_BUFFER_USAGE_PIXEL ); return region; @@ -335,6 +338,7 @@ xmesa_get_pipe_winsys(void) ws->region_alloc = xm_region_alloc; ws->region_release = xm_region_release; + ws->surface_pitch = xm_surface_pitch; ws->surface_alloc = xm_surface_alloc; ws->surface_release = xm_surface_release; diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index df41d5ce5b2..39c0cf6b327 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -455,10 +455,6 @@ alloc_mipmap_tree(struct st_context *st, cpp = st_sizeof_format(pipeFormat); - /* allocate texture region/storage */ - mt->region = st->pipe->winsys->region_alloc(st->pipe->winsys, - cpp, width, height, flags); - mt->target = PIPE_TEXTURE_2D; mt->internal_format = GL_RGBA; mt->format = pipeFormat; @@ -469,7 +465,10 @@ alloc_mipmap_tree(struct st_context *st, mt->depth0 = 1; mt->cpp = cpp; mt->compressed = 0; - mt->pitch = mt->region->pitch; + mt->pitch = st->pipe->winsys->surface_pitch(st->pipe->winsys, cpp, width, + flags); + mt->region = st->pipe->winsys->region_alloc(st->pipe->winsys, + mt->pitch * cpp * height, flags); mt->depth_pitch = 0; mt->total_height = height; mt->level[0].level_offset = 0; @@ -524,7 +523,7 @@ make_mipmap_tree(struct st_context *st, { static const GLuint dstImageOffsets = 0; GLboolean success; - GLuint pitch = mt->region->pitch; + GLuint pitch = mt->pitch; GLubyte *dest; const GLbitfield imageTransferStateSave = ctx->_ImageTransferState; @@ -569,7 +568,7 @@ make_mipmap_tree(struct st_context *st, mt->depth0 = 1; mt->cpp = cpp; mt->compressed = 0; - mt->pitch = mt->region->pitch; + mt->pitch = mt->pitch; mt->depth_pitch = 0; mt->total_height = height; mt->level[0].level_offset = 0; @@ -952,13 +951,13 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, switch (ps->format) { case PIPE_FORMAT_U_S8: { - ubyte *dest = stmap + spanY * ps->region->pitch + spanX; + ubyte *dest = stmap + spanY * ps->pitch + spanX; memcpy(dest, values, spanWidth); } break; case PIPE_FORMAT_S8_Z24: { - uint *dest = (uint *) stmap + spanY * ps->region->pitch + spanX; + uint *dest = (uint *) stmap + spanY * ps->pitch + spanX; GLint k; for (k = 0; k < spanWidth; k++) { uint p = dest[k]; @@ -1054,7 +1053,7 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, { struct pipe_context *pipe = ctx->st->pipe; const uint flags = PIPE_SURFACE_FLAG_TEXTURE; - uint format = 0, cpp, comp, pitch; + uint format = 0, cpp, comp; ubyte *dest; struct pipe_mipmap_tree *mt; int row, col; @@ -1091,9 +1090,9 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, /* allocate texture region/storage */ + mt->pitch = pipe->winsys->surface_pitch(pipe->winsys, cpp, width, flags); mt->region = pipe->winsys->region_alloc(pipe->winsys, - cpp, width, height, flags); - pitch = mt->region->pitch; + mt->pitch * cpp * height, flags); /* map texture region */ dest = pipe->region_map(pipe, mt->region); @@ -1110,7 +1109,7 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, for (row = 0; row < height; row++) { const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack, bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0); - ubyte *destRow = dest + row * pitch * cpp; + ubyte *destRow = dest + row * mt->pitch * cpp; if (unpack->LsbFirst) { /* Lsb first */ @@ -1172,7 +1171,6 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, mt->depth0 = 1; mt->cpp = cpp; mt->compressed = 0; - mt->pitch = mt->region->pitch; mt->depth_pitch = 0; mt->total_height = height; mt->level[0].level_offset = 0; @@ -1256,7 +1254,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, y = ctx->DrawBuffer->Height - y - 1; } - dst = drawMap + (y * psDraw->region->pitch + dstx) * psDraw->region->cpp; + dst = drawMap + (y * psDraw->pitch + dstx) * psDraw->cpp; src = buffer + i * width; switch (psDraw->format) { @@ -1297,6 +1295,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, struct st_vertex_program *stvp; struct st_fragment_program *stfp; struct pipe_surface *psRead; + struct pipe_surface *psTex; struct pipe_mipmap_tree *mt; GLfloat *color; uint format; @@ -1332,6 +1331,8 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, if (!mt) return; + psTex = pipe->get_tex_surface(pipe, mt, 0, 0, 0); + if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) { srcy = ctx->DrawBuffer->Height - srcy - height; } @@ -1342,21 +1343,16 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, */ if (st->haveFramebufferRegions) { /* copy source framebuffer region into mipmap/texture */ - pipe->region_copy(pipe, - mt->region, /* dest */ - 0, /* dest_offset */ - 0, 0, /* destx/y */ - psRead->region, - 0, /* src_offset */ - srcx, srcy, width, height); + pipe->surface_copy(pipe, + psTex, /* dest */ + 0, 0, /* destx/y */ + psRead, + srcx, srcy, width, height); } else { /* alternate path using get/put_tile() */ - struct pipe_surface *psTex; GLfloat *buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - psTex = pipe->get_tex_surface(pipe, mt, 0, 0, 0); - (void) pipe->region_map(pipe, psRead->region); (void) pipe->region_map(pipe, psTex->region); @@ -1366,17 +1362,15 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, pipe->region_unmap(pipe, psRead->region); pipe->region_unmap(pipe, psTex->region); - pipe_surface_reference(&psTex, NULL); - free(buf); } - /* draw textured quad */ draw_textured_quad(ctx, dstx, dsty, ctx->Current.RasterPos[2], width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY, mt, stvp, stfp, color, GL_TRUE); + pipe_surface_reference(&psTex, NULL); free_mipmap_tree(st->pipe, mt); } diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index f58d532b2a7..80c92e8b7a1 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -104,6 +104,9 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, assert(strb->surface); if (!strb->surface) return GL_FALSE; + strb->surface->cpp = cpp; + strb->surface->pitch = pipe->winsys->surface_pitch(pipe->winsys, cpp, + width, flags); } /* free old region */ @@ -115,8 +118,9 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, pipe->winsys->region_release(pipe->winsys, &strb->surface->region); } - strb->surface->region = pipe->winsys->region_alloc(pipe->winsys, cpp, - width, height, flags); + strb->surface->region = pipe->winsys->region_alloc(pipe->winsys, + strb->surface->pitch * + cpp * height, flags); if (!strb->surface->region) return GL_FALSE; /* out of memory, try s/w buffer? */ diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index 2e7c01672a1..83fe480af85 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -87,13 +87,13 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, switch (ps->format) { case PIPE_FORMAT_U_S8: { - const ubyte *src = stmap + srcY * ps->region->pitch + x; + const ubyte *src = stmap + srcY * ps->pitch + x; memcpy(values, src, width); } break; case PIPE_FORMAT_S8_Z24: { - const uint *src = (uint *) stmap + srcY * ps->region->pitch + x; + const uint *src = (uint *) stmap + srcY * ps->pitch + x; GLint k; for (k = 0; k < width; k++) { values[k] = src[k] >> 24; @@ -102,7 +102,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, break; case PIPE_FORMAT_Z24_S8: { - const uint *src = (uint *) stmap + srcY * ps->region->pitch + x; + const uint *src = (uint *) stmap + srcY * ps->pitch + x; GLint k; for (k = 0; k < width; k++) { values[k] = src[k] & 0xff; diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 4e6d4857f48..461705119fa 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -44,6 +44,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #define DBG if (0) printf @@ -1150,7 +1151,6 @@ do_copy_texsubimage(GLcontext *ctx, struct st_renderbuffer *strb; struct pipe_context *pipe = ctx->st->pipe; struct pipe_region *src_region, *dest_region; - uint dest_offset, src_offset; uint dest_format, src_format; (void) texImage; @@ -1185,12 +1185,13 @@ do_copy_texsubimage(GLcontext *ctx, ctx->_ImageTransferState == 0x0 && src_region && dest_region && - src_region->cpp == dest_region->cpp) { + strb->surface->cpp == stImage->mt->cpp) { /* do blit-style copy */ - src_offset = 0; - dest_offset = st_miptree_image_offset(stImage->mt, - stImage->face, - stImage->level); + struct pipe_surface *dest_surface = pipe->get_tex_surface(pipe, + stImage->mt, + stImage->face, + stImage->level, + destZ); /* XXX may need to invert image depending on window * vs. user-created FBO @@ -1213,18 +1214,18 @@ do_copy_texsubimage(GLcontext *ctx, GL_COPY); /* ? */ #else - pipe->region_copy(pipe, - /* dest */ - dest_region, - dest_offset, - destX, destY, - /* src */ - src_region, - src_offset, - srcX, srcY, - /* size */ - width, height); + pipe->surface_copy(pipe, + /* dest */ + dest_surface, + destX, destY, + /* src */ + strb->surface, + srcX, srcY, + /* size */ + width, height); #endif + + pipe_surface_reference(&dest_surface, NULL); } else { fallback_copy_texsubimage(ctx, target, level, diff --git a/src/mesa/state_tracker/st_mipmap_tree.c b/src/mesa/state_tracker/st_mipmap_tree.c index d1db590bee4..6ccf33105f1 100644 --- a/src/mesa/state_tracker/st_mipmap_tree.c +++ b/src/mesa/state_tracker/st_mipmap_tree.c @@ -31,6 +31,8 @@ #include "pipe/p_state.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" #include "pipe/p_winsys.h" @@ -87,10 +89,9 @@ st_miptree_create(struct pipe_context *pipe, ok = pipe->mipmap_tree_layout(pipe, mt); if (ok) { - /* note: it's OK to pass 'pitch' as 'width' here: */ - mt->region = pipe->winsys->region_alloc(pipe->winsys, mt->cpp, mt->pitch, - mt->total_height, flags); - mt->pitch = mt->region->pitch; /*XXX NEW */ + mt->region = pipe->winsys->region_alloc(pipe->winsys, + mt->pitch * mt->cpp * + mt->total_height, flags); } if (!mt->region) { @@ -266,24 +267,27 @@ st_miptree_image_data(struct pipe_context *pipe, GLuint src_row_pitch, GLuint src_image_pitch) { GLuint depth = dst->level[level].depth; - GLuint dst_offset = st_miptree_image_offset(dst, face, level); - const GLuint *dst_depth_offset = st_miptree_depth_offsets(dst, level); GLuint i; GLuint height = 0; const GLubyte *srcUB = src; + struct pipe_surface *dst_surface; DBG("%s\n", __FUNCTION__); for (i = 0; i < depth; i++) { height = dst->level[level].height; if(dst->compressed) height /= 4; - pipe->region_data(pipe, dst->region, - dst_offset + dst_depth_offset[i], /* dst_offset */ - 0, 0, /* dstx, dsty */ - srcUB, - src_row_pitch, - 0, 0, /* source x, y */ - dst->level[level].width, height); /* width, height */ + + dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); + + pipe->surface_data(pipe, dst_surface, + 0, 0, /* dstx, dsty */ + srcUB, + src_row_pitch, + 0, 0, /* source x, y */ + dst->level[level].width, height); /* width, height */ + + pipe_surface_reference(&dst_surface, NULL); srcUB += src_image_pitch * dst->cpp; } @@ -300,21 +304,25 @@ st_miptree_image_copy(struct pipe_context *pipe, GLuint width = src->level[level].width; GLuint height = src->level[level].height; GLuint depth = src->level[level].depth; - GLuint dst_offset = st_miptree_image_offset(dst, face, level); - GLuint src_offset = st_miptree_image_offset(src, face, level); - const GLuint *dst_depth_offset = st_miptree_depth_offsets(dst, level); - const GLuint *src_depth_offset = st_miptree_depth_offsets(src, level); + struct pipe_surface *src_surface; + struct pipe_surface *dst_surface; GLuint i; if (dst->compressed) height /= 4; for (i = 0; i < depth; i++) { - pipe->region_copy(pipe, - dst->region, dst_offset + dst_depth_offset[i], - 0, 0, /* destX, Y */ - src->region, src_offset + src_depth_offset[i], - 0, 0, /* srcX, Y */ - width, height); + dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); + src_surface = pipe->get_tex_surface(pipe, src, face, level, i); + + pipe->surface_copy(pipe, + dst_surface, + 0, 0, /* destX, Y */ + src_surface, + 0, 0, /* srcX, Y */ + width, height); + + pipe_surface_reference(&dst_surface, NULL); + pipe_surface_reference(&src_surface, NULL); } } -- cgit v1.2.3 From c989209b6a43f9c21b680e55db0bcda0769dccbf Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Sun, 2 Dec 2007 17:13:28 +0000 Subject: Cover state tracker and winsys driver. Tweak parameters. --- doxygen/gallium.doxy | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/doxygen/gallium.doxy b/doxygen/gallium.doxy index 17a0dadbfc8..b9781973887 100644 --- a/doxygen/gallium.doxy +++ b/doxygen/gallium.doxy @@ -136,7 +136,7 @@ SHORT_NAMES = NO # comments will behave just like regular Qt-style comments # (thus requiring an explicit @brief command for a brief description.) -JAVADOC_AUTOBRIEF = NO +JAVADOC_AUTOBRIEF = YES # If the QT_AUTOBRIEF tag is set to YES then Doxygen will # interpret the first line (until the first dot) of a Qt-style @@ -144,7 +144,7 @@ JAVADOC_AUTOBRIEF = NO # will behave just like regular Qt-style comments (thus requiring # an explicit \brief command for a brief description.) -QT_AUTOBRIEF = NO +QT_AUTOBRIEF = YES # The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make Doxygen # treat a multi-line C++ special comment block (i.e. a block of //! or /// @@ -255,7 +255,7 @@ EXTRACT_STATIC = YES # defined locally in source files will be included in the documentation. # If set to NO only classes defined in header files are included. -EXTRACT_LOCAL_CLASSES = NO +EXTRACT_LOCAL_CLASSES = YES # This flag is only useful for Objective-C code. When set to YES local # methods, which are defined in the implementation section but not in @@ -477,7 +477,10 @@ WARN_LOGFILE = # directories like "/usr/src/myproject". Separate the files or directories # with spaces. -INPUT = gallium.doc ../src/mesa/pipe +INPUT = gallium.doc \ + ../src/mesa/pipe \ + ../src/mesa/state_tracker \ + ../src/mesa/drivers/dri/intel_winsys # This tag can be used to specify the character encoding of the source files that # doxygen parses. Internally doxygen uses the UTF-8 encoding, which is also the default @@ -1020,7 +1023,7 @@ ENABLE_PREPROCESSING = YES # compilation will be performed. Macro expansion can be done in a controlled # way by setting EXPAND_ONLY_PREDEF to YES. -MACRO_EXPANSION = NO +MACRO_EXPANSION = YES # If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES # then the macro expansion is limited to the macros specified with the @@ -1054,7 +1057,7 @@ INCLUDE_FILE_PATTERNS = # undefined via #undef or recursively expanded use the := operator # instead of the = operator. -PREDEFINED = +PREDEFINED = INLINE=inline # If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then # this tag can be used to specify a list of macro names that should be expanded. -- cgit v1.2.3 From e5d8ee205a8d6ef83381de9eff5dbef92b1c1a7e Mon Sep 17 00:00:00 2001 From: Brian Date: Sun, 2 Dec 2007 15:20:13 -0700 Subject: glut doesn't need -lXt --- configs/default | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/configs/default b/configs/default index a7de3f813d6..92fc743ab46 100644 --- a/configs/default +++ b/configs/default @@ -71,7 +71,7 @@ PROGRAM_DIRS = demos redbook samples glsl xdemos GL_LIB_DEPS = $(EXTRA_LIB_PATH) -lX11 -lXext -lm -lpthread OSMESA_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GL_LIB) GLU_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GL_LIB) -lm -GLUT_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GLU_LIB) -l$(GL_LIB) $(EXTRA_LIB_PATH) -lX11 -lXmu -lXt -lXi -lm +GLUT_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GLU_LIB) -l$(GL_LIB) $(EXTRA_LIB_PATH) -lX11 -lXmu -lXi -lm GLW_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GL_LIB) $(EXTRA_LIB_PATH) -lXt -lX11 APP_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GLUT_LIB) -l$(GLU_LIB) -l$(GL_LIB) -lm -- cgit v1.2.3 From 1698cbde3029f047fe92107b3fd1176c258bd9a2 Mon Sep 17 00:00:00 2001 From: Brian Date: Sun, 2 Dec 2007 15:23:51 -0700 Subject: Initial Cell driver infrastructure. No real code yet. Just stand-ins and make/build infrastructure. --- Makefile | 1 + configs/linux-cell | 37 +++++++++++++++++ src/mesa/Makefile | 9 ++++- src/mesa/pipe/Makefile | 28 ++++++++++--- src/mesa/pipe/cell/Makefile | 12 ++++++ src/mesa/pipe/cell/common.h | 13 ++++++ src/mesa/pipe/cell/ppu/Makefile | 39 ++++++++++++++++++ src/mesa/pipe/cell/ppu/cell_context.c | 76 +++++++++++++++++++++++++++++++++++ src/mesa/pipe/cell/ppu/cell_context.h | 46 +++++++++++++++++++++ src/mesa/pipe/cell/ppu/cell_surface.c | 9 +++++ src/mesa/pipe/cell/spu/Makefile | 42 +++++++++++++++++++ src/mesa/pipe/cell/spu/main.c | 29 +++++++++++++ src/mesa/pipe/cell/spu/tri.c | 9 +++++ src/mesa/pipe/cell/spu/tri.h | 4 ++ 14 files changed, 346 insertions(+), 8 deletions(-) create mode 100644 configs/linux-cell create mode 100644 src/mesa/pipe/cell/Makefile create mode 100644 src/mesa/pipe/cell/common.h create mode 100644 src/mesa/pipe/cell/ppu/Makefile create mode 100644 src/mesa/pipe/cell/ppu/cell_context.c create mode 100644 src/mesa/pipe/cell/ppu/cell_context.h create mode 100644 src/mesa/pipe/cell/ppu/cell_surface.c create mode 100644 src/mesa/pipe/cell/spu/Makefile create mode 100644 src/mesa/pipe/cell/spu/main.c create mode 100644 src/mesa/pipe/cell/spu/tri.c create mode 100644 src/mesa/pipe/cell/spu/tri.h diff --git a/Makefile b/Makefile index c4ffea85a2b..6e92858ad50 100644 --- a/Makefile +++ b/Makefile @@ -97,6 +97,7 @@ irix6-o32-static \ linux \ linux-alpha \ linux-alpha-static \ +linux-cell \ linux-debug \ linux-directfb \ linux-dri \ diff --git a/configs/linux-cell b/configs/linux-cell new file mode 100644 index 00000000000..8bedca92467 --- /dev/null +++ b/configs/linux-cell @@ -0,0 +1,37 @@ +# linux-cell + +include $(TOP)/configs/default + +CONFIG_NAME = linux-cell + + +# Compiler and flags +CC = /usr/bin/ppu32-gcc +CXX = /usr/bin/ppu32-g++ + +# Cell SDK location +SDK = /opt/ibm/cell-sdk/prototype/sysroot/usr + + +CFLAGS = -g -W -Wall -Winline -m32 -mabi=altivec -maltivec -I. -I/usr/include -I$(SDK)/include -include altivec.h + +CXXFLAGS = $(CFLAGS) + +GL_LIB_DEPS = $(EXTRA_LIB_PATH) -lX11 -lXext -lm -lpthread \ + -L$(SDK)/lib -m32 -Wl,-m,elf32ppc -R$(SDK)/lib -lspe + + + +### SPU stuff + +SPU_CC = /usr/bin/spu-gcc + +SPU_CFLAGS = -g -W -Wall -Winline -Wno-main -I. -I $(SDK)/spu/include -include spu_intrinsics.h -I $(TOP)/src/mesa/ + +SPU_LFLAGS = -L$(SDK)/spu/lib -Wl,-N -lmisc + +SPU_AR = /usr/bin/ar +SPU_AR_FLAGS = -qcs + +SPU_EMBED = /usr/bin/embedspu +SPU_EMBED_FLAGS = -m32 diff --git a/src/mesa/Makefile b/src/mesa/Makefile index 521cbeb6064..6f0877d335b 100644 --- a/src/mesa/Makefile +++ b/src/mesa/Makefile @@ -13,6 +13,11 @@ GL_TINY = 0$(MESA_MAJOR)0$(MESA_MINOR)0$(MESA_TINY) SOFTPIPE_LIB = $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a +ifeq ($(CONFIG_NAME), linux-cell) +CELL_LIB = $(TOP)/src/mesa/pipe/cell/ppu/libcell.a +endif + + .SUFFIXES : .cpp .c.o: @@ -112,12 +117,12 @@ stand-alone: depend subdirs $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) $(TOP)/$(LIB_DIR)/$ osmesa-only: depend subdirs $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME) # Make the GL library -$(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(STAND_ALONE_OBJECTS) $(SOFTPIPE_LIB) +$(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(STAND_ALONE_OBJECTS) $(SOFTPIPE_LIB) $(CELL_LIB) @ $(TOP)/bin/mklib -o $(GL_LIB) \ -major $(GL_MAJOR) -minor $(GL_MINOR) -patch $(GL_TINY) \ -install $(TOP)/$(LIB_DIR) \ $(MKLIB_OPTIONS) $(STAND_ALONE_OBJECTS) \ - $(SOFTPIPE_LIB) $(GL_LIB_DEPS) + $(SOFTPIPE_LIB) $(CELL_LIB)$(GL_LIB_DEPS) # Make the OSMesa library $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME): $(OSMESA_DRIVER_OBJECTS) $(OSMESA16_OBJECTS) diff --git a/src/mesa/pipe/Makefile b/src/mesa/pipe/Makefile index 7394ad7f611..5099b65885e 100644 --- a/src/mesa/pipe/Makefile +++ b/src/mesa/pipe/Makefile @@ -1,9 +1,25 @@ -default: - cd softpipe ; make - cd i915simple ; make - cd failover ; make - cd nv40; make - cd nv50; make + +TOP = ../../.. +include $(TOP)/configs/current + + +ifeq ($(CONFIG_NAME), linux-cell) +CELL_DIR = cell +endif + +SUBDIRS = softpipe i915simple nv40 nv50 failover $(CELL_DIR) + + +default: subdirs + + +subdirs: + @for dir in $(SUBDIRS) ; do \ + if [ -d $$dir ] ; then \ + (cd $$dir && $(MAKE)) || exit 1 ; \ + fi \ + done + clean: rm -f `find . -name \*.[oa]` diff --git a/src/mesa/pipe/cell/Makefile b/src/mesa/pipe/cell/Makefile new file mode 100644 index 00000000000..47aef7b05f6 --- /dev/null +++ b/src/mesa/pipe/cell/Makefile @@ -0,0 +1,12 @@ +# Cell Gallium driver Makefile + + +default: + ( cd spu ; make ) + ( cd ppu ; make ) + + + +clean: + ( cd spu ; make clean ) + ( cd ppu ; make clean ) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h new file mode 100644 index 00000000000..c4bad4194a1 --- /dev/null +++ b/src/mesa/pipe/cell/common.h @@ -0,0 +1,13 @@ + +#ifndef CELL_COMMON_H +#define CELL_COMMON_H + + +struct init_info +{ + int foo; + int bar; +}; + + +#endif /* CELL_COMMON_H */ diff --git a/src/mesa/pipe/cell/ppu/Makefile b/src/mesa/pipe/cell/ppu/Makefile new file mode 100644 index 00000000000..ede341abca4 --- /dev/null +++ b/src/mesa/pipe/cell/ppu/Makefile @@ -0,0 +1,39 @@ +# Gallium3D Cell driver: PPU code + +# This makefile builds the g3dcell.a library which gets pulled into +# the main libGL.so library + + +TOP = ../../../../.. +include $(TOP)/configs/linux-cell + + +#PROG = gl4 + +CELL_LIB = libcell.a + +SPU_CODE_MODULE = ../spu/g3d_spu.a + +OBJECTS = cell_context.o cell_surface.o + +INCLUDE_DIRS = -I$(TOP)/src/mesa + + +.c.o: + $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $< -o $@ + + + +default: $(CELL_LIB) + + +$(CELL_LIB): $(OBJECTS) $(SPU_CODE_MODULE) + ar -ru $(CELL_LIB) $(OBJECTS) $(SPU_CODE_MODULE) + +#$(PROG): $(PPU_OBJECTS) +# $(CC) -o $(PROG) $(PPU_OBJECTS) $(SPU_CODE_MODULE) $(PPU_LFLAGS) + + + +clean: + rm -f *.o $(CELL_LIB) diff --git a/src/mesa/pipe/cell/ppu/cell_context.c b/src/mesa/pipe/cell/ppu/cell_context.c new file mode 100644 index 00000000000..a8f6cba2fa9 --- /dev/null +++ b/src/mesa/pipe/cell/ppu/cell_context.c @@ -0,0 +1,76 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * Authors + * Brian Paul + */ + + +#include +#include +#include +#include "pipe/cell/ppu/cell_context.h" +#include "pipe/cell/common.h" + +#define NUM_SPUS 6 + + +extern spe_program_handle_t g3d_spu; + +static speid_t speid[NUM_SPUS]; +static struct init_info inits[NUM_SPUS]; + + +static void +start_spus(void) +{ + int i; + + for (i = 0; i < NUM_SPUS; i++) { + inits[i].foo = i; + inits[i].bar = i * 10; + + speid[i] = spe_create_thread(0, /* gid */ + &g3d_spu, /* spe program handle */ + &inits[i], /* argp */ + NULL, /* envp */ + -1, /* mask */ + 0 ); /* flags */ + } +} + + + +void cell_create_context(void) +{ + printf("cell_create_context\n"); + + start_spus(); + + /* TODO: do something with the SPUs! */ +} diff --git a/src/mesa/pipe/cell/ppu/cell_context.h b/src/mesa/pipe/cell/ppu/cell_context.h new file mode 100644 index 00000000000..b4d93d14141 --- /dev/null +++ b/src/mesa/pipe/cell/ppu/cell_context.h @@ -0,0 +1,46 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef CELL_CONTEXT_H +#define CELL_CONTEXT_H + + +#include "pipe/p_context.h" + + + +struct cell_context +{ + struct pipe_context pipe; + + int spu_info; +}; + + + +#endif /* CELL_CONTEXT_H */ diff --git a/src/mesa/pipe/cell/ppu/cell_surface.c b/src/mesa/pipe/cell/ppu/cell_surface.c new file mode 100644 index 00000000000..2c4b6e640bd --- /dev/null +++ b/src/mesa/pipe/cell/ppu/cell_surface.c @@ -0,0 +1,9 @@ + +#include + + +void cell_create_surface(void) +{ + printf("cell_create_surface\n"); + +} diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile new file mode 100644 index 00000000000..00f931e1c19 --- /dev/null +++ b/src/mesa/pipe/cell/spu/Makefile @@ -0,0 +1,42 @@ +# Gallium3D Cell driver: PPU code + +# This makefile builds the g3d_spu.a file that's linked into the +# PPU code/library. + + +TOP = ../../../../.. +include $(TOP)/configs/linux-cell + + +PROG = g3d + +PROG_SPU = $(PROG)_spu +PROG_SPU_A = $(PROG)_spu.a +PROG_SPU_EMBED_O = $(PROG)_spu-embed.o + + +SPU_OBJECTS = main.o tri.o + + +# The .a file will be linked into the main/PPU executable +default: $(PROG_SPU_A) + +$(PROG_SPU_A): $(PROG_SPU_EMBED_O) + $(SPU_AR) $(SPU_AR_FLAGS) $(PROG_SPU_A) $(PROG_SPU_EMBED_O) + +$(PROG_SPU_EMBED_O): $(PROG_SPU) + $(SPU_EMBED) $(SPU_EMBED_FLAGS) $(PROG_SPU) $(PROG_SPU) $(PROG_SPU_EMBED_O) + +$(PROG_SPU): $(SPU_OBJECTS) + $(SPU_CC) -o $(PROG_SPU) $(SPU_OBJECTS) $(SPU_LFLAGS) + + +main.o: main.c + $(SPU_CC) $(SPU_CFLAGS) -c main.c + +tri.o: tri.c + $(SPU_CC) $(SPU_CFLAGS) -c tri.c + + +clean: + rm -f *.o *.a *.d $(PROG_SPU) diff --git a/src/mesa/pipe/cell/spu/main.c b/src/mesa/pipe/cell/spu/main.c new file mode 100644 index 00000000000..e8d5fdccbf7 --- /dev/null +++ b/src/mesa/pipe/cell/spu/main.c @@ -0,0 +1,29 @@ +/* main.c for cell SPU code */ + + +#include +#include +#include + +#include "tri.h" +#include "pipe/cell/common.h" + + +static struct init_info init; + + +int +main(unsigned long long speid, + unsigned long long argp, + unsigned long long envp) +{ + int tag = 0; + + mfc_get(&init, (unsigned int) argp, sizeof(struct init_info), tag, 0, 0); + + printf("Enter spu main(): init.foo=%d\n", init.foo); + + draw_triangle(0, 1, 2); + + return 0; +} diff --git a/src/mesa/pipe/cell/spu/tri.c b/src/mesa/pipe/cell/spu/tri.c new file mode 100644 index 00000000000..949c3b4c8e5 --- /dev/null +++ b/src/mesa/pipe/cell/spu/tri.c @@ -0,0 +1,9 @@ + +#include "tri.h" + +void +draw_triangle(int v1, int v2, int v3) +{ + + +} diff --git a/src/mesa/pipe/cell/spu/tri.h b/src/mesa/pipe/cell/spu/tri.h new file mode 100644 index 00000000000..6a915de60d5 --- /dev/null +++ b/src/mesa/pipe/cell/spu/tri.h @@ -0,0 +1,4 @@ + + +extern void +draw_triangle(int v1, int v2, int v3); -- cgit v1.2.3 From 0d2ba87b5430ddac97b5bf13ce41c0aed86d2120 Mon Sep 17 00:00:00 2001 From: Brian Date: Sun, 2 Dec 2007 15:34:14 -0700 Subject: don't build libglw --- configs/linux-cell | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/configs/linux-cell b/configs/linux-cell index 8bedca92467..151768583cb 100644 --- a/configs/linux-cell +++ b/configs/linux-cell @@ -17,6 +17,10 @@ CFLAGS = -g -W -Wall -Winline -m32 -mabi=altivec -maltivec -I. -I/usr/include -I CXXFLAGS = $(CFLAGS) +# Omitting glw here: +SRC_DIRS = mesa glu glut/glx + + GL_LIB_DEPS = $(EXTRA_LIB_PATH) -lX11 -lXext -lm -lpthread \ -L$(SDK)/lib -m32 -Wl,-m,elf32ppc -R$(SDK)/lib -lspe -- cgit v1.2.3 From 264f64fd89e5b6abb26d985c0e7787666cbce62d Mon Sep 17 00:00:00 2001 From: Brian Date: Sun, 2 Dec 2007 15:34:27 -0700 Subject: added -fPIC --- configs/linux-cell | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/configs/linux-cell b/configs/linux-cell index 151768583cb..4a327faccc9 100644 --- a/configs/linux-cell +++ b/configs/linux-cell @@ -13,7 +13,7 @@ CXX = /usr/bin/ppu32-g++ SDK = /opt/ibm/cell-sdk/prototype/sysroot/usr -CFLAGS = -g -W -Wall -Winline -m32 -mabi=altivec -maltivec -I. -I/usr/include -I$(SDK)/include -include altivec.h +CFLAGS = -g -W -Wall -Winline -fPIC -m32 -mabi=altivec -maltivec -I. -I/usr/include -I$(SDK)/include -include altivec.h CXXFLAGS = $(CFLAGS) -- cgit v1.2.3 From 7f9a64e2477e0bb96a558f13bd48dfac521fc4da Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:47:27 -0700 Subject: added B8G8R8A8 format --- src/mesa/pipe/p_format.h | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/mesa/pipe/p_format.h b/src/mesa/pipe/p_format.h index 70eb1247099..de0eb9539f8 100644 --- a/src/mesa/pipe/p_format.h +++ b/src/mesa/pipe/p_format.h @@ -154,6 +154,7 @@ static INLINE uint pf_get(pipe_format_rgbazs_t f, uint shift, uint mask) #define _PIPE_FORMAT_RGB0 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_A ) #define _PIPE_FORMAT_RGBA _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_A ) #define _PIPE_FORMAT_ARGB _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_A, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_B ) +#define _PIPE_FORMAT_BGRA _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_B, PIPE_FORMAT_COMP_G, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_A ) #define _PIPE_FORMAT_0000 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0 ) #define _PIPE_FORMAT_000R _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_0, PIPE_FORMAT_COMP_R ) #define _PIPE_FORMAT_RRR1 _PIPE_FORMAT_SWZ( PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_R, PIPE_FORMAT_COMP_1 ) @@ -199,6 +200,7 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f) #define PIPE_FORMAT_NONE _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_0000, 0, 0, 0, 0, PIPE_FORMAT_TYPE_UNKNOWN ) #define PIPE_FORMAT_A8R8G8B8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ARGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) +#define PIPE_FORMAT_B8G8R8A8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGRA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) #define PIPE_FORMAT_A1R5G5B5_UNORM _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 1, 5, 5, 5, PIPE_FORMAT_TYPE_UNORM ) #define PIPE_FORMAT_A4R4G4B4_UNORM _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ) #define PIPE_FORMAT_R5G6B5_UNORM _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_RGB0, 5, 6, 5, 0, PIPE_FORMAT_TYPE_UNORM ) @@ -275,6 +277,7 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f) * Duplicated formats: */ #define PIPE_FORMAT_U_A8_R8_G8_B8 PIPE_FORMAT_A8R8G8B8_UNORM +#define PIPE_FORMAT_U_B8_G8_R8_A8 PIPE_FORMAT_B8G8R8A8_UNORM #define PIPE_FORMAT_U_A1_R5_G5_B5 PIPE_FORMAT_A1R5G5B5_UNORM #define PIPE_FORMAT_U_A4_R4_G4_B4 PIPE_FORMAT_A4R4G4B4_UNORM #define PIPE_FORMAT_U_R8_G8_B8_A8 PIPE_FORMAT_R8G8B8A8_UNORM -- cgit v1.2.3 From 4f25c76735b678c5d2346198cb370d2d58443289 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:48:09 -0700 Subject: added tile funcs for B8G8R8A8 format --- src/mesa/pipe/softpipe/sp_surface.c | 70 +++++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index 2ddf3ab99c2..a6ab4046037 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -133,6 +133,70 @@ a8r8g8b8_put_tile(struct pipe_surface *ps, } +/*** PIPE_FORMAT_U_B8_G8_R8_A8 ***/ + +static void +b8g8r8a8_get_tile(struct pipe_surface *ps, + unsigned x, unsigned y, unsigned w, unsigned h, float *p) +{ + const unsigned *src + = ((const unsigned *) (ps->region->map + ps->offset)) + + y * ps->pitch + x; + unsigned i, j; + unsigned w0 = w; + + assert(ps->format == PIPE_FORMAT_U_B8_G8_R8_A8); + + CLIP_TILE; + + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++) { + const unsigned pixel = src[j]; + pRow[0] = UBYTE_TO_FLOAT((pixel >> 8) & 0xff); + pRow[1] = UBYTE_TO_FLOAT((pixel >> 16) & 0xff); + pRow[2] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); + pRow[3] = UBYTE_TO_FLOAT((pixel >> 0) & 0xff); + pRow += 4; + } + src += ps->pitch; + p += w0 * 4; + } +} + + +static void +b8g8r8a8_put_tile(struct pipe_surface *ps, + unsigned x, unsigned y, unsigned w, unsigned h, + const float *p) +{ + unsigned *dst + = ((unsigned *) (ps->region->map + ps->offset)) + + y * ps->pitch + x; + unsigned i, j; + unsigned w0 = w; + + assert(ps->format == PIPE_FORMAT_U_B8_G8_R8_A8); + + CLIP_TILE; + + for (i = 0; i < h; i++) { + const float *pRow = p; + for (j = 0; j < w; j++) { + unsigned r, g, b, a; + UNCLAMPED_FLOAT_TO_UBYTE(r, pRow[0]); + UNCLAMPED_FLOAT_TO_UBYTE(g, pRow[1]); + UNCLAMPED_FLOAT_TO_UBYTE(b, pRow[2]); + UNCLAMPED_FLOAT_TO_UBYTE(a, pRow[3]); + dst[j] = (b << 24) | (g << 16) | (r << 8) | a; + pRow += 4; + } + dst += ps->pitch; + p += w0 * 4; + } +} + + /*** PIPE_FORMAT_U_A1_R5_G5_B5 ***/ static void @@ -610,6 +674,9 @@ softpipe_get_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_U_A8_R8_G8_B8: a8r8g8b8_get_tile(ps, x, y, w, h, p); break; + case PIPE_FORMAT_U_B8_G8_R8_A8: + b8g8r8a8_get_tile(ps, x, y, w, h, p); + break; case PIPE_FORMAT_U_A1_R5_G5_B5: a1r5g5b5_get_tile(ps, x, y, w, h, p); break; @@ -657,6 +724,9 @@ softpipe_put_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_U_A8_R8_G8_B8: a8r8g8b8_put_tile(ps, x, y, w, h, p); break; + case PIPE_FORMAT_U_B8_G8_R8_A8: + b8g8r8a8_put_tile(ps, x, y, w, h, p); + break; case PIPE_FORMAT_U_A1_R5_G5_B5: /*a1r5g5b5_put_tile(ps, x, y, w, h, p);*/ break; -- cgit v1.2.3 From ecd0853a095c2df1eb0b3c72c1f7c570a23ef6a6 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:51:08 -0700 Subject: added B8G8R8A8 support and improved pixel format selection when doing remote display to X server of different endianness. --- src/mesa/pipe/xlib/xm_api.c | 10 +++++--- src/mesa/pipe/xlib/xm_buffer.c | 30 ++++++++++++++++++++++-- src/mesa/pipe/xlib/xm_surface.c | 51 +++++++++++++++++++++++++++++++++++++++-- src/mesa/pipe/xlib/xmesaP.h | 6 ++++- 4 files changed, 89 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/xlib/xm_api.c b/src/mesa/pipe/xlib/xm_api.c index e456ea2fd4d..f87d72d0c8c 100644 --- a/src/mesa/pipe/xlib/xm_api.c +++ b/src/mesa/pipe/xlib/xm_api.c @@ -329,7 +329,7 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type, /* * Front renderbuffer */ - b->frontxrb = xmesa_create_renderbuffer(winsys, 0, &vis->mesa_visual, GL_FALSE); + b->frontxrb = xmesa_create_renderbuffer(winsys, 0, vis, GL_FALSE); if (!b->frontxrb) { _mesa_free(b); return NULL; @@ -343,7 +343,7 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type, * Back renderbuffer */ if (vis->mesa_visual.doubleBufferMode) { - b->backxrb = xmesa_create_renderbuffer(winsys, 0, &vis->mesa_visual, GL_TRUE); + b->backxrb = xmesa_create_renderbuffer(winsys, 0, vis, GL_TRUE); if (!b->backxrb) { /* XXX free front xrb too */ _mesa_free(b); @@ -365,7 +365,11 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type, /* Visual has alpha, but pixel format doesn't support it. * We'll use an alpha renderbuffer wrapper. */ - b->swAlpha = GL_TRUE; +#if 0 + b->swAlpha = GL_TRUE; /* don't do any renderbuffer wrapping */ +#else + b->swAlpha = GL_FALSE; +#endif } else { b->swAlpha = GL_FALSE; diff --git a/src/mesa/pipe/xlib/xm_buffer.c b/src/mesa/pipe/xlib/xm_buffer.c index 71a2003e6df..449e90184b8 100644 --- a/src/mesa/pipe/xlib/xm_buffer.c +++ b/src/mesa/pipe/xlib/xm_buffer.c @@ -163,6 +163,16 @@ alloc_back_shm_ximage(XMesaBuffer b, GLuint width, GLuint height) #endif +/** + * \return LSBFirst or MSBFirst + */ +static int host_byte_order( void ) +{ + int i = 1; + char *cptr = (char *) &i; + return (*cptr==1) ? LSBFirst : MSBFirst; +} + /** * Setup an off-screen pixmap or Ximage to use as the back buffer. @@ -377,9 +387,11 @@ xmesa_alloc_back_storage(GLcontext *ctx, struct gl_renderbuffer *rb, */ struct xmesa_renderbuffer * xmesa_create_renderbuffer(struct pipe_winsys *winsys, - GLuint name, const GLvisual *visual, + GLuint name, XMesaVisual xmvis, GLboolean backBuffer) { + const GLvisual *visual = &xmvis->mesa_visual; + int byteOrder = ImageByteOrder(xmvis->display); /* LSBFirst or MSBFirst */ struct xmesa_renderbuffer *xrb = CALLOC_STRUCT(xmesa_renderbuffer); if (xrb) { GLuint name = 0; @@ -402,7 +414,21 @@ xmesa_create_renderbuffer(struct pipe_winsys *winsys, xrb->St.Base.GreenBits = visual->greenBits; xrb->St.Base.BlueBits = visual->blueBits; xrb->St.Base.AlphaBits = visual->alphaBits; - pipeFormat = PIPE_FORMAT_U_A8_R8_G8_B8; + if (visual->redMask == 0xff0000 && + visual->greenMask == 0x00ff00 && + visual->blueMask == 0x0000ff) { + if (host_byte_order() != byteOrder) { + pipeFormat = PIPE_FORMAT_U_B8_G8_R8_A8; + /*printf("Using format B8_G8_R8_A8 (LE dpy)\n");*/ + } + else { + pipeFormat = PIPE_FORMAT_U_A8_R8_G8_B8; + /*printf("Using format A8_R8_G8_B8 (BE dpy)\n");*/ + } + } + else { + assert(0); + } } else { xrb->St.Base.InternalFormat = GL_COLOR_INDEX; diff --git a/src/mesa/pipe/xlib/xm_surface.c b/src/mesa/pipe/xlib/xm_surface.c index 23945630959..58ca2e7fb3c 100644 --- a/src/mesa/pipe/xlib/xm_surface.c +++ b/src/mesa/pipe/xlib/xm_surface.c @@ -256,7 +256,7 @@ xmesa_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, /* put to ximage */ ximage = xms->ximage; char *dst; - int i; + uint i; /* this could be optimized/simplified */ switch (ps->format) { @@ -353,7 +353,7 @@ xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, for (j = 0; j < w; j++) { uint pix = src[j]; ubyte r = ((pix >> 16) & 0xff); - ubyte g = ((pix >> 8) & 0xff); + ubyte g = ((pix >> 8) & 0xff); ubyte b = ( pix & 0xff); ubyte a = ((pix >> 24) & 0xff); p[0] = UBYTE_TO_FLOAT(r); @@ -367,6 +367,29 @@ xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, } } break; + case PIPE_FORMAT_U_B8_G8_R8_A8: + { + const uint *src + = (uint *) (ximage->data + y * ximage->bytes_per_line + x * 4); + for (i = 0; i < h; i++) { + float *p = pRow; + for (j = 0; j < w; j++) { + uint pix = src[j]; + ubyte r = ((pix >> 8) & 0xff); + ubyte g = ((pix >> 16) & 0xff); + ubyte b = ((pix >> 24) & 0xff); + ubyte a = ( pix & 0xff); + p[0] = UBYTE_TO_FLOAT(r); + p[1] = UBYTE_TO_FLOAT(g); + p[2] = UBYTE_TO_FLOAT(b); + p[3] = UBYTE_TO_FLOAT(a); + p += 4; + } + src += ximage->width; + pRow += 4 * w0; + } + } + break; case PIPE_FORMAT_U_R5_G6_B5: { ushort *src @@ -390,6 +413,7 @@ xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, } break; default: + fprintf(stderr, "Bad format in xmesa_get_tile_rgba()\n"); assert(0); } @@ -461,6 +485,27 @@ xmesa_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, } } break; + case PIPE_FORMAT_U_B8_G8_R8_A8: + { + uint *dst + = (uint *) (ximage->data + y * ximage->bytes_per_line + x * 4); + const float *pRow = pixels; + for (i = 0; i < h; i++) { + const float *p = pRow; + for (j = 0; j < w; j++) { + ubyte r, g, b, a; + UNCLAMPED_FLOAT_TO_UBYTE(r, p[0]); + UNCLAMPED_FLOAT_TO_UBYTE(g, p[1]); + UNCLAMPED_FLOAT_TO_UBYTE(b, p[2]); + UNCLAMPED_FLOAT_TO_UBYTE(a, p[3]); + dst[j] = PACK_8B8G8R8A(r, g, b, a); + p += 4; + } + dst += ximage->width; + pRow += 4 * w0; + } + } + break; case PIPE_FORMAT_U_R5_G6_B5: { ushort *dst = @@ -483,6 +528,7 @@ xmesa_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, break; default: + fprintf(stderr, "Bad format in xmesa_put_tile_rgba()\n"); assert(0); } @@ -689,6 +735,7 @@ xmesa_clear(struct pipe_context *pipe, struct pipe_surface *ps, uint value) clear_16bit_ximage_surface(pipe, ps, value); break; case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_U_B8_G8_R8_A8: clear_32bit_ximage_surface(pipe, ps, value); break; default: diff --git a/src/mesa/pipe/xlib/xmesaP.h b/src/mesa/pipe/xlib/xmesaP.h index ba0ccdbcecb..a067dfd3e34 100644 --- a/src/mesa/pipe/xlib/xmesaP.h +++ b/src/mesa/pipe/xlib/xmesaP.h @@ -286,6 +286,10 @@ struct xmesa_buffer { */ #define PACK_8B8G8R( R, G, B ) ( ((B) << 16) | ((G) << 8) | (R) ) +#define PACK_8B8G8R8A( R, G, B, A ) \ + ( ((B) << 24) | ((G) << 16) | ((R) << 8) | (A) ) + + /** @@ -448,7 +452,7 @@ extern const int xmesa_kernel1[16]; extern struct xmesa_renderbuffer * xmesa_create_renderbuffer(struct pipe_winsys *winsys, - GLuint name, const GLvisual *visual, + GLuint name, XMesaVisual xmvis, GLboolean backBuffer); extern void -- cgit v1.2.3 From 64a1d5cbd170edfe4e99ca83ab1f60390311d69a Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:53:19 -0700 Subject: Fix endianness bug in _mesa_texstore_argb8888() On big-endian, storing in _mesa_texformat_argb8888 format produced wrong results. Also, clean-up nearby code to match. --- src/mesa/main/texstore.c | 28 +++++++++++++++------------- 1 file changed, 15 insertions(+), 13 deletions(-) diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c index 3b5151ed171..30be65525e0 100644 --- a/src/mesa/main/texstore.c +++ b/src/mesa/main/texstore.c @@ -1,8 +1,8 @@ /* * Mesa 3-D graphics library - * Version: 6.5.1 + * Version: 7.1 * - * Copyright (C) 1999-2006 Brian Paul All Rights Reserved. + * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -1443,7 +1443,6 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS) (baseInternalFormat == GL_RGBA || baseInternalFormat == GL_RGB) && srcType == GL_UNSIGNED_BYTE) { - int img, row, col; for (img = 0; img < srcDepth; img++) { const GLint srcRowStride = _mesa_image_row_stride(srcPacking, @@ -1455,11 +1454,12 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS) + dstYoffset * dstRowStride + dstXoffset * dstFormat->TexelBytes; for (row = 0; row < srcHeight; row++) { + GLuint *d4 = (GLuint *) dstRow; for (col = 0; col < srcWidth; col++) { - dstRow[col * 4 + 0] = srcRow[col * 3 + BCOMP]; - dstRow[col * 4 + 1] = srcRow[col * 3 + GCOMP]; - dstRow[col * 4 + 2] = srcRow[col * 3 + RCOMP]; - dstRow[col * 4 + 3] = 0xff; + d4[col] = ((0xff << 24) | + (srcRow[col * 3 + RCOMP] << 16) | + (srcRow[col * 3 + GCOMP] << 8) | + (srcRow[col * 3 + BCOMP] << 0)); } dstRow += dstRowStride; srcRow += srcRowStride; @@ -1471,7 +1471,9 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS) dstFormat == &_mesa_texformat_argb8888 && srcFormat == GL_RGBA && baseInternalFormat == GL_RGBA && - (srcType == GL_UNSIGNED_BYTE && littleEndian)) { + srcType == GL_UNSIGNED_BYTE && + littleEndian) { + /* same as above case, but src data has alpha too */ GLint img, row, col; /* For some reason, streaming copies to write-combined regions * are extremely sensitive to the characteristics of how the @@ -1488,13 +1490,13 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS) + dstImageOffsets[dstZoffset + img] * dstFormat->TexelBytes + dstYoffset * dstRowStride + dstXoffset * dstFormat->TexelBytes; - for (row = 0; row < srcHeight; row++) { + GLuint *d4 = (GLuint *) dstRow; for (col = 0; col < srcWidth; col++) { - *(GLuint *)(dstRow + col * 4) = (srcRow[col * 4 + RCOMP] << 16 | - srcRow[col * 4 + GCOMP] << 8 | - srcRow[col * 4 + BCOMP] << 0 | - srcRow[col * 4 + ACOMP] << 24); + d4[col] = ((srcRow[col * 4 + ACOMP] << 24) | + (srcRow[col * 4 + RCOMP] << 16) | + (srcRow[col * 4 + GCOMP] << 8) | + (srcRow[col * 4 + BCOMP] << 0)); } dstRow += dstRowStride; srcRow += srcRowStride; -- cgit v1.2.3 From dc76fe436c035f5ba6c9372584198a4d32e057f2 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:54:19 -0700 Subject: added PIPE_FORMAT_U_B8_G8_R8_A8 case in color_value() --- src/mesa/state_tracker/st_cb_clear.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 219a5afcbdd..8ac6699776b 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -69,9 +69,12 @@ color_value(GLuint pipeFormat, const GLfloat color[4]) return (r << 24) | (g << 16) | (b << 8) | a; case PIPE_FORMAT_U_A8_R8_G8_B8: return (a << 24) | (r << 16) | (g << 8) | b; + case PIPE_FORMAT_U_B8_G8_R8_A8: + return (b << 24) | (g << 16) | (r << 8) | a; case PIPE_FORMAT_U_R5_G6_B5: return ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3); default: + assert(0); return 0; } } -- cgit v1.2.3 From 5cb718673016159d06e08a832db594b99c20ad5a Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:54:29 -0700 Subject: silence warning --- src/mesa/state_tracker/st_draw.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 61ff034550d..b4c0d0cdd62 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -83,6 +83,7 @@ pipe_vertex_format(GLenum type, GLuint size) default: assert(0); } + return 0; /* silence compiler warning */ } -- cgit v1.2.3 From 6f44a21348cb70601fc138150bb603fa527d6b63 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 5 Dec 2007 14:57:19 -0700 Subject: Add dependency checking for Cell driver sources. --- configs/linux-cell | 3 +++ src/mesa/Makefile | 5 +++-- src/mesa/pipe/cell/ppu/Makefile | 22 ++++++++++++++++++++-- src/mesa/pipe/cell/spu/Makefile | 20 ++++++++++++++++++-- 4 files changed, 44 insertions(+), 6 deletions(-) diff --git a/configs/linux-cell b/configs/linux-cell index 4a327faccc9..f581b5a6fe3 100644 --- a/configs/linux-cell +++ b/configs/linux-cell @@ -21,6 +21,9 @@ CXXFLAGS = $(CFLAGS) SRC_DIRS = mesa glu glut/glx +MKDEP_OPTIONS = -fdepend -Y + + GL_LIB_DEPS = $(EXTRA_LIB_PATH) -lX11 -lXext -lm -lpthread \ -L$(SDK)/lib -m32 -Wl,-m,elf32ppc -R$(SDK)/lib -lspe diff --git a/src/mesa/Makefile b/src/mesa/Makefile index 6f0877d335b..6f090d9b3d6 100644 --- a/src/mesa/Makefile +++ b/src/mesa/Makefile @@ -15,6 +15,7 @@ SOFTPIPE_LIB = $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a ifeq ($(CONFIG_NAME), linux-cell) CELL_LIB = $(TOP)/src/mesa/pipe/cell/ppu/libcell.a +CELL_LIB_SPU = $(TOP)/src/mesa/pipe/cell/spu/g3d_spu.a endif @@ -117,12 +118,12 @@ stand-alone: depend subdirs $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) $(TOP)/$(LIB_DIR)/$ osmesa-only: depend subdirs $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME) # Make the GL library -$(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(STAND_ALONE_OBJECTS) $(SOFTPIPE_LIB) $(CELL_LIB) +$(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(STAND_ALONE_OBJECTS) $(SOFTPIPE_LIB) $(CELL_LIB) $(CELL_LIB_SPU) @ $(TOP)/bin/mklib -o $(GL_LIB) \ -major $(GL_MAJOR) -minor $(GL_MINOR) -patch $(GL_TINY) \ -install $(TOP)/$(LIB_DIR) \ $(MKLIB_OPTIONS) $(STAND_ALONE_OBJECTS) \ - $(SOFTPIPE_LIB) $(CELL_LIB)$(GL_LIB_DEPS) + $(SOFTPIPE_LIB) $(CELL_LIB) $(CELL_LIB_SPU) $(GL_LIB_DEPS) # Make the OSMesa library $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME): $(OSMESA_DRIVER_OBJECTS) $(OSMESA16_OBJECTS) diff --git a/src/mesa/pipe/cell/ppu/Makefile b/src/mesa/pipe/cell/ppu/Makefile index ede341abca4..c987d54e02f 100644 --- a/src/mesa/pipe/cell/ppu/Makefile +++ b/src/mesa/pipe/cell/ppu/Makefile @@ -14,7 +14,12 @@ CELL_LIB = libcell.a SPU_CODE_MODULE = ../spu/g3d_spu.a -OBJECTS = cell_context.o cell_surface.o + +SOURCES = \ + cell_context.c \ + cell_surface.c + +OBJECTS = $(SOURCES:.c=.o) \ INCLUDE_DIRS = -I$(TOP)/src/mesa @@ -28,7 +33,8 @@ default: $(CELL_LIB) $(CELL_LIB): $(OBJECTS) $(SPU_CODE_MODULE) - ar -ru $(CELL_LIB) $(OBJECTS) $(SPU_CODE_MODULE) +# ar -ru $(CELL_LIB) $(OBJECTS) $(SPU_CODE_MODULE) + ar -ru $(CELL_LIB) $(OBJECTS) #$(PROG): $(PPU_OBJECTS) # $(CC) -o $(PROG) $(PPU_OBJECTS) $(SPU_CODE_MODULE) $(PPU_LFLAGS) @@ -37,3 +43,15 @@ $(CELL_LIB): $(OBJECTS) $(SPU_CODE_MODULE) clean: rm -f *.o $(CELL_LIB) + + + +depend: $(SOURCES) + rm -f depend + touch depend + $(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(SOURCES) 2> /dev/null + +include depend + + + diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index 00f931e1c19..700a6c61cdc 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -1,4 +1,4 @@ -# Gallium3D Cell driver: PPU code +# Gallium3D Cell driver: SPU code # This makefile builds the g3d_spu.a file that's linked into the # PPU code/library. @@ -15,7 +15,13 @@ PROG_SPU_A = $(PROG)_spu.a PROG_SPU_EMBED_O = $(PROG)_spu-embed.o -SPU_OBJECTS = main.o tri.o +SOURCES = \ + main.c \ + tri.c + +SPU_OBJECTS = $(SOURCES:.c=.o) \ + +INCLUDE_DIRS = -I$(TOP)/src/mesa # The .a file will be linked into the main/PPU executable @@ -40,3 +46,13 @@ tri.o: tri.c clean: rm -f *.o *.a *.d $(PROG_SPU) + + + +depend: $(SOURCES) + rm -f depend + touch depend + $(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(SOURCES) 2> /dev/null + +include depend + -- cgit v1.2.3 From 274603c5c9553a13c44c6e3d48ac73c280cd3b64 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Fri, 30 Nov 2007 20:48:03 +0100 Subject: Hide texture layout details from the state tracker. pipe->get_tex_surface() has to be used for access to texture image data. --- src/mesa/drivers/dri/intel_winsys/intel_screen.c | 4 +- src/mesa/pipe/failover/fo_context.c | 3 +- src/mesa/pipe/failover/fo_context.h | 2 +- src/mesa/pipe/failover/fo_state.c | 2 +- src/mesa/pipe/i915simple/Makefile | 2 +- src/mesa/pipe/i915simple/i915_context.c | 9 +- src/mesa/pipe/i915simple/i915_context.h | 30 +- src/mesa/pipe/i915simple/i915_state.c | 4 +- src/mesa/pipe/i915simple/i915_state_sampler.c | 29 +- src/mesa/pipe/i915simple/i915_surface.c | 25 +- src/mesa/pipe/i915simple/i915_tex_layout.c | 474 -------------------- src/mesa/pipe/i915simple/i915_tex_layout.h | 16 - src/mesa/pipe/i915simple/i915_texture.c | 539 +++++++++++++++++++++++ src/mesa/pipe/i915simple/i915_texture.h | 16 + src/mesa/pipe/llvm/llvm_entry.c | 2 - src/mesa/pipe/p_context.h | 11 +- src/mesa/pipe/p_inlines.h | 20 + src/mesa/pipe/p_state.h | 42 +- src/mesa/pipe/softpipe/Makefile | 2 +- src/mesa/pipe/softpipe/sp_context.c | 17 +- src/mesa/pipe/softpipe/sp_context.h | 2 +- src/mesa/pipe/softpipe/sp_quad_fs.c | 2 +- src/mesa/pipe/softpipe/sp_state.h | 31 +- src/mesa/pipe/softpipe/sp_state_fs.c | 2 + src/mesa/pipe/softpipe/sp_state_sampler.c | 4 +- src/mesa/pipe/softpipe/sp_surface.c | 25 +- src/mesa/pipe/softpipe/sp_surface.h | 2 +- src/mesa/pipe/softpipe/sp_tex_layout.c | 361 --------------- src/mesa/pipe/softpipe/sp_tex_layout.h | 16 - src/mesa/pipe/softpipe/sp_tex_sample.c | 18 +- src/mesa/pipe/softpipe/sp_texture.c | 425 ++++++++++++++++++ src/mesa/pipe/softpipe/sp_texture.h | 18 + src/mesa/pipe/softpipe/sp_tile_cache.c | 4 +- src/mesa/pipe/softpipe/sp_tile_cache.h | 2 +- src/mesa/pipe/tgsi/exec/tgsi_exec.h | 2 +- src/mesa/sources | 2 +- src/mesa/state_tracker/st_atom_texture.c | 14 +- src/mesa/state_tracker/st_cb_drawpixels.c | 214 +++------ src/mesa/state_tracker/st_cb_fbo.c | 24 +- src/mesa/state_tracker/st_cb_texture.c | 427 +++++++++--------- src/mesa/state_tracker/st_cb_texture.h | 10 +- src/mesa/state_tracker/st_context.h | 22 +- src/mesa/state_tracker/st_mipmap_tree.c | 328 -------------- src/mesa/state_tracker/st_mipmap_tree.h | 118 ----- src/mesa/state_tracker/st_texture.c | 276 ++++++++++++ src/mesa/state_tracker/st_texture.h | 109 +++++ 46 files changed, 1877 insertions(+), 1830 deletions(-) delete mode 100644 src/mesa/pipe/i915simple/i915_tex_layout.c delete mode 100644 src/mesa/pipe/i915simple/i915_tex_layout.h create mode 100644 src/mesa/pipe/i915simple/i915_texture.c create mode 100644 src/mesa/pipe/i915simple/i915_texture.h delete mode 100644 src/mesa/pipe/softpipe/sp_tex_layout.c delete mode 100644 src/mesa/pipe/softpipe/sp_tex_layout.h create mode 100644 src/mesa/pipe/softpipe/sp_texture.c create mode 100644 src/mesa/pipe/softpipe/sp_texture.h delete mode 100644 src/mesa/state_tracker/st_mipmap_tree.c delete mode 100644 src/mesa/state_tracker/st_mipmap_tree.h create mode 100644 src/mesa/state_tracker/st_texture.c create mode 100644 src/mesa/state_tracker/st_texture.h diff --git a/src/mesa/drivers/dri/intel_winsys/intel_screen.c b/src/mesa/drivers/dri/intel_winsys/intel_screen.c index 01460e5be3c..1b520f71353 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_screen.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_screen.c @@ -328,8 +328,8 @@ intelSetTexOffset(__DRIcontext *pDRICtx, int texname, if (!stObj) return; - if (stObj->mt) - st_miptree_release(intel->st->pipe, &stObj->mt); + if (stObj->pt) + st->pipe->texture_release(intel->st->pipe, &stObj->pt); stObj->imageOverride = GL_TRUE; stObj->depthOverride = depth; diff --git a/src/mesa/pipe/failover/fo_context.c b/src/mesa/pipe/failover/fo_context.c index 0cc9cab408c..a25563d4511 100644 --- a/src/mesa/pipe/failover/fo_context.c +++ b/src/mesa/pipe/failover/fo_context.c @@ -145,7 +145,8 @@ struct pipe_context *failover_create( struct pipe_context *hw, failover->pipe.surface_data = hw->surface_data; failover->pipe.surface_copy = hw->surface_copy; failover->pipe.surface_fill = hw->surface_fill; - failover->pipe.mipmap_tree_layout = hw->mipmap_tree_layout; + failover->pipe.texture_create = hw->texture_create; + failover->pipe.texture_release = hw->texture_release; failover->pipe.flush = hw->flush; failover->dirty = 0; diff --git a/src/mesa/pipe/failover/fo_context.h b/src/mesa/pipe/failover/fo_context.h index 759b53ccbe4..7cf18c9ec19 100644 --- a/src/mesa/pipe/failover/fo_context.h +++ b/src/mesa/pipe/failover/fo_context.h @@ -85,7 +85,7 @@ struct failover_context { struct pipe_poly_stipple poly_stipple; struct pipe_scissor_state scissor; uint sampler_units[PIPE_MAX_SAMPLERS]; - struct pipe_mipmap_tree *texture[PIPE_MAX_SAMPLERS]; + struct pipe_texture *texture[PIPE_MAX_SAMPLERS]; struct pipe_viewport_state viewport; struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX]; diff --git a/src/mesa/pipe/failover/fo_state.c b/src/mesa/pipe/failover/fo_state.c index 2cd1a50b203..fd6137ba662 100644 --- a/src/mesa/pipe/failover/fo_state.c +++ b/src/mesa/pipe/failover/fo_state.c @@ -402,7 +402,7 @@ failover_delete_sampler_state(struct pipe_context *pipe, void *sampler) static void failover_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_mipmap_tree *texture) + struct pipe_texture *texture) { struct failover_context *failover = failover_context(pipe); diff --git a/src/mesa/pipe/i915simple/Makefile b/src/mesa/pipe/i915simple/Makefile index 391a084915d..1223b386a3d 100644 --- a/src/mesa/pipe/i915simple/Makefile +++ b/src/mesa/pipe/i915simple/Makefile @@ -22,7 +22,7 @@ DRIVER_SOURCES = \ i915_strings.c \ i915_prim_emit.c \ i915_prim_vbuf.c \ - i915_tex_layout.c \ + i915_texture.c \ i915_fpc_emit.c \ i915_fpc_translate.c \ i915_surface.c diff --git a/src/mesa/pipe/i915simple/i915_context.c b/src/mesa/pipe/i915simple/i915_context.c index a0ed3032b12..94649231cf3 100644 --- a/src/mesa/pipe/i915simple/i915_context.c +++ b/src/mesa/pipe/i915simple/i915_context.c @@ -29,7 +29,7 @@ #include "i915_winsys.h" #include "i915_state.h" #include "i915_batch.h" -#include "i915_tex_layout.h" +#include "i915_texture.h" #include "i915_reg.h" #include "pipe/draw/draw_context.h" @@ -357,11 +357,8 @@ struct pipe_context *i915_create( struct pipe_winsys *pipe_winsys, i915->pci_id = pci_id; i915->flags.is_i945 = is_i945; - if (i915->flags.is_i945) - i915->pipe.mipmap_tree_layout = i945_miptree_layout; - else - i915->pipe.mipmap_tree_layout = i915_miptree_layout; - + i915->pipe.texture_create = i915_texture_create; + i915->pipe.texture_release = i915_texture_release; i915->dirty = ~0; i915->hardware_dirty = ~0; diff --git a/src/mesa/pipe/i915simple/i915_context.h b/src/mesa/pipe/i915simple/i915_context.h index ee430ebc902..8ed3465be23 100644 --- a/src/mesa/pipe/i915simple/i915_context.h +++ b/src/mesa/pipe/i915simple/i915_context.h @@ -150,6 +150,34 @@ struct i915_alpha_test_state { unsigned LIS6; }; +struct i915_texture { + struct pipe_texture base; + + /* Derived from the above: + */ + unsigned pitch; + unsigned depth_pitch; /* per-image on i945? */ + unsigned total_height; + + unsigned nr_images[PIPE_MAX_TEXTURE_LEVELS]; + + /* Explicitly store the offset of each image for each cube face or + * depth value. Pretty much have to accept that hardware formats + * are going to be so diverse that there is no unified way to + * compute the offsets of depth/cube images within a mipmap level, + * so have to store them as a lookup table: + */ + unsigned *image_offset[PIPE_MAX_TEXTURE_LEVELS]; /**< array [depth] of offsets */ + + /* Includes image offset tables: + */ + unsigned level_offset[PIPE_MAX_TEXTURE_LEVELS]; + + /* The data is held here: + */ + struct pipe_region *region; +}; + struct i915_context { struct pipe_context pipe; @@ -174,7 +202,7 @@ struct i915_context struct pipe_poly_stipple poly_stipple; struct pipe_scissor_state scissor; uint sampler_units[PIPE_MAX_SAMPLERS]; - struct pipe_mipmap_tree *texture[PIPE_MAX_SAMPLERS]; + struct i915_texture *texture[PIPE_MAX_SAMPLERS]; struct pipe_viewport_state viewport; struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX]; diff --git a/src/mesa/pipe/i915simple/i915_state.c b/src/mesa/pipe/i915simple/i915_state.c index 468d0ce91b1..038fd623ea1 100644 --- a/src/mesa/pipe/i915simple/i915_state.c +++ b/src/mesa/pipe/i915simple/i915_state.c @@ -525,11 +525,11 @@ static void i915_set_constant_buffer(struct pipe_context *pipe, static void i915_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_mipmap_tree *texture) + struct pipe_texture *texture) { struct i915_context *i915 = i915_context(pipe); - i915->texture[unit] = texture; /* ptr, not struct */ + i915->texture[unit] = (struct i915_texture*)texture; /* ptr, not struct */ i915->dirty |= I915_NEW_TEXTURE; } diff --git a/src/mesa/pipe/i915simple/i915_state_sampler.c b/src/mesa/pipe/i915simple/i915_state_sampler.c index 0991e6ac0d1..1816d9abdb7 100644 --- a/src/mesa/pipe/i915simple/i915_state_sampler.c +++ b/src/mesa/pipe/i915simple/i915_state_sampler.c @@ -46,9 +46,11 @@ static void update_sampler(struct i915_context *i915, uint unit, const struct i915_sampler_state *sampler, - const struct pipe_mipmap_tree *mt, + const struct i915_texture *tex, unsigned state[3] ) { + const struct pipe_texture *pt = &tex->base; + /* Need to do this after updating the maps, which call the * intel_finalize_mipmap_tree and hence can update firstLevel: */ @@ -56,8 +58,8 @@ static void update_sampler(struct i915_context *i915, state[1] = sampler->state[1]; state[2] = sampler->state[2]; - if (mt->format == PIPE_FORMAT_YCBCR || - mt->format == PIPE_FORMAT_YCBCR_REV) + if (pt->format == PIPE_FORMAT_YCBCR || + pt->format == PIPE_FORMAT_YCBCR_REV) state[0] |= SS2_COLORSPACE_CONVERSION; /* 3D textures don't seem to respect the border color. @@ -75,7 +77,7 @@ static void update_sampler(struct i915_context *i915, const unsigned ws = sampler->templ->wrap_s; const unsigned wt = sampler->templ->wrap_t; const unsigned wr = sampler->templ->wrap_r; - if (mt->target == PIPE_TEXTURE_3D && + if (pt->target == PIPE_TEXTURE_3D && (sampler->templ->min_img_filter != PIPE_TEX_FILTER_NEAREST || sampler->templ->mag_img_filter != PIPE_TEX_FILTER_NEAREST) && (ws == PIPE_TEX_WRAP_CLAMP || @@ -105,13 +107,13 @@ void i915_update_samplers( struct i915_context *i915 ) i915->current.sampler_enable_flags = 0x0; for (unit = 0; unit < I915_TEX_UNITS; unit++) { - /* determine unit enable/disable by looking for a bound mipmap tree */ + /* determine unit enable/disable by looking for a bound texture */ /* could also examine the fragment program? */ if (i915->texture[unit]) { update_sampler( i915, unit, i915->sampler[unit], /* sampler state */ - i915->texture[unit], /* mipmap tree */ + i915->texture[unit], /* texture */ i915->current.sampler[unit] /* the result */ ); @@ -179,18 +181,19 @@ static void i915_update_texture(struct i915_context *i915, uint unit, uint state[6]) { - const struct pipe_mipmap_tree *mt = i915->texture[unit]; + const struct i915_texture *tex = i915->texture[unit]; + const struct pipe_texture *pt = &tex->base; uint format, pitch; - const uint width = mt->width0, height = mt->height0, depth = mt->depth0; - const uint num_levels = mt->last_level - mt->first_level; + const uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + const uint num_levels = pt->last_level - pt->first_level; - assert(mt); + assert(tex); assert(width); assert(height); assert(depth); - format = translate_texture_format(mt->format); - pitch = mt->pitch * mt->cpp; + format = translate_texture_format(pt->format); + pitch = tex->pitch * pt->cpp; assert(format); assert(pitch); @@ -217,7 +220,7 @@ i915_update_textures(struct i915_context *i915) uint unit; for (unit = 0; unit < I915_TEX_UNITS; unit++) { - /* determine unit enable/disable by looking for a bound mipmap tree */ + /* determine unit enable/disable by looking for a bound texture */ /* could also examine the fragment program? */ if (i915->texture[unit]) { i915_update_texture(i915, unit, i915->current.texbuffer[unit]); diff --git a/src/mesa/pipe/i915simple/i915_surface.c b/src/mesa/pipe/i915simple/i915_surface.c index e4a5de00d77..385202507db 100644 --- a/src/mesa/pipe/i915simple/i915_surface.c +++ b/src/mesa/pipe/i915simple/i915_surface.c @@ -187,34 +187,35 @@ i915_put_tile(struct pipe_context *pipe, */ static struct pipe_surface * i915_get_tex_surface(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt, + struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { + struct i915_texture *tex = (struct i915_texture *)pt; struct pipe_surface *ps; unsigned offset; /* in bytes */ - offset = mt->level[level].level_offset; + offset = tex->level_offset[level]; - if (mt->target == PIPE_TEXTURE_CUBE) { - offset += mt->level[level].image_offset[face] * mt->cpp; + if (pt->target == PIPE_TEXTURE_CUBE) { + offset += tex->image_offset[level][face] * pt->cpp; } - else if (mt->target == PIPE_TEXTURE_3D) { - offset += mt->level[level].image_offset[zslice] * mt->cpp; + else if (pt->target == PIPE_TEXTURE_3D) { + offset += tex->image_offset[level][zslice] * pt->cpp; } else { assert(face == 0); assert(zslice == 0); } - ps = pipe->winsys->surface_alloc(pipe->winsys, mt->format); + ps = pipe->winsys->surface_alloc(pipe->winsys, pt->format); if (ps) { assert(ps->format); assert(ps->refcount); - pipe_region_reference(&ps->region, mt->region); - ps->cpp = mt->cpp; - ps->width = mt->level[level].width; - ps->height = mt->level[level].height; - ps->pitch = mt->pitch; + pipe_region_reference(&ps->region, tex->region); + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = tex->pitch; ps->offset = offset; } return ps; diff --git a/src/mesa/pipe/i915simple/i915_tex_layout.c b/src/mesa/pipe/i915simple/i915_tex_layout.c deleted file mode 100644 index cb372a7170d..00000000000 --- a/src/mesa/pipe/i915simple/i915_tex_layout.c +++ /dev/null @@ -1,474 +0,0 @@ -/************************************************************************** - * - * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - /* - * Authors: - * Keith Whitwell - * Michel Dänzer - */ - -#include "pipe/p_state.h" -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" - -#include "i915_tex_layout.h" -#include "i915_debug.h" - - -static unsigned minify( unsigned d ) -{ - return MAX2(1, d>>1); -} - -static int align(int value, int alignment) -{ - return (value + alignment - 1) & ~(alignment - 1); -} - - -static void -i915_miptree_set_level_info(struct pipe_mipmap_tree *mt, - unsigned level, - unsigned nr_images, - unsigned x, unsigned y, unsigned w, unsigned h, unsigned d) -{ - assert(level < PIPE_MAX_TEXTURE_LEVELS); - - mt->level[level].width = w; - mt->level[level].height = h; - mt->level[level].depth = d; - mt->level[level].level_offset = (x + y * mt->pitch) * mt->cpp; - mt->level[level].nr_images = nr_images; - - /* - DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, - level, w, h, d, x, y, mt->level[level].level_offset); - */ - - /* Not sure when this would happen, but anyway: - */ - if (mt->level[level].image_offset) { - FREE(mt->level[level].image_offset); - mt->level[level].image_offset = NULL; - } - - assert(nr_images); - assert(!mt->level[level].image_offset); - - mt->level[level].image_offset = (unsigned *) MALLOC(nr_images * sizeof(unsigned)); - mt->level[level].image_offset[0] = 0; -} - - -static void -i915_miptree_set_image_offset(struct pipe_mipmap_tree *mt, - unsigned level, unsigned img, unsigned x, unsigned y) -{ - if (img == 0 && level == 0) - assert(x == 0 && y == 0); - - assert(img < mt->level[level].nr_images); - - mt->level[level].image_offset[img] = (x + y * mt->pitch); - - /* - DBG("%s level %d img %d pos %d,%d image_offset %x\n", - __FUNCTION__, level, img, x, y, mt->level[level].image_offset[img]); - */ -} - - -static void -i945_miptree_layout_2d( struct pipe_mipmap_tree *mt ) -{ - int align_h = 2, align_w = 4; - unsigned level; - unsigned x = 0; - unsigned y = 0; - unsigned width = mt->width0; - unsigned height = mt->height0; - - mt->pitch = mt->width0; - - /* May need to adjust pitch to accomodate the placement of - * the 2nd mipmap. This occurs when the alignment - * constraints of mipmap placement push the right edge of the - * 2nd mipmap out past the width of its parent. - */ - if (mt->first_level != mt->last_level) { - unsigned mip1_width = align(minify(mt->width0), align_w) - + minify(minify(mt->width0)); - - if (mip1_width > mt->width0) - mt->pitch = mip1_width; - } - - /* Pitch must be a whole number of dwords, even though we - * express it in texels. - */ - mt->pitch = align(mt->pitch * mt->cpp, 4) / mt->cpp; - mt->total_height = 0; - - for ( level = mt->first_level ; level <= mt->last_level ; level++ ) { - unsigned img_height; - - i915_miptree_set_level_info(mt, level, 1, x, y, width, height, 1); - - if (mt->compressed) - img_height = MAX2(1, height/4); - else - img_height = align(height, align_h); - - - /* Because the images are packed better, the final offset - * might not be the maximal one: - */ - mt->total_height = MAX2(mt->total_height, y + img_height); - - /* Layout_below: step right after second mipmap. - */ - if (level == mt->first_level + 1) { - x += align(width, align_w); - } - else { - y += img_height; - } - - width = minify(width); - height = minify(height); - } -} - - -static const int initial_offsets[6][2] = { - {0, 0}, - {0, 2}, - {1, 0}, - {1, 2}, - {1, 1}, - {1, 3} -}; - -static const int step_offsets[6][2] = { - {0, 2}, - {0, 2}, - {-1, 2}, - {-1, 2}, - {-1, 1}, - {-1, 1} -}; - - -boolean -i915_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree * mt) -{ - unsigned level; - - switch (mt->target) { - case PIPE_TEXTURE_CUBE: { - const unsigned dim = mt->width0; - unsigned face; - unsigned lvlWidth = mt->width0, lvlHeight = mt->height0; - - assert(lvlWidth == lvlHeight); /* cubemap images are square */ - - /* double pitch for cube layouts */ - mt->pitch = ((dim * mt->cpp * 2 + 3) & ~3) / mt->cpp; - mt->total_height = dim * 4; - - for (level = mt->first_level; level <= mt->last_level; level++) { - i915_miptree_set_level_info(mt, level, 6, - 0, 0, - /*OLD: mt->pitch, mt->total_height,*/ - lvlWidth, lvlHeight, - 1); - lvlWidth /= 2; - lvlHeight /= 2; - } - - for (face = 0; face < 6; face++) { - unsigned x = initial_offsets[face][0] * dim; - unsigned y = initial_offsets[face][1] * dim; - unsigned d = dim; - - for (level = mt->first_level; level <= mt->last_level; level++) { - i915_miptree_set_image_offset(mt, level, face, x, y); - d >>= 1; - x += step_offsets[face][0] * d; - y += step_offsets[face][1] * d; - } - } - break; - } - case PIPE_TEXTURE_3D:{ - unsigned width = mt->width0; - unsigned height = mt->height0; - unsigned depth = mt->depth0; - unsigned stack_height = 0; - - /* Calculate the size of a single slice. - */ - mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp; - - /* XXX: hardware expects/requires 9 levels at minimum. - */ - for (level = mt->first_level; level <= MAX2(8, mt->last_level); - level++) { - i915_miptree_set_level_info(mt, level, depth, 0, mt->total_height, - width, height, depth); - - - stack_height += MAX2(2, height); - - width = minify(width); - height = minify(height); - depth = minify(depth); - } - - /* Fixup depth image_offsets: - */ - depth = mt->depth0; - for (level = mt->first_level; level <= mt->last_level; level++) { - unsigned i; - for (i = 0; i < depth; i++) - i915_miptree_set_image_offset(mt, level, i, - 0, i * stack_height); - - depth = minify(depth); - } - - - /* Multiply slice size by texture depth for total size. It's - * remarkable how wasteful of memory the i915 texture layouts - * are. They are largely fixed in the i945. - */ - mt->total_height = stack_height * mt->depth0; - break; - } - - default:{ - unsigned width = mt->width0; - unsigned height = mt->height0; - unsigned img_height; - - mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp; - mt->total_height = 0; - - for (level = mt->first_level; level <= mt->last_level; level++) { - i915_miptree_set_level_info(mt, level, 1, - 0, mt->total_height, - width, height, 1); - - if (mt->compressed) - img_height = MAX2(1, height / 4); - else - img_height = (MAX2(2, height) + 1) & ~1; - - mt->total_height += img_height; - - width = minify(width); - height = minify(height); - } - break; - } - } - /* - DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, - mt->pitch, - mt->total_height, mt->cpp, mt->pitch * mt->total_height * mt->cpp); - */ - - return TRUE; -} - - -boolean -i945_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree * mt) -{ - unsigned level; - - switch (mt->target) { - case PIPE_TEXTURE_CUBE:{ - const unsigned dim = mt->width0; - unsigned face; - unsigned lvlWidth = mt->width0, lvlHeight = mt->height0; - - assert(lvlWidth == lvlHeight); /* cubemap images are square */ - - /* Depending on the size of the largest images, pitch can be - * determined either by the old-style packing of cubemap faces, - * or the final row of 4x4, 2x2 and 1x1 faces below this. - */ - if (dim > 32) - mt->pitch = ((dim * mt->cpp * 2 + 3) & ~3) / mt->cpp; - else - mt->pitch = 14 * 8; - - mt->total_height = dim * 4 + 4; - - /* Set all the levels to effectively occupy the whole rectangular region. - */ - for (level = mt->first_level; level <= mt->last_level; level++) { - i915_miptree_set_level_info(mt, level, 6, - 0, 0, - lvlWidth, lvlHeight, 1); - lvlWidth /= 2; - lvlHeight /= 2; - } - - - for (face = 0; face < 6; face++) { - unsigned x = initial_offsets[face][0] * dim; - unsigned y = initial_offsets[face][1] * dim; - unsigned d = dim; - - if (dim == 4 && face >= 4) { - y = mt->total_height - 4; - x = (face - 4) * 8; - } - else if (dim < 4 && (face > 0 || mt->first_level > 0)) { - y = mt->total_height - 4; - x = face * 8; - } - - for (level = mt->first_level; level <= mt->last_level; level++) { - i915_miptree_set_image_offset(mt, level, face, x, y); - - d >>= 1; - - switch (d) { - case 4: - switch (face) { - case PIPE_TEX_FACE_POS_X: - case PIPE_TEX_FACE_NEG_X: - x += step_offsets[face][0] * d; - y += step_offsets[face][1] * d; - break; - case PIPE_TEX_FACE_POS_Y: - case PIPE_TEX_FACE_NEG_Y: - y += 12; - x -= 8; - break; - case PIPE_TEX_FACE_POS_Z: - case PIPE_TEX_FACE_NEG_Z: - y = mt->total_height - 4; - x = (face - 4) * 8; - break; - } - - case 2: - y = mt->total_height - 4; - x = 16 + face * 8; - break; - - case 1: - x += 48; - break; - - default: - x += step_offsets[face][0] * d; - y += step_offsets[face][1] * d; - break; - } - } - } - break; - } - case PIPE_TEXTURE_3D:{ - unsigned width = mt->width0; - unsigned height = mt->height0; - unsigned depth = mt->depth0; - unsigned pack_x_pitch, pack_x_nr; - unsigned pack_y_pitch; - unsigned level; - - mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp; - mt->total_height = 0; - - pack_y_pitch = MAX2(mt->height0, 2); - pack_x_pitch = mt->pitch; - pack_x_nr = 1; - - for (level = mt->first_level; level <= mt->last_level; level++) { - unsigned nr_images = mt->target == PIPE_TEXTURE_3D ? depth : 6; - int x = 0; - int y = 0; - unsigned q, j; - - i915_miptree_set_level_info(mt, level, nr_images, - 0, mt->total_height, - width, height, depth); - - for (q = 0; q < nr_images;) { - for (j = 0; j < pack_x_nr && q < nr_images; j++, q++) { - i915_miptree_set_image_offset(mt, level, q, x, y); - x += pack_x_pitch; - } - - x = 0; - y += pack_y_pitch; - } - - - mt->total_height += y; - - if (pack_x_pitch > 4) { - pack_x_pitch >>= 1; - pack_x_nr <<= 1; - assert(pack_x_pitch * pack_x_nr <= mt->pitch); - } - - if (pack_y_pitch > 2) { - pack_y_pitch >>= 1; - } - - width = minify(width); - height = minify(height); - depth = minify(depth); - } - break; - } - - case PIPE_TEXTURE_1D: - case PIPE_TEXTURE_2D: -// case PIPE_TEXTURE_RECTANGLE: - i945_miptree_layout_2d(mt); - break; - default: - assert(0); - return FALSE; - } - - /* - DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, - mt->pitch, - mt->total_height, mt->cpp, mt->pitch * mt->total_height * mt->cpp); - */ - - return TRUE; -} - diff --git a/src/mesa/pipe/i915simple/i915_tex_layout.h b/src/mesa/pipe/i915simple/i915_tex_layout.h deleted file mode 100644 index e0337863815..00000000000 --- a/src/mesa/pipe/i915simple/i915_tex_layout.h +++ /dev/null @@ -1,16 +0,0 @@ - -#ifndef I915_TEX_LAYOUT_H -#define I915_TEX_LAYOUT_H - -struct pipe_context; -struct pipe_mipmap_tree; - - -extern boolean -i915_miptree_layout(struct pipe_context *, struct pipe_mipmap_tree *); - -extern boolean -i945_miptree_layout(struct pipe_context *, struct pipe_mipmap_tree *); - - -#endif /* I915_TEX_LAYOUT_H */ diff --git a/src/mesa/pipe/i915simple/i915_texture.c b/src/mesa/pipe/i915simple/i915_texture.c new file mode 100644 index 00000000000..3bfa806d9ec --- /dev/null +++ b/src/mesa/pipe/i915simple/i915_texture.c @@ -0,0 +1,539 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + /* + * Authors: + * Keith Whitwell + * Michel Dänzer + */ + +#include "pipe/p_state.h" +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" + +#include "i915_context.h" +#include "i915_texture.h" +#include "i915_debug.h" + + +static unsigned minify( unsigned d ) +{ + return MAX2(1, d>>1); +} + +static int align(int value, int alignment) +{ + return (value + alignment - 1) & ~(alignment - 1); +} + + +static void +i915_miptree_set_level_info(struct i915_texture *tex, + unsigned level, + unsigned nr_images, + unsigned x, unsigned y, unsigned w, unsigned h, unsigned d) +{ + struct pipe_texture *pt = &tex->base; + + assert(level < PIPE_MAX_TEXTURE_LEVELS); + + pt->width[level] = w; + pt->height[level] = h; + pt->depth[level] = d; + + tex->level_offset[level] = (x + y * tex->pitch) * pt->cpp; + tex->nr_images[level] = nr_images; + + /* + DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, + level, w, h, d, x, y, tex->level_offset[level]); + */ + + /* Not sure when this would happen, but anyway: + */ + if (tex->image_offset[level]) { + FREE(tex->image_offset[level]); + tex->image_offset[level] = NULL; + } + + assert(nr_images); + assert(!tex->image_offset[level]); + + tex->image_offset[level] = (unsigned *) MALLOC(nr_images * sizeof(unsigned)); + tex->image_offset[level][0] = 0; +} + + +static void +i915_miptree_set_image_offset(struct i915_texture *tex, + unsigned level, unsigned img, unsigned x, unsigned y) +{ + if (img == 0 && level == 0) + assert(x == 0 && y == 0); + + assert(img < tex->nr_images[level]); + + tex->image_offset[level][img] = (x + y * tex->pitch); + + /* + DBG("%s level %d img %d pos %d,%d image_offset %x\n", + __FUNCTION__, level, img, x, y, tex->image_offset[level][img]); + */ +} + + +static void +i945_miptree_layout_2d( struct i915_texture *tex ) +{ + struct pipe_texture *pt = &tex->base; + int align_h = 2, align_w = 4; + unsigned level; + unsigned x = 0; + unsigned y = 0; + unsigned width = pt->width[0]; + unsigned height = pt->height[0]; + + tex->pitch = pt->width[0]; + + /* May need to adjust pitch to accomodate the placement of + * the 2nd mipmap. This occurs when the alignment + * constraints of mipmap placement push the right edge of the + * 2nd mipmap out past the width of its parent. + */ + if (pt->first_level != pt->last_level) { + unsigned mip1_width = align(minify(pt->width[0]), align_w) + + minify(minify(pt->width[0])); + + if (mip1_width > pt->width[0]) + tex->pitch = mip1_width; + } + + /* Pitch must be a whole number of dwords, even though we + * express it in texels. + */ + tex->pitch = align(tex->pitch * pt->cpp, 4) / pt->cpp; + tex->total_height = 0; + + for ( level = pt->first_level ; level <= pt->last_level ; level++ ) { + unsigned img_height; + + i915_miptree_set_level_info(tex, level, 1, x, y, width, height, 1); + + if (pt->compressed) + img_height = MAX2(1, height/4); + else + img_height = align(height, align_h); + + + /* Because the images are packed better, the final offset + * might not be the maximal one: + */ + tex->total_height = MAX2(tex->total_height, y + img_height); + + /* Layout_below: step right after second mipmap. + */ + if (level == pt->first_level + 1) { + x += align(width, align_w); + } + else { + y += img_height; + } + + width = minify(width); + height = minify(height); + } +} + + +static const int initial_offsets[6][2] = { + {0, 0}, + {0, 2}, + {1, 0}, + {1, 2}, + {1, 1}, + {1, 3} +}; + +static const int step_offsets[6][2] = { + {0, 2}, + {0, 2}, + {-1, 2}, + {-1, 2}, + {-1, 1}, + {-1, 1} +}; + + +static boolean +i915_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) +{ + struct pipe_texture *pt = &tex->base; + unsigned level; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE: { + const unsigned dim = pt->width[0]; + unsigned face; + unsigned lvlWidth = pt->width[0], lvlHeight = pt->height[0]; + + assert(lvlWidth == lvlHeight); /* cubemap images are square */ + + /* double pitch for cube layouts */ + tex->pitch = ((dim * pt->cpp * 2 + 3) & ~3) / pt->cpp; + tex->total_height = dim * 4; + + for (level = pt->first_level; level <= pt->last_level; level++) { + i915_miptree_set_level_info(tex, level, 6, + 0, 0, + /*OLD: tex->pitch, tex->total_height,*/ + lvlWidth, lvlHeight, + 1); + lvlWidth /= 2; + lvlHeight /= 2; + } + + for (face = 0; face < 6; face++) { + unsigned x = initial_offsets[face][0] * dim; + unsigned y = initial_offsets[face][1] * dim; + unsigned d = dim; + + for (level = pt->first_level; level <= pt->last_level; level++) { + i915_miptree_set_image_offset(tex, level, face, x, y); + d >>= 1; + x += step_offsets[face][0] * d; + y += step_offsets[face][1] * d; + } + } + break; + } + case PIPE_TEXTURE_3D:{ + unsigned width = pt->width[0]; + unsigned height = pt->height[0]; + unsigned depth = pt->depth[0]; + unsigned stack_height = 0; + + /* Calculate the size of a single slice. + */ + tex->pitch = ((pt->width[0] * pt->cpp + 3) & ~3) / pt->cpp; + + /* XXX: hardware expects/requires 9 levels at minimum. + */ + for (level = pt->first_level; level <= MAX2(8, pt->last_level); + level++) { + i915_miptree_set_level_info(tex, level, depth, 0, tex->total_height, + width, height, depth); + + + stack_height += MAX2(2, height); + + width = minify(width); + height = minify(height); + depth = minify(depth); + } + + /* Fixup depth image_offsets: + */ + depth = pt->depth[0]; + for (level = pt->first_level; level <= pt->last_level; level++) { + unsigned i; + for (i = 0; i < depth; i++) + i915_miptree_set_image_offset(tex, level, i, + 0, i * stack_height); + + depth = minify(depth); + } + + + /* Multiply slice size by texture depth for total size. It's + * remarkable how wasteful of memory the i915 texture layouts + * are. They are largely fixed in the i945. + */ + tex->total_height = stack_height * pt->depth[0]; + break; + } + + default:{ + unsigned width = pt->width[0]; + unsigned height = pt->height[0]; + unsigned img_height; + + tex->pitch = ((pt->width[0] * pt->cpp + 3) & ~3) / pt->cpp; + tex->total_height = 0; + + for (level = pt->first_level; level <= pt->last_level; level++) { + i915_miptree_set_level_info(tex, level, 1, + 0, tex->total_height, + width, height, 1); + + if (pt->compressed) + img_height = MAX2(1, height / 4); + else + img_height = (MAX2(2, height) + 1) & ~1; + + tex->total_height += img_height; + + width = minify(width); + height = minify(height); + } + break; + } + } + /* + DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, + tex->pitch, + tex->total_height, pt->cpp, tex->pitch * tex->total_height * pt->cpp); + */ + + return TRUE; +} + + +static boolean +i945_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) +{ + struct pipe_texture *pt = &tex->base; + unsigned level; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE:{ + const unsigned dim = pt->width[0]; + unsigned face; + unsigned lvlWidth = pt->width[0], lvlHeight = pt->height[0]; + + assert(lvlWidth == lvlHeight); /* cubemap images are square */ + + /* Depending on the size of the largest images, pitch can be + * determined either by the old-style packing of cubemap faces, + * or the final row of 4x4, 2x2 and 1x1 faces below this. + */ + if (dim > 32) + tex->pitch = ((dim * pt->cpp * 2 + 3) & ~3) / pt->cpp; + else + tex->pitch = 14 * 8; + + tex->total_height = dim * 4 + 4; + + /* Set all the levels to effectively occupy the whole rectangular region. + */ + for (level = pt->first_level; level <= pt->last_level; level++) { + i915_miptree_set_level_info(tex, level, 6, + 0, 0, + lvlWidth, lvlHeight, 1); + lvlWidth /= 2; + lvlHeight /= 2; + } + + + for (face = 0; face < 6; face++) { + unsigned x = initial_offsets[face][0] * dim; + unsigned y = initial_offsets[face][1] * dim; + unsigned d = dim; + + if (dim == 4 && face >= 4) { + y = tex->total_height - 4; + x = (face - 4) * 8; + } + else if (dim < 4 && (face > 0 || pt->first_level > 0)) { + y = tex->total_height - 4; + x = face * 8; + } + + for (level = pt->first_level; level <= pt->last_level; level++) { + i915_miptree_set_image_offset(tex, level, face, x, y); + + d >>= 1; + + switch (d) { + case 4: + switch (face) { + case PIPE_TEX_FACE_POS_X: + case PIPE_TEX_FACE_NEG_X: + x += step_offsets[face][0] * d; + y += step_offsets[face][1] * d; + break; + case PIPE_TEX_FACE_POS_Y: + case PIPE_TEX_FACE_NEG_Y: + y += 12; + x -= 8; + break; + case PIPE_TEX_FACE_POS_Z: + case PIPE_TEX_FACE_NEG_Z: + y = tex->total_height - 4; + x = (face - 4) * 8; + break; + } + + case 2: + y = tex->total_height - 4; + x = 16 + face * 8; + break; + + case 1: + x += 48; + break; + + default: + x += step_offsets[face][0] * d; + y += step_offsets[face][1] * d; + break; + } + } + } + break; + } + case PIPE_TEXTURE_3D:{ + unsigned width = pt->width[0]; + unsigned height = pt->height[0]; + unsigned depth = pt->depth[0]; + unsigned pack_x_pitch, pack_x_nr; + unsigned pack_y_pitch; + unsigned level; + + tex->pitch = ((pt->width[0] * pt->cpp + 3) & ~3) / pt->cpp; + tex->total_height = 0; + + pack_y_pitch = MAX2(pt->height[0], 2); + pack_x_pitch = tex->pitch; + pack_x_nr = 1; + + for (level = pt->first_level; level <= pt->last_level; level++) { + unsigned nr_images = pt->target == PIPE_TEXTURE_3D ? depth : 6; + int x = 0; + int y = 0; + unsigned q, j; + + i915_miptree_set_level_info(tex, level, nr_images, + 0, tex->total_height, + width, height, depth); + + for (q = 0; q < nr_images;) { + for (j = 0; j < pack_x_nr && q < nr_images; j++, q++) { + i915_miptree_set_image_offset(tex, level, q, x, y); + x += pack_x_pitch; + } + + x = 0; + y += pack_y_pitch; + } + + + tex->total_height += y; + + if (pack_x_pitch > 4) { + pack_x_pitch >>= 1; + pack_x_nr <<= 1; + assert(pack_x_pitch * pack_x_nr <= tex->pitch); + } + + if (pack_y_pitch > 2) { + pack_y_pitch >>= 1; + } + + width = minify(width); + height = minify(height); + depth = minify(depth); + } + break; + } + + case PIPE_TEXTURE_1D: + case PIPE_TEXTURE_2D: +// case PIPE_TEXTURE_RECTANGLE: + i945_miptree_layout_2d(tex); + break; + default: + assert(0); + return FALSE; + } + + /* + DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, + tex->pitch, + tex->total_height, pt->cpp, tex->pitch * tex->total_height * pt->cpp); + */ + + return TRUE; +} + +void +i915_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct i915_texture *tex = REALLOC(*pt, sizeof(struct pipe_texture), + sizeof(struct i915_texture)); + + if (tex) { + struct i915_context *i915 = i915_context(pipe); + + memset(&tex->base + 1, 0, + sizeof(struct i915_texture) - sizeof(struct pipe_texture)); + + if (i915->flags.is_i945 ? i945_miptree_layout(pipe, tex) : + i915_miptree_layout(pipe, tex)) { + tex->region = pipe->winsys->region_alloc(pipe->winsys, + tex->pitch * tex->base.cpp * + tex->total_height, + PIPE_SURFACE_FLAG_TEXTURE); + } + + if (!tex->region) { + FREE(tex); + tex = NULL; + } + } + + *pt = &tex->base; +} + +void +i915_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + if (!*pt) + return; + + /* + DBG("%s %p refcount will be %d\n", + __FUNCTION__, (void *) *pt, (*pt)->refcount - 1); + */ + if (--(*pt)->refcount <= 0) { + struct i915_texture *tex = (struct i915_texture *)*pt; + uint i; + + /* + DBG("%s deleting %p\n", __FUNCTION__, (void *) tex); + */ + + pipe->winsys->region_release(pipe->winsys, &tex->region); + + for (i = 0; i < PIPE_MAX_TEXTURE_LEVELS; i++) + if (tex->image_offset[i]) + free(tex->image_offset[i]); + + free(tex); + } + *pt = NULL; +} diff --git a/src/mesa/pipe/i915simple/i915_texture.h b/src/mesa/pipe/i915simple/i915_texture.h new file mode 100644 index 00000000000..84a0502e81a --- /dev/null +++ b/src/mesa/pipe/i915simple/i915_texture.h @@ -0,0 +1,16 @@ + +#ifndef I915_TEXTURE_H +#define I915_TEXTURE_H + +struct pipe_context; +struct pipe_texture; + + +extern void +i915_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); + +extern void +i915_texture_release(struct pipe_context *pipe, struct pipe_texture **pt); + + +#endif /* I915_TEXTURE_H */ diff --git a/src/mesa/pipe/llvm/llvm_entry.c b/src/mesa/pipe/llvm/llvm_entry.c index 2459d14cb82..fe32e7810dc 100644 --- a/src/mesa/pipe/llvm/llvm_entry.c +++ b/src/mesa/pipe/llvm/llvm_entry.c @@ -194,7 +194,6 @@ void run_vertex_shader(float (*ainputs)[16][4], struct pipe_sampler_state; -struct pipe_mipmap_tree; struct softpipe_tile_cache; #define NUM_CHANNELS 4 /* R,G,B,A */ @@ -203,7 +202,6 @@ struct softpipe_tile_cache; struct tgsi_sampler { const struct pipe_sampler_state *state; - struct pipe_mipmap_tree *texture; /** Get samples for four fragments in a quad */ void (*get_samples)(struct tgsi_sampler *sampler, const float s[QUAD_SIZE], diff --git a/src/mesa/pipe/p_context.h b/src/mesa/pipe/p_context.h index e145b22f2ff..5033209323f 100644 --- a/src/mesa/pipe/p_context.h +++ b/src/mesa/pipe/p_context.h @@ -155,7 +155,7 @@ struct pipe_context { void (*set_texture_state)( struct pipe_context *, unsigned unit, - struct pipe_mipmap_tree * ); + struct pipe_texture * ); void (*set_viewport_state)( struct pipe_context *, const struct pipe_viewport_state * ); @@ -180,7 +180,7 @@ struct pipe_context { /** Get a surface which is a "view" into a texture */ struct pipe_surface *(*get_tex_surface)(struct pipe_context *pipe, - struct pipe_mipmap_tree *texture, + struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice); @@ -237,8 +237,11 @@ struct pipe_context { /* * Texture functions */ - boolean (*mipmap_tree_layout)( struct pipe_context *pipe, - struct pipe_mipmap_tree *mt ); + void (*texture_create)(struct pipe_context *pipe, + struct pipe_texture **pt); + + void (*texture_release)(struct pipe_context *pipe, + struct pipe_texture **pt); /* Flush rendering: diff --git a/src/mesa/pipe/p_inlines.h b/src/mesa/pipe/p_inlines.h index 2418d016e15..c04d46dddd8 100644 --- a/src/mesa/pipe/p_inlines.h +++ b/src/mesa/pipe/p_inlines.h @@ -80,4 +80,24 @@ pipe_surface_reference(struct pipe_surface **ptr, struct pipe_surface *surf) } } + +/** + * \sa pipe_region_reference + */ +static INLINE void +pipe_texture_reference(struct pipe_context *pipe, struct pipe_texture **ptr, + struct pipe_texture *pt) +{ + assert(ptr); + if (*ptr) { + pipe->texture_release(pipe, ptr); + assert(!*ptr); + } + if (pt) { + /* reference the new thing */ + pt->refcount++; + *ptr = pt; + } +} + #endif /* P_INLINES_H */ diff --git a/src/mesa/pipe/p_state.h b/src/mesa/pipe/p_state.h index 642734aeb80..077a8f5a06f 100644 --- a/src/mesa/pipe/p_state.h +++ b/src/mesa/pipe/p_state.h @@ -288,27 +288,11 @@ struct pipe_surface /** - * Describes the location of each texture image within a texture region. + * Texture. Represents one or several texture images on one or several mipmap + * levels. */ -struct pipe_mipmap_level -{ - unsigned level_offset; - unsigned width; - unsigned height; - unsigned depth; - unsigned nr_images; - - /* Explicitly store the offset of each image for each cube face or - * depth value. Pretty much have to accept that hardware formats - * are going to be so diverse that there is no unified way to - * compute the offsets of depth/cube images within a mipmap level, - * so have to store them as a lookup table: - */ - unsigned *image_offset; /**< array [depth] of offsets */ -}; - -struct pipe_mipmap_tree -{ +struct pipe_texture +{ /* Effectively the key: */ unsigned target; /* XXX convert to PIPE_TEXTURE_x */ @@ -318,25 +302,13 @@ struct pipe_mipmap_tree unsigned first_level; unsigned last_level; - unsigned width0, height0, depth0; /**< Level zero image dimensions */ + unsigned width[PIPE_MAX_TEXTURE_LEVELS]; + unsigned height[PIPE_MAX_TEXTURE_LEVELS]; + unsigned depth[PIPE_MAX_TEXTURE_LEVELS]; unsigned cpp; unsigned compressed:1; - /* Derived from the above: - */ - unsigned pitch; - unsigned depth_pitch; /* per-image on i945? */ - unsigned total_height; - - /* Includes image offset tables: - */ - struct pipe_mipmap_level level[PIPE_MAX_TEXTURE_LEVELS]; - - /* The data is held here: - */ - struct pipe_region *region; - /* These are also refcounted: */ unsigned refcount; diff --git a/src/mesa/pipe/softpipe/Makefile b/src/mesa/pipe/softpipe/Makefile index 59628531cc8..9978884c9bd 100644 --- a/src/mesa/pipe/softpipe/Makefile +++ b/src/mesa/pipe/softpipe/Makefile @@ -34,7 +34,7 @@ DRIVER_SOURCES = \ sp_state_rasterizer.c \ sp_state_surface.c \ sp_state_vertex.c \ - sp_tex_layout.c \ + sp_texture.c \ sp_tex_sample.c \ sp_tile_cache.c \ sp_surface.c diff --git a/src/mesa/pipe/softpipe/sp_context.c b/src/mesa/pipe/softpipe/sp_context.c index d5e68c189de..7a9fccce9ae 100644 --- a/src/mesa/pipe/softpipe/sp_context.c +++ b/src/mesa/pipe/softpipe/sp_context.c @@ -40,7 +40,7 @@ #include "sp_state.h" #include "sp_surface.h" #include "sp_tile_cache.h" -#include "sp_tex_layout.h" +#include "sp_texture.h" #include "sp_winsys.h" @@ -98,9 +98,9 @@ softpipe_map_texture_surfaces(struct softpipe_context *sp) uint i; for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - struct pipe_mipmap_tree *mt = sp->texture[i]; - if (mt) { - pipe->region_map(pipe, mt->region); + struct softpipe_texture *spt = sp->texture[i]; + if (spt) { + pipe->region_map(pipe, spt->region); } } } @@ -146,9 +146,9 @@ softpipe_unmap_texture_surfaces(struct softpipe_context *sp) struct pipe_context *pipe = &sp->pipe; uint i; for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - struct pipe_mipmap_tree *mt = sp->texture[i]; - if (mt) { - pipe->region_unmap(pipe, mt->region); + struct softpipe_texture *spt = sp->texture[i]; + if (spt) { + pipe->region_unmap(pipe, spt->region); } } } @@ -351,7 +351,8 @@ struct pipe_context *softpipe_create( struct pipe_winsys *pipe_winsys, softpipe->pipe.get_vendor = softpipe_get_vendor; /* textures */ - softpipe->pipe.mipmap_tree_layout = softpipe_mipmap_tree_layout; + softpipe->pipe.texture_create = softpipe_texture_create; + softpipe->pipe.texture_release = softpipe_texture_release; softpipe->pipe.get_tex_surface = softpipe_get_tex_surface; /* diff --git a/src/mesa/pipe/softpipe/sp_context.h b/src/mesa/pipe/softpipe/sp_context.h index 872766101d5..d4763a98c66 100644 --- a/src/mesa/pipe/softpipe/sp_context.h +++ b/src/mesa/pipe/softpipe/sp_context.h @@ -89,7 +89,7 @@ struct softpipe_context { struct pipe_framebuffer_state framebuffer; struct pipe_poly_stipple poly_stipple; struct pipe_scissor_state scissor; - struct pipe_mipmap_tree *texture[PIPE_MAX_SAMPLERS]; + struct softpipe_texture *texture[PIPE_MAX_SAMPLERS]; struct pipe_viewport_state viewport; struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX]; diff --git a/src/mesa/pipe/softpipe/sp_quad_fs.c b/src/mesa/pipe/softpipe/sp_quad_fs.c index 24c8a44c479..7184fcda521 100644 --- a/src/mesa/pipe/softpipe/sp_quad_fs.c +++ b/src/mesa/pipe/softpipe/sp_quad_fs.c @@ -277,7 +277,7 @@ static void shade_begin(struct quad_stage *qs) /* set TGSI sampler state that varies */ for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { qss->samplers[i].state = softpipe->sampler[i]; - qss->samplers[i].texture = softpipe->texture[i]; + qss->samplers[i].texture = &softpipe->texture[i]->base; } #ifdef MESA_LLVM diff --git a/src/mesa/pipe/softpipe/sp_state.h b/src/mesa/pipe/softpipe/sp_state.h index 2f096a9cc97..a543735b525 100644 --- a/src/mesa/pipe/softpipe/sp_state.h +++ b/src/mesa/pipe/softpipe/sp_state.h @@ -52,6 +52,35 @@ struct sp_fragment_shader_state { #endif }; +struct softpipe_texture +{ + struct pipe_texture base; + + /* Derived from the above: + */ + unsigned pitch; + unsigned depth_pitch; /* per-image on i945? */ + unsigned total_height; + + unsigned nr_images[PIPE_MAX_TEXTURE_LEVELS]; + + /* Explicitly store the offset of each image for each cube face or + * depth value. Pretty much have to accept that hardware formats + * are going to be so diverse that there is no unified way to + * compute the offsets of depth/cube images within a mipmap level, + * so have to store them as a lookup table: + */ + unsigned *image_offset[PIPE_MAX_TEXTURE_LEVELS]; /**< array [depth] of offsets */ + + /* Includes image offset tables: + */ + unsigned long level_offset[PIPE_MAX_TEXTURE_LEVELS]; + + /* The data is held here: + */ + struct pipe_region *region; +}; + void * softpipe_create_alpha_test_state(struct pipe_context *, const struct pipe_alpha_test_state *); @@ -125,7 +154,7 @@ void softpipe_set_scissor_state( struct pipe_context *, void softpipe_set_texture_state( struct pipe_context *, unsigned unit, - struct pipe_mipmap_tree * ); + struct pipe_texture * ); void softpipe_set_viewport_state( struct pipe_context *, const struct pipe_viewport_state * ); diff --git a/src/mesa/pipe/softpipe/sp_state_fs.c b/src/mesa/pipe/softpipe/sp_state_fs.c index 912f42d5688..a360b4f02b0 100644 --- a/src/mesa/pipe/softpipe/sp_state_fs.c +++ b/src/mesa/pipe/softpipe/sp_state_fs.c @@ -34,6 +34,8 @@ #include "pipe/draw/draw_context.h" #include "pipe/p_shader_tokens.h" #include "pipe/llvm/gallivm.h" +#include "pipe/tgsi/util/tgsi_dump.h" +#include "pipe/tgsi/exec/tgsi_sse2.h" void * softpipe_create_fs_state(struct pipe_context *pipe, diff --git a/src/mesa/pipe/softpipe/sp_state_sampler.c b/src/mesa/pipe/softpipe/sp_state_sampler.c index 246a7d6eda7..e71b9159e3f 100644 --- a/src/mesa/pipe/softpipe/sp_state_sampler.c +++ b/src/mesa/pipe/softpipe/sp_state_sampler.c @@ -68,12 +68,12 @@ softpipe_delete_sampler_state(struct pipe_context *pipe, void softpipe_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_mipmap_tree *texture) + struct pipe_texture *texture) { struct softpipe_context *softpipe = softpipe_context(pipe); assert(unit < PIPE_MAX_SAMPLERS); - softpipe->texture[unit] = texture; /* ptr, not struct */ + softpipe->texture[unit] = (struct softpipe_texture *)texture; /* ptr, not struct */ sp_tile_cache_set_texture(softpipe->tex_cache[unit], texture); diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index a6ab4046037..c41bbc59b90 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -565,34 +565,35 @@ z24s8_get_tile(struct pipe_surface *ps, */ struct pipe_surface * softpipe_get_tex_surface(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt, + struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { + struct softpipe_texture *spt = (struct softpipe_texture *)pt; struct pipe_surface *ps; unsigned offset; /* in bytes */ - offset = mt->level[level].level_offset; + offset = spt->level_offset[level]; - if (mt->target == PIPE_TEXTURE_CUBE) { - offset += mt->level[level].image_offset[face] * mt->cpp; + if (pt->target == PIPE_TEXTURE_CUBE) { + offset += spt->image_offset[level][face] * pt->cpp; } - else if (mt->target == PIPE_TEXTURE_3D) { - offset += mt->level[level].image_offset[zslice] * mt->cpp; + else if (pt->target == PIPE_TEXTURE_3D) { + offset += spt->image_offset[level][zslice] * pt->cpp; } else { assert(face == 0); assert(zslice == 0); } - ps = pipe->winsys->surface_alloc(pipe->winsys, mt->format); + ps = pipe->winsys->surface_alloc(pipe->winsys, pt->format); if (ps) { assert(ps->format); assert(ps->refcount); - pipe_region_reference(&ps->region, mt->region); - ps->cpp = mt->cpp; - ps->width = mt->level[level].width; - ps->height = mt->level[level].height; - ps->pitch = mt->pitch; + pipe_region_reference(&ps->region, spt->region); + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = spt->pitch; ps->offset = offset; } return ps; diff --git a/src/mesa/pipe/softpipe/sp_surface.h b/src/mesa/pipe/softpipe/sp_surface.h index cf87e1a92ce..b652e7598e8 100644 --- a/src/mesa/pipe/softpipe/sp_surface.h +++ b/src/mesa/pipe/softpipe/sp_surface.h @@ -41,7 +41,7 @@ struct softpipe_tile_cache; extern struct pipe_surface * softpipe_get_tex_surface(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt, + struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice); diff --git a/src/mesa/pipe/softpipe/sp_tex_layout.c b/src/mesa/pipe/softpipe/sp_tex_layout.c deleted file mode 100644 index 8156b00301f..00000000000 --- a/src/mesa/pipe/softpipe/sp_tex_layout.c +++ /dev/null @@ -1,361 +0,0 @@ -/************************************************************************** - * - * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - /* - * Authors: - * Keith Whitwell - * Michel Dänzer - */ - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "sp_tex_layout.h" - - -/* At the moment, just make softpipe use the same layout for its - * textures as the i945. Softpipe needs some sort of texture layout, - * this one was handy. May be worthwhile to simplify this code a - * little. - */ - -static unsigned minify( unsigned d ) -{ - return MAX2(1, d>>1); -} - -static int align(int value, int alignment) -{ - return (value + alignment - 1) & ~(alignment - 1); -} - - -static void -sp_miptree_set_level_info(struct pipe_mipmap_tree *mt, - unsigned level, - unsigned nr_images, - unsigned x, unsigned y, unsigned w, unsigned h, unsigned d) -{ - assert(level < PIPE_MAX_TEXTURE_LEVELS); - - mt->level[level].width = w; - mt->level[level].height = h; - mt->level[level].depth = d; - mt->level[level].level_offset = (x + y * mt->pitch) * mt->cpp; - mt->level[level].nr_images = nr_images; - - /* - DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, - level, w, h, d, x, y, mt->level[level].level_offset); - */ - - /* Not sure when this would happen, but anyway: - */ - if (mt->level[level].image_offset) { - FREE( mt->level[level].image_offset ); - mt->level[level].image_offset = NULL; - } - - assert(nr_images); - assert(!mt->level[level].image_offset); - - mt->level[level].image_offset = (unsigned *) MALLOC( nr_images * sizeof(unsigned) ); - mt->level[level].image_offset[0] = 0; -} - - -static void -sp_miptree_set_image_offset(struct pipe_mipmap_tree *mt, - unsigned level, unsigned img, unsigned x, unsigned y) -{ - if (img == 0 && level == 0) - assert(x == 0 && y == 0); - - assert(img < mt->level[level].nr_images); - - mt->level[level].image_offset[img] = (x + y * mt->pitch); - - /* - DBG("%s level %d img %d pos %d,%d image_offset %x\n", - __FUNCTION__, level, img, x, y, mt->level[level].image_offset[img]); - */ -} - - -static void -sp_miptree_layout_2d( struct pipe_mipmap_tree *mt ) -{ - int align_h = 2, align_w = 4; - unsigned level; - unsigned x = 0; - unsigned y = 0; - unsigned width = mt->width0; - unsigned height = mt->height0; - - mt->pitch = mt->width0; - /* XXX FIX THIS: - * we use alignment=64 bytes in sp_region_alloc(). If we change - * that, change this too. - */ - if (mt->pitch < 16) - mt->pitch = 16; - - /* May need to adjust pitch to accomodate the placement of - * the 2nd mipmap. This occurs when the alignment - * constraints of mipmap placement push the right edge of the - * 2nd mipmap out past the width of its parent. - */ - if (mt->first_level != mt->last_level) { - unsigned mip1_width = align(minify(mt->width0), align_w) - + minify(minify(mt->width0)); - - if (mip1_width > mt->width0) - mt->pitch = mip1_width; - } - - /* Pitch must be a whole number of dwords, even though we - * express it in texels. - */ - mt->pitch = align(mt->pitch * mt->cpp, 4) / mt->cpp; - mt->total_height = 0; - - for ( level = mt->first_level ; level <= mt->last_level ; level++ ) { - unsigned img_height; - - sp_miptree_set_level_info(mt, level, 1, x, y, width, height, 1); - - if (mt->compressed) - img_height = MAX2(1, height/4); - else - img_height = align(height, align_h); - - - /* Because the images are packed better, the final offset - * might not be the maximal one: - */ - mt->total_height = MAX2(mt->total_height, y + img_height); - - /* Layout_below: step right after second mipmap. - */ - if (level == mt->first_level + 1) { - x += align(width, align_w); - } - else { - y += img_height; - } - - width = minify(width); - height = minify(height); - } -} - - -static const int initial_offsets[6][2] = { - {0, 0}, - {0, 2}, - {1, 0}, - {1, 2}, - {1, 1}, - {1, 3} -}; - -static const int step_offsets[6][2] = { - {0, 2}, - {0, 2}, - {-1, 2}, - {-1, 2}, - {-1, 1}, - {-1, 1} -}; - - - -boolean -softpipe_mipmap_tree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree * mt) -{ - unsigned level; - - switch (mt->target) { - case PIPE_TEXTURE_CUBE:{ - const unsigned dim = mt->width0; - unsigned face; - unsigned lvlWidth = mt->width0, lvlHeight = mt->height0; - - assert(lvlWidth == lvlHeight); /* cubemap images are square */ - - /* Depending on the size of the largest images, pitch can be - * determined either by the old-style packing of cubemap faces, - * or the final row of 4x4, 2x2 and 1x1 faces below this. - */ - if (dim > 32) - mt->pitch = ((dim * mt->cpp * 2 + 3) & ~3) / mt->cpp; - else - mt->pitch = 14 * 8; - - mt->total_height = dim * 4 + 4; - - /* Set all the levels to effectively occupy the whole rectangular region. - */ - for (level = mt->first_level; level <= mt->last_level; level++) { - sp_miptree_set_level_info(mt, level, 6, - 0, 0, - lvlWidth, lvlHeight, 1); - lvlWidth /= 2; - lvlHeight /= 2; - } - - - for (face = 0; face < 6; face++) { - unsigned x = initial_offsets[face][0] * dim; - unsigned y = initial_offsets[face][1] * dim; - unsigned d = dim; - - if (dim == 4 && face >= 4) { - y = mt->total_height - 4; - x = (face - 4) * 8; - } - else if (dim < 4 && (face > 0 || mt->first_level > 0)) { - y = mt->total_height - 4; - x = face * 8; - } - - for (level = mt->first_level; level <= mt->last_level; level++) { - sp_miptree_set_image_offset(mt, level, face, x, y); - - d >>= 1; - - switch (d) { - case 4: - switch (face) { - case PIPE_TEX_FACE_POS_X: - case PIPE_TEX_FACE_NEG_X: - x += step_offsets[face][0] * d; - y += step_offsets[face][1] * d; - break; - case PIPE_TEX_FACE_POS_Y: - case PIPE_TEX_FACE_NEG_Y: - y += 12; - x -= 8; - break; - case PIPE_TEX_FACE_POS_Z: - case PIPE_TEX_FACE_NEG_Z: - y = mt->total_height - 4; - x = (face - 4) * 8; - break; - } - - case 2: - y = mt->total_height - 4; - x = 16 + face * 8; - break; - - case 1: - x += 48; - break; - - default: - x += step_offsets[face][0] * d; - y += step_offsets[face][1] * d; - break; - } - } - } - break; - } - case PIPE_TEXTURE_3D:{ - unsigned width = mt->width0; - unsigned height = mt->height0; - unsigned depth = mt->depth0; - unsigned pack_x_pitch, pack_x_nr; - unsigned pack_y_pitch; - unsigned level; - - mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp; - mt->total_height = 0; - - pack_y_pitch = MAX2(mt->height0, 2); - pack_x_pitch = mt->pitch; - pack_x_nr = 1; - - for (level = mt->first_level; level <= mt->last_level; level++) { - unsigned nr_images = mt->target == PIPE_TEXTURE_3D ? depth : 6; - int x = 0; - int y = 0; - unsigned q, j; - - sp_miptree_set_level_info(mt, level, nr_images, - 0, mt->total_height, - width, height, depth); - - for (q = 0; q < nr_images;) { - for (j = 0; j < pack_x_nr && q < nr_images; j++, q++) { - sp_miptree_set_image_offset(mt, level, q, x, y); - x += pack_x_pitch; - } - - x = 0; - y += pack_y_pitch; - } - - - mt->total_height += y; - - if (pack_x_pitch > 4) { - pack_x_pitch >>= 1; - pack_x_nr <<= 1; - assert(pack_x_pitch * pack_x_nr <= mt->pitch); - } - - if (pack_y_pitch > 2) { - pack_y_pitch >>= 1; - } - - width = minify(width); - height = minify(height); - depth = minify(depth); - } - break; - } - - case PIPE_TEXTURE_1D: - case PIPE_TEXTURE_2D: -// case PIPE_TEXTURE_RECTANGLE: - sp_miptree_layout_2d(mt); - break; - default: - assert(0); - break; - } - - /* - DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, - mt->pitch, - mt->total_height, mt->cpp, mt->pitch * mt->total_height * mt->cpp); - */ - - return TRUE; -} - diff --git a/src/mesa/pipe/softpipe/sp_tex_layout.h b/src/mesa/pipe/softpipe/sp_tex_layout.h deleted file mode 100644 index ea19c13b23f..00000000000 --- a/src/mesa/pipe/softpipe/sp_tex_layout.h +++ /dev/null @@ -1,16 +0,0 @@ -#ifndef SP_TEX_LAYOUT_H -#define SP_TEX_LAYOUT_H - - -struct pipe_context; -struct pipe_mipmap_tree; - - -extern boolean -softpipe_mipmap_tree_layout(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt); - - -#endif /* SP_TEX_LAYOUT_H */ - - diff --git a/src/mesa/pipe/softpipe/sp_tex_sample.c b/src/mesa/pipe/softpipe/sp_tex_sample.c index 92958400fc5..9e48ed0cd2c 100644 --- a/src/mesa/pipe/softpipe/sp_tex_sample.c +++ b/src/mesa/pipe/softpipe/sp_tex_sample.c @@ -422,7 +422,7 @@ compute_lambda(struct tgsi_sampler *sampler, dsdy = FABSF(dsdy); rho = MAX2(dsdx, dsdy); if (sampler->state->normalized_coords) - rho *= sampler->texture->width0; + rho *= sampler->texture->width[0]; } if (t) { float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]; @@ -432,7 +432,7 @@ compute_lambda(struct tgsi_sampler *sampler, dtdy = FABSF(dtdy); max = MAX2(dtdx, dtdy); if (sampler->state->normalized_coords) - max *= sampler->texture->height0; + max *= sampler->texture->height[0]; rho = MAX2(rho, max); } if (p) { @@ -443,7 +443,7 @@ compute_lambda(struct tgsi_sampler *sampler, dpdy = FABSF(dpdy); max = MAX2(dpdx, dpdy); if (sampler->state->normalized_coords) - max *= sampler->texture->depth0; + max *= sampler->texture->depth[0]; rho = MAX2(rho, max); } @@ -620,8 +620,8 @@ sp_get_samples_2d_common(struct tgsi_sampler *sampler, &level0, &level1, &levelBlend, &imgFilter); if (sampler->state->normalized_coords) { - width = sampler->texture->level[level0].width; - height = sampler->texture->level[level0].height; + width = sampler->texture->width[level0]; + height = sampler->texture->height[level0]; } else { width = height = 1; @@ -757,9 +757,9 @@ sp_get_samples_3d(struct tgsi_sampler *sampler, &level0, &level1, &levelBlend, &imgFilter); if (sampler->state->normalized_coords) { - width = sampler->texture->level[level0].width; - height = sampler->texture->level[level0].height; - depth = sampler->texture->level[level0].depth; + width = sampler->texture->width[level0]; + height = sampler->texture->height[level0]; + depth = sampler->texture->depth[level0]; } else { width = height = depth = 1; @@ -883,7 +883,7 @@ sp_get_samples_cube(struct tgsi_sampler *sampler, /** * Called via tgsi_sampler::get_samples() * Use the sampler's state setting to get a filtered RGBA value - * from the sampler's texture (mipmap tree). + * from the sampler's texture. * * XXX we can implement many versions of this function, each * tightly coded for a specific combination of sampler state diff --git a/src/mesa/pipe/softpipe/sp_texture.c b/src/mesa/pipe/softpipe/sp_texture.c new file mode 100644 index 00000000000..2f9a1e9837b --- /dev/null +++ b/src/mesa/pipe/softpipe/sp_texture.c @@ -0,0 +1,425 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + /* + * Authors: + * Keith Whitwell + * Michel Dänzer + */ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" + +#include "sp_context.h" +#include "sp_state.h" +#include "sp_texture.h" + + +/* At the moment, just make softpipe use the same layout for its + * textures as the i945. Softpipe needs some sort of texture layout, + * this one was handy. May be worthwhile to simplify this code a + * little. + */ + +static unsigned minify( unsigned d ) +{ + return MAX2(1, d>>1); +} + +static int align(int value, int alignment) +{ + return (value + alignment - 1) & ~(alignment - 1); +} + + +static void +sp_miptree_set_level_info(struct softpipe_texture *spt, + unsigned level, + unsigned nr_images, + unsigned x, unsigned y, unsigned w, unsigned h, + unsigned d) +{ + struct pipe_texture *pt = &spt->base; + + assert(level < PIPE_MAX_TEXTURE_LEVELS); + + pt->width[level] = w; + pt->height[level] = h; + pt->depth[level] = d; + + spt->nr_images[level] = nr_images; + spt->level_offset[level] = (x + y * spt->pitch) * pt->cpp; + + /* + DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, + level, w, h, d, x, y, spt->level_offset[level]); + */ + + /* Not sure when this would happen, but anyway: + */ + if (spt->image_offset[level]) { + FREE( spt->image_offset[level] ); + spt->image_offset[level] = NULL; + } + + assert(nr_images); + assert(!spt->image_offset[level]); + + spt->image_offset[level] = (unsigned *) MALLOC( nr_images * sizeof(unsigned) ); + spt->image_offset[level][0] = 0; +} + + +static void +sp_miptree_set_image_offset(struct softpipe_texture *spt, + unsigned level, unsigned img, unsigned x, unsigned y) +{ + if (img == 0 && level == 0) + assert(x == 0 && y == 0); + + assert(img < spt->nr_images[level]); + + spt->image_offset[level][img] = (x + y * spt->pitch); + + /* + DBG("%s level %d img %d pos %d,%d image_offset %x\n", + __FUNCTION__, level, img, x, y, spt->image_offset[level][img]); + */ +} + + +static void +sp_miptree_layout_2d( struct softpipe_texture *spt ) +{ + struct pipe_texture *pt = &spt->base; + int align_h = 2, align_w = 4; + unsigned level; + unsigned x = 0; + unsigned y = 0; + unsigned width = pt->width[0]; + unsigned height = pt->height[0]; + + spt->pitch = pt->width[0]; + /* XXX FIX THIS: + * we use alignment=64 bytes in sp_region_alloc(). If we change + * that, change this too. + */ + if (spt->pitch < 16) + spt->pitch = 16; + + /* May need to adjust pitch to accomodate the placement of + * the 2nd mipmap. This occurs when the alignment + * constraints of mipmap placement push the right edge of the + * 2nd mipmap out past the width of its parent. + */ + if (pt->first_level != pt->last_level) { + unsigned mip1_width = align(minify(pt->width[0]), align_w) + + minify(minify(pt->width[0])); + + if (mip1_width > pt->width[0]) + spt->pitch = mip1_width; + } + + /* Pitch must be a whole number of dwords, even though we + * express it in texels. + */ + spt->pitch = align(spt->pitch * pt->cpp, 4) / pt->cpp; + spt->total_height = 0; + + for ( level = pt->first_level ; level <= pt->last_level ; level++ ) { + unsigned img_height; + + sp_miptree_set_level_info(spt, level, 1, x, y, width, height, 1); + + if (pt->compressed) + img_height = MAX2(1, height/4); + else + img_height = align(height, align_h); + + + /* Because the images are packed better, the final offset + * might not be the maximal one: + */ + spt->total_height = MAX2(spt->total_height, y + img_height); + + /* Layout_below: step right after second mipmap. + */ + if (level == pt->first_level + 1) { + x += align(width, align_w); + } + else { + y += img_height; + } + + width = minify(width); + height = minify(height); + } +} + + +static const int initial_offsets[6][2] = { + {0, 0}, + {0, 2}, + {1, 0}, + {1, 2}, + {1, 1}, + {1, 3} +}; + +static const int step_offsets[6][2] = { + {0, 2}, + {0, 2}, + {-1, 2}, + {-1, 2}, + {-1, 1}, + {-1, 1} +}; + + + +static boolean +softpipe_mipmap_tree_layout(struct pipe_context *pipe, struct softpipe_texture * spt) +{ + struct pipe_texture *pt = &spt->base; + unsigned level; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE:{ + const unsigned dim = pt->width[0]; + unsigned face; + unsigned lvlWidth = pt->width[0], lvlHeight = pt->height[0]; + + assert(lvlWidth == lvlHeight); /* cubemap images are square */ + + /* Depending on the size of the largest images, pitch can be + * determined either by the old-style packing of cubemap faces, + * or the final row of 4x4, 2x2 and 1x1 faces below this. + */ + if (dim > 32) + spt->pitch = ((dim * pt->cpp * 2 + 3) & ~3) / pt->cpp; + else + spt->pitch = 14 * 8; + + spt->total_height = dim * 4 + 4; + + /* Set all the levels to effectively occupy the whole rectangular region. + */ + for (level = pt->first_level; level <= pt->last_level; level++) { + sp_miptree_set_level_info(spt, level, 6, + 0, 0, + lvlWidth, lvlHeight, 1); + lvlWidth /= 2; + lvlHeight /= 2; + } + + + for (face = 0; face < 6; face++) { + unsigned x = initial_offsets[face][0] * dim; + unsigned y = initial_offsets[face][1] * dim; + unsigned d = dim; + + if (dim == 4 && face >= 4) { + y = spt->total_height - 4; + x = (face - 4) * 8; + } + else if (dim < 4 && (face > 0 || pt->first_level > 0)) { + y = spt->total_height - 4; + x = face * 8; + } + + for (level = pt->first_level; level <= pt->last_level; level++) { + sp_miptree_set_image_offset(spt, level, face, x, y); + + d >>= 1; + + switch (d) { + case 4: + switch (face) { + case PIPE_TEX_FACE_POS_X: + case PIPE_TEX_FACE_NEG_X: + x += step_offsets[face][0] * d; + y += step_offsets[face][1] * d; + break; + case PIPE_TEX_FACE_POS_Y: + case PIPE_TEX_FACE_NEG_Y: + y += 12; + x -= 8; + break; + case PIPE_TEX_FACE_POS_Z: + case PIPE_TEX_FACE_NEG_Z: + y = spt->total_height - 4; + x = (face - 4) * 8; + break; + } + + case 2: + y = spt->total_height - 4; + x = 16 + face * 8; + break; + + case 1: + x += 48; + break; + + default: + x += step_offsets[face][0] * d; + y += step_offsets[face][1] * d; + break; + } + } + } + break; + } + case PIPE_TEXTURE_3D:{ + unsigned width = pt->width[0]; + unsigned height = pt->height[0]; + unsigned depth = pt->depth[0]; + unsigned pack_x_pitch, pack_x_nr; + unsigned pack_y_pitch; + unsigned level; + + spt->pitch = ((pt->width[0] * pt->cpp + 3) & ~3) / pt->cpp; + spt->total_height = 0; + + pack_y_pitch = MAX2(pt->height[0], 2); + pack_x_pitch = spt->pitch; + pack_x_nr = 1; + + for (level = pt->first_level; level <= pt->last_level; level++) { + unsigned nr_images = pt->target == PIPE_TEXTURE_3D ? depth : 6; + int x = 0; + int y = 0; + unsigned q, j; + + sp_miptree_set_level_info(spt, level, nr_images, + 0, spt->total_height, + width, height, depth); + + for (q = 0; q < nr_images;) { + for (j = 0; j < pack_x_nr && q < nr_images; j++, q++) { + sp_miptree_set_image_offset(spt, level, q, x, y); + x += pack_x_pitch; + } + + x = 0; + y += pack_y_pitch; + } + + + spt->total_height += y; + + if (pack_x_pitch > 4) { + pack_x_pitch >>= 1; + pack_x_nr <<= 1; + assert(pack_x_pitch * pack_x_nr <= spt->pitch); + } + + if (pack_y_pitch > 2) { + pack_y_pitch >>= 1; + } + + width = minify(width); + height = minify(height); + depth = minify(depth); + } + break; + } + + case PIPE_TEXTURE_1D: + case PIPE_TEXTURE_2D: +// case PIPE_TEXTURE_RECTANGLE: + sp_miptree_layout_2d(spt); + break; + default: + assert(0); + break; + } + + /* + DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, + spt->pitch, + spt->total_height, pt->cpp, spt->pitch * spt->total_height * pt->cpp); + */ + + return TRUE; +} + +void +softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct softpipe_texture *spt = REALLOC(*pt, sizeof(struct pipe_texture), + sizeof(struct softpipe_texture)); + + if (spt) { + memset(&spt->base + 1, 0, + sizeof(struct softpipe_texture) - sizeof(struct pipe_texture)); + + if (softpipe_mipmap_tree_layout(pipe, spt)) { + spt->region = pipe->winsys->region_alloc(pipe->winsys, + spt->pitch * (*pt)->cpp * + spt->total_height, + PIPE_SURFACE_FLAG_TEXTURE); + } + + if (!spt->region) { + FREE(spt); + spt = NULL; + } + } + + *pt = &spt->base; +} + +void +softpipe_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + if (!*pt) + return; + + /* + DBG("%s %p refcount will be %d\n", + __FUNCTION__, (void *) *pt, (*pt)->refcount - 1); + */ + if (--(*pt)->refcount <= 0) { + struct softpipe_texture *spt = (struct softpipe_texture *)*pt; + uint i; + + /* + DBG("%s deleting %p\n", __FUNCTION__, (void *) spt); + */ + + pipe->winsys->region_release(pipe->winsys, &spt->region); + + for (i = 0; i < PIPE_MAX_TEXTURE_LEVELS; i++) + if (spt->image_offset[i]) + free(spt->image_offset[i]); + + free(spt); + } + *pt = NULL; +} diff --git a/src/mesa/pipe/softpipe/sp_texture.h b/src/mesa/pipe/softpipe/sp_texture.h new file mode 100644 index 00000000000..2aca57bd1d2 --- /dev/null +++ b/src/mesa/pipe/softpipe/sp_texture.h @@ -0,0 +1,18 @@ +#ifndef SP_TEXTURE_H +#define SP_TEXTURE_H + + +struct pipe_context; +struct pipe_texture; + + +extern void +softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); + +extern void +softpipe_texture_release(struct pipe_context *pipe, struct pipe_texture **pt); + + +#endif /* SP_TEXTURE */ + + diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index ea0c8b8f91d..62ee6a27c93 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -51,7 +51,7 @@ struct softpipe_tile_cache { struct pipe_surface *surface; /**< the surface we're caching */ - struct pipe_mipmap_tree *texture; /**< if caching a texture */ + struct pipe_texture *texture; /**< if caching a texture */ struct softpipe_cached_tile entries[NUM_ENTRIES]; uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32]; float clear_value[4]; @@ -139,7 +139,7 @@ sp_tile_cache_get_surface(struct softpipe_tile_cache *tc) void sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, - struct pipe_mipmap_tree *texture) + struct pipe_texture *texture) { uint i; diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.h b/src/mesa/pipe/softpipe/sp_tile_cache.h index e66fec2e208..9967aa5044a 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.h +++ b/src/mesa/pipe/softpipe/sp_tile_cache.h @@ -71,7 +71,7 @@ sp_tile_cache_get_surface(struct softpipe_tile_cache *tc); extern void sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, - struct pipe_mipmap_tree *texture); + struct pipe_texture *texture); extern void sp_flush_tile_cache(struct softpipe_context *softpipe, diff --git a/src/mesa/pipe/tgsi/exec/tgsi_exec.h b/src/mesa/pipe/tgsi/exec/tgsi_exec.h index 1d497e97fb0..2c62b30f150 100644 --- a/src/mesa/pipe/tgsi/exec/tgsi_exec.h +++ b/src/mesa/pipe/tgsi/exec/tgsi_exec.h @@ -76,7 +76,7 @@ struct softpipe_tile_cache; /**< Opaque to TGSI */ struct tgsi_sampler { const struct pipe_sampler_state *state; - struct pipe_mipmap_tree *texture; + struct pipe_texture *texture; /** Get samples for four fragments in a quad */ void (*get_samples)(struct tgsi_sampler *sampler, const float s[QUAD_SIZE], diff --git a/src/mesa/sources b/src/mesa/sources index 2df60d1996c..596d3b89d5d 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -236,7 +236,7 @@ STATETRACKER_SOURCES = \ state_tracker/st_framebuffer.c \ state_tracker/st_mesa_to_tgsi.c \ state_tracker/st_program.c \ - state_tracker/st_mipmap_tree.c + state_tracker/st_texture.c SHADER_SOURCES = \ shader/arbprogparse.c \ diff --git a/src/mesa/state_tracker/st_atom_texture.c b/src/mesa/state_tracker/st_atom_texture.c index f25cfd386a3..c4e5af02d59 100644 --- a/src/mesa/state_tracker/st_atom_texture.c +++ b/src/mesa/state_tracker/st_atom_texture.c @@ -51,24 +51,24 @@ update_textures(struct st_context *st) for (u = 0; u < st->ctx->Const.MaxTextureImageUnits; u++) { struct gl_texture_object *texObj = st->ctx->Texture.Unit[u]._Current; - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; if (texObj) { GLboolean flush, retval; - retval = st_finalize_mipmap_tree(st->ctx, st->pipe, u, &flush); + retval = st_finalize_texture(st->ctx, st->pipe, u, &flush); #if 0 - printf("finalize_mipmap_tree returned %d, flush = %d\n", + printf("finalize_texture returned %d, flush = %d\n", retval, flush); #endif - mt = st_get_texobj_mipmap_tree(texObj); + pt = st_get_texobj_texture(texObj); } else { - mt = NULL; + pt = NULL; } - st->state.texture[u] = mt; - st->pipe->set_texture_state(st->pipe, u, mt); + st->state.texture[u] = pt; + st->pipe->set_texture_state(st->pipe, u, pt); } } diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 39c0cf6b327..c28ad15b294 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -33,6 +33,7 @@ #include "main/imports.h" #include "main/image.h" #include "main/macros.h" +#include "main/texformat.h" #include "shader/program.h" #include "shader/prog_parameter.h" #include "shader/prog_print.h" @@ -50,6 +51,7 @@ #include "st_draw.h" #include "st_format.h" #include "st_mesa_to_tgsi.h" +#include "st_texture.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_inlines.h" @@ -440,63 +442,20 @@ _mesa_base_format(GLenum format) } - -static struct pipe_mipmap_tree * -alloc_mipmap_tree(struct st_context *st, - GLsizei width, GLsizei height, uint pipeFormat) -{ - const GLbitfield flags = PIPE_SURFACE_FLAG_TEXTURE; - struct pipe_mipmap_tree *mt; - GLuint cpp; - - mt = CALLOC_STRUCT(pipe_mipmap_tree); - if (!mt) - return NULL; - - cpp = st_sizeof_format(pipeFormat); - - mt->target = PIPE_TEXTURE_2D; - mt->internal_format = GL_RGBA; - mt->format = pipeFormat; - mt->first_level = 0; - mt->last_level = 0; - mt->width0 = width; - mt->height0 = height; - mt->depth0 = 1; - mt->cpp = cpp; - mt->compressed = 0; - mt->pitch = st->pipe->winsys->surface_pitch(st->pipe->winsys, cpp, width, - flags); - mt->region = st->pipe->winsys->region_alloc(st->pipe->winsys, - mt->pitch * cpp * height, flags); - mt->depth_pitch = 0; - mt->total_height = height; - mt->level[0].level_offset = 0; - mt->level[0].width = width; - mt->level[0].height = height; - mt->level[0].depth = 1; - mt->level[0].nr_images = 1; - mt->level[0].image_offset = NULL; - mt->refcount = 1; - - return mt; -} - - /** - * Make mipmap tree containing an image for glDrawPixels image. + * Make texture containing an image for glDrawPixels image. * If 'pixels' is NULL, leave the texture image data undefined. */ -static struct pipe_mipmap_tree * -make_mipmap_tree(struct st_context *st, - GLsizei width, GLsizei height, GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels) +static struct pipe_texture * +make_texture(struct st_context *st, + GLsizei width, GLsizei height, GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels) { GLcontext *ctx = st->ctx; struct pipe_context *pipe = st->pipe; const struct gl_texture_format *mformat; - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; GLuint pipeFormat, cpp; GLenum baseFormat; @@ -509,29 +468,33 @@ make_mipmap_tree(struct st_context *st, assert(pipeFormat); cpp = st_sizeof_format(pipeFormat); - mt = alloc_mipmap_tree(st, width, height, pipeFormat); - if (!mt) + pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, baseFormat, 0, 0, + width, height, 1, 0); + if (!pt) return NULL; if (unpack->BufferObj && unpack->BufferObj->Name) { /* - mt->region = buffer_object_region(unpack->BufferObj); + pt->region = buffer_object_region(unpack->BufferObj); */ printf("st_DrawPixels (sourcing from PBO not implemented yet)\n"); } { + struct pipe_surface *surface; static const GLuint dstImageOffsets = 0; GLboolean success; - GLuint pitch = mt->pitch; GLubyte *dest; const GLbitfield imageTransferStateSave = ctx->_ImageTransferState; /* we'll do pixel transfer in a fragment shader */ ctx->_ImageTransferState = 0x0; + surface = pipe->get_tex_surface(pipe, pt, 0, 0, 0); + /* map texture region */ - dest = pipe->region_map(pipe, mt->region); + (void) pipe->region_map(pipe, surface->region); + dest = surface->region->map + surface->offset; /* Put image into texture region. * Note that the image is actually going to be upside down in @@ -542,7 +505,7 @@ make_mipmap_tree(struct st_context *st, mformat, /* gl_texture_format */ dest, /* dest */ 0, 0, 0, /* dstX/Y/Zoffset */ - pitch * cpp, /* dstRowStride, bytes */ + surface->pitch * cpp, /* dstRowStride, bytes */ &dstImageOffsets, /* dstImageOffsets */ width, height, 1, /* size */ format, type, /* src format/type */ @@ -550,44 +513,15 @@ make_mipmap_tree(struct st_context *st, unpack); /* unmap */ - pipe->region_unmap(pipe, mt->region); + pipe->region_unmap(pipe, surface->region); + pipe_surface_reference(&surface, NULL); assert(success); /* restore */ ctx->_ImageTransferState = imageTransferStateSave; } -#if 0 - mt->target = PIPE_TEXTURE_2D; - mt->internal_format = GL_RGBA; - mt->format = pipeFormat; - mt->first_level = 0; - mt->last_level = 0; - mt->width0 = width; - mt->height0 = height; - mt->depth0 = 1; - mt->cpp = cpp; - mt->compressed = 0; - mt->pitch = mt->pitch; - mt->depth_pitch = 0; - mt->total_height = height; - mt->level[0].level_offset = 0; - mt->level[0].width = width; - mt->level[0].height = height; - mt->level[0].depth = 1; - mt->level[0].nr_images = 1; - mt->level[0].image_offset = NULL; - mt->refcount = 1; -#endif - return mt; -} - - -static void -free_mipmap_tree(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) -{ - pipe->winsys->region_release(pipe->winsys, &mt->region); - free(mt); + return pt; } @@ -693,7 +627,7 @@ static void draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, GLsizei width, GLsizei height, GLfloat zoomX, GLfloat zoomY, - struct pipe_mipmap_tree *mt, + struct pipe_texture *pt, struct st_vertex_program *stvp, struct st_fragment_program *stfp, const GLfloat *color, @@ -761,9 +695,9 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, pipe->set_viewport_state(pipe, &vp); } - /* mipmap tree state: */ + /* texture state: */ { - pipe->set_texture_state(pipe, unit, mt); + pipe->set_texture_state(pipe, unit, pt); } /* Compute window coords (y=0=bottom) with pixel zoom. @@ -1025,14 +959,13 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, any_pixel_transfer_ops(st) || !compatible_formats(format, type, ps->format)) { /* textured quad */ - struct pipe_mipmap_tree *mt - = make_mipmap_tree(ctx->st, width, height, format, type, - unpack, pixels); - if (mt) { + struct pipe_texture *pt + = make_texture(ctx->st, width, height, format, type, unpack, pixels); + if (pt) { draw_textured_quad(ctx, x, y, ctx->Current.RasterPos[2], width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY, - mt, stvp, stfp, color, GL_FALSE); - free_mipmap_tree(st->pipe, mt); + pt, stvp, stfp, color, GL_FALSE); + st->pipe->texture_release(st->pipe, &pt); } } else { @@ -1046,26 +979,29 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, /** * Create a texture which represents a bitmap image. */ -static struct pipe_mipmap_tree * +static struct pipe_texture * make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap) { struct pipe_context *pipe = ctx->st->pipe; - const uint flags = PIPE_SURFACE_FLAG_TEXTURE; + struct pipe_surface *surface; uint format = 0, cpp, comp; + GLenum internal_format; ubyte *dest; - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; int row, col; /* find a texture format we know */ if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_I8 )) { format = PIPE_FORMAT_U_I8; + internal_format = GL_INTENSITY8; cpp = 1; comp = 0; } else if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A8_R8_G8_B8 )) { format = PIPE_FORMAT_U_A8_R8_G8_B8; + internal_format = GL_RGBA8; cpp = 4; comp = 3; /* alpha channel */ /*XXX little-endian dependency */ } @@ -1075,31 +1011,25 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, } /** - * Create a mipmap tree. + * Create a texture. */ - mt = CALLOC_STRUCT(pipe_mipmap_tree); - if (!mt) + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, internal_format, + 0, 0, width, height, 1, 0); + if (!pt) return NULL; if (unpack->BufferObj && unpack->BufferObj->Name) { /* - mt->region = buffer_object_region(unpack->BufferObj); + pt->region = buffer_object_region(unpack->BufferObj); */ printf("st_Bitmap (sourcing from PBO not implemented yet)\n"); } - - /* allocate texture region/storage */ - mt->pitch = pipe->winsys->surface_pitch(pipe->winsys, cpp, width, flags); - mt->region = pipe->winsys->region_alloc(pipe->winsys, - mt->pitch * cpp * height, flags); + surface = pipe->get_tex_surface(pipe, pt, 0, 0, 0); /* map texture region */ - dest = pipe->region_map(pipe, mt->region); - if (!dest) { - printf("st_Bitmap region_map() failed!?!"); - return NULL; - } + (void) pipe->region_map(pipe, surface->region); + dest = surface->region->map + surface->offset; /* Put image into texture region. * Note that the image is actually going to be upside down in @@ -1109,7 +1039,7 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, for (row = 0; row < height; row++) { const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack, bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0); - ubyte *destRow = dest + row * mt->pitch * cpp; + ubyte *destRow = dest + row * surface->pitch * cpp; if (unpack->LsbFirst) { /* Lsb first */ @@ -1158,30 +1088,13 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, } /* row */ - /* unmap */ - pipe->region_unmap(pipe, mt->region); - - mt->target = PIPE_TEXTURE_2D; - mt->internal_format = GL_RGBA; - mt->format = format; - mt->first_level = 0; - mt->last_level = 0; - mt->width0 = width; - mt->height0 = height; - mt->depth0 = 1; - mt->cpp = cpp; - mt->compressed = 0; - mt->depth_pitch = 0; - mt->total_height = height; - mt->level[0].level_offset = 0; - mt->level[0].width = width; - mt->level[0].height = height; - mt->level[0].depth = 1; - mt->level[0].nr_images = 1; - mt->level[0].image_offset = NULL; - mt->refcount = 1; - - return mt; + /* Release surface */ + pipe->region_unmap(pipe, surface->region); + pipe_surface_reference(&surface, NULL); + + pt->format = format; + + return pt; } @@ -1193,21 +1106,21 @@ st_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, struct st_fragment_program *stfp; struct st_vertex_program *stvp; struct st_context *st = ctx->st; - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; stvp = make_vertex_shader(ctx->st, GL_TRUE); stfp = combined_bitmap_fragment_program(ctx); st_validate_state(st); - mt = make_bitmap_texture(ctx, width, height, unpack, bitmap); - if (mt) { + pt = make_bitmap_texture(ctx, width, height, unpack, bitmap); + if (pt) { draw_textured_quad(ctx, x, y, ctx->Current.RasterPos[2], width, height, 1.0, 1.0, - mt, stvp, stfp, + pt, stvp, stfp, ctx->Current.RasterColor, GL_FALSE); - free_mipmap_tree(st->pipe, mt); + st->pipe->texture_release(st->pipe, &pt); } } @@ -1296,7 +1209,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, struct st_fragment_program *stfp; struct pipe_surface *psRead; struct pipe_surface *psTex; - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; GLfloat *color; uint format; @@ -1327,11 +1240,12 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, psRead = rbRead->surface; format = psRead->format; - mt = alloc_mipmap_tree(ctx->st, width, height, format); - if (!mt) + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, + rbRead->Base.InternalFormat, 0, 0, width, height, 1, 0); + if (!pt) return; - psTex = pipe->get_tex_surface(pipe, mt, 0, 0, 0); + psTex = pipe->get_tex_surface(pipe, pt, 0, 0, 0); if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) { srcy = ctx->DrawBuffer->Height - srcy - height; @@ -1368,10 +1282,10 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, /* draw textured quad */ draw_textured_quad(ctx, dstx, dsty, ctx->Current.RasterPos[2], width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY, - mt, stvp, stfp, color, GL_TRUE); + pt, stvp, stfp, color, GL_TRUE); pipe_surface_reference(&psTex, NULL); - free_mipmap_tree(st->pipe, mt); + st->pipe->texture_release(st->pipe, &pt); } diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index 80c92e8b7a1..43681b7f8a9 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -286,7 +286,7 @@ st_render_texture(GLcontext *ctx, struct st_renderbuffer *strb; struct gl_renderbuffer *rb; struct pipe_context *pipe = st->pipe; - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; assert(!att->Renderbuffer); @@ -302,26 +302,26 @@ st_render_texture(GLcontext *ctx, rb->AllocStorage = NULL; /* should not get called */ strb = st_renderbuffer(rb); - /* get the mipmap tree for the texture */ - mt = st_get_texobj_mipmap_tree(att->Texture); - assert(mt); - assert(mt->level[att->TextureLevel].width); + /* get the texture for the texture object */ + pt = st_get_texobj_texture(att->Texture); + assert(pt); + assert(pt->width[att->TextureLevel]); - rb->Width = mt->level[att->TextureLevel].width; - rb->Height = mt->level[att->TextureLevel].height; + rb->Width = pt->width[att->TextureLevel]; + rb->Height = pt->height[att->TextureLevel]; - /* the renderbuffer's surface is inside the mipmap_tree: */ - strb->surface = pipe->get_tex_surface(pipe, mt, + /* the renderbuffer's surface is inside the texture */ + strb->surface = pipe->get_tex_surface(pipe, pt, att->CubeMapFace, att->TextureLevel, att->Zoffset); assert(strb->surface); - init_renderbuffer_bits(strb, mt->format); + init_renderbuffer_bits(strb, pt->format); /* - printf("RENDER TO TEXTURE obj=%p mt=%p surf=%p %d x %d\n", - att->Texture, mt, strb->surface, rb->Width, rb->Height); + printf("RENDER TO TEXTURE obj=%p pt=%p surf=%p %d x %d\n", + att->Texture, pt, strb->surface, rb->Width, rb->Height); */ /* Invalidate buffer state so that the pipe's framebuffer state diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 461705119fa..f45d7e4275b 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -40,7 +40,7 @@ #include "state_tracker/st_cb_fbo.h" #include "state_tracker/st_cb_texture.h" #include "state_tracker/st_format.h" -#include "state_tracker/st_mipmap_tree.h" +#include "state_tracker/st_texture.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" @@ -54,8 +54,7 @@ struct st_texture_object { struct gl_texture_object base; /* The "parent" object */ - /* The mipmap tree must include at least these levels once - * validated: + /* The texture must include at least these levels once validated: */ GLuint firstLevel; GLuint lastLevel; @@ -67,7 +66,7 @@ struct st_texture_object /* On validation any active images held in main memory or in other * regions will be copied to this region and the old storage freed. */ - struct pipe_mipmap_tree *mt; + struct pipe_texture *pt; GLboolean imageOverride; GLint depthOverride; @@ -76,24 +75,6 @@ struct st_texture_object -struct st_texture_image -{ - struct gl_texture_image base; - - /* These aren't stored in gl_texture_image - */ - GLuint level; - GLuint face; - - /* If stImage->mt != NULL, image data is stored here. - * Else if stImage->base.Data != NULL, image is stored there. - * Else there is no image data. - */ - struct pipe_mipmap_tree *mt; -}; - - - static INLINE struct st_texture_object * st_texture_object(struct gl_texture_object *obj) @@ -108,11 +89,11 @@ st_texture_image(struct gl_texture_image *img) } -struct pipe_mipmap_tree * -st_get_texobj_mipmap_tree(struct gl_texture_object *texObj) +struct pipe_texture * +st_get_texobj_texture(struct gl_texture_object *texObj) { struct st_texture_object *stObj = st_texture_object(texObj); - return stObj->mt; + return stObj->pt; } @@ -174,9 +155,9 @@ st_IsTextureResident(GLcontext * ctx, struct gl_texture_object *texObj) struct st_texture_object *stObj = st_texture_object(texObj); return - stObj->mt && - stObj->mt->region && - intel_is_region_resident(intel, stObj->mt->region); + stObj->pt && + stObj->pt->region && + intel_is_region_resident(intel, stObj->pt->region); #endif return 1; } @@ -207,11 +188,10 @@ static void st_DeleteTextureObject(GLcontext *ctx, struct gl_texture_object *texObj) { - struct pipe_context *pipe = ctx->st->pipe; struct st_texture_object *stObj = st_texture_object(texObj); - if (stObj->mt) - st_miptree_release(pipe, &stObj->mt); + if (stObj->pt) + ctx->st->pipe->texture_release(ctx->st->pipe, &stObj->pt); _mesa_delete_texture_object(ctx, texObj); } @@ -220,13 +200,12 @@ st_DeleteTextureObject(GLcontext *ctx, static void st_FreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage) { - struct pipe_context *pipe = ctx->st->pipe; struct st_texture_image *stImage = st_texture_image(texImage); DBG("%s\n", __FUNCTION__); - if (stImage->mt) { - st_miptree_release(pipe, &stImage->mt); + if (stImage->pt) { + ctx->st->pipe->texture_release(ctx->st->pipe, &stImage->pt); } if (texImage->Data) { @@ -287,10 +266,10 @@ do_memcpy(void *dest, const void *src, size_t n) } -/* Functions to store texture images. Where possible, mipmap_tree's +/* Functions to store texture images. Where possible, textures * will be created or further instantiated with image data, otherwise * images will be stored in malloc'd memory. A validation step is - * required to pull those images into a mipmap tree, or otherwise + * required to pull those images into a texture, or otherwise * decide a fallback is required. */ @@ -313,17 +292,16 @@ logbase2(int n) /* Otherwise, store it in memory if (Border != 0) or (any dimension == * 1). * - * Otherwise, if max_level >= level >= min_level, create tree with - * space for textures from min_level down to max_level. + * Otherwise, if max_level >= level >= min_level, create texture with + * space for images from min_level down to max_level. * - * Otherwise, create tree with space for textures from (level - * 0)..(1x1). Consider pruning this tree at a validation if the - * saving is worth it. + * Otherwise, create texture with space for images from (level 0)..(1x1). + * Consider pruning this texture at a validation if the saving is worth it. */ static void -guess_and_alloc_mipmap_tree(struct pipe_context *pipe, - struct st_texture_object *stObj, - struct st_texture_image *stImage) +guess_and_alloc_texture(struct st_context *st, + struct st_texture_object *stObj, + struct st_texture_image *stImage) { GLuint firstLevel; GLuint lastLevel; @@ -382,23 +360,20 @@ guess_and_alloc_mipmap_tree(struct pipe_context *pipe, lastLevel = firstLevel + MAX2(MAX2(l2width, l2height), l2depth); } - assert(!stObj->mt); + assert(!stObj->pt); if (stImage->base.IsCompressed) comp_byte = compressed_num_bytes(stImage->base.TexFormat->MesaFormat); - stObj->mt = st_miptree_create(pipe, + stObj->pt = st_texture_create(st, gl_target_to_pipe(stObj->base.Target), - stImage->base.InternalFormat, + st_mesa_format_to_pipe_format(stImage->base.TexFormat->MesaFormat), + stImage->base.InternalFormat, firstLevel, lastLevel, width, height, depth, - stImage->base.TexFormat->TexelBytes, comp_byte); - stObj->mt->format - = st_mesa_format_to_pipe_format(stImage->base.TexFormat->MesaFormat); - DBG("%s - success\n", __FUNCTION__); } @@ -482,11 +457,11 @@ try_pbo_upload(GLcontext *ctx, else src_stride = width; - dst_offset = st_miptree_image_offset(stImage->mt, + dst_offset = st_texture_image_offset(stImage->pt, stImage->face, stImage->level); - dst_stride = stImage->mt->pitch; + dst_stride = stImage->pt->pitch; { struct _DriBufferObject *src_buffer = @@ -495,11 +470,11 @@ try_pbo_upload(GLcontext *ctx, /* Temporary hack: cast to _DriBufferObject: */ struct _DriBufferObject *dst_buffer = - (struct _DriBufferObject *)stImage->mt->region->buffer; + (struct _DriBufferObject *)stImage->pt->region->buffer; intelEmitCopyBlit(intel, - stImage->mt->cpp, + stImage->pt->cpp, src_stride, src_buffer, src_offset, dst_stride, dst_buffer, dst_offset, 0, 0, 0, 0, width, height, @@ -540,7 +515,6 @@ st_TexImage(GLcontext * ctx, struct gl_texture_object *texObj, struct gl_texture_image *texImage, GLsizei imageSize, int compressed) { - struct pipe_context *pipe = ctx->st->pipe; struct st_texture_object *stObj = st_texture_object(texObj); struct st_texture_image *stImage = st_texture_image(texImage); GLint postConvWidth = width; @@ -589,58 +563,58 @@ st_TexImage(GLcontext * ctx, /* Release the reference to a potentially orphaned buffer. * Release any old malloced memory. */ - if (stImage->mt) { - st_miptree_release(pipe, &stImage->mt); + if (stImage->pt) { + ctx->st->pipe->texture_release(ctx->st->pipe, &stImage->pt); assert(!texImage->Data); } else if (texImage->Data) { _mesa_align_free(texImage->Data); } - /* If this is the only texture image in the tree, could call + /* If this is the only texture image in the texture, could call * bmBufferData with NULL data to free the old block and avoid * waiting on any outstanding fences. */ - if (stObj->mt && - stObj->mt->first_level == level && - stObj->mt->last_level == level && - stObj->mt->target != PIPE_TEXTURE_CUBE && - !st_miptree_match_image(stObj->mt, &stImage->base, + if (stObj->pt && + stObj->pt->first_level == level && + stObj->pt->last_level == level && + stObj->pt->target != PIPE_TEXTURE_CUBE && + !st_texture_match_image(stObj->pt, &stImage->base, stImage->face, stImage->level)) { DBG("release it\n"); - st_miptree_release(pipe, &stObj->mt); - assert(!stObj->mt); + ctx->st->pipe->texture_release(ctx->st->pipe, &stObj->pt); + assert(!stObj->pt); } - if (!stObj->mt) { - guess_and_alloc_mipmap_tree(pipe, stObj, stImage); - if (!stObj->mt) { - DBG("guess_and_alloc_mipmap_tree: failed\n"); + if (!stObj->pt) { + guess_and_alloc_texture(ctx->st, stObj, stImage); + if (!stObj->pt) { + DBG("guess_and_alloc_texture: failed\n"); } } - assert(!stImage->mt); + assert(!stImage->pt); - if (stObj->mt && - st_miptree_match_image(stObj->mt, &stImage->base, + if (stObj->pt && + st_texture_match_image(stObj->pt, &stImage->base, stImage->face, stImage->level)) { - st_miptree_reference(&stImage->mt, stObj->mt); - assert(stImage->mt); + pipe_texture_reference(ctx->st->pipe, &stImage->pt, stObj->pt); + assert(stImage->pt); } - if (!stImage->mt) - DBG("XXX: Image did not fit into tree - storing in local memory!\n"); + if (!stImage->pt) + DBG("XXX: Image did not fit into texture - storing in local memory!\n"); #if 0 /* XXX FIX when st_buffer_objects are in place */ /* PBO fastpaths: */ if (dims <= 2 && - stImage->mt && + stImage->pt && intel_buffer_object(unpack->BufferObj) && check_pbo_format(internalFormat, format, - type, stImage->base.TexFormat)) { + type, texImage->TexFormat)) { DBG("trying pbo upload\n"); @@ -650,9 +624,9 @@ st_TexImage(GLcontext * ctx, * performance (in particular when pipe_region_cow() is * required). */ - if (stObj->mt == stImage->mt && - stObj->mt->first_level == level && - stObj->mt->last_level == level) { + if (stObj->pt == stImage->pt && + stObj->pt->first_level == level && + stObj->pt->last_level == level) { if (try_pbo_zcopy(intel, stImage, unpack, internalFormat, @@ -683,7 +657,7 @@ st_TexImage(GLcontext * ctx, /* intelCopyTexImage calls this function with pixels == NULL, with - * the expectation that the mipmap tree will be set up but nothing + * the expectation that the texture will be set up but nothing * more will be done. This is where those calls return: */ if (compressed) { @@ -698,13 +672,9 @@ st_TexImage(GLcontext * ctx, if (!pixels) return; - if (stImage->mt) { - texImage->Data = st_miptree_image_map(pipe, - stImage->mt, - stImage->face, - stImage->level, - &dstRowStride, - stImage->base.ImageOffsets); + if (stImage->pt) { + texImage->Data = st_texture_image_map(ctx->st, stImage, 0); + dstRowStride = stImage->surface->pitch * stImage->surface->cpp; } else { /* Allocate regular memory and store the image there temporarily. */ @@ -732,22 +702,36 @@ st_TexImage(GLcontext * ctx, if (compressed) { memcpy(texImage->Data, pixels, imageSize); } - else if (!texImage->TexFormat->StoreImage(ctx, dims, - texImage->_BaseFormat, - texImage->TexFormat, - texImage->Data, - 0, 0, 0, /* dstX/Y/Zoffset */ - dstRowStride, - texImage->ImageOffsets, - width, height, depth, - format, type, pixels, unpack)) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); + else { + GLuint srcImageStride = _mesa_image_image_stride(unpack, width, height, + format, type); + int i; + + for (i = 0; i++ < depth;) { + if (!texImage->TexFormat->StoreImage(ctx, dims, + texImage->_BaseFormat, + texImage->TexFormat, + texImage->Data, + 0, 0, 0, /* dstX/Y/Zoffset */ + dstRowStride, + texImage->ImageOffsets, + width, height, 1, + format, type, pixels, unpack)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); + } + + if (stImage->pt && i < depth) { + st_texture_image_unmap(ctx->st, stImage); + texImage->Data = st_texture_image_map(ctx->st, stImage, i); + pixels += srcImageStride; + } + } } _mesa_unmap_teximage_pbo(ctx, unpack); - if (stImage->mt) { - st_miptree_image_unmap(pipe, stImage->mt); + if (stImage->pt) { + st_texture_image_unmap(ctx->st, stImage); texImage->Data = NULL; } @@ -839,49 +823,58 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, /* struct intel_context *intel = intel_context(ctx); */ - struct pipe_context *pipe = ctx->st->pipe; struct st_texture_image *stImage = st_texture_image(texImage); + GLuint dstImageStride = _mesa_image_image_stride(&ctx->Pack, texImage->Width, + texImage->Height, format, + type); + GLuint depth; + int i; /* Map */ - if (stImage->mt) { + if (stImage->pt) { /* Image is stored in hardware format in a buffer managed by the * kernel. Need to explicitly map and unmap it. */ - stImage->base.Data = - st_miptree_image_map(pipe, - stImage->mt, - stImage->face, - stImage->level, - &stImage->base.RowStride, - stImage->base.ImageOffsets); - stImage->base.RowStride /= stImage->mt->cpp; + texImage->Data = st_texture_image_map(ctx->st, stImage, 0); + texImage->RowStride = stImage->surface->pitch; } else { /* Otherwise, the image should actually be stored in - * stImage->base.Data. This is pretty confusing for + * texImage->Data. This is pretty confusing for * everybody, I'd much prefer to separate the two functions of * texImage->Data - storage for texture images in main memory * and access (ie mappings) of images. In other words, we'd * create a new texImage->Map field and leave Data simply for * storage. */ - assert(stImage->base.Data); + assert(texImage->Data); } + depth = texImage->Depth; + texImage->Depth = 1; - if (compressed) { - _mesa_get_compressed_teximage(ctx, target, level, pixels, - texObj, texImage); - } else { - _mesa_get_teximage(ctx, target, level, format, type, pixels, - texObj, texImage); + for (i = 0; i++ < depth;) { + if (compressed) { + _mesa_get_compressed_teximage(ctx, target, level, pixels, + texObj, texImage); + } else { + _mesa_get_teximage(ctx, target, level, format, type, pixels, + texObj, texImage); + } + + if (stImage->pt && i < depth) { + st_texture_image_unmap(ctx->st, stImage); + texImage->Data = st_texture_image_map(ctx->st, stImage, i); + pixels += dstImageStride; + } } - + + texImage->Depth = depth; /* Unmap */ - if (stImage->mt) { - st_miptree_image_unmap(pipe, stImage->mt); - stImage->base.Data = NULL; + if (stImage->pt) { + st_texture_image_unmap(ctx->st, stImage); + texImage->Data = NULL; } } @@ -921,9 +914,11 @@ st_TexSubimage(GLcontext * ctx, struct gl_texture_object *texObj, struct gl_texture_image *texImage) { - struct pipe_context *pipe = ctx->st->pipe; struct st_texture_image *stImage = st_texture_image(texImage); GLuint dstRowStride; + GLuint srcImageStride = _mesa_image_image_stride(packing, width, height, + format, type); + int i; DBG("%s target %s level %d offset %d,%d %dx%d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), @@ -938,25 +933,28 @@ st_TexSubimage(GLcontext * ctx, /* Map buffer if necessary. Need to lock to prevent other contexts * from uploading the buffer under us. */ - if (stImage->mt) - texImage->Data = st_miptree_image_map(pipe, - stImage->mt, - stImage->face, - stImage->level, - &dstRowStride, - texImage->ImageOffsets); - - assert(dstRowStride); - - if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat, - texImage->TexFormat, - texImage->Data, - xoffset, yoffset, zoffset, - dstRowStride, - texImage->ImageOffsets, - width, height, depth, - format, type, pixels, packing)) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage"); + if (stImage->pt) { + texImage->Data = st_texture_image_map(ctx->st, stImage, zoffset); + dstRowStride = stImage->surface->pitch * stImage->surface->cpp; + } + + for (i = 0; i++ < depth;) { + if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat, + texImage->TexFormat, + texImage->Data, + xoffset, yoffset, 0, + dstRowStride, + texImage->ImageOffsets, + width, height, 1, + format, type, pixels, packing)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage"); + } + + if (stImage->pt && i < depth) { + st_texture_image_unmap(ctx->st, stImage); + texImage->Data = st_texture_image_map(ctx->st, stImage, zoffset + i); + pixels += srcImageStride; + } } #if 0 @@ -970,8 +968,8 @@ st_TexSubimage(GLcontext * ctx, _mesa_unmap_teximage_pbo(ctx, packing); - if (stImage->mt) { - st_miptree_image_unmap(pipe, stImage->mt); + if (stImage->pt) { + st_texture_image_unmap(ctx->st, stImage); texImage->Data = NULL; } } @@ -1074,7 +1072,7 @@ fallback_copy_texsubimage(GLcontext *ctx, { struct pipe_context *pipe = ctx->st->pipe; const uint face = texture_face(target); - struct pipe_mipmap_tree *mt = stImage->mt; + struct pipe_texture *pt = stImage->pt; struct pipe_surface *src_surf, *dest_surf; GLfloat *data; GLint row, yStep; @@ -1090,7 +1088,7 @@ fallback_copy_texsubimage(GLcontext *ctx, src_surf = strb->surface; - dest_surf = pipe->get_tex_surface(pipe, mt, + dest_surf = pipe->get_tex_surface(pipe, pt, face, level, destZ); (void) pipe->region_map(pipe, dest_surf->region); @@ -1150,7 +1148,7 @@ do_copy_texsubimage(GLcontext *ctx, struct gl_framebuffer *fb = ctx->ReadBuffer; struct st_renderbuffer *strb; struct pipe_context *pipe = ctx->st->pipe; - struct pipe_region *src_region, *dest_region; + struct pipe_surface *dest_surface; uint dest_format, src_format; (void) texImage; @@ -1169,29 +1167,24 @@ do_copy_texsubimage(GLcontext *ctx, assert(strb); assert(strb->surface); - assert(stImage->mt); + assert(stImage->pt); if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { srcY = strb->Base.Height - srcY - height; } src_format = strb->surface->format; - dest_format = stImage->mt->format; + dest_format = stImage->pt->format; - src_region = strb->surface->region; - dest_region = stImage->mt->region; + dest_surface = pipe->get_tex_surface(pipe, stImage->pt, stImage->face, + stImage->level, destZ); if (src_format == dest_format && ctx->_ImageTransferState == 0x0 && - src_region && - dest_region && - strb->surface->cpp == stImage->mt->cpp) { + strb->surface->region && + dest_surface->region && + strb->surface->cpp == stImage->pt->cpp) { /* do blit-style copy */ - struct pipe_surface *dest_surface = pipe->get_tex_surface(pipe, - stImage->mt, - stImage->face, - stImage->level, - destZ); /* XXX may need to invert image depending on window * vs. user-created FBO @@ -1203,12 +1196,12 @@ do_copy_texsubimage(GLcontext *ctx, * worth it: */ intelEmitCopyBlit(intel, - stImage->mt->cpp, + stImage->pt->cpp, -src->pitch, src->buffer, src->height * src->pitch * src->cpp, - stImage->mt->pitch, - stImage->mt->region->buffer, + stImage->pt->pitch, + stImage->pt->region->buffer, dest_offset, x, y + height, dstx, dsty, width, height, GL_COPY); /* ? */ @@ -1224,8 +1217,6 @@ do_copy_texsubimage(GLcontext *ctx, /* size */ width, height); #endif - - pipe_surface_reference(&dest_surface, NULL); } else { fallback_copy_texsubimage(ctx, target, level, @@ -1234,6 +1225,7 @@ do_copy_texsubimage(GLcontext *ctx, srcX, srcY, width, height); } + pipe_surface_reference(&dest_surface, NULL); #if 0 /* GL_SGIS_generate_mipmap -- this can be accelerated now. @@ -1267,7 +1259,7 @@ st_CopyTexImage1D(GLcontext * ctx, GLenum target, GLint level, goto fail; #endif - /* Setup or redefine the texture object, mipmap tree and texture + /* Setup or redefine the texture object, texture and texture * image. Don't populate yet. */ ctx->Driver.TexImage1D(ctx, target, level, internalFormat, @@ -1299,7 +1291,7 @@ st_CopyTexImage2D(GLcontext * ctx, GLenum target, GLint level, goto fail; #endif - /* Setup or redefine the texture object, mipmap tree and texture + /* Setup or redefine the texture object, texture and texture * image. Don't populate yet. */ ctx->Driver.TexImage2D(ctx, target, level, internalFormat, @@ -1407,28 +1399,28 @@ calculate_first_last_level(struct st_texture_object *stObj) static void -copy_image_data_to_tree(struct pipe_context *pipe, - struct st_texture_object *stObj, - struct st_texture_image *stImage) +copy_image_data_to_texture(struct st_context *st, + struct st_texture_object *stObj, + struct st_texture_image *stImage) { - if (stImage->mt) { + if (stImage->pt) { /* Copy potentially with the blitter: */ - st_miptree_image_copy(pipe, - stObj->mt, /* dest miptree */ + st_texture_image_copy(st->pipe, + stObj->pt, /* dest texture */ stImage->face, stImage->level, - stImage->mt /* src miptree */ + stImage->pt /* src texture */ ); - st_miptree_release(pipe, &stImage->mt); + st->pipe->texture_release(st->pipe, &stImage->pt); } else { assert(stImage->base.Data != NULL); /* More straightforward upload. */ - st_miptree_image_data(pipe, - stObj->mt, + st_texture_image_data(st->pipe, + stObj->pt, stImage->face, stImage->level, stImage->base.Data, @@ -1439,16 +1431,16 @@ copy_image_data_to_tree(struct pipe_context *pipe, stImage->base.Data = NULL; } - st_miptree_reference(&stImage->mt, stObj->mt); + pipe_texture_reference(st->pipe, &stImage->pt, stObj->pt); } /* */ GLboolean -st_finalize_mipmap_tree(GLcontext *ctx, - struct pipe_context *pipe, GLuint unit, - GLboolean *needFlush) +st_finalize_texture(GLcontext *ctx, + struct pipe_context *pipe, GLuint unit, + GLboolean *needFlush) { struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current; struct st_texture_object *stObj = st_texture_object(tObj); @@ -1465,7 +1457,7 @@ st_finalize_mipmap_tree(GLcontext *ctx, */ assert(stObj->base._Complete); - /* What levels must the tree include at a minimum? + /* What levels must the texture include at a minimum? */ calculate_first_last_level(stObj); firstImage = @@ -1474,27 +1466,27 @@ st_finalize_mipmap_tree(GLcontext *ctx, /* Fallback case: */ if (firstImage->base.Border) { - if (stObj->mt) { - st_miptree_release(pipe, &stObj->mt); + if (stObj->pt) { + ctx->st->pipe->texture_release(ctx->st->pipe, &stObj->pt); } return GL_FALSE; } - /* If both firstImage and stObj have a tree which can contain + /* If both firstImage and stObj point to a texture which can contain * all active images, favour firstImage. Note that because of the * completeness requirement, we know that the image dimensions * will match. */ - if (firstImage->mt && - firstImage->mt != stObj->mt && - firstImage->mt->first_level <= stObj->firstLevel && - firstImage->mt->last_level >= stObj->lastLevel) { + if (firstImage->pt && + firstImage->pt != stObj->pt && + firstImage->pt->first_level <= stObj->firstLevel && + firstImage->pt->last_level >= stObj->lastLevel) { - if (stObj->mt) - st_miptree_release(pipe, &stObj->mt); + if (stObj->pt) + ctx->st->pipe->texture_release(ctx->st->pipe, &stObj->pt); - st_miptree_reference(&stObj->mt, firstImage->mt); + pipe_texture_reference(ctx->st->pipe, &stObj->pt, firstImage->pt); } if (firstImage->base.IsCompressed) { @@ -1505,48 +1497,45 @@ st_finalize_mipmap_tree(GLcontext *ctx, cpp = firstImage->base.TexFormat->TexelBytes; } - /* Check tree can hold all active levels. Check tree matches + /* Check texture can hold all active levels. Check texture matches * target, imageFormat, etc. * * XXX: For some layouts (eg i945?), the test might have to be - * first_level == firstLevel, as the tree isn't valid except at the + * first_level == firstLevel, as the texture isn't valid except at the * original start level. Hope to get around this by * programming minLod, maxLod, baseLevel into the hardware and - * leaving the tree alone. + * leaving the texture alone. */ - if (stObj->mt && - (stObj->mt->target != gl_target_to_pipe(stObj->base.Target) || - stObj->mt->internal_format != firstImage->base.InternalFormat || - stObj->mt->first_level != stObj->firstLevel || - stObj->mt->last_level != stObj->lastLevel || - stObj->mt->width0 != firstImage->base.Width || - stObj->mt->height0 != firstImage->base.Height || - stObj->mt->depth0 != firstImage->base.Depth || - stObj->mt->cpp != cpp || - stObj->mt->compressed != firstImage->base.IsCompressed)) { - st_miptree_release(pipe, &stObj->mt); + if (stObj->pt && + (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) || + stObj->pt->internal_format != firstImage->base.InternalFormat || + stObj->pt->first_level != stObj->firstLevel || + stObj->pt->last_level != stObj->lastLevel || + stObj->pt->width[0] != firstImage->base.Width || + stObj->pt->height[0] != firstImage->base.Height || + stObj->pt->depth[0] != firstImage->base.Depth || + stObj->pt->cpp != cpp || + stObj->pt->compressed != firstImage->base.IsCompressed)) { + ctx->st->pipe->texture_release(ctx->st->pipe, &stObj->pt); } - /* May need to create a new tree: + /* May need to create a new texture: */ - if (!stObj->mt) { - stObj->mt = st_miptree_create(pipe, + if (!stObj->pt) { + stObj->pt = st_texture_create(ctx->st, gl_target_to_pipe(stObj->base.Target), - firstImage->base.InternalFormat, + st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat), + firstImage->base.InternalFormat, stObj->firstLevel, stObj->lastLevel, firstImage->base.Width, firstImage->base.Height, firstImage->base.Depth, - cpp, comp_byte); - - stObj->mt->format - = st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat); } - /* Pull in any images not in the object's tree: + /* Pull in any images not in the object's texture: */ nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1; for (face = 0; face < nr_faces; face++) { @@ -1554,10 +1543,10 @@ st_finalize_mipmap_tree(GLcontext *ctx, struct st_texture_image *stImage = st_texture_image(stObj->base.Image[face][i]); - /* Need to import images in main memory or held in other trees. + /* Need to import images in main memory or held in other textures. */ - if (stObj->mt != stImage->mt) { - copy_image_data_to_tree(pipe, stObj, stImage); + if (stObj->pt != stImage->pt) { + copy_image_data_to_texture(ctx->st, stObj, stImage); *needFlush = GL_TRUE; } } diff --git a/src/mesa/state_tracker/st_cb_texture.h b/src/mesa/state_tracker/st_cb_texture.h index 7a1867dc58a..7f8082b029d 100644 --- a/src/mesa/state_tracker/st_cb_texture.h +++ b/src/mesa/state_tracker/st_cb_texture.h @@ -2,14 +2,14 @@ #define ST_CB_TEXTURE_H -extern struct pipe_mipmap_tree * -st_get_texobj_mipmap_tree(struct gl_texture_object *texObj); +extern struct pipe_texture * +st_get_texobj_texture(struct gl_texture_object *texObj); extern GLboolean -st_finalize_mipmap_tree(GLcontext *ctx, - struct pipe_context *pipe, GLuint unit, - GLboolean *needFlush); +st_finalize_texture(GLcontext *ctx, + struct pipe_context *pipe, GLuint unit, + GLboolean *needFlush); extern void diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index a6045230a0f..db97014c5a7 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -36,7 +36,6 @@ struct st_context; struct st_region; struct st_texture_object; -struct st_texture_image; struct st_fragment_program; struct draw_context; struct draw_stage; @@ -61,6 +60,25 @@ struct st_tracked_state { +struct st_texture_image +{ + struct gl_texture_image base; + + /* These aren't stored in gl_texture_image + */ + GLuint level; + GLuint face; + + /* If stImage->pt != NULL, image data is stored here. + * Else if stImage->base.Data != NULL, image is stored there. + * Else there is no image data. + */ + struct pipe_texture *pt; + + struct pipe_surface *surface; +}; + + struct st_context { @@ -91,7 +109,7 @@ struct st_context struct pipe_constant_buffer constants[2]; struct pipe_feedback_state feedback; struct pipe_framebuffer_state framebuffer; - struct pipe_mipmap_tree *texture[PIPE_MAX_SAMPLERS]; + struct pipe_texture *texture[PIPE_MAX_SAMPLERS]; struct pipe_poly_stipple poly_stipple; struct pipe_scissor_state scissor; struct pipe_viewport_state viewport; diff --git a/src/mesa/state_tracker/st_mipmap_tree.c b/src/mesa/state_tracker/st_mipmap_tree.c deleted file mode 100644 index 6ccf33105f1..00000000000 --- a/src/mesa/state_tracker/st_mipmap_tree.c +++ /dev/null @@ -1,328 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "st_mipmap_tree.h" -#include "enums.h" - -#include "pipe/p_state.h" -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_inlines.h" -#include "pipe/p_winsys.h" - - -#define DBG if(0) printf - -#if 0 -static GLenum -target_to_target(GLenum target) -{ - switch (target) { - case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: - return GL_TEXTURE_CUBE_MAP_ARB; - default: - return target; - } -} -#endif - -struct pipe_mipmap_tree * -st_miptree_create(struct pipe_context *pipe, - unsigned target, - GLenum internal_format, - GLuint first_level, - GLuint last_level, - GLuint width0, - GLuint height0, - GLuint depth0, GLuint cpp, GLuint compress_byte) -{ - GLboolean ok; - struct pipe_mipmap_tree *mt = calloc(sizeof(*mt), 1); - GLbitfield flags = 0x0; - - assert(target <= PIPE_TEXTURE_CUBE); - - DBG("%s target %s format %s level %d..%d\n", __FUNCTION__, - _mesa_lookup_enum_by_nr(target), - _mesa_lookup_enum_by_nr(internal_format), first_level, last_level); - - mt->target = target; - mt->internal_format = internal_format; - mt->first_level = first_level; - mt->last_level = last_level; - mt->width0 = width0; - mt->height0 = height0; - mt->depth0 = depth0; - mt->cpp = compress_byte ? compress_byte : cpp; - mt->compressed = compress_byte ? 1 : 0; - mt->refcount = 1; - - ok = pipe->mipmap_tree_layout(pipe, mt); - if (ok) { - mt->region = pipe->winsys->region_alloc(pipe->winsys, - mt->pitch * mt->cpp * - mt->total_height, flags); - } - - if (!mt->region) { - free(mt); - return NULL; - } - - return mt; -} - - -void -st_miptree_reference(struct pipe_mipmap_tree **dst, - struct pipe_mipmap_tree *src) -{ - src->refcount++; - *dst = src; - DBG("%s %p refcount now %d\n", __FUNCTION__, (void *) src, src->refcount); -} - -void -st_miptree_release(struct pipe_context *pipe, - struct pipe_mipmap_tree **mt) -{ - if (!*mt) - return; - - DBG("%s %p refcount will be %d\n", - __FUNCTION__, (void *) *mt, (*mt)->refcount - 1); - if (--(*mt)->refcount <= 0) { - GLuint i; - - DBG("%s deleting %p\n", __FUNCTION__, (void *) *mt); - - pipe->winsys->region_release(pipe->winsys, &((*mt)->region)); - - for (i = 0; i < MAX_TEXTURE_LEVELS; i++) - if ((*mt)->level[i].image_offset) - free((*mt)->level[i].image_offset); - - free(*mt); - } - *mt = NULL; -} - - - - -/* Can the image be pulled into a unified mipmap tree. This mirrors - * the completeness test in a lot of ways. - * - * Not sure whether I want to pass gl_texture_image here. - */ -GLboolean -st_miptree_match_image(struct pipe_mipmap_tree *mt, - struct gl_texture_image *image, - GLuint face, GLuint level) -{ - /* Images with borders are never pulled into mipmap trees. - */ - if (image->Border) - return GL_FALSE; - - if (image->InternalFormat != mt->internal_format || - image->IsCompressed != mt->compressed) - return GL_FALSE; - - /* Test image dimensions against the base level image adjusted for - * minification. This will also catch images not present in the - * tree, changed targets, etc. - */ - if (image->Width != mt->level[level].width || - image->Height != mt->level[level].height || - image->Depth != mt->level[level].depth) - return GL_FALSE; - - return GL_TRUE; -} - - -/* Although we use the image_offset[] array to store relative offsets - * to cube faces, Mesa doesn't know anything about this and expects - * each cube face to be treated as a separate image. - * - * These functions present that view to mesa: - */ -const GLuint * -st_miptree_depth_offsets(struct pipe_mipmap_tree *mt, GLuint level) -{ - static const GLuint zero = 0; - - if (mt->target != PIPE_TEXTURE_3D || mt->level[level].nr_images == 1) - return &zero; - else - return mt->level[level].image_offset; -} - - -/** - * Return the offset to the given mipmap texture image within the - * texture memory buffer, in bytes. - */ -GLuint -st_miptree_image_offset(const struct pipe_mipmap_tree * mt, - GLuint face, GLuint level) -{ - if (mt->target == PIPE_TEXTURE_CUBE) - return (mt->level[level].level_offset + - mt->level[level].image_offset[face] * mt->cpp); - else - return mt->level[level].level_offset; -} - - -GLuint -st_miptree_texel_offset(const struct pipe_mipmap_tree * mt, - GLuint face, GLuint level, - GLuint col, GLuint row, GLuint img) -{ - GLuint imgOffset = st_miptree_image_offset(mt, face, level); - - return imgOffset + row * (mt->pitch + col) * mt->cpp; -} - - - -/** - * Map a teximage in a mipmap tree. - * \param row_stride returns row stride in bytes - * \param image_stride returns image stride in bytes (for 3D textures). - * \return address of mapping - */ -GLubyte * -st_miptree_image_map(struct pipe_context *pipe, - struct pipe_mipmap_tree * mt, - GLuint face, - GLuint level, - GLuint * row_stride, GLuint * image_offsets) -{ - GLubyte *ptr; - DBG("%s \n", __FUNCTION__); - - if (row_stride) - *row_stride = mt->pitch * mt->cpp; - - if (image_offsets) - memcpy(image_offsets, mt->level[level].image_offset, - mt->level[level].depth * sizeof(GLuint)); - - ptr = pipe->region_map(pipe, mt->region); - - return ptr + st_miptree_image_offset(mt, face, level); -} - -void -st_miptree_image_unmap(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt) -{ - DBG("%s\n", __FUNCTION__); - pipe->region_unmap(pipe, mt->region); -} - - - -/* Upload data for a particular image. - */ -void -st_miptree_image_data(struct pipe_context *pipe, - struct pipe_mipmap_tree *dst, - GLuint face, - GLuint level, - void *src, - GLuint src_row_pitch, GLuint src_image_pitch) -{ - GLuint depth = dst->level[level].depth; - GLuint i; - GLuint height = 0; - const GLubyte *srcUB = src; - struct pipe_surface *dst_surface; - - DBG("%s\n", __FUNCTION__); - for (i = 0; i < depth; i++) { - height = dst->level[level].height; - if(dst->compressed) - height /= 4; - - dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); - - pipe->surface_data(pipe, dst_surface, - 0, 0, /* dstx, dsty */ - srcUB, - src_row_pitch, - 0, 0, /* source x, y */ - dst->level[level].width, height); /* width, height */ - - pipe_surface_reference(&dst_surface, NULL); - - srcUB += src_image_pitch * dst->cpp; - } -} - -/* Copy mipmap image between trees - */ -void -st_miptree_image_copy(struct pipe_context *pipe, - struct pipe_mipmap_tree *dst, - GLuint face, GLuint level, - struct pipe_mipmap_tree *src) -{ - GLuint width = src->level[level].width; - GLuint height = src->level[level].height; - GLuint depth = src->level[level].depth; - struct pipe_surface *src_surface; - struct pipe_surface *dst_surface; - GLuint i; - - if (dst->compressed) - height /= 4; - for (i = 0; i < depth; i++) { - dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); - src_surface = pipe->get_tex_surface(pipe, src, face, level, i); - - pipe->surface_copy(pipe, - dst_surface, - 0, 0, /* destX, Y */ - src_surface, - 0, 0, /* srcX, Y */ - width, height); - - pipe_surface_reference(&dst_surface, NULL); - pipe_surface_reference(&src_surface, NULL); - } - -} diff --git a/src/mesa/state_tracker/st_mipmap_tree.h b/src/mesa/state_tracker/st_mipmap_tree.h deleted file mode 100644 index 3e7fd7fa0ce..00000000000 --- a/src/mesa/state_tracker/st_mipmap_tree.h +++ /dev/null @@ -1,118 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#ifndef ST_MIPMAP_TREE_H -#define ST_MIPMAP_TREE_H - - -#include "main/mtypes.h" - -struct pipe_context; -struct pipe_mipmap_tree; -struct pipe_region; - - -extern struct pipe_mipmap_tree * -st_miptree_create(struct pipe_context *pipe, - GLenum target, - GLenum internal_format, - GLuint first_level, - GLuint last_level, - GLuint width0, - GLuint height0, - GLuint depth0, - GLuint cpp, - GLuint compress_byte); - -extern void -st_miptree_reference(struct pipe_mipmap_tree **dst, - struct pipe_mipmap_tree *src); - -extern void -st_miptree_release(struct pipe_context *pipe, struct pipe_mipmap_tree **mt); - - -/* Check if an image fits an existing mipmap tree layout - */ -extern GLboolean -st_miptree_match_image(struct pipe_mipmap_tree *mt, - struct gl_texture_image *image, - GLuint face, GLuint level); - -/* Return a pointer to an image within a tree. Return image stride as - * well. - */ -extern GLubyte * -st_miptree_image_map(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt, - GLuint face, GLuint level, - GLuint * row_stride, GLuint * image_stride); - -extern void -st_miptree_image_unmap(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt); - - -/* Return pointers to each 2d slice within an image. Indexed by depth - * value. - */ -extern const GLuint * -st_miptree_depth_offsets(struct pipe_mipmap_tree *mt, GLuint level); - - -/* Return the linear offset of an image relative to the start of the - * tree: - */ -extern GLuint -st_miptree_image_offset(const struct pipe_mipmap_tree *mt, - GLuint face, GLuint level); - -extern GLuint -st_miptree_texel_offset(const struct pipe_mipmap_tree * mt, - GLuint face, GLuint level, - GLuint col, GLuint row, GLuint img); - - -/* Upload an image into a tree - */ -extern void -st_miptree_image_data(struct pipe_context *pipe, - struct pipe_mipmap_tree *dst, - GLuint face, GLuint level, void *src, - GLuint src_row_pitch, GLuint src_image_pitch); - - -/* Copy an image between two trees - */ -extern void -st_miptree_image_copy(struct pipe_context *pipe, - struct pipe_mipmap_tree *dst, - GLuint face, GLuint level, - struct pipe_mipmap_tree *src); - - -#endif diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c new file mode 100644 index 00000000000..a5582c31c0b --- /dev/null +++ b/src/mesa/state_tracker/st_texture.c @@ -0,0 +1,276 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "st_context.h" +#include "st_format.h" +#include "st_texture.h" +#include "enums.h" + +#include "pipe/p_state.h" +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" +#include "pipe/p_winsys.h" + + +#define DBG if(0) printf + +#if 0 +static GLenum +target_to_target(GLenum target) +{ + switch (target) { + case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: + return GL_TEXTURE_CUBE_MAP_ARB; + default: + return target; + } +} +#endif + +struct pipe_texture * +st_texture_create(struct st_context *st, + unsigned target, + unsigned format, + GLenum internal_format, + GLuint first_level, + GLuint last_level, + GLuint width0, + GLuint height0, + GLuint depth0, + GLuint compress_byte) +{ + struct pipe_texture *pt = CALLOC_STRUCT(pipe_texture); + + assert(target <= PIPE_TEXTURE_CUBE); + + DBG("%s target %s format %s level %d..%d\n", __FUNCTION__, + _mesa_lookup_enum_by_nr(target), + _mesa_lookup_enum_by_nr(internal_format), first_level, last_level); + + if (!pt) + return NULL; + + assert(format); + + pt->target = target; + pt->format = format; + pt->internal_format = internal_format; + pt->first_level = first_level; + pt->last_level = last_level; + pt->width[0] = width0; + pt->height[0] = height0; + pt->depth[0] = depth0; + pt->compressed = compress_byte ? 1 : 0; + pt->cpp = pt->compressed ? compress_byte : st_sizeof_format(format); + pt->refcount = 1; + + st->pipe->texture_create(st->pipe, &pt); + + return pt; +} + + + + +/* Can the image be pulled into a unified mipmap texture. This mirrors + * the completeness test in a lot of ways. + * + * Not sure whether I want to pass gl_texture_image here. + */ +GLboolean +st_texture_match_image(struct pipe_texture *pt, + struct gl_texture_image *image, + GLuint face, GLuint level) +{ + /* Images with borders are never pulled into mipmap textures. + */ + if (image->Border) + return GL_FALSE; + + if (image->InternalFormat != pt->internal_format || + image->IsCompressed != pt->compressed) + return GL_FALSE; + + /* Test image dimensions against the base level image adjusted for + * minification. This will also catch images not present in the + * texture, changed targets, etc. + */ + if (image->Width != pt->width[level] || + image->Height != pt->height[level] || + image->Depth != pt->depth[level]) + return GL_FALSE; + + return GL_TRUE; +} + + +#if 000 +/* Although we use the image_offset[] array to store relative offsets + * to cube faces, Mesa doesn't know anything about this and expects + * each cube face to be treated as a separate image. + * + * These functions present that view to mesa: + */ +const GLuint * +st_texture_depth_offsets(struct pipe_texture *pt, GLuint level) +{ + static const GLuint zero = 0; + + if (pt->target != PIPE_TEXTURE_3D || pt->level[level].nr_images == 1) + return &zero; + else + return pt->level[level].image_offset; +} + + +/** + * Return the offset to the given mipmap texture image within the + * texture memory buffer, in bytes. + */ +GLuint +st_texture_image_offset(const struct pipe_texture * pt, + GLuint face, GLuint level) +{ + if (pt->target == PIPE_TEXTURE_CUBE) + return (pt->level[level].level_offset + + pt->level[level].image_offset[face] * pt->cpp); + else + return pt->level[level].level_offset; +} +#endif + + +/** + * Map a teximage in a mipmap texture. + * \param row_stride returns row stride in bytes + * \param image_stride returns image stride in bytes (for 3D textures). + * \return address of mapping + */ +GLubyte * +st_texture_image_map(struct st_context *st, struct st_texture_image *stImage, + GLuint zoffset) +{ + struct pipe_texture *pt = stImage->pt; + DBG("%s \n", __FUNCTION__); + + stImage->surface = st->pipe->get_tex_surface(st->pipe, pt, stImage->face, + stImage->level, zoffset); + + (void) st->pipe->region_map(st->pipe, stImage->surface->region); + + return stImage->surface->region->map + stImage->surface->offset; +} + +void +st_texture_image_unmap(struct st_context *st, struct st_texture_image *stImage) +{ + DBG("%s\n", __FUNCTION__); + + st->pipe->region_unmap(st->pipe, stImage->surface->region); + + pipe_surface_reference(&stImage->surface, NULL); +} + + + +/* Upload data for a particular image. + */ +void +st_texture_image_data(struct pipe_context *pipe, + struct pipe_texture *dst, + GLuint face, + GLuint level, + void *src, + GLuint src_row_pitch, GLuint src_image_pitch) +{ + GLuint depth = dst->depth[level]; + GLuint i; + GLuint height = 0; + const GLubyte *srcUB = src; + struct pipe_surface *dst_surface; + + DBG("%s\n", __FUNCTION__); + for (i = 0; i < depth; i++) { + height = dst->height[level]; + if(dst->compressed) + height /= 4; + + dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); + + pipe->surface_data(pipe, dst_surface, + 0, 0, /* dstx, dsty */ + srcUB, + src_row_pitch, + 0, 0, /* source x, y */ + dst->width[level], height); /* width, height */ + + pipe_surface_reference(&dst_surface, NULL); + + srcUB += src_image_pitch * dst->cpp; + } +} + +/* Copy mipmap image between textures + */ +void +st_texture_image_copy(struct pipe_context *pipe, + struct pipe_texture *dst, + GLuint face, GLuint level, + struct pipe_texture *src) +{ + GLuint width = src->width[level]; + GLuint height = src->height[level]; + GLuint depth = src->depth[level]; + struct pipe_surface *src_surface; + struct pipe_surface *dst_surface; + GLuint i; + + if (dst->compressed) + height /= 4; + for (i = 0; i < depth; i++) { + dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); + src_surface = pipe->get_tex_surface(pipe, src, face, level, i); + + pipe->surface_copy(pipe, + dst_surface, + 0, 0, /* destX, Y */ + src_surface, + 0, 0, /* srcX, Y */ + width, height); + + pipe_surface_reference(&dst_surface, NULL); + pipe_surface_reference(&src_surface, NULL); + } + +} diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h new file mode 100644 index 00000000000..b25e3f3f3bd --- /dev/null +++ b/src/mesa/state_tracker/st_texture.h @@ -0,0 +1,109 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef ST_TEXTURE_H +#define ST_TEXTURE_H + + +#include "main/mtypes.h" + +struct pipe_context; +struct pipe_texture; +struct pipe_region; + + +extern struct pipe_texture * +st_texture_create(struct st_context *st, + unsigned target, + unsigned format, + GLenum internal_format, + GLuint first_level, + GLuint last_level, + GLuint width0, + GLuint height0, + GLuint depth0, + GLuint compress_byte); + + +/* Check if an image fits an existing texture + */ +extern GLboolean +st_texture_match_image(struct pipe_texture *pt, + struct gl_texture_image *image, + GLuint face, GLuint level); + +/* Return a pointer to an image within a texture. Return image stride as + * well. + */ +extern GLubyte * +st_texture_image_map(struct st_context *st, + struct st_texture_image *stImage, + GLuint zoffset); + +extern void +st_texture_image_unmap(struct st_context *st, + struct st_texture_image *stImage); + + +/* Return pointers to each 2d slice within an image. Indexed by depth + * value. + */ +extern const GLuint * +st_texture_depth_offsets(struct pipe_texture *pt, GLuint level); + + +/* Return the linear offset of an image relative to the start of its region: + */ +extern GLuint +st_texture_image_offset(const struct pipe_texture *pt, + GLuint face, GLuint level); + +extern GLuint +st_texture_texel_offset(const struct pipe_texture * pt, + GLuint face, GLuint level, + GLuint col, GLuint row, GLuint img); + + +/* Upload an image into a texture + */ +extern void +st_texture_image_data(struct pipe_context *pipe, + struct pipe_texture *dst, + GLuint face, GLuint level, void *src, + GLuint src_row_pitch, GLuint src_image_pitch); + + +/* Copy an image between two textures + */ +extern void +st_texture_image_copy(struct pipe_context *pipe, + struct pipe_texture *dst, + GLuint face, GLuint level, + struct pipe_texture *src); + + +#endif -- cgit v1.2.3 From b57632f04e72fb38305a963d25affd9c25b96fd2 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Thu, 6 Dec 2007 11:25:40 +0100 Subject: st_mesa_format_to_pipe_format: Handle MESA_FORMAT_ARGB4444. --- src/mesa/state_tracker/st_format.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index 6d056dbe832..b6b9375967e 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -256,6 +256,8 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat) case MESA_FORMAT_ARGB8888_REV: case MESA_FORMAT_ARGB8888: return PIPE_FORMAT_U_A8_R8_G8_B8; + case MESA_FORMAT_ARGB4444: + return PIPE_FORMAT_U_A4_R4_G4_B4; case MESA_FORMAT_AL88: return PIPE_FORMAT_U_A8_L8; case MESA_FORMAT_A8: -- cgit v1.2.3 From 7f984e1d2ab65a3d34ec0c0ef7487211dd644561 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Thu, 6 Dec 2007 11:30:34 +0100 Subject: Remove remnants of 'intel' from active state tracker code. --- src/mesa/state_tracker/st_cb_texture.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index f45d7e4275b..acc97df2e76 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -656,7 +656,7 @@ st_TexImage(GLcontext * ctx, #endif - /* intelCopyTexImage calls this function with pixels == NULL, with + /* st_CopyTexImage calls this function with pixels == NULL, with * the expectation that the texture will be set up but nothing * more will be done. This is where those calls return: */ @@ -820,9 +820,6 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, struct gl_texture_object *texObj, struct gl_texture_image *texImage, int compressed) { - /* - struct intel_context *intel = intel_context(ctx); - */ struct st_texture_image *stImage = st_texture_image(texImage); GLuint dstImageStride = _mesa_image_image_stride(&ctx->Pack, texImage->Width, texImage->Height, format, @@ -947,7 +944,7 @@ st_TexSubimage(GLcontext * ctx, texImage->ImageOffsets, width, height, 1, format, type, pixels, packing)) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "st_TexSubImage"); } if (stImage->pt && i < depth) { -- cgit v1.2.3 From 0a44a3fa2f9eb295106cd4c64549f55cc54f432f Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Fri, 7 Dec 2007 12:30:35 +0100 Subject: Eliminate struct pipe_region. Directly use struct pipe_buffer_handle for storage and struct pipe_surface for (un)mapping. --- .../drivers/dri/intel_winsys/intel_swapbuffers.c | 7 +- .../drivers/dri/intel_winsys/intel_winsys_pipe.c | 51 ++----------- src/mesa/pipe/failover/fo_context.c | 2 - src/mesa/pipe/i915simple/Makefile | 1 - src/mesa/pipe/i915simple/i915_context.c | 1 - src/mesa/pipe/i915simple/i915_context.h | 6 +- src/mesa/pipe/i915simple/i915_regions.c | 78 -------------------- src/mesa/pipe/i915simple/i915_state_emit.c | 6 +- src/mesa/pipe/i915simple/i915_surface.c | 35 ++++----- src/mesa/pipe/i915simple/i915_texture.c | 17 +++-- src/mesa/pipe/p_context.h | 10 +-- src/mesa/pipe/p_inlines.h | 53 +++++++------- src/mesa/pipe/p_state.h | 22 ++---- src/mesa/pipe/softpipe/Makefile | 1 - src/mesa/pipe/softpipe/sp_clear.c | 4 +- src/mesa/pipe/softpipe/sp_context.c | 84 ++++++---------------- src/mesa/pipe/softpipe/sp_draw_arrays.c | 2 - src/mesa/pipe/softpipe/sp_region.c | 79 -------------------- src/mesa/pipe/softpipe/sp_region.h | 40 ----------- src/mesa/pipe/softpipe/sp_state.h | 2 +- src/mesa/pipe/softpipe/sp_state_surface.c | 28 ++++---- src/mesa/pipe/softpipe/sp_surface.c | 61 ++++++++-------- src/mesa/pipe/softpipe/sp_texture.c | 18 +++-- src/mesa/pipe/softpipe/sp_tile_cache.c | 24 ++++++- src/mesa/pipe/softpipe/sp_tile_cache.h | 6 +- src/mesa/pipe/xlib/xm_api.c | 2 +- src/mesa/pipe/xlib/xm_buffer.c | 9 ++- src/mesa/pipe/xlib/xm_surface.c | 8 +-- src/mesa/pipe/xlib/xm_winsys.c | 50 +------------ src/mesa/state_tracker/st_cb_accum.c | 33 ++++----- src/mesa/state_tracker/st_cb_drawpixels.c | 48 ++++++------- src/mesa/state_tracker/st_cb_fbo.c | 28 ++++---- src/mesa/state_tracker/st_cb_readpixels.c | 13 ++-- src/mesa/state_tracker/st_cb_texture.c | 30 ++++---- src/mesa/state_tracker/st_context.c | 2 +- src/mesa/state_tracker/st_context.h | 5 +- src/mesa/state_tracker/st_texture.c | 9 ++- src/mesa/state_tracker/st_texture.h | 3 +- 38 files changed, 275 insertions(+), 603 deletions(-) delete mode 100644 src/mesa/pipe/i915simple/i915_regions.c delete mode 100644 src/mesa/pipe/softpipe/sp_region.c delete mode 100644 src/mesa/pipe/softpipe/sp_region.h diff --git a/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c b/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c index f96209d1b06..454cd71f6c0 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_swapbuffers.c @@ -88,12 +88,11 @@ intelDisplaySurface(__DRIdrawablePrivate *dPriv, const drm_clip_rect_t *pbox = dPriv->pClipRects; const int pitch = intelScreen->front.pitch / intelScreen->front.cpp; const int cpp = intelScreen->front.cpp; - const struct pipe_region *srcRegion = surf->region; const int srcpitch = surf->pitch; int BR13, CMD; int i; - ASSERT(srcRegion); + ASSERT(surf->buffer); ASSERT(surf->cpp == cpp); DBG(SWAP, "screen pitch %d src surface pitch %d\n", @@ -159,7 +158,7 @@ intelDisplaySurface(__DRIdrawablePrivate *dPriv, assert(box.x1 < box.x2); assert(box.y1 < box.y2); - /* XXX this could be done with pipe->region_copy() */ + /* XXX this could be done with pipe->surface_copy() */ BEGIN_BATCH(8, INTEL_BATCH_NO_CLIPRECTS); OUT_BATCH(CMD); OUT_BATCH(BR13); @@ -171,7 +170,7 @@ intelDisplaySurface(__DRIdrawablePrivate *dPriv, DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE, 0); OUT_BATCH((sbox.y1 << 16) | sbox.x1); OUT_BATCH((srcpitch * cpp) & 0xffff); - OUT_RELOC(dri_bo(srcRegion->buffer), + OUT_RELOC(dri_bo(surf->buffer), DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ, DRM_BO_MASK_MEM | DRM_BO_FLAG_READ, 0); diff --git a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c index 7b3aa994824..1b71d0ac107 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c @@ -135,13 +135,12 @@ static void intel_buffer_get_subdata(struct pipe_winsys *winsys, * for all buffers. */ static struct pipe_buffer_handle * -intel_buffer_create(struct pipe_winsys *winsys, - unsigned alignment) +intel_buffer_create(struct pipe_winsys *winsys, unsigned flags) { struct _DriBufferObject *buffer; struct intel_pipe_winsys *iws = intel_pipe_winsys(winsys); driGenBuffers( iws->regionPool, - "pipe buffer", 1, &buffer, alignment, 0, 0 ); + "pipe buffer", 1, &buffer, 64, 0, 0 ); return pipe_bo(buffer); } @@ -193,46 +192,6 @@ intel_i915_surface_pitch(struct pipe_winsys *winsys, } -static struct pipe_region * -intel_i915_region_alloc(struct pipe_winsys *winsys, - unsigned size, unsigned flags) -{ - struct pipe_region *region = calloc(sizeof(*region), 1); - const unsigned alignment = 64; - - region->refcount = 1; - - region->buffer = winsys->buffer_create( winsys, alignment ); - - winsys->buffer_data( winsys, - region->buffer, - size, - NULL, - PIPE_BUFFER_USAGE_PIXEL ); - - return region; -} - -static void -intel_i915_region_release(struct pipe_winsys *winsys, - struct pipe_region **region) -{ - if (!*region) - return; - - assert((*region)->refcount > 0); - (*region)->refcount--; - - if ((*region)->refcount == 0) { - assert((*region)->map_refcount == 0); - - winsys->buffer_reference( winsys, &((*region)->buffer), NULL ); - free(*region); - } - *region = NULL; -} - - static struct pipe_surface * intel_i915_surface_alloc(struct pipe_winsys *winsys, unsigned format) { @@ -252,8 +211,8 @@ intel_i915_surface_release(struct pipe_winsys *winsys, struct pipe_surface **s) struct pipe_surface *surf = *s; surf->refcount--; if (surf->refcount == 0) { - if (surf->region) - winsys->region_release(winsys, &surf->region); + if (surf->buffer) + winsys->buffer_reference(winsys, &surf->buffer, NULL); free(surf); } *s = NULL; @@ -300,8 +259,6 @@ intel_create_pipe_winsys( int fd ) iws->winsys.flush_frontbuffer = intel_flush_frontbuffer; iws->winsys.printf = intel_printf; iws->winsys.get_name = intel_get_name; - iws->winsys.region_alloc = intel_i915_region_alloc; - iws->winsys.region_release = intel_i915_region_release; iws->winsys.surface_pitch = intel_i915_surface_pitch; iws->winsys.surface_alloc = intel_i915_surface_alloc; iws->winsys.surface_release = intel_i915_surface_release; diff --git a/src/mesa/pipe/failover/fo_context.c b/src/mesa/pipe/failover/fo_context.c index a25563d4511..a1291f583b0 100644 --- a/src/mesa/pipe/failover/fo_context.c +++ b/src/mesa/pipe/failover/fo_context.c @@ -140,8 +140,6 @@ struct pipe_context *failover_create( struct pipe_context *hw, #endif failover->pipe.get_tex_surface = hw->get_tex_surface; - failover->pipe.region_map = hw->region_map; - failover->pipe.region_unmap = hw->region_unmap; failover->pipe.surface_data = hw->surface_data; failover->pipe.surface_copy = hw->surface_copy; failover->pipe.surface_fill = hw->surface_fill; diff --git a/src/mesa/pipe/i915simple/Makefile b/src/mesa/pipe/i915simple/Makefile index 1223b386a3d..2f91de3afc6 100644 --- a/src/mesa/pipe/i915simple/Makefile +++ b/src/mesa/pipe/i915simple/Makefile @@ -12,7 +12,6 @@ DRIVER_SOURCES = \ i915_context.c \ i915_debug.c \ i915_debug_fp.c \ - i915_regions.c \ i915_state.c \ i915_state_immediate.c \ i915_state_dynamic.c \ diff --git a/src/mesa/pipe/i915simple/i915_context.c b/src/mesa/pipe/i915simple/i915_context.c index 94649231cf3..07b331c5288 100644 --- a/src/mesa/pipe/i915simple/i915_context.c +++ b/src/mesa/pipe/i915simple/i915_context.c @@ -348,7 +348,6 @@ struct pipe_context *i915_create( struct pipe_winsys *pipe_winsys, draw_set_rasterize_stage(i915->draw, i915_draw_render_stage(i915)); } - i915_init_region_functions(i915); i915_init_surface_functions(i915); i915_init_state_functions(i915); i915_init_flush_functions(i915); diff --git a/src/mesa/pipe/i915simple/i915_context.h b/src/mesa/pipe/i915simple/i915_context.h index 8ed3465be23..dbf0c885cc8 100644 --- a/src/mesa/pipe/i915simple/i915_context.h +++ b/src/mesa/pipe/i915simple/i915_context.h @@ -175,7 +175,7 @@ struct i915_texture { /* The data is held here: */ - struct pipe_region *region; + struct pipe_buffer_handle *buffer; }; struct i915_context @@ -289,10 +289,10 @@ void i915_clear(struct pipe_context *pipe, struct pipe_surface *ps, /*********************************************************************** - * i915_region.c: + * i915_surface.c: */ -void i915_init_region_functions( struct i915_context *i915 ); void i915_init_surface_functions( struct i915_context *i915 ); + void i915_init_state_functions( struct i915_context *i915 ); void i915_init_flush_functions( struct i915_context *i915 ); void i915_init_string_functions( struct i915_context *i915 ); diff --git a/src/mesa/pipe/i915simple/i915_regions.c b/src/mesa/pipe/i915simple/i915_regions.c deleted file mode 100644 index 82fdec83d09..00000000000 --- a/src/mesa/pipe/i915simple/i915_regions.c +++ /dev/null @@ -1,78 +0,0 @@ -/************************************************************************** - * - * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/* Provide additional functionality on top of bufmgr buffers: - * - 2d semantics and blit operations (XXX: remove/simplify blits??) - * - refcounting of buffers for multiple images in a buffer. - * - refcounting of buffer mappings. - */ - -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "i915_context.h" - - - - -static ubyte * -i915_region_map(struct pipe_context *pipe, struct pipe_region *region) -{ - struct i915_context *i915 = i915_context( pipe ); - - if (!region->map_refcount++) { - region->map = i915->pipe.winsys->buffer_map( i915->pipe.winsys, - region->buffer, - PIPE_BUFFER_FLAG_WRITE | - PIPE_BUFFER_FLAG_READ); - } - - return region->map; -} - -static void -i915_region_unmap(struct pipe_context *pipe, struct pipe_region *region) -{ - struct i915_context *i915 = i915_context( pipe ); - - if (region->map_refcount > 0) { - assert(region->map); - if (!--region->map_refcount) { - i915->pipe.winsys->buffer_unmap( i915->pipe.winsys, - region->buffer ); - region->map = NULL; - } - } -} - - -void -i915_init_region_functions(struct i915_context *i915) -{ - i915->pipe.region_map = i915_region_map; - i915->pipe.region_unmap = i915_region_unmap; -} - diff --git a/src/mesa/pipe/i915simple/i915_state_emit.c b/src/mesa/pipe/i915simple/i915_state_emit.c index 900a91d8963..eda40d7c76c 100644 --- a/src/mesa/pipe/i915simple/i915_state_emit.c +++ b/src/mesa/pipe/i915simple/i915_state_emit.c @@ -220,7 +220,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) BUF_3D_PITCH(pitch) | /* pitch in bytes */ BUF_3D_USE_FENCE); - OUT_RELOC(cbuf_surface->region->buffer, + OUT_RELOC(cbuf_surface->buffer, I915_BUFFER_ACCESS_WRITE, 0); } @@ -236,7 +236,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) BUF_3D_PITCH(zpitch) | /* pitch in bytes */ BUF_3D_USE_FENCE); - OUT_RELOC(depth_surface->region->buffer, + OUT_RELOC(depth_surface->buffer, I915_BUFFER_ACCESS_WRITE, 0); } @@ -284,7 +284,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) for (unit = 0; unit < I915_TEX_UNITS; unit++) { if (enabled & (1 << unit)) { struct pipe_buffer_handle *buf = - i915->texture[unit]->region->buffer; + i915->texture[unit]->buffer; uint offset = 0; assert(buf); diff --git a/src/mesa/pipe/i915simple/i915_surface.c b/src/mesa/pipe/i915simple/i915_surface.c index 385202507db..ecbabe50eb5 100644 --- a/src/mesa/pipe/i915simple/i915_surface.c +++ b/src/mesa/pipe/i915simple/i915_surface.c @@ -29,6 +29,7 @@ #include "i915_blit.h" #include "i915_state.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" @@ -57,7 +58,7 @@ i915_get_tile_rgba(struct pipe_context *pipe, uint x, uint y, uint w, uint h, float *p) { const unsigned *src - = ((const unsigned *) (ps->region->map + ps->offset)) + = ((const unsigned *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -129,7 +130,7 @@ i915_get_tile(struct pipe_context *pipe, ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); CLIP_TILE; @@ -137,7 +138,7 @@ i915_get_tile(struct pipe_context *pipe, dst_stride = w0 * cpp; } - pSrc = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; + pSrc = ps->map + (y * ps->pitch + x) * cpp; pDest = (ubyte *) p; for (i = 0; i < h; i++) { @@ -163,7 +164,7 @@ i915_put_tile(struct pipe_context *pipe, ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); CLIP_TILE; @@ -172,7 +173,7 @@ i915_put_tile(struct pipe_context *pipe, } pSrc = (const ubyte *) p; - pDest = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; + pDest = ps->map + (y * ps->pitch + x) * cpp; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); @@ -211,7 +212,7 @@ i915_get_tex_surface(struct pipe_context *pipe, if (ps) { assert(ps->format); assert(ps->refcount); - pipe_region_reference(&ps->region, tex->region); + pipe->winsys->buffer_reference(pipe->winsys, &ps->buffer, tex->buffer); ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; @@ -274,12 +275,12 @@ i915_surface_data(struct pipe_context *pipe, const void *src, unsigned src_pitch, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - _mesa_copy_rect(pipe->region_map(pipe, dst->region) + dst->offset, + _mesa_copy_rect(pipe_surface_map(dst), dst->cpp, dst->pitch, dstx, dsty, width, height, src, src_pitch, srcx, srcy); - pipe->region_unmap(pipe, dst->region); + pipe_surface_unmap(dst); } @@ -297,23 +298,23 @@ i915_surface_copy(struct pipe_context *pipe, assert( dst->cpp == src->cpp ); if (0) { - _mesa_copy_rect(pipe->region_map(pipe, dst->region) + dst->offset, + _mesa_copy_rect(pipe_surface_map(dst), dst->cpp, dst->pitch, dstx, dsty, width, height, - pipe->region_map(pipe, src->region) + src->offset, + pipe_surface_map(src), src->pitch, srcx, srcy); - pipe->region_unmap(pipe, src->region); - pipe->region_unmap(pipe, dst->region); + pipe_surface_unmap(src); + pipe_surface_unmap(dst); } else { i915_copy_blit( i915_context(pipe), dst->cpp, - (short) src->pitch, src->region->buffer, src->offset, - (short) dst->pitch, dst->region->buffer, dst->offset, + (short) src->pitch, src->buffer, src->offset, + (short) dst->pitch, dst->buffer, dst->offset, (short) srcx, (short) srcy, (short) dstx, (short) dsty, (short) width, (short) height ); } } @@ -324,7 +325,7 @@ i915_surface_copy(struct pipe_context *pipe, static ubyte * get_pointer(struct pipe_surface *dst, unsigned x, unsigned y) { - return dst->region->map + (y * dst->pitch + x) * dst->cpp; + return dst->map + (y * dst->pitch + x) * dst->cpp; } @@ -337,7 +338,7 @@ i915_surface_fill(struct pipe_context *pipe, if (0) { unsigned i, j; - (void)pipe->region_map(pipe, dst->region); + (void)pipe_surface_map(dst); switch (dst->cpp) { case 1: { @@ -375,7 +376,7 @@ i915_surface_fill(struct pipe_context *pipe, i915_fill_blit( i915_context(pipe), dst->cpp, (short) dst->pitch, - dst->region->buffer, dst->offset, + dst->buffer, dst->offset, (short) dstx, (short) dsty, (short) width, (short) height, value ); diff --git a/src/mesa/pipe/i915simple/i915_texture.c b/src/mesa/pipe/i915simple/i915_texture.c index 3bfa806d9ec..1ca2815dfb9 100644 --- a/src/mesa/pipe/i915simple/i915_texture.c +++ b/src/mesa/pipe/i915simple/i915_texture.c @@ -33,6 +33,7 @@ #include "pipe/p_state.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #include "pipe/p_util.h" #include "pipe/p_winsys.h" @@ -494,13 +495,17 @@ i915_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) if (i915->flags.is_i945 ? i945_miptree_layout(pipe, tex) : i915_miptree_layout(pipe, tex)) { - tex->region = pipe->winsys->region_alloc(pipe->winsys, - tex->pitch * tex->base.cpp * - tex->total_height, - PIPE_SURFACE_FLAG_TEXTURE); + tex->buffer = pipe->winsys->buffer_create(pipe->winsys, + PIPE_SURFACE_FLAG_TEXTURE); + + if (tex->buffer) + pipe->winsys->buffer_data(pipe->winsys, tex->buffer, + tex->pitch * tex->base.cpp * + tex->total_height, NULL, + PIPE_BUFFER_USAGE_PIXEL); } - if (!tex->region) { + if (!tex->buffer) { FREE(tex); tex = NULL; } @@ -527,7 +532,7 @@ i915_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) DBG("%s deleting %p\n", __FUNCTION__, (void *) tex); */ - pipe->winsys->region_release(pipe->winsys, &tex->region); + pipe->winsys->buffer_reference(pipe->winsys, &tex->buffer, NULL); for (i = 0; i < PIPE_MAX_TEXTURE_LEVELS; i++) if (tex->image_offset[i]) diff --git a/src/mesa/pipe/p_context.h b/src/mesa/pipe/p_context.h index 5033209323f..2420d02213f 100644 --- a/src/mesa/pipe/p_context.h +++ b/src/mesa/pipe/p_context.h @@ -37,7 +37,7 @@ struct pipe_state_cache; * Gallium rendering context. Basically: * - state setting functions * - VBO drawing functions - * - memory region function + * - surface functions * - device queries */ struct pipe_context { @@ -201,14 +201,6 @@ struct pipe_context { uint x, uint y, uint w, uint h, const float *p); - /* - * Memory region functions - */ - ubyte *(*region_map)(struct pipe_context *pipe, struct pipe_region *r); - - void (*region_unmap)(struct pipe_context *pipe, struct pipe_region *r); - - /* * Surface functions */ diff --git a/src/mesa/pipe/p_inlines.h b/src/mesa/pipe/p_inlines.h index c04d46dddd8..1697284b049 100644 --- a/src/mesa/pipe/p_inlines.h +++ b/src/mesa/pipe/p_inlines.h @@ -29,40 +29,42 @@ #define P_INLINES_H #include "p_context.h" +#include "p_defines.h" #include "p_winsys.h" -/** - * Set 'ptr' to point to 'region' and update reference counting. - * The old thing pointed to, if any, will be unreferenced first. - * 'region' may be NULL. - */ +static INLINE ubyte * +pipe_surface_map(struct pipe_surface *surface) +{ + if (!surface->map_refcount++) { + surface->map = surface->winsys->buffer_map( surface->winsys, + surface->buffer, + PIPE_BUFFER_FLAG_WRITE | + PIPE_BUFFER_FLAG_READ ) + + surface->offset; + } + + return surface->map; +} + static INLINE void -pipe_region_reference(struct pipe_region **ptr, struct pipe_region *region) +pipe_surface_unmap(struct pipe_surface *surface) { - assert(ptr); - if (*ptr) { - /* unreference the old thing */ - struct pipe_region *oldReg = *ptr; - assert(oldReg->refcount > 0); - oldReg->refcount--; - if (oldReg->refcount == 0) { - /* free the old region */ - assert(oldReg->map_refcount == 0); - /* XXX dereference the region->buffer */ - FREE( oldReg ); + if (surface->map_refcount > 0) { + assert(surface->map); + if (!--surface->map_refcount) { + surface->winsys->buffer_unmap( surface->winsys, + surface->buffer ); + surface->map = NULL; } - *ptr = NULL; - } - if (region) { - /* reference the new thing */ - region->refcount++; - *ptr = region; } } + /** - * \sa pipe_region_reference + * Set 'ptr' to point to 'surf' and update reference counting. + * The old thing pointed to, if any, will be unreferenced first. + * 'surf' may be NULL. */ static INLINE void pipe_surface_reference(struct pipe_surface **ptr, struct pipe_surface *surf) @@ -82,7 +84,7 @@ pipe_surface_reference(struct pipe_surface **ptr, struct pipe_surface *surf) /** - * \sa pipe_region_reference + * \sa pipe_surface_reference */ static INLINE void pipe_texture_reference(struct pipe_context *pipe, struct pipe_texture **ptr, @@ -100,4 +102,5 @@ pipe_texture_reference(struct pipe_context *pipe, struct pipe_texture **ptr, } } + #endif /* P_INLINES_H */ diff --git a/src/mesa/pipe/p_state.h b/src/mesa/pipe/p_state.h index 077a8f5a06f..44dec9b7732 100644 --- a/src/mesa/pipe/p_state.h +++ b/src/mesa/pipe/p_state.h @@ -256,32 +256,20 @@ struct pipe_sampler_state }; -/*** - *** Resource Objects - ***/ - -struct pipe_region -{ - struct pipe_buffer_handle *buffer; /**< driver private buffer handle */ - - unsigned refcount; /**< Reference count for region */ - ubyte *map; /**< only non-NULL when region is actually mapped */ - unsigned map_refcount; /**< Reference count for mapping */ -}; - - /** - * 2D surface. This is basically a view into a pipe_region (memory buffer). + * 2D surface. This is basically a view into a memory buffer. * May be a renderbuffer, texture mipmap level, etc. */ struct pipe_surface { - struct pipe_region *region; + struct pipe_buffer_handle *buffer; /**< driver private buffer handle */ + ubyte *map; /**< only non-NULL when surface is actually mapped */ + unsigned map_refcount; /**< Reference count for mapping */ unsigned format; /**< PIPE_FORMAT_x */ unsigned cpp; /**< bytes per pixel */ unsigned width, height; unsigned pitch; /**< in pixels */ - unsigned offset; /**< offset from start of region, in bytes */ + unsigned offset; /**< offset from start of buffer, in bytes */ unsigned refcount; struct pipe_winsys *winsys; /**< winsys which owns/created the surface */ }; diff --git a/src/mesa/pipe/softpipe/Makefile b/src/mesa/pipe/softpipe/Makefile index 9978884c9bd..647cc05373b 100644 --- a/src/mesa/pipe/softpipe/Makefile +++ b/src/mesa/pipe/softpipe/Makefile @@ -24,7 +24,6 @@ DRIVER_SOURCES = \ sp_quad_output.c \ sp_quad_stencil.c \ sp_quad_stipple.c \ - sp_region.c \ sp_state_blend.c \ sp_state_clip.c \ sp_state_derived.c \ diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index a4276362b99..496b38fd5f1 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -50,12 +50,12 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, softpipe_update_derived(softpipe); /* not needed?? */ - if (ps == sp_tile_cache_get_surface(softpipe->zbuf_cache)) { + if (ps == sp_tile_cache_get_surface(softpipe, softpipe->zbuf_cache)) { float clear[4]; clear[0] = 1.0; /* XXX hack */ sp_tile_cache_clear(softpipe->zbuf_cache, clear); } - else if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[0])) { + else if (ps == sp_tile_cache_get_surface(softpipe, softpipe->cbuf_cache[0])) { float clear[4]; clear[0] = 0.2f; /* XXX hack */ clear[1] = 0.2f; /* XXX hack */ diff --git a/src/mesa/pipe/softpipe/sp_context.c b/src/mesa/pipe/softpipe/sp_context.c index 7a9fccce9ae..f7f0316cf2a 100644 --- a/src/mesa/pipe/softpipe/sp_context.c +++ b/src/mesa/pipe/softpipe/sp_context.c @@ -31,12 +31,12 @@ #include "pipe/draw/draw_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #include "pipe/p_util.h" #include "sp_clear.h" #include "sp_context.h" #include "sp_flush.h" #include "sp_prim_setup.h" -#include "sp_region.h" #include "sp_state.h" #include "sp_surface.h" #include "sp_tile_cache.h" @@ -65,44 +65,22 @@ softpipe_is_format_supported( struct pipe_context *pipe, uint format ) void softpipe_map_surfaces(struct softpipe_context *sp) { - struct pipe_context *pipe = &sp->pipe; + struct pipe_surface *ps; unsigned i; for (i = 0; i < sp->framebuffer.num_cbufs; i++) { - struct pipe_surface *ps = sp->framebuffer.cbufs[i]; - if (ps->region && !ps->region->map) { - pipe->region_map(pipe, ps->region); - } + ps = sp->framebuffer.cbufs[i]; + if (ps->buffer) + pipe_surface_map(ps); } - if (sp->framebuffer.zbuf) { - struct pipe_surface *ps = sp->framebuffer.zbuf; - if (ps->region && !ps->region->map) { - pipe->region_map(pipe, ps->region); - } - } - - if (sp->framebuffer.sbuf) { - struct pipe_surface *ps = sp->framebuffer.sbuf; - if (ps->region && !ps->region->map) { - pipe->region_map(pipe, ps->region); - } - } -} - + ps = sp->framebuffer.zbuf; + if (ps && ps->buffer) + pipe_surface_map(ps); -void -softpipe_map_texture_surfaces(struct softpipe_context *sp) -{ - struct pipe_context *pipe = &sp->pipe; - uint i; - - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - struct softpipe_texture *spt = sp->texture[i]; - if (spt) { - pipe->region_map(pipe, spt->region); - } - } + ps = sp->framebuffer.sbuf; + if (ps && ps->buffer) + pipe_surface_map(ps); } @@ -112,7 +90,7 @@ softpipe_map_texture_surfaces(struct softpipe_context *sp) void softpipe_unmap_surfaces(struct softpipe_context *sp) { - struct pipe_context *pipe = &sp->pipe; + struct pipe_surface *ps; uint i; for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) @@ -121,36 +99,18 @@ softpipe_unmap_surfaces(struct softpipe_context *sp) sp_flush_tile_cache(sp, sp->sbuf_cache); for (i = 0; i < sp->framebuffer.num_cbufs; i++) { - struct pipe_surface *ps = sp->framebuffer.cbufs[i]; - if (ps->region) - pipe->region_unmap(pipe, ps->region); + ps = sp->framebuffer.cbufs[i]; + if (ps->map) + pipe_surface_unmap(ps); } - if (sp->framebuffer.zbuf) { - struct pipe_surface *ps = sp->framebuffer.zbuf; - if (ps->region) - pipe->region_unmap(pipe, ps->region); - } - - if (sp->framebuffer.sbuf && sp->framebuffer.sbuf != sp->framebuffer.zbuf) { - struct pipe_surface *ps = sp->framebuffer.sbuf; - if (ps->region) - pipe->region_unmap(pipe, ps->region); - } -} - + ps = sp->framebuffer.zbuf; + if (ps && ps->map) + pipe_surface_unmap(ps); -void -softpipe_unmap_texture_surfaces(struct softpipe_context *sp) -{ - struct pipe_context *pipe = &sp->pipe; - uint i; - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - struct softpipe_texture *spt = sp->texture[i]; - if (spt) { - pipe->region_unmap(pipe, spt->region); - } - } + ps = sp->framebuffer.sbuf; + if (ps && ps->map) + pipe_surface_unmap(ps); } @@ -403,8 +363,6 @@ struct pipe_context *softpipe_create( struct pipe_winsys *pipe_winsys, draw_set_rasterize_stage(softpipe->draw, softpipe->setup); } - - sp_init_region_functions(softpipe); sp_init_surface_functions(softpipe); return &softpipe->pipe; diff --git a/src/mesa/pipe/softpipe/sp_draw_arrays.c b/src/mesa/pipe/softpipe/sp_draw_arrays.c index 8a82cdfe1a9..93eb68405db 100644 --- a/src/mesa/pipe/softpipe/sp_draw_arrays.c +++ b/src/mesa/pipe/softpipe/sp_draw_arrays.c @@ -112,7 +112,6 @@ softpipe_draw_elements(struct pipe_context *pipe, softpipe_update_derived( sp ); softpipe_map_surfaces(sp); - softpipe_map_texture_surfaces(sp); softpipe_map_constant_buffers(sp); /* @@ -184,7 +183,6 @@ softpipe_draw_elements(struct pipe_context *pipe, /* Note: leave drawing surfaces mapped */ - softpipe_unmap_texture_surfaces(sp); softpipe_unmap_constant_buffers(sp); return TRUE; diff --git a/src/mesa/pipe/softpipe/sp_region.c b/src/mesa/pipe/softpipe/sp_region.c deleted file mode 100644 index 58dc6bb96e4..00000000000 --- a/src/mesa/pipe/softpipe/sp_region.c +++ /dev/null @@ -1,79 +0,0 @@ -/************************************************************************** - * - * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/* Provide additional functionality on top of bufmgr buffers: - * - 2d semantics and blit operations (XXX: remove/simplify blits??) - * - refcounting of buffers for multiple images in a buffer. - * - refcounting of buffer mappings. - */ - -#include "sp_context.h" -#include "sp_region.h" -#include "pipe/p_util.h" -#include "pipe/p_winsys.h" -#include "pipe/p_defines.h" - - - -static ubyte * -sp_region_map(struct pipe_context *pipe, struct pipe_region *region) -{ - struct softpipe_context *sp = softpipe_context( pipe ); - - if (!region->map_refcount++) { - region->map = sp->pipe.winsys->buffer_map( sp->pipe.winsys, - region->buffer, - PIPE_BUFFER_FLAG_WRITE | - PIPE_BUFFER_FLAG_READ); - } - - return region->map; -} - -static void -sp_region_unmap(struct pipe_context *pipe, struct pipe_region *region) -{ - struct softpipe_context *sp = softpipe_context( pipe ); - - if (region->map_refcount > 0) { - assert(region->map); - if (!--region->map_refcount) { - sp->pipe.winsys->buffer_unmap( sp->pipe.winsys, - region->buffer ); - region->map = NULL; - } - } -} - - -void -sp_init_region_functions(struct softpipe_context *sp) -{ - sp->pipe.region_map = sp_region_map; - sp->pipe.region_unmap = sp_region_unmap; -} - diff --git a/src/mesa/pipe/softpipe/sp_region.h b/src/mesa/pipe/softpipe/sp_region.h deleted file mode 100644 index 432746b27f1..00000000000 --- a/src/mesa/pipe/softpipe/sp_region.h +++ /dev/null @@ -1,40 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - -#ifndef SP_REGION_H -#define SP_REGION_H - - -struct softpipe_context; - - -extern void -sp_init_region_functions(struct softpipe_context *sp); - - -#endif /* SP_REGION_H */ diff --git a/src/mesa/pipe/softpipe/sp_state.h b/src/mesa/pipe/softpipe/sp_state.h index a543735b525..daf9955ca5f 100644 --- a/src/mesa/pipe/softpipe/sp_state.h +++ b/src/mesa/pipe/softpipe/sp_state.h @@ -78,7 +78,7 @@ struct softpipe_texture /* The data is held here: */ - struct pipe_region *region; + struct pipe_buffer_handle *buffer; }; void * diff --git a/src/mesa/pipe/softpipe/sp_state_surface.c b/src/mesa/pipe/softpipe/sp_state_surface.c index d4e0bd1e157..9470ef485fe 100644 --- a/src/mesa/pipe/softpipe/sp_state_surface.c +++ b/src/mesa/pipe/softpipe/sp_state_surface.c @@ -27,6 +27,8 @@ /* Authors: Keith Whitwell */ +#include "p_inlines.h" + #include "sp_context.h" #include "sp_state.h" #include "sp_surface.h" @@ -54,17 +56,17 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, sp_flush_tile_cache(sp, sp->cbuf_cache[i]); /* unmap old */ ps = sp->framebuffer.cbufs[i]; - if (ps && ps->region) - pipe->region_unmap(pipe, ps->region); + if (ps && ps->map) + pipe_surface_unmap(ps); /* map new */ ps = fb->cbufs[i]; if (ps) - pipe->region_map(pipe, ps->region); + pipe_surface_map(ps); /* assign new */ sp->framebuffer.cbufs[i] = fb->cbufs[i]; /* update cache */ - sp_tile_cache_set_surface(sp->cbuf_cache[i], ps); + sp_tile_cache_set_surface(sp, sp->cbuf_cache[i], ps); } } @@ -76,8 +78,8 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, sp_flush_tile_cache(sp, sp->zbuf_cache); /* unmap old */ ps = sp->framebuffer.zbuf; - if (ps && ps->region) - pipe->region_unmap(pipe, ps->region); + if (ps && ps->map) + pipe_surface_unmap(ps); if (sp->framebuffer.sbuf == sp->framebuffer.zbuf) { /* combined z/stencil */ sp->framebuffer.sbuf = NULL; @@ -85,12 +87,12 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, /* map new */ ps = fb->zbuf; if (ps) - pipe->region_map(pipe, ps->region); + pipe_surface_map(ps); /* assign new */ sp->framebuffer.zbuf = fb->zbuf; /* update cache */ - sp_tile_cache_set_surface(sp->zbuf_cache, ps); + sp_tile_cache_set_surface(sp, sp->zbuf_cache, ps); } /* XXX combined depth/stencil here */ @@ -101,12 +103,12 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, sp_flush_tile_cache(sp, sp->sbuf_cache_sep); /* unmap old */ ps = sp->framebuffer.sbuf; - if (ps && ps->region) - pipe->region_unmap(pipe, ps->region); + if (ps && ps->map) + pipe_surface_unmap(ps); /* map new */ ps = fb->sbuf; if (ps && fb->sbuf != fb->zbuf) - pipe->region_map(pipe, ps->region); + pipe_surface_map(ps); /* assign new */ sp->framebuffer.sbuf = fb->sbuf; @@ -114,12 +116,12 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, if (fb->sbuf != fb->zbuf) { /* separate stencil buf */ sp->sbuf_cache = sp->sbuf_cache_sep; - sp_tile_cache_set_surface(sp->sbuf_cache, ps); + sp_tile_cache_set_surface(sp, sp->sbuf_cache, ps); } else { /* combined depth/stencil */ sp->sbuf_cache = sp->zbuf_cache; - sp_tile_cache_set_surface(sp->sbuf_cache, ps); + sp_tile_cache_set_surface(sp, sp->sbuf_cache, ps); } } diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index c41bbc59b90..c61e0842fc2 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -76,7 +76,7 @@ a8r8g8b8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const unsigned *src - = ((const unsigned *) (ps->region->map + ps->offset)) + = ((const unsigned *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -107,7 +107,7 @@ a8r8g8b8_put_tile(struct pipe_surface *ps, const float *p) { unsigned *dst - = ((unsigned *) (ps->region->map + ps->offset)) + = ((unsigned *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -140,7 +140,7 @@ b8g8r8a8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const unsigned *src - = ((const unsigned *) (ps->region->map + ps->offset)) + = ((const unsigned *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -171,7 +171,7 @@ b8g8r8a8_put_tile(struct pipe_surface *ps, const float *p) { unsigned *dst - = ((unsigned *) (ps->region->map + ps->offset)) + = ((unsigned *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -204,7 +204,7 @@ a1r5g5b5_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const ushort *src - = ((const ushort *) (ps->region->map + ps->offset)) + = ((const ushort *) (ps->map)) + y * ps->pitch + x; unsigned i, j; @@ -235,7 +235,7 @@ z16_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const ushort *src - = ((const ushort *) (ps->region->map + ps->offset)) + = ((const ushort *) (ps->map)) + y * ps->pitch + x; const float scale = 1.0f / 65535.0f; unsigned i, j; @@ -268,7 +268,7 @@ l8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const ubyte *src - = ((const ubyte *) (ps->region->map + ps->offset)) + = ((const ubyte *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -299,7 +299,7 @@ a8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const ubyte *src - = ((const ubyte *) (ps->region->map + ps->offset)) + = ((const ubyte *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -330,7 +330,7 @@ r16g16b16a16_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const short *src - = ((const short *) (ps->region->map + ps->offset)) + = ((const short *) (ps->map)) + (y * ps->pitch + x) * 4; unsigned i, j; unsigned w0 = w; @@ -362,7 +362,7 @@ r16g16b16a16_put_tile(struct pipe_surface *ps, const float *p) { short *dst - = ((short *) (ps->region->map + ps->offset)) + = ((short *) (ps->map)) + (y * ps->pitch + x) * 4; unsigned i, j; unsigned w0 = w; @@ -399,7 +399,7 @@ i8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const ubyte *src - = ((const ubyte *) (ps->region->map + ps->offset)) + = ((const ubyte *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -430,7 +430,7 @@ a8_l8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const ushort *src - = ((const ushort *) (ps->region->map + ps->offset)) + = ((const ushort *) (ps->map)) + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; @@ -467,7 +467,7 @@ z32_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const uint *src - = ((const uint *) (ps->region->map + ps->offset)) + = ((const uint *) (ps->map)) + y * ps->pitch + x; const double scale = 1.0 / (double) 0xffffffff; unsigned i, j; @@ -501,7 +501,7 @@ s8z24_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const uint *src - = ((const uint *) (ps->region->map + ps->offset)) + = ((const uint *) (ps->map)) + y * ps->pitch + x; const double scale = 1.0 / ((1 << 24) - 1); unsigned i, j; @@ -535,7 +535,7 @@ z24s8_get_tile(struct pipe_surface *ps, unsigned x, unsigned y, unsigned w, unsigned h, float *p) { const uint *src - = ((const uint *) (ps->region->map + ps->offset)) + = ((const uint *) (ps->map)) + y * ps->pitch + x; const double scale = 1.0 / ((1 << 24) - 1); unsigned i, j; @@ -589,7 +589,7 @@ softpipe_get_tex_surface(struct pipe_context *pipe, if (ps) { assert(ps->format); assert(ps->refcount); - pipe_region_reference(&ps->region, spt->region); + pipe->winsys->buffer_reference(pipe->winsys, &ps->buffer, spt->buffer); ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; @@ -613,7 +613,7 @@ softpipe_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); if (dst_stride == 0) { dst_stride = w * cpp; @@ -621,7 +621,7 @@ softpipe_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, CLIP_TILE; - pSrc = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; + pSrc = ps->map + (y * ps->pitch + x) * cpp; pDest = (ubyte *) p; for (i = 0; i < h; i++) { @@ -645,7 +645,7 @@ softpipe_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); if (src_stride == 0) { src_stride = w * cpp; @@ -654,7 +654,7 @@ softpipe_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, CLIP_TILE; pSrc = (const ubyte *) p; - pDest = ps->region->map + ps->offset + (y * ps->pitch + x) * cpp; + pDest = ps->map + (y * ps->pitch + x) * cpp; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w * cpp); @@ -817,12 +817,12 @@ sp_surface_data(struct pipe_context *pipe, const void *src, unsigned src_pitch, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - copy_rect(pipe->region_map(pipe, dst->region) + dst->offset, + copy_rect(pipe_surface_map(dst), dst->cpp, dst->pitch, dstx, dsty, width, height, src, src_pitch, srcx, srcy); - pipe->region_unmap(pipe, dst->region); + pipe_surface_unmap(dst); } /* Assumes all values are within bounds -- no checking at this level - @@ -835,29 +835,26 @@ sp_surface_copy(struct pipe_context *pipe, struct pipe_surface *src, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - ubyte *src_map, *dst_map; assert( dst->cpp == src->cpp ); - dst_map = pipe->region_map(pipe, dst->region); - src_map = pipe->region_map(pipe, src->region); - copy_rect(dst_map + dst->offset, + copy_rect(pipe_surface_map(dst), dst->cpp, dst->pitch, dstx, dsty, width, height, - src_map + src->offset, + pipe_surface_map(src), src->pitch, srcx, srcy); - pipe->region_unmap(pipe, src->region); - pipe->region_unmap(pipe, dst->region); + pipe_surface_unmap(src); + pipe_surface_unmap(dst); } static ubyte * get_pointer(struct pipe_surface *dst, unsigned x, unsigned y) { - return dst->region->map + (y * dst->pitch + x) * dst->cpp; + return dst->map + (y * dst->pitch + x) * dst->cpp; } @@ -879,7 +876,7 @@ sp_surface_fill(struct pipe_context *pipe, assert(dst->pitch > 0); assert(width <= dst->pitch); - (void)pipe->region_map(pipe, dst->region); + (void)pipe_surface_map(dst); switch (dst->cpp) { case 1: @@ -935,7 +932,7 @@ sp_surface_fill(struct pipe_context *pipe, break; } - pipe->region_unmap( pipe, dst->region ); + pipe_surface_unmap( dst ); } diff --git a/src/mesa/pipe/softpipe/sp_texture.c b/src/mesa/pipe/softpipe/sp_texture.c index 2f9a1e9837b..53486f9bba5 100644 --- a/src/mesa/pipe/softpipe/sp_texture.c +++ b/src/mesa/pipe/softpipe/sp_texture.c @@ -32,6 +32,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #include "pipe/p_util.h" #include "pipe/p_winsys.h" @@ -380,13 +381,18 @@ softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) sizeof(struct softpipe_texture) - sizeof(struct pipe_texture)); if (softpipe_mipmap_tree_layout(pipe, spt)) { - spt->region = pipe->winsys->region_alloc(pipe->winsys, - spt->pitch * (*pt)->cpp * - spt->total_height, - PIPE_SURFACE_FLAG_TEXTURE); + spt->buffer = pipe->winsys->buffer_create(pipe->winsys, + PIPE_SURFACE_FLAG_TEXTURE); + + if (spt->buffer) { + pipe->winsys->buffer_data(pipe->winsys, spt->buffer, + spt->pitch * (*pt)->cpp * + spt->total_height, NULL, + PIPE_BUFFER_USAGE_PIXEL); + } } - if (!spt->region) { + if (!spt->buffer) { FREE(spt); spt = NULL; } @@ -413,7 +419,7 @@ softpipe_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) DBG("%s deleting %p\n", __FUNCTION__, (void *) spt); */ - pipe->winsys->region_release(pipe->winsys, &spt->region); + pipe->winsys->buffer_reference(pipe->winsys, &spt->buffer, NULL); for (i = 0; i < PIPE_MAX_TEXTURE_LEVELS; i++) if (spt->image_offset[i]) diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 62ee6a27c93..08cd39cc555 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -123,16 +123,24 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc) void -sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, +sp_tile_cache_set_surface(struct softpipe_context *sp, + struct softpipe_tile_cache *tc, struct pipe_surface *ps) { + if (tc->surface && tc->surface->map) + pipe_surface_unmap(tc->surface); + pipe_surface_reference(&tc->surface, ps); } struct pipe_surface * -sp_tile_cache_get_surface(struct softpipe_tile_cache *tc) +sp_tile_cache_get_surface(struct softpipe_context *sp, + struct softpipe_tile_cache *tc) { + if (tc->surface && !tc->surface->map) + pipe_surface_map(tc->surface); + return tc->surface; } @@ -162,7 +170,7 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, boolean is_depth_stencil; int inuse = 0, pos; - if (!ps || !ps->region || !ps->region->map) + if (!ps || !ps->buffer) return; is_depth_stencil = (ps->format == PIPE_FORMAT_S8_Z24 || @@ -367,6 +375,16 @@ sp_get_cached_tile_tex(struct pipe_context *pipe, struct pipe_surface *ps = pipe->get_tex_surface(pipe, tc->texture, face, level, z); + if (ps != tc->surface) { + if (tc->surface && tc->surface->map) + pipe_surface_unmap(tc->surface); + + pipe_surface_reference(&tc->surface, ps); + + if (!tc->surface->map) + pipe_surface_map(tc->surface); + } + pipe->get_tile_rgba(pipe, ps, tile_x, tile_y, TILE_SIZE, TILE_SIZE, (float *) tile->data.color); diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.h b/src/mesa/pipe/softpipe/sp_tile_cache.h index 9967aa5044a..de5ff2c498b 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.h +++ b/src/mesa/pipe/softpipe/sp_tile_cache.h @@ -63,11 +63,13 @@ extern void sp_destroy_tile_cache(struct softpipe_tile_cache *tc); extern void -sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, +sp_tile_cache_set_surface(struct softpipe_context *sp, + struct softpipe_tile_cache *tc, struct pipe_surface *sps); extern struct pipe_surface * -sp_tile_cache_get_surface(struct softpipe_tile_cache *tc); +sp_tile_cache_get_surface(struct softpipe_context *sp, + struct softpipe_tile_cache *tc); extern void sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, diff --git a/src/mesa/pipe/xlib/xm_api.c b/src/mesa/pipe/xlib/xm_api.c index f87d72d0c8c..542b4aca49c 100644 --- a/src/mesa/pipe/xlib/xm_api.c +++ b/src/mesa/pipe/xlib/xm_api.c @@ -1512,7 +1512,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) pipe->get_tile_rgba = xmesa_get_tile_rgba; pipe->put_tile_rgba = xmesa_put_tile_rgba; - c->st->haveFramebufferRegions = GL_FALSE; + c->st->haveFramebufferSurfaces = GL_FALSE; /* special pipe->clear function */ pipe->clear = xmesa_clear; diff --git a/src/mesa/pipe/xlib/xm_buffer.c b/src/mesa/pipe/xlib/xm_buffer.c index 449e90184b8..6fc2f4ba630 100644 --- a/src/mesa/pipe/xlib/xm_buffer.c +++ b/src/mesa/pipe/xlib/xm_buffer.c @@ -269,9 +269,8 @@ static void finish_surface_init(GLcontext *ctx, struct xmesa_renderbuffer *xrb) { struct pipe_context *pipe = ctx->st->pipe; - if (!xrb->St.surface->region) { - xrb->St.surface->region = pipe->winsys->region_alloc(pipe->winsys, 1, - 0x0); + if (!xrb->St.surface->buffer) { + xrb->St.surface->buffer = pipe->winsys->buffer_create(pipe->winsys, 0x0); } } @@ -301,7 +300,7 @@ xmesa_alloc_front_storage(GLcontext *ctx, struct gl_renderbuffer *rb, rb->Height = height; rb->InternalFormat = internalFormat; - if (!xrb->St.surface || !xrb->St.surface->region) + if (!xrb->St.surface || !xrb->St.surface->buffer) finish_surface_init(ctx, xrb); /* surface info */ @@ -363,7 +362,7 @@ xmesa_alloc_back_storage(GLcontext *ctx, struct gl_renderbuffer *rb, xrb->origin4 = NULL; } - if (!xrb->St.surface || !xrb->St.surface->region) + if (!xrb->St.surface || !xrb->St.surface->buffer) finish_surface_init(ctx, xrb); xrb->St.surface->width = width; diff --git a/src/mesa/pipe/xlib/xm_surface.c b/src/mesa/pipe/xlib/xm_surface.c index 58ca2e7fb3c..4a54b5c7e1a 100644 --- a/src/mesa/pipe/xlib/xm_surface.c +++ b/src/mesa/pipe/xlib/xm_surface.c @@ -665,12 +665,12 @@ xmesa_new_color_surface(struct pipe_winsys *winsys, GLuint pipeFormat) xms->surface.refcount = 1; xms->surface.winsys = winsys; - /* Note, the region we allocate doesn't actually have any storage + /* Note, the buffer we allocate doesn't actually have any storage * since we're drawing into an XImage or Pixmap. - * The region's size will get set in the xmesa_alloc_front/back_storage() + * The surface's size will get set in the xmesa_alloc_front/back_storage() * functions. */ - xms->surface.region = winsys->region_alloc(winsys, 1, 0x0); + xms->surface.buffer = winsys->buffer_create(winsys, 0x0); return &xms->surface; } @@ -713,7 +713,7 @@ xmesa_clear(struct pipe_context *pipe, struct pipe_surface *ps, uint value) { struct softpipe_context *sp = softpipe_context(pipe); - if (ps == sp_tile_cache_get_surface(sp->cbuf_cache[0])) { + if (ps == sp_tile_cache_get_surface(sp, sp->cbuf_cache[0])) { float clear[4]; clear[0] = 0.2; /* XXX hack */ clear[1] = 0.2; diff --git a/src/mesa/pipe/xlib/xm_winsys.c b/src/mesa/pipe/xlib/xm_winsys.c index 99816a811d6..295174d4bba 100644 --- a/src/mesa/pipe/xlib/xm_winsys.c +++ b/src/mesa/pipe/xlib/xm_winsys.c @@ -186,7 +186,7 @@ xm_get_name(struct pipe_winsys *pws) static struct pipe_buffer_handle * -xm_buffer_create(struct pipe_winsys *pws, unsigned alignment) +xm_buffer_create(struct pipe_winsys *pws, unsigned flags) { struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer); buffer->refcount = 1; @@ -228,47 +228,6 @@ xm_surface_pitch(struct pipe_winsys *winsys, unsigned cpp, unsigned width, } -static struct pipe_region * -xm_region_alloc(struct pipe_winsys *winsys, unsigned size, unsigned flags) -{ - struct pipe_region *region = CALLOC_STRUCT(pipe_region); - const unsigned alignment = 64; - - region->refcount = 1; - - assert(size > 0); - - region->buffer = winsys->buffer_create( winsys, alignment ); - - /* NULL data --> just allocate the space */ - winsys->buffer_data( winsys, - region->buffer, - size, - NULL, - PIPE_BUFFER_USAGE_PIXEL ); - return region; -} - - -static void -xm_region_release(struct pipe_winsys *winsys, struct pipe_region **region) -{ - if (!*region) - return; - - assert((*region)->refcount > 0); - (*region)->refcount--; - - if ((*region)->refcount == 0) { - assert((*region)->map_refcount == 0); - - winsys->buffer_reference( winsys, &((*region)->buffer), NULL ); - free(*region); - } - *region = NULL; -} - - /** * Called via pipe->surface_alloc() to create new surfaces (textures, * renderbuffers, etc. @@ -301,8 +260,8 @@ xm_surface_release(struct pipe_winsys *winsys, struct pipe_surface **s) struct pipe_surface *surf = *s; surf->refcount--; if (surf->refcount == 0) { - if (surf->region) - winsys->region_release(winsys, &surf->region); + if (surf->buffer) + winsys->buffer_reference(winsys, &surf->buffer, NULL); free(surf); } *s = NULL; @@ -335,9 +294,6 @@ xmesa_get_pipe_winsys(void) ws->buffer_subdata = xm_buffer_subdata; ws->buffer_get_subdata = xm_buffer_get_subdata; - ws->region_alloc = xm_region_alloc; - ws->region_release = xm_region_release; - ws->surface_pitch = xm_surface_pitch; ws->surface_alloc = xm_surface_alloc; ws->surface_release = xm_surface_release; diff --git a/src/mesa/state_tracker/st_cb_accum.c b/src/mesa/state_tracker/st_cb_accum.c index 7a245b0ed6d..c8d9cba12f3 100644 --- a/src/mesa/state_tracker/st_cb_accum.c +++ b/src/mesa/state_tracker/st_cb_accum.c @@ -42,6 +42,7 @@ #include "st_format.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" /** @@ -68,7 +69,7 @@ st_clear_accum_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) GLfloat *accBuf; GLint i; - (void) pipe->region_map(pipe, acc_ps->region); + (void) pipe_surface_map(acc_ps); accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); @@ -83,7 +84,7 @@ st_clear_accum_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) free(accBuf); - pipe->region_unmap(pipe, acc_ps->region); + pipe_surface_unmap(acc_ps); } @@ -99,7 +100,7 @@ accum_mad(struct pipe_context *pipe, GLfloat scale, GLfloat bias, accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - (void) pipe->region_map(pipe, acc_ps->region); + (void) pipe_surface_map(acc_ps); pipe->get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); @@ -111,7 +112,7 @@ accum_mad(struct pipe_context *pipe, GLfloat scale, GLfloat bias, free(accBuf); - pipe->region_unmap(pipe, acc_ps->region); + pipe_surface_unmap(acc_ps); } @@ -128,8 +129,8 @@ accum_accum(struct pipe_context *pipe, GLfloat value, colorBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - colorMap = pipe->region_map(pipe, color_ps->region); - accMap = pipe->region_map(pipe, acc_ps->region); + colorMap = pipe_surface_map(color_ps); + accMap = pipe_surface_map(acc_ps); pipe->get_tile_rgba(pipe, color_ps, xpos, ypos, width, height, colorBuf); pipe->get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf); @@ -143,8 +144,8 @@ accum_accum(struct pipe_context *pipe, GLfloat value, free(colorBuf); free(accBuf); - pipe->region_unmap(pipe, color_ps->region); - pipe->region_unmap(pipe, acc_ps->region); + pipe_surface_unmap(color_ps); + pipe_surface_unmap(acc_ps); } @@ -159,8 +160,8 @@ accum_load(struct pipe_context *pipe, GLfloat value, buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - (void) pipe->region_map(pipe, color_ps->region); - (void) pipe->region_map(pipe, acc_ps->region); + (void) pipe_surface_map(color_ps); + (void) pipe_surface_map(acc_ps); pipe->get_tile_rgba(pipe, color_ps, xpos, ypos, width, height, buf); @@ -172,8 +173,8 @@ accum_load(struct pipe_context *pipe, GLfloat value, free(buf); - pipe->region_unmap(pipe, color_ps->region); - pipe->region_unmap(pipe, acc_ps->region); + pipe_surface_unmap(color_ps); + pipe_surface_unmap(acc_ps); } @@ -190,8 +191,8 @@ accum_return(GLcontext *ctx, GLfloat value, abuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - (void) pipe->region_map(pipe, color_ps->region); - (void) pipe->region_map(pipe, acc_ps->region); + (void) pipe_surface_map(color_ps); + (void) pipe_surface_map(acc_ps); pipe->get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, abuf); @@ -218,8 +219,8 @@ accum_return(GLcontext *ctx, GLfloat value, if (cbuf) free(cbuf); - pipe->region_unmap(pipe, color_ps->region); - pipe->region_unmap(pipe, acc_ps->region); + pipe_surface_unmap(color_ps); + pipe_surface_unmap(acc_ps); } diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index c28ad15b294..a61daffb20a 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -492,11 +492,10 @@ make_texture(struct st_context *st, surface = pipe->get_tex_surface(pipe, pt, 0, 0, 0); - /* map texture region */ - (void) pipe->region_map(pipe, surface->region); - dest = surface->region->map + surface->offset; + /* map texture surface */ + dest = pipe_surface_map(surface); - /* Put image into texture region. + /* Put image into texture surface. * Note that the image is actually going to be upside down in * the texture. We deal with that with texcoords. */ @@ -513,7 +512,7 @@ make_texture(struct st_context *st, unpack); /* unmap */ - pipe->region_unmap(pipe, surface->region); + pipe_surface_unmap(surface); pipe_surface_reference(&surface, NULL); assert(success); @@ -852,7 +851,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, pipe->flush(pipe, 0); /* map the stencil buffer */ - stmap = pipe->region_map(pipe, ps->region); + stmap = pipe_surface_map(ps); /* if width > MAX_WIDTH, have to process image in chunks */ skipPixels = 0; @@ -909,7 +908,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, } /* unmap the stencil buffer */ - pipe->region_unmap(pipe, ps->region); + pipe_surface_unmap(ps); } @@ -1027,11 +1026,10 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, surface = pipe->get_tex_surface(pipe, pt, 0, 0, 0); - /* map texture region */ - (void) pipe->region_map(pipe, surface->region); - dest = surface->region->map + surface->offset; + /* map texture surface */ + dest = pipe_surface_map(surface); - /* Put image into texture region. + /* Put image into texture surface. * Note that the image is actually going to be upside down in * the texture. We deal with that with texcoords. */ @@ -1089,7 +1087,7 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, } /* row */ /* Release surface */ - pipe->region_unmap(pipe, surface->region); + pipe_surface_unmap(surface); pipe_surface_reference(&surface, NULL); pt->format = format; @@ -1130,8 +1128,6 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, GLsizei width, GLsizei height, GLint dstx, GLint dsty) { - struct st_context *st = ctx->st; - struct pipe_context *pipe = st->pipe; struct st_renderbuffer *rbRead = st_renderbuffer(ctx->ReadBuffer->_StencilBuffer); struct st_renderbuffer *rbDraw = st_renderbuffer(ctx->DrawBuffer->_StencilBuffer); struct pipe_surface *psRead = rbRead->surface; @@ -1147,8 +1143,8 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, } /* map the stencil buffers */ - readMap = pipe->region_map(pipe, psRead->region); - drawMap = pipe->region_map(pipe, psDraw->region); + readMap = pipe_surface_map(psRead); + drawMap = pipe_surface_map(psDraw); /* this will do stencil pixel transfer ops */ st_read_stencil_pixels(ctx, srcx, srcy, width, height, GL_UNSIGNED_BYTE, @@ -1192,8 +1188,8 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, free(buffer); /* unmap the stencil buffers */ - pipe->region_unmap(pipe, psRead->region); - pipe->region_unmap(pipe, psDraw->region); + pipe_surface_unmap(psRead); + pipe_surface_unmap(psDraw); } @@ -1252,11 +1248,11 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, } /* For some drivers (like Xlib) it's not possible to treat the - * front/back color buffers as regions (they're XImages and Pixmaps). - * So, this var tells us if we can use region_copy here... + * front/back color buffers as surfaces (they're XImages and Pixmaps). + * So, this var tells us if we can use surface_copy here... */ - if (st->haveFramebufferRegions) { - /* copy source framebuffer region into mipmap/texture */ + if (st->haveFramebufferSurfaces) { + /* copy source framebuffer surface into mipmap/texture */ pipe->surface_copy(pipe, psTex, /* dest */ 0, 0, /* destx/y */ @@ -1267,14 +1263,14 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, /* alternate path using get/put_tile() */ GLfloat *buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - (void) pipe->region_map(pipe, psRead->region); - (void) pipe->region_map(pipe, psTex->region); + (void) pipe_surface_map(psRead); + (void) pipe_surface_map(psTex); pipe->get_tile_rgba(pipe, psRead, srcx, srcy, width, height, buf); pipe->put_tile_rgba(pipe, psTex, 0, 0, width, height, buf); - pipe->region_unmap(pipe, psRead->region); - pipe->region_unmap(pipe, psTex->region); + pipe_surface_unmap(psRead); + pipe_surface_unmap(psTex); free(buf); } diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index 43681b7f8a9..6b9023c4104 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -109,22 +109,22 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, width, flags); } - /* free old region */ - if (strb->surface->region) { - /* loop here since mapping is refcounted */ - struct pipe_region *r = strb->surface->region; - while (r->map) - pipe->region_unmap(pipe, r); - pipe->winsys->region_release(pipe->winsys, &strb->surface->region); - } - - strb->surface->region = pipe->winsys->region_alloc(pipe->winsys, - strb->surface->pitch * - cpp * height, flags); - if (!strb->surface->region) + /* loop here since mapping is refcounted */ + while (strb->surface->map) + pipe_surface_unmap(strb->surface); + if (strb->surface->buffer) + pipe->winsys->buffer_reference(pipe->winsys, &strb->surface->buffer, + NULL); + + strb->surface->buffer = pipe->winsys->buffer_create(pipe->winsys, flags); + if (!strb->surface->buffer) return GL_FALSE; /* out of memory, try s/w buffer? */ - ASSERT(strb->surface->region->buffer); + pipe->winsys->buffer_data(pipe->winsys, strb->surface->buffer, + strb->surface->pitch * cpp * height, NULL, + PIPE_BUFFER_USAGE_PIXEL); + + ASSERT(strb->surface->buffer); ASSERT(strb->surface->format); strb->Base.Width = strb->surface->width = width; diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index 83fe480af85..e2243e7de49 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -39,6 +39,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #include "st_context.h" #include "st_cb_readpixels.h" #include "st_cb_fbo.h" @@ -56,8 +57,6 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, const struct gl_pixelstore_attrib *packing, GLvoid *pixels) { - struct st_context *st = ctx->st; - struct pipe_context *pipe = st->pipe; struct gl_framebuffer *fb = ctx->ReadBuffer; struct st_renderbuffer *strb = st_renderbuffer(fb->_StencilBuffer); struct pipe_surface *ps = strb->surface; @@ -65,7 +64,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, GLint j; /* map the stencil buffer */ - stmap = pipe->region_map(pipe, ps->region); + stmap = pipe_surface_map(ps); /* width should never be > MAX_WIDTH since we did clipping earlier */ ASSERT(width <= MAX_WIDTH); @@ -122,7 +121,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, /* unmap the stencil buffer */ - pipe->region_unmap(pipe, ps->region); + pipe_surface_unmap(ps); } @@ -151,7 +150,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, /* Do all needed clipping here, so that we can forget about it later */ if (!_mesa_clip_readpixels(ctx, &x, &y, &width, &height, &clippedPacking)) { - /* The ReadPixels region is totally outside the window bounds */ + /* The ReadPixels surface is totally outside the window bounds */ return; } @@ -180,7 +179,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, if (!strb) return; - pipe->region_map(pipe, strb->surface->region); + pipe_surface_map(strb->surface); if (format == GL_RGBA && type == GL_FLOAT) { /* write tile(row) directly into user's buffer */ @@ -231,7 +230,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, } } - pipe->region_unmap(pipe, strb->surface->region); + pipe_surface_unmap(strb->surface); } diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index acc97df2e76..1fcef746c1f 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -64,7 +64,7 @@ struct st_texture_object GLuint textureOffset; /* On validation any active images held in main memory or in other - * regions will be copied to this region and the old storage freed. + * textures will be copied to this texture and the old storage freed. */ struct pipe_texture *pt; @@ -721,7 +721,7 @@ st_TexImage(GLcontext * ctx, } if (stImage->pt && i < depth) { - st_texture_image_unmap(ctx->st, stImage); + st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, i); pixels += srcImageStride; } @@ -731,7 +731,7 @@ st_TexImage(GLcontext * ctx, _mesa_unmap_teximage_pbo(ctx, unpack); if (stImage->pt) { - st_texture_image_unmap(ctx->st, stImage); + st_texture_image_unmap(stImage); texImage->Data = NULL; } @@ -860,7 +860,7 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, } if (stImage->pt && i < depth) { - st_texture_image_unmap(ctx->st, stImage); + st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, i); pixels += dstImageStride; } @@ -870,7 +870,7 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, /* Unmap */ if (stImage->pt) { - st_texture_image_unmap(ctx->st, stImage); + st_texture_image_unmap(stImage); texImage->Data = NULL; } } @@ -948,7 +948,7 @@ st_TexSubimage(GLcontext * ctx, } if (stImage->pt && i < depth) { - st_texture_image_unmap(ctx->st, stImage); + st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, zoffset + i); pixels += srcImageStride; } @@ -966,7 +966,7 @@ st_TexSubimage(GLcontext * ctx, _mesa_unmap_teximage_pbo(ctx, packing); if (stImage->pt) { - st_texture_image_unmap(ctx->st, stImage); + st_texture_image_unmap(stImage); texImage->Data = NULL; } } @@ -1051,8 +1051,8 @@ texture_face(GLenum target) /** - * Do a CopyTexSubImage operation by mapping the source region and - * dest region and using get_tile()/put_tile() to access the pixels/texels. + * Do a CopyTexSubImage operation by mapping the source surface and + * dest surface and using get_tile()/put_tile() to access the pixels/texels. * * Note: srcY=0=TOP of renderbuffer */ @@ -1088,8 +1088,8 @@ fallback_copy_texsubimage(GLcontext *ctx, dest_surf = pipe->get_tex_surface(pipe, pt, face, level, destZ); - (void) pipe->region_map(pipe, dest_surf->region); - (void) pipe->region_map(pipe, src_surf->region); + (void) pipe_surface_map(dest_surf); + (void) pipe_surface_map(src_surf); /* buffer for one row */ data = (GLfloat *) malloc(width * 4 * sizeof(GLfloat)); @@ -1110,8 +1110,8 @@ fallback_copy_texsubimage(GLcontext *ctx, } - (void) pipe->region_unmap(pipe, dest_surf->region); - (void) pipe->region_unmap(pipe, src_surf->region); + (void) pipe_surface_unmap(dest_surf); + (void) pipe_surface_unmap(src_surf); free(data); } @@ -1178,8 +1178,8 @@ do_copy_texsubimage(GLcontext *ctx, if (src_format == dest_format && ctx->_ImageTransferState == 0x0 && - strb->surface->region && - dest_surface->region && + strb->surface->buffer && + dest_surface->buffer && strb->surface->cpp == stImage->pt->cpp) { /* do blit-style copy */ diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index a4ec3721bef..abde6d64b01 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -103,7 +103,7 @@ st_create_context_priv( GLcontext *ctx, struct pipe_context *pipe ) st->ctx->VertexProgram._MaintainTnlProgram = GL_TRUE; - st->haveFramebufferRegions = GL_TRUE; + st->haveFramebufferSurfaces = GL_TRUE; st->pixel_xfer.cache = _mesa_new_program_cache(); diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index db97014c5a7..c31b76c63f3 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -34,7 +34,6 @@ struct st_context; -struct st_region; struct st_texture_object; struct st_fragment_program; struct draw_context; @@ -131,11 +130,11 @@ struct st_context char vendor[100]; char renderer[100]; - /** Can we access the front/back color buffers as pipe_regions? + /** Can we access the front/back color buffers as pipe_surfaces? * We can't with the Xlib driver... * This is a hack that should be fixed someday. */ - GLboolean haveFramebufferRegions; + GLboolean haveFramebufferSurfaces; /* State to be validated: */ diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index a5582c31c0b..1ec45148736 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -33,6 +33,7 @@ #include "pipe/p_state.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" #include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" @@ -187,17 +188,15 @@ st_texture_image_map(struct st_context *st, struct st_texture_image *stImage, stImage->surface = st->pipe->get_tex_surface(st->pipe, pt, stImage->face, stImage->level, zoffset); - (void) st->pipe->region_map(st->pipe, stImage->surface->region); - - return stImage->surface->region->map + stImage->surface->offset; + return pipe_surface_map(stImage->surface); } void -st_texture_image_unmap(struct st_context *st, struct st_texture_image *stImage) +st_texture_image_unmap(struct st_texture_image *stImage) { DBG("%s\n", __FUNCTION__); - st->pipe->region_unmap(st->pipe, stImage->surface->region); + pipe_surface_unmap(stImage->surface); pipe_surface_reference(&stImage->surface, NULL); } diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index b25e3f3f3bd..f49d91a0692 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -65,8 +65,7 @@ st_texture_image_map(struct st_context *st, GLuint zoffset); extern void -st_texture_image_unmap(struct st_context *st, - struct st_texture_image *stImage); +st_texture_image_unmap(struct st_texture_image *stImage); /* Return pointers to each 2d slice within an image. Indexed by depth -- cgit v1.2.3 From a775689036fc86ea2cfaf529fba6c368fa6d3502 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 07:48:33 -0700 Subject: silence void ptr arithmetic warning --- src/mesa/pipe/p_inlines.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/p_inlines.h b/src/mesa/pipe/p_inlines.h index 1697284b049..e7303c45c1b 100644 --- a/src/mesa/pipe/p_inlines.h +++ b/src/mesa/pipe/p_inlines.h @@ -37,11 +37,12 @@ static INLINE ubyte * pipe_surface_map(struct pipe_surface *surface) { if (!surface->map_refcount++) { - surface->map = surface->winsys->buffer_map( surface->winsys, + surface->map + = (ubyte *) surface->winsys->buffer_map( surface->winsys, surface->buffer, PIPE_BUFFER_FLAG_WRITE | PIPE_BUFFER_FLAG_READ ) - + surface->offset; + + surface->offset; } return surface->map; -- cgit v1.2.3 From 6cb181dc15116fb495a8f68d4088ecb15dc39c0f Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 07:52:35 -0700 Subject: fix void pointer arithmetic warnings --- src/mesa/state_tracker/st_cb_texture.c | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 1fcef746c1f..14f8d38def8 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -706,6 +706,7 @@ st_TexImage(GLcontext * ctx, GLuint srcImageStride = _mesa_image_image_stride(unpack, width, height, format, type); int i; + const GLubyte *src = (const GLubyte *) pixels; for (i = 0; i++ < depth;) { if (!texImage->TexFormat->StoreImage(ctx, dims, @@ -716,14 +717,14 @@ st_TexImage(GLcontext * ctx, dstRowStride, texImage->ImageOffsets, width, height, 1, - format, type, pixels, unpack)) { + format, type, src, unpack)) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); } if (stImage->pt && i < depth) { st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, i); - pixels += srcImageStride; + src += srcImageStride; } } } @@ -826,6 +827,7 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, type); GLuint depth; int i; + GLubyte *dest; /* Map */ if (stImage->pt) { @@ -850,19 +852,21 @@ st_get_tex_image(GLcontext * ctx, GLenum target, GLint level, depth = texImage->Depth; texImage->Depth = 1; + dest = (GLubyte *) pixels; + for (i = 0; i++ < depth;) { if (compressed) { - _mesa_get_compressed_teximage(ctx, target, level, pixels, + _mesa_get_compressed_teximage(ctx, target, level, dest, texObj, texImage); } else { - _mesa_get_teximage(ctx, target, level, format, type, pixels, + _mesa_get_teximage(ctx, target, level, format, type, dest, texObj, texImage); } if (stImage->pt && i < depth) { st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, i); - pixels += dstImageStride; + dest += dstImageStride; } } @@ -916,6 +920,7 @@ st_TexSubimage(GLcontext * ctx, GLuint srcImageStride = _mesa_image_image_stride(packing, width, height, format, type); int i; + const GLubyte *src; DBG("%s target %s level %d offset %d,%d %dx%d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), @@ -935,6 +940,8 @@ st_TexSubimage(GLcontext * ctx, dstRowStride = stImage->surface->pitch * stImage->surface->cpp; } + src = (const GLubyte *) pixels; + for (i = 0; i++ < depth;) { if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat, texImage->TexFormat, @@ -943,14 +950,14 @@ st_TexSubimage(GLcontext * ctx, dstRowStride, texImage->ImageOffsets, width, height, 1, - format, type, pixels, packing)) { + format, type, src, packing)) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "st_TexSubImage"); } if (stImage->pt && i < depth) { st_texture_image_unmap(stImage); texImage->Data = st_texture_image_map(ctx->st, stImage, zoffset + i); - pixels += srcImageStride; + src += srcImageStride; } } -- cgit v1.2.3 From a09224258a205dbb99eae2c9f94fce8d4f3f73f6 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 07:53:06 -0700 Subject: move surface pitch calculation so it always gets updated --- src/mesa/state_tracker/st_cb_fbo.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index 6b9023c4104..0ee5f45aae3 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -105,10 +105,11 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, if (!strb->surface) return GL_FALSE; strb->surface->cpp = cpp; - strb->surface->pitch = pipe->winsys->surface_pitch(pipe->winsys, cpp, - width, flags); } + strb->surface->pitch = pipe->winsys->surface_pitch(pipe->winsys, cpp, + width, flags); + /* loop here since mapping is refcounted */ while (strb->surface->map) pipe_surface_unmap(strb->surface); -- cgit v1.2.3 From 6397bef076271d36443ce9db5f80fa20b05ee60e Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 07:53:21 -0700 Subject: added ST_SURFACE_DEPTH --- src/mesa/state_tracker/st_framebuffer.c | 2 +- src/mesa/state_tracker/st_public.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c index 4ae2837f0a0..8633f431b29 100644 --- a/src/mesa/state_tracker/st_framebuffer.c +++ b/src/mesa/state_tracker/st_framebuffer.c @@ -155,7 +155,7 @@ st_get_framebuffer_surface(struct st_framebuffer *stfb, uint surfIndex) { struct st_renderbuffer *strb; - assert(surfIndex <= ST_SURFACE_BACK_RIGHT); + assert(surfIndex <= ST_SURFACE_DEPTH); /* sanity checks, ST tokens should match Mesa tokens */ assert(ST_SURFACE_FRONT_LEFT == BUFFER_FRONT_LEFT); diff --git a/src/mesa/state_tracker/st_public.h b/src/mesa/state_tracker/st_public.h index 408e1927e5b..307635936bc 100644 --- a/src/mesa/state_tracker/st_public.h +++ b/src/mesa/state_tracker/st_public.h @@ -35,6 +35,7 @@ #define ST_SURFACE_BACK_LEFT 1 #define ST_SURFACE_FRONT_RIGHT 2 #define ST_SURFACE_BACK_RIGHT 3 +#define ST_SURFACE_DEPTH 8 struct st_context; -- cgit v1.2.3 From 2a9c33f9fe4b78710bfa36f1309d791c294d7cc8 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 07:57:54 -0700 Subject: Overhaul the Xlib winsys layer. Front/back color buffers are now allocated with ordinary malloc() via the winsys buffer functions. To display surfaces in SwapBuffers() or flush_frontbuffer() we create an XImage that wraps the surface, then use XPutImage to copy to the window. Shared memory transport disabled for now. --- src/mesa/pipe/xlib/fakeglx.c | 17 +- src/mesa/pipe/xlib/xm_api.c | 1282 ++++----------------------------------- src/mesa/pipe/xlib/xm_buffer.c | 20 +- src/mesa/pipe/xlib/xm_surface.c | 683 --------------------- src/mesa/pipe/xlib/xm_winsys.c | 157 +++-- src/mesa/pipe/xlib/xmesaP.h | 457 +------------- src/mesa/sources | 4 +- 7 files changed, 257 insertions(+), 2363 deletions(-) diff --git a/src/mesa/pipe/xlib/fakeglx.c b/src/mesa/pipe/xlib/fakeglx.c index d7e7b292d1f..6965ee0794c 100644 --- a/src/mesa/pipe/xlib/fakeglx.c +++ b/src/mesa/pipe/xlib/fakeglx.c @@ -1433,7 +1433,6 @@ Fake_glXCreateContext( Display *dpy, XVisualInfo *visinfo, return NULL; } - glxCtx->xmesaContext->direct = GL_FALSE; glxCtx->glxContext.isDirect = GL_FALSE; glxCtx->glxContext.currentDpy = dpy; glxCtx->glxContext.xid = (XID) glxCtx; /* self pointer */ @@ -1565,7 +1564,7 @@ Fake_glXCreateGLXPixmap( Display *dpy, XVisualInfo *visinfo, Pixmap pixmap ) if (!b) { return 0; } - return b->frontxrb->pixmap; + return b->drawable; } @@ -1591,7 +1590,7 @@ Fake_glXCreateGLXPixmapMESA( Display *dpy, XVisualInfo *visinfo, if (!b) { return 0; } - return b->frontxrb->pixmap; + return b->drawable; } @@ -1662,9 +1661,9 @@ Fake_glXDestroyContext( Display *dpy, GLXContext ctx ) static Bool Fake_glXIsDirect( Display *dpy, GLXContext ctx ) { - struct fake_glx_context *glxCtx = (struct fake_glx_context *) ctx; (void) dpy; - return glxCtx->xmesaContext->direct; + (void) ctx; + return False; } @@ -2360,7 +2359,7 @@ Fake_glXCreatePbuffer( Display *dpy, GLXFBConfig config, * glXMakeCurrent takes. */ if (xmbuf) - return (GLXPbuffer) xmbuf->frontxrb->pixmap; + return (GLXPbuffer) xmbuf->drawable; else return 0; } @@ -2444,7 +2443,6 @@ Fake_glXCreateNewContext( Display *dpy, GLXFBConfig config, return NULL; } - glxCtx->xmesaContext->direct = GL_FALSE; glxCtx->glxContext.isDirect = GL_FALSE; glxCtx->glxContext.currentDpy = dpy; glxCtx->glxContext.xid = (XID) glxCtx; /* self pointer */ @@ -2642,7 +2640,7 @@ Fake_glXCreateGLXPixmapWithConfigSGIX(Display *dpy, GLXFBConfigSGIX config, Pixm { XMesaVisual xmvis = (XMesaVisual) config; XMesaBuffer xmbuf = XMesaCreatePixmapBuffer(xmvis, pixmap, 0); - return xmbuf->frontxrb->pixmap; /* need to return an X ID */ + return xmbuf->drawable; /* need to return an X ID */ } @@ -2667,7 +2665,6 @@ Fake_glXCreateContextWithConfigSGIX(Display *dpy, GLXFBConfigSGIX config, int re return NULL; } - glxCtx->xmesaContext->direct = GL_FALSE; glxCtx->glxContext.isDirect = GL_FALSE; glxCtx->glxContext.currentDpy = dpy; glxCtx->glxContext.xid = (XID) glxCtx; /* self pointer */ @@ -2736,7 +2733,7 @@ Fake_glXCreateGLXPbufferSGIX(Display *dpy, GLXFBConfigSGIX config, /* A GLXPbuffer handle must be an X Drawable because that's what * glXMakeCurrent takes. */ - return (GLXPbuffer) xmbuf->frontxrb->pixmap; + return (GLXPbuffer) xmbuf->drawable; } diff --git a/src/mesa/pipe/xlib/xm_api.c b/src/mesa/pipe/xlib/xm_api.c index 542b4aca49c..ce7e3cb6c2d 100644 --- a/src/mesa/pipe/xlib/xm_api.c +++ b/src/mesa/pipe/xlib/xm_api.c @@ -48,12 +48,7 @@ * according to the display's visual red_mask, green_mask, and blue_mask. * If XPutPixel is used to put a pixel into an XImage then XPutPixel will * do byte swapping if needed. If one wants to directly "poke" the pixel - * into the XImage's buffer then the pixel must be byte swapped first. In - * Mesa, when byte swapping is needed we use the PF_TRUECOLOR pixel format - * and use XPutPixel everywhere except in the implementation of - * glClear(GL_COLOR_BUFFER_BIT). We want this function to be fast so - * instead of using XPutPixel we "poke" our values after byte-swapping - * the clear pixel value if needed. + * into the XImage's buffer then the pixel must be byte swapped first. * */ @@ -77,8 +72,8 @@ #include "state_tracker/st_public.h" #include "state_tracker/st_context.h" -#include "pipe/softpipe/sp_context.h" #include "pipe/p_defines.h" +#include "pipe/p_context.h" /** * Global X driver lock @@ -135,24 +130,6 @@ static int check_for_xshm( XMesaDisplay *display ) } -/** - * Apply gamma correction to an intensity value in [0..max]. Return the - * new intensity value. - */ -static GLint -gamma_adjust( GLfloat gamma, GLint value, GLint max ) -{ - if (gamma == 1.0) { - return value; - } - else { - double x = (double) value / (double) max; - return IROUND_POS((GLfloat) max * _mesa_pow(x, 1.0F/gamma)); - } -} - - - /** * Return the true number of bits per pixel for XImages. * For example, if we request a 24-bit deep visual we may actually need/get @@ -252,19 +229,19 @@ get_drawable_size( XMesaDisplay *dpy, Drawable d, GLuint *width, GLuint *height * \param width returns width in pixels * \param height returns height in pixels */ -void +static void xmesa_get_window_size(XMesaDisplay *dpy, XMesaBuffer b, GLuint *width, GLuint *height) { #ifdef XFree86Server - *width = MIN2(b->frontxrb->drawable->width, MAX_WIDTH); - *height = MIN2(b->frontxrb->drawable->height, MAX_HEIGHT); + *width = MIN2(b->drawable->width, MAX_WIDTH); + *height = MIN2(b->drawable->height, MAX_HEIGHT); #else Status stat; _glthread_LOCK_MUTEX(_xmesa_lock); XSync(b->xm_visual->display, 0); /* added for Chromium */ - stat = get_drawable_size(dpy, b->frontxrb->pixmap, width, height); + stat = get_drawable_size(dpy, b->drawable, width, height); _glthread_UNLOCK_MUTEX(_xmesa_lock); if (!stat) { @@ -301,7 +278,6 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type, { XMesaBuffer b; GLframebuffer *fb; - struct pipe_winsys *winsys = xmesa_get_pipe_winsys(); ASSERT(type == WINDOW || type == PIXMAP || type == PBUFFER); @@ -309,7 +285,8 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type, if (!b) return NULL; - b->display = vis->display; + b->drawable = d; + b->xm_visual = vis; b->type = type; b->cmap = cmap; @@ -317,113 +294,20 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type, /* * Create framebuffer, but we'll plug in our own renderbuffers below. */ - b->stfb = st_create_framebuffer(&vis->mesa_visual, GL_FALSE, (void *) b); + b->stfb = st_create_framebuffer(&vis->mesa_visual, GL_TRUE, (void *) b); fb = &b->stfb->Base; - fb->Delete = xmesa_delete_framebuffer; - - /* - * XXX we want to create surfaces for pipe, not renderbuffers for Mesa. - */ - - /* - * Front renderbuffer - */ - b->frontxrb = xmesa_create_renderbuffer(winsys, 0, vis, GL_FALSE); - if (!b->frontxrb) { - _mesa_free(b); - return NULL; - } - b->frontxrb->Parent = b; - b->frontxrb->drawable = d; - b->frontxrb->pixmap = (XMesaPixmap) d; - _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &b->frontxrb->St.Base); - - /* - * Back renderbuffer - */ - if (vis->mesa_visual.doubleBufferMode) { - b->backxrb = xmesa_create_renderbuffer(winsys, 0, vis, GL_TRUE); - if (!b->backxrb) { - /* XXX free front xrb too */ - _mesa_free(b); - return NULL; - } - b->backxrb->Parent = b; - /* determine back buffer implementation */ - b->db_mode = vis->ximage_flag ? BACK_XIMAGE : BACK_PIXMAP; - - _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &b->backxrb->St.Base); - } - - /* - * Software alpha planes - */ - if (vis->mesa_visual.alphaBits > 0 - && vis->undithered_pf != PF_8A8B8G8R - && vis->undithered_pf != PF_8A8R8G8B) { - /* Visual has alpha, but pixel format doesn't support it. - * We'll use an alpha renderbuffer wrapper. - */ -#if 0 - b->swAlpha = GL_TRUE; /* don't do any renderbuffer wrapping */ -#else - b->swAlpha = GL_FALSE; -#endif - } - else { - b->swAlpha = GL_FALSE; - } - - if (vis->mesa_visual.depthBits > 0 && - vis->mesa_visual.stencilBits > 0) { - /* combined depth/stencil */ - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(GL_DEPTH24_STENCIL8_EXT); - _mesa_add_renderbuffer(fb, BUFFER_DEPTH, rb); - _mesa_add_renderbuffer(fb, BUFFER_STENCIL, rb); - } - else { - if (vis->mesa_visual.depthBits > 24) { - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(GL_DEPTH_COMPONENT32); - _mesa_add_renderbuffer(fb, BUFFER_DEPTH, rb); - } - else if (vis->mesa_visual.depthBits > 16) { - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(GL_DEPTH24_STENCIL8_EXT); - _mesa_add_renderbuffer(fb, BUFFER_DEPTH, rb); - } - else if (vis->mesa_visual.depthBits > 1) { - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(GL_DEPTH_COMPONENT16); - _mesa_add_renderbuffer(fb, BUFFER_DEPTH, rb); - } - - if (vis->mesa_visual.stencilBits > 0) { - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(GL_STENCIL_INDEX8_EXT); - _mesa_add_renderbuffer(fb, BUFFER_STENCIL, rb); - } - } - - if (vis->mesa_visual.accumRedBits > 0) { - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(GL_RGBA16); - _mesa_add_renderbuffer(fb, BUFFER_ACCUM, rb); - } - - /* - * Other renderbuffer (depth, stencil, etc) + * Create scratch XImage for xmesa_display_surface() */ - _mesa_add_soft_renderbuffers(fb, - GL_FALSE, /* color */ - GL_FALSE, /*vis->mesa_visual.haveDepthBuffer,*/ - GL_FALSE, /* stencil */ - GL_FALSE, /* accum */ - b->swAlpha, - vis->mesa_visual.numAuxBuffers > 0 ); + b->tempImage = XCreateImage(vis->display, + vis->visinfo->visual, + vis->visinfo->depth, + ZPixmap, 0, /* format, offset */ + NULL, /* data */ + 0, 0, /* size */ + 32, /* bitmap_pad */ + 0); /* bytes_per_line */ /* GLX_EXT_texture_from_pixmap */ b->TextureTarget = 0; @@ -446,8 +330,10 @@ XMesaBuffer xmesa_find_buffer(XMesaDisplay *dpy, XMesaColormap cmap, XMesaBuffer notThis) { XMesaBuffer b; - for (b=XMesaBufferList; b; b=b->Next) { - if (b->display==dpy && b->cmap==cmap && b!=notThis) { + for (b = XMesaBufferList; b; b = b->Next) { + if (b->xm_visual->display == dpy && + b->cmap == cmap && + b != notThis) { return b; } } @@ -479,7 +365,10 @@ xmesa_free_buffer(XMesaBuffer buffer) /* Since the X window for the XMesaBuffer is going away, we don't * want to dereference this pointer in the future. */ - b->frontxrb->drawable = 0; + b->drawable = 0; + + buffer->tempImage->data = NULL; + XDestroyImage(buffer->tempImage); /* Unreference. If count = zero we'll really delete the buffer */ _mesa_unreference_framebuffer(&fb); @@ -494,22 +383,6 @@ xmesa_free_buffer(XMesaBuffer buffer) } -/** - * Copy X color table stuff from one XMesaBuffer to another. - */ -static void -copy_colortable_info(XMesaBuffer dst, const XMesaBuffer src) -{ - MEMCPY(dst->color_table, src->color_table, sizeof(src->color_table)); - MEMCPY(dst->pixel_to_r, src->pixel_to_r, sizeof(src->pixel_to_r)); - MEMCPY(dst->pixel_to_g, src->pixel_to_g, sizeof(src->pixel_to_g)); - MEMCPY(dst->pixel_to_b, src->pixel_to_b, sizeof(src->pixel_to_b)); - dst->num_alloced = src->num_alloced; - MEMCPY(dst->alloced_colors, src->alloced_colors, - sizeof(src->alloced_colors)); -} - - /**********************************************************************/ /***** Misc Private Functions *****/ @@ -517,450 +390,49 @@ copy_colortable_info(XMesaBuffer dst, const XMesaBuffer src) /** - * A replacement for XAllocColor. This function should never - * fail to allocate a color. When XAllocColor fails, we return - * the nearest matching color. If we have to allocate many colors - * this function isn't too efficient; the XQueryColors() could be - * done just once. - * Written by Michael Pichler, Brian Paul, Mark Kilgard - * Input: dpy - X display - * cmap - X colormap - * cmapSize - size of colormap - * In/Out: color - the XColor struct - * Output: exact - 1=exact color match, 0=closest match - * alloced - 1=XAlloc worked, 0=XAlloc failed - */ -static void -noFaultXAllocColor( int client, - XMesaDisplay *dpy, - XMesaColormap cmap, - int cmapSize, - XMesaColor *color, - int *exact, int *alloced ) -{ -#ifdef XFree86Server - Pixel *ppixIn; - xrgb *ctable; -#else - /* we'll try to cache ctable for better remote display performance */ - static Display *prevDisplay = NULL; - static XMesaColormap prevCmap = 0; - static int prevCmapSize = 0; - static XMesaColor *ctable = NULL; -#endif - XMesaColor subColor; - int i, bestmatch; - double mindist; /* 3*2^16^2 exceeds long int precision. */ - - (void) client; - - /* First try just using XAllocColor. */ -#ifdef XFree86Server - if (AllocColor(cmap, - &color->red, &color->green, &color->blue, - &color->pixel, - client) == Success) -#else - if (XAllocColor(dpy, cmap, color)) -#endif - { - *exact = 1; - *alloced = 1; - return; - } - - /* Alloc failed, search for closest match */ - - /* Retrieve color table entries. */ - /* XXX alloca candidate. */ -#ifdef XFree86Server - ppixIn = (Pixel *) MALLOC(cmapSize * sizeof(Pixel)); - ctable = (xrgb *) MALLOC(cmapSize * sizeof(xrgb)); - for (i = 0; i < cmapSize; i++) { - ppixIn[i] = i; - } - QueryColors(cmap, cmapSize, ppixIn, ctable); -#else - if (prevDisplay != dpy || prevCmap != cmap - || prevCmapSize != cmapSize || !ctable) { - /* free previously cached color table */ - if (ctable) - _mesa_free(ctable); - /* Get the color table from X */ - ctable = (XMesaColor *) MALLOC(cmapSize * sizeof(XMesaColor)); - assert(ctable); - for (i = 0; i < cmapSize; i++) { - ctable[i].pixel = i; - } - XQueryColors(dpy, cmap, ctable, cmapSize); - prevDisplay = dpy; - prevCmap = cmap; - prevCmapSize = cmapSize; - } -#endif - - /* Find best match. */ - bestmatch = -1; - mindist = 0.0; - for (i = 0; i < cmapSize; i++) { - double dr = 0.30 * ((double) color->red - (double) ctable[i].red); - double dg = 0.59 * ((double) color->green - (double) ctable[i].green); - double db = 0.11 * ((double) color->blue - (double) ctable[i].blue); - double dist = dr * dr + dg * dg + db * db; - if (bestmatch < 0 || dist < mindist) { - bestmatch = i; - mindist = dist; - } - } - - /* Return result. */ - subColor.red = ctable[bestmatch].red; - subColor.green = ctable[bestmatch].green; - subColor.blue = ctable[bestmatch].blue; - /* Try to allocate the closest match color. This should only - * fail if the cell is read/write. Otherwise, we're incrementing - * the cell's reference count. - */ -#ifdef XFree86Server - if (AllocColor(cmap, - &subColor.red, &subColor.green, &subColor.blue, - &subColor.pixel, - client) == Success) { -#else - if (XAllocColor(dpy, cmap, &subColor)) { -#endif - *alloced = 1; - } - else { - /* do this to work around a problem reported by Frank Ortega */ - subColor.pixel = (unsigned long) bestmatch; - subColor.red = ctable[bestmatch].red; - subColor.green = ctable[bestmatch].green; - subColor.blue = ctable[bestmatch].blue; - subColor.flags = DoRed | DoGreen | DoBlue; - *alloced = 0; - } -#ifdef XFree86Server - _mesa_free(ppixIn); - _mesa_free(ctable); -#else - /* don't free table, save it for next time */ -#endif - - *color = subColor; - *exact = 0; -} - - - -/** - * Do setup for PF_GRAYSCALE pixel format. - * Note that buffer may be NULL. - */ -static GLboolean -setup_grayscale(int client, XMesaVisual v, - XMesaBuffer buffer, XMesaColormap cmap) -{ - if (GET_VISUAL_DEPTH(v)<4 || GET_VISUAL_DEPTH(v)>16) { - return GL_FALSE; - } - - if (buffer) { - XMesaBuffer prevBuffer; - - if (!cmap) { - return GL_FALSE; - } - - prevBuffer = xmesa_find_buffer(v->display, cmap, buffer); - if (prevBuffer && - (buffer->xm_visual->mesa_visual.rgbMode == - prevBuffer->xm_visual->mesa_visual.rgbMode)) { - /* Copy colormap stuff from previous XMesaBuffer which uses same - * X colormap. Do this to avoid time spent in noFaultXAllocColor. - */ - copy_colortable_info(buffer, prevBuffer); - } - else { - /* Allocate 256 shades of gray */ - int gray; - int colorsfailed = 0; - for (gray=0;gray<256;gray++) { - GLint r = gamma_adjust( v->RedGamma, gray, 255 ); - GLint g = gamma_adjust( v->GreenGamma, gray, 255 ); - GLint b = gamma_adjust( v->BlueGamma, gray, 255 ); - int exact, alloced; - XMesaColor xcol; - xcol.red = (r << 8) | r; - xcol.green = (g << 8) | g; - xcol.blue = (b << 8) | b; - noFaultXAllocColor( client, v->display, - cmap, GET_COLORMAP_SIZE(v), - &xcol, &exact, &alloced ); - if (!exact) { - colorsfailed++; - } - if (alloced) { - assert(buffer->num_alloced<256); - buffer->alloced_colors[buffer->num_alloced] = xcol.pixel; - buffer->num_alloced++; - } - - /*OLD - assert(gray < 576); - buffer->color_table[gray*3+0] = xcol.pixel; - buffer->color_table[gray*3+1] = xcol.pixel; - buffer->color_table[gray*3+2] = xcol.pixel; - assert(xcol.pixel < 65536); - buffer->pixel_to_r[xcol.pixel] = gray * 30 / 100; - buffer->pixel_to_g[xcol.pixel] = gray * 59 / 100; - buffer->pixel_to_b[xcol.pixel] = gray * 11 / 100; - */ - buffer->color_table[gray] = xcol.pixel; - assert(xcol.pixel < 65536); - buffer->pixel_to_r[xcol.pixel] = gray; - buffer->pixel_to_g[xcol.pixel] = gray; - buffer->pixel_to_b[xcol.pixel] = gray; - } - - if (colorsfailed && _mesa_getenv("MESA_DEBUG")) { - _mesa_warning(NULL, - "Note: %d out of 256 needed colors do not match exactly.\n", - colorsfailed ); - } - } - } - - v->dithered_pf = PF_Grayscale; - v->undithered_pf = PF_Grayscale; - return GL_TRUE; -} - - - -/** - * Setup RGB rendering for a window with a PseudoColor, StaticColor, - * or 8-bit TrueColor visual visual. We try to allocate a palette of 225 - * colors (5 red, 9 green, 5 blue) and dither to approximate a 24-bit RGB - * color. While this function was originally designed just for 8-bit - * visuals, it has also proven to work from 4-bit up to 16-bit visuals. - * Dithering code contributed by Bob Mercier. - */ -static GLboolean -setup_dithered_color(int client, XMesaVisual v, - XMesaBuffer buffer, XMesaColormap cmap) -{ - if (GET_VISUAL_DEPTH(v)<4 || GET_VISUAL_DEPTH(v)>16) { - return GL_FALSE; - } - - if (buffer) { - XMesaBuffer prevBuffer; - - if (!cmap) { - return GL_FALSE; - } - - prevBuffer = xmesa_find_buffer(v->display, cmap, buffer); - if (prevBuffer && - (buffer->xm_visual->mesa_visual.rgbMode == - prevBuffer->xm_visual->mesa_visual.rgbMode)) { - /* Copy colormap stuff from previous, matching XMesaBuffer. - * Do this to avoid time spent in noFaultXAllocColor. - */ - copy_colortable_info(buffer, prevBuffer); - } - else { - /* Allocate X colors and initialize color_table[], red_table[], etc */ - int r, g, b, i; - int colorsfailed = 0; - for (r = 0; r < DITH_R; r++) { - for (g = 0; g < DITH_G; g++) { - for (b = 0; b < DITH_B; b++) { - XMesaColor xcol; - int exact, alloced; - xcol.red =gamma_adjust(v->RedGamma, r*65535/(DITH_R-1),65535); - xcol.green=gamma_adjust(v->GreenGamma, g*65535/(DITH_G-1),65535); - xcol.blue =gamma_adjust(v->BlueGamma, b*65535/(DITH_B-1),65535); - noFaultXAllocColor( client, v->display, - cmap, GET_COLORMAP_SIZE(v), - &xcol, &exact, &alloced ); - if (!exact) { - colorsfailed++; - } - if (alloced) { - assert(buffer->num_alloced<256); - buffer->alloced_colors[buffer->num_alloced] = xcol.pixel; - buffer->num_alloced++; - } - i = DITH_MIX( r, g, b ); - assert(i < 576); - buffer->color_table[i] = xcol.pixel; - assert(xcol.pixel < 65536); - buffer->pixel_to_r[xcol.pixel] = r * 255 / (DITH_R-1); - buffer->pixel_to_g[xcol.pixel] = g * 255 / (DITH_G-1); - buffer->pixel_to_b[xcol.pixel] = b * 255 / (DITH_B-1); - } - } - } - - if (colorsfailed && _mesa_getenv("MESA_DEBUG")) { - _mesa_warning(NULL, - "Note: %d out of %d needed colors do not match exactly.\n", - colorsfailed, DITH_R * DITH_G * DITH_B ); - } - } - } - - v->dithered_pf = PF_Dither; - v->undithered_pf = PF_Lookup; - return GL_TRUE; -} - - -/** - * Setup RGB rendering for a window with a True/DirectColor visual. + * Choose the pixel format for the given visual. + * This will tell the gallium driver how to pack pixel data into + * drawing surfaces. */ -static void -setup_truecolor(XMesaVisual v, XMesaBuffer buffer, XMesaColormap cmap) +static GLuint +choose_pixel_format(XMesaVisual v) { - unsigned long rmask, gmask, bmask; - (void) buffer; - (void) cmap; - - /* Compute red multiplier (mask) and bit shift */ - v->rshift = 0; - rmask = GET_REDMASK(v); - while ((rmask & 1)==0) { - v->rshift++; - rmask = rmask >> 1; - } - - /* Compute green multiplier (mask) and bit shift */ - v->gshift = 0; - gmask = GET_GREENMASK(v); - while ((gmask & 1)==0) { - v->gshift++; - gmask = gmask >> 1; - } - - /* Compute blue multiplier (mask) and bit shift */ - v->bshift = 0; - bmask = GET_BLUEMASK(v); - while ((bmask & 1)==0) { - v->bshift++; - bmask = bmask >> 1; - } - - /* - * Compute component-to-pixel lookup tables and dithering kernel - */ - { - static GLubyte kernel[16] = { - 0*16, 8*16, 2*16, 10*16, - 12*16, 4*16, 14*16, 6*16, - 3*16, 11*16, 1*16, 9*16, - 15*16, 7*16, 13*16, 5*16, - }; - GLint rBits = _mesa_bitcount(rmask); - GLint gBits = _mesa_bitcount(gmask); - GLint bBits = _mesa_bitcount(bmask); - GLint maxBits; - GLuint i; - - /* convert pixel components in [0,_mask] to RGB values in [0,255] */ - for (i=0; i<=rmask; i++) - v->PixelToR[i] = (unsigned char) ((i * 255) / rmask); - for (i=0; i<=gmask; i++) - v->PixelToG[i] = (unsigned char) ((i * 255) / gmask); - for (i=0; i<=bmask; i++) - v->PixelToB[i] = (unsigned char) ((i * 255) / bmask); - - /* convert RGB values from [0,255] to pixel components */ - - for (i=0;i<256;i++) { - GLint r = gamma_adjust(v->RedGamma, i, 255); - GLint g = gamma_adjust(v->GreenGamma, i, 255); - GLint b = gamma_adjust(v->BlueGamma, i, 255); - v->RtoPixel[i] = (r >> (8-rBits)) << v->rshift; - v->GtoPixel[i] = (g >> (8-gBits)) << v->gshift; - v->BtoPixel[i] = (b >> (8-bBits)) << v->bshift; - } - /* overflow protection */ - for (i=256;i<512;i++) { - v->RtoPixel[i] = v->RtoPixel[255]; - v->GtoPixel[i] = v->GtoPixel[255]; - v->BtoPixel[i] = v->BtoPixel[255]; - } - - /* setup dithering kernel */ - maxBits = rBits; - if (gBits > maxBits) maxBits = gBits; - if (bBits > maxBits) maxBits = bBits; - for (i=0;i<16;i++) { - v->Kernel[i] = kernel[i] >> maxBits; - } - - v->undithered_pf = PF_Truecolor; - v->dithered_pf = (GET_VISUAL_DEPTH(v)<24) ? PF_Dither_True : PF_Truecolor; - } - - /* - * Now check for TrueColor visuals which we can optimize. - */ - if ( GET_REDMASK(v) ==0x0000ff - && GET_GREENMASK(v)==0x00ff00 - && GET_BLUEMASK(v) ==0xff0000 + if ( GET_REDMASK(v) == 0x0000ff + && GET_GREENMASK(v) == 0x00ff00 + && GET_BLUEMASK(v) == 0xff0000 && CHECK_BYTE_ORDER(v) - && v->BitsPerPixel==32 - && v->RedGamma==1.0 && v->GreenGamma==1.0 && v->BlueGamma==1.0) { + && v->BitsPerPixel == 32) { /* common 32 bpp config used on SGI, Sun */ - v->undithered_pf = v->dithered_pf = PF_8A8B8G8R; /* ABGR */ - } - else if (GET_REDMASK(v) == 0xff0000 - && GET_GREENMASK(v)== 0x00ff00 - && GET_BLUEMASK(v) == 0x0000ff - && CHECK_BYTE_ORDER(v) - && v->RedGamma == 1.0 && v->GreenGamma == 1.0 && v->BlueGamma == 1.0){ - if (v->BitsPerPixel==32) { - /* if 32 bpp, and visual indicates 8 bpp alpha channel */ - if (GET_VISUAL_DEPTH(v) == 32 && v->mesa_visual.alphaBits == 8) - v->undithered_pf = v->dithered_pf = PF_8A8R8G8B; /* ARGB */ - else - v->undithered_pf = v->dithered_pf = PF_8R8G8B; /* xRGB */ + assert(0); + return 0 /* PIXEL_FORMAT_U_A8_B8_G8_A8 */; + } + else if ( GET_REDMASK(v) == 0xff0000 + && GET_GREENMASK(v) == 0x00ff00 + && GET_BLUEMASK(v) == 0x0000ff + && CHECK_BYTE_ORDER(v)) { + if (v->BitsPerPixel == 32) { + return PIPE_FORMAT_U_A8_R8_G8_B8; } else if (v->BitsPerPixel == 24) { - v->undithered_pf = v->dithered_pf = PF_8R8G8B24; /* RGB */ + /* not supported yet */ + abort(); + return 0; } } - else if (GET_REDMASK(v) ==0xf800 - && GET_GREENMASK(v)==0x07e0 - && GET_BLUEMASK(v) ==0x001f - && CHECK_BYTE_ORDER(v) - && v->BitsPerPixel==16 - && v->RedGamma==1.0 && v->GreenGamma==1.0 && v->BlueGamma==1.0) { + else if ( GET_REDMASK(v) == 0xf800 + && GET_GREENMASK(v) == 0x07e0 + && GET_BLUEMASK(v) == 0x001f + && CHECK_BYTE_ORDER(v) + && v->BitsPerPixel == 16) { /* 5-6-5 RGB */ - v->undithered_pf = PF_5R6G5B; - v->dithered_pf = PF_Dither_5R6G5B; + return PIPE_FORMAT_U_R5_G6_B5; } -} - - -/** - * Setup RGB rendering for a window with a monochrome visual. - */ -static void -setup_monochrome( XMesaVisual v, XMesaBuffer b ) -{ - (void) b; - v->dithered_pf = v->undithered_pf = PF_1Bit; - /* if black=1 then we must flip pixel values */ - v->bitFlip = (GET_BLACK_PIXEL(v) != 0); + assert(0); + return 0; } - /** * When a context is bound for the first time, we can finally finish * initializing the context's visual and buffer information. @@ -976,10 +448,8 @@ initialize_visual_and_buffer(XMesaVisual v, XMesaBuffer b, GLboolean rgb_flag, XMesaDrawable window, XMesaColormap cmap) { - int client = 0; - #ifdef XFree86Server - client = (window) ? CLIENT_ID(window->id) : 0; + int client = (window) ? CLIENT_ID(window->id) : 0; #endif ASSERT(!b || b->xm_visual == v); @@ -989,47 +459,22 @@ initialize_visual_and_buffer(XMesaVisual v, XMesaBuffer b, assert(v->BitsPerPixel > 0); if (rgb_flag == GL_FALSE) { - /* COLOR-INDEXED WINDOW: - * Even if the visual is TrueColor or DirectColor we treat it as - * being color indexed. This is weird but might be useful to someone. - */ - v->dithered_pf = v->undithered_pf = PF_Index; - v->mesa_visual.indexBits = GET_VISUAL_DEPTH(v); + /* COLOR-INDEXED WINDOW: not supported*/ + return GL_FALSE; } else { /* RGB WINDOW: * We support RGB rendering into almost any kind of visual. */ const int xclass = v->mesa_visual.visualType; - if (xclass == GLX_TRUE_COLOR || xclass == GLX_DIRECT_COLOR) { - setup_truecolor( v, b, cmap ); - } - else if (xclass == GLX_STATIC_GRAY && GET_VISUAL_DEPTH(v) == 1) { - setup_monochrome( v, b ); - } - else if (xclass == GLX_GRAY_SCALE || xclass == GLX_STATIC_GRAY) { - if (!setup_grayscale( client, v, b, cmap )) { - return GL_FALSE; - } - } - else if ((xclass == GLX_PSEUDO_COLOR || xclass == GLX_STATIC_COLOR) - && GET_VISUAL_DEPTH(v)>=4 && GET_VISUAL_DEPTH(v)<=16) { - if (!setup_dithered_color( client, v, b, cmap )) { - return GL_FALSE; - } - } - else { - _mesa_warning(NULL, "XMesa: RGB mode rendering not supported in given visual.\n"); + if (xclass != GLX_TRUE_COLOR && xclass == !GLX_DIRECT_COLOR) { + _mesa_warning(NULL, + "XMesa: RGB mode rendering not supported in given visual.\n"); return GL_FALSE; } v->mesa_visual.indexBits = 0; - - if (_mesa_getenv("MESA_NO_DITHER")) { - v->dithered_pf = v->undithered_pf; - } } - /* * If MESA_INFO env var is set print out some debugging info * which can help Brian figure out what's going on when a user @@ -1037,21 +482,14 @@ initialize_visual_and_buffer(XMesaVisual v, XMesaBuffer b, */ if (_mesa_getenv("MESA_INFO")) { _mesa_printf("X/Mesa visual = %p\n", (void *) v); - _mesa_printf("X/Mesa dithered pf = %u\n", v->dithered_pf); - _mesa_printf("X/Mesa undithered pf = %u\n", v->undithered_pf); _mesa_printf("X/Mesa level = %d\n", v->mesa_visual.level); _mesa_printf("X/Mesa depth = %d\n", GET_VISUAL_DEPTH(v)); _mesa_printf("X/Mesa bits per pixel = %d\n", v->BitsPerPixel); } if (b && window) { - char *data; - - /* Do window-specific initializations */ - /* these should have been set in create_xmesa_buffer */ - ASSERT(b->frontxrb->drawable == window); - ASSERT(b->frontxrb->pixmap == (XMesaPixmap) window); + ASSERT(b->drawable == window); /* Setup for single/double buffering */ if (v->mesa_visual.doubleBufferMode) { @@ -1059,59 +497,13 @@ initialize_visual_and_buffer(XMesaVisual v, XMesaBuffer b, b->shm = check_for_xshm( v->display ); } - /* X11 graphics contexts */ + /* X11 graphics context */ #ifdef XFree86Server b->gc = CreateScratchGC(v->display, window->depth); #else b->gc = XCreateGC( v->display, window, 0, NULL ); #endif XMesaSetFunction( v->display, b->gc, GXcopy ); - - /* cleargc - for glClear() */ -#ifdef XFree86Server - b->cleargc = CreateScratchGC(v->display, window->depth); -#else - b->cleargc = XCreateGC( v->display, window, 0, NULL ); -#endif - XMesaSetFunction( v->display, b->cleargc, GXcopy ); - - /* - * Don't generate Graphics Expose/NoExpose events in swapbuffers(). - * Patch contributed by Michael Pichler May 15, 1995. - */ -#ifdef XFree86Server - b->swapgc = CreateScratchGC(v->display, window->depth); - { - CARD32 v[1]; - v[0] = FALSE; - dixChangeGC(NullClient, b->swapgc, GCGraphicsExposures, v, NULL); - } -#else - { - XGCValues gcvalues; - gcvalues.graphics_exposures = False; - b->swapgc = XCreateGC(v->display, window, - GCGraphicsExposures, &gcvalues); - } -#endif - XMesaSetFunction( v->display, b->swapgc, GXcopy ); - - /* Initialize the row buffer XImage for use in write_color_span() */ - data = (char*) MALLOC(MAX_WIDTH*4); -#ifdef XFree86Server - b->rowimage = XMesaCreateImage(GET_VISUAL_DEPTH(v), MAX_WIDTH, 1, data); -#else - b->rowimage = XCreateImage( v->display, - v->visinfo->visual, - v->visinfo->depth, - ZPixmap, 0, /*format, offset*/ - data, /*data*/ - MAX_WIDTH, 1, /*width, height*/ - 32, /*bitmap_pad*/ - 0 /*bytes_per_line*/ ); -#endif - if (!b->rowimage) - return GL_FALSE; } return GL_TRUE; @@ -1119,65 +511,6 @@ initialize_visual_and_buffer(XMesaVisual v, XMesaBuffer b, -/* - * Convert an RGBA color to a pixel value. - */ -unsigned long -xmesa_color_to_pixel(XMesaContext xmesa, - GLubyte r, GLubyte g, GLubyte b, GLubyte a, - GLuint pixelFormat) -{ - GLcontext *ctx = xmesa->st->ctx; - - switch (pixelFormat) { - case PF_Index: - return 0; - case PF_Truecolor: - { - unsigned long p; - PACK_TRUECOLOR( p, r, g, b ); - return p; - } - case PF_8A8B8G8R: - return PACK_8A8B8G8R( r, g, b, a ); - case PF_8A8R8G8B: - return PACK_8A8R8G8B( r, g, b, a ); - case PF_8R8G8B: - /* fall through */ - case PF_8R8G8B24: - return PACK_8R8G8B( r, g, b ); - case PF_5R6G5B: - return PACK_5R6G5B( r, g, b ); - case PF_Dither: - { - DITHER_SETUP; - return DITHER( 1, 0, r, g, b ); - } - case PF_1Bit: - /* 382 = (3*255)/2 */ - return ((r+g+b) > 382) ^ xmesa->xm_visual->bitFlip; - case PF_Lookup: - { - LOOKUP_SETUP; - return LOOKUP( r, g, b ); - } - case PF_Grayscale: - return GRAY_RGB( r, g, b ); - case PF_Dither_True: - /* fall through */ - case PF_Dither_5R6G5B: - { - unsigned long p; - PACK_TRUEDITHER(p, 1, 0, r, g, b); - return p; - } - default: - _mesa_problem(ctx, "Bad pixel format in xmesa_color_to_pixel"); - } - return 0; -} - - #define NUM_VISUAL_TYPES 6 /** @@ -1252,7 +585,6 @@ XMesaVisual XMesaCreateVisual( XMesaDisplay *display, GLint level, GLint visualCaveat ) { - char *gamma; XMesaVisual v; GLint red_bits, green_bits, blue_bits, alpha_bits; @@ -1287,19 +619,6 @@ XMesaVisual XMesaCreateVisual( XMesaDisplay *display, MEMCPY(v->visinfo, visinfo, sizeof(*visinfo)); #endif - /* check for MESA_GAMMA environment variable */ - gamma = _mesa_getenv("MESA_GAMMA"); - if (gamma) { - v->RedGamma = v->GreenGamma = v->BlueGamma = 0.0; - sscanf( gamma, "%f %f %f", &v->RedGamma, &v->GreenGamma, &v->BlueGamma ); - if (v->RedGamma<=0.0) v->RedGamma = 1.0; - if (v->GreenGamma<=0.0) v->GreenGamma = v->RedGamma; - if (v->BlueGamma<=0.0) v->BlueGamma = v->RedGamma; - } - else { - v->RedGamma = v->GreenGamma = v->BlueGamma = 1.0; - } - v->ximage_flag = ximage_flag; #ifdef XFree86Server @@ -1388,66 +707,6 @@ void XMesaDestroyVisual( XMesaVisual v ) -static void -finish_or_flush( GLcontext *ctx ) -{ -#ifdef XFree86Server - /* NOT_NEEDED */ -#else - const XMesaContext xmesa = XMESA_CONTEXT(ctx); - ctx->st->pipe->flush(ctx->st->pipe, 0); - if (xmesa) { - _glthread_LOCK_MUTEX(_xmesa_lock); - XSync( xmesa->display, False ); - _glthread_UNLOCK_MUTEX(_xmesa_lock); - } -#endif -} - - -/** - * Called by glViewport. - * This is a good time for us to poll the current X window size and adjust - * our renderbuffers to match the current window size. - * Remember, we have no opportunity to respond to conventional - * X Resize/StructureNotify events since the X driver has no event loop. - * Thus, we poll. - * Note that this trick isn't fool-proof. If the application never calls - * glViewport, our notion of the current window size may be incorrect. - * That problem led to the GLX_MESA_resize_buffers extension. - */ -static void -xmesa_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h) -{ - XMesaContext xmctx = XMESA_CONTEXT(ctx); - XMesaBuffer xmdrawbuf = XMESA_BUFFER(ctx->WinSysDrawBuffer); - XMesaBuffer xmreadbuf = XMESA_BUFFER(ctx->WinSysReadBuffer); - xmesa_check_and_update_buffer_size(xmctx, xmdrawbuf); - xmesa_check_and_update_buffer_size(xmctx, xmreadbuf); - (void) x; - (void) y; - (void) w; - (void) h; -} - - -/** - * Initialize the device driver function table with the functions - * we implement in this driver. - */ -static void -xmesa_init_driver_functions(struct dd_function_table *driver) -{ -#if 0 /* not needed for now */ - driver->Flush = finish_or_flush; - driver->Finish = finish_or_flush; -#else - (void) finish_or_flush; -#endif - driver->Viewport = xmesa_viewport; -} - - /** * Create a new XMesaContext. * \param v the XMesaVisual @@ -1462,6 +721,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) struct pipe_context *pipe; XMesaContext c; GLcontext *mesaCtx; + uint pf; if (firstTime) { _glthread_INIT_MUTEX(_xmesa_lock); @@ -1473,7 +733,9 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) if (!c) return NULL; - pipe = xmesa_create_context( c ); + pf = choose_pixel_format(v); + + pipe = xmesa_create_pipe_context( c, pf ); c->st = st_create_context(pipe, &v->mesa_visual, share_list ? share_list->st : NULL); @@ -1495,27 +757,11 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) mesaCtx->Const.CheckArrayBounds = GL_TRUE; #endif - xmesa_init_driver_functions(&mesaCtx->Driver); - /* finish up xmesa context initializations */ - c->swapbytes = CHECK_BYTE_ORDER(v) ? GL_FALSE : GL_TRUE; c->xm_visual = v; c->xm_buffer = NULL; /* set later by XMesaMakeCurrent */ - c->display = v->display; - c->pixelformat = v->dithered_pf; /* Dithering is enabled by default */ - - /* override these functions, as if the xlib driver were derived from - * the softpipe driver. - */ - pipe->get_tile = xmesa_get_tile; - pipe->put_tile = xmesa_put_tile; - pipe->get_tile_rgba = xmesa_get_tile_rgba; - pipe->put_tile_rgba = xmesa_put_tile_rgba; - c->st->haveFramebufferSurfaces = GL_FALSE; - - /* special pipe->clear function */ - pipe->clear = xmesa_clear; + c->st->haveFramebufferSurfaces = GL_TRUE; return c; } @@ -1735,8 +981,8 @@ XMesaDestroyBuffer(XMesaBuffer b) * and all attached renderbuffers. * Called when: * 1. the first time a buffer is bound to a context. - * 2. from glViewport to poll for window size changes - * 3. from the XMesaResizeBuffers() API function. + * 2. from the XMesaResizeBuffers() API function. + * 3. SwapBuffers. XXX probabaly from xm_flush_frontbuffer() too... * Note: it's possible (and legal) for xmctx to be NULL. That can happen * when resizing a buffer when no rendering context is bound. */ @@ -1744,7 +990,7 @@ void xmesa_check_and_update_buffer_size(XMesaContext xmctx, XMesaBuffer drawBuffer) { GLuint width, height; - xmesa_get_window_size(drawBuffer->display, drawBuffer, &width, &height); + xmesa_get_window_size(drawBuffer->xm_visual->display, drawBuffer, &width, &height); st_resize_framebuffer(drawBuffer->stfb, width, height); } @@ -1774,7 +1020,7 @@ GLboolean XMesaMakeCurrent2( XMesaContext c, XMesaBuffer drawBuffer, if (&(c->mesa) == _mesa_get_current_context() && c->mesa.DrawBuffer == &drawBuffer->mesa_buffer && c->mesa.ReadBuffer == &readBuffer->mesa_buffer - && XMESA_BUFFER(c->mesa.DrawBuffer)->wasCurrent) { + && xmesa_buffer(c->mesa.DrawBuffer)->wasCurrent) { /* same context and buffer, do nothing */ return GL_TRUE; } @@ -1793,22 +1039,6 @@ GLboolean XMesaMakeCurrent2( XMesaContext c, XMesaBuffer drawBuffer, st_make_current(c->st, drawBuffer->stfb, readBuffer->stfb); -#if 0 - if (c->xm_visual->mesa_visual.rgbMode) { - /* - * Must recompute and set these pixel values because colormap - * can be different for different windows. - */ - c->clearpixel = xmesa_color_to_pixel( c, - c->clearcolor[0], - c->clearcolor[1], - c->clearcolor[2], - c->clearcolor[3], - c->xm_visual->undithered_pf); - XMesaSetForeground(c->display, drawBuffer->cleargc, c->clearpixel); - } -#endif - /* Solution to Stephane Rehel's problem with glXReleaseBuffersMESA(): */ drawBuffer->wasCurrent = GL_TRUE; } @@ -1834,7 +1064,7 @@ XMesaContext XMesaGetCurrentContext( void ) { GET_CURRENT_CONTEXT(ctx); if (ctx) { - XMesaContext xmesa = XMESA_CONTEXT(ctx); + XMesaContext xmesa = xmesa_context(ctx); return xmesa; } else { @@ -1847,7 +1077,7 @@ XMesaBuffer XMesaGetCurrentBuffer( void ) { GET_CURRENT_CONTEXT(ctx); if (ctx) { - XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer); + XMesaBuffer xmbuf = xmesa_buffer(ctx->DrawBuffer); return xmbuf; } else { @@ -1861,7 +1091,7 @@ XMesaBuffer XMesaGetCurrentReadBuffer( void ) { GET_CURRENT_CONTEXT(ctx); if (ctx) { - return XMESA_BUFFER(ctx->ReadBuffer); + return xmesa_buffer(ctx->ReadBuffer); } else { return 0; @@ -1922,62 +1152,19 @@ GLboolean XMesaSetFXmode( GLint mode ) PUBLIC void XMesaSwapBuffers( XMesaBuffer b ) { - if (!b->backxrb) { - /* single buffered */ - return; - } + struct pipe_surface *surf; /* If we're swapping the buffer associated with the current context * we have to flush any pending rendering commands first. */ st_notify_swapbuffers(b->stfb); - if (b->db_mode) { - if (b->backxrb->ximage) { - /* Copy Ximage (back buf) from client memory to server window */ -#if defined(USE_XSHM) && !defined(XFree86Server) - if (b->shm) { - /*_glthread_LOCK_MUTEX(_xmesa_lock);*/ - XShmPutImage( b->xm_visual->display, b->frontxrb->drawable, - b->swapgc, - b->backxrb->ximage, 0, 0, - 0, 0, xmesa_buffer_width(b), xmesa_buffer_height(b), - False ); - /*_glthread_UNLOCK_MUTEX(_xmesa_lock);*/ - } - else -#endif - { - /*_glthread_LOCK_MUTEX(_xmesa_lock);*/ - XMesaPutImage( b->xm_visual->display, b->frontxrb->drawable, - b->swapgc, - b->backxrb->ximage, 0, 0, - 0, 0, - xmesa_buffer_width(b), xmesa_buffer_height(b)); - /*_glthread_UNLOCK_MUTEX(_xmesa_lock);*/ - } - } - else if (b->backxrb->pixmap) { - /* Copy pixmap (back buf) to window (front buf) on server */ - /*_glthread_LOCK_MUTEX(_xmesa_lock);*/ - XMesaCopyArea( b->xm_visual->display, - b->backxrb->pixmap, /* source drawable */ - b->frontxrb->drawable, /* dest. drawable */ - b->swapgc, - 0, 0, xmesa_buffer_width(b), xmesa_buffer_height(b), - 0, 0 /* dest region */ - ); - /*_glthread_UNLOCK_MUTEX(_xmesa_lock);*/ - } - - if (b->swAlpha) { - GET_CURRENT_CONTEXT(ctx); - _mesa_copy_soft_alpha_renderbuffers(ctx, &b->stfb->Base); - } + surf = st_get_framebuffer_surface(b->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + xmesa_display_surface(b, surf); } -#if !defined(XFree86Server) - XSync( b->xm_visual->display, False ); -#endif + + xmesa_check_and_update_buffer_size(NULL, b); } @@ -1987,82 +1174,22 @@ void XMesaSwapBuffers( XMesaBuffer b ) */ void XMesaCopySubBuffer( XMesaBuffer b, int x, int y, int width, int height ) { - if (!b->backxrb) { - /* single buffered */ - return; - } + struct pipe_surface *surf_front + = st_get_framebuffer_surface(b->stfb, ST_SURFACE_FRONT_LEFT); + struct pipe_surface *surf_back + = st_get_framebuffer_surface(b->stfb, ST_SURFACE_BACK_LEFT); + struct pipe_context *pipe = NULL; /* XXX fix */ - /* If we're swapping the buffer associated with the current context - * we have to flush any pending rendering commands first. - */ - st_notify_swapbuffers(b->stfb); + if (!surf_front || !surf_back) + return; - if (b->db_mode) { - int yTop = xmesa_buffer_height(b) - y - height; - if (b->backxrb->ximage) { - /* Copy Ximage from host's memory to server's window */ -#if defined(USE_XSHM) && !defined(XFree86Server) - if (b->shm) { - /* XXX assuming width and height aren't too large! */ - XShmPutImage( b->xm_visual->display, b->frontxrb->drawable, - b->swapgc, - b->backxrb->ximage, x, yTop, - x, yTop, width, height, False ); - /* wait for finished event??? */ - } - else -#endif - { - /* XXX assuming width and height aren't too large! */ - XMesaPutImage( b->xm_visual->display, b->frontxrb->drawable, - b->swapgc, - b->backxrb->ximage, x, yTop, - x, yTop, width, height ); - } - } - else { - /* Copy pixmap to window on server */ - XMesaCopyArea( b->xm_visual->display, - b->backxrb->pixmap, /* source drawable */ - b->frontxrb->drawable, /* dest. drawable */ - b->swapgc, - x, yTop, width, height, /* source region */ - x, yTop /* dest region */ - ); - } - } + pipe->surface_copy(pipe, + surf_front, x, y, /* dest */ + surf_back, x, y, /* src */ + width, height); } -/* - * Return a pointer to the XMesa backbuffer Pixmap or XImage. This function - * is a way to get "under the hood" of X/Mesa so one can manipulate the - * back buffer directly. - * Output: pixmap - pointer to back buffer's Pixmap, or 0 - * ximage - pointer to back buffer's XImage, or NULL - * Return: GL_TRUE = context is double buffered - * GL_FALSE = context is single buffered - */ -#ifndef XFree86Server -GLboolean XMesaGetBackBuffer( XMesaBuffer b, - XMesaPixmap *pixmap, - XMesaImage **ximage ) -{ - if (b->db_mode) { - if (pixmap) - *pixmap = b->backxrb->pixmap; - if (ximage) - *ximage = b->backxrb->ximage; - return GL_TRUE; - } - else { - *pixmap = 0; - *ximage = NULL; - return GL_FALSE; - } -} -#endif /* XFree86Server */ - /* * Return the depth buffer associated with an XMesaBuffer. @@ -2075,33 +1202,32 @@ GLboolean XMesaGetBackBuffer( XMesaBuffer b, GLboolean XMesaGetDepthBuffer( XMesaBuffer b, GLint *width, GLint *height, GLint *bytesPerValue, void **buffer ) { - struct gl_renderbuffer *rb - = b->stfb->Base.Attachment[BUFFER_DEPTH].Renderbuffer; - if (!rb || !rb->Data) { + struct pipe_surface *surf + = st_get_framebuffer_surface(b->stfb, ST_SURFACE_DEPTH); + if (surf) { + *width = surf->width; + *height = surf->pitch; + *bytesPerValue = surf->cpp; + *buffer = surf->map; + return GL_TRUE; + } + else { *width = 0; *height = 0; *bytesPerValue = 0; *buffer = 0; return GL_FALSE; } - else { - *width = xmesa_buffer_width(b); - *height = xmesa_buffer_height(b); - *bytesPerValue = b->stfb->Base.Visual.depthBits <= 16 - ? sizeof(GLushort) : sizeof(GLuint); - *buffer = rb->Data; - return GL_TRUE; - } } void XMesaFlush( XMesaContext c ) { - if (c && c->display) { + if (c && c->xm_visual->display) { #ifdef XFree86Server /* NOT_NEEDED */ #else - XSync( c->display, False ); + XSync( c->xm_visual->display, False ); #endif } } @@ -2128,7 +1254,7 @@ XMesaBuffer XMesaFindBuffer( XMesaDisplay *dpy, XMesaDrawable d ) { XMesaBuffer b; for (b=XMesaBufferList; b; b=b->Next) { - if (b->frontxrb->drawable == d && b->display == dpy) { + if (b->drawable == d && b->xm_visual->display == dpy) { return b; } } @@ -2144,7 +1270,7 @@ void xmesa_destroy_buffers_on_display(XMesaDisplay *dpy) XMesaBuffer b, next; for (b = XMesaBufferList; b; b = next) { next = b->Next; - if (b->display == dpy) { + if (b->xm_visual->display == dpy) { xmesa_free_buffer(b); } } @@ -2160,12 +1286,15 @@ void XMesaGarbageCollect( void ) XMesaBuffer b, next; for (b=XMesaBufferList; b; b=next) { next = b->Next; - if (b->display && b->frontxrb->drawable && b->type == WINDOW) { + if (b->xm_visual && + b->xm_visual->display && + b->drawable && + b->type == WINDOW) { #ifdef XFree86Server /* NOT_NEEDED */ #else - XSync(b->display, False); - if (!window_exists( b->display, b->frontxrb->drawable )) { + XSync(b->xm_visual->display, False); + if (!window_exists( b->xm_visual->display, b->drawable )) { /* found a dead window, free the ancillary info */ XMesaDestroyBuffer( b ); } @@ -2179,55 +1308,7 @@ unsigned long XMesaDitherColor( XMesaContext xmesa, GLint x, GLint y, GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha ) { - GLcontext *ctx = xmesa->st->ctx; - GLint r = (GLint) (red * 255.0F); - GLint g = (GLint) (green * 255.0F); - GLint b = (GLint) (blue * 255.0F); - GLint a = (GLint) (alpha * 255.0F); - - switch (xmesa->pixelformat) { - case PF_Index: - return 0; - case PF_Truecolor: - { - unsigned long p; - PACK_TRUECOLOR( p, r, g, b ); - return p; - } - case PF_8A8B8G8R: - return PACK_8A8B8G8R( r, g, b, a ); - case PF_8A8R8G8B: - return PACK_8A8R8G8B( r, g, b, a ); - case PF_8R8G8B: - return PACK_8R8G8B( r, g, b ); - case PF_5R6G5B: - return PACK_5R6G5B( r, g, b ); - case PF_Dither: - { - DITHER_SETUP; - return DITHER( x, y, r, g, b ); - } - case PF_1Bit: - /* 382 = (3*255)/2 */ - return ((r+g+b) > 382) ^ xmesa->xm_visual->bitFlip; - case PF_Lookup: - { - LOOKUP_SETUP; - return LOOKUP( r, g, b ); - } - case PF_Grayscale: - return GRAY_RGB( r, g, b ); - case PF_Dither_5R6G5B: - /* fall through */ - case PF_Dither_True: - { - unsigned long p; - PACK_TRUEDITHER(p, x, y, r, g, b); - return p; - } - default: - _mesa_problem(NULL, "Bad pixel format in XMesaDitherColor"); - } + /* no longer supported */ return 0; } @@ -2240,149 +1321,19 @@ PUBLIC void XMesaResizeBuffers( XMesaBuffer b ) { GET_CURRENT_CONTEXT(ctx); - XMesaContext xmctx = XMESA_CONTEXT(ctx); + XMesaContext xmctx = xmesa_context(ctx); if (!xmctx) return; xmesa_check_and_update_buffer_size(xmctx, b); } -static GLint -xbuffer_to_renderbuffer(int buffer) -{ - assert(MAX_AUX_BUFFERS <= 4); - - switch (buffer) { - case GLX_FRONT_LEFT_EXT: - return BUFFER_FRONT_LEFT; - case GLX_FRONT_RIGHT_EXT: - return BUFFER_FRONT_RIGHT; - case GLX_BACK_LEFT_EXT: - return BUFFER_BACK_LEFT; - case GLX_BACK_RIGHT_EXT: - return BUFFER_BACK_RIGHT; - case GLX_AUX0_EXT: - return BUFFER_AUX0; - case GLX_AUX1_EXT: - return BUFFER_AUX1; - case GLX_AUX2_EXT: - return BUFFER_AUX2; - case GLX_AUX3_EXT: - return BUFFER_AUX3; - case GLX_AUX4_EXT: - case GLX_AUX5_EXT: - case GLX_AUX6_EXT: - case GLX_AUX7_EXT: - case GLX_AUX8_EXT: - case GLX_AUX9_EXT: - default: - /* BadValue error */ - return -1; - } -} PUBLIC void XMesaBindTexImage(XMesaDisplay *dpy, XMesaBuffer drawable, int buffer, const int *attrib_list) { -#if 0 - GET_CURRENT_CONTEXT(ctx); - const GLuint unit = ctx->Texture.CurrentUnit; - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; - struct gl_texture_object *texObj; -#endif - struct gl_renderbuffer *rb; - struct xmesa_renderbuffer *xrb; - GLint b; - XMesaImage *img = NULL; - GLboolean freeImg = GL_FALSE; - - b = xbuffer_to_renderbuffer(buffer); - if (b < 0) - return; - - if (drawable->TextureFormat == GLX_TEXTURE_FORMAT_NONE_EXT) - return; /* BadMatch error */ - - rb = drawable->stfb->Base.Attachment[b].Renderbuffer; - if (!rb) { - /* invalid buffer */ - return; - } - xrb = xmesa_renderbuffer(rb); - -#if 0 - switch (drawable->TextureTarget) { - case GLX_TEXTURE_1D_EXT: - texObj = texUnit->Current1D; - break; - case GLX_TEXTURE_2D_EXT: - texObj = texUnit->Current2D; - break; - case GLX_TEXTURE_RECTANGLE_EXT: - texObj = texUnit->CurrentRect; - break; - default: - return; /* BadMatch error */ - } -#endif - - /* - * The following is a quick and simple way to implement - * BindTexImage. The better way is to write some new FetchTexel() - * functions which would extract texels from XImages. We'd still - * need to use GetImage when texturing from a Pixmap (front buffer) - * but texturing from a back buffer (XImage) would avoid an image - * copy. - */ - - /* get XImage */ - if (xrb->pixmap) { - img = XMesaGetImage(dpy, xrb->pixmap, 0, 0, rb->Width, rb->Height, ~0L, - ZPixmap); - freeImg = GL_TRUE; - } - else if (xrb->ximage) { - img = xrb->ximage; - } - - /* store the XImage as a new texture image */ - if (img) { - GLenum format, type, intFormat; - if (img->bits_per_pixel == 32) { - format = GL_BGRA; - type = GL_UNSIGNED_BYTE; - intFormat = GL_RGBA; - } - else if (img->bits_per_pixel == 24) { - format = GL_BGR; - type = GL_UNSIGNED_BYTE; - intFormat = GL_RGB; - } - else if (img->bits_per_pixel == 16) { - format = GL_BGR; - type = GL_UNSIGNED_SHORT_5_6_5; - intFormat = GL_RGB; - } - else { - _mesa_problem(NULL, "Unexpected XImage format in XMesaBindTexImage"); - return; - } - if (drawable->TextureFormat == GLX_TEXTURE_FORMAT_RGBA_EXT) { - intFormat = GL_RGBA; - } - else if (drawable->TextureFormat == GLX_TEXTURE_FORMAT_RGB_EXT) { - intFormat = GL_RGB; - } - - _mesa_TexImage2D(GL_TEXTURE_2D, 0, intFormat, rb->Width, rb->Height, 0, - format, type, img->data); - - if (freeImg) { - XMesaDestroyImage(img); - } - } } @@ -2390,10 +1341,5 @@ XMesaBindTexImage(XMesaDisplay *dpy, XMesaBuffer drawable, int buffer, PUBLIC void XMesaReleaseTexImage(XMesaDisplay *dpy, XMesaBuffer drawable, int buffer) { - const GLint b = xbuffer_to_renderbuffer(buffer); - if (b < 0) - return; - - /* no-op for now */ } diff --git a/src/mesa/pipe/xlib/xm_buffer.c b/src/mesa/pipe/xlib/xm_buffer.c index 6fc2f4ba630..e49eafbb092 100644 --- a/src/mesa/pipe/xlib/xm_buffer.c +++ b/src/mesa/pipe/xlib/xm_buffer.c @@ -41,7 +41,6 @@ #include "state_tracker/st_context.h" -#if defined(USE_XSHM) && !defined(XFree86Server) static volatile int mesaXErrorFlag = 0; /** @@ -458,8 +457,8 @@ xmesa_delete_framebuffer(struct gl_framebuffer *fb) if (!xmesa_find_buffer(b->display, b->cmap, b)) { #ifdef XFree86Server int client = 0; - if (b->frontxrb->drawable) - client = CLIENT_ID(b->frontxrb->drawable->id); + if (b->drawable) + client = CLIENT_ID(b->drawable->id); (void)FreeColors(b->cmap, client, b->num_alloced, b->alloced_colors, 0); #else @@ -478,21 +477,6 @@ xmesa_delete_framebuffer(struct gl_framebuffer *fb) if (fb->Visual.doubleBufferMode) { /* free back ximage/pixmap/shmregion */ - if (b->backxrb->ximage) { -#if defined(USE_XSHM) && !defined(XFree86Server) - if (b->shm) { - XShmDetach( b->display, &b->shminfo ); - XDestroyImage( b->backxrb->ximage ); - shmdt( b->shminfo.shmaddr ); - } - else -#endif - XMesaDestroyImage( b->backxrb->ximage ); - b->backxrb->ximage = NULL; - } - if (b->backxrb->pixmap) { - XMesaFreePixmap( b->display, b->backxrb->pixmap ); - } } if (b->rowimage) { diff --git a/src/mesa/pipe/xlib/xm_surface.c b/src/mesa/pipe/xlib/xm_surface.c index 4a54b5c7e1a..6f5c7ac6ec1 100644 --- a/src/mesa/pipe/xlib/xm_surface.c +++ b/src/mesa/pipe/xlib/xm_surface.c @@ -70,689 +70,6 @@ const int xmesa_kernel1[16] = { -#define CLIP_TILE \ - do { \ - if (x >= ps->width) \ - return; \ - if (y >= ps->height) \ - return; \ - if (x + w > ps->width) \ - w = ps->width - x; \ - if (y + h > ps->height) \ - h = ps->height -y; \ - } while(0) - - - -/* - * The following functions are used to trap XGetImage() calls which - * generate BadMatch errors if the drawable isn't mapped. - */ - -#ifndef XFree86Server -static int caught_xgetimage_error = 0; -static int (*old_xerror_handler)( XMesaDisplay *dpy, XErrorEvent *ev ); -static unsigned long xgetimage_serial; - -/* - * This is the error handler which will be called if XGetImage fails. - */ -static int xgetimage_error_handler( XMesaDisplay *dpy, XErrorEvent *ev ) -{ - if (ev->serial==xgetimage_serial && ev->error_code==BadMatch) { - /* caught the expected error */ - caught_xgetimage_error = 0; - } - else { - /* call the original X error handler, if any. otherwise ignore */ - if (old_xerror_handler) { - (*old_xerror_handler)( dpy, ev ); - } - } - return 0; -} - - -/* - * Call this right before XGetImage to setup error trap. - */ -static void catch_xgetimage_errors( XMesaDisplay *dpy ) -{ - xgetimage_serial = NextRequest( dpy ); - old_xerror_handler = XSetErrorHandler( xgetimage_error_handler ); - caught_xgetimage_error = 0; -} - - -/* - * Call this right after XGetImage to check if an error occured. - */ -static int check_xgetimage_errors( void ) -{ - /* restore old handler */ - (void) XSetErrorHandler( old_xerror_handler ); - /* return 0=no error, 1=error caught */ - return caught_xgetimage_error; -} -#endif - - -/** - * Wrapper for XGetImage() that catches BadMatch errors that can occur - * when the window is unmapped or the x/y/w/h extend beyond the window - * bounds. - * If build into xserver, wrap the internal GetImage method. - */ -static XMesaImage * -xget_image(XMesaDisplay *dpy, Drawable d, int x, int y, uint w, uint h) -{ -#ifdef XFree86Server - uint bpp = 4; /* XXX fix this */ - XMesaImage *ximage = (XMesaImage *) malloc(sizeof(XMesaImage)); - if (ximage) { - ximage->data = malloc(width * height * bpp); - } - (*dpy->GetImage)(d, x, y, w, h, ZPixmap, ~0L, (pointer)ximage->data); - ximage->width = w; - ximage->height = h; - ximage->bytes_per_row = w * bpp; - return ximage; -#else - int error; - XMesaImage *ximage; - catch_xgetimage_errors(dpy); - ximage = XGetImage(dpy, d, x, y, w, h, AllPlanes, ZPixmap); - error = check_xgetimage_errors(); - return ximage; -#endif -} - - - -/** - * Return raw pixels from pixmap or XImage. - */ -void -xmesa_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, void *p, int dst_stride) -{ - const uint w0 = w; - struct xmesa_surface *xms = xmesa_surface(ps); - XMesaImage *ximage = NULL; - ubyte *dst = (ubyte *) p; - uint i; - - if (!xms->drawable && !xms->ximage) { - /* not an X surface */ - softpipe_get_tile(pipe, ps, x, y, w, h, p, dst_stride); - return; - } - - CLIP_TILE; - - if (!xms->ximage) { - /* XImage = pixmap data */ - assert(xms->drawable); - ximage = xget_image(xms->display, xms->drawable, x, y, w, h); - if (!ximage) - return; - x = y = 0; - } - else { - ximage = xms->ximage; - } - - /* this could be optimized/simplified */ - switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - if (!dst_stride) { - dst_stride = w0 * 4; - } - for (i = 0; i < h; i++) { - memcpy(dst, ximage->data + y * ximage->bytes_per_line + x * 4, 4 * w); - dst += dst_stride; - } - break; - case PIPE_FORMAT_U_R5_G6_B5: - if (!dst_stride) { - dst_stride = w0 * 2; - } - for (i = 0; i < h; i++) { - memcpy(dst, ximage->data + y * ximage->bytes_per_line + x * 2, 4 * 2); - dst += dst_stride; - } - break; - default: - assert(0); - } - - if (!xms->ximage) { - XMesaDestroyImage(ximage); - } -} - - -/** - * Put raw pixels into pixmap or XImage. - */ -void -xmesa_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, const void *p, int src_stride) -{ - const uint w0 = w; - struct xmesa_surface *xms = xmesa_surface(ps); - const ubyte *src = (const ubyte *) p; - XMesaImage *ximage; - - if (!xms->drawable && !xms->ximage) { - /* not an X surface */ - softpipe_put_tile(pipe, ps, x, y, w, h, p, src_stride); - return; - } - - CLIP_TILE; - - if (xms->ximage) { - /* put to ximage */ - ximage = xms->ximage; - char *dst; - uint i; - - /* this could be optimized/simplified */ - switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - if (!src_stride) { - src_stride = w0 * 4; - } - dst = ximage->data + y * ximage->bytes_per_line + x * 4; - for (i = 0; i < h; i++) { - memcpy(dst, src, w * 4); - dst += ximage->bytes_per_line; - src += src_stride; - } - break; - case PIPE_FORMAT_U_R5_G6_B5: - if (!src_stride) { - src_stride = w0 * 2; - } - dst = ximage->data + y * ximage->bytes_per_line + x * 2; - for (i = 0; i < h; i++) { - memcpy(dst, src, w * 2); - dst += ximage->bytes_per_line; - src += src_stride; - } - break; - default: - assert(0); - } - } - else { - /* put to pixmap/window */ - /* Create temp XImage for data */ -#ifdef XFree86Server - ximage = XMesaCreateImage(GET_VISUAL_DEPTH(v), w, h, p); -#else - XVisualInfo *visinfo = xms->xrb->Parent->xm_visual->visinfo; - ximage = XCreateImage(xms->display, - visinfo->visual, - visinfo->depth, - ZPixmap, 0, /* format, offset */ - (char *) p, /* data */ - w, h, /* width, height */ - 32, /* bitmap_pad */ - 0); /* bytes_per_line */ -#endif - - /* send XImage data to pixmap */ - XPutImage(xms->display, xms->drawable, xms->gc, - ximage, 0, 0, x, y, w, h); - /* clean-up */ - ximage->data = NULL; /* prevents freeing user data at 'p' */ - XMesaDestroyImage(ximage); - } -} - - -void -xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, float *pixels) -{ - const uint w0 = w; - struct xmesa_surface *xms = xmesa_surface(ps); - XMesaImage *ximage = NULL; - float *pRow = pixels; - uint i, j; - - if (!xms->drawable && !xms->ximage) { - /* not an X surface */ - softpipe_get_tile_rgba(pipe, ps, x, y, w, h, pixels); - return; - } - - CLIP_TILE; - - if (!xms->ximage) { - /* XImage = pixmap data */ - assert(xms->drawable); - ximage = xget_image(xms->display, xms->drawable, x, y, w, h); - if (!ximage) - return; - x = y = 0; - } - else { - ximage = xms->ximage; - } - - switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - { - const uint *src - = (uint *) (ximage->data + y * ximage->bytes_per_line + x * 4); - for (i = 0; i < h; i++) { - float *p = pRow; - for (j = 0; j < w; j++) { - uint pix = src[j]; - ubyte r = ((pix >> 16) & 0xff); - ubyte g = ((pix >> 8) & 0xff); - ubyte b = ( pix & 0xff); - ubyte a = ((pix >> 24) & 0xff); - p[0] = UBYTE_TO_FLOAT(r); - p[1] = UBYTE_TO_FLOAT(g); - p[2] = UBYTE_TO_FLOAT(b); - p[3] = UBYTE_TO_FLOAT(a); - p += 4; - } - src += ximage->width; - pRow += 4 * w0; - } - } - break; - case PIPE_FORMAT_U_B8_G8_R8_A8: - { - const uint *src - = (uint *) (ximage->data + y * ximage->bytes_per_line + x * 4); - for (i = 0; i < h; i++) { - float *p = pRow; - for (j = 0; j < w; j++) { - uint pix = src[j]; - ubyte r = ((pix >> 8) & 0xff); - ubyte g = ((pix >> 16) & 0xff); - ubyte b = ((pix >> 24) & 0xff); - ubyte a = ( pix & 0xff); - p[0] = UBYTE_TO_FLOAT(r); - p[1] = UBYTE_TO_FLOAT(g); - p[2] = UBYTE_TO_FLOAT(b); - p[3] = UBYTE_TO_FLOAT(a); - p += 4; - } - src += ximage->width; - pRow += 4 * w0; - } - } - break; - case PIPE_FORMAT_U_R5_G6_B5: - { - ushort *src - = (ushort *) (ximage->data + y * ximage->bytes_per_line + x * 2); - for (i = 0; i < h; i++) { - float *p = pRow; - for (j = 0; j < w; j++) { - uint pix = src[j]; - ubyte r = (pix >> 8) | ((pix >> 13) & 0x7); - ubyte g = (pix >> 3) | ((pix >> 9) & 0x3); - ubyte b = ((pix & 0x1f) << 3) | ((pix >> 2) & 0x3); - p[0] = UBYTE_TO_FLOAT(r); - p[1] = UBYTE_TO_FLOAT(g); - p[2] = UBYTE_TO_FLOAT(b); - p[3] = 1.0; - p += 4; - } - src += ximage->width; - pRow += 4 * w0; - } - } - break; - default: - fprintf(stderr, "Bad format in xmesa_get_tile_rgba()\n"); - assert(0); - } - - if (!xms->ximage) { - XMesaDestroyImage(ximage); - } -} - - -void -xmesa_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, const float *pixels) -{ - const uint x0 = x, y0 = y, w0 = w; - struct xmesa_surface *xms = xmesa_surface(ps); - XMesaImage *ximage; - uint i, j; - - if (!xms->drawable && !xms->ximage) { - /* not an X surface */ - softpipe_put_tile_rgba(pipe, ps, x, y, w, h, pixels); - return; - } - - CLIP_TILE; - - if (!xms->ximage) { - /* create temp XImage */ - char *data = (char *) malloc(w * h * 4); -#ifdef XFree86Server - ximage = XMesaCreateImage(GET_VISUAL_DEPTH(v), w, h, data); -#else - XVisualInfo *visinfo = xms->xrb->Parent->xm_visual->visinfo; - ximage = XCreateImage(xms->display, - visinfo->visual, - visinfo->depth, - ZPixmap, 0, /* format, offset */ - data, /* data */ - w, h, /* width, height */ - 32, /* bitmap_pad */ - 0); /* bytes_per_line */ -#endif - x = y = 0; - } - else { - ximage = xms->ximage; - } - - /* convert floats to ximage's format */ - switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - { - uint *dst - = (uint *) (ximage->data + y * ximage->bytes_per_line + x * 4); - const float *pRow = pixels; - for (i = 0; i < h; i++) { - const float *p = pRow; - for (j = 0; j < w; j++) { - ubyte r, g, b, a; - UNCLAMPED_FLOAT_TO_UBYTE(r, p[0]); - UNCLAMPED_FLOAT_TO_UBYTE(g, p[1]); - UNCLAMPED_FLOAT_TO_UBYTE(b, p[2]); - UNCLAMPED_FLOAT_TO_UBYTE(a, p[3]); - dst[j] = PACK_8A8R8G8B(r, g, b, a); - p += 4; - } - dst += ximage->width; - pRow += 4 * w0; - } - } - break; - case PIPE_FORMAT_U_B8_G8_R8_A8: - { - uint *dst - = (uint *) (ximage->data + y * ximage->bytes_per_line + x * 4); - const float *pRow = pixels; - for (i = 0; i < h; i++) { - const float *p = pRow; - for (j = 0; j < w; j++) { - ubyte r, g, b, a; - UNCLAMPED_FLOAT_TO_UBYTE(r, p[0]); - UNCLAMPED_FLOAT_TO_UBYTE(g, p[1]); - UNCLAMPED_FLOAT_TO_UBYTE(b, p[2]); - UNCLAMPED_FLOAT_TO_UBYTE(a, p[3]); - dst[j] = PACK_8B8G8R8A(r, g, b, a); - p += 4; - } - dst += ximage->width; - pRow += 4 * w0; - } - } - break; - case PIPE_FORMAT_U_R5_G6_B5: - { - ushort *dst = - (ushort *) (ximage->data + y * ximage->bytes_per_line + x * 2); - const float *pRow = pixels; - for (i = 0; i < h; i++) { - const float *p = pRow; - for (j = 0; j < w; j++) { - ubyte r, g, b; - UNCLAMPED_FLOAT_TO_UBYTE(r, p[0]); - UNCLAMPED_FLOAT_TO_UBYTE(g, p[1]); - UNCLAMPED_FLOAT_TO_UBYTE(b, p[2]); - dst[j] = PACK_5R6G5B(r, g, b); - p += 4; - } - dst += ximage->width; - pRow += 4 * w0; - } - } - break; - - default: - fprintf(stderr, "Bad format in xmesa_put_tile_rgba()\n"); - assert(0); - } - - if (!xms->ximage) { - /* send XImage data to pixmap */ - XPutImage(xms->display, xms->drawable, xms->gc, - ximage, 0, 0, x0, y0, w, h); - /* clean-up */ - free(ximage->data); - ximage->data = NULL; - XMesaDestroyImage(ximage); - } -} - - -static void -clear_pixmap_surface(struct pipe_context *pipe, struct pipe_surface *ps, - uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - assert(xms); - assert(xms->display); - assert(xms->drawable); - assert(xms->gc); - XMesaSetForeground( xms->display, xms->gc, value ); - XMesaFillRectangle( xms->display, xms->drawable, xms->gc, - 0, 0, ps->width, ps->height); -} - -static void -clear_nbit_ximage_surface(struct pipe_context *pipe, struct pipe_surface *ps, - uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - int width = xms->surface.width; - int height = xms->surface.height; - int i, j; - for (j = 0; j < height; j++) { - for (i = 0; i < width; i++) { - XMesaPutPixel(xms->ximage, i, j, value); - } - } -} - -static void -clear_8bit_ximage_surface(struct pipe_context *pipe, struct pipe_surface *ps, - uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - memset(xms->ximage->data, - value, - xms->ximage->bytes_per_line * xms->ximage->height); -} - -static void -clear_16bit_ximage_surface(struct pipe_context *pipe, struct pipe_surface *ps, - uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - const int n = xms->ximage->width * xms->ximage->height; - ushort *dst = (ushort *) xms->ximage->data; - int i; - for (i = 0; i < n; i++) { - dst[i] = value; - } -} - - -/* Optimized code provided by Nozomi Ytow */ -static void -clear_24bit_ximage_surface(struct pipe_context *pipe, struct pipe_surface *ps, - uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - const ubyte r = (value ) & 0xff; - const ubyte g = (value >> 8) & 0xff; - const ubyte b = (value >> 16) & 0xff; - - if (r == g && g == b) { - /* same value for all three components (gray) */ - memset(xms->ximage->data, r, - xms->ximage->bytes_per_line * xms->ximage->height); - } - else { - /* non-gray clear color */ - const int n = xms->ximage->width * xms->ximage->height; - int i; - bgr_t *ptr3 = (bgr_t *) xms->ximage->data; - for (i = 0; i < n; i++) { - ptr3->r = r; - ptr3->g = g; - ptr3->b = b; - ptr3++; - } - } -} - -static void -clear_32bit_ximage_surface(struct pipe_context *pipe, struct pipe_surface *ps, - uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - - if (value == 0) { - /* common case */ - memset(xms->ximage->data, value, - xms->ximage->bytes_per_line * xms->ximage->height); - } - else { - const int n = xms->ximage->width * xms->ximage->height; - uint *dst = (uint *) xms->ximage->data; - int i; - for (i = 0; i < n; i++) - dst[i] = value; - } -} - - - - -/** - * Called to create a pipe_surface for each X renderbuffer. - * Note: this is being used instead of pipe->surface_alloc() since we - * have special/unique quad read/write functions for X. - */ -struct pipe_surface * -xmesa_new_color_surface(struct pipe_winsys *winsys, GLuint pipeFormat) -{ - struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface); - - assert(pipeFormat); - - xms->surface.format = pipeFormat; - xms->surface.refcount = 1; - xms->surface.winsys = winsys; - - /* Note, the buffer we allocate doesn't actually have any storage - * since we're drawing into an XImage or Pixmap. - * The surface's size will get set in the xmesa_alloc_front/back_storage() - * functions. - */ - xms->surface.buffer = winsys->buffer_create(winsys, 0x0); - - return &xms->surface; -} - - -/** - * Called via pipe->surface_alloc() to create new surfaces (textures, - * renderbuffers, etc. - */ -struct pipe_surface * -xmesa_surface_alloc(struct pipe_context *pipe, GLuint pipeFormat) -{ - struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface); - - assert(pipe); - assert(pipeFormat); - - xms->surface.format = pipeFormat; - xms->surface.refcount = 1; - xms->surface.winsys = pipe->winsys; - - return &xms->surface; -} - - -/** - * Called via pipe->clear() to clear entire surface to a certain value. - * If the surface is not an X pixmap or XImage, pass the call to - * softpipe_clear(). - */ -void -xmesa_clear(struct pipe_context *pipe, struct pipe_surface *ps, uint value) -{ - struct xmesa_surface *xms = xmesa_surface(ps); - - /* XXX actually, we should just discard any cached tiles from this - * surface since we don't want to accidentally re-use them after clearing. - */ - pipe->flush(pipe, 0); - - { - struct softpipe_context *sp = softpipe_context(pipe); - if (ps == sp_tile_cache_get_surface(sp, sp->cbuf_cache[0])) { - float clear[4]; - clear[0] = 0.2; /* XXX hack */ - clear[1] = 0.2; - clear[2] = 0.2; - clear[3] = 0.2; - sp_tile_cache_clear(sp->cbuf_cache[0], clear); - } - } - -#if 1 - (void) clear_8bit_ximage_surface; - (void) clear_24bit_ximage_surface; -#endif - - if (xms->ximage) { - /* back color buffer */ - switch (xms->surface.format) { - case PIPE_FORMAT_U_R5_G6_B5: - clear_16bit_ximage_surface(pipe, ps, value); - break; - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_U_B8_G8_R8_A8: - clear_32bit_ximage_surface(pipe, ps, value); - break; - default: - clear_nbit_ximage_surface(pipe, ps, value); - break; - } - } - else if (xms->drawable) { - /* front color buffer */ - clear_pixmap_surface(pipe, ps, value); - } - else { - /* other kind of buffer */ - softpipe_clear(pipe, ps, value); - } -} - /** XXX unfinished sketch... */ struct pipe_surface * diff --git a/src/mesa/pipe/xlib/xm_winsys.c b/src/mesa/pipe/xlib/xm_winsys.c index 295174d4bba..b9dd3e3b6ee 100644 --- a/src/mesa/pipe/xlib/xm_winsys.c +++ b/src/mesa/pipe/xlib/xm_winsys.c @@ -38,6 +38,8 @@ #include "main/macros.h" #include "pipe/p_winsys.h" +#include "pipe/p_format.h" +#include "pipe/p_context.h" #include "pipe/softpipe/sp_winsys.h" @@ -54,17 +56,52 @@ struct xm_buffer }; +struct xmesa_surface +{ + struct pipe_surface surface; + /* no extra fields for now */ +}; + + +/** + * Derived from softpipe_winsys. + * We just need one extra field which indicates the pixel format to use for + * drawing surfaces so that we're compatible with the XVisual/window format. + */ +struct xmesa_softpipe_winsys +{ + struct softpipe_winsys spws; + uint pixelformat; +}; + + -/* Turn the softpipe opaque buffer pointer into a dri_bufmgr opaque +/** Cast wrapper */ +static INLINE struct xmesa_surface * +xmesa_surface(struct pipe_surface *ps) +{ + assert(0); + return (struct xmesa_surface *) ps; +} + +/** cast wrapper */ +static INLINE struct xmesa_softpipe_winsys * +xmesa_softpipe_winsys(struct softpipe_winsys *spws) +{ + return (struct xmesa_softpipe_winsys *) spws; +} + +/** + * Turn the softpipe opaque buffer pointer into a dri_bufmgr opaque * buffer pointer... */ -static inline struct xm_buffer * +static INLINE struct xm_buffer * xm_bo( struct pipe_buffer_handle *bo ) { return (struct xm_buffer *) bo; } -static inline struct pipe_buffer_handle * +static INLINE struct pipe_buffer_handle * pipe_bo( struct xm_buffer *bo ) { return (struct pipe_buffer_handle *) bo; @@ -156,6 +193,33 @@ xm_buffer_get_subdata(struct pipe_winsys *pws, struct pipe_buffer_handle *buf, memcpy(data, b + offset, size); } + +/** + * Display/copy the image in the surface into the X window specified + * by the XMesaBuffer. + */ +void +xmesa_display_surface(XMesaBuffer b, const struct pipe_surface *surf) +{ + XImage *ximage = b->tempImage; + struct xm_buffer *xm_buf = xm_bo(surf->buffer); + + /* check that the XImage has been previously initialized */ + assert(ximage->format); + assert(ximage->bitmap_unit); + + /* update XImage's fields */ + ximage->width = surf->width; + ximage->height = surf->height; + ximage->bytes_per_line = surf->pitch * (ximage->bits_per_pixel / 8); + ximage->data = xm_buf->data; + + /* display image in Window */ + XPutImage(b->xm_visual->display, b->drawable, b->gc, + ximage, 0, 0, 0, 0, surf->width, surf->height); +} + + static void xm_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, @@ -166,9 +230,13 @@ xm_flush_frontbuffer(struct pipe_winsys *pws, * If we instead did front buffer rendering to a temporary XImage, * this would be the place to copy the Ximage to the on-screen Window. */ + XMesaContext xmctx = (XMesaContext) context_private; + + xmesa_display_surface(xmctx->xm_buffer, surf); } + static void xm_printf(struct pipe_winsys *pws, const char *fmtString, ...) { @@ -178,6 +246,7 @@ xm_printf(struct pipe_winsys *pws, const char *fmtString, ...) va_end( args ); } + static const char * xm_get_name(struct pipe_winsys *pws) { @@ -219,7 +288,6 @@ round_up(unsigned n, unsigned multiple) return (n + multiple - 1) & ~(multiple - 1); } - static unsigned xm_surface_pitch(struct pipe_winsys *winsys, unsigned cpp, unsigned width, unsigned flags) @@ -243,12 +311,7 @@ xm_surface_alloc(struct pipe_winsys *ws, GLuint pipeFormat) xms->surface.format = pipeFormat; xms->surface.refcount = 1; xms->surface.winsys = ws; -#if 0 - /* - * This is really just a softpipe surface, not an XImage/Pixmap surface. - */ - softpipe_init_surface_funcs(&xms->surface); -#endif + return &xms->surface; } @@ -261,7 +324,7 @@ xm_surface_release(struct pipe_winsys *winsys, struct pipe_surface **s) surf->refcount--; if (surf->refcount == 0) { if (surf->buffer) - winsys->buffer_reference(winsys, &surf->buffer, NULL); + winsys->buffer_reference(winsys, &surf->buffer, NULL); free(surf); } *s = NULL; @@ -274,7 +337,7 @@ xm_surface_release(struct pipe_winsys *winsys, struct pipe_surface **s) * For Xlib, this is a singleton object. * Nothing special for the Xlib driver so no subclassing or anything. */ -struct pipe_winsys * +static struct pipe_winsys * xmesa_get_pipe_winsys(void) { static struct pipe_winsys *ws = NULL; @@ -308,59 +371,65 @@ xmesa_get_pipe_winsys(void) /** - * XXX this depends on the depths supported by the screen (8/16/32/etc). - * Maybe when we're about to create a context/drawable we create a new - * softpipe_winsys object that corresponds to the specified screen... + * The winsys being queried will have been created at glXCreateContext + * time, with a pixel format corresponding to the context's visual. * - * Also, this query only really matters for on-screen drawables. - * For textures and FBOs we (softpipe) can support any format.o + * XXX we should pass a flag indicating if the format is going to be + * use for a drawing surface vs. a texture. In the later case, we + * can support any format. */ static boolean xmesa_is_format_supported(struct softpipe_winsys *sws, uint format) { - /* Any format supported by softpipe can be listed here. - * This query is not used for allocating window-system color buffers - * (which would depend on the screen depth/bpp). - */ - switch (format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_S_R16_G16_B16_A16: - case PIPE_FORMAT_S8_Z24: - case PIPE_FORMAT_U_S8: - case PIPE_FORMAT_U_Z16: - case PIPE_FORMAT_U_Z32: + struct xmesa_softpipe_winsys *xmws = xmesa_softpipe_winsys(sws); + + if (format == xmws->pixelformat) { return TRUE; - default: - return FALSE; - }; + } + else { + /* non-color / window surface format */ + switch (format) { + case PIPE_FORMAT_S_R16_G16_B16_A16: + case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_U_S8: + case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_U_Z32: + return TRUE; + default: + return FALSE; + }; + } } /** * Return pointer to a softpipe_winsys object. - * For Xlib, this is a singleton object. */ static struct softpipe_winsys * -xmesa_get_softpipe_winsys(void) +xmesa_get_softpipe_winsys(uint pixelformat) { - static struct softpipe_winsys *spws = NULL; + struct xmesa_softpipe_winsys *xmws + = CALLOC_STRUCT(xmesa_softpipe_winsys); + if (!xmws) + return NULL; - if (!spws) { - spws = CALLOC_STRUCT(softpipe_winsys); - if (spws) { - spws->is_format_supported = xmesa_is_format_supported; - } - } + xmws->spws.is_format_supported = xmesa_is_format_supported; + xmws->pixelformat = pixelformat; - return spws; + return &xmws->spws; } struct pipe_context * -xmesa_create_context(XMesaContext xmesa) +xmesa_create_pipe_context(XMesaContext xmesa, uint pixelformat) { struct pipe_winsys *pws = xmesa_get_pipe_winsys(); - struct softpipe_winsys *spws = xmesa_get_softpipe_winsys(); + struct softpipe_winsys *spws = xmesa_get_softpipe_winsys(pixelformat); + struct pipe_context *pipe; - return softpipe_create( pws, spws ); + pipe = softpipe_create( pws, spws ); + if (pipe) + pipe->priv = xmesa; + + return pipe; } diff --git a/src/mesa/pipe/xlib/xmesaP.h b/src/mesa/pipe/xlib/xmesaP.h index a067dfd3e34..e8c2a597a69 100644 --- a/src/mesa/pipe/xlib/xmesaP.h +++ b/src/mesa/pipe/xlib/xmesaP.h @@ -32,42 +32,15 @@ #ifdef XFree86Server #include "xm_image.h" #endif -#include "state_tracker/st_cb_fbo.h" + #include "state_tracker/st_context.h" #include "state_tracker/st_public.h" -#include "pipe/softpipe/sp_context.h" -#include "pipe/softpipe/sp_surface.h" extern _glthread_Mutex _xmesa_lock; extern XMesaBuffer XMesaBufferList; -/* for PF_8R8G8B24 pixel format */ -typedef struct { - GLubyte b; - GLubyte g; - GLubyte r; -} bgr_t; - - -/** Framebuffer pixel formats */ -enum pixel_format { - PF_Index, /**< Color Index mode */ - PF_Truecolor, /**< TrueColor or DirectColor, any depth */ - PF_Dither_True, /**< TrueColor with dithering */ - PF_8A8R8G8B, /**< 32-bit TrueColor: 8-A, 8-R, 8-G, 8-B bits */ - PF_8A8B8G8R, /**< 32-bit TrueColor: 8-A, 8-B, 8-G, 8-R bits */ - PF_8R8G8B, /**< 32-bit TrueColor: 8-R, 8-G, 8-B bits */ - PF_8R8G8B24, /**< 24-bit TrueColor: 8-R, 8-G, 8-B bits */ - PF_5R6G5B, /**< 16-bit TrueColor: 5-R, 6-G, 5-B bits */ - PF_Dither, /**< Color-mapped RGB with dither */ - PF_Lookup, /**< Color-mapped RGB without dither */ - PF_1Bit, /**< monochrome dithering of RGB */ - PF_Grayscale, /**< Grayscale or StaticGray */ - PF_Dither_5R6G5B /**< 16-bit dithered TrueColor: 5-R, 6-G, 5-B */ -}; - /** * Visual inforation, derived from GLvisual. @@ -86,47 +59,17 @@ struct xmesa_visual { GLint BitsPerPixel; /* True bits per pixel for XImages */ GLboolean ximage_flag; /* Use XImage for back buffer (not pixmap)? */ - - enum pixel_format dithered_pf; /* Pixel format when dithering */ - enum pixel_format undithered_pf;/* Pixel format when not dithering */ - - GLfloat RedGamma; /* Gamma values, 1.0 is default */ - GLfloat GreenGamma; - GLfloat BlueGamma; - - /* For PF_TRUECOLOR */ - GLint rshift, gshift, bshift;/* Pixel color component shifts */ - GLubyte Kernel[16]; /* Dither kernel */ - unsigned long RtoPixel[512]; /* RGB to pixel conversion */ - unsigned long GtoPixel[512]; - unsigned long BtoPixel[512]; - GLubyte PixelToR[256]; /* Pixel to RGB conversion */ - GLubyte PixelToG[256]; - GLubyte PixelToB[256]; - - /* For PF_1BIT */ - int bitFlip; }; /** - * Context info, derived from GLcontext. + * Context info, derived from st_context. * Basically corresponds to a GLXContext. */ struct xmesa_context { struct st_context *st; - - XMesaVisual xm_visual; /* Describes the buffers */ - XMesaBuffer xm_buffer; /* current span/point/line/triangle buffer */ - - XMesaDisplay *display; /* == xm_visual->display */ - GLboolean swapbytes; /* Host byte order != display byte order? */ - GLboolean direct; /* Direct rendering context? */ - - enum pixel_format pixelformat; - - GLubyte clearcolor[4]; /* current clearing color */ - unsigned long clearpixel; /* current clearing pixel value */ + XMesaVisual xm_visual; /** pixel format info */ + XMesaBuffer xm_buffer; /** current drawbuffer */ }; @@ -141,46 +84,6 @@ typedef enum { } BufferType; -/** Values for db_mode: */ -/*@{*/ -#define BACK_PIXMAP 1 -#define BACK_XIMAGE 2 -/*@}*/ - - -/** - * An xmesa_renderbuffer represents the back or front color buffer. - * For the front color buffer: - * is the X window - * For the back color buffer: - * Either or will be used, never both. - * In any case, always equals . - * For stand-alone Mesa, we could merge and into one - * field. We don't do that for the server-side GLcore module because - * pixmaps and drawables are different and we'd need a bunch of casts. - */ -struct xmesa_renderbuffer -{ - struct st_renderbuffer St; /**< Base class (XXX temporary?) */ - - XMesaBuffer Parent; /**< The XMesaBuffer this renderbuffer belongs to */ - XMesaDrawable drawable; /* Usually the X window ID */ - XMesaPixmap pixmap; /* Back color buffer */ - XMesaImage *ximage; /* The back buffer, if not using a Pixmap */ - - GLubyte *origin1; /* used for PIXEL_ADDR1 macro */ - GLint width1; - GLushort *origin2; /* used for PIXEL_ADDR2 macro */ - GLint width2; - GLubyte *origin3; /* used for PIXEL_ADDR3 macro */ - GLint width3; - GLuint *origin4; /* used for PIXEL_ADDR4 macro */ - GLint width4; - - GLint bottom; /* used for FLIP macro, equals height - 1 */ -}; - - /** * Framebuffer information, derived from. * Basically corresponds to a GLXDrawable. @@ -190,22 +93,13 @@ struct xmesa_buffer { GLboolean wasCurrent; /* was ever the current buffer? */ XMesaVisual xm_visual; /* the X/Mesa visual */ - - XMesaDisplay *display; - BufferType type; /* window, pixmap, pbuffer or glxwindow */ - - struct xmesa_renderbuffer *frontxrb; /* front color renderbuffer */ - struct xmesa_renderbuffer *backxrb; /* back color renderbuffer */ - + XMesaDrawable drawable; /* Usually the X window ID */ XMesaColormap cmap; /* the X colormap */ + BufferType type; /* window, pixmap, pbuffer or glxwindow */ + XMesaImage *tempImage; unsigned long selectedEvents;/* for pbuffers only */ - GLint db_mode; /* 0 = single buffered */ - /* BACK_PIXMAP = use Pixmap for back buffer */ - /* BACK_XIMAGE = use XImage for back buffer */ - GLboolean swAlpha; - GLuint shm; /* X Shared Memory extension status: */ /* 0 = not available */ /* 1 = XImage support available */ @@ -214,33 +108,7 @@ struct xmesa_buffer { XShmSegmentInfo shminfo; #endif - XMesaImage *rowimage; /* Used for optimized span writing */ - XMesaPixmap stipple_pixmap; /* For polygon stippling */ - XMesaGC stipple_gc; /* For polygon stippling */ - XMesaGC gc; /* scratch GC for span, line, tri drawing */ - XMesaGC cleargc; /* GC for clearing the color buffer */ - XMesaGC swapgc; /* GC for swapping the color buffers */ - - /* The following are here instead of in the XMesaVisual - * because they depend on the window's colormap. - */ - - /* For PF_DITHER, PF_LOOKUP, PF_GRAYSCALE */ - unsigned long color_table[576]; /* RGB -> pixel value */ - - /* For PF_DITHER, PF_LOOKUP, PF_GRAYSCALE */ - GLubyte pixel_to_r[65536]; /* pixel value -> red */ - GLubyte pixel_to_g[65536]; /* pixel value -> green */ - GLubyte pixel_to_b[65536]; /* pixel value -> blue */ - - /* Used to do XAllocColor/XFreeColors accounting: */ - int num_alloced; -#if defined(XFree86Server) - Pixel alloced_colors[256]; -#else - unsigned long alloced_colors[256]; -#endif /* GLX_EXT_texture_from_pixmap */ GLint TextureTarget; /** GLX_TEXTURE_1D_EXT, for example */ @@ -251,325 +119,38 @@ struct xmesa_buffer { }; -/** - * If pixelformat==PF_TRUECOLOR: - */ -#define PACK_TRUECOLOR( PIXEL, R, G, B ) \ - PIXEL = xmesa->xm_visual->RtoPixel[R] \ - | xmesa->xm_visual->GtoPixel[G] \ - | xmesa->xm_visual->BtoPixel[B]; \ - - -/** - * If pixelformat==PF_TRUEDITHER: - */ -#define PACK_TRUEDITHER( PIXEL, X, Y, R, G, B ) \ -{ \ - int d = xmesa->xm_visual->Kernel[((X)&3) | (((Y)&3)<<2)]; \ - PIXEL = xmesa->xm_visual->RtoPixel[(R)+d] \ - | xmesa->xm_visual->GtoPixel[(G)+d] \ - | xmesa->xm_visual->BtoPixel[(B)+d]; \ -} - - - -/** - * If pixelformat==PF_8A8B8G8R: - */ -#define PACK_8A8B8G8R( R, G, B, A ) \ - ( ((A) << 24) | ((B) << 16) | ((G) << 8) | (R) ) - -/** - * Like PACK_8A8B8G8R() but don't use alpha. This is usually an acceptable - * shortcut. - */ -#define PACK_8B8G8R( R, G, B ) ( ((B) << 16) | ((G) << 8) | (R) ) - -#define PACK_8B8G8R8A( R, G, B, A ) \ - ( ((B) << 24) | ((G) << 16) | ((R) << 8) | (A) ) - - - - -/** - * If pixelformat==PF_8R8G8B: - */ -#define PACK_8R8G8B( R, G, B) ( ((R) << 16) | ((G) << 8) | (B) ) - - -/** - * If pixelformat==PF_5R6G5B: - */ -#define PACK_5R6G5B( R, G, B) ( (((R) & 0xf8) << 8) | (((G) & 0xfc) << 3) | ((B) >> 3) ) - - -/** - * If pixelformat==PF_8A8R8G8B: - */ -#define PACK_8A8R8G8B( R, G, B, A ) \ - ( ((A) << 24) | ((R) << 16) | ((G) << 8) | (B) ) - - - -/** - * If pixelformat==PF_DITHER: - * - * Improved 8-bit RGB dithering code contributed by Bob Mercier - * (mercier@hollywood.cinenet.net). Thanks Bob! - */ -#ifdef DITHER666 -# define DITH_R 6 -# define DITH_G 6 -# define DITH_B 6 -# define DITH_MIX(r,g,b) (((r) * DITH_G + (g)) * DITH_B + (b)) -#else -# define DITH_R 5 -# define DITH_G 9 -# define DITH_B 5 -# define DITH_MIX(r,g,b) (((g) << 6) | ((b) << 3) | (r)) -#endif -#define DITH_DX 4 -#define DITH_DY 4 -#define DITH_N (DITH_DX * DITH_DY) - -#define _dither(C, c, d) (((unsigned)((DITH_N * (C - 1) + 1) * c + d)) >> 12) - -#define MAXC 256 -extern const int xmesa_kernel8[DITH_DY * DITH_DX]; - -/* Dither for random X,Y */ -#define DITHER_SETUP \ - int __d; \ - unsigned long *ctable = XMESA_BUFFER(ctx->DrawBuffer)->color_table; - -#define DITHER( X, Y, R, G, B ) \ - (__d = xmesa_kernel8[(((Y)&3)<<2) | ((X)&3)], \ - ctable[DITH_MIX(_dither(DITH_R, (R), __d), \ - _dither(DITH_G, (G), __d), \ - _dither(DITH_B, (B), __d))]) - -/* Dither for random X, fixed Y */ -#define XDITHER_SETUP(Y) \ - int __d; \ - unsigned long *ctable = XMESA_BUFFER(ctx->DrawBuffer)->color_table; \ - const int *kernel = &xmesa_kernel8[ ((Y)&3) << 2 ]; - -#define XDITHER( X, R, G, B ) \ - (__d = kernel[(X)&3], \ - ctable[DITH_MIX(_dither(DITH_R, (R), __d), \ - _dither(DITH_G, (G), __d), \ - _dither(DITH_B, (B), __d))]) - - - -/* - * Dithering for flat-shaded triangles. Precompute all 16 possible - * pixel values given the triangle's RGB color. Contributed by Martin Shenk. - */ -#define FLAT_DITHER_SETUP( R, G, B ) \ - GLushort ditherValues[16]; \ - { \ - unsigned long *ctable = XMESA_BUFFER(ctx->DrawBuffer)->color_table; \ - int msdr = (DITH_N*((DITH_R)-1)+1) * (R); \ - int msdg = (DITH_N*((DITH_G)-1)+1) * (G); \ - int msdb = (DITH_N*((DITH_B)-1)+1) * (B); \ - int i; \ - for (i=0;i<16;i++) { \ - int k = xmesa_kernel8[i]; \ - int j = DITH_MIX( (msdr+k)>>12, (msdg+k)>>12, (msdb+k)>>12 );\ - ditherValues[i] = (GLushort) ctable[j]; \ - } \ - } - -#define FLAT_DITHER_ROW_SETUP(Y) \ - GLushort *ditherRow = ditherValues + ( ((Y)&3) << 2); - -#define FLAT_DITHER(X) ditherRow[(X)&3] - - - -/** - * If pixelformat==PF_LOOKUP: - */ -#define _dither_lookup(C, c) (((unsigned)((DITH_N * (C - 1) + 1) * c)) >> 12) - -#define LOOKUP_SETUP \ - unsigned long *ctable = XMESA_BUFFER(ctx->DrawBuffer)->color_table - -#define LOOKUP( R, G, B ) \ - ctable[DITH_MIX(_dither_lookup(DITH_R, (R)), \ - _dither_lookup(DITH_G, (G)), \ - _dither_lookup(DITH_B, (B)))] - - - -/** - * If pixelformat==PF_1BIT: - */ -extern const int xmesa_kernel1[16]; - -#define SETUP_1BIT int bitFlip = xmesa->xm_visual->bitFlip -#define DITHER_1BIT( X, Y, R, G, B ) \ - (( ((int)(R)+(int)(G)+(int)(B)) > xmesa_kernel1[(((Y)&3) << 2) | ((X)&3)] ) ^ bitFlip) - - - -/** - * If pixelformat==PF_GRAYSCALE: - */ -#define GRAY_RGB( R, G, B ) XMESA_BUFFER(ctx->DrawBuffer)->color_table[((R) + (G) + (B))/3] - - - -/** - * Converts a GL window Y coord to an X window Y coord: - */ -#define YFLIP(XRB, Y) ((XRB)->bottom - (Y)) - - -/** - * Return the address of a 1, 2 or 4-byte pixel in the buffer's XImage: - * X==0 is left, Y==0 is bottom. - */ -#define PIXEL_ADDR1(XRB, X, Y) \ - ( (XRB)->origin1 - (Y) * (XRB)->width1 + (X) ) - -#define PIXEL_ADDR2(XRB, X, Y) \ - ( (XRB)->origin2 - (Y) * (XRB)->width2 + (X) ) - -#define PIXEL_ADDR3(XRB, X, Y) \ - ( (bgr_t *) ( (XRB)->origin3 - (Y) * (XRB)->width3 + 3 * (X) )) - -#define PIXEL_ADDR4(XRB, X, Y) \ - ( (XRB)->origin4 - (Y) * (XRB)->width4 + (X) ) - - - -/* - * External functions: - */ - -extern struct xmesa_renderbuffer * -xmesa_create_renderbuffer(struct pipe_winsys *winsys, - GLuint name, XMesaVisual xmvis, - GLboolean backBuffer); - -extern void -xmesa_delete_framebuffer(struct gl_framebuffer *fb); - -extern XMesaBuffer -xmesa_find_buffer(XMesaDisplay *dpy, XMesaColormap cmap, XMesaBuffer notThis); - -extern unsigned long -xmesa_color_to_pixel( XMesaContext xmesa, - GLubyte r, GLubyte g, GLubyte b, GLubyte a, - GLuint pixelFormat ); - -extern void -xmesa_get_window_size(XMesaDisplay *dpy, XMesaBuffer b, - GLuint *width, GLuint *height); - -extern void -xmesa_check_and_update_buffer_size(XMesaContext xmctx, XMesaBuffer drawBuffer); - - -extern void -xmesa_destroy_buffers_on_display(XMesaDisplay *dpy); - - -/** - * Using a function instead of an ordinary cast is safer. - */ -static INLINE struct xmesa_renderbuffer * -xmesa_renderbuffer(struct gl_renderbuffer *rb) -{ - return (struct xmesa_renderbuffer *) rb; -} - - -/** - * Return pointer to XMesaContext corresponding to a Mesa GLcontext. - * Since we're using structure containment, it's just a cast!. - * XXX should use inlined function for better type safety. - */ +/** cast wrapper */ static INLINE XMesaContext -XMESA_CONTEXT(GLcontext *ctx) +xmesa_context(GLcontext *ctx) { return (XMesaContext) ctx->DriverCtx; } -/** - * Return pointer to XMesaBuffer corresponding to a Mesa GLframebuffer. - * Since we're using structure containment, it's just a cast!. - * XXX should use inlined function for better type safety. - */ +/** cast wrapper */ static INLINE XMesaBuffer -XMESA_BUFFER(GLframebuffer *fb) +xmesa_buffer(GLframebuffer *fb) { struct st_framebuffer *stfb = (struct st_framebuffer *) fb; return (XMesaBuffer) st_framebuffer_private(stfb); } - -struct pipe_context; - -struct xmesa_surface -{ - struct pipe_surface surface; - struct xmesa_renderbuffer *xrb; - XMesaDisplay *display; - BufferType type; - XMesaDrawable drawable; - XMesaImage *ximage; - XMesaGC gc; -}; - - -/** Cast wrapper */ -static INLINE struct xmesa_surface * -xmesa_surface(struct pipe_surface *ps) -{ - return (struct xmesa_surface *) ps; -} - - extern void -xmesa_clear(struct pipe_context *pipe, struct pipe_surface *ps, uint value); - -extern struct pipe_context * -xmesa_create_context(XMesaContext xm); - -extern struct pipe_surface * -xmesa_surface_alloc(struct pipe_context *pipe, GLuint format); - -extern struct pipe_surface * -xmesa_new_color_surface(struct pipe_winsys *winsys, GLuint format); - -extern struct pipe_winsys * -xmesa_get_pipe_winsys(void); - -extern void -xmesa_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, void *p, int dst_stride); +xmesa_delete_framebuffer(struct gl_framebuffer *fb); -extern void -xmesa_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, const void *p, int src_stride); +extern XMesaBuffer +xmesa_find_buffer(XMesaDisplay *dpy, XMesaColormap cmap, XMesaBuffer notThis); extern void -xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, float *p); +xmesa_check_and_update_buffer_size(XMesaContext xmctx, XMesaBuffer drawBuffer); extern void -xmesa_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps, - uint x, uint y, uint w, uint h, const float *p); - +xmesa_destroy_buffers_on_display(XMesaDisplay *dpy); -extern struct pipe_surface * -xmesa_create_front_surface(XMesaVisual vis, Window win); +extern struct pipe_context * +xmesa_create_pipe_context(XMesaContext xm, uint pixelformat); static INLINE GLuint xmesa_buffer_width(XMesaBuffer b) @@ -583,5 +164,7 @@ xmesa_buffer_height(XMesaBuffer b) return b->stfb->Base.Height; } +extern void +xmesa_display_surface(XMesaBuffer b, const struct pipe_surface *surf); #endif diff --git a/src/mesa/sources b/src/mesa/sources index 596d3b89d5d..8286e558d6c 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -334,9 +334,7 @@ X11_DRIVER_SOURCES = \ pipe/xlib/fakeglx.c \ pipe/xlib/xfonts.c \ pipe/xlib/xm_api.c \ - pipe/xlib/xm_buffer.c \ - pipe/xlib/xm_winsys.c \ - pipe/xlib/xm_surface.c + pipe/xlib/xm_winsys.c OSMESA_DRIVER_SOURCES = \ drivers/osmesa/osmesa.c -- cgit v1.2.3 From 5f512d94235593467129144306104d2054a055ac Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 07:58:31 -0700 Subject: Remove obsolete xlib driver files --- src/mesa/pipe/xlib/xm_buffer.c | 490 ---------------------------------------- src/mesa/pipe/xlib/xm_surface.c | 94 -------- 2 files changed, 584 deletions(-) delete mode 100644 src/mesa/pipe/xlib/xm_buffer.c delete mode 100644 src/mesa/pipe/xlib/xm_surface.c diff --git a/src/mesa/pipe/xlib/xm_buffer.c b/src/mesa/pipe/xlib/xm_buffer.c deleted file mode 100644 index e49eafbb092..00000000000 --- a/src/mesa/pipe/xlib/xm_buffer.c +++ /dev/null @@ -1,490 +0,0 @@ -/* - * Mesa 3-D graphics library - * Version: 6.5.2 - * - * Copyright (C) 1999-2006 Brian Paul All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included - * in all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN - * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - */ - - -/** - * \file xm_buffer.h - * Framebuffer and renderbuffer-related functions. - */ - - -#include "GL/xmesa.h" -#include "glxheader.h" -#include "xmesaP.h" -#include "main/imports.h" -#include "main/framebuffer.h" -#include "main/renderbuffer.h" -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "state_tracker/st_context.h" - - -static volatile int mesaXErrorFlag = 0; - -/** - * Catches potential Xlib errors. - */ -static int -mesaHandleXError(XMesaDisplay *dpy, XErrorEvent *event) -{ - (void) dpy; - (void) event; - mesaXErrorFlag = 1; - return 0; -} - -/** - * Allocate a shared memory XImage back buffer for the given XMesaBuffer. - * Return: GL_TRUE if success, GL_FALSE if error - */ -static GLboolean -alloc_back_shm_ximage(XMesaBuffer b, GLuint width, GLuint height) -{ - /* - * We have to do a _lot_ of error checking here to be sure we can - * really use the XSHM extension. It seems different servers trigger - * errors at different points if the extension won't work. Therefore - * we have to be very careful... - */ - GC gc; - int (*old_handler)(XMesaDisplay *, XErrorEvent *); - - if (width == 0 || height == 0) { - /* this will be true the first time we're called on 'b' */ - return GL_FALSE; - } - - b->backxrb->ximage = XShmCreateImage(b->xm_visual->display, - b->xm_visual->visinfo->visual, - b->xm_visual->visinfo->depth, - ZPixmap, NULL, &b->shminfo, - width, height); - if (b->backxrb->ximage == NULL) { - _mesa_warning(NULL, "alloc_back_buffer: Shared memory error (XShmCreateImage), disabling.\n"); - b->shm = 0; - return GL_FALSE; - } - - b->shminfo.shmid = shmget(IPC_PRIVATE, b->backxrb->ximage->bytes_per_line - * b->backxrb->ximage->height, IPC_CREAT|0777); - if (b->shminfo.shmid < 0) { - _mesa_warning(NULL, "shmget failed while allocating back buffer.\n"); - XDestroyImage(b->backxrb->ximage); - b->backxrb->ximage = NULL; - _mesa_warning(NULL, "alloc_back_buffer: Shared memory error (shmget), disabling.\n"); - b->shm = 0; - return GL_FALSE; - } - - b->shminfo.shmaddr = b->backxrb->ximage->data - = (char*)shmat(b->shminfo.shmid, 0, 0); - if (b->shminfo.shmaddr == (char *) -1) { - _mesa_warning(NULL, "shmat() failed while allocating back buffer.\n"); - XDestroyImage(b->backxrb->ximage); - shmctl(b->shminfo.shmid, IPC_RMID, 0); - b->backxrb->ximage = NULL; - _mesa_warning(NULL, "alloc_back_buffer: Shared memory error (shmat), disabling.\n"); - b->shm = 0; - return GL_FALSE; - } - - b->shminfo.readOnly = False; - mesaXErrorFlag = 0; - old_handler = XSetErrorHandler(mesaHandleXError); - /* This may trigger the X protocol error we're ready to catch: */ - XShmAttach(b->xm_visual->display, &b->shminfo); - XSync(b->xm_visual->display, False); - - if (mesaXErrorFlag) { - /* we are on a remote display, this error is normal, don't print it */ - XFlush(b->xm_visual->display); - mesaXErrorFlag = 0; - XDestroyImage(b->backxrb->ximage); - shmdt(b->shminfo.shmaddr); - shmctl(b->shminfo.shmid, IPC_RMID, 0); - b->backxrb->ximage = NULL; - b->shm = 0; - (void) XSetErrorHandler(old_handler); - return GL_FALSE; - } - - shmctl(b->shminfo.shmid, IPC_RMID, 0); /* nobody else needs it */ - - /* Finally, try an XShmPutImage to be really sure the extension works */ - gc = XCreateGC(b->xm_visual->display, b->frontxrb->drawable, 0, NULL); - XShmPutImage(b->xm_visual->display, b->frontxrb->drawable, gc, - b->backxrb->ximage, 0, 0, 0, 0, 1, 1 /*one pixel*/, False); - XSync(b->xm_visual->display, False); - XFreeGC(b->xm_visual->display, gc); - (void) XSetErrorHandler(old_handler); - if (mesaXErrorFlag) { - XFlush(b->xm_visual->display); - mesaXErrorFlag = 0; - XDestroyImage(b->backxrb->ximage); - shmdt(b->shminfo.shmaddr); - shmctl(b->shminfo.shmid, IPC_RMID, 0); - b->backxrb->ximage = NULL; - b->shm = 0; - return GL_FALSE; - } - - return GL_TRUE; -} -#else -static GLboolean -alloc_back_shm_ximage(XMesaBuffer b, GLuint width, GLuint height) -{ - /* Can't compile XSHM support */ - return GL_FALSE; -} -#endif - - -/** - * \return LSBFirst or MSBFirst - */ -static int host_byte_order( void ) -{ - int i = 1; - char *cptr = (char *) &i; - return (*cptr==1) ? LSBFirst : MSBFirst; -} - - -/** - * Setup an off-screen pixmap or Ximage to use as the back buffer. - * Input: b - the X/Mesa buffer - */ -static void -alloc_back_buffer(XMesaBuffer b, GLuint width, GLuint height) -{ - if (b->db_mode == BACK_XIMAGE) { - /* Deallocate the old backxrb->ximage, if any */ - if (b->backxrb->ximage) { -#if defined(USE_XSHM) && !defined(XFree86Server) - if (b->shm) { - XShmDetach(b->xm_visual->display, &b->shminfo); - XDestroyImage(b->backxrb->ximage); - shmdt(b->shminfo.shmaddr); - } - else -#endif - XMesaDestroyImage(b->backxrb->ximage); - b->backxrb->ximage = NULL; - } - - if (width == 0 || height == 0) - return; - - /* Allocate new back buffer */ - if (b->shm == 0 || !alloc_back_shm_ximage(b, width, height)) { - /* Allocate a regular XImage for the back buffer. */ -#ifdef XFree86Server - b->backxrb->ximage = XMesaCreateImage(b->xm_visual->BitsPerPixel, - width, height, NULL); -#else - b->backxrb->ximage = XCreateImage(b->xm_visual->display, - b->xm_visual->visinfo->visual, - GET_VISUAL_DEPTH(b->xm_visual), - ZPixmap, 0, /* format, offset */ - NULL, - width, height, - 8, 0); /* pad, bytes_per_line */ -#endif - if (!b->backxrb->ximage) { - _mesa_warning(NULL, "alloc_back_buffer: XCreateImage failed.\n"); - return; - } - b->backxrb->ximage->data = (char *) MALLOC(b->backxrb->ximage->height - * b->backxrb->ximage->bytes_per_line); - if (!b->backxrb->ximage->data) { - _mesa_warning(NULL, "alloc_back_buffer: MALLOC failed.\n"); - XMesaDestroyImage(b->backxrb->ximage); - b->backxrb->ximage = NULL; - } - } - b->backxrb->pixmap = None; - } - else if (b->db_mode == BACK_PIXMAP) { - /* Free the old back pixmap */ - if (b->backxrb->pixmap) { - XMesaFreePixmap(b->xm_visual->display, b->backxrb->pixmap); - b->backxrb->pixmap = 0; - } - - if (width > 0 && height > 0) { - /* Allocate new back pixmap */ - b->backxrb->pixmap = XMesaCreatePixmap(b->xm_visual->display, - b->frontxrb->drawable, - width, height, - GET_VISUAL_DEPTH(b->xm_visual)); - } - - b->backxrb->ximage = NULL; - } -} - - -static void -xmesa_delete_renderbuffer(struct gl_renderbuffer *rb) -{ - struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb); - if (xrb->St.surface) { - struct pipe_winsys *ws = xrb->St.surface->winsys; - ws->surface_release(ws, &xrb->St.surface); - } - - /* XXX Note: the ximage or Pixmap attached to this renderbuffer - * should probably get freed here, but that's currently done in - * XMesaDestroyBuffer(). - */ - _mesa_free(rb); -} - - -static void -finish_surface_init(GLcontext *ctx, struct xmesa_renderbuffer *xrb) -{ - struct pipe_context *pipe = ctx->st->pipe; - if (!xrb->St.surface->buffer) { - xrb->St.surface->buffer = pipe->winsys->buffer_create(pipe->winsys, 0x0); - } -} - - -/** - * Reallocate renderbuffer storage for front color buffer. - * Called via gl_renderbuffer::AllocStorage() - */ -static GLboolean -xmesa_alloc_front_storage(GLcontext *ctx, struct gl_renderbuffer *rb, - GLenum internalFormat, GLuint width, GLuint height) -{ - const XMesaContext xmesa = XMESA_CONTEXT(ctx); - struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb); - struct xmesa_surface *xms = xmesa_surface(xrb->St.surface); - - /* just clear these to be sure we don't accidentally use them */ - xrb->origin1 = NULL; - xrb->origin2 = NULL; - xrb->origin3 = NULL; - xrb->origin4 = NULL; - - /* for the FLIP macro: */ - xrb->bottom = height - 1; - - rb->Width = width; - rb->Height = height; - rb->InternalFormat = internalFormat; - - if (!xrb->St.surface || !xrb->St.surface->buffer) - finish_surface_init(ctx, xrb); - - /* surface info */ - xms->surface.width = width; - xms->surface.height = height; - xms->display = xmesa->display; - xms->drawable = xrb->drawable; - xms->gc = xrb->Parent->cleargc; - xms->ximage = NULL; - - return GL_TRUE; -} - - -/** - * Reallocate renderbuffer storage for back color buffer. - * Called via gl_renderbuffer::AllocStorage() - */ -static GLboolean -xmesa_alloc_back_storage(GLcontext *ctx, struct gl_renderbuffer *rb, - GLenum internalFormat, GLuint width, GLuint height) -{ - const XMesaContext xmesa = XMESA_CONTEXT(ctx); - struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb); - struct xmesa_surface *xms = xmesa_surface(xrb->St.surface); - - /* reallocate the back buffer XImage or Pixmap */ - assert(xrb->Parent); - alloc_back_buffer(xrb->Parent, width, height); - - /* same as front buffer */ - /* XXX why is this here? */ - (void) xmesa_alloc_front_storage(ctx, rb, internalFormat, width, height); - - /* plus... */ - if (xrb->ximage) { - /* Needed by PIXELADDR1 macro */ - xrb->width1 = xrb->ximage->bytes_per_line; - xrb->origin1 = (GLubyte *) xrb->ximage->data + xrb->width1 * (height - 1); - - /* Needed by PIXELADDR2 macro */ - xrb->width2 = xrb->ximage->bytes_per_line / 2; - xrb->origin2 = (GLushort *) xrb->ximage->data + xrb->width2 * (height - 1); - - /* Needed by PIXELADDR3 macro */ - xrb->width3 = xrb->ximage->bytes_per_line; - xrb->origin3 = (GLubyte *) xrb->ximage->data + xrb->width3 * (height - 1); - - /* Needed by PIXELADDR4 macro */ - xrb->width4 = xrb->ximage->width; - xrb->origin4 = (GLuint *) xrb->ximage->data + xrb->width4 * (height - 1); - } - else { - /* out of memory or buffer size is 0 x 0 */ - xrb->width1 = xrb->width2 = xrb->width3 = xrb->width4 = 0; - xrb->origin1 = NULL; - xrb->origin2 = NULL; - xrb->origin3 = NULL; - xrb->origin4 = NULL; - } - - if (!xrb->St.surface || !xrb->St.surface->buffer) - finish_surface_init(ctx, xrb); - - xrb->St.surface->width = width; - xrb->St.surface->height = height; - - /* surface info */ - xms->surface.width = width; - xms->surface.height = height; - xms->display = xmesa->display; - xms->drawable = xrb->drawable; - xms->gc = xrb->Parent->cleargc; - xms->ximage = xrb->ximage; - - return GL_TRUE; -} - - -/** - * Called to create the front/back color renderbuffers, not user-created - * renderbuffers. - */ -struct xmesa_renderbuffer * -xmesa_create_renderbuffer(struct pipe_winsys *winsys, - GLuint name, XMesaVisual xmvis, - GLboolean backBuffer) -{ - const GLvisual *visual = &xmvis->mesa_visual; - int byteOrder = ImageByteOrder(xmvis->display); /* LSBFirst or MSBFirst */ - struct xmesa_renderbuffer *xrb = CALLOC_STRUCT(xmesa_renderbuffer); - if (xrb) { - GLuint name = 0; - GLuint pipeFormat = 0; - struct xmesa_surface *xms; - - _mesa_init_renderbuffer(&xrb->St.Base, name); - - xrb->St.Base.Delete = xmesa_delete_renderbuffer; - if (backBuffer) - xrb->St.Base.AllocStorage = xmesa_alloc_back_storage; - else - xrb->St.Base.AllocStorage = xmesa_alloc_front_storage; - - if (visual->rgbMode) { - xrb->St.Base.InternalFormat = GL_RGBA; - xrb->St.Base._BaseFormat = GL_RGBA; - xrb->St.Base.DataType = GL_UNSIGNED_BYTE; - xrb->St.Base.RedBits = visual->redBits; - xrb->St.Base.GreenBits = visual->greenBits; - xrb->St.Base.BlueBits = visual->blueBits; - xrb->St.Base.AlphaBits = visual->alphaBits; - if (visual->redMask == 0xff0000 && - visual->greenMask == 0x00ff00 && - visual->blueMask == 0x0000ff) { - if (host_byte_order() != byteOrder) { - pipeFormat = PIPE_FORMAT_U_B8_G8_R8_A8; - /*printf("Using format B8_G8_R8_A8 (LE dpy)\n");*/ - } - else { - pipeFormat = PIPE_FORMAT_U_A8_R8_G8_B8; - /*printf("Using format A8_R8_G8_B8 (BE dpy)\n");*/ - } - } - else { - assert(0); - } - } - else { - xrb->St.Base.InternalFormat = GL_COLOR_INDEX; - xrb->St.Base._BaseFormat = GL_COLOR_INDEX; - xrb->St.Base.DataType = GL_UNSIGNED_INT; - xrb->St.Base.IndexBits = visual->indexBits; - } - /* only need to set Red/Green/EtcBits fields for user-created RBs */ - xrb->St.surface = xmesa_new_color_surface(winsys, pipeFormat); - xms = (struct xmesa_surface *) xrb->St.surface; - xms->xrb = xrb; - } - return xrb; -} - - -/** - * Called via gl_framebuffer::Delete() method when this buffer - * is _really_ being deleted. - */ -void -xmesa_delete_framebuffer(struct gl_framebuffer *fb) -{ - XMesaBuffer b = XMESA_BUFFER(fb); - - if (b->num_alloced > 0) { - /* If no other buffer uses this X colormap then free the colors. */ - if (!xmesa_find_buffer(b->display, b->cmap, b)) { -#ifdef XFree86Server - int client = 0; - if (b->drawable) - client = CLIENT_ID(b->drawable->id); - (void)FreeColors(b->cmap, client, - b->num_alloced, b->alloced_colors, 0); -#else - XFreeColors(b->display, b->cmap, - b->alloced_colors, b->num_alloced, 0); -#endif - } - } - - if (b->gc) - XMesaFreeGC(b->display, b->gc); - if (b->cleargc) - XMesaFreeGC(b->display, b->cleargc); - if (b->swapgc) - XMesaFreeGC(b->display, b->swapgc); - - if (fb->Visual.doubleBufferMode) { - /* free back ximage/pixmap/shmregion */ - } - - if (b->rowimage) { - _mesa_free( b->rowimage->data ); - b->rowimage->data = NULL; - XMesaDestroyImage( b->rowimage ); - } - - _mesa_free_framebuffer_data(fb); - _mesa_free(fb); -} diff --git a/src/mesa/pipe/xlib/xm_surface.c b/src/mesa/pipe/xlib/xm_surface.c deleted file mode 100644 index 6f5c7ac6ec1..00000000000 --- a/src/mesa/pipe/xlib/xm_surface.c +++ /dev/null @@ -1,94 +0,0 @@ -/* - * Mesa 3-D graphics library - * Version: 7.1 - * - * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included - * in all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN - * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - */ - - -/** - * \file xm_surface.c - * Code to allow the softpipe code to write to X windows/buffers. - * This is a bit of a hack for now. We've basically got two different - * abstractions for color buffers: gl_renderbuffer and pipe_surface. - * They'll need to get merged someday... - * For now, they're separate things that point to each other. - */ - - -#include "GL/xmesa.h" -#include "glxheader.h" -#include "xmesaP.h" -#include "main/context.h" -#include "main/imports.h" -#include "main/macros.h" - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "pipe/softpipe/sp_context.h" -#include "pipe/softpipe/sp_clear.h" -#include "pipe/softpipe/sp_tile_cache.h" -#include "pipe/softpipe/sp_surface.h" -#include "state_tracker/st_context.h" - - -/* - * Dithering kernels and lookup tables. - */ - -const int xmesa_kernel8[DITH_DY * DITH_DX] = { - 0 * MAXC, 8 * MAXC, 2 * MAXC, 10 * MAXC, - 12 * MAXC, 4 * MAXC, 14 * MAXC, 6 * MAXC, - 3 * MAXC, 11 * MAXC, 1 * MAXC, 9 * MAXC, - 15 * MAXC, 7 * MAXC, 13 * MAXC, 5 * MAXC, -}; - -const int xmesa_kernel1[16] = { - 0*47, 9*47, 4*47, 12*47, /* 47 = (255*3)/16 */ - 6*47, 2*47, 14*47, 8*47, - 10*47, 1*47, 5*47, 11*47, - 7*47, 13*47, 3*47, 15*47 -}; - - - - -/** XXX unfinished sketch... */ -struct pipe_surface * -xmesa_create_front_surface(XMesaVisual vis, Window win) -{ - struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface); - if (!xms) { - return NULL; - } - - xms->display = vis->display; - xms->drawable = win; - - xms->surface.format = PIPE_FORMAT_U_A8_R8_G8_B8; - xms->surface.refcount = 1; -#if 0 - xms->surface.region = pipe->winsys->region_alloc(pipe->winsys, - 1, 0, 0, 0x0); -#endif - return &xms->surface; -} - -- cgit v1.2.3 From 5002d1eb8872ce9f6b8da73367bf559178a64006 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 08:24:33 -0700 Subject: remove stray ; --- src/mesa/pipe/xlib/xm_winsys.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/xlib/xm_winsys.c b/src/mesa/pipe/xlib/xm_winsys.c index b9dd3e3b6ee..e8cb36eced4 100644 --- a/src/mesa/pipe/xlib/xm_winsys.c +++ b/src/mesa/pipe/xlib/xm_winsys.c @@ -397,7 +397,7 @@ xmesa_is_format_supported(struct softpipe_winsys *sws, uint format) return TRUE; default: return FALSE; - }; + } } } -- cgit v1.2.3 From f1285f238a957d9ac90be12f537f8ef8f2e5d897 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 08:24:56 -0700 Subject: byteswapping fixes in choose_pixel_format() --- src/mesa/pipe/xlib/xm_api.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) diff --git a/src/mesa/pipe/xlib/xm_api.c b/src/mesa/pipe/xlib/xm_api.c index ce7e3cb6c2d..5cb9ca0d363 100644 --- a/src/mesa/pipe/xlib/xm_api.c +++ b/src/mesa/pipe/xlib/xm_api.c @@ -400,23 +400,25 @@ choose_pixel_format(XMesaVisual v) if ( GET_REDMASK(v) == 0x0000ff && GET_GREENMASK(v) == 0x00ff00 && GET_BLUEMASK(v) == 0xff0000 - && CHECK_BYTE_ORDER(v) && v->BitsPerPixel == 32) { - /* common 32 bpp config used on SGI, Sun */ - assert(0); - return 0 /* PIXEL_FORMAT_U_A8_B8_G8_A8 */; + if (CHECK_BYTE_ORDER(v)) { + /* no byteswapping needed */ + return 0 /* PIXEL_FORMAT_U_A8_B8_G8_R8 */; + } + else { + return PIPE_FORMAT_U_R8_G8_B8_A8; + } } else if ( GET_REDMASK(v) == 0xff0000 && GET_GREENMASK(v) == 0x00ff00 && GET_BLUEMASK(v) == 0x0000ff - && CHECK_BYTE_ORDER(v)) { - if (v->BitsPerPixel == 32) { + && v->BitsPerPixel == 32) { + if (CHECK_BYTE_ORDER(v)) { + /* no byteswapping needed */ return PIPE_FORMAT_U_A8_R8_G8_B8; } - else if (v->BitsPerPixel == 24) { - /* not supported yet */ - abort(); - return 0; + else { + return PIPE_FORMAT_U_B8_G8_R8_A8; } } else if ( GET_REDMASK(v) == 0xf800 @@ -734,6 +736,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) return NULL; pf = choose_pixel_format(v); + assert(pf); pipe = xmesa_create_pipe_context( c, pf ); -- cgit v1.2.3 From 03dcde4caaf44f8afca0d7f40e6378b77d6de39d Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 08:28:59 -0700 Subject: Add PIPE_FORMAT_U_B8_G8_R8_A8 in default_rgba_format(). Also, rewrite/simplify default_rgba_format() and default_depth_format(). --- src/mesa/state_tracker/st_format.c | 50 +++++++++++++++++++++----------------- 1 file changed, 28 insertions(+), 22 deletions(-) diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index b6b9375967e..f93f7f48d94 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -275,20 +275,23 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat) } /** - * Search list of formats for first RGBA format. + * Find an RGBA format supported by the context/winsys. */ static GLuint default_rgba_format( struct pipe_context *pipe ) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_R8_G8_B8_A8 )) { - return PIPE_FORMAT_U_R8_G8_B8_A8; - } - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A8_R8_G8_B8 )) { - return PIPE_FORMAT_U_A8_R8_G8_B8; - } - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_R5_G6_B5 )) { - return PIPE_FORMAT_U_R5_G6_B5; + static const uint colorFormats[] = { + PIPE_FORMAT_U_R8_G8_B8_A8, + PIPE_FORMAT_U_A8_R8_G8_B8, + PIPE_FORMAT_U_B8_G8_R8_A8, + PIPE_FORMAT_U_R5_G6_B5 + }; + uint i; + for (i = 0; i < Elements(colorFormats); i++) { + if (pipe->is_format_supported( pipe, colorFormats[i] )) { + return colorFormats[i]; + } } return PIPE_FORMAT_NONE; } @@ -309,27 +312,28 @@ default_deep_rgba_format( /** - * Search list of formats for first depth/Z format. + * Find an Z format supported by the context/winsys. */ static GLuint default_depth_format( struct pipe_context *pipe ) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_Z16 )) { - return PIPE_FORMAT_U_Z16; - } - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_Z32 )) { - return PIPE_FORMAT_U_Z32; - } - if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) { - return PIPE_FORMAT_S8_Z24; - } - if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) { - return PIPE_FORMAT_Z24_S8; + static const uint zFormats[] = { + PIPE_FORMAT_U_Z16, + PIPE_FORMAT_U_Z32, + PIPE_FORMAT_S8_Z24, + PIPE_FORMAT_Z24_S8 + }; + uint i; + for (i = 0; i < Elements(zFormats); i++) { + if (pipe->is_format_supported( pipe, zFormats[i] )) { + return zFormats[i]; + } } return PIPE_FORMAT_NONE; } + /** * Choose the PIPE_FORMAT_ to use for storing a texture image based * on the user's internalFormat, format and type parameters. @@ -541,7 +545,7 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, */ const struct gl_texture_format * st_ChooseTextureFormat(GLcontext * ctx, GLint internalFormat, - GLenum format, GLenum type) + GLenum format, GLenum type) { #if 0 struct intel_context *intel = intel_context(ctx); @@ -550,6 +554,8 @@ st_ChooseTextureFormat(GLcontext * ctx, GLint internalFormat, const GLboolean do32bpt = 1; #endif + (void) ctx; + switch (internalFormat) { case 4: case GL_RGBA: -- cgit v1.2.3 From 99823680a51ccc46a76d0abe6a24e7e4465a4cc6 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 10:18:34 -0700 Subject: minor code movement --- src/mesa/pipe/softpipe/sp_context.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_context.c b/src/mesa/pipe/softpipe/sp_context.c index f7f0316cf2a..696b991334b 100644 --- a/src/mesa/pipe/softpipe/sp_context.c +++ b/src/mesa/pipe/softpipe/sp_context.c @@ -255,6 +255,8 @@ struct pipe_context *softpipe_create( struct pipe_winsys *pipe_winsys, /* queries */ softpipe->pipe.is_format_supported = softpipe_is_format_supported; + softpipe->pipe.get_name = softpipe_get_name; + softpipe->pipe.get_vendor = softpipe_get_vendor; softpipe->pipe.get_param = softpipe_get_param; softpipe->pipe.get_paramf = softpipe_get_paramf; @@ -307,9 +309,6 @@ struct pipe_context *softpipe_create( struct pipe_winsys *pipe_winsys, softpipe->pipe.end_query = softpipe_end_query; softpipe->pipe.wait_query = softpipe_wait_query; - softpipe->pipe.get_name = softpipe_get_name; - softpipe->pipe.get_vendor = softpipe_get_vendor; - /* textures */ softpipe->pipe.texture_create = softpipe_texture_create; softpipe->pipe.texture_release = softpipe_texture_release; -- cgit v1.2.3 From 0d74d51790d81f6a19145ddb2f95d83c607c89dd Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 10:21:56 -0700 Subject: Move struct softpipe_texture definition into sp_texture.h Also, added softpipe_texture() cast wrapper. --- src/mesa/pipe/softpipe/sp_quad_fs.c | 1 + src/mesa/pipe/softpipe/sp_state.h | 29 ----------------------- src/mesa/pipe/softpipe/sp_state_sampler.c | 3 ++- src/mesa/pipe/softpipe/sp_surface.c | 3 ++- src/mesa/pipe/softpipe/sp_texture.c | 2 +- src/mesa/pipe/softpipe/sp_texture.h | 39 +++++++++++++++++++++++++++++++ 6 files changed, 45 insertions(+), 32 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_quad_fs.c b/src/mesa/pipe/softpipe/sp_quad_fs.c index 7184fcda521..1aba54d12aa 100644 --- a/src/mesa/pipe/softpipe/sp_quad_fs.c +++ b/src/mesa/pipe/softpipe/sp_quad_fs.c @@ -49,6 +49,7 @@ #include "sp_state.h" #include "sp_headers.h" #include "sp_quad.h" +#include "sp_texture.h" #include "sp_tex_sample.h" diff --git a/src/mesa/pipe/softpipe/sp_state.h b/src/mesa/pipe/softpipe/sp_state.h index daf9955ca5f..ea9d2e62be6 100644 --- a/src/mesa/pipe/softpipe/sp_state.h +++ b/src/mesa/pipe/softpipe/sp_state.h @@ -52,35 +52,6 @@ struct sp_fragment_shader_state { #endif }; -struct softpipe_texture -{ - struct pipe_texture base; - - /* Derived from the above: - */ - unsigned pitch; - unsigned depth_pitch; /* per-image on i945? */ - unsigned total_height; - - unsigned nr_images[PIPE_MAX_TEXTURE_LEVELS]; - - /* Explicitly store the offset of each image for each cube face or - * depth value. Pretty much have to accept that hardware formats - * are going to be so diverse that there is no unified way to - * compute the offsets of depth/cube images within a mipmap level, - * so have to store them as a lookup table: - */ - unsigned *image_offset[PIPE_MAX_TEXTURE_LEVELS]; /**< array [depth] of offsets */ - - /* Includes image offset tables: - */ - unsigned long level_offset[PIPE_MAX_TEXTURE_LEVELS]; - - /* The data is held here: - */ - struct pipe_buffer_handle *buffer; -}; - void * softpipe_create_alpha_test_state(struct pipe_context *, const struct pipe_alpha_test_state *); diff --git a/src/mesa/pipe/softpipe/sp_state_sampler.c b/src/mesa/pipe/softpipe/sp_state_sampler.c index e71b9159e3f..173901f04e6 100644 --- a/src/mesa/pipe/softpipe/sp_state_sampler.c +++ b/src/mesa/pipe/softpipe/sp_state_sampler.c @@ -32,6 +32,7 @@ #include "pipe/p_util.h" #include "sp_context.h" #include "sp_state.h" +#include "sp_texture.h" #include "sp_tile_cache.h" @@ -73,7 +74,7 @@ softpipe_set_texture_state(struct pipe_context *pipe, struct softpipe_context *softpipe = softpipe_context(pipe); assert(unit < PIPE_MAX_SAMPLERS); - softpipe->texture[unit] = (struct softpipe_texture *)texture; /* ptr, not struct */ + softpipe->texture[unit] = softpipe_texture(texture); /* ptr, not struct */ sp_tile_cache_set_texture(softpipe->tex_cache[unit], texture); diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index c61e0842fc2..16f0209eee4 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -32,6 +32,7 @@ #include "sp_context.h" #include "sp_state.h" #include "sp_surface.h" +#include "sp_texture.h" #include "sp_tile_cache.h" /** @@ -568,7 +569,7 @@ softpipe_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { - struct softpipe_texture *spt = (struct softpipe_texture *)pt; + struct softpipe_texture *spt = softpipe_texture(pt); struct pipe_surface *ps; unsigned offset; /* in bytes */ diff --git a/src/mesa/pipe/softpipe/sp_texture.c b/src/mesa/pipe/softpipe/sp_texture.c index 53486f9bba5..e8cdd67435b 100644 --- a/src/mesa/pipe/softpipe/sp_texture.c +++ b/src/mesa/pipe/softpipe/sp_texture.c @@ -412,7 +412,7 @@ softpipe_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) __FUNCTION__, (void *) *pt, (*pt)->refcount - 1); */ if (--(*pt)->refcount <= 0) { - struct softpipe_texture *spt = (struct softpipe_texture *)*pt; + struct softpipe_texture *spt = softpipe_texture(*pt); uint i; /* diff --git a/src/mesa/pipe/softpipe/sp_texture.h b/src/mesa/pipe/softpipe/sp_texture.h index 2aca57bd1d2..732064d9867 100644 --- a/src/mesa/pipe/softpipe/sp_texture.h +++ b/src/mesa/pipe/softpipe/sp_texture.h @@ -6,6 +6,45 @@ struct pipe_context; struct pipe_texture; +struct softpipe_texture +{ + struct pipe_texture base; + + /* Derived from the above: + */ + unsigned pitch; + unsigned depth_pitch; /* per-image on i945? */ + unsigned total_height; + + unsigned nr_images[PIPE_MAX_TEXTURE_LEVELS]; + + /* Explicitly store the offset of each image for each cube face or + * depth value. Pretty much have to accept that hardware formats + * are going to be so diverse that there is no unified way to + * compute the offsets of depth/cube images within a mipmap level, + * so have to store them as a lookup table: + */ + unsigned *image_offset[PIPE_MAX_TEXTURE_LEVELS]; /**< array [depth] of offsets */ + + /* Includes image offset tables: + */ + unsigned long level_offset[PIPE_MAX_TEXTURE_LEVELS]; + + /* The data is held here: + */ + struct pipe_buffer_handle *buffer; +}; + + +/** cast wrapper */ +static INLINE struct softpipe_texture * +softpipe_texture(struct pipe_texture *pt) +{ + return (struct softpipe_texture *) pt; +} + + + extern void softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); -- cgit v1.2.3 From 1eb78225c4759f4295f0700f25f77c52d6acb2eb Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 13:08:28 -0700 Subject: add some whitespace --- src/mesa/pipe/softpipe/sp_context.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/mesa/pipe/softpipe/sp_context.c b/src/mesa/pipe/softpipe/sp_context.c index 696b991334b..54b2076b4a1 100644 --- a/src/mesa/pipe/softpipe/sp_context.c +++ b/src/mesa/pipe/softpipe/sp_context.c @@ -264,21 +264,27 @@ struct pipe_context *softpipe_create( struct pipe_winsys *pipe_winsys, softpipe->pipe.create_alpha_test_state = softpipe_create_alpha_test_state; softpipe->pipe.bind_alpha_test_state = softpipe_bind_alpha_test_state; softpipe->pipe.delete_alpha_test_state = softpipe_delete_alpha_test_state; + softpipe->pipe.create_blend_state = softpipe_create_blend_state; softpipe->pipe.bind_blend_state = softpipe_bind_blend_state; softpipe->pipe.delete_blend_state = softpipe_delete_blend_state; + softpipe->pipe.create_sampler_state = softpipe_create_sampler_state; softpipe->pipe.bind_sampler_state = softpipe_bind_sampler_state; softpipe->pipe.delete_sampler_state = softpipe_delete_sampler_state; + softpipe->pipe.create_depth_stencil_state = softpipe_create_depth_stencil_state; softpipe->pipe.bind_depth_stencil_state = softpipe_bind_depth_stencil_state; softpipe->pipe.delete_depth_stencil_state = softpipe_delete_depth_stencil_state; + softpipe->pipe.create_rasterizer_state = softpipe_create_rasterizer_state; softpipe->pipe.bind_rasterizer_state = softpipe_bind_rasterizer_state; softpipe->pipe.delete_rasterizer_state = softpipe_delete_rasterizer_state; + softpipe->pipe.create_fs_state = softpipe_create_fs_state; softpipe->pipe.bind_fs_state = softpipe_bind_fs_state; softpipe->pipe.delete_fs_state = softpipe_delete_fs_state; + softpipe->pipe.create_vs_state = softpipe_create_vs_state; softpipe->pipe.bind_vs_state = softpipe_bind_vs_state; softpipe->pipe.delete_vs_state = softpipe_delete_vs_state; -- cgit v1.2.3 From f36258d7e1668afcbd5b1deb6a405b9f211bf3ee Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 13:19:00 -0700 Subject: don't dereference pt after realloc - fixes valgrind error --- src/mesa/pipe/softpipe/sp_texture.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/softpipe/sp_texture.c b/src/mesa/pipe/softpipe/sp_texture.c index e8cdd67435b..2288c343bf1 100644 --- a/src/mesa/pipe/softpipe/sp_texture.c +++ b/src/mesa/pipe/softpipe/sp_texture.c @@ -386,7 +386,7 @@ softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) if (spt->buffer) { pipe->winsys->buffer_data(pipe->winsys, spt->buffer, - spt->pitch * (*pt)->cpp * + spt->pitch * spt->base.cpp * spt->total_height, NULL, PIPE_BUFFER_USAGE_PIXEL); } -- cgit v1.2.3 From dfc21b2efbc5ca9022fa82d7a97ec1b247e27cf9 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 15:32:13 -0700 Subject: Some tile cache improvements, clean-ups. Code for optimized clears (per-tile flag indicating clear status) in place but not enabled yet. --- src/mesa/pipe/softpipe/sp_clear.c | 4 +- src/mesa/pipe/softpipe/sp_state_surface.c | 8 +- src/mesa/pipe/softpipe/sp_tile_cache.c | 330 +++++++++++++++++------------- src/mesa/pipe/softpipe/sp_tile_cache.h | 12 +- 4 files changed, 206 insertions(+), 148 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index 496b38fd5f1..a4276362b99 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -50,12 +50,12 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, softpipe_update_derived(softpipe); /* not needed?? */ - if (ps == sp_tile_cache_get_surface(softpipe, softpipe->zbuf_cache)) { + if (ps == sp_tile_cache_get_surface(softpipe->zbuf_cache)) { float clear[4]; clear[0] = 1.0; /* XXX hack */ sp_tile_cache_clear(softpipe->zbuf_cache, clear); } - else if (ps == sp_tile_cache_get_surface(softpipe, softpipe->cbuf_cache[0])) { + else if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[0])) { float clear[4]; clear[0] = 0.2f; /* XXX hack */ clear[1] = 0.2f; /* XXX hack */ diff --git a/src/mesa/pipe/softpipe/sp_state_surface.c b/src/mesa/pipe/softpipe/sp_state_surface.c index 9470ef485fe..30bedc74bc0 100644 --- a/src/mesa/pipe/softpipe/sp_state_surface.c +++ b/src/mesa/pipe/softpipe/sp_state_surface.c @@ -66,7 +66,7 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, sp->framebuffer.cbufs[i] = fb->cbufs[i]; /* update cache */ - sp_tile_cache_set_surface(sp, sp->cbuf_cache[i], ps); + sp_tile_cache_set_surface(sp->cbuf_cache[i], ps); } } @@ -92,7 +92,7 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, sp->framebuffer.zbuf = fb->zbuf; /* update cache */ - sp_tile_cache_set_surface(sp, sp->zbuf_cache, ps); + sp_tile_cache_set_surface(sp->zbuf_cache, ps); } /* XXX combined depth/stencil here */ @@ -116,12 +116,12 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, if (fb->sbuf != fb->zbuf) { /* separate stencil buf */ sp->sbuf_cache = sp->sbuf_cache_sep; - sp_tile_cache_set_surface(sp, sp->sbuf_cache, ps); + sp_tile_cache_set_surface(sp->sbuf_cache, ps); } else { /* combined depth/stencil */ sp->sbuf_cache = sp->zbuf_cache; - sp_tile_cache_set_surface(sp, sp->sbuf_cache, ps); + sp_tile_cache_set_surface(sp->sbuf_cache, ps); } } diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 08cd39cc555..92e43d63bc9 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -38,9 +38,7 @@ #include "sp_surface.h" #include "sp_tile_cache.h" -#define CLEAR_OPTIMIZATION 0 - -#define NUM_ENTRIES 20 +#define NUM_ENTRIES 30 /** XXX move these */ @@ -69,7 +67,7 @@ struct softpipe_tile_cache -static uint +static INLINE uint is_clear_flag_set(const uint *bitvec, int x, int y) { int pos, bit; @@ -82,7 +80,7 @@ is_clear_flag_set(const uint *bitvec, int x, int y) } -static void +static INLINE void clear_clear_flag(uint *bitvec, int x, int y) { int pos; @@ -123,24 +121,22 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc) void -sp_tile_cache_set_surface(struct softpipe_context *sp, - struct softpipe_tile_cache *tc, +sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *ps) { if (tc->surface && tc->surface->map) pipe_surface_unmap(tc->surface); pipe_surface_reference(&tc->surface, ps); + + if (!ps->map) + pipe_surface_map(ps); } struct pipe_surface * -sp_tile_cache_get_surface(struct softpipe_context *sp, - struct softpipe_tile_cache *tc) +sp_tile_cache_get_surface(struct softpipe_tile_cache *tc) { - if (tc->surface && !tc->surface->map) - pipe_surface_map(tc->surface); - return tc->surface; } @@ -161,46 +157,200 @@ sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, } +static INLINE boolean +is_depth_stencil_surface(struct pipe_surface *ps) +{ + return (ps && + (ps->format == PIPE_FORMAT_S8_Z24 || + ps->format == PIPE_FORMAT_U_Z16 || + ps->format == PIPE_FORMAT_U_Z32 || + ps->format == PIPE_FORMAT_U_S8)); +} + + +/** + * Set pixels in a tile to the given clear color/value. + */ +static void +clear_tile(struct softpipe_cached_tile *tile, uint format, + const float clear_value[4]) +{ + uint i, j; + + switch (format) { + case PIPE_FORMAT_U_Z16: + { + ushort clear_val = (ushort) (clear_value[0] * 0xffff); + for (i = 0; i < TILE_SIZE; i++) { + for (j = 0; j < TILE_SIZE; j++) { + tile->data.depth16[i][j] = clear_val; + } + } + } + break; + case PIPE_FORMAT_U_Z32: + { + uint clear_val = (uint) (clear_value[0] * 0xffffffff); + for (i = 0; i < TILE_SIZE; i++) { + for (j = 0; j < TILE_SIZE; j++) { + tile->data.depth32[i][j] = clear_val; + } + } + } + break; + case PIPE_FORMAT_S8_Z24: + { + uint clear_val = (uint) (clear_value[0] * 0xffffff); + clear_val |= ((uint) clear_value[1]) << 24; + for (i = 0; i < TILE_SIZE; i++) { + for (j = 0; j < TILE_SIZE; j++) { + tile->data.depth32[i][j] = clear_val; + } + } + } + break; + case PIPE_FORMAT_U_S8: + { + ubyte clear_val = (uint) clear_value[0]; + for (i = 0; i < TILE_SIZE; i++) { + for (j = 0; j < TILE_SIZE; j++) { + tile->data.stencil8[i][j] = clear_val; + } + } + } + break; + default: + /* color */ + if (clear_value[0] == 0.0 && + clear_value[1] == 0.0 && + clear_value[2] == 0.0 && + clear_value[3] == 0.0) { + memset(tile->data.color, 0, sizeof(tile->data.color)); + } + else { + for (i = 0; i < TILE_SIZE; i++) { + for (j = 0; j < TILE_SIZE; j++) { + tile->data.color[i][j][0] = clear_value[0]; + tile->data.color[i][j][1] = clear_value[1]; + tile->data.color[i][j][2] = clear_value[2]; + tile->data.color[i][j][3] = clear_value[3]; + } + } + } + } +} + + +static INLINE void +put_tile(struct pipe_context *pipe, + struct pipe_surface *ps, + struct softpipe_cached_tile *tile) +{ + assert(tile->x >= 0); + assert(tile->x % TILE_SIZE == 0); + assert(tile->y >= 0); + assert(tile->y % TILE_SIZE == 0); + + if (is_depth_stencil_surface(ps)) { + pipe->put_tile(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); + } + else { + pipe->put_tile_rgba(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + } +} + + +static INLINE void +get_tile(struct pipe_context *pipe, + struct pipe_surface *ps, + struct softpipe_cached_tile *tile) +{ + assert(tile->x >= 0); + assert(tile->x % TILE_SIZE == 0); + assert(tile->y >= 0); + assert(tile->y % TILE_SIZE == 0); + + if (is_depth_stencil_surface(ps)) { + pipe->get_tile(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); + } + else { + pipe->get_tile_rgba(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + } +} + + void sp_flush_tile_cache(struct softpipe_context *softpipe, struct softpipe_tile_cache *tc) { struct pipe_context *pipe = &softpipe->pipe; struct pipe_surface *ps = tc->surface; - boolean is_depth_stencil; int inuse = 0, pos; if (!ps || !ps->buffer) return; - is_depth_stencil = (ps->format == PIPE_FORMAT_S8_Z24 || - ps->format == PIPE_FORMAT_Z24_S8 || - ps->format == PIPE_FORMAT_U_Z16 || - ps->format == PIPE_FORMAT_U_Z32 || - ps->format == PIPE_FORMAT_U_S8); - for (pos = 0; pos < NUM_ENTRIES; pos++) { struct softpipe_cached_tile *tile = tc->entries + pos; if (tile->x >= 0) { - if (is_depth_stencil) { - pipe->put_tile(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); - } - else { - pipe->put_tile_rgba(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - } - + put_tile(pipe, ps, tile); tile->x = tile->y = -1; /* mark as empty */ inuse++; } } - /* +#if TILE_CLEAR_OPTIMIZATION + sp_tile_cache_flush_clear(&softpipe->pipe, tc); +#endif +#if 0 printf("flushed tiles in use: %d\n", inuse); - */ +#endif +} + + +/** + * Actually clear the tiles which were flagged as being cleared. + */ +void +sp_tile_cache_flush_clear(struct pipe_context *pipe, + struct softpipe_tile_cache *tc) +{ + struct pipe_surface *ps = tc->surface; + const uint w = tc->surface->width; + const uint h = tc->surface->height; + uint x, y; + struct softpipe_cached_tile tile; + uint numCleared = 0; + + /* clear one tile to the clear value */ + clear_tile(&tile, ps->format, tc->clear_value); + + /* push the tile to all positions marked as clear */ + for (y = 0; y < h; y += TILE_SIZE) { + for (x = 0; x < w; x += TILE_SIZE) { + if (is_clear_flag_set(tc->clear_flags, x, y)) { + tile.x = x; + tile.y = y; + put_tile(pipe, ps, &tile); + + /* do this? */ + clear_clear_flag(tc->clear_flags, x, y); + + numCleared++; + } + } + } +#if 0 + printf("num cleared: %u\n", numCleared); +#endif } @@ -210,12 +360,6 @@ sp_get_cached_tile(struct softpipe_context *softpipe, { struct pipe_context *pipe = &softpipe->pipe; struct pipe_surface *ps = tc->surface; - boolean is_depth_stencil - = (ps->format == PIPE_FORMAT_S8_Z24 || - ps->format == PIPE_FORMAT_Z24_S8 || - ps->format == PIPE_FORMAT_U_Z16 || - ps->format == PIPE_FORMAT_U_Z32 || - ps->format == PIPE_FORMAT_U_S8); /* tile pos in framebuffer: */ const int tile_x = x & ~(TILE_SIZE - 1); @@ -230,104 +374,21 @@ sp_get_cached_tile(struct softpipe_context *softpipe, if (tile->x != -1) { /* put dirty tile back in framebuffer */ - if (is_depth_stencil) { - pipe->put_tile(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0 /*STRIDE*/); - } - else { - pipe->put_tile_rgba(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - } + put_tile(pipe, ps, tile); } + tile->x = tile_x; + tile->y = tile_y; + if (is_clear_flag_set(tc->clear_flags, x, y)) { /* don't get tile from framebuffer, just clear it */ - uint i, j; - /* XXX these loops could be optimized */ - switch (ps->format) { - case PIPE_FORMAT_U_Z16: - { - ushort clear_val = (ushort) (tc->clear_value[0] * 0xffff); - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth16[i][j] = clear_val; - } - } - } - break; - case PIPE_FORMAT_U_Z32: - { - uint clear_val = (uint) (tc->clear_value[0] * 0xffffffff); - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth32[i][j] = clear_val; - } - } - } - break; - case PIPE_FORMAT_S8_Z24: - { - uint clear_val = (uint) (tc->clear_value[0] * 0xffffff); - clear_val |= ((uint) tc->clear_value[1]) << 24; - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth32[i][j] = clear_val; - } - } - } - break; - case PIPE_FORMAT_Z24_S8: - { - uint clear_val = ((uint) (tc->clear_value[0] * 0xffffff)) << 8; - clear_val |= ((uint) tc->clear_value[1]) & 0xff; - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth32[i][j] = clear_val; - } - } - } - break; - case PIPE_FORMAT_U_S8: - { - ubyte clear_val = (uint) tc->clear_value[0]; - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.stencil8[i][j] = clear_val; - } - } - } - break; - default: - /* color */ - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.color[i][j][0] = tc->clear_value[0]; - tile->data.color[i][j][1] = tc->clear_value[1]; - tile->data.color[i][j][2] = tc->clear_value[2]; - tile->data.color[i][j][3] = tc->clear_value[3]; - } - } - } + clear_tile(tile, ps->format, tc->clear_value); clear_clear_flag(tc->clear_flags, x, y); } else { - /* get new tile from framebuffer */ - if (is_depth_stencil) { - pipe->get_tile(pipe, ps, - tile_x, tile_y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); - } - else { - pipe->get_tile_rgba(pipe, ps, - tile_x, tile_y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - } + /* get new tile data from surface */ + get_tile(pipe, ps, tile); } - - tile->x = tile_x; - tile->y = tile_y; } return tile; @@ -375,20 +436,13 @@ sp_get_cached_tile_tex(struct pipe_context *pipe, struct pipe_surface *ps = pipe->get_tex_surface(pipe, tc->texture, face, level, z); - if (ps != tc->surface) { - if (tc->surface && tc->surface->map) - pipe_surface_unmap(tc->surface); - - pipe_surface_reference(&tc->surface, ps); - - if (!tc->surface->map) - pipe_surface_map(tc->surface); - } + pipe_surface_map(ps); pipe->get_tile_rgba(pipe, ps, tile_x, tile_y, TILE_SIZE, TILE_SIZE, (float *) tile->data.color); + pipe_surface_unmap(ps); pipe_surface_reference(&ps, NULL); tile->x = tile_x; @@ -415,7 +469,7 @@ sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float value[4]) tc->clear_value[2] = value[2]; tc->clear_value[3] = value[3]; -#if CLEAR_OPTIMIZATION +#if TILE_CLEAR_OPTIMIZATION memset(tc->clear_flags, 255, sizeof(tc->clear_flags)); #else memset(tc->clear_flags, 0, sizeof(tc->clear_flags)); diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.h b/src/mesa/pipe/softpipe/sp_tile_cache.h index de5ff2c498b..b2a0e011e9c 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.h +++ b/src/mesa/pipe/softpipe/sp_tile_cache.h @@ -28,6 +28,8 @@ #ifndef SP_TILE_CACHE_H #define SP_TILE_CACHE_H +#define TILE_CLEAR_OPTIMIZATION 0 + #include "pipe/p_compiler.h" @@ -63,13 +65,11 @@ extern void sp_destroy_tile_cache(struct softpipe_tile_cache *tc); extern void -sp_tile_cache_set_surface(struct softpipe_context *sp, - struct softpipe_tile_cache *tc, +sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *sps); extern struct pipe_surface * -sp_tile_cache_get_surface(struct softpipe_context *sp, - struct softpipe_tile_cache *tc); +sp_tile_cache_get_surface(struct softpipe_tile_cache *tc); extern void sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, @@ -79,6 +79,10 @@ extern void sp_flush_tile_cache(struct softpipe_context *softpipe, struct softpipe_tile_cache *tc); +extern void +sp_tile_cache_flush_clear(struct pipe_context *pipe, + struct softpipe_tile_cache *tc); + extern void sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float value[4]); -- cgit v1.2.3 From 7655486c888f51dfb8ad8a7b7eab1d2021ee46dd Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 15:44:58 -0700 Subject: Convert the clearValue back to floats for sp_tile_cache_clear() It might be nice if sp->clear() took the color/value as floats... --- src/mesa/pipe/softpipe/sp_clear.c | 36 ++++++++++++++++++++++++++++++++---- 1 file changed, 32 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index a4276362b99..2b24ea63ab7 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -57,10 +57,38 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, } else if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[0])) { float clear[4]; - clear[0] = 0.2f; /* XXX hack */ - clear[1] = 0.2f; /* XXX hack */ - clear[2] = 0.2f; /* XXX hack */ - clear[3] = 0.2f; /* XXX hack */ + /* XXX it sure would be nice if the clear color was passed to + * this function as float[4].... + */ + uint r, g, b, a; + switch (ps->format) { + case PIPE_FORMAT_U_R8_G8_B8_A8: + r = (clearValue >> 24) & 0xff; + g = (clearValue >> 16) & 0xff; + g = (clearValue >> 8) & 0xff; + a = (clearValue ) & 0xff; + break; + case PIPE_FORMAT_U_A8_R8_G8_B8: + r = (clearValue >> 16) & 0xff; + g = (clearValue >> 8) & 0xff; + b = (clearValue ) & 0xff; + a = (clearValue >> 24) & 0xff; + break; + case PIPE_FORMAT_U_B8_G8_R8_A8: + r = (clearValue >> 8) & 0xff; + g = (clearValue >> 16) & 0xff; + b = (clearValue >> 24) & 0xff; + a = (clearValue ) & 0xff; + break; + default: + assert(0); + } + + clear[0] = r / 255.0; + clear[1] = g / 255.0; + clear[2] = b / 255.0; + clear[3] = a / 255.0; + sp_tile_cache_clear(softpipe->cbuf_cache[0], clear); } -- cgit v1.2.3 From 7bbc3ef7ec3c616ebbc05a92c4ef39d8e3e5a008 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 15:46:10 -0700 Subject: Make sp_tile_cache_flush_clear() static. Also, misc clean-ups, comments. --- src/mesa/pipe/softpipe/sp_tile_cache.c | 80 ++++++++++++++++++++-------------- src/mesa/pipe/softpipe/sp_tile_cache.h | 4 -- 2 files changed, 48 insertions(+), 36 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 92e43d63bc9..93930b18f28 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -124,8 +124,10 @@ void sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *ps) { - if (tc->surface && tc->surface->map) + if (tc->surface && tc->surface->map) { + assert(tc->surface != ps); pipe_surface_unmap(tc->surface); + } pipe_surface_reference(&tc->surface, ps); @@ -287,39 +289,10 @@ get_tile(struct pipe_context *pipe, } -void -sp_flush_tile_cache(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc) -{ - struct pipe_context *pipe = &softpipe->pipe; - struct pipe_surface *ps = tc->surface; - int inuse = 0, pos; - - if (!ps || !ps->buffer) - return; - - for (pos = 0; pos < NUM_ENTRIES; pos++) { - struct softpipe_cached_tile *tile = tc->entries + pos; - if (tile->x >= 0) { - put_tile(pipe, ps, tile); - tile->x = tile->y = -1; /* mark as empty */ - inuse++; - } - } - -#if TILE_CLEAR_OPTIMIZATION - sp_tile_cache_flush_clear(&softpipe->pipe, tc); -#endif -#if 0 - printf("flushed tiles in use: %d\n", inuse); -#endif -} - - /** - * Actually clear the tiles which were flagged as being cleared. + * Actually clear the tiles which were flagged as being in a clear state. */ -void +static void sp_tile_cache_flush_clear(struct pipe_context *pipe, struct softpipe_tile_cache *tc) { @@ -354,6 +327,47 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, } +/** + * Flush the tile cache: write all dirty tiles back to the surface. + * any tiles "flagged" as cleared will be "really" cleared. + */ +void +sp_flush_tile_cache(struct softpipe_context *softpipe, + struct softpipe_tile_cache *tc) +{ + struct pipe_context *pipe = &softpipe->pipe; + struct pipe_surface *ps = tc->surface; + int inuse = 0, pos; + + if (!ps || !ps->buffer) + return; + + if (!ps->map) + pipe_surface_map(ps); + + for (pos = 0; pos < NUM_ENTRIES; pos++) { + struct softpipe_cached_tile *tile = tc->entries + pos; + if (tile->x >= 0) { + put_tile(pipe, ps, tile); + tile->x = tile->y = -1; /* mark as empty */ + inuse++; + } + } + +#if TILE_CLEAR_OPTIMIZATION + sp_tile_cache_flush_clear(&softpipe->pipe, tc); +#endif + +#if 0 + printf("flushed tiles in use: %d\n", inuse); +#endif +} + + +/** + * Get a tile from the cache. + * \param x, y position of tile, in pixels + */ struct softpipe_cached_tile * sp_get_cached_tile(struct softpipe_context *softpipe, struct softpipe_tile_cache *tc, int x, int y) @@ -470,8 +484,10 @@ sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float value[4]) tc->clear_value[3] = value[3]; #if TILE_CLEAR_OPTIMIZATION + /* set flags to indicate all the tiles are cleared */ memset(tc->clear_flags, 255, sizeof(tc->clear_flags)); #else + /* disable the optimization */ memset(tc->clear_flags, 0, sizeof(tc->clear_flags)); #endif } diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.h b/src/mesa/pipe/softpipe/sp_tile_cache.h index b2a0e011e9c..12d5815db28 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.h +++ b/src/mesa/pipe/softpipe/sp_tile_cache.h @@ -79,10 +79,6 @@ extern void sp_flush_tile_cache(struct softpipe_context *softpipe, struct softpipe_tile_cache *tc); -extern void -sp_tile_cache_flush_clear(struct pipe_context *pipe, - struct softpipe_tile_cache *tc); - extern void sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float value[4]); -- cgit v1.2.3 From 382ced1600888c0de5fce06c55f2235d68dfece7 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 15:47:39 -0700 Subject: Enable the tile clear "optimization" Not really faster yet because we're using the float-valued put/get_tile() functions so we're doing quite a bit of float/int conversion. That can be fixed though... --- src/mesa/pipe/softpipe/sp_tile_cache.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.h b/src/mesa/pipe/softpipe/sp_tile_cache.h index 12d5815db28..c4f2e0f1840 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.h +++ b/src/mesa/pipe/softpipe/sp_tile_cache.h @@ -28,7 +28,7 @@ #ifndef SP_TILE_CACHE_H #define SP_TILE_CACHE_H -#define TILE_CLEAR_OPTIMIZATION 0 +#define TILE_CLEAR_OPTIMIZATION 1 #include "pipe/p_compiler.h" -- cgit v1.2.3 From 2edf1b8e7172f7025895f4378cdea76a7c8254a9 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 16:01:31 -0700 Subject: Get rid of "duplicate" formats. For example, replace PIPE_FORMAT_U_A8_R8_G8_B8 with PIPE_FORMAT_A8R8G8B8_UNORM --- .../dri/intel_winsys/intel_winsys_softpipe.c | 6 +- src/mesa/pipe/draw/draw_vertex_fetch.c | 2 +- src/mesa/pipe/i915simple/i915_context.c | 24 ++++---- src/mesa/pipe/i915simple/i915_state_emit.c | 10 ++-- src/mesa/pipe/i915simple/i915_state_sampler.c | 12 ++-- src/mesa/pipe/i915simple/i915_surface.c | 4 +- src/mesa/pipe/p_format.h | 12 ---- src/mesa/pipe/softpipe/sp_clear.c | 6 +- src/mesa/pipe/softpipe/sp_quad_depth_test.c | 16 ++--- src/mesa/pipe/softpipe/sp_quad_stencil.c | 8 +-- src/mesa/pipe/softpipe/sp_surface.c | 70 +++++++++++----------- src/mesa/pipe/softpipe/sp_tile_cache.c | 12 ++-- src/mesa/pipe/xlib/xm_api.c | 8 +-- src/mesa/pipe/xlib/xm_winsys.c | 8 +-- 14 files changed, 93 insertions(+), 105 deletions(-) diff --git a/src/mesa/drivers/dri/intel_winsys/intel_winsys_softpipe.c b/src/mesa/drivers/dri/intel_winsys/intel_winsys_softpipe.c index 7a93546bc22..f9ed6253f1d 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_winsys_softpipe.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_winsys_softpipe.c @@ -48,9 +48,9 @@ static boolean intel_is_format_supported(struct softpipe_winsys *sws, uint format) { switch(format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_U_R5_G6_B5: - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_S8Z24_UNORM: return TRUE; default: return FALSE; diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index 5510b3674eb..77df5c6427e 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -79,7 +79,7 @@ fetch_attrib4(const void *ptr, unsigned format, float attrib[4]) attrib[0] = (float) ((int *) ptr)[0]; break; - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: attrib[0] = (float) ((unsigned char *) ptr)[2] / 255.0f; attrib[1] = (float) ((unsigned char *) ptr)[1] / 255.0f; attrib[2] = (float) ((unsigned char *) ptr)[0] / 255.0f; diff --git a/src/mesa/pipe/i915simple/i915_context.c b/src/mesa/pipe/i915simple/i915_context.c index 07b331c5288..cfce116920d 100644 --- a/src/mesa/pipe/i915simple/i915_context.c +++ b/src/mesa/pipe/i915simple/i915_context.c @@ -50,32 +50,32 @@ i915_is_format_supported( struct pipe_context *pipe, #if 0 /* XXX: This is broken -- rewrite if still needed. */ static const unsigned tex_supported[] = { - PIPE_FORMAT_U_R8_G8_B8_A8, - PIPE_FORMAT_U_A8_R8_G8_B8, - PIPE_FORMAT_U_R5_G6_B5, + PIPE_FORMAT_R8G8B8A8_UNORM, + PIPE_FORMAT_A8R8G8B8_UNORM, + PIPE_FORMAT_R5G6B5_UNORM, PIPE_FORMAT_U_L8, PIPE_FORMAT_U_A8, PIPE_FORMAT_U_I8, PIPE_FORMAT_U_L8_A8, PIPE_FORMAT_YCBCR, PIPE_FORMAT_YCBCR_REV, - PIPE_FORMAT_S8_Z24, + PIPE_FORMAT_S8Z24_UNORM, }; /* Actually a lot more than this - add later: */ static const unsigned render_supported[] = { - PIPE_FORMAT_U_A8_R8_G8_B8, - PIPE_FORMAT_U_R5_G6_B5, + PIPE_FORMAT_A8R8G8B8_UNORM, + PIPE_FORMAT_R5G6B5_UNORM, }; /* */ static const unsigned z_stencil_supported[] = { - PIPE_FORMAT_U_Z16, - PIPE_FORMAT_U_Z32, - PIPE_FORMAT_S8_Z24, + PIPE_FORMAT_Z16_UNORM, + PIPE_FORMAT_Z32_UNORM, + PIPE_FORMAT_S8Z24_UNORM, }; switch (type) { @@ -97,9 +97,9 @@ i915_is_format_supported( struct pipe_context *pipe, } #else switch( format ) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_U_R5_G6_B5: - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_S8Z24_UNORM: return TRUE; }; return FALSE; diff --git a/src/mesa/pipe/i915simple/i915_state_emit.c b/src/mesa/pipe/i915simple/i915_state_emit.c index eda40d7c76c..1e0db91024d 100644 --- a/src/mesa/pipe/i915simple/i915_state_emit.c +++ b/src/mesa/pipe/i915simple/i915_state_emit.c @@ -38,9 +38,9 @@ static unsigned translate_format( unsigned format ) { switch (format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: return COLOR_BUF_ARGB8888; - case PIPE_FORMAT_U_R5_G6_B5: + case PIPE_FORMAT_R5G6B5_UNORM: return COLOR_BUF_RGB565; default: assert(0); @@ -51,9 +51,9 @@ static unsigned translate_format( unsigned format ) static unsigned translate_depth_format( unsigned zformat ) { switch (zformat) { - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: return DEPTH_FRMT_24_FIXED_8_OTHER; - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: return DEPTH_FRMT_16_FIXED; default: assert(0); @@ -247,7 +247,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) if (cbuf_surface) cformat = cbuf_surface->format; else - cformat = PIPE_FORMAT_U_A8_R8_G8_B8; /* arbitrary */ + cformat = PIPE_FORMAT_A8R8G8B8_UNORM; /* arbitrary */ cformat = translate_format(cformat); if (depth_surface) diff --git a/src/mesa/pipe/i915simple/i915_state_sampler.c b/src/mesa/pipe/i915simple/i915_state_sampler.c index 1816d9abdb7..dd2d51d9f60 100644 --- a/src/mesa/pipe/i915simple/i915_state_sampler.c +++ b/src/mesa/pipe/i915simple/i915_state_sampler.c @@ -138,13 +138,13 @@ translate_texture_format(uint pipeFormat) return MAPSURF_8BIT | MT_8BIT_A8; case PIPE_FORMAT_U_A8_L8: return MAPSURF_16BIT | MT_16BIT_AY88; - case PIPE_FORMAT_U_R5_G6_B5: + case PIPE_FORMAT_R5G6B5_UNORM: return MAPSURF_16BIT | MT_16BIT_RGB565; - case PIPE_FORMAT_U_A1_R5_G5_B5: + case PIPE_FORMAT_A1R5G5B5_UNORM: return MAPSURF_16BIT | MT_16BIT_ARGB1555; - case PIPE_FORMAT_U_A4_R4_G4_B4: + case PIPE_FORMAT_A4R4G4B4_UNORM: return MAPSURF_16BIT | MT_16BIT_ARGB4444; - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: return MAPSURF_32BIT | MT_32BIT_ARGB8888; case PIPE_FORMAT_YCBCR_REV: return (MAPSURF_422 | MT_422_YCRCB_NORMAL); @@ -155,7 +155,7 @@ translate_texture_format(uint pipeFormat) case PIPE_FORMAT_RGBA_FXT1: return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1); #endif - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: return (MAPSURF_16BIT | MT_16BIT_L16); #if 0 case PIPE_FORMAT_RGBA_DXT1: @@ -166,7 +166,7 @@ translate_texture_format(uint pipeFormat) case PIPE_FORMAT_RGBA_DXT5: return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5); #endif - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: return (MAPSURF_32BIT | MT_32BIT_xL824); default: fprintf(stderr, "i915: translate_texture_format() bad image format %x\n", diff --git a/src/mesa/pipe/i915simple/i915_surface.c b/src/mesa/pipe/i915simple/i915_surface.c index ecbabe50eb5..ea48c1bd8ae 100644 --- a/src/mesa/pipe/i915simple/i915_surface.c +++ b/src/mesa/pipe/i915simple/i915_surface.c @@ -66,7 +66,7 @@ i915_get_tile_rgba(struct pipe_context *pipe, CLIP_TILE; switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: for (i = 0; i < h; i++) { float *pRow = p; for (j = 0; j < w; j++) { @@ -81,7 +81,7 @@ i915_get_tile_rgba(struct pipe_context *pipe, p += w0 * 4; } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: { const float scale = 1.0f / (float) 0xffffff; for (i = 0; i < h; i++) { diff --git a/src/mesa/pipe/p_format.h b/src/mesa/pipe/p_format.h index de0eb9539f8..86728f77a23 100644 --- a/src/mesa/pipe/p_format.h +++ b/src/mesa/pipe/p_format.h @@ -276,18 +276,6 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f) /** * Duplicated formats: */ -#define PIPE_FORMAT_U_A8_R8_G8_B8 PIPE_FORMAT_A8R8G8B8_UNORM -#define PIPE_FORMAT_U_B8_G8_R8_A8 PIPE_FORMAT_B8G8R8A8_UNORM -#define PIPE_FORMAT_U_A1_R5_G5_B5 PIPE_FORMAT_A1R5G5B5_UNORM -#define PIPE_FORMAT_U_A4_R4_G4_B4 PIPE_FORMAT_A4R4G4B4_UNORM -#define PIPE_FORMAT_U_R8_G8_B8_A8 PIPE_FORMAT_R8G8B8A8_UNORM -#define PIPE_FORMAT_U_R5_G6_B5 PIPE_FORMAT_R5G6B5_UNORM -#define PIPE_FORMAT_S_R16_G16_B16_A16 PIPE_FORMAT_R16G16B16A16_SNORM -#define PIPE_FORMAT_U_Z16 PIPE_FORMAT_Z16_UNORM -#define PIPE_FORMAT_U_Z32 PIPE_FORMAT_Z32_UNORM -#define PIPE_FORMAT_F_Z32 PIPE_FORMAT_Z32_FLOAT -#define PIPE_FORMAT_S8_Z24 PIPE_FORMAT_S8Z24_UNORM -#define PIPE_FORMAT_Z24_S8 PIPE_FORMAT_Z24S8_UNORM #define PIPE_FORMAT_U_S8 PIPE_FORMAT_S8_UNORM /** diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index 2b24ea63ab7..a6352e06161 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -62,19 +62,19 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, */ uint r, g, b, a; switch (ps->format) { - case PIPE_FORMAT_U_R8_G8_B8_A8: + case PIPE_FORMAT_R8G8B8A8_UNORM: r = (clearValue >> 24) & 0xff; g = (clearValue >> 16) & 0xff; g = (clearValue >> 8) & 0xff; a = (clearValue ) & 0xff; break; - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: r = (clearValue >> 16) & 0xff; g = (clearValue >> 8) & 0xff; b = (clearValue ) & 0xff; a = (clearValue >> 24) & 0xff; break; - case PIPE_FORMAT_U_B8_G8_R8_A8: + case PIPE_FORMAT_B8G8R8A8_UNORM: r = (clearValue >> 8) & 0xff; g = (clearValue >> 16) & 0xff; b = (clearValue >> 24) & 0xff; diff --git a/src/mesa/pipe/softpipe/sp_quad_depth_test.c b/src/mesa/pipe/softpipe/sp_quad_depth_test.c index 33181896218..00128fa5284 100644 --- a/src/mesa/pipe/softpipe/sp_quad_depth_test.c +++ b/src/mesa/pipe/softpipe/sp_quad_depth_test.c @@ -74,7 +74,7 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) * Also, get the zbuffer values (bzzzz) from the cached tile. */ switch (format) { - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: { float scale = 65535.0; @@ -89,7 +89,7 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) } } break; - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z32_UNORM: { double scale = (double) (uint) ~0UL; @@ -104,7 +104,7 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) } } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: { float scale = (float) ((1 << 24) - 1); @@ -119,7 +119,7 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) } } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: { float scale = (float) ((1 << 24) - 1); @@ -202,21 +202,21 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) /* put updated Z values back into cached tile */ switch (format) { - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); tile->data.depth16[y][x] = (ushort) bzzzz[j]; } break; - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z32_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); tile->data.depth32[y][x] = bzzzz[j]; } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); @@ -225,7 +225,7 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) tile->data.depth32[y][x] = s8z24; } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); diff --git a/src/mesa/pipe/softpipe/sp_quad_stencil.c b/src/mesa/pipe/softpipe/sp_quad_stencil.c index 0149b20f48f..b8c199204df 100644 --- a/src/mesa/pipe/softpipe/sp_quad_stencil.c +++ b/src/mesa/pipe/softpipe/sp_quad_stencil.c @@ -234,14 +234,14 @@ stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) /* get stencil values from cached tile */ switch (ps->format) { - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); stencilVals[j] = tile->data.depth32[y][x] >> 24; } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); @@ -298,7 +298,7 @@ stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) /* put new stencil values into cached tile */ switch (ps->format) { - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); @@ -307,7 +307,7 @@ stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) tile->data.depth32[y][x] = s8z24; } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: for (j = 0; j < QUAD_SIZE; j++) { int x = quad->x0 % TILE_SIZE + (j & 1); int y = quad->y0 % TILE_SIZE + (j >> 1); diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index 16f0209eee4..cdbffe54000 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -70,7 +70,7 @@ #endif -/*** PIPE_FORMAT_U_A8_R8_G8_B8 ***/ +/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/ static void a8r8g8b8_get_tile(struct pipe_surface *ps, @@ -82,7 +82,7 @@ a8r8g8b8_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_U_A8_R8_G8_B8); + assert(ps->format == PIPE_FORMAT_A8R8G8B8_UNORM); CLIP_TILE; @@ -113,7 +113,7 @@ a8r8g8b8_put_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_U_A8_R8_G8_B8); + assert(ps->format == PIPE_FORMAT_A8R8G8B8_UNORM); CLIP_TILE; @@ -134,7 +134,7 @@ a8r8g8b8_put_tile(struct pipe_surface *ps, } -/*** PIPE_FORMAT_U_B8_G8_R8_A8 ***/ +/*** PIPE_FORMAT_B8G8R8A8_UNORM ***/ static void b8g8r8a8_get_tile(struct pipe_surface *ps, @@ -146,7 +146,7 @@ b8g8r8a8_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_U_B8_G8_R8_A8); + assert(ps->format == PIPE_FORMAT_B8G8R8A8_UNORM); CLIP_TILE; @@ -177,7 +177,7 @@ b8g8r8a8_put_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_U_B8_G8_R8_A8); + assert(ps->format == PIPE_FORMAT_B8G8R8A8_UNORM); CLIP_TILE; @@ -198,7 +198,7 @@ b8g8r8a8_put_tile(struct pipe_surface *ps, } -/*** PIPE_FORMAT_U_A1_R5_G5_B5 ***/ +/*** PIPE_FORMAT_A1R5G5B5_UNORM ***/ static void a1r5g5b5_get_tile(struct pipe_surface *ps, @@ -209,7 +209,7 @@ a1r5g5b5_get_tile(struct pipe_surface *ps, + y * ps->pitch + x; unsigned i, j; - assert(ps->format == PIPE_FORMAT_U_A1_R5_G5_B5); + assert(ps->format == PIPE_FORMAT_A1R5G5B5_UNORM); for (i = 0; i < h; i++) { for (j = 0; j < w; j++) { @@ -226,7 +226,7 @@ a1r5g5b5_get_tile(struct pipe_surface *ps, -/*** PIPE_FORMAT_U_Z16 ***/ +/*** PIPE_FORMAT_Z16_UNORM ***/ /** * Return each Z value as four floats in [0,1]. @@ -242,7 +242,7 @@ z16_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_U_Z16); + assert(ps->format == PIPE_FORMAT_Z16_UNORM); CLIP_TILE; @@ -324,7 +324,7 @@ a8_get_tile(struct pipe_surface *ps, } -/*** PIPE_FORMAT_S_R16_G16_B16_A16 ***/ +/*** PIPE_FORMAT_R16G16B16A16_SNORM ***/ static void r16g16b16a16_get_tile(struct pipe_surface *ps, @@ -336,7 +336,7 @@ r16g16b16a16_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_S_R16_G16_B16_A16); + assert(ps->format == PIPE_FORMAT_R16G16B16A16_SNORM); CLIP_TILE; @@ -368,7 +368,7 @@ r16g16b16a16_put_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_S_R16_G16_B16_A16); + assert(ps->format == PIPE_FORMAT_R16G16B16A16_SNORM); CLIP_TILE; @@ -458,7 +458,7 @@ a8_l8_get_tile(struct pipe_surface *ps, -/*** PIPE_FORMAT_U_Z32 ***/ +/*** PIPE_FORMAT_Z32_UNORM ***/ /** * Return each Z value as four floats in [0,1]. @@ -474,7 +474,7 @@ z32_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_U_Z16); + assert(ps->format == PIPE_FORMAT_Z16_UNORM); CLIP_TILE; @@ -492,7 +492,7 @@ z32_get_tile(struct pipe_surface *ps, } -/*** PIPE_FORMAT_S8_Z24 ***/ +/*** PIPE_FORMAT_S8Z24_UNORM ***/ /** * Return Z component as four float in [0,1]. Stencil part ignored. @@ -508,7 +508,7 @@ s8z24_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_S8_Z24); + assert(ps->format == PIPE_FORMAT_S8Z24_UNORM); CLIP_TILE; @@ -526,7 +526,7 @@ s8z24_get_tile(struct pipe_surface *ps, } -/*** PIPE_FORMAT_Z24_S8 ***/ +/*** PIPE_FORMAT_Z24S8_UNORM ***/ /** * Return Z component as four float in [0,1]. Stencil part ignored. @@ -542,7 +542,7 @@ z24s8_get_tile(struct pipe_surface *ps, unsigned i, j; unsigned w0 = w; - assert(ps->format == PIPE_FORMAT_Z24_S8); + assert(ps->format == PIPE_FORMAT_Z24S8_UNORM); CLIP_TILE; @@ -673,13 +673,13 @@ softpipe_get_tile_rgba(struct pipe_context *pipe, float *p) { switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: a8r8g8b8_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_B8_G8_R8_A8: + case PIPE_FORMAT_B8G8R8A8_UNORM: b8g8r8a8_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_A1_R5_G5_B5: + case PIPE_FORMAT_A1R5G5B5_UNORM: a1r5g5b5_get_tile(ps, x, y, w, h, p); break; case PIPE_FORMAT_U_L8: @@ -694,19 +694,19 @@ softpipe_get_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_U_A8_L8: a8_l8_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_S_R16_G16_B16_A16: + case PIPE_FORMAT_R16G16B16A16_SNORM: r16g16b16a16_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: z16_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z32_UNORM: z32_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: s8z24_get_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: z24s8_get_tile(ps, x, y, w, h, p); break; default: @@ -723,13 +723,13 @@ softpipe_put_tile_rgba(struct pipe_context *pipe, const float *p) { switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: a8r8g8b8_put_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_B8_G8_R8_A8: + case PIPE_FORMAT_B8G8R8A8_UNORM: b8g8r8a8_put_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_A1_R5_G5_B5: + case PIPE_FORMAT_A1R5G5B5_UNORM: /*a1r5g5b5_put_tile(ps, x, y, w, h, p);*/ break; case PIPE_FORMAT_U_L8: @@ -744,19 +744,19 @@ softpipe_put_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_U_A8_L8: /*a8_l8_put_tile(ps, x, y, w, h, p);*/ break; - case PIPE_FORMAT_S_R16_G16_B16_A16: + case PIPE_FORMAT_R16G16B16A16_SNORM: r16g16b16a16_put_tile(ps, x, y, w, h, p); break; - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: /*z16_put_tile(ps, x, y, w, h, p);*/ break; - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z32_UNORM: /*z32_put_tile(ps, x, y, w, h, p);*/ break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: /*s8z24_put_tile(ps, x, y, w, h, p);*/ break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: /*z24s8_put_tile(ps, x, y, w, h, p);*/ break; default: diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 93930b18f28..be5dd5c2890 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -163,9 +163,9 @@ static INLINE boolean is_depth_stencil_surface(struct pipe_surface *ps) { return (ps && - (ps->format == PIPE_FORMAT_S8_Z24 || - ps->format == PIPE_FORMAT_U_Z16 || - ps->format == PIPE_FORMAT_U_Z32 || + (ps->format == PIPE_FORMAT_S8Z24_UNORM || + ps->format == PIPE_FORMAT_Z16_UNORM || + ps->format == PIPE_FORMAT_Z32_UNORM || ps->format == PIPE_FORMAT_U_S8)); } @@ -180,7 +180,7 @@ clear_tile(struct softpipe_cached_tile *tile, uint format, uint i, j; switch (format) { - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: { ushort clear_val = (ushort) (clear_value[0] * 0xffff); for (i = 0; i < TILE_SIZE; i++) { @@ -190,7 +190,7 @@ clear_tile(struct softpipe_cached_tile *tile, uint format, } } break; - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z32_UNORM: { uint clear_val = (uint) (clear_value[0] * 0xffffffff); for (i = 0; i < TILE_SIZE; i++) { @@ -200,7 +200,7 @@ clear_tile(struct softpipe_cached_tile *tile, uint format, } } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: { uint clear_val = (uint) (clear_value[0] * 0xffffff); clear_val |= ((uint) clear_value[1]) << 24; diff --git a/src/mesa/pipe/xlib/xm_api.c b/src/mesa/pipe/xlib/xm_api.c index 5cb9ca0d363..9c7713790b8 100644 --- a/src/mesa/pipe/xlib/xm_api.c +++ b/src/mesa/pipe/xlib/xm_api.c @@ -406,7 +406,7 @@ choose_pixel_format(XMesaVisual v) return 0 /* PIXEL_FORMAT_U_A8_B8_G8_R8 */; } else { - return PIPE_FORMAT_U_R8_G8_B8_A8; + return PIPE_FORMAT_R8G8B8A8_UNORM; } } else if ( GET_REDMASK(v) == 0xff0000 @@ -415,10 +415,10 @@ choose_pixel_format(XMesaVisual v) && v->BitsPerPixel == 32) { if (CHECK_BYTE_ORDER(v)) { /* no byteswapping needed */ - return PIPE_FORMAT_U_A8_R8_G8_B8; + return PIPE_FORMAT_A8R8G8B8_UNORM; } else { - return PIPE_FORMAT_U_B8_G8_R8_A8; + return PIPE_FORMAT_B8G8R8A8_UNORM; } } else if ( GET_REDMASK(v) == 0xf800 @@ -427,7 +427,7 @@ choose_pixel_format(XMesaVisual v) && CHECK_BYTE_ORDER(v) && v->BitsPerPixel == 16) { /* 5-6-5 RGB */ - return PIPE_FORMAT_U_R5_G6_B5; + return PIPE_FORMAT_R5G6B5_UNORM; } assert(0); diff --git a/src/mesa/pipe/xlib/xm_winsys.c b/src/mesa/pipe/xlib/xm_winsys.c index e8cb36eced4..408797c688c 100644 --- a/src/mesa/pipe/xlib/xm_winsys.c +++ b/src/mesa/pipe/xlib/xm_winsys.c @@ -389,11 +389,11 @@ xmesa_is_format_supported(struct softpipe_winsys *sws, uint format) else { /* non-color / window surface format */ switch (format) { - case PIPE_FORMAT_S_R16_G16_B16_A16: - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_R16G16B16A16_SNORM: + case PIPE_FORMAT_S8Z24_UNORM: case PIPE_FORMAT_U_S8: - case PIPE_FORMAT_U_Z16: - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z32_UNORM: return TRUE; default: return FALSE; -- cgit v1.2.3 From cfe0ebe1c0871c9dd5d51f30219cfe686b467ff8 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 16:15:14 -0700 Subject: Replace "duplicate" formats --- src/mesa/state_tracker/st_cb_clear.c | 24 ++++----- src/mesa/state_tracker/st_cb_drawpixels.c | 22 ++++---- src/mesa/state_tracker/st_cb_readpixels.c | 4 +- src/mesa/state_tracker/st_format.c | 88 +++++++++++++++---------------- 4 files changed, 69 insertions(+), 69 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 8ac6699776b..1f0d1e64260 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -65,13 +65,13 @@ color_value(GLuint pipeFormat, const GLfloat color[4]) UNCLAMPED_FLOAT_TO_UBYTE(a, color[3]); switch (pipeFormat) { - case PIPE_FORMAT_U_R8_G8_B8_A8: + case PIPE_FORMAT_R8G8B8A8_UNORM: return (r << 24) | (g << 16) | (b << 8) | a; - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: return (a << 24) | (r << 16) | (g << 8) | b; - case PIPE_FORMAT_U_B8_G8_R8_A8: + case PIPE_FORMAT_B8G8R8A8_UNORM: return (b << 24) | (g << 16) | (r << 8) | a; - case PIPE_FORMAT_U_R5_G6_B5: + case PIPE_FORMAT_R5G6B5_UNORM: return ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3); default: assert(0); @@ -84,17 +84,17 @@ static uint depth_value(GLuint pipeFormat, GLfloat value) { switch (pipeFormat) { - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: return (uint) (value * 0xffff); - case PIPE_FORMAT_U_Z32: + case PIPE_FORMAT_Z32_UNORM: /* special-case to avoid overflow */ if (value == 1.0) return 0xffffffff; else return (uint) (value * 0xffffffff); - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: return (uint) (value * 0xffffff); - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: return ((uint) (value * 0xffffff)) << 8; default: assert(0); @@ -107,8 +107,8 @@ static GLboolean is_depth_stencil_format(GLuint pipeFormat) { switch (pipeFormat) { - case PIPE_FORMAT_S8_Z24: - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: return GL_TRUE; default: return GL_FALSE; @@ -523,10 +523,10 @@ clear_depth_stencil_buffer(GLcontext *ctx, struct gl_renderbuffer *rb) GLuint clearValue = depth_value(strb->surface->format, ctx->Depth.Clear); switch (strb->surface->format) { - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: clearValue |= ctx->Stencil.Clear << 24; break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: clearValue |= clearValue | ctx->Stencil.Clear; break; default: diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index a61daffb20a..464cf5ced34 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -734,42 +734,42 @@ compatible_formats(GLenum format, GLenum type, GLuint pipeFormat) static const GLuint one = 1; GLubyte littleEndian = *((GLubyte *) &one); - if (pipeFormat == PIPE_FORMAT_U_R8_G8_B8_A8 && + if (pipeFormat == PIPE_FORMAT_R8G8B8A8_UNORM && format == GL_RGBA && type == GL_UNSIGNED_BYTE && !littleEndian) { return GL_TRUE; } - else if (pipeFormat == PIPE_FORMAT_U_R8_G8_B8_A8 && + else if (pipeFormat == PIPE_FORMAT_R8G8B8A8_UNORM && format == GL_ABGR_EXT && type == GL_UNSIGNED_BYTE && littleEndian) { return GL_TRUE; } - else if (pipeFormat == PIPE_FORMAT_U_A8_R8_G8_B8 && + else if (pipeFormat == PIPE_FORMAT_A8R8G8B8_UNORM && format == GL_BGRA && type == GL_UNSIGNED_BYTE && littleEndian) { return GL_TRUE; } - else if (pipeFormat == PIPE_FORMAT_U_R5_G6_B5 && + else if (pipeFormat == PIPE_FORMAT_R5G6B5_UNORM && format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5) { /* endian don't care */ return GL_TRUE; } - else if (pipeFormat == PIPE_FORMAT_U_R5_G6_B5 && + else if (pipeFormat == PIPE_FORMAT_R5G6B5_UNORM && format == GL_BGR && type == GL_UNSIGNED_SHORT_5_6_5_REV) { /* endian don't care */ return GL_TRUE; } - else if (pipeFormat == PIPE_FORMAT_U_S8 && + else if (pipeFormat == PIPE_FORMAT_S8_UNORM && format == GL_STENCIL_INDEX && type == GL_UNSIGNED_BYTE) { return GL_TRUE; } - else if (pipeFormat == PIPE_FORMAT_U_Z32 && + else if (pipeFormat == PIPE_FORMAT_Z32_UNORM && format == GL_DEPTH_COMPONENT && type == GL_UNSIGNED_INT) { return GL_TRUE; @@ -888,7 +888,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, memcpy(dest, values, spanWidth); } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: { uint *dest = (uint *) stmap + spanY * ps->pitch + spanX; GLint k; @@ -998,8 +998,8 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, cpp = 1; comp = 0; } - else if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A8_R8_G8_B8 )) { - format = PIPE_FORMAT_U_A8_R8_G8_B8; + else if (pipe->is_format_supported( pipe, PIPE_FORMAT_A8R8G8B8_UNORM )) { + format = PIPE_FORMAT_A8R8G8B8_UNORM; internal_format = GL_RGBA8; cpp = 4; comp = 3; /* alpha channel */ /*XXX little-endian dependency */ @@ -1167,7 +1167,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, src = buffer + i * width; switch (psDraw->format) { - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: { uint *dst4 = (uint *) dst; int j; diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index e2243e7de49..96829fcfa0f 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -90,7 +90,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, memcpy(values, src, width); } break; - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_S8Z24_UNORM: { const uint *src = (uint *) stmap + srcY * ps->pitch + x; GLint k; @@ -99,7 +99,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, } } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: { const uint *src = (uint *) stmap + srcY * ps->pitch + x; GLint k; diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index f93f7f48d94..9e249a47a60 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -120,7 +120,7 @@ st_get_format_info( #endif /* Data type */ - if (format == PIPE_FORMAT_U_A1_R5_G5_B5 || format == PIPE_FORMAT_U_R5_G6_B5) { + if (format == PIPE_FORMAT_A1R5G5B5_UNORM || format == PIPE_FORMAT_R5G6B5_UNORM) { pinfo->datatype = GL_UNSIGNED_SHORT; } else { @@ -255,9 +255,9 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat) /* fix this */ case MESA_FORMAT_ARGB8888_REV: case MESA_FORMAT_ARGB8888: - return PIPE_FORMAT_U_A8_R8_G8_B8; + return PIPE_FORMAT_A8R8G8B8_UNORM; case MESA_FORMAT_ARGB4444: - return PIPE_FORMAT_U_A4_R4_G4_B4; + return PIPE_FORMAT_A4R4G4B4_UNORM; case MESA_FORMAT_AL88: return PIPE_FORMAT_U_A8_L8; case MESA_FORMAT_A8: @@ -267,7 +267,7 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat) case MESA_FORMAT_I8: return PIPE_FORMAT_U_I8; case MESA_FORMAT_Z16: - return PIPE_FORMAT_U_Z16; + return PIPE_FORMAT_Z16_UNORM; default: assert(0); return 0; @@ -282,10 +282,10 @@ default_rgba_format( struct pipe_context *pipe ) { static const uint colorFormats[] = { - PIPE_FORMAT_U_R8_G8_B8_A8, - PIPE_FORMAT_U_A8_R8_G8_B8, - PIPE_FORMAT_U_B8_G8_R8_A8, - PIPE_FORMAT_U_R5_G6_B5 + PIPE_FORMAT_R8G8B8A8_UNORM, + PIPE_FORMAT_A8R8G8B8_UNORM, + PIPE_FORMAT_B8G8R8A8_UNORM, + PIPE_FORMAT_R5G6B5_UNORM }; uint i; for (i = 0; i < Elements(colorFormats); i++) { @@ -304,8 +304,8 @@ static GLuint default_deep_rgba_format( struct pipe_context *pipe ) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_S_R16_G16_B16_A16 )) { - return PIPE_FORMAT_S_R16_G16_B16_A16; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_R16G16B16A16_SNORM )) { + return PIPE_FORMAT_R16G16B16A16_SNORM; } return PIPE_FORMAT_NONE; } @@ -319,10 +319,10 @@ default_depth_format( struct pipe_context *pipe ) { static const uint zFormats[] = { - PIPE_FORMAT_U_Z16, - PIPE_FORMAT_U_Z32, - PIPE_FORMAT_S8_Z24, - PIPE_FORMAT_Z24_S8 + PIPE_FORMAT_Z16_UNORM, + PIPE_FORMAT_Z32_UNORM, + PIPE_FORMAT_S8Z24_UNORM, + PIPE_FORMAT_Z24S8_UNORM }; uint i; for (i = 0; i < Elements(zFormats); i++) { @@ -358,16 +358,16 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, case GL_COMPRESSED_RGBA: if (format == GL_BGRA) { if (type == GL_UNSIGNED_BYTE || type == GL_UNSIGNED_INT_8_8_8_8_REV) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A8_R8_G8_B8 )) - return PIPE_FORMAT_U_A8_R8_G8_B8; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_A8R8G8B8_UNORM )) + return PIPE_FORMAT_A8R8G8B8_UNORM; } else if (type == GL_UNSIGNED_SHORT_4_4_4_4_REV) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A4_R4_G4_B4 )) - return PIPE_FORMAT_U_A4_R4_G4_B4; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_A4R4G4B4_UNORM )) + return PIPE_FORMAT_A4R4G4B4_UNORM; } else if (type == GL_UNSIGNED_SHORT_1_5_5_5_REV) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A1_R5_G5_B5 )) - return PIPE_FORMAT_U_A1_R5_G5_B5; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_A1R5G5B5_UNORM )) + return PIPE_FORMAT_A1R5G5B5_UNORM; } } return default_rgba_format( pipe ); @@ -376,8 +376,8 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, case GL_RGB: case GL_COMPRESSED_RGB: if (format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5) { - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_R5_G6_B5 )) - return PIPE_FORMAT_U_R5_G6_B5; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_R5G6B5_UNORM )) + return PIPE_FORMAT_R5G6B5_UNORM; } return default_rgba_format( pipe ); @@ -390,13 +390,13 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, case GL_RGBA4: case GL_RGBA2: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A4_R4_G4_B4 )) - return PIPE_FORMAT_U_A4_R4_G4_B4; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_A4R4G4B4_UNORM )) + return PIPE_FORMAT_A4R4G4B4_UNORM; return default_rgba_format( pipe ); case GL_RGB5_A1: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A1_R5_G5_B5 )) - return PIPE_FORMAT_U_A1_R5_G5_B5; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_A1R5G5B5_UNORM )) + return PIPE_FORMAT_A1R5G5B5_UNORM; return default_rgba_format( pipe ); case GL_RGB8: @@ -408,8 +408,8 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, case GL_RGB5: case GL_RGB4: case GL_R3_G3_B2: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_A1_R5_G5_B5 )) - return PIPE_FORMAT_U_A1_R5_G5_B5; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_A1R5G5B5_UNORM )) + return PIPE_FORMAT_A1R5G5B5_UNORM; return default_rgba_format( pipe ); case GL_ALPHA: @@ -491,18 +491,18 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, #endif case GL_DEPTH_COMPONENT16: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_Z16 )) - return PIPE_FORMAT_U_Z16; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z16_UNORM )) + return PIPE_FORMAT_Z16_UNORM; /* fall-through */ case GL_DEPTH_COMPONENT24: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) - return PIPE_FORMAT_S8_Z24; - if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) - return PIPE_FORMAT_Z24_S8; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8Z24_UNORM )) + return PIPE_FORMAT_S8Z24_UNORM; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24S8_UNORM )) + return PIPE_FORMAT_Z24S8_UNORM; /* fall-through */ case GL_DEPTH_COMPONENT32: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_Z32 )) - return PIPE_FORMAT_U_Z32; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z32_UNORM )) + return PIPE_FORMAT_Z32_UNORM; /* fall-through */ case GL_DEPTH_COMPONENT: return default_depth_format( pipe ); @@ -514,18 +514,18 @@ st_choose_pipe_format(struct pipe_context *pipe, GLint internalFormat, case GL_STENCIL_INDEX16_EXT: if (pipe->is_format_supported( pipe, PIPE_FORMAT_U_S8 )) return PIPE_FORMAT_U_S8; - if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) - return PIPE_FORMAT_S8_Z24; - if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) - return PIPE_FORMAT_Z24_S8; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8Z24_UNORM )) + return PIPE_FORMAT_S8Z24_UNORM; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24S8_UNORM )) + return PIPE_FORMAT_Z24S8_UNORM; return PIPE_FORMAT_NONE; case GL_DEPTH_STENCIL_EXT: case GL_DEPTH24_STENCIL8_EXT: - if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8_Z24 )) - return PIPE_FORMAT_S8_Z24; - if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24_S8 )) - return PIPE_FORMAT_Z24_S8; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_S8Z24_UNORM )) + return PIPE_FORMAT_S8Z24_UNORM; + if (pipe->is_format_supported( pipe, PIPE_FORMAT_Z24S8_UNORM )) + return PIPE_FORMAT_Z24S8_UNORM; return PIPE_FORMAT_NONE; default: -- cgit v1.2.3 From 144b70b0b583892ffa5c96b009ee91b3ccb202a6 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 16:22:39 -0700 Subject: include tgsi_dump.h --- src/mesa/state_tracker/st_program.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index 4e8bdc820e9..e64bf14d56a 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -38,6 +38,7 @@ #include "pipe/p_defines.h" #include "pipe/p_shader_tokens.h" #include "pipe/draw/draw_context.h" +#include "pipe/tgsi/util/tgsi_dump.h" #include "st_context.h" #include "st_cache.h" -- cgit v1.2.3 From 61d1972dd038ba94924a57891408e19d9e5da205 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 16:46:30 -0700 Subject: Define PIPE_FORMAT_ tokens as an enum set, rather than #defines. This makes debugging a _lot_ easier. In gdb, "print format" used to display 613570600, now you see PIPE_FORMAT_A8R8G8B8_UNORM. --- .../drivers/dri/intel_winsys/intel_winsys_pipe.c | 2 +- src/mesa/pipe/draw/draw_vertex_fetch.c | 2 +- src/mesa/pipe/draw/draw_vertex_shader_llvm.c | 2 +- src/mesa/pipe/i915simple/i915_context.c | 5 +- src/mesa/pipe/i915simple/i915_state_emit.c | 4 +- src/mesa/pipe/i915simple/i915_state_sampler.c | 2 +- src/mesa/pipe/p_context.h | 2 +- src/mesa/pipe/p_format.h | 154 +++++++++++---------- src/mesa/pipe/p_state.h | 7 +- src/mesa/pipe/p_winsys.h | 4 +- src/mesa/pipe/softpipe/sp_context.c | 3 +- src/mesa/pipe/softpipe/sp_quad_depth_test.c | 2 +- src/mesa/pipe/softpipe/sp_tile_cache.c | 3 +- src/mesa/pipe/xlib/xm_winsys.c | 5 +- src/mesa/state_tracker/st_cb_clear.c | 6 +- src/mesa/state_tracker/st_cb_drawpixels.c | 5 +- src/mesa/state_tracker/st_cb_fbo.c | 5 +- src/mesa/state_tracker/st_format.c | 14 +- src/mesa/state_tracker/st_format.h | 18 ++- 19 files changed, 127 insertions(+), 118 deletions(-) diff --git a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c index 1b71d0ac107..1799e9b901b 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c @@ -193,7 +193,7 @@ intel_i915_surface_pitch(struct pipe_winsys *winsys, static struct pipe_surface * -intel_i915_surface_alloc(struct pipe_winsys *winsys, unsigned format) +intel_i915_surface_alloc(struct pipe_winsys *winsys, enum pipe_format format) { struct pipe_surface *surf = CALLOC_STRUCT(pipe_surface); if (surf) { diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index 77df5c6427e..e0759c2e9a9 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -46,7 +46,7 @@ * XXX this might be a temporary thing. */ static void -fetch_attrib4(const void *ptr, unsigned format, float attrib[4]) +fetch_attrib4(const void *ptr, enum pipe_format format, float attrib[4]) { /* defaults */ attrib[1] = 0.0; diff --git a/src/mesa/pipe/draw/draw_vertex_shader_llvm.c b/src/mesa/pipe/draw/draw_vertex_shader_llvm.c index 53a1776ffcb..29de437595c 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader_llvm.c +++ b/src/mesa/pipe/draw/draw_vertex_shader_llvm.c @@ -43,7 +43,7 @@ #define DBG 0 static INLINE void -fetch_attrib4(const void *ptr, unsigned format, float attrib[4]) +fetch_attrib4(const void *ptr, enum pipe_format format, float attrib[4]) { /* defaults */ attrib[1] = 0.0; diff --git a/src/mesa/pipe/i915simple/i915_context.c b/src/mesa/pipe/i915simple/i915_context.c index cfce116920d..b915a677908 100644 --- a/src/mesa/pipe/i915simple/i915_context.c +++ b/src/mesa/pipe/i915simple/i915_context.c @@ -45,7 +45,7 @@ */ static boolean i915_is_format_supported( struct pipe_context *pipe, - uint format ) + enum pipe_format format ) { #if 0 /* XXX: This is broken -- rewrite if still needed. */ @@ -101,8 +101,9 @@ i915_is_format_supported( struct pipe_context *pipe, case PIPE_FORMAT_R5G6B5_UNORM: case PIPE_FORMAT_S8Z24_UNORM: return TRUE; + default: + return FALSE; }; - return FALSE; #endif } diff --git a/src/mesa/pipe/i915simple/i915_state_emit.c b/src/mesa/pipe/i915simple/i915_state_emit.c index 1e0db91024d..09bf1fa2d68 100644 --- a/src/mesa/pipe/i915simple/i915_state_emit.c +++ b/src/mesa/pipe/i915simple/i915_state_emit.c @@ -35,7 +35,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" -static unsigned translate_format( unsigned format ) +static unsigned translate_format( enum pipe_format format ) { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: @@ -48,7 +48,7 @@ static unsigned translate_format( unsigned format ) } } -static unsigned translate_depth_format( unsigned zformat ) +static unsigned translate_depth_format( enum pipe_format zformat ) { switch (zformat) { case PIPE_FORMAT_S8Z24_UNORM: diff --git a/src/mesa/pipe/i915simple/i915_state_sampler.c b/src/mesa/pipe/i915simple/i915_state_sampler.c index dd2d51d9f60..59408b6ba06 100644 --- a/src/mesa/pipe/i915simple/i915_state_sampler.c +++ b/src/mesa/pipe/i915simple/i915_state_sampler.c @@ -127,7 +127,7 @@ void i915_update_samplers( struct i915_context *i915 ) static uint -translate_texture_format(uint pipeFormat) +translate_texture_format(enum pipe_format pipeFormat) { switch (pipeFormat) { case PIPE_FORMAT_U_L8: diff --git a/src/mesa/pipe/p_context.h b/src/mesa/pipe/p_context.h index 2420d02213f..b3a2122ade9 100644 --- a/src/mesa/pipe/p_context.h +++ b/src/mesa/pipe/p_context.h @@ -51,7 +51,7 @@ struct pipe_context { * Queries */ boolean (*is_format_supported)( struct pipe_context *pipe, - uint format ); + enum pipe_format format ); const char *(*get_name)( struct pipe_context *pipe ); diff --git a/src/mesa/pipe/p_format.h b/src/mesa/pipe/p_format.h index 86728f77a23..8f11bfab76b 100644 --- a/src/mesa/pipe/p_format.h +++ b/src/mesa/pipe/p_format.h @@ -198,86 +198,90 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f) * z24s8, compressed textures, ycbcr, etc that won't fit that model. */ -#define PIPE_FORMAT_NONE _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_0000, 0, 0, 0, 0, PIPE_FORMAT_TYPE_UNKNOWN ) -#define PIPE_FORMAT_A8R8G8B8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ARGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_B8G8R8A8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGRA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_A1R5G5B5_UNORM _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 1, 5, 5, 5, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_A4R4G4B4_UNORM _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R5G6B5_UNORM _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_RGB0, 5, 6, 5, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_U_L8 _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) /**< ubyte luminance */ -#define PIPE_FORMAT_U_A8 _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_000R, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) /**< ubyte alpha */ -#define PIPE_FORMAT_U_I8 _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRR, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) /**< ubyte intensity */ -#define PIPE_FORMAT_U_A8_L8 _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) /**< ubyte alpha, luminance */ -#define PIPE_FORMAT_YCBCR _PIPE_FORMAT_YCBCR( 0 ) -#define PIPE_FORMAT_YCBCR_REV _PIPE_FORMAT_YCBCR( 1 ) -#define PIPE_FORMAT_Z16_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_Z32_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_Z32_FLOAT _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_S8Z24_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_SZ00, 1, 3, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_Z24S8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ZS00, 3, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_S8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_S000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R64_FLOAT _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R64G64_FLOAT _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R64G64B64_FLOAT _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R64G64B64A64_FLOAT _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R32_FLOAT _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R32G32_FLOAT _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R32G32B32_FLOAT _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R32G32B32A32_FLOAT _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_FLOAT ) -#define PIPE_FORMAT_R32_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R32G32_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R32G32B32_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R32G32B32A32_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R32_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R32G32_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R32G32B32_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R32G32B32A32_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R32_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R32G32_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R32G32B32_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R32G32B32A32_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R32_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R32G32_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R32G32B32_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R32G32B32A32_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R16_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R16G16_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R16G16B16_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R16G16B16A16_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R16_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R16G16_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R16G16B16_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R16G16B16A16_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R16_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R16G16_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R16G16B16_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R16G16B16A16_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R16_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R16G16_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R16G16B16_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R16G16B16A16_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R8G8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R8G8B8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R8G8B8A8_UNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ) -#define PIPE_FORMAT_R8_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R8G8_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R8G8B8_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R8G8B8A8_USCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_USCALED ) -#define PIPE_FORMAT_R8_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R8G8_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R8G8B8_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R8G8B8A8_SNORM _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SNORM ) -#define PIPE_FORMAT_R8_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R8G8_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R8G8B8_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SSCALED ) -#define PIPE_FORMAT_R8G8B8A8_SSCALED _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SSCALED ) +enum pipe_format { + PIPE_FORMAT_NONE = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_0000, 0, 0, 0, 0, PIPE_FORMAT_TYPE_UNKNOWN ), + PIPE_FORMAT_A8R8G8B8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ARGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_B8G8R8A8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGRA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_A1R5G5B5_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 1, 5, 5, 5, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_A4R4G4B4_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_ARGB, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R5G6B5_UNORM = _PIPE_FORMAT_RGBAZS_1 ( _PIPE_FORMAT_RGB0, 5, 6, 5, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_U_L8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte luminance */ + PIPE_FORMAT_U_A8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_000R, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte alpha */ + PIPE_FORMAT_U_I8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRR, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte intensity */ + PIPE_FORMAT_U_A8_L8 = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte alpha, luminance */ + PIPE_FORMAT_YCBCR = _PIPE_FORMAT_YCBCR( 0 ), + PIPE_FORMAT_YCBCR_REV = _PIPE_FORMAT_YCBCR( 1 ), + PIPE_FORMAT_Z16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_Z32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_Z32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_Z000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_S8Z24_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_SZ00, 1, 3, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_Z24S8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ZS00, 3, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_S8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_S000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), /**< ubyte stencil */ + PIPE_FORMAT_R64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R64G64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R64G64B64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R64G64B64A64_FLOAT = _PIPE_FORMAT_RGBAZS_64( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32G32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32G32B32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32G32B32A32_FLOAT = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_FLOAT ), + PIPE_FORMAT_R32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32G32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32G32B32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32G32B32A32_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32G32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32G32B32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32G32B32A32_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32G32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32G32B32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32G32B32A32_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 4, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R32G32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 4, 4, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R32G32B32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 4, 4, 4, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R32G32B32A32_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16G16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16G16B16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16G16B16A16_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16G16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16G16B16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16G16B16A16_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16G16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16G16B16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16G16B16A16_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 2, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16G16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 2, 2, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16G16B16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 2, 2, 2, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R16G16B16A16_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 2, 2, 2, 2, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8G8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8G8B8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8G8B8A8_UNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_UNORM ), + PIPE_FORMAT_R8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8G8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8G8B8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8G8B8A8_USCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_USCALED ), + PIPE_FORMAT_R8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8G8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8G8B8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8G8B8A8_SNORM = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SNORM ), + PIPE_FORMAT_R8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_R000, 1, 0, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RG00, 1, 1, 0, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8B8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB0, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SSCALED ), + PIPE_FORMAT_R8G8B8A8_SSCALED = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SSCALED ) +}; + /** - * Duplicated formats: + * XXX should remove this, but S8_UNORM is a poor name */ #define PIPE_FORMAT_U_S8 PIPE_FORMAT_S8_UNORM + /** * Builds pipe format name from format token. */ diff --git a/src/mesa/pipe/p_state.h b/src/mesa/pipe/p_state.h index 44dec9b7732..6db9bbc9534 100644 --- a/src/mesa/pipe/p_state.h +++ b/src/mesa/pipe/p_state.h @@ -39,6 +39,7 @@ #define PIPE_STATE_H #include "p_compiler.h" +#include "p_format.h" /** * Implementation limits @@ -265,7 +266,7 @@ struct pipe_surface struct pipe_buffer_handle *buffer; /**< driver private buffer handle */ ubyte *map; /**< only non-NULL when surface is actually mapped */ unsigned map_refcount; /**< Reference count for mapping */ - unsigned format; /**< PIPE_FORMAT_x */ + enum pipe_format format; /**< PIPE_FORMAT_x */ unsigned cpp; /**< bytes per pixel */ unsigned width, height; unsigned pitch; /**< in pixels */ @@ -286,7 +287,7 @@ struct pipe_texture unsigned target; /* XXX convert to PIPE_TEXTURE_x */ unsigned internal_format; /* XXX convert to PIPE_FORMAT_x */ - unsigned format; /**< PIPE_FORMAT_x */ + enum pipe_format format; /**< PIPE_FORMAT_x */ unsigned first_level; unsigned last_level; @@ -331,7 +332,7 @@ struct pipe_vertex_element unsigned vertex_buffer_index:5; unsigned dst_offset:8; - unsigned src_format; /**< PIPE_FORMAT_* */ + enum pipe_format src_format; /**< PIPE_FORMAT_* */ }; diff --git a/src/mesa/pipe/p_winsys.h b/src/mesa/pipe/p_winsys.h index 5adca1d6fd6..438e8bdb63b 100644 --- a/src/mesa/pipe/p_winsys.h +++ b/src/mesa/pipe/p_winsys.h @@ -29,6 +29,8 @@ #define P_WINSYS_H +#include "p_format.h" + /** * \file * This is the interface that Gallium3D requires any window system @@ -90,7 +92,7 @@ struct pipe_winsys /** allocate a new surface (no context dependency) */ struct pipe_surface *(*surface_alloc)(struct pipe_winsys *ws, - unsigned format); + enum pipe_format format); void (*surface_release)(struct pipe_winsys *ws, struct pipe_surface **s); diff --git a/src/mesa/pipe/softpipe/sp_context.c b/src/mesa/pipe/softpipe/sp_context.c index 54b2076b4a1..bdfd6228ef8 100644 --- a/src/mesa/pipe/softpipe/sp_context.c +++ b/src/mesa/pipe/softpipe/sp_context.c @@ -51,7 +51,8 @@ * parameter or another function. */ static boolean -softpipe_is_format_supported( struct pipe_context *pipe, uint format ) +softpipe_is_format_supported( struct pipe_context *pipe, + enum pipe_format format ) { struct softpipe_context *softpipe = softpipe_context( pipe ); /* ask winsys if the format is supported */ diff --git a/src/mesa/pipe/softpipe/sp_quad_depth_test.c b/src/mesa/pipe/softpipe/sp_quad_depth_test.c index 00128fa5284..93ea1196f99 100644 --- a/src/mesa/pipe/softpipe/sp_quad_depth_test.c +++ b/src/mesa/pipe/softpipe/sp_quad_depth_test.c @@ -54,7 +54,7 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) { struct softpipe_context *softpipe = qs->softpipe; struct pipe_surface *ps = softpipe->framebuffer.zbuf; - const uint format = ps->format; + const enum pipe_format format = ps->format; unsigned bzzzz[QUAD_SIZE]; /**< Z values fetched from depth buffer */ unsigned qzzzz[QUAD_SIZE]; /**< Z values from the quad */ unsigned zmask = 0; diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index be5dd5c2890..602b78ebe6b 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -174,7 +174,8 @@ is_depth_stencil_surface(struct pipe_surface *ps) * Set pixels in a tile to the given clear color/value. */ static void -clear_tile(struct softpipe_cached_tile *tile, uint format, +clear_tile(struct softpipe_cached_tile *tile, + enum pipe_format format, const float clear_value[4]) { uint i, j; diff --git a/src/mesa/pipe/xlib/xm_winsys.c b/src/mesa/pipe/xlib/xm_winsys.c index 408797c688c..976884ad520 100644 --- a/src/mesa/pipe/xlib/xm_winsys.c +++ b/src/mesa/pipe/xlib/xm_winsys.c @@ -301,7 +301,7 @@ xm_surface_pitch(struct pipe_winsys *winsys, unsigned cpp, unsigned width, * renderbuffers, etc. */ static struct pipe_surface * -xm_surface_alloc(struct pipe_winsys *ws, GLuint pipeFormat) +xm_surface_alloc(struct pipe_winsys *ws, enum pipe_format pipeFormat) { struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface); @@ -379,7 +379,8 @@ xmesa_get_pipe_winsys(void) * can support any format. */ static boolean -xmesa_is_format_supported(struct softpipe_winsys *sws, uint format) +xmesa_is_format_supported(struct softpipe_winsys *sws, + enum pipe_format format) { struct xmesa_softpipe_winsys *xmws = xmesa_softpipe_winsys(sws); diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 1f0d1e64260..b4b2429a2aa 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -55,7 +55,7 @@ static GLuint -color_value(GLuint pipeFormat, const GLfloat color[4]) +color_value(enum pipe_format pipeFormat, const GLfloat color[4]) { GLubyte r, g, b, a; @@ -81,7 +81,7 @@ color_value(GLuint pipeFormat, const GLfloat color[4]) static uint -depth_value(GLuint pipeFormat, GLfloat value) +depth_value(enum pipe_format pipeFormat, GLfloat value) { switch (pipeFormat) { case PIPE_FORMAT_Z16_UNORM: @@ -104,7 +104,7 @@ depth_value(GLuint pipeFormat, GLfloat value) static GLboolean -is_depth_stencil_format(GLuint pipeFormat) +is_depth_stencil_format(enum pipe_format pipeFormat) { switch (pipeFormat) { case PIPE_FORMAT_S8Z24_UNORM: diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 464cf5ced34..c0e41dbeec5 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -456,7 +456,8 @@ make_texture(struct st_context *st, struct pipe_context *pipe = st->pipe; const struct gl_texture_format *mformat; struct pipe_texture *pt; - GLuint pipeFormat, cpp; + enum pipe_format pipeFormat; + GLuint cpp; GLenum baseFormat; baseFormat = _mesa_base_format(format); @@ -729,7 +730,7 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, * XXX probably move this to a re-usable place. */ static GLboolean -compatible_formats(GLenum format, GLenum type, GLuint pipeFormat) +compatible_formats(GLenum format, GLenum type, enum pipe_format pipeFormat) { static const GLuint one = 1; GLubyte littleEndian = *((GLubyte *) &one); diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index 0ee5f45aae3..cbda56b5c3d 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -55,7 +55,8 @@ * Compute the renderbuffer's Red/Green/EtcBit fields from the pipe format. */ static int -init_renderbuffer_bits(struct st_renderbuffer *strb, uint pipeFormat) +init_renderbuffer_bits(struct st_renderbuffer *strb, + enum pipe_format pipeFormat) { struct pipe_format_info info; @@ -86,7 +87,7 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb, { struct pipe_context *pipe = ctx->st->pipe; struct st_renderbuffer *strb = st_renderbuffer(rb); - const uint pipeFormat + const enum pipe_format pipeFormat = st_choose_pipe_format(pipe, internalFormat, GL_NONE, GL_NONE); GLuint cpp; GLbitfield flags = PIPE_SURFACE_FLAG_RENDER; /* want to render to surface */ diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index 9e249a47a60..8d39e1bec69 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -99,9 +99,7 @@ format_size( * XXX temporary here */ GLboolean -st_get_format_info( - GLuint format, - struct pipe_format_info *pinfo ) +st_get_format_info(enum pipe_format format, struct pipe_format_info *pinfo) { if (pf_layout(format) == PIPE_FORMAT_LAYOUT_RGBAZS) { pipe_format_rgbazs_t info; @@ -222,10 +220,10 @@ st_get_format_info( * Return bytes per pixel for the given format. */ GLuint -st_sizeof_format(GLuint pipeFormat) +st_sizeof_format(enum pipe_format format) { struct pipe_format_info info; - if (!st_get_format_info( pipeFormat, &info )) { + if (!st_get_format_info( format, &info )) { assert( 0 ); return 0; } @@ -237,10 +235,10 @@ st_sizeof_format(GLuint pipeFormat) * Return bytes per pixel for the given format. */ GLenum -st_format_datatype(GLuint pipeFormat) +st_format_datatype(enum pipe_format format) { struct pipe_format_info info; - if (!st_get_format_info( pipeFormat, &info )) { + if (!st_get_format_info( format, &info )) { assert( 0 ); return 0; } @@ -248,7 +246,7 @@ st_format_datatype(GLuint pipeFormat) } -GLuint +enum pipe_format st_mesa_format_to_pipe_format(GLuint mesaFormat) { switch (mesaFormat) { diff --git a/src/mesa/state_tracker/st_format.h b/src/mesa/state_tracker/st_format.h index 23abc36edff..6ccf5536f9d 100644 --- a/src/mesa/state_tracker/st_format.h +++ b/src/mesa/state_tracker/st_format.h @@ -26,13 +26,13 @@ **************************************************************************/ -#ifndef ST_CB_TEXIMAGE_H -#define ST_CB_TEXIMAGE_H +#ifndef ST_FORMAT_H +#define ST_FORMAT_H struct pipe_format_info { - GLuint format; + enum pipe_format format; GLenum base_format; GLenum datatype; GLubyte red_bits; @@ -47,21 +47,19 @@ struct pipe_format_info }; -extern GLboolean -st_get_format_info( - GLuint format, - struct pipe_format_info *pinfo ); +GLboolean +st_get_format_info(enum pipe_format format, struct pipe_format_info *pinfo); extern GLuint -st_sizeof_format(GLuint pipeFormat); +st_sizeof_format(enum pipe_format format); extern GLenum -st_format_datatype(GLuint pipeFormat); +st_format_datatype(enum pipe_format format); -extern GLuint +extern enum pipe_format st_mesa_format_to_pipe_format(GLuint mesaFormat); -- cgit v1.2.3 From 2636ea5758d8f5c2845e9e9f9d317be35d058e1f Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 17:11:30 -0700 Subject: Move _mesa_init_glsl_driver_functions() into shader_api.c This allows making a bunch of functions static, and removes a state tracker dependency on driverfuncs.c --- src/mesa/drivers/common/driverfuncs.c | 37 -------- src/mesa/drivers/common/driverfuncs.h | 4 - src/mesa/shader/shader_api.c | 153 +++++++++++++++++++++------------- src/mesa/shader/shader_api.h | 103 +---------------------- src/mesa/state_tracker/st_context.c | 2 +- 5 files changed, 100 insertions(+), 199 deletions(-) diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c index 96e5037fa57..ea0cc51d4a9 100644 --- a/src/mesa/drivers/common/driverfuncs.c +++ b/src/mesa/drivers/common/driverfuncs.c @@ -259,43 +259,6 @@ _mesa_init_driver_functions(struct dd_function_table *driver) } -/** - * Plug in Mesa's GLSL functions. - */ -void -_mesa_init_glsl_driver_functions(struct dd_function_table *driver) -{ - driver->AttachShader = _mesa_attach_shader; - driver->BindAttribLocation = _mesa_bind_attrib_location; - driver->CompileShader = _mesa_compile_shader; - driver->CreateProgram = _mesa_create_program; - driver->CreateShader = _mesa_create_shader; - driver->DeleteProgram2 = _mesa_delete_program2; - driver->DeleteShader = _mesa_delete_shader; - driver->DetachShader = _mesa_detach_shader; - driver->GetActiveAttrib = _mesa_get_active_attrib; - driver->GetActiveUniform = _mesa_get_active_uniform; - driver->GetAttachedShaders = _mesa_get_attached_shaders; - driver->GetAttribLocation = _mesa_get_attrib_location; - driver->GetHandle = _mesa_get_handle; - driver->GetProgramiv = _mesa_get_programiv; - driver->GetProgramInfoLog = _mesa_get_program_info_log; - driver->GetShaderiv = _mesa_get_shaderiv; - driver->GetShaderInfoLog = _mesa_get_shader_info_log; - driver->GetShaderSource = _mesa_get_shader_source; - driver->GetUniformfv = _mesa_get_uniformfv; - driver->GetUniformLocation = _mesa_get_uniform_location; - driver->IsProgram = _mesa_is_program; - driver->IsShader = _mesa_is_shader; - driver->LinkProgram = _mesa_link_program; - driver->ShaderSource = _mesa_shader_source; - driver->Uniform = _mesa_uniform; - driver->UniformMatrix = _mesa_uniform_matrix; - driver->UseProgram = _mesa_use_program; - driver->ValidateProgram = _mesa_validate_program; -} - - /** * Call the ctx->Driver.* state functions with current values to initialize * driver state. diff --git a/src/mesa/drivers/common/driverfuncs.h b/src/mesa/drivers/common/driverfuncs.h index 6ed23c4520e..4c90ed12f60 100644 --- a/src/mesa/drivers/common/driverfuncs.h +++ b/src/mesa/drivers/common/driverfuncs.h @@ -30,10 +30,6 @@ extern void _mesa_init_driver_functions(struct dd_function_table *driver); -extern void -_mesa_init_glsl_driver_functions(struct dd_function_table *driver); - - extern void _mesa_init_driver_state(GLcontext *ctx); diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c index bc5ecdaa156..4470a87af0a 100644 --- a/src/mesa/shader/shader_api.c +++ b/src/mesa/shader/shader_api.c @@ -52,7 +52,7 @@ /** * Allocate a new gl_shader_program object, initialize it. */ -struct gl_shader_program * +static struct gl_shader_program * _mesa_new_shader_program(GLcontext *ctx, GLuint name) { struct gl_shader_program *shProg; @@ -371,7 +371,7 @@ copy_string(GLchar *dst, GLsizei maxLength, GLsizei *length, const GLchar *src) /** * Called via ctx->Driver.AttachShader() */ -void +static void _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader) { struct gl_shader_program *shProg @@ -410,7 +410,38 @@ _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader) } -void +static GLint +_mesa_get_attrib_location(GLcontext *ctx, GLuint program, + const GLchar *name) +{ + struct gl_shader_program *shProg + = _mesa_lookup_shader_program(ctx, program); + + if (!shProg) { + _mesa_error(ctx, GL_INVALID_VALUE, "glGetAttribLocation"); + return -1; + } + + if (!shProg->LinkStatus) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glGetAttribLocation(program not linked)"); + return -1; + } + + if (!name) + return -1; + + if (shProg->Attributes) { + GLint i = _mesa_lookup_parameter_index(shProg->Attributes, -1, name); + if (i >= 0) { + return shProg->Attributes->Parameters[i].StateIndexes[0]; + } + } + return -1; +} + + +static void _mesa_bind_attrib_location(GLcontext *ctx, GLuint program, GLuint index, const GLchar *name) { @@ -456,7 +487,7 @@ _mesa_bind_attrib_location(GLcontext *ctx, GLuint program, GLuint index, } -GLuint +static GLuint _mesa_create_shader(GLcontext *ctx, GLenum type) { struct gl_shader *sh; @@ -480,7 +511,7 @@ _mesa_create_shader(GLcontext *ctx, GLenum type) } -GLuint +static GLuint _mesa_create_program(GLcontext *ctx) { GLuint name; @@ -501,7 +532,7 @@ _mesa_create_program(GLcontext *ctx) * Named w/ "2" to indicate OpenGL 2.x vs GL_ARB_fragment_programs's * DeleteProgramARB. */ -void +static void _mesa_delete_program2(GLcontext *ctx, GLuint name) { /* @@ -527,7 +558,7 @@ _mesa_delete_program2(GLcontext *ctx, GLuint name) } -void +static void _mesa_delete_shader(GLcontext *ctx, GLuint shader) { struct gl_shader *sh = _mesa_lookup_shader(ctx, shader); @@ -542,7 +573,7 @@ _mesa_delete_shader(GLcontext *ctx, GLuint shader) } -void +static void _mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader) { struct gl_shader_program *shProg @@ -602,7 +633,7 @@ _mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader) } -void +static void _mesa_get_active_attrib(GLcontext *ctx, GLuint program, GLuint index, GLsizei maxLength, GLsizei *length, GLint *size, GLenum *type, GLchar *nameOut) @@ -637,7 +668,7 @@ _mesa_get_active_attrib(GLcontext *ctx, GLuint program, GLuint index, /** * Called via ctx->Driver.GetActiveUniform(). */ -void +static void _mesa_get_active_uniform(GLcontext *ctx, GLuint program, GLuint index, GLsizei maxLength, GLsizei *length, GLint *size, GLenum *type, GLchar *nameOut) @@ -681,7 +712,7 @@ _mesa_get_active_uniform(GLcontext *ctx, GLuint program, GLuint index, /** * Called via ctx->Driver.GetAttachedShaders(). */ -void +static void _mesa_get_attached_shaders(GLcontext *ctx, GLuint program, GLsizei maxCount, GLsizei *count, GLuint *obj) { @@ -701,38 +732,7 @@ _mesa_get_attached_shaders(GLcontext *ctx, GLuint program, GLsizei maxCount, } -GLint -_mesa_get_attrib_location(GLcontext *ctx, GLuint program, - const GLchar *name) -{ - struct gl_shader_program *shProg - = _mesa_lookup_shader_program(ctx, program); - - if (!shProg) { - _mesa_error(ctx, GL_INVALID_VALUE, "glGetAttribLocation"); - return -1; - } - - if (!shProg->LinkStatus) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "glGetAttribLocation(program not linked)"); - return -1; - } - - if (!name) - return -1; - - if (shProg->Attributes) { - GLint i = _mesa_lookup_parameter_index(shProg->Attributes, -1, name); - if (i >= 0) { - return shProg->Attributes->Parameters[i].StateIndexes[0]; - } - } - return -1; -} - - -GLuint +static GLuint _mesa_get_handle(GLcontext *ctx, GLenum pname) { #if 0 @@ -756,7 +756,7 @@ _mesa_get_handle(GLcontext *ctx, GLenum pname) } -void +static void _mesa_get_programiv(GLcontext *ctx, GLuint program, GLenum pname, GLint *params) { @@ -810,7 +810,7 @@ _mesa_get_programiv(GLcontext *ctx, GLuint program, } -void +static void _mesa_get_shaderiv(GLcontext *ctx, GLuint name, GLenum pname, GLint *params) { struct gl_shader *shader = _mesa_lookup_shader(ctx, name); @@ -843,7 +843,7 @@ _mesa_get_shaderiv(GLcontext *ctx, GLuint name, GLenum pname, GLint *params) } -void +static void _mesa_get_program_info_log(GLcontext *ctx, GLuint program, GLsizei bufSize, GLsizei *length, GLchar *infoLog) { @@ -857,7 +857,7 @@ _mesa_get_program_info_log(GLcontext *ctx, GLuint program, GLsizei bufSize, } -void +static void _mesa_get_shader_info_log(GLcontext *ctx, GLuint shader, GLsizei bufSize, GLsizei *length, GLchar *infoLog) { @@ -873,7 +873,7 @@ _mesa_get_shader_info_log(GLcontext *ctx, GLuint shader, GLsizei bufSize, /** * Called via ctx->Driver.GetShaderSource(). */ -void +static void _mesa_get_shader_source(GLcontext *ctx, GLuint shader, GLsizei maxLength, GLsizei *length, GLchar *sourceOut) { @@ -889,7 +889,7 @@ _mesa_get_shader_source(GLcontext *ctx, GLuint shader, GLsizei maxLength, /** * Called via ctx->Driver.GetUniformfv(). */ -void +static void _mesa_get_uniformfv(GLcontext *ctx, GLuint program, GLint location, GLfloat *params) { @@ -915,7 +915,7 @@ _mesa_get_uniformfv(GLcontext *ctx, GLuint program, GLint location, /** * Called via ctx->Driver.GetUniformLocation(). */ -GLint +static GLint _mesa_get_uniform_location(GLcontext *ctx, GLuint program, const GLchar *name) { struct gl_shader_program *shProg @@ -940,7 +940,7 @@ _mesa_get_uniform_location(GLcontext *ctx, GLuint program, const GLchar *name) } -GLboolean +static GLboolean _mesa_is_program(GLcontext *ctx, GLuint name) { struct gl_shader_program *shProg = _mesa_lookup_shader_program(ctx, name); @@ -948,7 +948,7 @@ _mesa_is_program(GLcontext *ctx, GLuint name) } -GLboolean +static GLboolean _mesa_is_shader(GLcontext *ctx, GLuint name) { struct gl_shader *shader = _mesa_lookup_shader(ctx, name); @@ -960,7 +960,7 @@ _mesa_is_shader(GLcontext *ctx, GLuint name) /** * Called via ctx->Driver.ShaderSource() */ -void +static void _mesa_shader_source(GLcontext *ctx, GLuint shader, const GLchar *source) { struct gl_shader *sh = _mesa_lookup_shader(ctx, shader); @@ -981,7 +981,7 @@ _mesa_shader_source(GLcontext *ctx, GLuint shader, const GLchar *source) /** * Called via ctx->Driver.CompileShader() */ -void +static void _mesa_compile_shader(GLcontext *ctx, GLuint shaderObj) { struct gl_shader *sh = _mesa_lookup_shader(ctx, shaderObj); @@ -998,7 +998,7 @@ _mesa_compile_shader(GLcontext *ctx, GLuint shaderObj) /** * Called via ctx->Driver.LinkProgram() */ -void +static void _mesa_link_program(GLcontext *ctx, GLuint program) { struct gl_shader_program *shProg; @@ -1070,7 +1070,7 @@ update_textures_used(struct gl_program *prog) /** * Called via ctx->Driver.Uniform(). */ -void +static void _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count, const GLvoid *values, GLenum type) { @@ -1182,7 +1182,7 @@ _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count, /** * Called by ctx->Driver.UniformMatrix(). */ -void +static void _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows, GLenum matrixType, GLint location, GLsizei count, GLboolean transpose, const GLfloat *values) @@ -1230,7 +1230,7 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows, } -void +static void _mesa_validate_program(GLcontext *ctx, GLuint program) { struct gl_shader_program *shProg; @@ -1256,3 +1256,40 @@ _mesa_validate_program(GLcontext *ctx, GLuint program) image units allowed. */ } + + +/** + * Plug in Mesa's GLSL functions into the device driver function table. + */ +void +_mesa_init_glsl_driver_functions(struct dd_function_table *driver) +{ + driver->AttachShader = _mesa_attach_shader; + driver->BindAttribLocation = _mesa_bind_attrib_location; + driver->CompileShader = _mesa_compile_shader; + driver->CreateProgram = _mesa_create_program; + driver->CreateShader = _mesa_create_shader; + driver->DeleteProgram2 = _mesa_delete_program2; + driver->DeleteShader = _mesa_delete_shader; + driver->DetachShader = _mesa_detach_shader; + driver->GetActiveAttrib = _mesa_get_active_attrib; + driver->GetActiveUniform = _mesa_get_active_uniform; + driver->GetAttachedShaders = _mesa_get_attached_shaders; + driver->GetAttribLocation = _mesa_get_attrib_location; + driver->GetHandle = _mesa_get_handle; + driver->GetProgramiv = _mesa_get_programiv; + driver->GetProgramInfoLog = _mesa_get_program_info_log; + driver->GetShaderiv = _mesa_get_shaderiv; + driver->GetShaderInfoLog = _mesa_get_shader_info_log; + driver->GetShaderSource = _mesa_get_shader_source; + driver->GetUniformfv = _mesa_get_uniformfv; + driver->GetUniformLocation = _mesa_get_uniform_location; + driver->IsProgram = _mesa_is_program; + driver->IsShader = _mesa_is_shader; + driver->LinkProgram = _mesa_link_program; + driver->ShaderSource = _mesa_shader_source; + driver->Uniform = _mesa_uniform; + driver->UniformMatrix = _mesa_uniform_matrix; + driver->UseProgram = _mesa_use_program; + driver->ValidateProgram = _mesa_validate_program; +} diff --git a/src/mesa/shader/shader_api.h b/src/mesa/shader/shader_api.h index 27e5870d705..5521c585b53 100644 --- a/src/mesa/shader/shader_api.h +++ b/src/mesa/shader/shader_api.h @@ -41,9 +41,10 @@ _mesa_init_shader_state(GLcontext * ctx); extern void _mesa_free_shader_state(GLcontext *ctx); +/* extern struct gl_shader_program * _mesa_new_shader_program(GLcontext *ctx, GLuint name); - +*/ extern void _mesa_clear_shader_program_data(GLcontext *ctx, struct gl_shader_program *shProg); @@ -78,108 +79,12 @@ extern struct gl_shader * _mesa_lookup_shader(GLcontext *ctx, GLuint name); -/** - * API/Driver functions - */ - -extern void -_mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader); - -extern void -_mesa_bind_attrib_location(GLcontext *ctx, GLuint program, GLuint index, - const GLchar *name); - -extern void -_mesa_compile_shader(GLcontext *ctx, GLuint shaderObj); - -extern GLuint -_mesa_create_shader(GLcontext *ctx, GLenum type); - -extern GLuint -_mesa_create_program(GLcontext *ctx); - -extern void -_mesa_delete_program2(GLcontext *ctx, GLuint name); - -extern void -_mesa_delete_shader(GLcontext *ctx, GLuint shader); - -extern void -_mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader); - -extern void -_mesa_get_active_attrib(GLcontext *ctx, GLuint program, GLuint index, - GLsizei maxLength, GLsizei *length, GLint *size, - GLenum *type, GLchar *name); - -extern void -_mesa_get_active_uniform(GLcontext *ctx, GLuint program, GLuint index, - GLsizei maxLength, GLsizei *length, GLint *size, - GLenum *type, GLchar *name); - -extern void -_mesa_get_attached_shaders(GLcontext *ctx, GLuint program, GLsizei maxCount, - GLsizei *count, GLuint *obj); - -extern GLint -_mesa_get_attrib_location(GLcontext *ctx, GLuint program, - const GLchar *name); - -extern GLuint -_mesa_get_handle(GLcontext *ctx, GLenum pname); - -extern void -_mesa_get_programiv(GLcontext *ctx, GLuint program, - GLenum pname, GLint *params); - -extern void -_mesa_get_program_info_log(GLcontext *ctx, GLuint program, GLsizei bufSize, - GLsizei *length, GLchar *infoLog); - -extern void -_mesa_get_shaderiv(GLcontext *ctx, GLuint shader, GLenum pname, GLint *params); - -extern void -_mesa_get_shader_info_log(GLcontext *ctx, GLuint shader, GLsizei bufSize, - GLsizei *length, GLchar *infoLog); - -extern void -_mesa_get_shader_source(GLcontext *ctx, GLuint shader, GLsizei maxLength, - GLsizei *length, GLchar *sourceOut); - -extern void -_mesa_get_uniformfv(GLcontext *ctx, GLuint program, GLint location, - GLfloat *params); - -extern GLint -_mesa_get_uniform_location(GLcontext *ctx, GLuint program, const GLchar *name); - -extern GLboolean -_mesa_is_program(GLcontext *ctx, GLuint name); - -extern GLboolean -_mesa_is_shader(GLcontext *ctx, GLuint name); - -extern void -_mesa_link_program(GLcontext *ctx, GLuint program); - -extern void -_mesa_shader_source(GLcontext *ctx, GLuint shader, const GLchar *source); - -extern void -_mesa_uniform(GLcontext *ctx, GLint location, GLsizei count, - const GLvoid *values, GLenum type); - -void -_mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows, - GLenum matrixType, GLint location, GLsizei count, - GLboolean transpose, const GLfloat *values); - extern void _mesa_use_program(GLcontext *ctx, GLuint program); + extern void -_mesa_validate_program(GLcontext *ctx, GLuint program); +_mesa_init_glsl_driver_functions(struct dd_function_table *driver); #endif /* SHADER_API_H */ diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index abde6d64b01..524e06fb00a 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -29,7 +29,7 @@ #include "main/context.h" #include "main/extensions.h" #include "vbo/vbo.h" -#include "drivers/common/driverfuncs.h" +#include "shader/shader_api.h" #include "st_public.h" #include "st_context.h" #include "st_cb_accum.h" -- cgit v1.2.3 From 75ec482001b0c8f386327c2fe5ff8ac4aee43582 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 17:14:02 -0700 Subject: Disable swrast, swrast_setup and driverfuncs in the build. The gallium/state tracker no longer uses them and this speeds up the build a bit. --- src/mesa/sources | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/mesa/sources b/src/mesa/sources index 8286e558d6c..2f16af3aa46 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -80,7 +80,7 @@ MATH_SOURCES = \ math/m_vector.c \ math/m_xform.c -SWRAST_SOURCES = \ +__SWRAST_SOURCES = \ swrast/s_aaline.c \ swrast/s_aatriangle.c \ swrast/s_accum.c \ @@ -111,7 +111,7 @@ SWRAST_SOURCES = \ swrast/s_triangle.c \ swrast/s_zoom.c -SWRAST_SETUP_SOURCES = \ +__SWRAST_SETUP_SOURCES = \ swrast_setup/ss_context.c \ swrast_setup/ss_triangle.c @@ -326,7 +326,7 @@ SPARC_SOURCES = \ SPARC_API = \ sparc/glapi_sparc.S -COMMON_DRIVER_SOURCES = \ +__COMMON_DRIVER_SOURCES = \ drivers/common/driverfuncs.c X11_DRIVER_SOURCES = \ -- cgit v1.2.3 From 5e46b252b923de820f65ed63d379b9b4578e5fa8 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 17:24:42 -0700 Subject: New vbo_set_draw_func() to keep vbo context opaque to state tracker and tnl module. --- src/mesa/state_tracker/st_cb_feedback.c | 8 +++----- src/mesa/state_tracker/st_draw.c | 6 +----- src/mesa/tnl/t_context.c | 3 +++ src/mesa/vbo/vbo.h | 3 +++ src/mesa/vbo/vbo_context.c | 12 ++++++++++++ 5 files changed, 22 insertions(+), 10 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_feedback.c b/src/mesa/state_tracker/st_cb_feedback.c index a9fd2579a22..ea775b9452d 100644 --- a/src/mesa/state_tracker/st_cb_feedback.c +++ b/src/mesa/state_tracker/st_cb_feedback.c @@ -43,7 +43,6 @@ #include "main/macros.h" #include "vbo/vbo.h" -#include "vbo/vbo_context.h" #include "st_context.h" #include "st_atom.h" @@ -281,26 +280,25 @@ static void st_RenderMode(GLcontext *ctx, GLenum newMode ) { struct st_context *st = ctx->st; - struct vbo_context *vbo = (struct vbo_context *) ctx->swtnl_im; struct draw_context *draw = st->draw; if (newMode == GL_RENDER) { /* restore normal VBO draw function */ - vbo->draw_prims = st_draw_vbo; + vbo_set_draw_func(ctx, st_draw_vbo); } else if (newMode == GL_SELECT) { if (!st->selection_stage) st->selection_stage = draw_glselect_stage(ctx, draw); draw_set_rasterize_stage(draw, st->selection_stage); /* Plug in new vbo draw function */ - vbo->draw_prims = st_feedback_draw_vbo; + vbo_set_draw_func(ctx, st_feedback_draw_vbo); } else { if (!st->feedback_stage) st->feedback_stage = draw_glfeedback_stage(ctx, draw); draw_set_rasterize_stage(draw, st->feedback_stage); /* Plug in new vbo draw function */ - vbo->draw_prims = st_feedback_draw_vbo; + vbo_set_draw_func(ctx, st_feedback_draw_vbo); /* need to generate/use a vertex program that emits pos/color/tex */ st->dirty.st |= ST_NEW_VERTEX_PROGRAM; } diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index b4c0d0cdd62..32dcd73c46c 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -34,7 +34,6 @@ #include "main/image.h" #include "vbo/vbo.h" -#include "vbo/vbo_context.h" #include "st_atom.h" #include "st_cache.h" @@ -519,14 +518,11 @@ st_feedback_draw_vbo(GLcontext *ctx, void st_init_draw( struct st_context *st ) { GLcontext *ctx = st->ctx; - struct vbo_context *vbo = (struct vbo_context *) ctx->swtnl_im; /* actually, not used here, but elsewhere */ create_default_attribs_buffer(st); - assert(vbo); - assert(vbo->draw_prims); - vbo->draw_prims = st_draw_vbo; + vbo_set_draw_func(ctx, st_draw_vbo); } diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c index b87452d2eb4..0ace5c2d6f1 100644 --- a/src/mesa/tnl/t_context.c +++ b/src/mesa/tnl/t_context.c @@ -78,6 +78,9 @@ _tnl_CreateContext( GLcontext *ctx ) tnl->nr_blocks = 0; + /* plug in the VBO drawing function */ + vbo_set_draw_func(ctx, _tnl_draw_prims); + return GL_TRUE; } diff --git a/src/mesa/vbo/vbo.h b/src/mesa/vbo/vbo.h index 4c51b44cdad..79d33d09c16 100644 --- a/src/mesa/vbo/vbo.h +++ b/src/mesa/vbo/vbo.h @@ -117,4 +117,7 @@ void vbo_rebase_prims( GLcontext *ctx, void vbo_use_buffer_objects(GLcontext *ctx); +void vbo_set_draw_func(GLcontext *ctx, vbo_draw_func func); + + #endif diff --git a/src/mesa/vbo/vbo_context.c b/src/mesa/vbo/vbo_context.c index ad4556c500b..60d0b76ab62 100644 --- a/src/mesa/vbo/vbo_context.c +++ b/src/mesa/vbo/vbo_context.c @@ -31,6 +31,7 @@ #include "vbo.h" #include "vbo_context.h" +#if 0 /* Reach out and grab this to use as the default: */ extern void _tnl_draw_prims( GLcontext *ctx, @@ -40,6 +41,7 @@ extern void _tnl_draw_prims( GLcontext *ctx, const struct _mesa_index_buffer *ib, GLuint min_index, GLuint max_index ); +#endif @@ -214,7 +216,9 @@ GLboolean _vbo_CreateContext( GLcontext *ctx ) /* By default: */ +#if 0 /* dead - see vbo_set_draw_func() */ vbo->draw_prims = _tnl_draw_prims; +#endif /* Hook our functions into exec and compile dispatch tables. These * will pretty much be permanently installed, which means that the @@ -245,3 +249,11 @@ void _vbo_DestroyContext( GLcontext *ctx ) FREE(vbo_context(ctx)); ctx->swtnl_im = NULL; } + + +void vbo_set_draw_func(GLcontext *ctx, vbo_draw_func func) +{ + struct vbo_context *vbo = vbo_context(ctx); + vbo->draw_prims = func; +} + -- cgit v1.2.3 From 77c7fd3cec458511ea0789130a72e443a8d16d2f Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 17:26:06 -0700 Subject: Disable, remove tnl module from the Gallium build since no longer used. --- src/mesa/sources | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/sources b/src/mesa/sources index 2f16af3aa46..32e1805ecab 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -115,7 +115,7 @@ __SWRAST_SETUP_SOURCES = \ swrast_setup/ss_context.c \ swrast_setup/ss_triangle.c -TNL_SOURCES = \ +__TNL_SOURCES = \ tnl/t_context.c \ tnl/t_pipeline.c \ tnl/t_draw.c \ -- cgit v1.2.3 From 7c8cc71a28e5d02c01c0a528f1c35f29281e88f2 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 20:13:10 -0700 Subject: Tile clearing optimizations. Clear using int values rather than floats when possible. Better performance now. --- src/mesa/pipe/softpipe/sp_clear.c | 42 +----- src/mesa/pipe/softpipe/sp_tile_cache.c | 252 ++++++++++++++++++--------------- src/mesa/pipe/softpipe/sp_tile_cache.h | 4 +- 3 files changed, 147 insertions(+), 151 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index a6352e06161..bfe4a291448 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -51,49 +51,15 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, softpipe_update_derived(softpipe); /* not needed?? */ if (ps == sp_tile_cache_get_surface(softpipe->zbuf_cache)) { - float clear[4]; - clear[0] = 1.0; /* XXX hack */ - sp_tile_cache_clear(softpipe->zbuf_cache, clear); + sp_tile_cache_clear(softpipe->zbuf_cache, clearValue); } else if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[0])) { - float clear[4]; - /* XXX it sure would be nice if the clear color was passed to - * this function as float[4].... - */ - uint r, g, b, a; - switch (ps->format) { - case PIPE_FORMAT_R8G8B8A8_UNORM: - r = (clearValue >> 24) & 0xff; - g = (clearValue >> 16) & 0xff; - g = (clearValue >> 8) & 0xff; - a = (clearValue ) & 0xff; - break; - case PIPE_FORMAT_A8R8G8B8_UNORM: - r = (clearValue >> 16) & 0xff; - g = (clearValue >> 8) & 0xff; - b = (clearValue ) & 0xff; - a = (clearValue >> 24) & 0xff; - break; - case PIPE_FORMAT_B8G8R8A8_UNORM: - r = (clearValue >> 8) & 0xff; - g = (clearValue >> 16) & 0xff; - b = (clearValue >> 24) & 0xff; - a = (clearValue ) & 0xff; - break; - default: - assert(0); - } - - clear[0] = r / 255.0; - clear[1] = g / 255.0; - clear[2] = b / 255.0; - clear[3] = a / 255.0; - - sp_tile_cache_clear(softpipe->cbuf_cache[0], clear); + sp_tile_cache_clear(softpipe->cbuf_cache[0], clearValue); } +#if !TILE_CLEAR_OPTIMIZATION pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); - +#endif #if 0 sp_clear_tile_cache(ps, clearValue); diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 602b78ebe6b..13774e6c8fb 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -52,7 +52,9 @@ struct softpipe_tile_cache struct pipe_texture *texture; /**< if caching a texture */ struct softpipe_cached_tile entries[NUM_ENTRIES]; uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32]; - float clear_value[4]; + float clear_color[4]; + uint clear_val; + boolean depth_stencil; /** Is the surface a depth/stencil format? */ }; @@ -67,6 +69,9 @@ struct softpipe_tile_cache +/** + * Is the tile at (x,y) in cleared state? + */ static INLINE uint is_clear_flag_set(const uint *bitvec, int x, int y) { @@ -80,6 +85,9 @@ is_clear_flag_set(const uint *bitvec, int x, int y) } +/** + * Mark the tile at (x,y) as not cleared. + */ static INLINE void clear_clear_flag(uint *bitvec, int x, int y) { @@ -120,10 +128,15 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc) } +/** + * Specify the surface to cache. + */ void sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *ps) { + assert(!tc->texture); + if (tc->surface && tc->surface->map) { assert(tc->surface != ps); pipe_surface_unmap(tc->surface); @@ -133,9 +146,19 @@ sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, if (!ps->map) pipe_surface_map(ps); + + if (ps) { + tc->depth_stencil = (ps->format == PIPE_FORMAT_S8Z24_UNORM || + ps->format == PIPE_FORMAT_Z16_UNORM || + ps->format == PIPE_FORMAT_Z32_UNORM || + ps->format == PIPE_FORMAT_U_S8); + } } +/** + * Return the surface being cached. + */ struct pipe_surface * sp_tile_cache_get_surface(struct softpipe_tile_cache *tc) { @@ -143,12 +166,17 @@ sp_tile_cache_get_surface(struct softpipe_tile_cache *tc) } +/** + * Specify the texture to cache. + */ void sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, struct pipe_texture *texture) { uint i; + assert(!tc->surface); + tc->texture = texture; /* mark as entries as invalid/empty */ @@ -159,84 +187,71 @@ sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, } -static INLINE boolean -is_depth_stencil_surface(struct pipe_surface *ps) +/** + * Set pixels in a tile to the given clear color/value, float. + */ +static void +clear_tile_rgba(struct softpipe_cached_tile *tile, + enum pipe_format format, + const float clear_value[4]) { - return (ps && - (ps->format == PIPE_FORMAT_S8Z24_UNORM || - ps->format == PIPE_FORMAT_Z16_UNORM || - ps->format == PIPE_FORMAT_Z32_UNORM || - ps->format == PIPE_FORMAT_U_S8)); + if (clear_value[0] == 0.0 && + clear_value[1] == 0.0 && + clear_value[2] == 0.0 && + clear_value[3] == 0.0) { + memset(tile->data.color, 0, sizeof(tile->data.color)); + } + else { + uint i, j; + for (i = 0; i < TILE_SIZE; i++) { + for (j = 0; j < TILE_SIZE; j++) { + tile->data.color[i][j][0] = clear_value[0]; + tile->data.color[i][j][1] = clear_value[1]; + tile->data.color[i][j][2] = clear_value[2]; + tile->data.color[i][j][3] = clear_value[3]; + } + } + } } /** - * Set pixels in a tile to the given clear color/value. + * Set a tile to a solid value/color. */ static void clear_tile(struct softpipe_cached_tile *tile, enum pipe_format format, - const float clear_value[4]) + uint clear_value) { uint i, j; switch (format) { - case PIPE_FORMAT_Z16_UNORM: - { - ushort clear_val = (ushort) (clear_value[0] * 0xffff); - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth16[i][j] = clear_val; - } - } - } - break; - case PIPE_FORMAT_Z32_UNORM: - { - uint clear_val = (uint) (clear_value[0] * 0xffffffff); - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth32[i][j] = clear_val; - } - } - } + case PIPE_FORMAT_U_S8: + /* 8 bpp */ + memset(tile->data.any, 0, TILE_SIZE * TILE_SIZE); break; - case PIPE_FORMAT_S8Z24_UNORM: - { - uint clear_val = (uint) (clear_value[0] * 0xffffff); - clear_val |= ((uint) clear_value[1]) << 24; - for (i = 0; i < TILE_SIZE; i++) { - for (j = 0; j < TILE_SIZE; j++) { - tile->data.depth32[i][j] = clear_val; - } - } + case PIPE_FORMAT_Z16_UNORM: + /* 16 bpp */ + if (clear_value == 0) { + memset(tile->data.any, 0, 2 * TILE_SIZE * TILE_SIZE); } - break; - case PIPE_FORMAT_U_S8: - { - ubyte clear_val = (uint) clear_value[0]; + else { for (i = 0; i < TILE_SIZE; i++) { for (j = 0; j < TILE_SIZE; j++) { - tile->data.stencil8[i][j] = clear_val; + tile->data.depth16[i][j] = clear_value; } } } break; default: - /* color */ - if (clear_value[0] == 0.0 && - clear_value[1] == 0.0 && - clear_value[2] == 0.0 && - clear_value[3] == 0.0) { - memset(tile->data.color, 0, sizeof(tile->data.color)); + /* 32 bpp */ + if (clear_value == 0) { + memset(tile->data.any, 0, 4 * TILE_SIZE * TILE_SIZE); } else { for (i = 0; i < TILE_SIZE; i++) { for (j = 0; j < TILE_SIZE; j++) { - tile->data.color[i][j][0] = clear_value[0]; - tile->data.color[i][j][1] = clear_value[1]; - tile->data.color[i][j][2] = clear_value[2]; - tile->data.color[i][j][3] = clear_value[3]; + tile->data.color32[i][j] = clear_value; } } } @@ -244,52 +259,6 @@ clear_tile(struct softpipe_cached_tile *tile, } -static INLINE void -put_tile(struct pipe_context *pipe, - struct pipe_surface *ps, - struct softpipe_cached_tile *tile) -{ - assert(tile->x >= 0); - assert(tile->x % TILE_SIZE == 0); - assert(tile->y >= 0); - assert(tile->y % TILE_SIZE == 0); - - if (is_depth_stencil_surface(ps)) { - pipe->put_tile(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); - } - else { - pipe->put_tile_rgba(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - } -} - - -static INLINE void -get_tile(struct pipe_context *pipe, - struct pipe_surface *ps, - struct softpipe_cached_tile *tile) -{ - assert(tile->x >= 0); - assert(tile->x % TILE_SIZE == 0); - assert(tile->y >= 0); - assert(tile->y % TILE_SIZE == 0); - - if (is_depth_stencil_surface(ps)) { - pipe->get_tile(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); - } - else { - pipe->get_tile_rgba(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - } -} - - /** * Actually clear the tiles which were flagged as being in a clear state. */ @@ -305,15 +274,15 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, uint numCleared = 0; /* clear one tile to the clear value */ - clear_tile(&tile, ps->format, tc->clear_value); + clear_tile(&tile, ps->format, tc->clear_val); /* push the tile to all positions marked as clear */ for (y = 0; y < h; y += TILE_SIZE) { for (x = 0; x < w; x += TILE_SIZE) { if (is_clear_flag_set(tc->clear_flags, x, y)) { - tile.x = x; - tile.y = y; - put_tile(pipe, ps, &tile); + pipe->put_tile(pipe, ps, + x, y, TILE_SIZE, TILE_SIZE, + tile.data.color32, 0/*STRIDE*/); /* do this? */ clear_clear_flag(tc->clear_flags, x, y); @@ -349,7 +318,16 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, for (pos = 0; pos < NUM_ENTRIES; pos++) { struct softpipe_cached_tile *tile = tc->entries + pos; if (tile->x >= 0) { - put_tile(pipe, ps, tile); + if (tc->depth_stencil) { + pipe->put_tile(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); + } + else { + pipe->put_tile_rgba(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + } tile->x = tile->y = -1; /* mark as empty */ inuse++; } @@ -389,7 +367,16 @@ sp_get_cached_tile(struct softpipe_context *softpipe, if (tile->x != -1) { /* put dirty tile back in framebuffer */ - put_tile(pipe, ps, tile); + if (tc->depth_stencil) { + pipe->put_tile(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); + } + else { + pipe->put_tile_rgba(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + } } tile->x = tile_x; @@ -397,12 +384,26 @@ sp_get_cached_tile(struct softpipe_context *softpipe, if (is_clear_flag_set(tc->clear_flags, x, y)) { /* don't get tile from framebuffer, just clear it */ - clear_tile(tile, ps->format, tc->clear_value); + if (tc->depth_stencil) { + clear_tile(tile, ps->format, tc->clear_val); + } + else { + clear_tile_rgba(tile, ps->format, tc->clear_color); + } clear_clear_flag(tc->clear_flags, x, y); } else { /* get new tile data from surface */ - get_tile(pipe, ps, tile); + if (tc->depth_stencil) { + pipe->get_tile(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); + } + else { + pipe->get_tile_rgba(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + } } } @@ -477,12 +478,39 @@ sp_get_cached_tile_tex(struct pipe_context *pipe, * Save the color and set a 'clearflag' for each tile of the screen. */ void -sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float value[4]) +sp_tile_cache_clear(struct softpipe_tile_cache *tc, uint clearValue) { - tc->clear_value[0] = value[0]; - tc->clear_value[1] = value[1]; - tc->clear_value[2] = value[2]; - tc->clear_value[3] = value[3]; + uint r, g, b, a; + + tc->clear_val = clearValue; + + switch (tc->surface->format) { + case PIPE_FORMAT_R8G8B8A8_UNORM: + r = (clearValue >> 24) & 0xff; + g = (clearValue >> 16) & 0xff; + b = (clearValue >> 8) & 0xff; + a = (clearValue ) & 0xff; + break; + case PIPE_FORMAT_A8R8G8B8_UNORM: + r = (clearValue >> 16) & 0xff; + g = (clearValue >> 8) & 0xff; + b = (clearValue ) & 0xff; + a = (clearValue >> 24) & 0xff; + break; + case PIPE_FORMAT_B8G8R8A8_UNORM: + r = (clearValue >> 8) & 0xff; + g = (clearValue >> 16) & 0xff; + b = (clearValue >> 24) & 0xff; + a = (clearValue ) & 0xff; + break; + default: + r = g = b = a = 0; + } + + tc->clear_color[0] = r / 255.0; + tc->clear_color[1] = g / 255.0; + tc->clear_color[2] = b / 255.0; + tc->clear_color[3] = a / 255.0; #if TILE_CLEAR_OPTIMIZATION /* set flags to indicate all the tiles are cleared */ diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.h b/src/mesa/pipe/softpipe/sp_tile_cache.h index c4f2e0f1840..91fb2795a26 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.h +++ b/src/mesa/pipe/softpipe/sp_tile_cache.h @@ -51,9 +51,11 @@ struct softpipe_cached_tile int z, face, level; /**< Extra texture indexes */ union { float color[TILE_SIZE][TILE_SIZE][4]; + uint color32[TILE_SIZE][TILE_SIZE]; uint depth32[TILE_SIZE][TILE_SIZE]; ushort depth16[TILE_SIZE][TILE_SIZE]; ubyte stencil8[TILE_SIZE][TILE_SIZE]; + ubyte any[1]; } data; }; @@ -80,7 +82,7 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, struct softpipe_tile_cache *tc); extern void -sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float value[4]); +sp_tile_cache_clear(struct softpipe_tile_cache *tc, uint clearValue); extern struct softpipe_cached_tile * sp_get_cached_tile(struct softpipe_context *softpipe, -- cgit v1.2.3 From 92e7a02e5ac8f8589033954c7df188fdc4c330cb Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 20:24:24 -0700 Subject: clean-ups --- src/mesa/pipe/softpipe/sp_surface.c | 20 ++++++-------------- 1 file changed, 6 insertions(+), 14 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_surface.c b/src/mesa/pipe/softpipe/sp_surface.c index cdbffe54000..55b8b85ef2b 100644 --- a/src/mesa/pipe/softpipe/sp_surface.c +++ b/src/mesa/pipe/softpipe/sp_surface.c @@ -30,10 +30,9 @@ #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" #include "sp_context.h" -#include "sp_state.h" #include "sp_surface.h" #include "sp_texture.h" -#include "sp_tile_cache.h" + /** * Softpipe surface functions. @@ -48,14 +47,6 @@ -#if 0 -#define CLIP_TILE \ - do { \ - assert(x + w <= ps->width); \ - assert(y + h <= ps->height); \ - } while(0) - -#else #define CLIP_TILE \ do { \ if (x >= ps->width) \ @@ -65,9 +56,8 @@ if (x + w > ps->width) \ w = ps->width - x; \ if (y + h > ps->height) \ - h = ps->height -y; \ + h = ps->height - y; \ } while(0) -#endif /*** PIPE_FORMAT_A8R8G8B8_UNORM ***/ @@ -611,6 +601,7 @@ softpipe_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, { const uint cpp = ps->cpp; const ubyte *pSrc; + const uint src_stride = ps->pitch * cpp; ubyte *pDest; uint i; @@ -628,7 +619,7 @@ softpipe_get_tile(struct pipe_context *pipe, struct pipe_surface *ps, for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w * cpp); pDest += dst_stride; - pSrc += ps->pitch * cpp; + pSrc += src_stride; } } @@ -643,6 +634,7 @@ softpipe_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, { const uint cpp = ps->cpp; const ubyte *pSrc; + const uint dst_stride = ps->pitch * cpp; ubyte *pDest; uint i; @@ -659,7 +651,7 @@ softpipe_put_tile(struct pipe_context *pipe, struct pipe_surface *ps, for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w * cpp); - pDest += ps->pitch * cpp; + pDest += dst_stride; pSrc += src_stride; } } -- cgit v1.2.3 From 0ee512aaf03a2091f0cc5eee26712fe1f0992159 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 20:31:26 -0700 Subject: code re-org in softpipe_clear() --- src/mesa/pipe/softpipe/sp_clear.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index bfe4a291448..f32e834ac0e 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -47,19 +47,28 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { struct softpipe_context *softpipe = softpipe_context(pipe); + uint i; +#if 0 softpipe_update_derived(softpipe); /* not needed?? */ +#endif +#if TILE_CLEAR_OPTIMIZATION if (ps == sp_tile_cache_get_surface(softpipe->zbuf_cache)) { sp_tile_cache_clear(softpipe->zbuf_cache, clearValue); + return; } - else if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[0])) { - sp_tile_cache_clear(softpipe->cbuf_cache[0], clearValue); + + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { + if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[i])) { + sp_tile_cache_clear(softpipe->cbuf_cache[i], clearValue); + return; + } } +#endif -#if !TILE_CLEAR_OPTIMIZATION + /* non-cached surface */ pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); -#endif #if 0 sp_clear_tile_cache(ps, clearValue); -- cgit v1.2.3 From 8c1c222b6ded06e40c044f4dee4550005ba620f9 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 20:58:01 -0700 Subject: shorten loops over color bufs --- src/mesa/pipe/softpipe/sp_clear.c | 2 +- src/mesa/pipe/softpipe/sp_context.c | 2 +- src/mesa/pipe/softpipe/sp_flush.c | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index f32e834ac0e..5b3857145d6 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -59,7 +59,7 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, return; } - for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { + for (i = 0; i < softpipe->framebuffer.num_cbufs; i++) { if (ps == sp_tile_cache_get_surface(softpipe->cbuf_cache[i])) { sp_tile_cache_clear(softpipe->cbuf_cache[i], clearValue); return; diff --git a/src/mesa/pipe/softpipe/sp_context.c b/src/mesa/pipe/softpipe/sp_context.c index bdfd6228ef8..df4e0cbd5ae 100644 --- a/src/mesa/pipe/softpipe/sp_context.c +++ b/src/mesa/pipe/softpipe/sp_context.c @@ -94,7 +94,7 @@ softpipe_unmap_surfaces(struct softpipe_context *sp) struct pipe_surface *ps; uint i; - for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) + for (i = 0; i < sp->framebuffer.num_cbufs; i++) sp_flush_tile_cache(sp, sp->cbuf_cache[i]); sp_flush_tile_cache(sp, sp->zbuf_cache); sp_flush_tile_cache(sp, sp->sbuf_cache); diff --git a/src/mesa/pipe/softpipe/sp_flush.c b/src/mesa/pipe/softpipe/sp_flush.c index 1010924bf61..47b11803ce1 100644 --- a/src/mesa/pipe/softpipe/sp_flush.c +++ b/src/mesa/pipe/softpipe/sp_flush.c @@ -55,7 +55,7 @@ softpipe_flush( struct pipe_context *pipe, * - flush the render cache */ - for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) + for (i = 0; i < softpipe->framebuffer.num_cbufs; i++) if (softpipe->cbuf_cache[i]) sp_flush_tile_cache(softpipe, softpipe->cbuf_cache[i]); -- cgit v1.2.3 From 274a3d9cb6a196c1b191769f581915cf5f8dccfe Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 7 Dec 2007 21:21:58 -0700 Subject: Try to reduce the frequency of calls to pipe->get_tex_surface() Save the surface info in the tile cache and re-use whenever possible. --- src/mesa/pipe/softpipe/sp_tile_cache.c | 41 ++++++++++++++++++++++++++-------- 1 file changed, 32 insertions(+), 9 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 13774e6c8fb..fadd169f5df 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -55,6 +55,9 @@ struct softpipe_tile_cache float clear_color[4]; uint clear_val; boolean depth_stencil; /** Is the surface a depth/stencil format? */ + + struct pipe_surface *tex_surf; + int tex_face, tex_level, tex_z; }; @@ -179,11 +182,17 @@ sp_tile_cache_set_texture(struct softpipe_tile_cache *tc, tc->texture = texture; + if (tc->tex_surf && tc->tex_surf->map) + pipe_surface_unmap(tc->tex_surf); + pipe_surface_reference(&tc->tex_surf, NULL); + /* mark as entries as invalid/empty */ /* XXX we should try to avoid this when the teximage hasn't changed */ for (i = 0; i < NUM_ENTRIES; i++) { tc->entries[i].x = -1; } + + tc->tex_face = -1; /* any invalid value here */ } @@ -448,19 +457,33 @@ sp_get_cached_tile_tex(struct pipe_context *pipe, z != tile->z || face != tile->face || level != tile->level) { - /* XXX this call is a bit heavier than we'd like: */ - struct pipe_surface *ps - = pipe->get_tex_surface(pipe, tc->texture, face, level, z); + /* cache miss */ - pipe_surface_map(ps); + /* check if we need to get a new surface */ + if (!tc->tex_surf || + tc->tex_face != face || + tc->tex_level != level || + tc->tex_z != z) { + /* get new surface (view into texture) */ + struct pipe_surface *ps; - pipe->get_tile_rgba(pipe, ps, - tile_x, tile_y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); + if (tc->tex_surf && tc->tex_surf->map) + pipe_surface_unmap(tc->tex_surf); - pipe_surface_unmap(ps); - pipe_surface_reference(&ps, NULL); + ps = pipe->get_tex_surface(pipe, tc->texture, face, level, z); + pipe_surface_reference(&tc->tex_surf, ps); + pipe_surface_map(ps); + + tc->tex_face = face; + tc->tex_level = level; + tc->tex_z = z; + } + + /* get tile from the surface (view into texture) */ + pipe->get_tile_rgba(pipe, tc->tex_surf, + tile_x, tile_y, TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); tile->x = tile_x; tile->y = tile_y; tile->z = z; -- cgit v1.2.3 From b8965bee404cb36ccd97ac089fbd3ffc63268080 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 9 Dec 2007 03:22:14 +1100 Subject: nouveau: adapt to gallium interface changes --- src/mesa/drivers/dri/Makefile.template | 1 - src/mesa/drivers/dri/nouveau_winsys/Makefile | 4 +- .../drivers/dri/nouveau_winsys/nouveau_context.c | 36 ++- .../drivers/dri/nouveau_winsys/nouveau_context.h | 44 ++-- src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c | 15 +- .../drivers/dri/nouveau_winsys/nouveau_screen.c | 6 - .../dri/nouveau_winsys/nouveau_swapbuffers.c | 10 +- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 42 ++-- .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 61 ++--- .../dri/nouveau_winsys/nouveau_winsys_softpipe.c | 6 +- src/mesa/drivers/dri/nouveau_winsys/nv04_region.c | 249 --------------------- src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 244 ++++++++++++++++++++ src/mesa/drivers/dri/nouveau_winsys/nv50_region.c | 185 --------------- src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c | 182 +++++++++++++++ src/mesa/pipe/nouveau/nouveau_winsys.h | 17 +- src/mesa/pipe/nv40/Makefile | 1 - src/mesa/pipe/nv40/nv40_clear.c | 3 +- src/mesa/pipe/nv40/nv40_context.c | 15 +- src/mesa/pipe/nv40/nv40_context.h | 16 +- src/mesa/pipe/nv40/nv40_miptree.c | 116 +++++++--- src/mesa/pipe/nv40/nv40_query.c | 39 ++-- src/mesa/pipe/nv40/nv40_region.c | 85 ------- src/mesa/pipe/nv40/nv40_state.c | 59 +++-- src/mesa/pipe/nv40/nv40_state.h | 12 + src/mesa/pipe/nv40/nv40_state_tex.c | 50 +++-- src/mesa/pipe/nv40/nv40_surface.c | 114 ++++++---- src/mesa/pipe/nv50/Makefile | 1 - src/mesa/pipe/nv50/nv50_clear.c | 3 +- src/mesa/pipe/nv50/nv50_context.c | 13 +- src/mesa/pipe/nv50/nv50_context.h | 12 +- src/mesa/pipe/nv50/nv50_miptree.c | 17 +- src/mesa/pipe/nv50/nv50_query.c | 13 +- src/mesa/pipe/nv50/nv50_region.c | 85 ------- src/mesa/pipe/nv50/nv50_state.c | 2 +- src/mesa/pipe/nv50/nv50_surface.c | 93 ++++---- 35 files changed, 849 insertions(+), 1002 deletions(-) delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv04_region.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv50_region.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c delete mode 100644 src/mesa/pipe/nv40/nv40_region.c delete mode 100644 src/mesa/pipe/nv50/nv50_region.c diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template index 37bf0a49772..1d0b0726db1 100644 --- a/src/mesa/drivers/dri/Makefile.template +++ b/src/mesa/drivers/dri/Makefile.template @@ -3,7 +3,6 @@ MESA_MODULES = $(TOP)/src/mesa/libmesa.a COMMON_SOURCES = \ - ../../common/driverfuncs.c \ ../common/utils.c \ ../common/texmem.c \ ../common/vblank.c \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index 292ec5e0a43..9b8a33b725c 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -25,8 +25,8 @@ DRIVER_SOURCES = \ nouveau_winsys.c \ nouveau_winsys_pipe.c \ nouveau_winsys_softpipe.c \ - nv04_region.c \ - nv50_region.c + nv04_surface.c \ + nv50_surface.c C_SOURCES = \ $(COMMON_SOURCES) \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c index 70655a1a7d9..03f1816c323 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c @@ -1,18 +1,8 @@ -#include "glheader.h" -#include "context.h" -#include "extensions.h" - -#include "tnl/tnl.h" -#include "tnl/t_pipeline.h" -#include "tnl/t_vertex.h" - -#include "drivers/common/driverfuncs.h" - -#include "drirenderbuffer.h" +#include "main/glheader.h" +#include "glapi/glthread.h" +#include #include "state_tracker/st_public.h" -#include "state_tracker/st_context.h" - #include "pipe/p_defines.h" #include "pipe/p_context.h" @@ -130,7 +120,7 @@ nouveau_context_create(const __GLcontextModes *glVis, * of the front buffer handle passed to us by the DDX. */ { - struct pipe_region *fb_region; + struct pipe_surface *fb_surf; struct nouveau_bo_priv *fb_bo; fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); @@ -145,14 +135,14 @@ nouveau_context_create(const __GLcontextModes *glVis, fb_bo->base.size = fb_bo->drm.size; fb_bo->base.device = nv_screen->device; - fb_region = calloc(1, sizeof(struct pipe_region)); - fb_region->cpp = nv_screen->front_cpp; - fb_region->pitch = nv_screen->front_pitch / fb_region->cpp; - fb_region->height = nv_screen->front_height; - fb_region->refcount = 1; - fb_region->buffer = (void *)fb_bo; + fb_surf = calloc(1, sizeof(struct pipe_surface)); + fb_surf->cpp = nv_screen->front_cpp; + fb_surf->pitch = nv_screen->front_pitch / fb_surf->cpp; + fb_surf->height = nv_screen->front_height; + fb_surf->refcount = 1; + fb_surf->buffer = (void *)fb_bo; - nv->frontbuffer = fb_region; + nv->frontbuffer = fb_surf; } if ((ret = nouveau_grobj_alloc(nv->channel, 0x00000000, 0x30, @@ -169,9 +159,9 @@ nouveau_context_create(const __GLcontextModes *glVis, } if (nv->chipset < 0x50) - ret = nouveau_region_init_nv04(nv); + ret = nouveau_surface_init_nv04(nv); else - ret = nouveau_region_init_nv50(nv); + ret = nouveau_surface_init_nv50(nv); if (ret) { return GL_FALSE; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index 70110f97683..7033ff7ae96 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -1,9 +1,6 @@ #ifndef __NOUVEAU_CONTEXT_H__ #define __NOUVEAU_CONTEXT_H__ -#include "glheader.h" -#include "context.h" - #include "dri_util.h" #include "xmlconfig.h" @@ -31,7 +28,7 @@ struct nouveau_context { drmLock drm_lock; GLboolean locked; struct nouveau_screen *nv_screen; - struct pipe_region *frontbuffer; + struct pipe_surface *frontbuffer; /* Bufmgr */ struct { @@ -58,29 +55,22 @@ struct nouveau_context { uint32_t next_handle; uint32_t next_sequence; - /* pipe_region accel */ - struct pipe_region *region_src, *region_dst; - unsigned region_src_offset, region_dst_offset; - int (*region_copy_prep)(struct nouveau_context *, - struct pipe_region *dst, uint32_t dst_offset, - struct pipe_region *src, uint32_t src_offset); - void (*region_copy)(struct nouveau_context *, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h); - void (*region_copy_done)(struct nouveau_context *); - int (*region_fill)(struct nouveau_context *, struct pipe_region *, - unsigned, unsigned, unsigned, unsigned, unsigned, - unsigned); - int (*region_data)(struct nouveau_context *, struct pipe_region *, - unsigned, unsigned, unsigned, const void *, - unsigned, unsigned, unsigned, unsigned, unsigned); + /* pipe_surface accel */ + struct pipe_surface *surf_src, *surf_dst; + unsigned surf_src_offset, surf_dst_offset; + int (*surface_copy_prep)(struct nouveau_context *, + struct pipe_surface *dst, + struct pipe_surface *src); + void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h); + void (*surface_copy_done)(struct nouveau_context *); + int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned, unsigned); + int (*surface_data)(struct nouveau_context *, struct pipe_surface *, + unsigned, unsigned, const void *, unsigned, + unsigned, unsigned, unsigned, unsigned); }; -static INLINE struct nouveau_context * -nouveau_context(GLcontext *ctx) -{ - return (struct nouveau_context *)ctx; -} - extern GLboolean nouveau_context_create(const __GLcontextModes *, __DRIcontextPrivate *, void *); extern void nouveau_context_destroy(__DRIcontextPrivate *); @@ -105,7 +95,7 @@ extern int __nouveau_debug; extern void LOCK_HARDWARE(struct nouveau_context *); extern void UNLOCK_HARDWARE(struct nouveau_context *); -extern int nouveau_region_init_nv04(struct nouveau_context *); -extern int nouveau_region_init_nv50(struct nouveau_context *); +extern int nouveau_surface_init_nv04(struct nouveau_context *); +extern int nouveau_surface_init_nv50(struct nouveau_context *); #endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c index ffe7d01d667..9adb9ac8547 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c @@ -25,18 +25,9 @@ * **************************************************************************/ - -#include "glheader.h" -#include "context.h" -#include "extensions.h" - -#include "state_tracker/st_public.h" -#include "state_tracker/st_context.h" - -#include "drirenderbuffer.h" -#include "vblank.h" -#include "utils.h" -#include "xmlpool.h" /* for symbolic values of enum-type options */ +#include "main/glheader.h" +#include "glapi/glthread.h" +#include #include "nouveau_context.h" #include "nouveau_screen.h" diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index 6554082e622..a1eed2e44aa 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -1,9 +1,3 @@ -#include "glheader.h" -#include "context.h" -#include "framebuffer.h" -#include "matrix.h" -#include "renderbuffer.h" -#include "simple_list.h" #include "utils.h" #include "vblank.h" #include "xmlpool.h" diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c index 5b2e76b942f..91bf243f424 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c @@ -1,7 +1,10 @@ -#include "context.h" +#include "main/glheader.h" +#include "glapi/glthread.h" +#include #include "pipe/p_context.h" #include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" #include "state_tracker/st_cb_fbo.h" #include "nouveau_context.h" @@ -14,7 +17,6 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, const drm_clip_rect_t *rect) { struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; - struct pipe_region *p_region = surf->region; drm_clip_rect_t *pbox; int nbox, i; @@ -26,7 +28,7 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, pbox = dPriv->pClipRects; nbox = dPriv->numClipRects; - nv->region_copy_prep(nv, nv->frontbuffer, 0, p_region, 0); + nv->surface_copy_prep(nv, nv->frontbuffer, surf); for (i = 0; i < nbox; i++, pbox++) { int sx, sy, dx, dy, w, h; @@ -37,7 +39,7 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, w = pbox->x2 - pbox->x1; h = pbox->y2 - pbox->y1; - nv->region_copy(nv, dx, dy, sx, sy, w, h); + nv->surface_copy(nv, dx, dy, sx, sy, w, h); } FIRE_RING(); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index e35b6009e34..6ef81ccb86a 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -1,3 +1,5 @@ +#include "pipe/p_util.h" + #include "nouveau_context.h" #include "nouveau_winsys_pipe.h" @@ -128,40 +130,38 @@ nouveau_pipe_dma_kickoff(struct nouveau_channel *userchan) } static int -nouveau_pipe_region_copy(struct nouveau_winsys *nvws, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - struct pipe_region *src, unsigned src_offset, - unsigned sx, unsigned sy, unsigned w, unsigned h) +nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, struct pipe_surface *src, + unsigned sx, unsigned sy, unsigned w, unsigned h) { struct nouveau_context *nv = nvws->nv; - if (nv->region_copy_prep(nv, dst, dst_offset, src, src_offset)) + if (nv->surface_copy_prep(nv, dst, src)) return 1; - nv->region_copy(nv, dx, dy, sx, sy, w, h); - nv->region_copy_done(nv); + nv->surface_copy(nv, dx, dy, sx, sy, w, h); + nv->surface_copy_done(nv); return 0; } static int -nouveau_pipe_region_fill(struct nouveau_winsys *nvws, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - unsigned w, unsigned h, unsigned value) +nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) { - if (nvws->nv->region_fill(nvws->nv, dst, dst_offset, dx, dy, - w, h, value)) + if (nvws->nv->surface_fill(nvws->nv, dst, dx, dy, w, h, value)) return 1; return 0; } static int -nouveau_pipe_region_data(struct nouveau_winsys *nvws, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - const void *src, unsigned src_pitch, - unsigned sx, unsigned sy, unsigned w, unsigned h) +nouveau_pipe_surface_data(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, const void *src, + unsigned src_pitch, unsigned sx, unsigned sy, + unsigned w, unsigned h) { - if (nvws->nv->region_data(nvws->nv, dst, dst_offset, dx, dy, src, - src_pitch, sx, sy, w, h)) + if (nvws->nv->surface_data(nvws->nv, dst, dx, dy, src, src_pitch, sx, + sy, w, h)) return 1; return 0; } @@ -211,9 +211,9 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->notifier_retval = nouveau_notifier_return_val; nvws->notifier_wait = nouveau_notifier_wait_status; - nvws->region_copy = nouveau_pipe_region_copy; - nvws->region_fill = nouveau_pipe_region_fill; - nvws->region_data = nouveau_pipe_region_data; + nvws->surface_copy = nouveau_pipe_surface_copy; + nvws->surface_fill = nouveau_pipe_surface_fill; + nvws->surface_data = nouveau_pipe_surface_data; return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index 0dd2cec2ca1..adbf6b28e8d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -1,5 +1,6 @@ #include "pipe/p_winsys.h" #include "pipe/p_defines.h" +#include "pipe/p_util.h" #include "nouveau_context.h" #include "nouveau_device.h" @@ -33,46 +34,18 @@ nouveau_get_name(struct pipe_winsys *pws) return "Nouveau/DRI"; } -static struct pipe_region * -nouveau_region_alloc(struct pipe_winsys *ws, unsigned cpp, - unsigned width, unsigned height, unsigned flags) +static unsigned +nouveau_surface_pitch(struct pipe_winsys *ws, unsigned cpp, unsigned width, + unsigned flags) { - struct pipe_region *region; - - region = calloc(sizeof(*region), 1); - region->cpp = cpp; - region->pitch = ((cpp * width + 63) & ~63) / cpp; - region->height = height; - region->refcount = 1; - region->buffer = ws->buffer_create(ws, 64); - - ws->buffer_data(ws, region->buffer, region->pitch * cpp * height, NULL, - PIPE_BUFFER_USAGE_PIXEL); - return region; -} - -static void -nouveau_region_release(struct pipe_winsys *pws, struct pipe_region **pregion) -{ - struct pipe_region *region; - - if (!pregion || !*pregion) - return; - region = *pregion; - *pregion = NULL; - - assert(region->refcount > 0); - region->refcount--; + unsigned pitch = width * cpp; - if (region->refcount == 0) { - assert(region->map_refcount == 0); - pws->buffer_reference(pws, ®ion->buffer, NULL); - free(region); - } + pitch = (pitch + 63) & ~63; + return pitch / cpp; } static struct pipe_surface * -nouveau_surface_alloc(struct pipe_winsys *pws, unsigned format) +nouveau_surface_alloc(struct pipe_winsys *ws, unsigned format) { struct pipe_surface *surf; @@ -82,19 +55,19 @@ nouveau_surface_alloc(struct pipe_winsys *pws, unsigned format) surf->format = format; surf->refcount = 1; - surf->winsys = pws; - + surf->winsys = ws; return surf; } static void -nouveau_surface_release(struct pipe_winsys *pws, struct pipe_surface **s) +nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) { - struct pipe_surface *surf = *s; *s = NULL; + struct pipe_surface *surf = *s; - if (surf->refcount-- == 0) { - if (surf->region) - pws->region_release(pws, &surf->region); + *s = NULL; + if (--surf->refcount <= 0) { + if (surf->buffer) + ws->buffer_reference(ws, &surf->buffer, NULL); free(surf); } } @@ -229,9 +202,7 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws->flush_frontbuffer = nouveau_flush_frontbuffer; pws->printf = nouveau_printf; - pws->region_alloc = nouveau_region_alloc; - pws->region_release = nouveau_region_release; - + pws->surface_pitch = nouveau_surface_pitch; pws->surface_alloc = nouveau_surface_alloc; pws->surface_release = nouveau_surface_release; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c index c86575f1875..ebce4ddede1 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c @@ -49,9 +49,9 @@ static boolean nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) { switch (format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_U_R5_G6_B5: - case PIPE_FORMAT_S8_Z24: + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_S8Z24_UNORM: return TRUE; default: break; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c deleted file mode 100644 index d41588d8907..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_region.c +++ /dev/null @@ -1,249 +0,0 @@ -#include "pipe/p_context.h" - -#include "nouveau_context.h" - -static INLINE int -nv04_surface_format(int cpp) -{ - switch (cpp) { - case 1: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; - case 2: return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; - case 4: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; - default: - return -1; - } -} - -static INLINE int -nv04_rect_format(int cpp) -{ - switch (cpp) { - case 1: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - case 2: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; - case 4: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - default: - return -1; - } -} - -static void -nv04_region_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct pipe_region *dst = nv->region_dst; - struct pipe_region *src = nv->region_dst; - unsigned dst_offset, src_offset; - - dst_offset = nv->region_dst_offset + (dy * dst->pitch + dx) * dst->cpp; - src_offset = nv->region_src_offset + (sy * src->pitch + sx) * src->cpp; - - while (h) { - int count = (h > 2047) ? 2047 : h; - - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); - OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_WR); - OUT_RING (src->pitch * src->cpp); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (w * src->cpp); - OUT_RING (count); - OUT_RING (0x0101); - OUT_RING (0); - - h -= count; - src_offset += src->pitch * count; - dst_offset += dst->pitch * count; - } -} - -static void -nv04_region_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - BEGIN_RING(NvImageBlit, 0x0300, 3); - OUT_RING ((sy << 16) | sx); - OUT_RING ((dy << 16) | dx); - OUT_RING (( h << 16) | w); -} - -static int -nv04_region_copy_prep(struct nouveau_context *nv, - struct pipe_region *dst, unsigned dst_offset, - struct pipe_region *src, unsigned src_offset) -{ - int format; - - if (src->cpp != dst->cpp) - return 1; - - /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback - * to NV_MEMORY_TO_MEMORY_FORMAT in this case. - */ - if ((src_offset & 63) || (dst_offset & 63)) { - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_WR); - - nv->region_copy = nv04_region_copy_m2mf; - nv->region_dst = dst; - nv->region_dst_offset = dst_offset; - nv->region_src = src; - nv->region_src_offset = src_offset; - return 0; - - } - - if ((format = nv04_surface_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); - return 1; - } - nv->region_copy = nv04_region_copy_blit; - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (format); - OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); - OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - return 0; -} - -static void -nv04_region_copy_done(struct nouveau_context *nv) -{ - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(NvGdiRect, 0x104, 1); - OUT_RING (0); - BEGIN_RING(NvGdiRect, 0x100, 1); - OUT_RING (0); - FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); -} - -static int -nv04_region_fill(struct nouveau_context *nv, - struct pipe_region *dst, unsigned dst_offset, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - int cs2d_format, gdirect_format; - - if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); - return 1; - } - - if ((gdirect_format = nv04_rect_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); - return 1; - } - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (cs2d_format); - OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); - OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); - OUT_RING (gdirect_format); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); - OUT_RING (value); - BEGIN_RING(NvGdiRect, - NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); - OUT_RING ((dx << 16) | dy); - OUT_RING (( w << 16) | h); - - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(NvGdiRect, 0x104, 1); - OUT_RING (0); - BEGIN_RING(NvGdiRect, 0x100, 1); - OUT_RING (0); - FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); - - return 0; -} - -static int -nv04_region_data(struct nouveau_context *nv, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - const void *src, unsigned src_pitch, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - NOUVEAU_ERR("unimplemented!!\n"); - return 0; -} - -int -nouveau_region_init_nv04(struct nouveau_context *nv) -{ - unsigned class; - int ret; - - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, - &nv->NvM2MF))) { - NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); - return 1; - } - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - - class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : - NV10_CONTEXT_SURFACES_2D; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvCtxSurf2D))) { - NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); - return 1; - } - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RING (nv->channel->vram->handle); - OUT_RING (nv->channel->vram->handle); - - class = nv->chipset < 0x10 ? NV_IMAGE_BLIT : - NV12_IMAGE_BLIT; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvImageBlit))) { - NOUVEAU_ERR("Error creating blit object: %d\n", ret); - return 1; - } - BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_SURFACE, 1); - OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_OPERATION, 1); - OUT_RING (NV_IMAGE_BLIT_OPERATION_SRCCOPY); - - class = NV04_GDI_RECTANGLE_TEXT; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvGdiRect))) { - NOUVEAU_ERR("Error creating rect object: %d\n", ret); - return 1; - } - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); - OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); - OUT_RING (NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); - OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); - - nv->region_copy_prep = nv04_region_copy_prep; - nv->region_copy = nv04_region_copy_blit; - nv->region_copy_done = nv04_region_copy_done; - nv->region_fill = nv04_region_fill; - nv->region_data = nv04_region_data; - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c new file mode 100644 index 00000000000..893f292e232 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c @@ -0,0 +1,244 @@ +#include "pipe/p_context.h" + +#include "nouveau_context.h" + +static INLINE int +nv04_surface_format(int cpp) +{ + switch (cpp) { + case 1: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; + case 2: return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; + case 4: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; + default: + return -1; + } +} + +static INLINE int +nv04_rect_format(int cpp) +{ + switch (cpp) { + case 1: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + case 2: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; + case 4: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + default: + return -1; + } +} + +static void +nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct pipe_surface *dst = nv->surf_dst; + struct pipe_surface *src = nv->surf_dst; + unsigned dst_offset, src_offset; + + dst_offset = dst->offset + (dy * dst->pitch + dx) * dst->cpp; + src_offset = src->offset + (sy * src->pitch + sx) * src->cpp; + + while (h) { + int count = (h > 2047) ? 2047 : h; + + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_WR); + OUT_RING (src->pitch * src->cpp); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (w * src->cpp); + OUT_RING (count); + OUT_RING (0x0101); + OUT_RING (0); + + h -= count; + src_offset += src->pitch * count; + dst_offset += dst->pitch * count; + } +} + +static void +nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + BEGIN_RING(NvImageBlit, 0x0300, 3); + OUT_RING ((sy << 16) | sx); + OUT_RING ((dy << 16) | dx); + OUT_RING (( h << 16) | w); +} + +static int +nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, + struct pipe_surface *src) +{ + int format; + + if (src->cpp != dst->cpp) + return 1; + + /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback + * to NV_MEMORY_TO_MEMORY_FORMAT in this case. + */ + if ((src->offset & 63) || (dst->offset & 63)) { + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_WR); + + nv->surface_copy = nv04_surface_copy_m2mf; + nv->surf_dst = dst; + nv->surf_src = src; + return 0; + + } + + if ((format = nv04_surface_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + nv->surface_copy = nv04_surface_copy_blit; + + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (format); + OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); + OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + return 0; +} + +static void +nv04_surface_copy_done(struct nouveau_context *nv) +{ + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(NvGdiRect, 0x104, 1); + OUT_RING (0); + BEGIN_RING(NvGdiRect, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); +} + +static int +nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + int cs2d_format, gdirect_format; + + if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + if ((gdirect_format = nv04_rect_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (cs2d_format); + OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); + OUT_RING (gdirect_format); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); + OUT_RING (value); + BEGIN_RING(NvGdiRect, + NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); + OUT_RING ((dx << 16) | dy); + OUT_RING (( w << 16) | h); + + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(NvGdiRect, 0x104, 1); + OUT_RING (0); + BEGIN_RING(NvGdiRect, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); + + return 0; +} + +static int +nv04_surface_data(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, const void *src, unsigned src_pitch, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + NOUVEAU_ERR("unimplemented!!\n"); + return 0; +} + +int +nouveau_surface_init_nv04(struct nouveau_context *nv) +{ + unsigned class; + int ret; + + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, + &nv->NvM2MF))) { + NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); + return 1; + } + BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + + class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : + NV10_CONTEXT_SURFACES_2D; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, + &nv->NvCtxSurf2D))) { + NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); + return 1; + } + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RING (nv->channel->vram->handle); + OUT_RING (nv->channel->vram->handle); + + class = nv->chipset < 0x10 ? NV_IMAGE_BLIT : + NV12_IMAGE_BLIT; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, + &nv->NvImageBlit))) { + NOUVEAU_ERR("Error creating blit object: %d\n", ret); + return 1; + } + BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_SURFACE, 1); + OUT_RING (nv->NvCtxSurf2D->handle); + BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_OPERATION, 1); + OUT_RING (NV_IMAGE_BLIT_OPERATION_SRCCOPY); + + class = NV04_GDI_RECTANGLE_TEXT; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, + &nv->NvGdiRect))) { + NOUVEAU_ERR("Error creating rect object: %d\n", ret); + return 1; + } + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); + OUT_RING (nv->NvCtxSurf2D->handle); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); + OUT_RING (NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); + OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); + + nv->surface_copy_prep = nv04_surface_copy_prep; + nv->surface_copy = nv04_surface_copy_blit; + nv->surface_copy_done = nv04_surface_copy_done; + nv->surface_fill = nv04_surface_fill; + nv->surface_data = nv04_surface_data; + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c deleted file mode 100644 index 168cbdfb7c2..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_region.c +++ /dev/null @@ -1,185 +0,0 @@ -#include "pipe/p_context.h" - -#include "nouveau_context.h" - -static INLINE int -nv50_format(int cpp) -{ - switch (cpp) { - case 4: return NV50_2D_DST_FORMAT_32BPP; - case 3: return NV50_2D_DST_FORMAT_24BPP; - case 2: return NV50_2D_DST_FORMAT_16BPP; - case 1: return NV50_2D_DST_FORMAT_8BPP; - default: - return -1; - } -} - -static int -nv50_region_copy_prep(struct nouveau_context *nv, - struct pipe_region *dst, unsigned dst_offset, - struct pipe_region *src, unsigned src_offset) -{ - int surf_format; - - assert(src->cpp == dst->cpp); - - surf_format = nv50_format(dst->cpp); - assert(surf_format >= 0); - - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - OUT_RELOCh(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); - OUT_RING (0); - OUT_RING (0); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - - BEGIN_RING(Nv2D, NV50_2D_SRC_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_SRC_PITCH, 5); - OUT_RING (src->pitch * src->cpp); - OUT_RING (src->pitch); - OUT_RING (src->height); - OUT_RELOCh(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - - return 0; -} - -static void -nv50_region_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - BEGIN_RING(Nv2D, 0x0110, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, NV50_2D_BLIT_DST_X, 12); - OUT_RING (dx); - OUT_RING (dy); - OUT_RING (w); - OUT_RING (h); - OUT_RING (0); - OUT_RING (1); - OUT_RING (0); - OUT_RING (1); - OUT_RING (0); - OUT_RING (sx); - OUT_RING (0); - OUT_RING (sy); -} - -static void -nv50_region_copy_done(struct nouveau_context *nv) -{ - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(Nv2D, 0x104, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, 0x100, 1); - OUT_RING (0); - FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); -} - -static int -nv50_region_fill(struct nouveau_context *nv, - struct pipe_region *dst, unsigned dst_offset, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - int surf_format, rect_format; - - surf_format = nv50_format(dst->cpp); - if (surf_format < 0) - return 1; - - rect_format = nv50_format(dst->cpp); - if (rect_format < 0) - return 1; - - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY1, 1); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - OUT_RELOCh(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); - OUT_RING (0); - OUT_RING (0); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - - BEGIN_RING(Nv2D, 0x0580, 3); - OUT_RING (4); - OUT_RING (rect_format); - OUT_RING (value); - - BEGIN_RING(Nv2D, NV50_2D_RECT_X1, 4); - OUT_RING (dx); - OUT_RING (dy); - OUT_RING (dx + w); - OUT_RING (dy + h); - - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(Nv2D, 0x104, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, 0x100, 1); - OUT_RING (0); - FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); - - return 0; -} - -static int -nv50_region_data(struct nouveau_context *nv, struct pipe_region *dst, - unsigned dst_offset, unsigned dx, unsigned dy, - const void *src, unsigned src_pitch, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - NOUVEAU_ERR("unimplemented!!\n"); - return 0; -} - -int -nouveau_region_init_nv50(struct nouveau_context *nv) -{ - int ret; - - ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, NV50_2D, - &nv->Nv2D); - if (ret) - return ret; - - BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RING (nv->channel->vram->handle); - OUT_RING (nv->channel->vram->handle); - BEGIN_RING(Nv2D, NV50_2D_OPERATION, 1); - OUT_RING (NV50_2D_OPERATION_SRCCOPY); - - nv->region_copy_prep = nv50_region_copy_prep; - nv->region_copy = nv50_region_copy; - nv->region_copy_done = nv50_region_copy_done; - nv->region_fill = nv50_region_fill; - nv->region_data = nv50_region_data; - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c new file mode 100644 index 00000000000..2525476a792 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c @@ -0,0 +1,182 @@ +#include "pipe/p_context.h" + +#include "nouveau_context.h" + +static INLINE int +nv50_format(int cpp) +{ + switch (cpp) { + case 4: return NV50_2D_DST_FORMAT_32BPP; + case 3: return NV50_2D_DST_FORMAT_24BPP; + case 2: return NV50_2D_DST_FORMAT_16BPP; + case 1: return NV50_2D_DST_FORMAT_8BPP; + default: + return -1; + } +} + +static int +nv50_surface_copy_prep(struct nouveau_context *nv, + struct pipe_surface *dst, struct pipe_surface *src) +{ + int surf_format; + + assert(src->cpp == dst->cpp); + + surf_format = nv50_format(dst->cpp); + assert(surf_format >= 0); + + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); + OUT_RING (0); + OUT_RING (0); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + + BEGIN_RING(Nv2D, NV50_2D_SRC_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_SRC_PITCH, 5); + OUT_RING (src->pitch * src->cpp); + OUT_RING (src->pitch); + OUT_RING (src->height); + OUT_RELOCh(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + + return 0; +} + +static void +nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + BEGIN_RING(Nv2D, 0x0110, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, NV50_2D_BLIT_DST_X, 12); + OUT_RING (dx); + OUT_RING (dy); + OUT_RING (w); + OUT_RING (h); + OUT_RING (0); + OUT_RING (1); + OUT_RING (0); + OUT_RING (1); + OUT_RING (0); + OUT_RING (sx); + OUT_RING (0); + OUT_RING (sy); +} + +static void +nv50_surface_copy_done(struct nouveau_context *nv) +{ + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(Nv2D, 0x104, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); +} + +static int +nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + int surf_format, rect_format; + + surf_format = nv50_format(dst->cpp); + if (surf_format < 0) + return 1; + + rect_format = nv50_format(dst->cpp); + if (rect_format < 0) + return 1; + + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY1, 1); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); + OUT_RING (0); + OUT_RING (0); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + + BEGIN_RING(Nv2D, 0x0580, 3); + OUT_RING (4); + OUT_RING (rect_format); + OUT_RING (value); + + BEGIN_RING(Nv2D, NV50_2D_RECT_X1, 4); + OUT_RING (dx); + OUT_RING (dy); + OUT_RING (dx + w); + OUT_RING (dy + h); + + nouveau_notifier_reset(nv->sync_notifier, 0); + BEGIN_RING(Nv2D, 0x104, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, 0x100, 1); + OUT_RING (0); + FIRE_RING(); + nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); + + return 0; +} + +static int +nv50_surface_data(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, const void *src, unsigned src_pitch, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + NOUVEAU_ERR("unimplemented!!\n"); + return 0; +} + +int +nouveau_surface_init_nv50(struct nouveau_context *nv) +{ + int ret; + + ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, NV50_2D, + &nv->Nv2D); + if (ret) + return ret; + + BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RING (nv->channel->vram->handle); + OUT_RING (nv->channel->vram->handle); + BEGIN_RING(Nv2D, NV50_2D_OPERATION, 1); + OUT_RING (NV50_2D_OPERATION_SRCCOPY); + + nv->surface_copy_prep = nv50_surface_copy_prep; + nv->surface_copy = nv50_surface_copy; + nv->surface_copy_done = nv50_surface_copy_done; + nv->surface_fill = nv50_surface_fill; + nv->surface_data = nv50_surface_data; + return 0; +} + diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 40e4bfae54e..a274e23af5f 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -53,15 +53,14 @@ struct nouveau_winsys { int (*notifier_wait)(struct nouveau_notifier *, int id, int status, int timeout); - int (*region_copy)(struct nouveau_winsys *, struct pipe_region *, - unsigned, unsigned, unsigned, struct pipe_region *, - unsigned, unsigned, unsigned, unsigned, unsigned); - int (*region_fill)(struct nouveau_winsys *, struct pipe_region *, - unsigned, unsigned, unsigned, unsigned, unsigned, - unsigned); - int (*region_data)(struct nouveau_winsys *, struct pipe_region *, - unsigned, unsigned, unsigned, const void *, - unsigned, unsigned, unsigned, unsigned, unsigned); + int (*surface_copy)(struct nouveau_winsys *, struct pipe_surface *, + unsigned, unsigned, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned); + int (*surface_fill)(struct nouveau_winsys *, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned, unsigned); + int (*surface_data)(struct nouveau_winsys *, struct pipe_surface *, + unsigned, unsigned, const void *, unsigned, + unsigned, unsigned, unsigned, unsigned); }; extern struct pipe_context * diff --git a/src/mesa/pipe/nv40/Makefile b/src/mesa/pipe/nv40/Makefile index 90c8542da45..9818889ee32 100644 --- a/src/mesa/pipe/nv40/Makefile +++ b/src/mesa/pipe/nv40/Makefile @@ -10,7 +10,6 @@ DRIVER_SOURCES = \ nv40_fragprog.c \ nv40_miptree.c \ nv40_query.c \ - nv40_region.c \ nv40_state.c \ nv40_state_emit.c \ nv40_state_tex.c \ diff --git a/src/mesa/pipe/nv40/nv40_clear.c b/src/mesa/pipe/nv40/nv40_clear.c index 7890107c957..380a2a642fc 100644 --- a/src/mesa/pipe/nv40/nv40_clear.c +++ b/src/mesa/pipe/nv40/nv40_clear.c @@ -10,6 +10,5 @@ void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { - pipe->region_fill(pipe, ps->region, 0, 0, 0, ps->width, ps->height, - clearValue); + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); } diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 8706ed1c07c..7b77c70e81d 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -10,9 +10,9 @@ static boolean nv40_is_format_supported(struct pipe_context *pipe, uint format) { switch (format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_U_R5_G6_B5: - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: return TRUE; default: break; @@ -267,17 +267,12 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, nv40->pipe.draw_elements = nv40_draw_elements; nv40->pipe.clear = nv40_clear; - nv40->pipe.begin_query = nv40_query_begin; - nv40->pipe.end_query = nv40_query_end; - nv40->pipe.wait_query = nv40_query_wait; - - nv40->pipe.mipmap_tree_layout = nv40_miptree_layout; - nv40->pipe.flush = nv40_flush; - nv40_init_region_functions(nv40); + nv40_init_query_functions(nv40); nv40_init_surface_functions(nv40); nv40_init_state_functions(nv40); + nv40_init_miptree_functions(nv40); nv40->draw = draw_create(); assert(nv40->draw); diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 65a197ae2f1..3a1d3f076c6 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -41,7 +41,7 @@ struct nv40_context { uint32_t dirty; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct pipe_mipmap_tree *tex_miptree[PIPE_MAX_SAMPLERS]; + struct pipe_texture *tex_miptree[PIPE_MAX_SAMPLERS]; uint32_t tex_dirty; struct { @@ -66,17 +66,14 @@ struct nv40_context { }; -extern void nv40_init_region_functions(struct nv40_context *nv40); -extern void nv40_init_surface_functions(struct nv40_context *nv40); extern void nv40_init_state_functions(struct nv40_context *nv40); +extern void nv40_init_surface_functions(struct nv40_context *nv40); +extern void nv40_init_miptree_functions(struct nv40_context *nv40); +extern void nv40_init_query_functions(struct nv40_context *nv40); /* nv40_draw.c */ extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); -/* nv40_miptree.c */ -extern boolean nv40_miptree_layout(struct pipe_context *, - struct pipe_mipmap_tree *); - /* nv40_vertprog.c */ extern void nv40_vertprog_translate(struct nv40_context *, struct nv40_vertex_program *); @@ -107,9 +104,4 @@ extern void nv40_vbo_arrays_update(struct nv40_context *nv40); extern void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue); -/* nv40_query.c */ -extern void nv40_query_begin(struct pipe_context *, struct pipe_query_object *); -extern void nv40_query_end(struct pipe_context *, struct pipe_query_object *); -extern void nv40_query_wait(struct pipe_context *, struct pipe_query_object *); - #endif diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index f5c4206f402..45711f064d6 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -4,63 +4,105 @@ #include "nv40_context.h" -boolean -nv40_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) +static void +nv40_miptree_layout(struct nv40_miptree *nv40mt) { + struct pipe_texture *pt = &nv40mt->base; boolean swizzled = FALSE; - uint width = mt->width0, height = mt->height0, depth = mt->depth0; - uint offset; + uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint offset = 0; int nr_faces, l, f; - mt->pitch = mt->width0; - mt->total_height = 0; - - if (mt->target == PIPE_TEXTURE_CUBE) { + if (pt->target == PIPE_TEXTURE_CUBE) { nr_faces = 6; } else - if (mt->target == PIPE_TEXTURE_3D) { - nr_faces = mt->depth0; + if (pt->target == PIPE_TEXTURE_3D) { + nr_faces = pt->depth[0]; } else { nr_faces = 1; } + + for (l = pt->first_level; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->depth[l] = depth; - for (l = mt->first_level; l <= mt->last_level; l++) { - mt->level[l].width = width; - mt->level[l].height = height; - mt->level[l].depth = depth; - mt->level[l].level_offset = 0; + if (swizzled) + nv40mt->level[l].pitch = pt->width[l] * pt->cpp; + else + nv40mt->level[l].pitch = pt->width[0] * pt->cpp; + nv40mt->level[l].pitch = (nv40mt->level[l].pitch + 63) & ~63; - mt->level[l].nr_images = nr_faces; - mt->level[l].image_offset = malloc(nr_faces * sizeof(unsigned)); - for (f = 0; f < nr_faces; f++) - mt->total_height += height; + nv40mt->level[l].image_offset = + calloc(nr_faces, sizeof(unsigned)); width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); depth = MAX2(1, depth >> 1); + } - offset = 0; for (f = 0; f < nr_faces; f++) { - for (l = mt->first_level; l <= mt->last_level; l++) { - if (f == 0) { - mt->level[l].level_offset = offset; - } - - uint pitch; - - if (swizzled) - pitch = mt->level[l].width * mt->cpp; - else - pitch = mt->width0 * mt->cpp; - pitch = (pitch + 63) & ~63; - - mt->level[l].image_offset[f] = - (offset - mt->level[l].level_offset) / mt->cpp; - offset += pitch * mt->level[l].height; + for (l = pt->first_level; l <= pt->last_level; l++) { + nv40mt->level[l].image_offset[f] = offset; + offset += nv40mt->level[l].pitch * pt->height[l]; + } + } + + nv40mt->total_size = offset; +} + +static void +nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_texture *mt = *pt; + struct pipe_winsys *ws = pipe->winsys; + struct nv40_miptree *nv40mt; + + *pt = NULL; + + nv40mt = calloc(1, sizeof(struct nv40_miptree)); + if (!nv40mt) + return; + + memcpy(&nv40mt->base, mt, sizeof(struct pipe_texture)); + nv40_miptree_layout(nv40mt); + + nv40mt->buffer = ws->buffer_create(ws, PIPE_SURFACE_FLAG_TEXTURE); + if (!nv40mt->buffer) { + free(nv40mt); + return; + } + + ws->buffer_data(ws, nv40mt->buffer, nv40mt->total_size, NULL, + PIPE_BUFFER_USAGE_PIXEL); + *pt = &nv40mt->base; +} + +static void +nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv40_miptree *nv40mt = (struct nv40_miptree *)mt; + int l; + + ws->buffer_reference(ws, &nv40mt->buffer, NULL); + for (l = 0; l < PIPE_MAX_TEXTURE_LEVELS; l++) { + if (nv40mt->level[l].image_offset) + free(nv40mt->level[l].image_offset); } + free(nv40mt); } +} - return TRUE; +void +nv40_init_miptree_functions(struct nv40_context *nv40) +{ + nv40->pipe.texture_create = nv40_miptree_create; + nv40->pipe.texture_release = nv40_miptree_release; } diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index efd81e66408..bcd6fe0cf49 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -16,7 +16,7 @@ nv40_query_object_find(struct nv40_context *nv40, struct pipe_query_object *q) return -1; } -void +static void nv40_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) { struct nv40_context *nv40 = (struct nv40_context *)pipe; @@ -53,7 +53,24 @@ nv40_query_update(struct pipe_context *pipe, struct pipe_query_object *q) } } -void +static void +nv40_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) +{ + nv40_query_update(pipe, q); + if (!q->ready) { + struct nv40_context *nv40 = (struct nv40_context *)pipe; + int id; + + id = nv40_query_object_find(nv40, q); + assert(id >= 0); + + nv40->nvws->notifier_wait(nv40->query, id, 0, 0); + nv40_query_update(pipe, q); + assert(q->ready); + } +} + +static void nv40_query_end(struct pipe_context *pipe, struct pipe_query_object *q) { struct nv40_context *nv40 = (struct nv40_context *)pipe; @@ -80,19 +97,9 @@ nv40_query_end(struct pipe_context *pipe, struct pipe_query_object *q) } void -nv40_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) +nv40_init_query_functions(struct nv40_context *nv40) { - nv40_query_update(pipe, q); - if (!q->ready) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; - int id; - - id = nv40_query_object_find(nv40, q); - assert(id >= 0); - - nv40->nvws->notifier_wait(nv40->query, id, 0, 0); - nv40_query_update(pipe, q); - assert(q->ready); - } + nv40->pipe.begin_query = nv40_query_begin; + nv40->pipe.end_query = nv40_query_end; + nv40->pipe.wait_query = nv40_query_wait; } - diff --git a/src/mesa/pipe/nv40/nv40_region.c b/src/mesa/pipe/nv40/nv40_region.c deleted file mode 100644 index 572b82bca3a..00000000000 --- a/src/mesa/pipe/nv40/nv40_region.c +++ /dev/null @@ -1,85 +0,0 @@ -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" - -#include "nv40_context.h" -#include "nv40_dma.h" - -static ubyte * -nv40_region_map(struct pipe_context *pipe, struct pipe_region *region) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct pipe_winsys *ws = nv40->pipe.winsys; - - if (!region->map_refcount++) { - region->map = ws->buffer_map(ws, region->buffer, - PIPE_BUFFER_FLAG_WRITE | - PIPE_BUFFER_FLAG_READ); - } - - return region->map; -} - -static void -nv40_region_unmap(struct pipe_context *pipe, struct pipe_region *region) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct pipe_winsys *ws = nv40->pipe.winsys; - - if (!--region->map_refcount) { - ws->buffer_unmap(ws, region->buffer); - region->map = NULL; - } -} - -static void -nv40_region_data(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - const void *src, unsigned src_pitch, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nouveau_winsys *nvws = nv40->nvws; - - nvws->region_data(nvws, dst, dst_offset, dstx, dsty, - src, src_pitch, srcx, srcy, width, height); -} - - -static void -nv40_region_copy(struct pipe_context *pipe, struct pipe_region *dst, - unsigned dst_offset, unsigned dstx, unsigned dsty, - struct pipe_region *src, unsigned src_offset, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nouveau_winsys *nvws = nv40->nvws; - - nvws->region_copy(nvws, dst, dst_offset, dstx, dsty, - src, src_offset, srcx, srcy, width, height); -} - -static void -nv40_region_fill(struct pipe_context *pipe, - struct pipe_region *dst, unsigned dst_offset, - unsigned dstx, unsigned dsty, - unsigned width, unsigned height, unsigned value) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nouveau_winsys *nvws = nv40->nvws; - - nvws->region_fill(nvws, dst, dst_offset, dstx, dsty, - width, height, value); -} - -void -nv40_init_region_functions(struct nv40_context *nv40) -{ - nv40->pipe.region_map = nv40_region_map; - nv40->pipe.region_unmap = nv40_region_unmap; - nv40->pipe.region_data = nv40_region_data; - nv40->pipe.region_copy = nv40_region_copy; - nv40->pipe.region_fill = nv40_region_fill; -} - diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 937de7face1..e4915794830 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -549,7 +549,7 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct pipe_region *region[4], *zregion; + struct pipe_surface *rt[4], *zeta; uint32_t rt_enable, rt_format, w, h; int i, colour_format = 0, zeta_format = 0; @@ -567,7 +567,7 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, h = fb->cbufs[i]->height; colour_format = fb->cbufs[i]->format; rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); - region[i] = fb->cbufs[i]->region; + rt[i] = fb->cbufs[i]; } } @@ -585,7 +585,7 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } zeta_format = fb->zbuf->format; - zregion = fb->zbuf->region; + zeta = fb->zbuf; } if (fb->sbuf) { @@ -599,21 +599,21 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, if (zeta_format) { assert(fb->sbuf->format == zeta_format); - assert(fb->sbuf->region == zregion); + assert(fb->sbuf == zeta); } else { zeta_format = fb->sbuf->format; - zregion = fb->sbuf->region; + zeta = fb->sbuf; } } rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; switch (colour_format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: case 0: rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; break; - case PIPE_FORMAT_U_R5_G6_B5: + case PIPE_FORMAT_R5G6B5_UNORM: rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; break; default: @@ -621,13 +621,12 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } switch (zeta_format) { - case PIPE_FORMAT_U_Z16: + case PIPE_FORMAT_Z16_UNORM: rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; break; - case PIPE_FORMAT_Z24_S8: - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; - break; + case PIPE_FORMAT_Z24S8_UNORM: case 0: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; break; default: assert(0); @@ -635,51 +634,47 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(region[0]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(rt[0]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); - OUT_RING (region[0]->pitch * region[0]->cpp); - OUT_RELOCl(region[0]->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (rt[0]->pitch * rt[0]->cpp); + OUT_RELOCl(rt[0]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(region[1]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(rt[1]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); - OUT_RELOCl(region[1]->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RING (region[1]->pitch * region[1]->cpp); + OUT_RELOCl(rt[1]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(region[2]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(rt[2]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(region[2]->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(rt[2]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); - OUT_RING (region[2]->pitch * region[2]->cpp); + OUT_RING (rt[2]->pitch * rt[2]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(region[3]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(rt[3]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(region[3]->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(rt[3]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); - OUT_RING (region[3]->pitch * region[3]->cpp); + OUT_RING (rt[3]->pitch * rt[3]->cpp); } if (zeta_format) { BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(zregion->buffer, + OUT_RELOCo(zeta->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(zregion->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); + OUT_RELOCl(zeta->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | + NOUVEAU_BO_RD); BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); - OUT_RING (zregion->pitch * zregion->cpp); + OUT_RING (zeta->pitch * zeta->cpp); } BEGIN_RING(curie, NV40TCL_RT_ENABLE, 1); @@ -725,7 +720,7 @@ nv40_set_scissor_state(struct pipe_context *pipe, static void nv40_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_mipmap_tree *miptree) + struct pipe_texture *miptree) { struct nv40_context *nv40 = (struct nv40_context *)pipe; diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index c6d22ceaa86..80c76cd25b6 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -124,4 +124,16 @@ struct nv40_depth_stencil_state { } stencil; }; +struct nv40_miptree { + struct pipe_texture base; + + struct pipe_buffer_handle *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + #endif diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 6902b41ba71..30462a2a3d6 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -21,18 +21,18 @@ struct nv40_texture_format { static struct nv40_texture_format nv40_texture_formats[] = { - _(U_A8_R8_G8_B8, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), - _(U_A1_R5_G5_B5, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), - _(U_A4_R4_G4_B4, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), - _(U_R5_G6_B5 , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), - _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, Z, W, X, Y), -// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), + _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), + _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, Z, W, X, Y), +// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), {}, }; @@ -50,30 +50,32 @@ nv40_tex_format(uint pipe_format) return NULL; } + static void nv40_tex_unit_enable(struct nv40_context *nv40, int unit) { struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; - struct pipe_mipmap_tree *mt = nv40->tex_miptree[unit]; + struct pipe_texture *pt = nv40->tex_miptree[unit]; + struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; struct nv40_texture_format *tf; uint32_t txf, txs, txp; int swizzled = 0; /*XXX: implement in region code? */ - tf = nv40_tex_format(mt->format); + tf = nv40_tex_format(pt->format); if (!tf || !tf->defined) { - NOUVEAU_ERR("Unsupported texture format: 0x%x\n", mt->format); + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); return; } txf = ps->fmt; txf |= tf->format | 0x8000; - txf |= ((mt->last_level - mt->first_level + 1) << + txf |= ((pt->last_level - pt->first_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); if (1) /* XXX */ txf |= NV40TCL_TEX_FORMAT_NO_BORDER; - switch (mt->target) { + switch (pt->target) { case PIPE_TEXTURE_CUBE: txf |= NV40TCL_TEX_FORMAT_CUBIC; /* fall-through */ @@ -87,25 +89,25 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) txf |= NV40TCL_TEX_FORMAT_DIMS_1D; break; default: - NOUVEAU_ERR("Unknown target %d\n", mt->target); + NOUVEAU_ERR("Unknown target %d\n", pt->target); return; } if (swizzled) { txp = 0; } else { - txp = mt->pitch * mt->cpp; + txp = nv40mt->level[0].pitch; txf |= NV40TCL_TEX_FORMAT_LINEAR; } txs = tf->swizzle; - if (mt->format == PIPE_FORMAT_U_A8_L8) + if (pt->format == PIPE_FORMAT_U_A8_L8) txs |= (1<<16); /*nfi*/ BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); - OUT_RELOCl(mt->region->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + OUT_RELOCl(nv40mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(mt->region->buffer, txf, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + OUT_RELOCd(nv40mt->buffer, txf, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); OUT_RING (ps->wrap); @@ -113,10 +115,10 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) (0x00078000) /* mipmap related? */); OUT_RING (txs); OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); - OUT_RING ((mt->width0 << NV40TCL_TEX_SIZE0_W_SHIFT) | mt->height0); + OUT_RING ((pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); OUT_RING (ps->bcol); BEGIN_RING(curie, NV40TCL_TEX_SIZE1(unit), 1); - OUT_RING ((mt->depth0 << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); + OUT_RING ((pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); } void diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index 5292b5171ff..b4562f26e65 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -56,15 +56,15 @@ nv40_get_tile_rgba(struct pipe_context *pipe, uint x, uint y, uint w, uint h, float *p) { const unsigned *src - = ((const unsigned *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + = ((const unsigned *) (ps->map + ps->offset)) + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; CLIP_TILE; switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: for (i = 0; i < h; i++) { float *pRow = p; for (j = 0; j < w; j++) { @@ -75,11 +75,11 @@ nv40_get_tile_rgba(struct pipe_context *pipe, pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: { const float scale = 1.0 / (float) 0xffffff; for (i = 0; i < h; i++) { @@ -92,7 +92,7 @@ nv40_get_tile_rgba(struct pipe_context *pipe, pRow[3] = ((pixel & 0xffffff) >> 8) * scale; pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -122,13 +122,13 @@ nv40_get_tile(struct pipe_context *pipe, uint x, uint y, uint w, uint h, void *p, int dst_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const uint w0 = w; const ubyte *pSrc; ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); CLIP_TILE; @@ -136,13 +136,13 @@ nv40_get_tile(struct pipe_context *pipe, dst_stride = w0 * cpp; } - pSrc = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pSrc = ps->map + ps->offset + (y * ps->pitch + x) * cpp; pDest = (ubyte *) p; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); pDest += dst_stride; - pSrc += ps->region->pitch * cpp; + pSrc += ps->pitch * cpp; } } @@ -156,13 +156,13 @@ nv40_put_tile(struct pipe_context *pipe, uint x, uint y, uint w, uint h, const void *p, int src_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const uint w0 = w; const ubyte *pSrc; ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); CLIP_TILE; @@ -171,52 +171,81 @@ nv40_put_tile(struct pipe_context *pipe, } pSrc = (const ubyte *) p; - pDest = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pDest = ps->map + ps->offset + (y * ps->pitch + x) * cpp; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); - pDest += ps->region->pitch * cpp; + pDest += ps->pitch * cpp; pSrc += src_stride; } } -/* - * XXX note: same as code in sp_surface.c - */ static struct pipe_surface * nv40_get_tex_surface(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt, + struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { - struct pipe_surface *ps; - unsigned offset; /* in bytes */ + struct pipe_winsys *ws = pipe->winsys; + struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; + struct pipe_surface *ps; - offset = mt->level[level].level_offset; + ps = ws->surface_alloc(ws, pt->format); + if (!ps) + return NULL; + ws->buffer_reference(ws, &ps->buffer, nv40mt->buffer); + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv40mt->level[level].pitch / ps->cpp; - if (mt->target == PIPE_TEXTURE_CUBE) { - offset += mt->level[level].image_offset[face] * mt->cpp; - } - else if (mt->target == PIPE_TEXTURE_3D) { - offset += mt->level[level].image_offset[zslice] * mt->cpp; - } - else { - assert(face == 0); - assert(zslice == 0); - } + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv40mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv40mt->level[level].image_offset[zslice]; + } else { + ps->offset = nv40mt->level[level].image_offset[0]; + } - ps = pipe->winsys->surface_alloc(pipe->winsys, mt->format); - if (ps) { - assert(ps->format); - assert(ps->refcount); - pipe_region_reference(&ps->region, mt->region); - ps->width = mt->level[level].width; - ps->height = mt->level[level].height; - ps->offset = offset; - } - return ps; + return ps; } +static void +nv40_surface_data(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, const void *src, + unsigned src_stride, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->surface_data(nvws, dest, destx, desty, src, src_stride, + srcx, srcy, width, height); +} + +static void +nv40_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, struct pipe_surface *src, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} void nv40_init_surface_functions(struct nv40_context *nv40) @@ -226,4 +255,7 @@ nv40_init_surface_functions(struct nv40_context *nv40) nv40->pipe.put_tile = nv40_put_tile; nv40->pipe.get_tile_rgba = nv40_get_tile_rgba; nv40->pipe.put_tile_rgba = nv40_put_tile_rgba; + nv40->pipe.surface_data = nv40_surface_data; + nv40->pipe.surface_copy = nv40_surface_copy; + nv40->pipe.surface_fill = nv40_surface_fill; } diff --git a/src/mesa/pipe/nv50/Makefile b/src/mesa/pipe/nv50/Makefile index af3fa0e7ecd..d3d011b14b7 100644 --- a/src/mesa/pipe/nv50/Makefile +++ b/src/mesa/pipe/nv50/Makefile @@ -9,7 +9,6 @@ DRIVER_SOURCES = \ nv50_draw.c \ nv50_miptree.c \ nv50_query.c \ - nv50_region.c \ nv50_state.c \ nv50_surface.c \ nv50_vbo.c diff --git a/src/mesa/pipe/nv50/nv50_clear.c b/src/mesa/pipe/nv50/nv50_clear.c index 85af1af78dd..2b453a49d15 100644 --- a/src/mesa/pipe/nv50/nv50_clear.c +++ b/src/mesa/pipe/nv50/nv50_clear.c @@ -10,6 +10,5 @@ void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { - pipe->region_fill(pipe, ps->region, 0, 0, 0, ps->width, ps->height, - clearValue); + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); } diff --git a/src/mesa/pipe/nv50/nv50_context.c b/src/mesa/pipe/nv50/nv50_context.c index 6fceb13174a..25ead1b5646 100644 --- a/src/mesa/pipe/nv50/nv50_context.c +++ b/src/mesa/pipe/nv50/nv50_context.c @@ -10,8 +10,8 @@ static boolean nv50_is_format_supported(struct pipe_context *pipe, uint format) { switch (format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: return TRUE; default: break; @@ -193,17 +193,12 @@ nv50_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, nv50->pipe.draw_elements = nv50_draw_elements; nv50->pipe.clear = nv50_clear; - nv50->pipe.begin_query = nv50_query_begin; - nv50->pipe.end_query = nv50_query_end; - nv50->pipe.wait_query = nv50_query_wait; - - nv50->pipe.mipmap_tree_layout = nv50_miptree_layout; - nv50->pipe.flush = nv50_flush; - nv50_init_region_functions(nv50); + nv50_init_miptree_functions(nv50); nv50_init_surface_functions(nv50); nv50_init_state_functions(nv50); + nv50_init_query_functions(nv50); nv50->draw = draw_create(); assert(nv50->draw); diff --git a/src/mesa/pipe/nv50/nv50_context.h b/src/mesa/pipe/nv50/nv50_context.h index f7fa48553d0..5be4e5cb1be 100644 --- a/src/mesa/pipe/nv50/nv50_context.h +++ b/src/mesa/pipe/nv50/nv50_context.h @@ -30,16 +30,14 @@ struct nv50_context { }; -extern void nv50_init_region_functions(struct nv50_context *nv50); +extern void nv50_init_miptree_functions(struct nv50_context *nv50); extern void nv50_init_surface_functions(struct nv50_context *nv50); extern void nv50_init_state_functions(struct nv50_context *nv50); +extern void nv50_init_query_functions(struct nv50_context *nv50); /* nv50_draw.c */ extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50); -/* nv50_miptree.c */ -extern boolean nv50_miptree_layout(struct pipe_context *, - struct pipe_mipmap_tree *); /* nv50_vbo.c */ extern boolean nv50_draw_arrays(struct pipe_context *, unsigned mode, unsigned start, unsigned count); @@ -53,10 +51,4 @@ extern boolean nv50_draw_elements(struct pipe_context *pipe, extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue); -/* nv50_query.c */ -extern void nv50_query_begin(struct pipe_context *, struct pipe_query_object *); -extern void nv50_query_end(struct pipe_context *, struct pipe_query_object *); -extern void nv50_query_wait(struct pipe_context *, struct pipe_query_object *); - - #endif diff --git a/src/mesa/pipe/nv50/nv50_miptree.c b/src/mesa/pipe/nv50/nv50_miptree.c index cf0e530f317..51442d64f3e 100644 --- a/src/mesa/pipe/nv50/nv50_miptree.c +++ b/src/mesa/pipe/nv50/nv50_miptree.c @@ -4,10 +4,21 @@ #include "nv50_context.h" -boolean -nv50_miptree_layout(struct pipe_context *pipe, struct pipe_mipmap_tree *mt) +static void +nv50_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) { NOUVEAU_ERR("unimplemented\n"); - return TRUE; } +static void +nv50_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +void +nv50_init_miptree_functions(struct nv50_context *nv50) +{ + nv50->pipe.texture_create = nv50_miptree_create; + nv50->pipe.texture_release = nv50_miptree_release; +} diff --git a/src/mesa/pipe/nv50/nv50_query.c b/src/mesa/pipe/nv50/nv50_query.c index 1ce48943b99..51cf6804d7c 100644 --- a/src/mesa/pipe/nv50/nv50_query.c +++ b/src/mesa/pipe/nv50/nv50_query.c @@ -3,21 +3,28 @@ #include "nv50_context.h" #include "nv50_dma.h" -void +static void nv50_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) { NOUVEAU_ERR("unimplemented\n"); } -void +static void nv50_query_end(struct pipe_context *pipe, struct pipe_query_object *q) { NOUVEAU_ERR("unimplemented\n"); } -void +static void nv50_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) { NOUVEAU_ERR("unimplemented\n"); } +void +nv50_init_query_functions(struct nv50_context *nv50) +{ + nv50->pipe.begin_query = nv50_query_begin; + nv50->pipe.end_query = nv50_query_end; + nv50->pipe.wait_query = nv50_query_wait; +} diff --git a/src/mesa/pipe/nv50/nv50_region.c b/src/mesa/pipe/nv50/nv50_region.c deleted file mode 100644 index d5a071c8b26..00000000000 --- a/src/mesa/pipe/nv50/nv50_region.c +++ /dev/null @@ -1,85 +0,0 @@ -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" - -#include "nv50_context.h" -#include "nv50_dma.h" - -static ubyte * -nv50_region_map(struct pipe_context *pipe, struct pipe_region *region) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct pipe_winsys *ws = nv50->pipe.winsys; - - if (!region->map_refcount++) { - region->map = ws->buffer_map(ws, region->buffer, - PIPE_BUFFER_FLAG_WRITE | - PIPE_BUFFER_FLAG_READ); - } - - return region->map; -} - -static void -nv50_region_unmap(struct pipe_context *pipe, struct pipe_region *region) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct pipe_winsys *ws = nv50->pipe.winsys; - - if (!--region->map_refcount) { - ws->buffer_unmap(ws, region->buffer); - region->map = NULL; - } -} - -static void -nv50_region_data(struct pipe_context *pipe, - struct pipe_region *dst, - unsigned dst_offset, - unsigned dstx, unsigned dsty, - const void *src, unsigned src_pitch, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - nvws->region_data(nvws, dst, dst_offset, dstx, dsty, - src, src_pitch, srcx, srcy, width, height); -} - - -static void -nv50_region_copy(struct pipe_context *pipe, struct pipe_region *dst, - unsigned dst_offset, unsigned dstx, unsigned dsty, - struct pipe_region *src, unsigned src_offset, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - nvws->region_copy(nvws, dst, dst_offset, dstx, dsty, - src, src_offset, srcx, srcy, width, height); -} - -static void -nv50_region_fill(struct pipe_context *pipe, - struct pipe_region *dst, unsigned dst_offset, - unsigned dstx, unsigned dsty, - unsigned width, unsigned height, unsigned value) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - nvws->region_fill(nvws, dst, dst_offset, dstx, dsty, - width, height, value); -} - -void -nv50_init_region_functions(struct nv50_context *nv50) -{ - nv50->pipe.region_map = nv50_region_map; - nv50->pipe.region_unmap = nv50_region_unmap; - nv50->pipe.region_data = nv50_region_data; - nv50->pipe.region_copy = nv50_region_copy; - nv50->pipe.region_fill = nv50_region_fill; -} - diff --git a/src/mesa/pipe/nv50/nv50_state.c b/src/mesa/pipe/nv50/nv50_state.c index 0674c30dc80..778b9f5c9ac 100644 --- a/src/mesa/pipe/nv50/nv50_state.c +++ b/src/mesa/pipe/nv50/nv50_state.c @@ -169,7 +169,7 @@ nv50_set_scissor_state(struct pipe_context *pipe, static void nv50_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_mipmap_tree *miptree) + struct pipe_texture *pt) { } diff --git a/src/mesa/pipe/nv50/nv50_surface.c b/src/mesa/pipe/nv50/nv50_surface.c index 68013c0c1da..7e294cdfdfc 100644 --- a/src/mesa/pipe/nv50/nv50_surface.c +++ b/src/mesa/pipe/nv50/nv50_surface.c @@ -56,15 +56,15 @@ nv50_get_tile_rgba(struct pipe_context *pipe, uint x, uint y, uint w, uint h, float *p) { const unsigned *src - = ((const unsigned *) (ps->region->map + ps->offset)) - + y * ps->region->pitch + x; + = ((const unsigned *) (ps->map + ps->offset)) + + y * ps->pitch + x; unsigned i, j; unsigned w0 = w; CLIP_TILE; switch (ps->format) { - case PIPE_FORMAT_U_A8_R8_G8_B8: + case PIPE_FORMAT_A8R8G8B8_UNORM: for (i = 0; i < h; i++) { float *pRow = p; for (j = 0; j < w; j++) { @@ -75,11 +75,11 @@ nv50_get_tile_rgba(struct pipe_context *pipe, pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } break; - case PIPE_FORMAT_Z24_S8: + case PIPE_FORMAT_Z24S8_UNORM: { const float scale = 1.0 / (float) 0xffffff; for (i = 0; i < h; i++) { @@ -92,7 +92,7 @@ nv50_get_tile_rgba(struct pipe_context *pipe, pRow[3] = ((pixel & 0xffffff) >> 8) * scale; pRow += 4; } - src += ps->region->pitch; + src += ps->pitch; p += w0 * 4; } } @@ -122,13 +122,13 @@ nv50_get_tile(struct pipe_context *pipe, uint x, uint y, uint w, uint h, void *p, int dst_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const uint w0 = w; const ubyte *pSrc; ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); CLIP_TILE; @@ -136,13 +136,13 @@ nv50_get_tile(struct pipe_context *pipe, dst_stride = w0 * cpp; } - pSrc = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pSrc = ps->map + ps->offset + (y * ps->pitch + x) * cpp; pDest = (ubyte *) p; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); pDest += dst_stride; - pSrc += ps->region->pitch * cpp; + pSrc += ps->pitch * cpp; } } @@ -156,13 +156,13 @@ nv50_put_tile(struct pipe_context *pipe, uint x, uint y, uint w, uint h, const void *p, int src_stride) { - const uint cpp = ps->region->cpp; + const uint cpp = ps->cpp; const uint w0 = w; const ubyte *pSrc; ubyte *pDest; uint i; - assert(ps->region->map); + assert(ps->map); CLIP_TILE; @@ -171,52 +171,60 @@ nv50_put_tile(struct pipe_context *pipe, } pSrc = (const ubyte *) p; - pDest = ps->region->map + ps->offset + (y * ps->region->pitch + x) * cpp; + pDest = ps->map + ps->offset + (y * ps->pitch + x) * cpp; for (i = 0; i < h; i++) { memcpy(pDest, pSrc, w0 * cpp); - pDest += ps->region->pitch * cpp; + pDest += ps->pitch * cpp; pSrc += src_stride; } } -/* - * XXX note: same as code in sp_surface.c - */ static struct pipe_surface * nv50_get_tex_surface(struct pipe_context *pipe, - struct pipe_mipmap_tree *mt, + struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { - struct pipe_surface *ps; - unsigned offset; /* in bytes */ + NOUVEAU_ERR("unimplemented\n"); + return NULL; +} - offset = mt->level[level].level_offset; +static void +nv50_surface_data(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, const void *src, + unsigned src_stride, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; - if (mt->target == PIPE_TEXTURE_CUBE) { - offset += mt->level[level].image_offset[face] * mt->cpp; - } - else if (mt->target == PIPE_TEXTURE_3D) { - offset += mt->level[level].image_offset[zslice] * mt->cpp; - } - else { - assert(face == 0); - assert(zslice == 0); - } + nvws->surface_data(nvws, dest, destx, desty, src, src_stride, + srcx, srcy, width, height); +} - ps = pipe->winsys->surface_alloc(pipe->winsys, mt->format); - if (ps) { - assert(ps->format); - assert(ps->refcount); - pipe_region_reference(&ps->region, mt->region); - ps->width = mt->level[level].width; - ps->height = mt->level[level].height; - ps->offset = offset; - } - return ps; +static void +nv50_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, struct pipe_surface *src, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); } +static void +nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} void nv50_init_surface_functions(struct nv50_context *nv50) @@ -226,4 +234,7 @@ nv50_init_surface_functions(struct nv50_context *nv50) nv50->pipe.put_tile = nv50_put_tile; nv50->pipe.get_tile_rgba = nv50_get_tile_rgba; nv50->pipe.put_tile_rgba = nv50_put_tile_rgba; + nv50->pipe.surface_data = nv50_surface_data; + nv50->pipe.surface_copy = nv50_surface_copy; + nv50->pipe.surface_fill = nv50_surface_fill; } -- cgit v1.2.3 From 62bbf8db873716c56ca089dbba59d1d3980563b5 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 9 Dec 2007 20:38:08 +1100 Subject: nouveau: use "indirect" push buffers --- src/mesa/drivers/dri/nouveau_winsys/Makefile | 2 + .../drivers/dri/nouveau_winsys/nouveau_channel.c | 1 + .../drivers/dri/nouveau_winsys/nouveau_context.h | 4 + src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 45 ++---- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 76 +++++++++ .../drivers/dri/nouveau_winsys/nouveau_fence.c | 135 ++++++++++++++++ .../drivers/dri/nouveau_winsys/nouveau_local.h | 18 +-- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 177 +++++++++++++++++++++ .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 34 ++-- src/mesa/pipe/nouveau/nouveau_resource.h | 37 +++++ src/mesa/pipe/nouveau/nouveau_winsys.h | 12 +- 11 files changed, 474 insertions(+), 67 deletions(-) create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c create mode 100644 src/mesa/pipe/nouveau/nouveau_resource.h diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index 9b8a33b725c..fe9a9668b11 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -17,9 +17,11 @@ DRIVER_SOURCES = \ nouveau_context.c \ nouveau_device.c \ nouveau_dma.c \ + nouveau_fence.c \ nouveau_grobj.c \ nouveau_lock.c \ nouveau_notifier.c \ + nouveau_pushbuf.c \ nouveau_screen.c \ nouveau_swapbuffers.c \ nouveau_winsys.c \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c index ef41020b727..64136775da0 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c @@ -91,6 +91,7 @@ nouveau_channel_alloc(struct nouveau_device *userdev, uint32_t fb_ctxdma, malloc(sizeof(struct nouveau_bo_reloc) * chan->max_relocs); nouveau_dma_channel_init(&chan->base); + nouveau_pushbuf_init(&chan->base); *userchan = &chan->base; return 0; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index 7033ff7ae96..cd59b7cc436 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -44,6 +44,7 @@ struct nouveau_context { struct nouveau_bo *reloc_head; /* Hardware context */ + uint32_t *pushbuf; struct nouveau_channel *channel; struct nouveau_notifier *sync_notifier; struct nouveau_grobj *NvNull; @@ -98,4 +99,7 @@ extern void UNLOCK_HARDWARE(struct nouveau_context *); extern int nouveau_surface_init_nv04(struct nouveau_context *); extern int nouveau_surface_init_nv50(struct nouveau_context *); +extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int); +extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *); + #endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 6e123c44734..6e32d739abc 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -141,35 +141,6 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) if (chan->dma.cur == chan->dma.put) return; - if (chan->num_relocs) { - nouveau_bo_validate(userchan); - - for (i = 0; i < chan->num_relocs; i++) { - struct nouveau_bo_reloc *r = &chan->relocs[i]; - uint32_t push; - - if (r->flags & NOUVEAU_BO_LOW) { - push = r->bo->base.offset + r->data; - } else - if (r->flags & NOUVEAU_BO_HIGH) { - push = (r->bo->base.offset + r->data) >> 32; - } else { - push = r->data; - } - - if (r->flags & NOUVEAU_BO_OR) { - if (r->bo->base.flags & NOUVEAU_BO_VRAM) - push |= r->vor; - else - push |= r->tor; - } - - *r->ptr = push; - } - - chan->num_relocs = 0; - } - #ifdef NOUVEAU_DMA_DEBUG if (chan->dma.push_free) { NOUVEAU_ERR("Packet incomplete: %d left\n", chan->dma.push_free); @@ -178,8 +149,20 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) #endif #ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF - for (i = chan->dma.put; i < chan->dma.cur; i++) - NOUVEAU_MSG("0x%08x\n", chan->pushbuf[i]); + for (i = chan->dma.put; i < chan->dma.cur; i++) { + NOUVEAU_MSG("0x%08x 0x%08x\n", (i<<2)+chan->dma.base, + chan->pushbuf[i]); + if ((chan->pushbuf[i] & 0xf0000000) == 0x20000000) { + int n = (((chan->pushbuf[i] & 0x0fffffff) - + chan->dma.base) / 4); + + do { + NOUVEAU_MSG("\t0x%08x 0x%08x\n", + (n<<2)+chan->dma.base, + chan->pushbuf[n]); + } while ((chan->pushbuf[n++]&0xf0000000) != 0x20000000); + } + } #endif put_offset = (chan->dma.cur << 2) + chan->dma.base; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index df98890f8e2..4f467406ddb 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -32,6 +32,7 @@ #include "pipe/nouveau/nouveau_grobj.h" #include "pipe/nouveau/nouveau_notifier.h" #include "pipe/nouveau/nouveau_bo.h" +#include "pipe/nouveau/nouveau_resource.h" struct nouveau_device_priv { struct nouveau_device base; @@ -59,6 +60,61 @@ nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v); extern int nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val); +struct nouveau_fence { + struct nouveau_channel *channel; +}; + +struct nouveau_fence_priv { + struct nouveau_fence base; + int refcount; + + struct nouveau_fence *next; + + uint32_t sequence; + int emitted; + int signalled; +}; +#define nouveau_fence(n) ((struct nouveau_fence_priv *)(n)) + +extern int +nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **); + +extern int +nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); + +extern void +nouveau_fence_del(struct nouveau_fence **); + +extern void +nouveau_fence_emit(struct nouveau_fence *); + +extern int +nouveau_fence_wait(struct nouveau_fence **); + +extern void +nouveau_fence_flush(struct nouveau_channel *); + +struct nouveau_pushbuf { + struct nouveau_channel *channel; + unsigned remaining; + uint32_t *cur; +}; + +struct nouveau_pushbuf_priv { + struct nouveau_pushbuf base; + struct nouveau_pushbuf *next; + + struct nouveau_resource *res; + struct nouveau_fence *fence; +}; +#define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) + +extern int +nouveau_pushbuf_init(struct nouveau_channel *); + +extern int +nouveau_pushbuf_flush(struct nouveau_channel *); + struct nouveau_channel_priv { struct nouveau_channel base; @@ -95,6 +151,14 @@ struct nouveau_channel_priv { struct nouveau_bo_reloc *relocs; int num_relocs; int max_relocs; + + struct nouveau_fence *fence_head; + struct nouveau_fence *fence_tail; + uint32_t fence_sequence; + + struct nouveau_resource *pb_heap; + struct nouveau_pushbuf *pb_head; + struct nouveau_pushbuf *pb_tail; }; #define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) @@ -148,6 +212,8 @@ nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, struct nouveau_bo_priv { struct nouveau_bo base; + struct nouveau_fence *fence; + struct drm_nouveau_mem_alloc drm; void *map; @@ -203,4 +269,14 @@ nouveau_bo_emit_reloc(struct nouveau_channel *chan, void *ptr, extern void nouveau_bo_validate(struct nouveau_channel *); +extern int +nouveau_resource_init(struct nouveau_resource **heap, int size); + +extern int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **); + +extern void +nouveau_resource_free(struct nouveau_resource **); + #endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c new file mode 100644 index 00000000000..842b38f913b --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c @@ -0,0 +1,135 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +int +nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!chan || !fence || *fence) + return -EINVAL; + + nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); + if (!nvfence) + return -ENOMEM; + nvfence->base.channel = chan; + nvfence->refcount = 1; + + *fence = &nvfence->base; + return 0; +} + +int +nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!ref || !fence || *fence) + return -EINVAL; + nvfence = nouveau_fence(ref); + nvfence->refcount++; + + *fence = &nvfence->base; + return 0; +} + +void +nouveau_fence_del(struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence || !*fence) + return; + nvfence = nouveau_fence(*fence); + *fence = NULL; + + if (--nvfence->refcount <= 0) { + if (nvfence->emitted && !nvfence->signalled) + nouveau_fence_wait((void *)&nvfence); + free(nvfence); + } +} + +void +nouveau_fence_emit(struct nouveau_fence *fence) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); + struct nouveau_fence_priv *nvfence = nouveau_fence(fence); + + nvfence->emitted = 1; + nvfence->sequence = ++nvchan->fence_sequence; + if (nvfence->sequence == 0xffffffff) + NOUVEAU_ERR("AII wrap unhandled\n"); + + BEGIN_RING_CH(&nvchan->base, nvchan->subchannel[0].grobj, 0x50, 1); + OUT_RING_CH (&nvchan->base, nvfence->sequence); + + if (nvchan->fence_tail) { + nouveau_fence(nvchan->fence_tail)->next = fence; + } else { + nvchan->fence_head = fence; + } + nvchan->fence_tail = fence; +} + +void +nouveau_fence_flush(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_fence_priv *nvfence = nouveau_fence(nvchan->fence_head); + uint32_t sequence = *nvchan->ref_cnt; + + while (nvchan->fence_head && nvfence->sequence <= sequence) { + nvfence->signalled = 1; + + nvchan->fence_head = nvfence->next; + if (nvchan->fence_head == NULL) + nvchan->fence_tail = NULL; + nvfence = nouveau_fence(nvchan->fence_head); + } +} + +int +nouveau_fence_wait(struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence || !*fence) + return -EINVAL; + nvfence = nouveau_fence(*fence); + + if (nvfence->emitted) { + while (!nvfence->signalled) + nouveau_fence_flush(nvfence->base.channel); + } + nouveau_fence_del(fence); + + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index 4ada5d014bf..2534a8ea6f9 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -26,15 +26,16 @@ /* Push buffer access macros */ #define BEGIN_RING(obj,mthd,size) do { \ - BEGIN_RING_CH(nv->channel, nv->obj, (mthd), (size)); \ + nv->pushbuf = nouveau_pipe_dma_beginp(nv->obj, (mthd), (size)); \ } while(0) #define OUT_RING(data) do { \ - OUT_RING_CH(nv->channel, (data)); \ + (*nv->pushbuf++) = (data); \ } while(0) #define OUT_RINGp(src,size) do { \ - OUT_RINGp_CH(nv->channel, (src), (size)); \ + memcpy(nv->pushbuf, (src), (size)<<2); \ + nv->pushbuf += (size); \ } while(0) #define OUT_RINGf(data) do { \ @@ -43,18 +44,13 @@ OUT_RING(c.u); \ } while(0) -#define WAIT_RING(size) do { \ - WAIT_RING_CH(nv->channel, (size)); \ -} while(0) - #define FIRE_RING() do { \ - FIRE_RING_CH(nv->channel); \ + nouveau_pipe_dma_kickoff(nv->channel); \ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ - struct nouveau_channel_priv *chan = nouveau_channel(nv->channel); \ - nouveau_bo_emit_reloc(nv->channel, &chan->pushbuf[chan->dma.cur], \ - (void*)(bo), (data), (flags), (vor), (tor)); \ + nouveau_bo_emit_reloc(nv->channel, nv->pushbuf, (void*)(bo), (data), \ + (flags), (vor), (tor)); \ OUT_RING(0); \ } while(0) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c new file mode 100644 index 00000000000..75d41d6c147 --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -0,0 +1,177 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +int +nouveau_pushbuf_init(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + + if (!nvchan) + return -EINVAL; + + /* Everything except first 4KiB of the push buffer is managed by us */ + if (nouveau_resource_init(&nvchan->pb_heap, + nvchan->drm.cmdbuf_size - 4096)) + return -EINVAL; + + /* Shrink master ring to 4KiB */ + assert(nvchan->dma.cur <= (4096/4)); + nvchan->dma.max = (4096 / 4) - 2; + nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; + + return 0; +} + +/* This would be our TTM "superioctl" */ +int +nouveau_pushbuf_flush(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); + int i, ret; + + if (!nvpb) + goto out_realloc; + + if (nvpb->base.remaining == nvpb->res->size / 4) + return 0; + + ret = nouveau_fence_new(chan, &nvpb->fence); + if (ret) + return ret; + + /* Validate buffers */ + nouveau_bo_validate(chan); + + /* Apply relocations */ + if (nvchan->num_relocs) { + for (i = 0; i < nvchan->num_relocs; i++) { + struct nouveau_bo_reloc *r = &nvchan->relocs[i]; + uint32_t push; + + if (r->flags & NOUVEAU_BO_LOW) { + push = r->bo->base.offset + r->data; + } else + if (r->flags & NOUVEAU_BO_HIGH) { + push = (r->bo->base.offset + r->data) >> 32; + } else { + push = r->data; + } + + if (r->flags & NOUVEAU_BO_OR) { + if (r->bo->base.flags & NOUVEAU_BO_VRAM) + push |= r->vor; + else + push |= r->tor; + } + + *r->ptr = push; + } + + nvchan->num_relocs = 0; + } + + /* Emit JMP to indirect pushbuf */ + if (nvchan->dma.free < 1) + WAIT_RING_CH(chan, 1); + nvchan->dma.free -= 1; + OUT_RING_CH(chan, 0x20000000 | (nvpb->res->start + 4096)); + + /* Add JMP back to master pushbuf from indirect pushbuf */ + (*nvpb->base.cur++) = + 0x20000000 | ((nvchan->dma.cur << 2) + nvchan->dma.base); + + /* Fence */ + nouveau_fence_emit(nvpb->fence); + + /* Kickoff */ + FIRE_RING_CH(chan); + + /* Allocate space for next push buffer */ +out_realloc: + nvpb = calloc(1, sizeof(struct nouveau_pushbuf_priv)); + if (!nvpb) + return -ENOMEM; + + if (nouveau_resource_alloc(nvchan->pb_heap, 0x2000, NULL, &nvpb->res)) { + struct nouveau_pushbuf_priv *e; + int nr = 0; + + /* Update fences */ + nouveau_fence_flush(chan); + + /* Free any push buffers that have already been executed */ + e = nouveau_pushbuf(nvchan->pb_head); + while (e && e->fence) { + if (!e->fence || !nouveau_fence(e->fence)->signalled) + break; + nouveau_fence_del(&e->fence); + nouveau_resource_free(&e->res); + nr++; + + nvchan->pb_head = e->next; + if (nvchan->pb_head == NULL) + nvchan->pb_tail = NULL; + free(e); + e = nouveau_pushbuf(nvchan->pb_head); + } + + /* We didn't free any buffers above. As a last resort, busy + * wait on the oldest buffer becoming available. + */ + if (!nr) { + e = nouveau_pushbuf(nvchan->pb_head); + nouveau_fence_wait(&e->fence); + nouveau_resource_free(&e->res); + + nvchan->pb_head = e->next; + if (nvchan->pb_head == NULL) + nvchan->pb_tail = NULL; + free(e); + } + + if (nouveau_resource_alloc(nvchan->pb_heap, 0x2000, nvpb, + &nvpb->res)) + assert(0); + } + + nvpb->base.channel = chan; + nvpb->base.remaining = nvpb->res->size / 4; + nvpb->base.cur = &nvchan->pushbuf[(nvpb->res->start + 4096)/4]; + + if (nvchan->pb_tail) { + nouveau_pushbuf(nvchan->pb_tail)->next = & nvpb->base; + } else { + nvchan->pb_head = &nvpb->base; + } + nvchan->pb_tail = &nvpb->base; + + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index 6ef81ccb86a..77d9e710dbe 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -5,7 +5,7 @@ #include "pipe/nouveau/nouveau_winsys.h" -static int +int nouveau_resource_init(struct nouveau_resource **heap, int size) { struct nouveau_resource *r; @@ -20,7 +20,7 @@ nouveau_resource_init(struct nouveau_resource **heap, int size) return 0; } -static int +int nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, struct nouveau_resource **res) { @@ -58,7 +58,7 @@ nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, return 1; } -static void +void nouveau_resource_free(struct nouveau_resource **res) { struct nouveau_resource *r; @@ -108,25 +108,31 @@ nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, grclass, grobj); } -static uint32_t * +uint32_t * nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) { - struct nouveau_channel_priv *chan = nouveau_channel(grobj->channel); + struct nouveau_channel_priv *nvchan = nouveau_channel(grobj->channel); uint32_t *pushbuf; - BEGIN_RING_CH(&chan->base, grobj, mthd, size); - pushbuf = &chan->pushbuf[chan->dma.cur]; - chan->dma.cur += size; -#ifdef NOUVEAU_DMA_DEBUG - chan->dma.push_free -= size; -#endif + if (!nvchan->pb_tail || nvchan->pb_tail->remaining < (size + 1)) + nouveau_pushbuf_flush(grobj->channel); + + if (grobj->bound == NOUVEAU_GROBJ_UNBOUND) + nouveau_dma_subc_bind(grobj); + nvchan->subchannel[grobj->subc].seq = nvchan->subc_sequence++; + + pushbuf = nvchan->pb_tail->cur; + nvchan->pb_tail->cur += (size + 1); + nvchan->pb_tail->remaining -= (size + 1); + + (*pushbuf++) = ((grobj->subc << 13) | (size << 18) | mthd); return pushbuf; } -static void -nouveau_pipe_dma_kickoff(struct nouveau_channel *userchan) +void +nouveau_pipe_dma_kickoff(struct nouveau_channel *chan) { - FIRE_RING_CH(userchan); + nouveau_pushbuf_flush(chan); } static int diff --git a/src/mesa/pipe/nouveau/nouveau_resource.h b/src/mesa/pipe/nouveau/nouveau_resource.h new file mode 100644 index 00000000000..1af7961d301 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_resource.h @@ -0,0 +1,37 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_RESOURCE_H__ +#define __NOUVEAU_RESOURCE_H__ + +struct nouveau_resource { + struct nouveau_resource *prev; + struct nouveau_resource *next; + + int in_use; + void *priv; + + unsigned int start; + unsigned int size; +}; + +#endif diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index a274e23af5f..ce1744c2eec 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -10,17 +10,7 @@ #include "pipe/nouveau/nouveau_class.h" #include "pipe/nouveau/nouveau_grobj.h" #include "pipe/nouveau/nouveau_notifier.h" - -struct nouveau_resource { - struct nouveau_resource *prev; - struct nouveau_resource *next; - - boolean in_use; - void *priv; - - uint start; - uint size; -}; +#include "pipe/nouveau/nouveau_resource.h" struct nouveau_winsys { struct nouveau_context *nv; -- cgit v1.2.3 From f58f4e0f0ecc6c7079ef52f4a417707843507fab Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 9 Dec 2007 21:44:46 +1100 Subject: nouveau: relocs are per-pushbuf, not per-channel --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 90 ++++++---------------- .../drivers/dri/nouveau_winsys/nouveau_channel.c | 8 -- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 60 +++++++++------ .../drivers/dri/nouveau_winsys/nouveau_local.h | 4 +- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 87 ++++++++++++++++++--- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 2 +- src/mesa/pipe/nouveau/nouveau_winsys.h | 2 +- 7 files changed, 137 insertions(+), 116 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index f1981b97770..fd6d05b7fab 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -232,47 +232,6 @@ nouveau_bo_unmap(struct nouveau_bo *userbo) userbo->map = NULL; } -void -nouveau_bo_emit_reloc(struct nouveau_channel *userchan, void *ptr, - struct nouveau_bo *userbo, uint32_t data, uint32_t flags, - uint32_t vor, uint32_t tor) -{ - struct nouveau_channel_priv *chan = nouveau_channel(userchan); - struct nouveau_bo_priv *bo = nouveau_bo(userbo); - struct nouveau_bo_reloc *r; - int i, on_list = 0; - - for (i = 0; i < chan->nr_buffers; i++) { - if (chan->buffers[i].bo == bo) { - on_list = 1; - break; - } - } - - if (i >= 128) - return; - - if (on_list) { - chan->buffers[i].flags &= (flags | NOUVEAU_BO_RDWR); - chan->buffers[i].flags |= (flags & NOUVEAU_BO_RDWR); - } else { - chan->buffers[i].bo = bo; - chan->buffers[i].flags = flags; - chan->nr_buffers++; - } - - if (chan->num_relocs >= chan->max_relocs) - FIRE_RING_CH(userchan); - r = &chan->relocs[chan->num_relocs++]; - - r->ptr = ptr; - r->bo = bo; - r->data = data; - r->flags = flags; - r->vor = vor; - r->tor = tor; -} - static int nouveau_bo_upload(struct nouveau_bo_priv *bo) { @@ -280,35 +239,30 @@ nouveau_bo_upload(struct nouveau_bo_priv *bo) return 0; } -void -nouveau_bo_validate(struct nouveau_channel *userchan) +int +nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, + uint32_t flags) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); - int i; - - for (i = 0; i < chan->nr_buffers; i++) { - struct nouveau_bo_priv *bo = chan->buffers[i].bo; - - if (!bo->drm.size) { - nouveau_bo_realloc_gpu(bo, chan->buffers[i].flags, - bo->base.size); - nouveau_bo_upload(bo); - } else - if (bo->user || bo->base.map) - nouveau_bo_upload(bo); - - if (!bo->user && !bo->base.map) { - free(bo->sysmem); - bo->sysmem = NULL; - } + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + + if (!nvbo->drm.size) { + nouveau_bo_realloc_gpu(nvbo, flags, nvbo->base.size); + nouveau_bo_upload(nvbo); + } else + if (nvbo->user || nvbo->base.map) + nouveau_bo_upload(nvbo); + + if (!nvbo->user && !nvbo->base.map) { + free(nvbo->sysmem); + nvbo->sysmem = NULL; + } + nvbo->base.offset = nvbo->drm.offset; + if (nvbo->drm.flags & NOUVEAU_MEM_AGP) + nvbo->base.flags = NOUVEAU_BO_GART; + else + nvbo->base.flags = NOUVEAU_BO_VRAM; - bo->base.offset = bo->drm.offset; - if (bo->drm.flags & NOUVEAU_MEM_AGP) - bo->base.flags = NOUVEAU_BO_GART; - else - bo->base.flags = NOUVEAU_BO_VRAM; - } - chan->nr_buffers = 0; + return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c index 64136775da0..c4558e5573a 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c @@ -85,11 +85,6 @@ nouveau_channel_alloc(struct nouveau_device *userdev, uint32_t fb_ctxdma, return ret; } - chan->max_relocs = chan->drm.cmdbuf_size / 4; - chan->num_relocs = 0; - chan->relocs = - malloc(sizeof(struct nouveau_bo_reloc) * chan->max_relocs); - nouveau_dma_channel_init(&chan->base); nouveau_pushbuf_init(&chan->base); @@ -115,9 +110,6 @@ nouveau_channel_free(struct nouveau_channel **userchan) FIRE_RING_CH(&chan->base); - if (chan->relocs) - free(chan->relocs); - nouveau_grobj_free(&chan->base.vram); nouveau_grobj_free(&chan->base.gart); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 4f467406ddb..8af0b9d5cd3 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -94,6 +94,24 @@ nouveau_fence_wait(struct nouveau_fence **); extern void nouveau_fence_flush(struct nouveau_channel *); +struct nouveau_pushbuf_reloc { + uint64_t next; + uint64_t handle; + uint32_t *ptr; + uint32_t flags; + uint32_t data; + uint32_t vor; + uint32_t tor; +}; + +struct nouveau_pushbuf_bo { + uint64_t next; + uint64_t handle; + uint64_t flags; + uint64_t relocs; + int nr_relocs; +}; + struct nouveau_pushbuf { struct nouveau_channel *channel; unsigned remaining; @@ -106,8 +124,17 @@ struct nouveau_pushbuf_priv { struct nouveau_resource *res; struct nouveau_fence *fence; + + uint64_t buffers; + int nr_buffers; }; #define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) +#define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h)) +#define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_pbrel(h) ((struct nouveau_pushbuf_reloc *)(unsigned long)(h)) +#define bo_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_bo(h) ((struct nouveau_bo_priv *)(unsigned long)(h)) extern int nouveau_pushbuf_init(struct nouveau_channel *); @@ -115,6 +142,11 @@ nouveau_pushbuf_init(struct nouveau_channel *); extern int nouveau_pushbuf_flush(struct nouveau_channel *); +extern int +nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr, + struct nouveau_bo *, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor); + struct nouveau_channel_priv { struct nouveau_channel base; @@ -142,16 +174,6 @@ struct nouveau_channel_priv { int push_free; } dma; - struct { - struct nouveau_bo_priv *bo; - uint32_t flags; - } buffers[128]; - int nr_buffers; - - struct nouveau_bo_reloc *relocs; - int num_relocs; - int max_relocs; - struct nouveau_fence *fence_head; struct nouveau_fence *fence_tail; uint32_t fence_sequence; @@ -222,14 +244,6 @@ struct nouveau_bo_priv { int refcount; }; - -struct nouveau_bo_reloc { - struct nouveau_bo_priv *bo; - uint32_t *ptr; - uint32_t flags; - uint32_t data, vor, tor; -}; - #define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) extern int @@ -261,13 +275,9 @@ nouveau_bo_map(struct nouveau_bo *, uint32_t flags); extern void nouveau_bo_unmap(struct nouveau_bo *); -extern void -nouveau_bo_emit_reloc(struct nouveau_channel *chan, void *ptr, - struct nouveau_bo *, uint32_t data, uint32_t flags, - uint32_t vor, uint32_t tor); - -extern void -nouveau_bo_validate(struct nouveau_channel *); +extern int +nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, + uint32_t flags); extern int nouveau_resource_init(struct nouveau_resource **heap, int size); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index 2534a8ea6f9..1978edea02d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -49,8 +49,8 @@ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nouveau_bo_emit_reloc(nv->channel, nv->pushbuf, (void*)(bo), (data), \ - (flags), (vor), (tor)); \ + nouveau_pushbuf_emit_reloc(nv->channel, nv->pushbuf, (void*)(bo), \ + (data), (flags), (vor), (tor)); \ OUT_RING(0); \ } while(0) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 75d41d6c147..4a964cb82b5 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -54,7 +54,8 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); - int i, ret; + struct nouveau_pushbuf_bo *pbbo; + int ret; if (!nvpb) goto out_realloc; @@ -66,36 +67,42 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) if (ret) return ret; - /* Validate buffers */ - nouveau_bo_validate(chan); + /* Validate buffers + apply relocations */ + while ((pbbo = ptr_to_pbbo(nvpb->buffers))) { + struct nouveau_pushbuf_reloc *r; + struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; - /* Apply relocations */ - if (nvchan->num_relocs) { - for (i = 0; i < nvchan->num_relocs; i++) { - struct nouveau_bo_reloc *r = &nvchan->relocs[i]; + ret = nouveau_bo_validate(chan, bo, pbbo->flags); + assert (ret == 0); + + while ((r = ptr_to_pbrel(pbbo->relocs))) { uint32_t push; if (r->flags & NOUVEAU_BO_LOW) { - push = r->bo->base.offset + r->data; + push = bo->offset + r->data; } else if (r->flags & NOUVEAU_BO_HIGH) { - push = (r->bo->base.offset + r->data) >> 32; + push = (bo->offset + r->data) >> 32; } else { push = r->data; } if (r->flags & NOUVEAU_BO_OR) { - if (r->bo->base.flags & NOUVEAU_BO_VRAM) + if (bo->flags & NOUVEAU_BO_VRAM) push |= r->vor; else push |= r->tor; } *r->ptr = push; + pbbo->relocs = r->next; + free(r); } - nvchan->num_relocs = 0; + nvpb->buffers = pbbo->next; + free(pbbo); } + nvpb->nr_buffers = 0; /* Emit JMP to indirect pushbuf */ if (nvchan->dma.free < 1) @@ -175,3 +182,61 @@ out_realloc: return 0; } +static struct nouveau_pushbuf_bo * +nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); + struct nouveau_pushbuf_bo *pbbo = ptr_to_pbbo(nvpb->buffers); + + while (pbbo) { + if (pbbo->handle == bo->handle) + return pbbo; + pbbo = ptr_to_pbbo(pbbo->next); + } + + pbbo = malloc(sizeof(struct nouveau_pushbuf_bo)); + pbbo->next = nvpb->buffers; + nvpb->buffers = pbbo_to_ptr(pbbo); + nvpb->nr_buffers++; + + pbbo->handle = bo_to_ptr(bo); + pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; + pbbo->relocs = 0; + pbbo->nr_relocs = 0; + return pbbo; +} + +int +nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, + struct nouveau_bo *bo, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor) +{ + struct nouveau_pushbuf_bo *pbbo; + struct nouveau_pushbuf_reloc *r; + + if (!chan) + return -EINVAL; + + pbbo = nouveau_pushbuf_emit_buffer(chan, bo); + if (!pbbo) + return -EFAULT; + + r = malloc(sizeof(struct nouveau_pushbuf_reloc)); + r->next = pbbo->relocs; + pbbo->relocs = pbrel_to_ptr(r); + pbbo->nr_relocs++; + + pbbo->flags |= (flags & NOUVEAU_BO_RDWR); + pbbo->flags &= (flags | NOUVEAU_BO_RDWR); + + r->handle = bo_to_ptr(r); + r->ptr = ptr; + r->flags = flags; + r->data = data; + r->vor = vor; + r->tor = tor; + + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index 77d9e710dbe..9919272ebe1 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -204,7 +204,7 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->res_free = nouveau_resource_free; nvws->begin_ring = nouveau_pipe_dma_beginp; - nvws->out_reloc = nouveau_bo_emit_reloc; + nvws->out_reloc = nouveau_pushbuf_emit_reloc; nvws->fire_ring = nouveau_pipe_dma_kickoff; nvws->grobj_alloc = nouveau_pipe_grobj_alloc; diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index ce1744c2eec..4525f7113c2 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -25,7 +25,7 @@ struct nouveau_winsys { /*XXX: this is crappy, and bound to be slow.. however, it's nice and * simple, it'll do for the moment*/ uint32_t *(*begin_ring)(struct nouveau_grobj *, int mthd, int size); - void (*out_reloc)(struct nouveau_channel *, void *ptr, + int (*out_reloc)(struct nouveau_channel *, void *ptr, struct nouveau_bo *, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor); void (*fire_ring)(struct nouveau_channel *); -- cgit v1.2.3 From 897c96aacc6fbc08936eaf48955c15259e36caf7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 9 Dec 2007 22:25:35 +1100 Subject: nouveau: downgrade hacked syncs to hacked flushes Unfortunately we still need to keep the forced flushes as mesa's vbo path is still insane. But, at least this is a little less brutal. The bo code will detect when it needs to do a sync. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 18 ++++++++++++++++-- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 1 - src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 ++- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 21 ++++++++++++++++++--- src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 13 ------------- src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c | 12 ------------ src/mesa/pipe/nv40/nv40_vbo.c | 4 ++-- 7 files changed, 38 insertions(+), 34 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index fd6d05b7fab..d684ab4d7c4 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -205,6 +205,9 @@ nouveau_bo_del(struct nouveau_bo **userbo) if (--bo->refcount) return; + if (bo->fence) + nouveau_fence_wait(&bo->fence); + nouveau_bo_realloc_gpu(bo, 0, 0); if (bo->sysmem && !bo->user) free(bo->sysmem); @@ -235,13 +238,15 @@ nouveau_bo_unmap(struct nouveau_bo *userbo) static int nouveau_bo_upload(struct nouveau_bo_priv *bo) { + if (bo->fence) + nouveau_fence_wait(&bo->fence); memcpy(bo->map, bo->sysmem, bo->drm.size); return 0; } int nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, - uint32_t flags) + struct nouveau_fence *fence, uint32_t flags) { struct nouveau_bo_priv *nvbo = nouveau_bo(bo); @@ -249,14 +254,23 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, nouveau_bo_realloc_gpu(nvbo, flags, nvbo->base.size); nouveau_bo_upload(nvbo); } else - if (nvbo->user || nvbo->base.map) + if (nvbo->user) { + nouveau_bo_upload(nvbo); + } else + if (nvbo->base.map) { nouveau_bo_upload(nvbo); + nvbo->sync_hack = 1; + } if (!nvbo->user && !nvbo->base.map) { free(nvbo->sysmem); nvbo->sysmem = NULL; } + if (nvbo->fence) + nouveau_fence_del(&nvbo->fence); + nouveau_fence_ref(fence, &nvbo->fence); + nvbo->base.offset = nvbo->drm.offset; if (nvbo->drm.flags & NOUVEAU_MEM_AGP) nvbo->base.flags = NOUVEAU_BO_GART; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 6e32d739abc..d7bddbcfe20 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -155,7 +155,6 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) if ((chan->pushbuf[i] & 0xf0000000) == 0x20000000) { int n = (((chan->pushbuf[i] & 0x0fffffff) - chan->dma.base) / 4); - do { NOUVEAU_MSG("\t0x%08x 0x%08x\n", (n<<2)+chan->dma.base, diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 8af0b9d5cd3..f176767f136 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -243,6 +243,7 @@ struct nouveau_bo_priv { int user; int refcount; + int sync_hack; }; #define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) @@ -277,7 +278,7 @@ nouveau_bo_unmap(struct nouveau_bo *); extern int nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, - uint32_t flags); + struct nouveau_fence *fence, uint32_t flags); extern int nouveau_resource_init(struct nouveau_resource **heap, int size); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 4a964cb82b5..6cb06f8c37b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -55,6 +55,8 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) struct nouveau_channel_priv *nvchan = nouveau_channel(chan); struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); struct nouveau_pushbuf_bo *pbbo; + struct nouveau_fence *fence = NULL; + int sync_hack = 0; int ret; if (!nvpb) @@ -63,7 +65,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) if (nvpb->base.remaining == nvpb->res->size / 4) return 0; - ret = nouveau_fence_new(chan, &nvpb->fence); + ret = nouveau_fence_new(chan, &fence); if (ret) return ret; @@ -72,9 +74,12 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) struct nouveau_pushbuf_reloc *r; struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; - ret = nouveau_bo_validate(chan, bo, pbbo->flags); + ret = nouveau_bo_validate(chan, bo, fence, pbbo->flags); assert (ret == 0); + sync_hack |= nouveau_bo(bo)->sync_hack; + nouveau_bo(bo)->sync_hack = 0; + while ((r = ptr_to_pbrel(pbbo->relocs))) { uint32_t push; @@ -108,6 +113,9 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) if (nvchan->dma.free < 1) WAIT_RING_CH(chan, 1); nvchan->dma.free -= 1; +#ifdef NOUVEAU_DMA_DEBUG + nvchan->dma.push_free = 1; +#endif OUT_RING_CH(chan, 0x20000000 | (nvpb->res->start + 4096)); /* Add JMP back to master pushbuf from indirect pushbuf */ @@ -115,6 +123,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) 0x20000000 | ((nvchan->dma.cur << 2) + nvchan->dma.base); /* Fence */ + nvpb->fence = fence; nouveau_fence_emit(nvpb->fence); /* Kickoff */ @@ -173,12 +182,18 @@ out_realloc: nvpb->base.cur = &nvchan->pushbuf[(nvpb->res->start + 4096)/4]; if (nvchan->pb_tail) { - nouveau_pushbuf(nvchan->pb_tail)->next = & nvpb->base; + nouveau_pushbuf(nvchan->pb_tail)->next = &nvpb->base; } else { nvchan->pb_head = &nvpb->base; } nvchan->pb_tail = &nvpb->base; + if (sync_hack) { + struct nouveau_fence *f = NULL; + nouveau_fence_ref(nvpb->fence, &f); + nouveau_fence_wait(&f); + } + return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c index 893f292e232..67e2514026b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c @@ -116,13 +116,7 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, static void nv04_surface_copy_done(struct nouveau_context *nv) { - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(NvGdiRect, 0x104, 1); - OUT_RING (0); - BEGIN_RING(NvGdiRect, 0x100, 1); - OUT_RING (0); FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); } static int @@ -160,14 +154,7 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, OUT_RING ((dx << 16) | dy); OUT_RING (( w << 16) | h); - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(NvGdiRect, 0x104, 1); - OUT_RING (0); - BEGIN_RING(NvGdiRect, 0x100, 1); - OUT_RING (0); FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); - return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c index 2525476a792..2a95b1fcd36 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c @@ -82,13 +82,7 @@ nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, static void nv50_surface_copy_done(struct nouveau_context *nv) { - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(Nv2D, 0x104, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, 0x100, 1); - OUT_RING (0); FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); } static int @@ -134,13 +128,7 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, OUT_RING (dx + w); OUT_RING (dy + h); - nouveau_notifier_reset(nv->sync_notifier, 0); - BEGIN_RING(Nv2D, 0x104, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, 0x100, 1); - OUT_RING (0); FIRE_RING(); - nouveau_notifier_wait_status(nv->sync_notifier, 0, 0, 2000); return 0; } diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 5e919c5ca97..8491f2a3ec8 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -41,7 +41,7 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (0); - pipe->flush(pipe, PIPE_FLUSH_WAIT); + pipe->flush(pipe, 0); return TRUE; } @@ -153,7 +153,7 @@ nv40_draw_elements(struct pipe_context *pipe, OUT_RING (0); pipe->winsys->buffer_unmap(pipe->winsys, ib); - pipe->flush(pipe, PIPE_FLUSH_WAIT); + pipe->flush(pipe, 0); return TRUE; } -- cgit v1.2.3 From 1a3987240a547ba6e625c864f10a033858de4c65 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Dec 2007 02:07:47 +1100 Subject: nouveau: give resources a start property --- src/mesa/drivers/dri/nouveau_winsys/Makefile | 1 + .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 +- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 6 +- .../drivers/dri/nouveau_winsys/nouveau_resource.c | 111 +++++++++++++++++++++ .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 83 --------------- src/mesa/pipe/nouveau/nouveau_winsys.h | 3 +- src/mesa/pipe/nv40/nv40_context.c | 4 +- 7 files changed, 121 insertions(+), 90 deletions(-) create mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index fe9a9668b11..f2490c823d4 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -22,6 +22,7 @@ DRIVER_SOURCES = \ nouveau_lock.c \ nouveau_notifier.c \ nouveau_pushbuf.c \ + nouveau_resource.c \ nouveau_screen.c \ nouveau_swapbuffers.c \ nouveau_winsys.c \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index f176767f136..1c2a2c12cff 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -281,7 +281,8 @@ nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, struct nouveau_fence *fence, uint32_t flags); extern int -nouveau_resource_init(struct nouveau_resource **heap, int size); +nouveau_resource_init(struct nouveau_resource **heap, unsigned start, + unsigned size); extern int nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 6cb06f8c37b..551889f94c7 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -36,7 +36,7 @@ nouveau_pushbuf_init(struct nouveau_channel *chan) return -EINVAL; /* Everything except first 4KiB of the push buffer is managed by us */ - if (nouveau_resource_init(&nvchan->pb_heap, + if (nouveau_resource_init(&nvchan->pb_heap, 4096, nvchan->drm.cmdbuf_size - 4096)) return -EINVAL; @@ -116,7 +116,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) #ifdef NOUVEAU_DMA_DEBUG nvchan->dma.push_free = 1; #endif - OUT_RING_CH(chan, 0x20000000 | (nvpb->res->start + 4096)); + OUT_RING_CH(chan, 0x20000000 | nvpb->res->start); /* Add JMP back to master pushbuf from indirect pushbuf */ (*nvpb->base.cur++) = @@ -179,7 +179,7 @@ out_realloc: nvpb->base.channel = chan; nvpb->base.remaining = nvpb->res->size / 4; - nvpb->base.cur = &nvchan->pushbuf[(nvpb->res->start + 4096)/4]; + nvpb->base.cur = &nvchan->pushbuf[nvpb->res->start/4]; if (nvchan->pb_tail) { nouveau_pushbuf(nvchan->pb_tail)->next = &nvpb->base; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c new file mode 100644 index 00000000000..5d9d578b4fe --- /dev/null +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c @@ -0,0 +1,111 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +int +nouveau_resource_init(struct nouveau_resource **heap, + unsigned start, unsigned size) +{ + struct nouveau_resource *r; + + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = start; + r->size = size; + *heap = r; + return 0; +} + +int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!heap || !size || !res || *res) + return 1; + + while (heap) { + if (!heap->in_use && heap->size >= size) { + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = (heap->start + heap->size) - size; + r->size = size; + r->in_use = 1; + r->priv = priv; + + heap->size -= size; + + r->next = heap->next; + if (heap->next) + heap->next->prev = r; + r->prev = heap; + heap->next = r; + + *res = r; + return 0; + } + + heap = heap->next; + } + + return 1; +} + +void +nouveau_resource_free(struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!res || !*res) + return; + r = *res; + + if (r->prev && !r->prev->in_use) { + r->prev->next = r->next; + if (r->next) + r->next->prev = r->prev; + r->prev->size += r->size; + free(r); + } else + if (r->next && !r->next->in_use) { + r->next->prev = r->prev; + if (r->prev) + r->prev->next = r->next; + r->next->size += r->size; + r->next->start = r->start; + free(r); + } else { + r->in_use = 0; + } + + *res = NULL; +} diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index 9919272ebe1..e6481a20f2c 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -5,89 +5,6 @@ #include "pipe/nouveau/nouveau_winsys.h" -int -nouveau_resource_init(struct nouveau_resource **heap, int size) -{ - struct nouveau_resource *r; - - r = calloc(1, sizeof(struct nouveau_resource)); - if (!r) - return 1; - - r->start = 0; - r->size = size; - *heap = r; - return 0; -} - -int -nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, - struct nouveau_resource **res) -{ - struct nouveau_resource *r; - - if (!heap || !size || !res || *res) - return 1; - - while (heap) { - if (!heap->in_use && heap->size >= size) { - r = calloc(1, sizeof(struct nouveau_resource)); - if (!r) - return 1; - - r->start = (heap->start + heap->size) - size; - r->size = size; - r->in_use = TRUE; - r->priv = priv; - - heap->size -= size; - - r->next = heap->next; - if (heap->next) - heap->next->prev = r; - r->prev = heap; - heap->next = r; - - *res = r; - return 0; - } - - heap = heap->next; - } - - return 1; -} - -void -nouveau_resource_free(struct nouveau_resource **res) -{ - struct nouveau_resource *r; - - if (!res || !*res) - return; - r = *res; - - if (r->prev && !r->prev->in_use) { - r->prev->next = r->next; - if (r->next) - r->next->prev = r->prev; - r->prev->size += r->size; - free(r); - } else - if (r->next && !r->next->in_use) { - r->next->prev = r->prev; - if (r->prev) - r->prev->next = r->next; - r->next->size += r->size; - r->next->start = r->start; - free(r); - } else { - r->in_use = FALSE; - } - - *res = NULL; -} - static int nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, struct nouveau_notifier **notify) diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 4525f7113c2..7b80027c3a6 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -17,7 +17,8 @@ struct nouveau_winsys { struct nouveau_channel *channel; - int (*res_init)(struct nouveau_resource **heap, int size); + int (*res_init)(struct nouveau_resource **heap, unsigned start, + unsigned size); int (*res_alloc)(struct nouveau_resource *heap, int size, void *priv, struct nouveau_resource **); void (*res_free)(struct nouveau_resource **); diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 7b77c70e81d..518ba3574f7 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -241,8 +241,8 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, return NULL; } - if (nvws->res_init(&nv40->vertprog.exec_heap, 512) || - nvws->res_init(&nv40->vertprog.data_heap, 256)) { + if (nvws->res_init(&nv40->vertprog.exec_heap, 0, 512) || + nvws->res_init(&nv40->vertprog.data_heap, 0, 256)) { nvws->res_free(&nv40->vertprog.exec_heap); nvws->res_free(&nv40->vertprog.data_heap); free(nv40); -- cgit v1.2.3 From 9ba3890c6f35701df5c7ea94cfac9954e9cbec2e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Dec 2007 12:13:00 +1100 Subject: nouveau: fix build --- .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 21 ++++++++++++++------- src/mesa/pipe/nv40/nv40_fragprog.c | 2 +- src/mesa/pipe/nv40/nv40_miptree.c | 2 +- 3 files changed, 16 insertions(+), 9 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index adbf6b28e8d..fdfad4a5399 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -73,7 +73,8 @@ nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) } static struct pipe_buffer_handle * -nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment) +nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, + unsigned flags, unsigned hint) { struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; struct nouveau_device *dev = nvpws->nv->nv_screen->device; @@ -144,7 +145,7 @@ nouveau_pipe_bo_reference(struct pipe_winsys *pws, } } -static void +static int nouveau_pipe_bo_data(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, unsigned size, const void *data, unsigned usage) { @@ -155,13 +156,15 @@ nouveau_pipe_bo_data(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, if (data) { if (nouveau_bo_map(nvbo, NOUVEAU_BO_WR)) - return; + return 1; memcpy(nvbo->map, data, size); nouveau_bo_unmap(nvbo); } + + return 0; } -static void +static int nouveau_pipe_bo_subdata(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, unsigned long offset, unsigned long size, const void *data) @@ -169,12 +172,14 @@ nouveau_pipe_bo_subdata(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; if (nouveau_bo_map(nvbo, NOUVEAU_BO_WR)) - return; + return 1; memcpy(nvbo->map + offset, data, size); nouveau_bo_unmap(nvbo); + + return 0; } -static void +static int nouveau_pipe_bo_get_subdata(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, unsigned long offset, unsigned long size, void *data) @@ -182,9 +187,11 @@ nouveau_pipe_bo_get_subdata(struct pipe_winsys *pws, struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; if (nouveau_bo_map(nvbo, NOUVEAU_BO_RD)) - return; + return 1; memcpy(data, nvbo->map + offset, size); nouveau_bo_unmap(nvbo); + + return 0; } struct pipe_winsys * diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 8f3eeb1f481..cc3cb970430 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -695,7 +695,7 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) if (!fp->on_hw) { if (!fp->buffer) - fp->buffer = ws->buffer_create(ws, 0x100); + fp->buffer = ws->buffer_create(ws, 0x100, 0, 0); #if 0 int i; diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 45711f064d6..3844dd9dda0 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -68,7 +68,7 @@ nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) memcpy(&nv40mt->base, mt, sizeof(struct pipe_texture)); nv40_miptree_layout(nv40mt); - nv40mt->buffer = ws->buffer_create(ws, PIPE_SURFACE_FLAG_TEXTURE); + nv40mt->buffer = ws->buffer_create(ws, 0, PIPE_SURFACE_FLAG_TEXTURE, 0); if (!nv40mt->buffer) { free(nv40mt); return; -- cgit v1.2.3 From 7d8368790fabc19e51add0fd9f1d1f85b7803cdf Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Dec 2007 13:19:47 +1100 Subject: nouveau: add callbacks for signalled fences --- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 9 +++ .../drivers/dri/nouveau_winsys/nouveau_fence.c | 46 ++++++++++++-- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 72 +++++++--------------- 3 files changed, 72 insertions(+), 55 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 1c2a2c12cff..6d6633fac30 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -64,11 +64,18 @@ struct nouveau_fence { struct nouveau_channel *channel; }; +struct nouveau_fence_cb { + struct nouveau_fence_cb *next; + void (*func)(void *); + void *priv; +}; + struct nouveau_fence_priv { struct nouveau_fence base; int refcount; struct nouveau_fence *next; + struct nouveau_fence_cb *signal_cb; uint32_t sequence; int emitted; @@ -85,6 +92,8 @@ nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); extern void nouveau_fence_del(struct nouveau_fence **); +extern int +nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *); extern void nouveau_fence_emit(struct nouveau_fence *); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c index 842b38f913b..7861b6f84d3 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c @@ -76,6 +76,27 @@ nouveau_fence_del(struct nouveau_fence **fence) } } +int +nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *), + void *priv) +{ + struct nouveau_fence_priv *nvfence = nouveau_fence(fence); + struct nouveau_fence_cb *cb; + + if (!nvfence || !func) + return -EINVAL; + + cb = malloc(sizeof(struct nouveau_fence_cb)); + if (!cb) + return -ENOMEM; + + cb->func = func; + cb->priv = priv; + cb->next = nvfence->signal_cb; + nvfence->signal_cb = cb; + return 0; +} + void nouveau_fence_emit(struct nouveau_fence *fence) { @@ -102,16 +123,33 @@ void nouveau_fence_flush(struct nouveau_channel *chan) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_fence_priv *nvfence = nouveau_fence(nvchan->fence_head); uint32_t sequence = *nvchan->ref_cnt; - while (nvchan->fence_head && nvfence->sequence <= sequence) { - nvfence->signalled = 1; + while (nvchan->fence_head) { + struct nouveau_fence *fence = NULL; + struct nouveau_fence_priv *nvfence; + + nouveau_fence_ref(nvchan->fence_head, &fence); + nvfence = nouveau_fence(nvchan->fence_head); + + if (nvfence->sequence > sequence) { + nouveau_fence_del(&fence); + break; + } nvchan->fence_head = nvfence->next; if (nvchan->fence_head == NULL) nvchan->fence_tail = NULL; - nvfence = nouveau_fence(nvchan->fence_head); + nvfence->signalled = 1; + + while (nvfence->signal_cb) { + struct nouveau_fence_cb *cb = nvfence->signal_cb; + nvfence->signal_cb = cb->next; + cb->func(cb->priv); + free(cb); + } + + nouveau_fence_del(&fence); } } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 551889f94c7..a922300ff5e 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -48,6 +48,16 @@ nouveau_pushbuf_init(struct nouveau_channel *chan) return 0; } +static void +nouveau_pushbuf_fence_signalled(void *priv) +{ + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(priv); + + nouveau_fence_del(&nvpb->fence); + nouveau_resource_free(&nvpb->res); + free(nvpb); +} + /* This would be our TTM "superioctl" */ int nouveau_pushbuf_flush(struct nouveau_channel *chan) @@ -64,6 +74,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) if (nvpb->base.remaining == nvpb->res->size / 4) return 0; + nvchan->pb_tail = NULL; ret = nouveau_fence_new(chan, &fence); if (ret) @@ -124,76 +135,35 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) /* Fence */ nvpb->fence = fence; + nouveau_fence_signal_cb(nvpb->fence, nouveau_pushbuf_fence_signalled, + nvpb); nouveau_fence_emit(nvpb->fence); /* Kickoff */ FIRE_RING_CH(chan); + if (sync_hack) { + struct nouveau_fence *f = NULL; + nouveau_fence_ref(nvpb->fence, &f); + nouveau_fence_wait(&f); + } + /* Allocate space for next push buffer */ out_realloc: nvpb = calloc(1, sizeof(struct nouveau_pushbuf_priv)); if (!nvpb) return -ENOMEM; - if (nouveau_resource_alloc(nvchan->pb_heap, 0x2000, NULL, &nvpb->res)) { - struct nouveau_pushbuf_priv *e; - int nr = 0; - - /* Update fences */ + while (nouveau_resource_alloc(nvchan->pb_heap, 0x2000, NULL, + &nvpb->res)) { nouveau_fence_flush(chan); - - /* Free any push buffers that have already been executed */ - e = nouveau_pushbuf(nvchan->pb_head); - while (e && e->fence) { - if (!e->fence || !nouveau_fence(e->fence)->signalled) - break; - nouveau_fence_del(&e->fence); - nouveau_resource_free(&e->res); - nr++; - - nvchan->pb_head = e->next; - if (nvchan->pb_head == NULL) - nvchan->pb_tail = NULL; - free(e); - e = nouveau_pushbuf(nvchan->pb_head); - } - - /* We didn't free any buffers above. As a last resort, busy - * wait on the oldest buffer becoming available. - */ - if (!nr) { - e = nouveau_pushbuf(nvchan->pb_head); - nouveau_fence_wait(&e->fence); - nouveau_resource_free(&e->res); - - nvchan->pb_head = e->next; - if (nvchan->pb_head == NULL) - nvchan->pb_tail = NULL; - free(e); - } - - if (nouveau_resource_alloc(nvchan->pb_heap, 0x2000, nvpb, - &nvpb->res)) - assert(0); } nvpb->base.channel = chan; nvpb->base.remaining = nvpb->res->size / 4; nvpb->base.cur = &nvchan->pushbuf[nvpb->res->start/4]; - - if (nvchan->pb_tail) { - nouveau_pushbuf(nvchan->pb_tail)->next = &nvpb->base; - } else { - nvchan->pb_head = &nvpb->base; - } nvchan->pb_tail = &nvpb->base; - if (sync_hack) { - struct nouveau_fence *f = NULL; - nouveau_fence_ref(nvpb->fence, &f); - nouveau_fence_wait(&f); - } - return 0; } -- cgit v1.2.3 From 3b2598c70bce098ea8c43ff37862bcce1663514b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Dec 2007 15:16:05 +1100 Subject: nouveau: fix elusive dma bug In some situations WAIT_RING would get called while the GPU was processing data from outside the "master" ring, which caused dma.free to be updated incorrectly and much fun was had. WAIT_RING will now wait until it reads GET values from within the main ring buffer before calculating free space. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 54 ++++++++++++++--------- 1 file changed, 32 insertions(+), 22 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 6ec390ec769..5739e0010d4 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -28,14 +28,29 @@ #include "nouveau_dma.h" #include "nouveau_local.h" -#define READ_GET(ch) ((*(ch)->get - (ch)->dma.base) >> 2) -#define WRITE_PUT(ch, val) do { \ - volatile int dum; \ - NOUVEAU_DMA_BARRIER; \ - dum=READ_GET(ch); \ - *(ch)->put = (((val) << 2) + (ch)->dma.base); \ - NOUVEAU_DMA_BARRIER; \ -} while(0) +static __inline__ uint32_t +READ_GET(struct nouveau_channel_priv *nvchan) +{ + return ((*nvchan->get - nvchan->dma.base) >> 2); +} + +static __inline__ void +WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) +{ + uint32_t put = ((val << 2) + nvchan->dma.base); + volatile int dum; + + NOUVEAU_DMA_BARRIER; + dum = READ_GET(nvchan); + + *nvchan->put = put; + nvchan->dma.put = val; +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put); +#endif + + NOUVEAU_DMA_BARRIER; +} void nouveau_dma_channel_init(struct nouveau_channel *userchan) @@ -57,6 +72,8 @@ nouveau_dma_channel_init(struct nouveau_channel *userchan) return - EBUSY; \ } while(0) +#define IN_MASTER_RING(chan, ptr) ((ptr) <= (chan)->dma.max) + int nouveau_dma_wait(struct nouveau_channel *userchan, int size) { @@ -67,7 +84,11 @@ nouveau_dma_wait(struct nouveau_channel *userchan, int size) t_start = NOUVEAU_TIME_MSEC(); while (chan->dma.free < size) { + CHECK_TIMEOUT(); + get = READ_GET(chan); + if (!IN_MASTER_RING(chan, get)) + continue; if (chan->dma.put >= get) { chan->dma.free = chan->dma.max - chan->dma.cur; @@ -86,6 +107,8 @@ nouveau_dma_wait(struct nouveau_channel *userchan, int size) do { CHECK_TIMEOUT(); get = READ_GET(chan); + if (!IN_MASTER_RING(chan, get)) + continue; } while (get <= RING_SKIPS); } @@ -96,8 +119,6 @@ nouveau_dma_wait(struct nouveau_channel *userchan, int size) } else { chan->dma.free = get - chan->dma.cur - 1; } - - CHECK_TIMEOUT(); } return 0; @@ -135,9 +156,6 @@ void nouveau_dma_kickoff(struct nouveau_channel *userchan) { struct nouveau_channel_priv *chan = nouveau_channel(userchan); - uint32_t put_offset; - int i; - volatile int dum; if (chan->dma.cur == chan->dma.put) return; @@ -165,13 +183,5 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) } #endif - put_offset = (chan->dma.cur << 2) + chan->dma.base; -#ifdef NOUVEAU_DMA_TRACE - NOUVEAU_MSG("FIRE_RING %d/0x%08x\n", chan->drm.channel, put_offset); -#endif - chan->dma.put = chan->dma.cur; - NOUVEAU_DMA_BARRIER; - dum = READ_GET(chan); - *chan->put = put_offset; - NOUVEAU_DMA_BARRIER; + WRITE_PUT(chan, chan->dma.cur); } -- cgit v1.2.3 From aab6f8f9485e73eb4098e7c4b497da5c30f1cf06 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Dec 2007 16:47:47 +1100 Subject: nv40: more flexible vbo format translation --- src/mesa/pipe/nv40/nv40_vbo.c | 32 ++++++++++++++++++++++---------- 1 file changed, 22 insertions(+), 10 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 8491f2a3ec8..b911ba5a7fc 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -158,16 +158,28 @@ nv40_draw_elements(struct pipe_context *pipe, } static INLINE int -nv40_vbo_format_to_ncomp(uint format) +nv40_vbo_ncomp(uint format) { - switch (format) { - case PIPE_FORMAT_R32G32B32A32_FLOAT: return 4; - case PIPE_FORMAT_R32G32B32_FLOAT: return 3; - case PIPE_FORMAT_R32G32_FLOAT: return 2; - case PIPE_FORMAT_R32_FLOAT: return 1; + int ncomp = 0; + + if (pf_size_x(format)) ncomp++; + if (pf_size_y(format)) ncomp++; + if (pf_size_z(format)) ncomp++; + if (pf_size_w(format)) ncomp++; + + return ncomp; +} + +static INLINE int +nv40_vbo_type(uint format) +{ + switch (pf_type(format)) { + case PIPE_FORMAT_TYPE_FLOAT: + return NV40TCL_VTXFMT_TYPE_FLOAT; + case PIPE_FORMAT_TYPE_UNORM: + return NV40TCL_VTXFMT_TYPE_UBYTE; default: - NOUVEAU_ERR("AII, unknown vbo format %d\n", format); - return 1; + assert(0); } } @@ -207,9 +219,9 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, NV40TCL_VTXBUF_ADDRESS_DMA1); vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_format_to_ncomp(ve->src_format) << + (nv40_vbo_ncomp(ve->src_format) << NV40TCL_VTXFMT_SIZE_SHIFT) | - NV40TCL_VTXFMT_TYPE_FLOAT); + nv40_vbo_type(ve->src_format)); } BEGIN_RING(curie, 0x1710, 1); -- cgit v1.2.3 From 4bc3cfaaf065f373f2e2d1ca4361c3d4ffdca4da Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 11 Dec 2007 10:04:39 +1100 Subject: nouveau: fix build --- src/mesa/drivers/dri/nouveau_winsys/Makefile | 2 +- src/mesa/pipe/nv40/nv40_miptree.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index f2490c823d4..f547ec43766 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -32,7 +32,7 @@ DRIVER_SOURCES = \ nv50_surface.c C_SOURCES = \ - $(COMMON_SOURCES) \ + $(COMMON_GALLIUM_SOURCES) \ $(DRIVER_SOURCES) ASM_SOURCES = diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 3844dd9dda0..9d4fbb68264 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -68,7 +68,7 @@ nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) memcpy(&nv40mt->base, mt, sizeof(struct pipe_texture)); nv40_miptree_layout(nv40mt); - nv40mt->buffer = ws->buffer_create(ws, 0, PIPE_SURFACE_FLAG_TEXTURE, 0); + nv40mt->buffer = ws->buffer_create(ws, 256, 0, 0); if (!nv40mt->buffer) { free(nv40mt); return; -- cgit v1.2.3 From ce0e274c7cdf7ceb5ffd46d545b5cd18311dc6e0 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 11 Dec 2007 14:08:42 +1100 Subject: nv40: some forgotten adaptions --- src/mesa/pipe/nv40/nv40_context.c | 37 ++++++++++++++++++++++++++++++------- src/mesa/pipe/nv40/nv40_draw.c | 7 +++++++ 2 files changed, 37 insertions(+), 7 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 518ba3574f7..6d12312d518 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -7,15 +7,38 @@ #include "nv40_dma.h" static boolean -nv40_is_format_supported(struct pipe_context *pipe, uint format) +nv40_is_format_supported(struct pipe_context *pipe, enum pipe_format format, + uint type) { - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: + switch (type) { + case PIPE_SURFACE: + case PIPE_SCREEN_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + return TRUE; + default: + break; + } break; + default: + assert(0); }; return FALSE; diff --git a/src/mesa/pipe/nv40/nv40_draw.c b/src/mesa/pipe/nv40/nv40_draw.c index 52ce493ea2d..6ad2b66cce8 100644 --- a/src/mesa/pipe/nv40/nv40_draw.c +++ b/src/mesa/pipe/nv40/nv40_draw.c @@ -44,6 +44,12 @@ nv40_draw_reset_stipple_counter(struct draw_stage *draw) NOUVEAU_ERR("\n"); } +static void +nv40_draw_destroy(struct draw_stage *draw) +{ + free(draw); +} + struct draw_stage * nv40_draw_render_stage(struct nv40_context *nv40) { @@ -57,6 +63,7 @@ nv40_draw_render_stage(struct nv40_context *nv40) nv40draw->draw.tri = nv40_draw_tri; nv40draw->draw.end = nv40_draw_end; nv40draw->draw.reset_stipple_counter = nv40_draw_reset_stipple_counter; + nv40draw->draw.destroy = nv40_draw_destroy; return &nv40draw->draw; } -- cgit v1.2.3 From b34952c758cf009927e7d7091205e7c13052efad Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 11 Dec 2007 14:53:37 +1100 Subject: nv40: depth textures --- src/mesa/pipe/nv40/nv40_context.c | 5 ++++- src/mesa/pipe/nv40/nv40_state.c | 31 +++++++++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_state_tex.c | 5 +++++ 3 files changed, 40 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 6d12312d518..b1358a8cfa5 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -17,6 +17,7 @@ nv40_is_format_supported(struct pipe_context *pipe, enum pipe_format format, case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: return TRUE; default: break; @@ -32,6 +33,8 @@ nv40_is_format_supported(struct pipe_context *pipe, enum pipe_format format, case PIPE_FORMAT_U_A8: case PIPE_FORMAT_U_I8: case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: return TRUE; default: break; @@ -83,7 +86,7 @@ nv40_get_param(struct pipe_context *pipe, int param) case PIPE_CAP_OCCLUSION_QUERY: return 1; case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 0; + return 1; case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: return 13; case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index e4915794830..1d4cbb454d4 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -244,6 +244,37 @@ nv40_sampler_state_create(struct pipe_context *pipe, ps->filt = filter; + if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= (0x0 << 28); + break; + case PIPE_FUNC_GREATER: + ps->wrap |= (0x1 << 28); + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= (0x2 << 28); + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= (0x3 << 28); + break; + case PIPE_FUNC_LESS: + ps->wrap |= (0x4 << 28); + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= (0x5 << 28); + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= (0x6 << 28); + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= (0x7 << 28); + break; + default: + break; + } + } + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | (float_to_ubyte(cso->border_color[0]) << 16) | (float_to_ubyte(cso->border_color[1]) << 8) | diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 30462a2a3d6..05c136fcc19 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -12,6 +12,9 @@ NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ } +#define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x1200 +#define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x1000 + struct nv40_texture_format { boolean defined; uint pipe; @@ -29,6 +32,8 @@ nv40_texture_formats[] = { _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), _(U_A8_L8 , A8L8 , S1, S1, S1, S1, Z, W, X, Y), + _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), + _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), // _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), // _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), // _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -- cgit v1.2.3 From c23d0f4c501b9e80ba18d6ccf09d4c95353c1a28 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 11 Dec 2007 16:02:14 +1100 Subject: nv40: allow reading from fragprog result regs --- src/mesa/pipe/nv40/nv40_fragprog.c | 16 +++++++++++++++- src/mesa/pipe/nv40/nv40_shader.h | 8 ++++---- 2 files changed, 19 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index cc3cb970430..105d7937f34 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -85,6 +85,9 @@ emit_src(struct nv40_fpc *fpc, uint32_t *hw, int pos, struct nv40_sreg src) sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); hw[0] |= (src.index << NV40_FP_OP_INPUT_SRC_SHIFT); break; + case NV40SR_OUTPUT: + sr |= NV40_FP_REG_SRC_HALF; + /* fall-through */ case NV40SR_TEMP: sr |= (NV40_FP_REG_TYPE_TEMP << NV40_FP_REG_TYPE_SHIFT); sr |= (src.index << NV40_FP_REG_SRC_SHIFT); @@ -129,7 +132,7 @@ emit_dst(struct nv40_fpc *fpc, uint32_t *hw, struct nv40_sreg dst) if (dst.index == 1) { fp->writes_depth = 1; } else { - hw[0] |= NV40_FP_OP_UNK0_7; + hw[0] |= NV40_FP_OP_OUT_REG_HALF; } break; case NV40SR_NONE: @@ -213,6 +216,15 @@ tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) if (fpc->high_temp < src.index) fpc->high_temp = src.index; break; + /* This is clearly insane, but gallium hands us shaders like this. + * Luckily fragprog results are just temp regs.. + */ + case TGSI_FILE_OUTPUT: + if (fsrc->SrcRegister.Index == fpc->colour_id) + return nv40_sr(NV40SR_OUTPUT, 0); + else + return nv40_sr(NV40SR_OUTPUT, 1); + break; default: NOUVEAU_ERR("bad src file\n"); break; @@ -391,6 +403,8 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, case TGSI_FILE_SAMPLER: unit = fsrc->SrcRegister.Index; break; + case TGSI_FILE_OUTPUT: + break; default: NOUVEAU_ERR("bad src file\n"); return FALSE; diff --git a/src/mesa/pipe/nv40/nv40_shader.h b/src/mesa/pipe/nv40/nv40_shader.h index 207dafd7483..01c0652b4d8 100644 --- a/src/mesa/pipe/nv40/nv40_shader.h +++ b/src/mesa/pipe/nv40/nv40_shader.h @@ -292,9 +292,9 @@ //== Opcode / Destination selection == #define NV40_FP_OP_PROGRAM_END (1 << 0) #define NV40_FP_OP_OUT_REG_SHIFT 1 -#define NV40_FP_OP_OUT_REG_MASK (31 << 1) +#define NV40_FP_OP_OUT_REG_MASK (63 << 1) /* Needs to be set when writing outputs to get expected result.. */ -#define NV40_FP_OP_UNK0_7 (1 << 7) +#define NV40_FP_OP_OUT_REG_HALF (1 << 7) #define NV40_FP_OP_COND_WRITE_ENABLE (1 << 8) #define NV40_FP_OP_OUTMASK_SHIFT 9 #define NV40_FP_OP_OUTMASK_MASK (0xF << 9) @@ -456,8 +456,8 @@ # define NV40_FP_REG_TYPE_INPUT 1 # define NV40_FP_REG_TYPE_CONST 2 #define NV40_FP_REG_SRC_SHIFT 2 -#define NV40_FP_REG_SRC_MASK (31 << 2) -#define NV40_FP_REG_UNK_0 (1 << 8) +#define NV40_FP_REG_SRC_MASK (63 << 2) +#define NV40_FP_REG_SRC_HALF (1 << 8) #define NV40_FP_REG_SWZ_ALL_SHIFT 9 #define NV40_FP_REG_SWZ_ALL_MASK (255 << 9) #define NV40_FP_REG_SWZ_X_SHIFT 9 -- cgit v1.2.3 From e282d22d512d2a5871d0fabb7d855a16b4593c50 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 11 Dec 2007 19:06:00 +1100 Subject: nv40: insanely stupid bug.. --- src/mesa/pipe/nv40/nv40_vbo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index b911ba5a7fc..19ae6ca3238 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -152,7 +152,7 @@ nv40_draw_elements(struct pipe_context *pipe, BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (0); - pipe->winsys->buffer_unmap(pipe->winsys, ib); + pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); pipe->flush(pipe, 0); return TRUE; } -- cgit v1.2.3 From 79bca7dd884da33c06ecd3dabb893f9cfed1aaed Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Dec 2007 14:19:59 +1100 Subject: nouveau: adapt Some things that worked before are now broken, there's an "XXX:" around one of the culprits in the GL state tracker so hopefully it'll get fixed soon! --- .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 42 ++++--- src/mesa/pipe/nv40/nv40_context.h | 4 +- src/mesa/pipe/nv40/nv40_query.c | 126 ++++++++++----------- src/mesa/pipe/nv40/nv40_state.c | 50 +++----- src/mesa/pipe/nv40/nv40_surface.c | 3 +- src/mesa/pipe/nv50/nv50_context.c | 11 +- src/mesa/pipe/nv50/nv50_query.c | 26 ++++- src/mesa/pipe/nv50/nv50_state.c | 36 ++---- 8 files changed, 145 insertions(+), 153 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index fdfad4a5399..1e160f04edd 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -34,18 +34,8 @@ nouveau_get_name(struct pipe_winsys *pws) return "Nouveau/DRI"; } -static unsigned -nouveau_surface_pitch(struct pipe_winsys *ws, unsigned cpp, unsigned width, - unsigned flags) -{ - unsigned pitch = width * cpp; - - pitch = (pitch + 63) & ~63; - return pitch / cpp; -} - static struct pipe_surface * -nouveau_surface_alloc(struct pipe_winsys *ws, unsigned format) +nouveau_surface_alloc(struct pipe_winsys *ws) { struct pipe_surface *surf; @@ -53,12 +43,38 @@ nouveau_surface_alloc(struct pipe_winsys *ws, unsigned format) if (!surf) return NULL; - surf->format = format; surf->refcount = 1; surf->winsys = ws; return surf; } +static int +nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, + unsigned width, unsigned height, + enum pipe_format format, unsigned flags) +{ + unsigned pitch = ((width * pf_get_size(format)) + 63) & ~63; + int ret; + + surf->format = format; + surf->width = width; + surf->height = height; + surf->cpp = pf_get_size(format); + surf->pitch = pitch / surf->cpp; + + surf->buffer = ws->buffer_create(ws, 256, 0, 0); + if (!surf->buffer) + return 1; + + ret = ws->buffer_data(ws, surf->buffer, pitch * height, NULL, 0); + if (ret) { + ws->buffer_reference(ws, &surf->buffer, NULL); + return ret; + } + + return 0; +} + static void nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) { @@ -209,8 +225,8 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws->flush_frontbuffer = nouveau_flush_frontbuffer; pws->printf = nouveau_printf; - pws->surface_pitch = nouveau_surface_pitch; pws->surface_alloc = nouveau_surface_alloc; + pws->surface_alloc_storage = nouveau_surface_alloc_storage; pws->surface_release = nouveau_surface_release; pws->buffer_create = nouveau_pipe_bo_create; diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 3a1d3f076c6..83d3f7b676e 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -35,7 +35,7 @@ struct nv40_context { /* query objects */ struct nouveau_notifier *query; - struct pipe_query_object **query_objects; + boolean *query_objects; uint num_query_objects; uint32_t dirty; @@ -64,7 +64,7 @@ struct nv40_context { struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; - +#define nv40_context(ctx) ((struct nv40_context *)(ctx)) extern void nv40_init_state_functions(struct nv40_context *nv40); extern void nv40_init_surface_functions(struct nv40_context *nv40); diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index bcd6fe0cf49..f51b34b119d 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -3,33 +3,57 @@ #include "nv40_context.h" #include "nv40_dma.h" -static uint -nv40_query_object_find(struct nv40_context *nv40, struct pipe_query_object *q) +/*XXX: Maybe go notifier per-query one day? not sure if PRAMIN space is + * plentiful enough however.. */ +struct nv40_query { + unsigned type; + int id; +}; +#define nv40_query(o) ((struct nv40_query *)(o)) + +static struct pipe_query * +nv40_query_create(struct pipe_context *pipe, unsigned query_type) { + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *nv40query; int id; for (id = 0; id < nv40->num_query_objects; id++) { - if (nv40->query_objects[id] == q) - return id; + if (nv40->query_objects[id] == 0) + break; } + + if (id == nv40->num_query_objects) + return NULL; + nv40->query_objects[id] = TRUE; - return -1; + nv40query = malloc(sizeof(struct nv40_query)); + nv40query->type = query_type; + nv40query->id = id; + + return (struct pipe_query *)nv40query; } static void -nv40_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) +nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *q) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; - int id; + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *nv40query = nv40_query(q); - assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); + assert(nv40->query_objects[nv40query->id]); + nv40->query_objects[nv40query->id] = FALSE; + free(nv40query); +} - id = nv40_query_object_find(nv40, NULL); - assert(id >= 0); - nv40->query_objects[id] = q; +static void +nv40_query_begin(struct pipe_context *pipe, struct pipe_query *q) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *nv40query = nv40_query(q); + + assert(nv40query->type == PIPE_QUERY_OCCLUSION_COUNTER); - nv40->nvws->notifier_reset(nv40->query, id); - q->ready = 0; + nv40->nvws->notifier_reset(nv40->query, nv40query->id); BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); OUT_RING (1); @@ -38,68 +62,44 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) } static void -nv40_query_update(struct pipe_context *pipe, struct pipe_query_object *q) +nv40_query_end(struct pipe_context *pipe, struct pipe_query *q) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - int id; + struct nv40_query *nv40query = nv40_query(q); - id = nv40_query_object_find(nv40, q); - assert(id >= 0); - - if (nv40->nvws->notifier_status(nv40->query, id) == 0) { - q->ready = 1; - q->count = nv40->nvws->notifier_retval(nv40->query, id); - nv40->query_objects[id] = NULL; - } -} - -static void -nv40_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) -{ - nv40_query_update(pipe, q); - if (!q->ready) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; - int id; - - id = nv40_query_object_find(nv40, q); - assert(id >= 0); - - nv40->nvws->notifier_wait(nv40->query, id, 0, 0); - nv40_query_update(pipe, q); - assert(q->ready); - } + BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); + OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | + ((nv40query->id * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); + FIRE_RING(); } -static void -nv40_query_end(struct pipe_context *pipe, struct pipe_query_object *q) +static boolean +nv40_query_result(struct pipe_context *pipe, struct pipe_query *q, + boolean wait, uint64_t *result) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - int id; - - id = nv40_query_object_find(nv40, q); - assert(id >= 0); + struct nv40_query *nv40query = nv40_query(q); + struct nouveau_winsys *nvws = nv40->nvws; + boolean status; + + status = nvws->notifier_status(nv40->query, nv40query->id); + if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { + if (wait == FALSE) + return FALSE; + nvws->notifier_wait(nv40->query, nv40query->id, + NV_NOTIFY_STATE_STATUS_COMPLETED, 0); + } - BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); - OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | - ((id * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); - FIRE_RING (); - - /*XXX: Some apps spin waiting for GL_QUERY_RESULT_AVAILABLE_ARB. - * Core mesa won't ask the driver to update the query object's - * status in this case, so the app waits forever.. fix this some - * day. - */ -#if 0 - nv40_query_update(pipe, q); -#else - nv40_query_wait(pipe, q); -#endif + *result = nvws->notifier_retval(nv40->query, nv40query->id); + return TRUE; } void nv40_init_query_functions(struct nv40_context *nv40) { + nv40->pipe.create_query = nv40_query_create; + nv40->pipe.destroy_query = nv40_query_destroy; nv40->pipe.begin_query = nv40_query_begin; nv40->pipe.end_query = nv40_query_end; - nv40->pipe.wait_query = nv40_query_wait; + nv40->pipe.get_query_result = nv40_query_result; } diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 1d4cbb454d4..5aeba684ee0 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -302,6 +302,18 @@ nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso) free(hwcso); } +static void +nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, + struct pipe_texture *miptree) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + + nv40->tex_miptree[unit] = miptree; + nv40->tex_dirty |= unit; + + nv40->dirty |= NV40_NEW_TEXTURE; +} + static void * nv40_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) @@ -546,19 +558,6 @@ nv40_set_clip_state(struct pipe_context *pipe, nv40->dirty |= NV40_NEW_VERTPROG; } -static void -nv40_set_clear_color_state(struct pipe_context *pipe, - const struct pipe_clear_color_state *ccol) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - - BEGIN_RING(curie, NV40TCL_CLEAR_VALUE_COLOR, 1); - OUT_RING ((float_to_ubyte(ccol->color[3]) << 24) | - (float_to_ubyte(ccol->color[0]) << 16) | - (float_to_ubyte(ccol->color[1]) << 8) | - (float_to_ubyte(ccol->color[2]) << 0)); -} - static void nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, const struct pipe_constant_buffer *buf ) @@ -732,12 +731,6 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, OUT_RINGp ((uint32_t *)stipple->stipple, 32); } -static void -nv40_set_sampler_units(struct pipe_context *pipe, - uint num_samplers, const uint *units) -{ -} - static void nv40_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) @@ -749,18 +742,6 @@ nv40_set_scissor_state(struct pipe_context *pipe, OUT_RING (((s->maxy - s->miny) << 16) | s->miny); } -static void -nv40_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *miptree) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - - nv40->tex_miptree[unit] = miptree; - nv40->tex_dirty |= unit; - - nv40->dirty |= NV40_NEW_TEXTURE; -} - static void nv40_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) @@ -814,6 +795,7 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.create_sampler_state = nv40_sampler_state_create; nv40->pipe.bind_sampler_state = nv40_sampler_state_bind; nv40->pipe.delete_sampler_state = nv40_sampler_state_delete; + nv40->pipe.set_sampler_texture = nv40_set_sampler_texture; nv40->pipe.create_rasterizer_state = nv40_rasterizer_state_create; nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind; @@ -833,19 +815,13 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.set_blend_color = nv40_set_blend_color; nv40->pipe.set_clip_state = nv40_set_clip_state; - nv40->pipe.set_clear_color_state = nv40_set_clear_color_state; nv40->pipe.set_constant_buffer = nv40_set_constant_buffer; nv40->pipe.set_framebuffer_state = nv40_set_framebuffer_state; nv40->pipe.set_polygon_stipple = nv40_set_polygon_stipple; - nv40->pipe.set_sampler_units = nv40_set_sampler_units; nv40->pipe.set_scissor_state = nv40_set_scissor_state; - nv40->pipe.set_texture_state = nv40_set_texture_state; nv40->pipe.set_viewport_state = nv40_set_viewport_state; nv40->pipe.set_vertex_buffer = nv40_set_vertex_buffer; nv40->pipe.set_vertex_element = nv40_set_vertex_element; - -// nv40->pipe.set_feedback_state = nv40_set_feedback_state; -// nv40->pipe.set_feedback_buffer = nv40_set_feedback_buffer; } diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index b4562f26e65..f140b764cac 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -190,10 +190,11 @@ nv40_get_tex_surface(struct pipe_context *pipe, struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; struct pipe_surface *ps; - ps = ws->surface_alloc(ws, pt->format); + ps = ws->surface_alloc(ws); if (!ps) return NULL; ws->buffer_reference(ws, &ps->buffer, nv40mt->buffer); + ps->format = pt->format; ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; diff --git a/src/mesa/pipe/nv50/nv50_context.c b/src/mesa/pipe/nv50/nv50_context.c index 25ead1b5646..ed12c6c0d93 100644 --- a/src/mesa/pipe/nv50/nv50_context.c +++ b/src/mesa/pipe/nv50/nv50_context.c @@ -7,16 +7,9 @@ #include "nv50_dma.h" static boolean -nv50_is_format_supported(struct pipe_context *pipe, uint format) +nv50_is_format_supported(struct pipe_context *pipe, enum pipe_format format, + uint type) { - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - }; - return FALSE; } diff --git a/src/mesa/pipe/nv50/nv50_query.c b/src/mesa/pipe/nv50/nv50_query.c index 51cf6804d7c..69eb0e700b4 100644 --- a/src/mesa/pipe/nv50/nv50_query.c +++ b/src/mesa/pipe/nv50/nv50_query.c @@ -3,28 +3,46 @@ #include "nv50_context.h" #include "nv50_dma.h" +static struct pipe_query * +nv50_query_create(struct pipe_context *pipe, unsigned type) +{ + NOUVEAU_ERR("unimplemented\n"); + return NULL; +} + static void -nv50_query_begin(struct pipe_context *pipe, struct pipe_query_object *q) +nv50_query_destroy(struct pipe_context *pipe, struct pipe_query *q) { NOUVEAU_ERR("unimplemented\n"); } static void -nv50_query_end(struct pipe_context *pipe, struct pipe_query_object *q) +nv50_query_begin(struct pipe_context *pipe, struct pipe_query *q) { NOUVEAU_ERR("unimplemented\n"); } static void -nv50_query_wait(struct pipe_context *pipe, struct pipe_query_object *q) +nv50_query_end(struct pipe_context *pipe, struct pipe_query *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +static boolean +nv50_query_result(struct pipe_context *pipe, struct pipe_query *q, + boolean wait, uint64_t *result) { NOUVEAU_ERR("unimplemented\n"); + *result = 0xdeadcafe; + return TRUE; } void nv50_init_query_functions(struct nv50_context *nv50) { + nv50->pipe.create_query = nv50_query_create; + nv50->pipe.destroy_query = nv50_query_destroy; nv50->pipe.begin_query = nv50_query_begin; nv50->pipe.end_query = nv50_query_end; - nv50->pipe.wait_query = nv50_query_wait; + nv50->pipe.get_query_result = nv50_query_result; } diff --git a/src/mesa/pipe/nv50/nv50_state.c b/src/mesa/pipe/nv50/nv50_state.c index 778b9f5c9ac..0a27a4676eb 100644 --- a/src/mesa/pipe/nv50/nv50_state.c +++ b/src/mesa/pipe/nv50/nv50_state.c @@ -10,6 +10,7 @@ static void * nv50_alpha_test_state_create(struct pipe_context *pipe, const struct pipe_alpha_test_state *cso) { + return NULL; } static void @@ -26,6 +27,7 @@ static void * nv50_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { + return NULL; } static void @@ -42,6 +44,7 @@ static void * nv50_sampler_state_create(struct pipe_context *pipe, const struct pipe_sampler_state *cso) { + return NULL; } static void @@ -55,10 +58,17 @@ nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { } +static void +nv50_set_sampler_texture(struct pipe_context *pipe, unsigned unit, + struct pipe_texture *pt) +{ +} + static void * nv50_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { + return NULL; } static void @@ -75,6 +85,7 @@ static void * nv50_depth_stencil_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_state *cso) { + return NULL; } static void @@ -131,12 +142,6 @@ nv50_set_clip_state(struct pipe_context *pipe, { } -static void -nv50_set_clear_color_state(struct pipe_context *pipe, - const struct pipe_clear_color_state *ccol) -{ -} - static void nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, const struct pipe_constant_buffer *buf ) @@ -155,24 +160,12 @@ nv50_set_polygon_stipple(struct pipe_context *pipe, { } -static void -nv50_set_sampler_units(struct pipe_context *pipe, - uint num_samplers, const uint *units) -{ -} - static void nv50_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { } -static void -nv50_set_texture_state(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *pt) -{ -} - static void nv50_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) @@ -205,6 +198,7 @@ nv50_init_state_functions(struct nv50_context *nv50) nv50->pipe.create_sampler_state = nv50_sampler_state_create; nv50->pipe.bind_sampler_state = nv50_sampler_state_bind; nv50->pipe.delete_sampler_state = nv50_sampler_state_delete; + nv50->pipe.set_sampler_texture = nv50_set_sampler_texture; nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create; nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; @@ -224,19 +218,13 @@ nv50_init_state_functions(struct nv50_context *nv50) nv50->pipe.set_blend_color = nv50_set_blend_color; nv50->pipe.set_clip_state = nv50_set_clip_state; - nv50->pipe.set_clear_color_state = nv50_set_clear_color_state; nv50->pipe.set_constant_buffer = nv50_set_constant_buffer; nv50->pipe.set_framebuffer_state = nv50_set_framebuffer_state; nv50->pipe.set_polygon_stipple = nv50_set_polygon_stipple; - nv50->pipe.set_sampler_units = nv50_set_sampler_units; nv50->pipe.set_scissor_state = nv50_set_scissor_state; - nv50->pipe.set_texture_state = nv50_set_texture_state; nv50->pipe.set_viewport_state = nv50_set_viewport_state; nv50->pipe.set_vertex_buffer = nv50_set_vertex_buffer; nv50->pipe.set_vertex_element = nv50_set_vertex_element; - -// nv50->pipe.set_feedback_state = nv50_set_feedback_state; -// nv50->pipe.set_feedback_buffer = nv50_set_feedback_buffer; } -- cgit v1.2.3 From 9114324b0b27d9d66f2acfb6e072f027bc216406 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Dec 2007 23:31:16 +1100 Subject: nv40: small bug --- src/mesa/pipe/nv40/nv40_fragprog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 105d7937f34..bad5f0020eb 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -312,7 +312,7 @@ src_native_swz(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc, neg_mask |= (1 << c); } - if (mask == MASK_ALL) + if (mask == MASK_ALL && !neg_mask) return TRUE; *src = temp(fpc); -- cgit v1.2.3 From 805c5993a3d442c6b87733416654677e61fa36a2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Dec 2007 23:37:52 +1100 Subject: nv40: Fix bug in vtxprog + a few tweaks --- src/mesa/pipe/nv40/nv40_vertprog.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index 66722bafa58..b6ebaee2f2a 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -184,7 +184,7 @@ emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) hw[3] |= (dst.index << NV40_VP_INST_DEST_SHIFT); if (slot == 0) { hw[0] |= NV40_VP_INST_VEC_RESULT; - hw[0] |= NV40_VP_INST_VEC_DEST_TEMP_MASK; + hw[0] |= NV40_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); } else { hw[3] |= NV40_VP_INST_SCA_RESULT; hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; @@ -212,12 +212,11 @@ nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op, if (slot == 0) { hw[1] |= (op << NV40_VP_INST_VEC_OPCODE_SHIFT); - hw[3] |= (NV40_VP_INST_SCA_RESULT | - NV40_VP_INST_SCA_DEST_TEMP_MASK); + hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; hw[3] |= (mask << NV40_VP_INST_VEC_WRITEMASK_SHIFT); } else { hw[1] |= (op << NV40_VP_INST_SCA_OPCODE_SHIFT); - hw[1] |= (NV40_VP_INST_VEC_DEST_TEMP_MASK | (1 << 20)); + hw[0] |= (NV40_VP_INST_VEC_DEST_TEMP_MASK | (1 << 20)); hw[3] |= (mask << NV40_VP_INST_SCA_WRITEMASK_SHIFT); } -- cgit v1.2.3 From 1fb97368989184c26fe37e98ed6916e6793a0171 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 13 Dec 2007 01:04:12 +1100 Subject: nouveau: take jumps into account properly when debugging pushbuf --- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 71 ++++++++++++++++++----- 1 file changed, 58 insertions(+), 13 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 5739e0010d4..0c04a59c744 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -152,6 +152,63 @@ nouveau_dma_subc_bind(struct nouveau_grobj *grobj) } #endif +#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF +static void +nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + unsigned mthd_count = 0; + + while (get != put) { + uint32_t gpuget = (get << 2) + nvchan->dma.base; + uint32_t data; + + if (get < 0 || get >= nvchan->drm.cmdbuf_size) { + NOUVEAU_ERR("DMA_PT 0x%08x\n", gpuget); + assert(0); + } + data = nvchan->pushbuf[get++]; + + if (mthd_count) { + NOUVEAU_MSG("0x%08x 0x%08x\n", gpuget, data); + mthd_count--; + continue; + } + + switch (data & 0x60000000) { + case 0x00000000: + mthd_count = (data >> 18) & 0x7ff; + NOUVEAU_MSG("0x%08x 0x%08x MTHD " + "Sc %d Mthd 0x%04x Size %d\n", + gpuget, data, (data>>13) & 7, data & 0x1ffc, + mthd_count); + break; + case 0x20000000: + get = (data & 0x1ffffffc) >> 2; + NOUVEAU_MSG("0x%08x 0x%08x JUMP 0x%08x\n", + gpuget, data, data & 0x1ffffffc); + continue; + case 0x40000000: + mthd_count = (data >> 18) & 0x7ff; + NOUVEAU_MSG("0x%08x 0x%08x NINC " + "Sc %d Mthd 0x%04x Size %d\n", + gpuget, data, (data>>13) & 7, data & 0x1ffc, + mthd_count); + break; + case 0x60000000: + /* DMA_OPCODE_CALL apparently, doesn't seem to work on + * my NV40 at least.. + */ + /* fall-through */ + default: + NOUVEAU_MSG("DMA_PUSHER 0x%08x 0x%08x\n", + gpuget, data); + assert(0); + } + } +} +#endif + void nouveau_dma_kickoff(struct nouveau_channel *userchan) { @@ -168,19 +225,7 @@ nouveau_dma_kickoff(struct nouveau_channel *userchan) #endif #ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF - for (i = chan->dma.put; i < chan->dma.cur; i++) { - NOUVEAU_MSG("0x%08x 0x%08x\n", (i<<2)+chan->dma.base, - chan->pushbuf[i]); - if ((chan->pushbuf[i] & 0xf0000000) == 0x20000000) { - int n = (((chan->pushbuf[i] & 0x0fffffff) - - chan->dma.base) / 4); - do { - NOUVEAU_MSG("\t0x%08x 0x%08x\n", - (n<<2)+chan->dma.base, - chan->pushbuf[n]); - } while ((chan->pushbuf[n++]&0xf0000000) != 0x20000000); - } - } + nouveau_dma_parse_pushbuf(userchan, chan->dma.put, chan->dma.cur); #endif WRITE_PUT(chan, chan->dma.cur); -- cgit v1.2.3 From 72d47f1949f99231f18cc5144b1c45c58b35f890 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 13 Dec 2007 11:55:35 +1100 Subject: nouveau: more adaption.. --- .../drivers/dri/nouveau_winsys/nouveau_screen.c | 21 ++++++++++++++++++++- src/mesa/pipe/nv40/nv40_context.c | 1 - src/mesa/pipe/nv40/nv40_surface.c | 5 +++-- 3 files changed, 23 insertions(+), 4 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index a1eed2e44aa..5a957f48eb8 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -76,6 +76,7 @@ nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, const __GLcontextModes *glVis, GLboolean pixmapBuffer) { struct nouveau_framebuffer *nvfb; + enum pipe_format colour, depth, stencil; if (pixmapBuffer) return GL_FALSE; @@ -84,7 +85,25 @@ nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, if (!nvfb) return GL_FALSE; - nvfb->stfb = st_create_framebuffer(glVis, GL_TRUE, (void*)nvfb); + if (glVis->redBits == 5) + colour = PIPE_FORMAT_R5G6B5_UNORM; + else + colour = PIPE_FORMAT_A8R8G8B8_UNORM; + + if (glVis->depthBits == 16) + depth = PIPE_FORMAT_Z16_UNORM; + else if (glVis->depthBits == 24) + depth = PIPE_FORMAT_Z24S8_UNORM; + else + depth = PIPE_FORMAT_NONE; + + if (glVis->stencilBits == 8) + stencil = PIPE_FORMAT_Z24S8_UNORM; + else + stencil = PIPE_FORMAT_NONE; + + nvfb->stfb = st_create_framebuffer(glVis, GL_TRUE, colour, depth, + stencil, (void*)nvfb); if (!nvfb->stfb) { free(nvfb); return GL_FALSE; diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index b1358a8cfa5..e7d0a186378 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -12,7 +12,6 @@ nv40_is_format_supported(struct pipe_context *pipe, enum pipe_format format, { switch (type) { case PIPE_SURFACE: - case PIPE_SCREEN_SURFACE: switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index f140b764cac..da9e18ac136 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -31,6 +31,7 @@ #include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" +#include "pipe/softpipe/sp_rgba_tile.h" #define CLIP_TILE \ @@ -254,8 +255,8 @@ nv40_init_surface_functions(struct nv40_context *nv40) nv40->pipe.get_tex_surface = nv40_get_tex_surface; nv40->pipe.get_tile = nv40_get_tile; nv40->pipe.put_tile = nv40_put_tile; - nv40->pipe.get_tile_rgba = nv40_get_tile_rgba; - nv40->pipe.put_tile_rgba = nv40_put_tile_rgba; + nv40->pipe.get_tile_rgba = softpipe_get_tile_rgba; + nv40->pipe.put_tile_rgba = softpipe_put_tile_rgba; nv40->pipe.surface_data = nv40_surface_data; nv40->pipe.surface_copy = nv40_surface_copy; nv40->pipe.surface_fill = nv40_surface_fill; -- cgit v1.2.3 From 00f0f0d0a4b474757becb99109cb2dd6574178f8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 14 Dec 2007 01:14:23 +1100 Subject: nouveau: mass renaming for consistency, no functional changes --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 194 ++++++++++----------- .../drivers/dri/nouveau_winsys/nouveau_channel.c | 105 ++++++----- .../drivers/dri/nouveau_winsys/nouveau_device.c | 68 ++++---- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 85 ++++----- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h | 72 ++++---- .../drivers/dri/nouveau_winsys/nouveau_grobj.c | 84 ++++----- .../drivers/dri/nouveau_winsys/nouveau_notifier.c | 79 ++++----- 7 files changed, 342 insertions(+), 345 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index d684ab4d7c4..d31026b52d5 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -29,62 +29,62 @@ #include "nouveau_local.h" int -nouveau_bo_init(struct nouveau_device *userdev) +nouveau_bo_init(struct nouveau_device *dev) { return 0; } void -nouveau_bo_takedown(struct nouveau_device *userdev) +nouveau_bo_takedown(struct nouveau_device *dev) { } static int -nouveau_bo_realloc_gpu(struct nouveau_bo_priv *bo, uint32_t flags, int size) +nouveau_bo_realloc_gpu(struct nouveau_bo_priv *nvbo, uint32_t flags, int size) { - struct nouveau_device_priv *nv = nouveau_device(bo->base.device); + struct nouveau_device_priv *nvdev = nouveau_device(nvbo->base.device); int ret; - if (bo->drm.size && bo->drm.size != size) { + if (nvbo->drm.size && nvbo->drm.size != size) { struct drm_nouveau_mem_free f; - if (bo->map) { - drmUnmap(bo->map, bo->drm.size); - bo->map = NULL; + if (nvbo->map) { + drmUnmap(nvbo->map, nvbo->drm.size); + nvbo->map = NULL; } - f.flags = bo->drm.flags; - f.offset = bo->drm.offset; - drmCommandWrite(nv->fd, DRM_NOUVEAU_MEM_FREE, &f, sizeof(f)); + f.flags = nvbo->drm.flags; + f.offset = nvbo->drm.offset; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, &f, sizeof(f)); - bo->drm.size = 0; + nvbo->drm.size = 0; } - if (size && !bo->drm.size) { + if (size && !nvbo->drm.size) { if (flags) { - bo->drm.flags = 0; + nvbo->drm.flags = 0; if (flags & NOUVEAU_BO_VRAM) - bo->drm.flags |= NOUVEAU_MEM_FB; + nvbo->drm.flags |= NOUVEAU_MEM_FB; if (flags & NOUVEAU_BO_GART) - bo->drm.flags |= (NOUVEAU_MEM_AGP | - NOUVEAU_MEM_PCI); - bo->drm.flags |= NOUVEAU_MEM_MAPPED; + nvbo->drm.flags |= (NOUVEAU_MEM_AGP | + NOUVEAU_MEM_PCI); + nvbo->drm.flags |= NOUVEAU_MEM_MAPPED; } - bo->drm.size = size; + nvbo->drm.size = size; - ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_MEM_ALLOC, - &bo->drm, sizeof(bo->drm)); + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, + &nvbo->drm, sizeof(nvbo->drm)); if (ret) { - free(bo); + free(nvbo); return ret; } - ret = drmMap(nv->fd, bo->drm.map_handle, bo->drm.size, - &bo->map); + ret = drmMap(nvdev->fd, nvbo->drm.map_handle, nvbo->drm.size, + &nvbo->map); if (ret) { - bo->map = NULL; - nouveau_bo_del((void *)&bo); + nvbo->map = NULL; + nouveau_bo_del((void *)&nvbo); return ret; } } @@ -93,154 +93,154 @@ nouveau_bo_realloc_gpu(struct nouveau_bo_priv *bo, uint32_t flags, int size) } int -nouveau_bo_new(struct nouveau_device *userdev, uint32_t flags, int align, - int size, struct nouveau_bo **userbo) +nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, + int size, struct nouveau_bo **bo) { - struct nouveau_bo_priv *bo; + struct nouveau_bo_priv *nvbo; int ret; - if (!userdev || !userbo || *userbo) + if (!dev || !bo || *bo) return -EINVAL; - bo = calloc(1, sizeof(*bo)); - if (!bo) + nvbo = calloc(1, sizeof(struct nouveau_bo_priv)); + if (!nvbo) return -ENOMEM; - bo->base.device = userdev; - bo->drm.alignment = align; + nvbo->base.device = dev; + nvbo->drm.alignment = align; if (flags & NOUVEAU_BO_PIN) { - ret = nouveau_bo_realloc_gpu(bo, flags, size); + ret = nouveau_bo_realloc_gpu(nvbo, flags, size); if (ret) { - free(bo); + free(nvbo); return ret; } } else { - bo->sysmem = malloc(size); - if (!bo->sysmem) { - free(bo); + nvbo->sysmem = malloc(size); + if (!nvbo->sysmem) { + free(nvbo); return -ENOMEM; } } - bo->base.size = size; - bo->base.offset = bo->drm.offset; - bo->base.handle = (unsigned long)bo; - bo->refcount = 1; - *userbo = &bo->base; + nvbo->base.size = size; + nvbo->base.offset = nvbo->drm.offset; + nvbo->base.handle = bo_to_ptr(nvbo); + nvbo->refcount = 1; + *bo = &nvbo->base; return 0; } int -nouveau_bo_user(struct nouveau_device *userdev, void *ptr, int size, - struct nouveau_bo **userbo) +nouveau_bo_user(struct nouveau_device *dev, void *ptr, int size, + struct nouveau_bo **bo) { - struct nouveau_bo_priv *bo; + struct nouveau_bo_priv *nvbo; - if (!userdev || !userbo || *userbo) + if (!dev || !bo || *bo) return -EINVAL; - bo = calloc(1, sizeof(*bo)); - if (!bo) + nvbo = calloc(1, sizeof(*nvbo)); + if (!nvbo) return -ENOMEM; - bo->base.device = userdev; + nvbo->base.device = dev; - bo->sysmem = ptr; - bo->user = 1; - - bo->base.size = size; - bo->base.offset = bo->drm.offset; - bo->base.handle = (unsigned long)bo; - bo->refcount = 1; - *userbo = &bo->base; + nvbo->sysmem = ptr; + nvbo->user = 1; + + nvbo->base.size = size; + nvbo->base.offset = nvbo->drm.offset; + nvbo->base.handle = bo_to_ptr(nvbo); + nvbo->refcount = 1; + *bo = &nvbo->base; return 0; } int -nouveau_bo_ref(struct nouveau_device *userdev, uint64_t handle, - struct nouveau_bo **userbo) +nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, + struct nouveau_bo **bo) { - struct nouveau_bo_priv *bo = (void *)(unsigned long)handle; + struct nouveau_bo_priv *nvbo = ptr_to_bo(handle); - if (!userdev || !userbo || *userbo) + if (!dev || !bo || *bo) return -EINVAL; - bo->refcount++; - *userbo = &bo->base; + nvbo->refcount++; + *bo = &nvbo->base; return 0; } int -nouveau_bo_resize(struct nouveau_bo *userbo, int size) +nouveau_bo_resize(struct nouveau_bo *bo, int size) { - struct nouveau_bo_priv *bo = nouveau_bo(userbo); + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); int ret; - if (!bo || bo->user) + if (!nvbo || nvbo->user) return -EINVAL; - if (bo->sysmem) { - bo->sysmem = realloc(bo->sysmem, size); - if (!bo->sysmem) + if (nvbo->sysmem) { + nvbo->sysmem = realloc(nvbo->sysmem, size); + if (!nvbo->sysmem) return -ENOMEM; } else { - ret = nouveau_bo_realloc_gpu(bo, 0, size); + ret = nouveau_bo_realloc_gpu(nvbo, 0, size); if (ret) return ret; } - bo->base.size = size; + nvbo->base.size = size; return 0; } void -nouveau_bo_del(struct nouveau_bo **userbo) +nouveau_bo_del(struct nouveau_bo **bo) { - struct nouveau_bo_priv *bo; + struct nouveau_bo_priv *nvbo; - if (!userbo || !*userbo) + if (!bo || !*bo) return; - bo = nouveau_bo(*userbo); - *userbo = NULL; + nvbo = nouveau_bo(*bo); + *bo = NULL; - if (--bo->refcount) + if (--nvbo->refcount) return; - if (bo->fence) - nouveau_fence_wait(&bo->fence); + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); - nouveau_bo_realloc_gpu(bo, 0, 0); - if (bo->sysmem && !bo->user) - free(bo->sysmem); - free(bo); + nouveau_bo_realloc_gpu(nvbo, 0, 0); + if (nvbo->sysmem && !nvbo->user) + free(nvbo->sysmem); + free(nvbo); } int -nouveau_bo_map(struct nouveau_bo *userbo, uint32_t flags) +nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) { - struct nouveau_bo_priv *bo = nouveau_bo(userbo); + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - if (!bo) + if (!nvbo) return -EINVAL; - if (bo->sysmem) - userbo->map = bo->sysmem; + if (nvbo->sysmem) + bo->map = nvbo->sysmem; else - userbo->map = bo->map; + bo->map = nvbo->map; return 0; } void -nouveau_bo_unmap(struct nouveau_bo *userbo) +nouveau_bo_unmap(struct nouveau_bo *bo) { - userbo->map = NULL; + bo->map = NULL; } static int -nouveau_bo_upload(struct nouveau_bo_priv *bo) +nouveau_bo_upload(struct nouveau_bo_priv *nvbo) { - if (bo->fence) - nouveau_fence_wait(&bo->fence); - memcpy(bo->map, bo->sysmem, bo->drm.size); + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + memcpy(nvbo->map, nvbo->sysmem, nvbo->drm.size); return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c index c4558e5573a..df80d04add5 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c @@ -28,96 +28,91 @@ #include "nouveau_dma.h" int -nouveau_channel_alloc(struct nouveau_device *userdev, uint32_t fb_ctxdma, - uint32_t tt_ctxdma, struct nouveau_channel **userchan) +nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma, + uint32_t tt_ctxdma, struct nouveau_channel **chan) { - struct nouveau_device_priv *nv = nouveau_device(userdev); - struct nouveau_channel_priv *chan; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct nouveau_channel_priv *nvchan; int ret; - if (!nv || !userchan || *userchan) + if (!nvdev || !chan || *chan) return -EINVAL; - chan = calloc(1, sizeof(*chan)); - if (!chan) + nvchan = calloc(1, sizeof(struct nouveau_channel_priv)); + if (!nvchan) return -ENOMEM; - chan->base.device = userdev; + nvchan->base.device = dev; - chan->drm.fb_ctxdma_handle = fb_ctxdma; - chan->drm.tt_ctxdma_handle = tt_ctxdma; - ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_CHANNEL_ALLOC, - &chan->drm, sizeof(chan->drm)); + nvchan->drm.fb_ctxdma_handle = fb_ctxdma; + nvchan->drm.tt_ctxdma_handle = tt_ctxdma; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_CHANNEL_ALLOC, + &nvchan->drm, sizeof(nvchan->drm)); if (ret) { - free(chan); + free(nvchan); return ret; } - chan->base.id = chan->drm.channel; - if (nouveau_grobj_ref(&chan->base, chan->drm.fb_ctxdma_handle, - &chan->base.vram) || - nouveau_grobj_ref(&chan->base, chan->drm.tt_ctxdma_handle, - &chan->base.gart)) { - nouveau_channel_free((void *)&chan); + nvchan->base.id = nvchan->drm.channel; + if (nouveau_grobj_ref(&nvchan->base, nvchan->drm.fb_ctxdma_handle, + &nvchan->base.vram) || + nouveau_grobj_ref(&nvchan->base, nvchan->drm.tt_ctxdma_handle, + &nvchan->base.gart)) { + nouveau_channel_free((void *)&nvchan); return -EINVAL; } - ret = drmMap(nv->fd, chan->drm.ctrl, chan->drm.ctrl_size, - (void*)&chan->user); + ret = drmMap(nvdev->fd, nvchan->drm.ctrl, nvchan->drm.ctrl_size, + (void*)&nvchan->user); if (ret) { - nouveau_channel_free((void *)&chan); + nouveau_channel_free((void *)&nvchan); return ret; } - chan->put = &chan->user[0x40/4]; - chan->get = &chan->user[0x44/4]; - chan->ref_cnt = &chan->user[0x48/4]; + nvchan->put = &nvchan->user[0x40/4]; + nvchan->get = &nvchan->user[0x44/4]; + nvchan->ref_cnt = &nvchan->user[0x48/4]; - ret = drmMap(nv->fd, chan->drm.notifier, chan->drm.notifier_size, - (drmAddressPtr)&chan->notifier_block); + ret = drmMap(nvdev->fd, nvchan->drm.notifier, nvchan->drm.notifier_size, + (drmAddressPtr)&nvchan->notifier_block); if (ret) { - nouveau_channel_free((void *)&chan); + nouveau_channel_free((void *)&nvchan); return ret; } - ret = drmMap(nv->fd, chan->drm.cmdbuf, chan->drm.cmdbuf_size, - (void*)&chan->pushbuf); + ret = drmMap(nvdev->fd, nvchan->drm.cmdbuf, nvchan->drm.cmdbuf_size, + (void*)&nvchan->pushbuf); if (ret) { - nouveau_channel_free((void *)&chan); + nouveau_channel_free((void *)&nvchan); return ret; } - nouveau_dma_channel_init(&chan->base); - nouveau_pushbuf_init(&chan->base); + nouveau_dma_channel_init(&nvchan->base); + nouveau_pushbuf_init(&nvchan->base); - *userchan = &chan->base; + *chan = &nvchan->base; return 0; } void -nouveau_channel_free(struct nouveau_channel **userchan) +nouveau_channel_free(struct nouveau_channel **chan) { - struct nouveau_channel_priv *chan; - - if (!userchan) - return; - chan = nouveau_channel(*userchan); - - if (chan) { - struct nouveau_device_priv *nv; - struct drm_nouveau_channel_free cf; - - nv = nouveau_device((*userchan)->device); - *userchan = NULL; + struct nouveau_channel_priv *nvchan; + struct nouveau_device_priv *nvdev; + struct drm_nouveau_channel_free cf; - FIRE_RING_CH(&chan->base); + if (!chan || !*chan) + return; + nvchan = nouveau_channel(*chan); + *chan = NULL; + nvdev = nouveau_device(nvchan->base.device); + + FIRE_RING_CH(&nvchan->base); - nouveau_grobj_free(&chan->base.vram); - nouveau_grobj_free(&chan->base.gart); + nouveau_grobj_free(&nvchan->base.vram); + nouveau_grobj_free(&nvchan->base.gart); - cf.channel = chan->drm.channel; - drmCommandWrite(nv->fd, DRM_NOUVEAU_CHANNEL_FREE, - &cf, sizeof(cf)); - free(chan); - } + cf.channel = nvchan->drm.channel; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf)); + free(nvchan); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c index a3f55135504..409e4415f76 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c @@ -27,40 +27,40 @@ #include "nouveau_drmif.h" int -nouveau_device_open_existing(struct nouveau_device **userdev, int close, +nouveau_device_open_existing(struct nouveau_device **dev, int close, int fd, drm_context_t ctx) { - struct nouveau_device_priv *nv; + struct nouveau_device_priv *nvdev; int ret; - if (!userdev || *userdev) + if (!dev || *dev) return -EINVAL; - nv = calloc(1, sizeof(*nv)); - if (!nv) + nvdev = calloc(1, sizeof(*nvdev)); + if (!nvdev) return -ENOMEM; - nv->fd = fd; - nv->ctx = ctx; - nv->needs_close = close; + nvdev->fd = fd; + nvdev->ctx = ctx; + nvdev->needs_close = close; - drmCommandNone(nv->fd, DRM_NOUVEAU_CARD_INIT); + drmCommandNone(nvdev->fd, DRM_NOUVEAU_CARD_INIT); - if ((ret = nouveau_bo_init(&nv->base))) { - nouveau_device_close((void *)&nv); + if ((ret = nouveau_bo_init(&nvdev->base))) { + nouveau_device_close((void *)&nvdev); return ret; } - *userdev = &nv->base; + *dev = &nvdev->base; return 0; } int -nouveau_device_open(struct nouveau_device **userdev, const char *busid) +nouveau_device_open(struct nouveau_device **dev, const char *busid) { drm_context_t ctx; int fd, ret; - if (!userdev || *userdev) + if (!dev || *dev) return -EINVAL; fd = drmOpen("nouveau", busid); @@ -73,7 +73,7 @@ nouveau_device_open(struct nouveau_device **userdev, const char *busid) return ret; } - ret = nouveau_device_open_existing(userdev, 1, fd, ctx); + ret = nouveau_device_open_existing(dev, 1, fd, ctx); if (ret) { drmDestroyContext(fd, ctx); drmClose(fd); @@ -84,37 +84,38 @@ nouveau_device_open(struct nouveau_device **userdev, const char *busid) } void -nouveau_device_close(struct nouveau_device **userdev) +nouveau_device_close(struct nouveau_device **dev) { - struct nouveau_device_priv *nv; + struct nouveau_device_priv *nvdev; - if (userdev || !*userdev) + if (dev || !*dev) return; - nv = (struct nouveau_device_priv *)*userdev; - *userdev = NULL; + nvdev = nouveau_device(*dev); + *dev = NULL; - nouveau_bo_takedown(&nv->base); + nouveau_bo_takedown(&nvdev->base); - if (nv->needs_close) { - drmDestroyContext(nv->fd, nv->ctx); - drmClose(nv->fd); + if (nvdev->needs_close) { + drmDestroyContext(nvdev->fd, nvdev->ctx); + drmClose(nvdev->fd); } - free(nv); + free(nvdev); } int -nouveau_device_get_param(struct nouveau_device *userdev, +nouveau_device_get_param(struct nouveau_device *dev, uint64_t param, uint64_t *value) { - struct nouveau_device_priv *nv = (struct nouveau_device_priv *)userdev; + struct nouveau_device_priv *nvdev = nouveau_device(dev); struct drm_nouveau_getparam g; int ret; - if (!nv || !value) + if (!nvdev || !value) return -EINVAL; g.param = param; - ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_GETPARAM, &g, sizeof(g)); + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_GETPARAM, + &g, sizeof(g)); if (ret) return ret; @@ -123,19 +124,20 @@ nouveau_device_get_param(struct nouveau_device *userdev, } int -nouveau_device_set_param(struct nouveau_device *userdev, +nouveau_device_set_param(struct nouveau_device *dev, uint64_t param, uint64_t value) { - struct nouveau_device_priv *nv = (struct nouveau_device_priv *)userdev; + struct nouveau_device_priv *nvdev = nouveau_device(dev); struct drm_nouveau_setparam s; int ret; - if (!nv) + if (!nvdev) return -EINVAL; s.param = param; s.value = value; - ret = drmCommandWriteRead(nv->fd, DRM_NOUVEAU_SETPARAM, &s, sizeof(s)); + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_SETPARAM, + &s, sizeof(s)); if (ret) return ret; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 0c04a59c744..b89f1b5cf5f 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -53,18 +53,18 @@ WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) } void -nouveau_dma_channel_init(struct nouveau_channel *userchan) +nouveau_dma_channel_init(struct nouveau_channel *chan) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); int i; - chan->dma.base = chan->drm.put_base; - chan->dma.cur = chan->dma.put = RING_SKIPS; - chan->dma.max = (chan->drm.cmdbuf_size >> 2) - 2; - chan->dma.free = chan->dma.max - chan->dma.cur; + nvchan->dma.base = nvchan->drm.put_base; + nvchan->dma.cur = nvchan->dma.put = RING_SKIPS; + nvchan->dma.max = (nvchan->drm.cmdbuf_size >> 2) - 2; + nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; for (i = 0; i < RING_SKIPS; i++) - chan->pushbuf[i] = 0x00000000; + nvchan->pushbuf[i] = 0x00000000; } #define CHECK_TIMEOUT() do { \ @@ -75,49 +75,51 @@ nouveau_dma_channel_init(struct nouveau_channel *userchan) #define IN_MASTER_RING(chan, ptr) ((ptr) <= (chan)->dma.max) int -nouveau_dma_wait(struct nouveau_channel *userchan, int size) +nouveau_dma_wait(struct nouveau_channel *chan, int size) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); uint32_t get, t_start; - FIRE_RING_CH(userchan); + FIRE_RING_CH(chan); t_start = NOUVEAU_TIME_MSEC(); - while (chan->dma.free < size) { + while (nvchan->dma.free < size) { CHECK_TIMEOUT(); - get = READ_GET(chan); - if (!IN_MASTER_RING(chan, get)) + get = READ_GET(nvchan); + if (!IN_MASTER_RING(nvchan, get)) continue; - if (chan->dma.put >= get) { - chan->dma.free = chan->dma.max - chan->dma.cur; + if (nvchan->dma.put >= get) { + nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; - if (chan->dma.free < size) { + if (nvchan->dma.free < size) { #ifdef NOUVEAU_DMA_DEBUG - chan->dma.push_free = 1; + nvchan->dma.push_free = 1; #endif - OUT_RING_CH(userchan, - 0x20000000 | chan->dma.base); + OUT_RING_CH(chan, + 0x20000000 | nvchan->dma.base); if (get <= RING_SKIPS) { /*corner case - will be idle*/ - if (chan->dma.put <= RING_SKIPS) - WRITE_PUT(chan, RING_SKIPS + 1); + if (nvchan->dma.put <= RING_SKIPS) + WRITE_PUT(nvchan, + RING_SKIPS + 1); do { CHECK_TIMEOUT(); - get = READ_GET(chan); - if (!IN_MASTER_RING(chan, get)) + get = READ_GET(nvchan); + if (!IN_MASTER_RING(nvchan, + get)) continue; } while (get <= RING_SKIPS); } - WRITE_PUT(chan, RING_SKIPS); - chan->dma.cur = chan->dma.put = RING_SKIPS; - chan->dma.free = get - (RING_SKIPS + 1); + WRITE_PUT(nvchan, RING_SKIPS); + nvchan->dma.cur = nvchan->dma.put = RING_SKIPS; + nvchan->dma.free = get - (RING_SKIPS + 1); } } else { - chan->dma.free = get - chan->dma.cur - 1; + nvchan->dma.free = get - nvchan->dma.cur - 1; } } @@ -128,22 +130,22 @@ nouveau_dma_wait(struct nouveau_channel *userchan, int size) void nouveau_dma_subc_bind(struct nouveau_grobj *grobj) { - struct nouveau_channel_priv *chan = nouveau_channel(grobj->channel); + struct nouveau_channel_priv *nvchan = nouveau_channel(grobj->channel); int subc = -1, i; for (i = 0; i < 8; i++) { - if (chan->subchannel[i].grobj && - chan->subchannel[i].grobj->bound == + if (nvchan->subchannel[i].grobj && + nvchan->subchannel[i].grobj->bound == NOUVEAU_GROBJ_EXPLICIT_BIND) continue; - if (chan->subchannel[i].seq < chan->subchannel[subc].seq) + if (nvchan->subchannel[i].seq < nvchan->subchannel[subc].seq) subc = i; } assert(subc >= 0); - if (chan->subchannel[subc].grobj) - chan->subchannel[subc].grobj->bound = 0; - chan->subchannel[subc].grobj = grobj; + if (nvchan->subchannel[subc].grobj) + nvchan->subchannel[subc].grobj->bound = 0; + nvchan->subchannel[subc].grobj = grobj; grobj->subc = subc; grobj->bound = NOUVEAU_GROBJ_BOUND; @@ -210,23 +212,24 @@ nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) #endif void -nouveau_dma_kickoff(struct nouveau_channel *userchan) +nouveau_dma_kickoff(struct nouveau_channel *chan) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - if (chan->dma.cur == chan->dma.put) + if (nvchan->dma.cur == nvchan->dma.put) return; #ifdef NOUVEAU_DMA_DEBUG - if (chan->dma.push_free) { - NOUVEAU_ERR("Packet incomplete: %d left\n", chan->dma.push_free); + if (nvchan->dma.push_free) { + NOUVEAU_ERR("Packet incomplete: %d left\n", + nvchan->dma.push_free); return; } #endif #ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF - nouveau_dma_parse_pushbuf(userchan, chan->dma.put, chan->dma.cur); + nouveau_dma_parse_pushbuf(chan, nvchan->dma.put, nvchan->dma.cur); #endif - WRITE_PUT(chan, chan->dma.cur); + WRITE_PUT(nvchan, nvchan->dma.cur); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h index 927841c4eb4..d4dd3054f3d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h @@ -39,112 +39,112 @@ static char faulty[1024]; #endif static inline void -nouveau_dma_out(struct nouveau_channel *userchan, uint32_t data) +nouveau_dma_out(struct nouveau_channel *chan, uint32_t data) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); #ifdef NOUVEAU_DMA_DEBUG - if (chan->dma.push_free == 0) { + if (nvchan->dma.push_free == 0) { NOUVEAU_ERR("No space left in packet. Error at %s\n",faulty); return; } - chan->dma.push_free--; + nvchan->dma.push_free--; #endif #ifdef NOUVEAU_DMA_TRACE { - uint32_t offset = (chan->dma.cur << 2) + chan->dma.base; + uint32_t offset = (nvchan->dma.cur << 2) + nvchan->dma.base; NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", - chan->drm.channel, offset, data); + nvchan->drm.channel, offset, data); } #endif - chan->pushbuf[chan->dma.cur++] = data; + nvchan->pushbuf[nvchan->dma.cur++] = data; } static inline void -nouveau_dma_outp(struct nouveau_channel *userchan, uint32_t *ptr, int size) +nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); (void)chan; #ifdef NOUVEAU_DMA_DEBUG - if (chan->dma.push_free < size) { + if (nvchan->dma.push_free < size) { NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", - chan->dma.push_free, size); + nvchan->dma.push_free, size); return; } #endif #ifdef NOUVEAU_DMA_TRACE while (size--) { - nouveau_dma_out(userchan, *ptr); + nouveau_dma_out(chan, *ptr); ptr++; } #else - memcpy(&chan->pushbuf[chan->dma.cur], ptr, size << 2); + memcpy(&nvchan->pushbuf[nvchan->dma.cur], ptr, size << 2); #ifdef NOUVEAU_DMA_DEBUG - chan->dma.push_free -= size; + nvchan->dma.push_free -= size; #endif - chan->dma.cur += size; + nvchan->dma.cur += size; #endif } static inline void -nouveau_dma_begin(struct nouveau_channel *userchan, struct nouveau_grobj *grobj, +nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, int method, int size, const char* file, int line) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); int push_size = size + 1; #ifdef NOUVEAU_DMA_SUBCHAN_LRU if (grobj->bound == NOUVEAU_GROBJ_UNBOUND) nouveau_dma_subc_bind(grobj); - chan->subchannel[grobj->subc].seq = chan->subc_sequence++; + nvchan->subchannel[grobj->subc].seq = nvchan->subc_sequence++; #endif #ifdef NOUVEAU_DMA_TRACE - NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", chan->drm.channel, + NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, grobj->handle, grobj->subc, method, size); #endif #ifdef NOUVEAU_DMA_DEBUG - if (chan->dma.push_free) { + if (nvchan->dma.push_free) { NOUVEAU_ERR("Previous packet incomplete: %d left. Error at %s\n", - chan->dma.push_free,faulty); + nvchan->dma.push_free,faulty); return; } sprintf(faulty,"%s:%d",file,line); #endif - if (chan->dma.free < push_size) { - if (nouveau_dma_wait(userchan, push_size) && - userchan->hang_notify) { - userchan->hang_notify(userchan); + if (nvchan->dma.free < push_size) { + if (nouveau_dma_wait(chan, push_size) && + chan->hang_notify) { + chan->hang_notify(chan); } } - chan->dma.free -= push_size; + nvchan->dma.free -= push_size; #ifdef NOUVEAU_DMA_DEBUG - chan->dma.push_free = push_size; + nvchan->dma.push_free = push_size; #endif - nouveau_dma_out(userchan, (size << 18) | (grobj->subc << 13) | method); + nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); } static inline void -nouveau_dma_bind(struct nouveau_channel *userchan, struct nouveau_grobj *grobj, +nouveau_dma_bind(struct nouveau_channel *chan, struct nouveau_grobj *grobj, int subc) { - struct nouveau_channel_priv *chan = nouveau_channel(userchan); + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - if (chan->subchannel[subc].grobj == grobj) + if (nvchan->subchannel[subc].grobj == grobj) return; - if (chan->subchannel[subc].grobj) - chan->subchannel[subc].grobj->bound = NOUVEAU_GROBJ_UNBOUND; - chan->subchannel[subc].grobj = grobj; + if (nvchan->subchannel[subc].grobj) + nvchan->subchannel[subc].grobj->bound = NOUVEAU_GROBJ_UNBOUND; + nvchan->subchannel[subc].grobj = grobj; grobj->subc = subc; grobj->bound = NOUVEAU_GROBJ_EXPLICIT_BIND; - nouveau_dma_begin(userchan, grobj, 0x0000, 1, __FUNCTION__, __LINE__); - nouveau_dma_out (userchan, grobj->handle); + nouveau_dma_begin(chan, grobj, 0x0000, 1, __FUNCTION__, __LINE__); + nouveau_dma_out (chan, grobj->handle); } #define BIND_RING_CH(ch,gr,sc) nouveau_dma_bind((ch), (gr), (sc)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c index 8dab2023956..55dfeb99aa7 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c @@ -26,82 +26,82 @@ #include "nouveau_drmif.h" int -nouveau_grobj_alloc(struct nouveau_channel *userchan, uint32_t handle, - int class, struct nouveau_grobj **usergrobj) +nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle, + int class, struct nouveau_grobj **grobj) { - struct nouveau_device_priv *nv = nouveau_device(userchan->device); - struct nouveau_grobj_priv *gr; + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_grobj_priv *nvgrobj; struct drm_nouveau_grobj_alloc g; int ret; - if (!nv || !usergrobj || *usergrobj) + if (!nvdev || !grobj || *grobj) return -EINVAL; - gr = calloc(1, sizeof(*gr)); - if (!gr) + nvgrobj = calloc(1, sizeof(*nvgrobj)); + if (!nvgrobj) return -ENOMEM; - gr->base.channel = userchan; - gr->base.handle = handle; - gr->base.grclass = class; + nvgrobj->base.channel = chan; + nvgrobj->base.handle = handle; + nvgrobj->base.grclass = class; - g.channel = userchan->id; + g.channel = chan->id; g.handle = handle; g.class = class; - ret = drmCommandWrite(nv->fd, DRM_NOUVEAU_GROBJ_ALLOC, &g, sizeof(g)); + ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC, + &g, sizeof(g)); if (ret) { - nouveau_grobj_free((void *)&gr); + nouveau_grobj_free((void *)&grobj); return ret; } - *usergrobj = &gr->base; + *grobj = &nvgrobj->base; return 0; } int -nouveau_grobj_ref(struct nouveau_channel *userchan, uint32_t handle, - struct nouveau_grobj **usergr) +nouveau_grobj_ref(struct nouveau_channel *chan, uint32_t handle, + struct nouveau_grobj **grobj) { - struct nouveau_grobj_priv *gr; + struct nouveau_grobj_priv *nvgrobj; - if (!userchan || !usergr || *usergr) + if (!chan || !grobj || *grobj) return -EINVAL; - gr = calloc(1, sizeof(*gr)); - if (!gr) + nvgrobj = calloc(1, sizeof(struct nouveau_grobj_priv)); + if (!nvgrobj) return -ENOMEM; - gr->base.channel = userchan; - gr->base.handle = handle; - gr->base.grclass = 0; + nvgrobj->base.channel = chan; + nvgrobj->base.handle = handle; + nvgrobj->base.grclass = 0; - *usergr = &gr->base; + *grobj = &nvgrobj->base; return 0; } void -nouveau_grobj_free(struct nouveau_grobj **usergrobj) +nouveau_grobj_free(struct nouveau_grobj **grobj) { - struct nouveau_grobj_priv *gr; + struct nouveau_device_priv *nvdev; + struct nouveau_channel_priv *chan; + struct nouveau_grobj_priv *nvgrobj; - if (!usergrobj) + if (!grobj || !*grobj) return; - gr = nouveau_grobj(*usergrobj); - *usergrobj = NULL; + nvgrobj = nouveau_grobj(*grobj); + *grobj = NULL; - if (gr) { - struct nouveau_channel_priv *chan; - struct nouveau_device_priv *nv; - struct drm_nouveau_gpuobj_free f; - chan = nouveau_channel(gr->base.channel); - nv = nouveau_device(chan->base.device); + chan = nouveau_channel(nvgrobj->base.channel); + nvdev = nouveau_device(chan->base.device); + + if (nvgrobj->base.grclass) { + struct drm_nouveau_gpuobj_free f; - if (gr->base.grclass) { - f.channel = chan->drm.channel; - f.handle = gr->base.handle; - drmCommandWrite(nv->fd, DRM_NOUVEAU_GPUOBJ_FREE, - &f, sizeof(f)); - } - free(gr); + f.channel = chan->drm.channel; + f.handle = nvgrobj->base.handle; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, + &f, sizeof(f)); } + free(nvgrobj); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c index e423d59b1df..01e8f38440e 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c @@ -27,70 +27,67 @@ #include "nouveau_local.h" #define NOTIFIER(__v) \ - struct nouveau_notifier_priv *notifier = nouveau_notifier(user); \ - volatile uint32_t *n = (void*)notifier->map + (id * 32) + struct nouveau_notifier_priv *nvnotify = nouveau_notifier(notifier); \ + volatile uint32_t *__v = (void*)nvnotify->map + (id * 32) int -nouveau_notifier_alloc(struct nouveau_channel *userchan, uint32_t handle, - int count, struct nouveau_notifier **usernotifier) +nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, + int count, struct nouveau_notifier **notifier) { - struct nouveau_notifier_priv *notifier; + struct nouveau_notifier_priv *nvnotify; int ret; - if (!userchan || !usernotifier || *usernotifier) + if (!chan || !notifier || *notifier) return -EINVAL; - notifier = calloc(1, sizeof(*notifier)); - if (!notifier) + nvnotify = calloc(1, sizeof(struct nouveau_notifier_priv)); + if (!nvnotify) return -ENOMEM; - notifier->base.channel = userchan; - notifier->base.handle = handle; + nvnotify->base.channel = chan; + nvnotify->base.handle = handle; - notifier->drm.channel = userchan->id; - notifier->drm.handle = handle; - notifier->drm.count = count; - if ((ret = drmCommandWriteRead(nouveau_device(userchan->device)->fd, + nvnotify->drm.channel = chan->id; + nvnotify->drm.handle = handle; + nvnotify->drm.count = count; + if ((ret = drmCommandWriteRead(nouveau_device(chan->device)->fd, DRM_NOUVEAU_NOTIFIEROBJ_ALLOC, - ¬ifier->drm, - sizeof(notifier->drm)))) { - nouveau_notifier_free((void *)¬ifier); + &nvnotify->drm, + sizeof(nvnotify->drm)))) { + nouveau_notifier_free((void *)&nvnotify); return ret; } - notifier->map = (void *)nouveau_channel(userchan)->notifier_block + - notifier->drm.offset; - *usernotifier = ¬ifier->base; + nvnotify->map = (void *)nouveau_channel(chan)->notifier_block + + nvnotify->drm.offset; + *notifier = &nvnotify->base; return 0; } void -nouveau_notifier_free(struct nouveau_notifier **usernotifier) +nouveau_notifier_free(struct nouveau_notifier **notifier) { - struct nouveau_notifier_priv *notifier; + struct nouveau_notifier_priv *nvnotify; + struct nouveau_channel_priv *nvchan; + struct nouveau_device_priv *nvdev; + struct drm_nouveau_gpuobj_free f; - if (!usernotifier) + if (!notifier || !*notifier) return; - notifier = nouveau_notifier(*usernotifier); - *usernotifier = NULL; + nvnotify = nouveau_notifier(*notifier); + *notifier = NULL; - if (notifier) { - struct nouveau_channel_priv *chan; - struct nouveau_device_priv *nv; - struct drm_nouveau_gpuobj_free f; + nvchan = nouveau_channel(nvnotify->base.channel); + nvdev = nouveau_device(nvchan->base.device); - chan = nouveau_channel(notifier->base.channel); - nv = nouveau_device(chan->base.device); - - f.channel = chan->drm.channel; - f.handle = notifier->base.handle; - drmCommandWrite(nv->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); - free(notifier); - } + f.channel = nvchan->drm.channel; + f.handle = nvnotify->base.handle; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); + free(nvnotify); } void -nouveau_notifier_reset(struct nouveau_notifier *user, int id) +nouveau_notifier_reset(struct nouveau_notifier *notifier, int id) { NOTIFIER(n); @@ -102,7 +99,7 @@ nouveau_notifier_reset(struct nouveau_notifier *user, int id) } uint32_t -nouveau_notifier_status(struct nouveau_notifier *user, int id) +nouveau_notifier_status(struct nouveau_notifier *notifier, int id) { NOTIFIER(n); @@ -110,7 +107,7 @@ nouveau_notifier_status(struct nouveau_notifier *user, int id) } uint32_t -nouveau_notifier_return_val(struct nouveau_notifier *user, int id) +nouveau_notifier_return_val(struct nouveau_notifier *notifier, int id) { NOTIFIER(n); @@ -118,7 +115,7 @@ nouveau_notifier_return_val(struct nouveau_notifier *user, int id) } int -nouveau_notifier_wait_status(struct nouveau_notifier *user, int id, +nouveau_notifier_wait_status(struct nouveau_notifier *notifier, int id, int status, int timeout) { NOTIFIER(n); -- cgit v1.2.3 From 868048fcc77ec954e2823959285bfa7b8f82b13c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 14 Dec 2007 01:51:43 +1100 Subject: nv40: Do all fp setup in nv40_fragprog.c --- src/mesa/pipe/nv40/nv40_fragprog.c | 29 ++++++++++++++++++--- src/mesa/pipe/nv40/nv40_state_emit.c | 49 ++++-------------------------------- 2 files changed, 31 insertions(+), 47 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index bad5f0020eb..a0d98ae4dcc 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -701,10 +701,34 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) { struct pipe_winsys *ws = nv40->pipe.winsys; uint32_t fp_control; + int i; if (!fp->translated) { - NOUVEAU_ERR("fragprog invalid, using passthrough shader\n"); - fp = &passthrough_fp; + nv40_fragprog_translate(nv40, fp); + if (!fp->translated) { + NOUVEAU_ERR("invalid, using passthrough shader\n"); + fp = &passthrough_fp; + } + } + + if (fp->num_consts) { + float *map = ws->buffer_map(ws, nv40->fragprog.constant_buf, + PIPE_BUFFER_FLAG_READ); + for (i = 0; i < fp->num_consts; i++) { + uint pid = fp->consts[i].pipe_id; + + if (pid == -1) + continue; + + if (!memcmp(&fp->insn[fp->consts[i].hw_id], &map[pid*4], + 4 * sizeof(float))) + continue; + + memcpy(&fp->insn[fp->consts[i].hw_id], &map[pid*4], + 4 * sizeof(float)); + fp->on_hw = 0; + } + ws->buffer_unmap(ws, nv40->fragprog.constant_buf); } if (!fp->on_hw) { @@ -712,7 +736,6 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->buffer = ws->buffer_create(ws, 0x100, 0, 0); #if 0 - int i; for (i = 0; i < fp->insn_len; i++) NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); #endif diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 55c7d36b129..fb2a3cb98e9 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -2,53 +2,12 @@ #include "nv40_dma.h" #include "nv40_state.h" -static INLINE void -nv40_state_update_fragprog(struct nv40_context *nv40) -{ - struct pipe_context *pipe = (struct pipe_context *)nv40; - struct nv40_fragment_program *fp = nv40->fragprog.current; - float *map; - int i; - - if (!fp->translated) - nv40_fragprog_translate(nv40, fp); - - if (fp->num_consts) { - map = pipe->winsys->buffer_map(pipe->winsys, - nv40->fragprog.constant_buf, - PIPE_BUFFER_FLAG_READ); - for (i = 0; i < fp->num_consts; i++) { - uint pid = fp->consts[i].pipe_id; - - if (pid == -1) - continue; - - if (!memcmp(&fp->insn[fp->consts[i].hw_id], &map[pid*4], - 4 * sizeof(float))) - continue; - - memcpy(&fp->insn[fp->consts[i].hw_id], &map[pid*4], - 4 * sizeof(float)); - fp->on_hw = 0; - } - pipe->winsys->buffer_unmap(pipe->winsys, - nv40->fragprog.constant_buf); - } -} - void nv40_emit_hw_state(struct nv40_context *nv40) { if (nv40->dirty & NV40_NEW_FRAGPROG) { - struct nv40_fragment_program *cur = nv40->fragprog.current; - - nv40_state_update_fragprog(nv40); - - if (cur->on_hw) - nv40->dirty &= ~NV40_NEW_FRAGPROG; - - if (!cur->on_hw || cur != nv40->fragprog.active) - nv40_fragprog_bind(nv40, cur); + nv40_fragprog_bind(nv40, nv40->fragprog.current); + /*XXX: clear NV40_NEW_FRAGPROG if no now program uploaded */ } if (nv40->dirty & NV40_NEW_TEXTURE) @@ -62,8 +21,10 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->dirty &= ~(NV40_NEW_TEXTURE | NV40_NEW_FRAGPROG); } - if (nv40->dirty & NV40_NEW_VERTPROG) + if (nv40->dirty & NV40_NEW_VERTPROG) { nv40_vertprog_bind(nv40, nv40->vertprog.current); + nv40->dirty &= ~NV40_NEW_VERTPROG; + } if (nv40->dirty & NV40_NEW_ARRAYS) { nv40_vbo_arrays_update(nv40); -- cgit v1.2.3 From 7f89c776e19b400c0adf647fc9dfb392efe88dbd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 15 Dec 2007 01:50:15 +1100 Subject: nv40: less dodgy vp const/insn handling --- src/mesa/pipe/nv40/nv40_shader.h | 6 +- src/mesa/pipe/nv40/nv40_state.h | 29 +++--- src/mesa/pipe/nv40/nv40_vertprog.c | 191 +++++++++++++++++++++---------------- 3 files changed, 132 insertions(+), 94 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_shader.h b/src/mesa/pipe/nv40/nv40_shader.h index 01c0652b4d8..5909c70713c 100644 --- a/src/mesa/pipe/nv40/nv40_shader.h +++ b/src/mesa/pipe/nv40/nv40_shader.h @@ -90,8 +90,8 @@ # define NV40_VP_INST_OP_ADD 0x03 # define NV40_VP_INST_OP_MAD 0x04 # define NV40_VP_INST_OP_DP3 0x05 -# define NV40_VP_INST_OP_DP4 0x07 # define NV40_VP_INST_OP_DPH 0x06 +# define NV40_VP_INST_OP_DP4 0x07 # define NV40_VP_INST_OP_DST 0x08 # define NV40_VP_INST_OP_MIN 0x09 # define NV40_VP_INST_OP_MAX 0x0A @@ -109,9 +109,11 @@ # define NV40_VP_INST_OP_SSG 0x16 # define NV40_VP_INST_OP_ARR 0x17 # define NV40_VP_INST_OP_ARA 0x18 -# define NV40_VP_INST_OP_TXWHAT 0x19 +# define NV40_VP_INST_OP_TXL 0x19 #define NV40_VP_INST_SCA_OPCODE_SHIFT 27 #define NV40_VP_INST_SCA_OPCODE_MASK (0x1F << 27) +# define NV40_VP_INST_OP_NOP 0x00 +# define NV40_VP_INST_OP_MOV 0x01 # define NV40_VP_INST_OP_RCP 0x02 # define NV40_VP_INST_OP_RCC 0x03 # define NV40_VP_INST_OP_RSQ 0x04 diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 80c76cd25b6..8ab334d2675 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -54,24 +54,31 @@ struct nv40_rasterizer_state { uint32_t point_sprite; }; +struct nv40_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv40_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + struct nv40_vertex_program { const struct pipe_shader_state *pipe; boolean translated; + struct nv40_vertex_program_exec *insns; + unsigned nr_insns; + struct nv40_vertex_program_data *consts; + unsigned nr_consts; struct nouveau_resource *exec; - uint32_t *insn; - uint insn_len; - + unsigned exec_start; struct nouveau_resource *data; - uint data_start; - - struct { - int pipe_id; - int hw_id; - float value[4]; - } consts[256]; - int num_consts; + unsigned data_start; + unsigned data_start_min; uint32_t ir; uint32_t or; diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index b6ebaee2f2a..c9e1f251e88 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -9,6 +9,18 @@ #include "nv40_dma.h" #include "nv40_state.h" +/* TODO (at least...): + * 1. Indexed consts + ARL + * 2. Arb. swz/negation + * 3. NV_vp11, NV_vp2, NV_vp3 features + * - extra arith opcodes + * - branching + * - texture sampling + * - indexed attribs + * - indexed results + * 4. bugs + */ + #define SWZ_X 0 #define SWZ_Y 1 #define SWZ_Z 2 @@ -26,28 +38,12 @@ #define neg(s) nv40_sr_neg((s)) #define abs(s) nv40_sr_abs((s)) -static uint32_t -passthrough_vp_data[] = { - 0x40041c6c, 0x0040010d, 0x8106c083, 0x6041ff84, - 0x40041c6c, 0x0040000d, 0x8106c083, 0x6041ff81, -}; - -static struct nv40_vertex_program -passthrough_vp = { - .pipe = NULL, - .translated = TRUE, - - .insn = passthrough_vp_data, - .insn_len = sizeof(passthrough_vp_data) / sizeof(uint32_t), - - .ir = 0x00000003, - .or = 0x00000001, -}; - struct nv40_vpc { struct nv40_vertex_program *vp; - uint output_map[PIPE_MAX_SHADER_OUTPUTS]; + struct nv40_vertex_program_exec *vpi; + + unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; int high_temp; int temp_temp_count; @@ -59,7 +55,7 @@ temp(struct nv40_vpc *vpc) int idx; idx = vpc->temp_temp_count++; - idx += vpc->high_temp; + idx += vpc->high_temp + 1; return nv40_sr(NV40SR_TEMP, idx); } @@ -67,16 +63,25 @@ static INLINE struct nv40_sreg constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w) { struct nv40_vertex_program *vp = vpc->vp; - int idx = vp->num_consts; + struct nv40_vertex_program_data *vpd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < vp->nr_consts; idx++) { + if (vp->consts[idx].index == pipe) + return nv40_sr(NV40SR_CONST, idx); + } + } - vp->consts[idx].pipe_id = pipe; - vp->consts[idx].hw_id = idx; - vp->consts[idx].value[0] = x; - vp->consts[idx].value[1] = y; - vp->consts[idx].value[2] = z; - vp->consts[idx].value[3] = w; - vp->num_consts++; + idx = vp->nr_consts++; + vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); + vpd = &vp->consts[idx]; + vpd->index = pipe; + vpd->value[0] = x; + vpd->value[1] = y; + vpd->value[2] = z; + vpd->value[3] = w; return nv40_sr(NV40SR_CONST, idx); } @@ -103,7 +108,9 @@ emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src) case NV40SR_CONST: sr |= (NV40_VP_SRC_REG_TYPE_CONST << NV40_VP_SRC_REG_TYPE_SHIFT); - hw[1] |= (src.index << NV40_VP_INST_CONST_SRC_SHIFT); + assert(vpc->vpi->const_index == -1 || + vpc->vpi->const_index == src.index); + vpc->vpi->const_index = src.index; break; case NV40SR_NONE: sr |= (NV40_VP_SRC_REG_TYPE_INPUT << @@ -202,7 +209,14 @@ nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op, struct nv40_sreg s2) { struct nv40_vertex_program *vp = vpc->vp; - uint32_t *hw = &vp->insn[vp->insn_len]; + uint32_t *hw; + + vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); + vpc->vpi = &vp->insns[vp->nr_insns - 1]; + memset(vpc->vpi, 0, sizeof(*vpc->vpi)); + vpc->vpi->const_index = -1; + + hw = vpc->vpi->data; hw[0] |= (NV40_VP_INST_COND_TR << NV40_VP_INST_COND_SHIFT); hw[0] |= ((0 << NV40_VP_INST_COND_SWZ_X_SHIFT) | @@ -224,8 +238,6 @@ nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op, emit_src(vpc, hw, 0, s0); emit_src(vpc, hw, 1, s1); emit_src(vpc, hw, 2, s2); - - vp->insn_len += 4; } static INLINE struct nv40_sreg @@ -326,8 +338,6 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, ai = fsrc->SrcRegister.Index; src[i] = tgsi_src(vpc, fsrc); } else { - NOUVEAU_MSG("extra src attr %d\n", - fsrc->SrcRegister.Index); src[i] = temp(vpc); arith(vpc, 0, OP_MOV, src[i], MASK_ALL, tgsi_src(vpc, fsrc), none, none); @@ -518,7 +528,6 @@ nv40_vertprog_translate(struct nv40_context *nv40, vpc = calloc(1, sizeof(struct nv40_vpc)); if (!vpc) return; - vp->insn = calloc(1, 128*4*sizeof(uint32_t)); vpc->vp = vp; vpc->high_temp = -1; @@ -547,7 +556,6 @@ nv40_vertprog_translate(struct nv40_context *nv40, case TGSI_TOKEN_TYPE_INSTRUCTION: { const struct tgsi_full_instruction *finst; - finst = &parse.FullToken.FullInstruction; if (!nv40_vertprog_parse_instruction(vpc, finst)) goto out_err; @@ -558,14 +566,7 @@ nv40_vertprog_translate(struct nv40_context *nv40, } } - vp->insn[vp->insn_len - 1] |= NV40_VP_INST_LAST; -#if 0 - { - int i; - for (i = 0; i < vp->insn_len; i++) - NOUVEAU_ERR("inst[%d] = 0x%08x\n", i, vp->insn[i]); - } -#endif + vp->insns[vp->nr_insns - 1].data[3] |= NV40_VP_INST_LAST; vp->translated = TRUE; out_err: tgsi_parse_free(&parse); @@ -576,9 +577,8 @@ void nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) { struct nouveau_winsys *nvws = nv40->nvws; - struct pipe_context *pipe = &nv40->pipe; + struct pipe_winsys *ws = nv40->pipe.winsys; boolean upload_code = FALSE, upload_data = FALSE; - float *map; int i; /* Translate TGSI shader into hw bytecode */ @@ -589,11 +589,9 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) } /* Allocate hw vtxprog exec slots */ - /*XXX: when we do branching, need to patch targets if program moves. - */ if (!vp->exec) { struct nouveau_resource *heap = nv40->vertprog.exec_heap; - uint vplen = vp->insn_len / 4; + uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { while (heap->next && heap->size < vplen) { @@ -611,75 +609,106 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) } /* Allocate hw vtxprog const slots */ - if (vp->num_consts && !vp->data) { + if (vp->nr_consts && !vp->data) { struct nouveau_resource *heap = nv40->vertprog.data_heap; - int count = vp->num_consts; - if (nvws->res_alloc(heap, count, vp, &vp->data)) { - while (heap->next && heap->size < count) { + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { + while (heap->next && heap->size < vp->nr_consts) { struct nv40_vertex_program *evict; evict = heap->next->priv; nvws->res_free(&evict->data); } - if (nvws->res_alloc(heap, count, vp, &vp->data)) + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) assert(0); } + /*XXX: handle this some day */ + assert(vp->data->start >= vp->data_start_min); + upload_data = TRUE; + if (vp->data_start != vp->data->start) + upload_code = TRUE; } - /* If constants moved, patch the vtxprog to fix the offsets */ - if (vp->num_consts && vp->data_start != vp->data->start) { - for (i = 0; i < vp->insn_len; i += 4) { - int id; + /* If exec or data segments moved we need to patch the program to + * fixup offsets and register IDs. + */ + if (vp->exec_start != vp->exec->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv40_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->has_branch_offset) { + assert(0); + } + } - id = (vp->insn[i + 1] & NV40_VP_INST_CONST_SRC_MASK) >> - NV40_VP_INST_CONST_SRC_SHIFT; - id -= vp->data_start; - id += vp->data->start; + vp->exec_start = vp->exec->start; + } + + if (vp->nr_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv40_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->const_index >= 0) { + vpi->data[1] &= ~NV40_VP_INST_CONST_SRC_MASK; + vpi->data[1] |= + (vpi->const_index + vp->data->start) << + NV40_VP_INST_CONST_SRC_SHIFT; - vp->insn[i + 1] &= ~NV40_VP_INST_CONST_SRC_MASK; - vp->insn[i + 1] |= (id << NV40_VP_INST_CONST_SRC_SHIFT); + } } vp->data_start = vp->data->start; - upload_code = TRUE; } /* Update + Upload constant values */ - if (vp->num_consts) { - map = pipe->winsys->buffer_map(pipe->winsys, - nv40->vertprog.constant_buf, - PIPE_BUFFER_FLAG_READ); - for (i = 0; i < vp->num_consts; i++) { - uint pid = vp->consts[i].pipe_id; - - if (pid >= 0) { + if (vp->nr_consts) { + float *map = NULL; + + if (nv40->vertprog.constant_buf) { + map = ws->buffer_map(ws, nv40->vertprog.constant_buf, + PIPE_BUFFER_FLAG_READ); + } + + for (i = 0; i < vp->nr_consts; i++) { + struct nv40_vertex_program_data *vpd = &vp->consts[i]; + + if (vpd->index >= 0) { if (!upload_data && - !memcmp(vp->consts[i].value, &map[pid*4], + !memcmp(vpd->value, &map[vpd->index * 4], 4 * sizeof(float))) continue; - memcpy(vp->consts[i].value, &map[pid*4], + memcpy(vpd->value, &map[vpd->index * 4], 4 * sizeof(float)); } BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5); - OUT_RING (vp->consts[i].hw_id + vp->data->start); - OUT_RINGp ((uint32_t *)vp->consts[i].value, 4); + OUT_RING (i + vp->data->start); + OUT_RINGp ((uint32_t *)vpd->value, 4); + } + + if (map) { + ws->buffer_unmap(ws, nv40->vertprog.constant_buf); } - pipe->winsys->buffer_unmap(pipe->winsys, - nv40->vertprog.constant_buf); } /* Upload vtxprog */ if (upload_code) { +#if 0 + for (i = 0; i < vp->nr_insns; i++) { + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); + } +#endif BEGIN_RING(curie, NV40TCL_VP_UPLOAD_FROM_ID, 1); OUT_RING (vp->exec->start); - for (i = 0; i < vp->insn_len; i += 4) { + for (i = 0; i < vp->nr_insns; i++) { BEGIN_RING(curie, NV40TCL_VP_UPLOAD_INST(0), 4); - OUT_RINGp (&vp->insn[i], 4); + OUT_RINGp (vp->insns[i].data, 4); } } -- cgit v1.2.3 From 210bf673c2fee08fa1fabb84439a2e09a70036bf Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 00:31:26 +1100 Subject: nouveau: move extension stuff into nouveau_screen.c --- .../drivers/dri/nouveau_winsys/nouveau_context.c | 50 +--------------------- .../drivers/dri/nouveau_winsys/nouveau_screen.c | 42 +++++++++++++++++- 2 files changed, 41 insertions(+), 51 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c index 03f1816c323..9c524fa0017 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c @@ -1,6 +1,7 @@ #include "main/glheader.h" #include "glapi/glthread.h" #include +#include "utils.h" #include "state_tracker/st_public.h" #include "pipe/p_defines.h" @@ -12,55 +13,6 @@ #include "nouveau_screen.h" #include "nouveau_winsys_pipe.h" -#define need_GL_ARB_fragment_program -#define need_GL_ARB_multisample -#define need_GL_ARB_occlusion_query -#define need_GL_ARB_point_parameters -#define need_GL_ARB_texture_compression -#define need_GL_ARB_vertex_program -#define need_GL_ARB_vertex_buffer_object -#define need_GL_EXT_compiled_vertex_array -#define need_GL_EXT_fog_coord -#define need_GL_EXT_secondary_color -#define need_GL_EXT_framebuffer_object -#include "extension_helper.h" - -const struct dri_extension common_extensions[] = -{ - { NULL, 0 } -}; - -const struct dri_extension nv40_extensions[] = -{ - { "GL_ARB_fragment_program", NULL }, - { "GL_ARB_multisample", GL_ARB_multisample_functions }, - { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, - { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, - { "GL_ARB_texture_border_clamp", NULL }, - { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, - { "GL_ARB_texture_cube_map", NULL }, - { "GL_ARB_texture_env_add", NULL }, - { "GL_ARB_texture_env_combine", NULL }, - { "GL_ARB_texture_env_crossbar", NULL }, - { "GL_ARB_texture_env_dot3", NULL }, - { "GL_ARB_texture_mirrored_repeat", NULL }, - { "GL_ARB_texture_non_power_of_two", NULL }, - { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, - { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, - { "GL_ATI_texture_env_combine3", NULL }, - { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, - { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, - { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, - { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, - { "GL_EXT_texture_edge_clamp", NULL }, - { "GL_EXT_texture_env_add", NULL }, - { "GL_EXT_texture_env_combine", NULL }, - { "GL_EXT_texture_env_dot3", NULL }, - { "GL_EXT_texture_mirror_clamp", NULL }, - { "GL_NV_texture_rectangle", NULL }, - { NULL, 0 } -}; - #ifdef DEBUG static const struct dri_debug_control debug_control[] = { { "bo", DEBUG_BO }, diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index 5a957f48eb8..6324a591311 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -18,6 +18,45 @@ #error nouveau_drm.h version does not match expected version #endif +/* Extension stuff, enabling of extensions handled by Gallium's GL state + * tracker. But, we still need to define the entry points we want. + */ +#define need_GL_ARB_fragment_program +#define need_GL_ARB_multisample +#define need_GL_ARB_occlusion_query +#define need_GL_ARB_point_parameters +#define need_GL_ARB_shader_objects +#define need_GL_ARB_texture_compression +#define need_GL_ARB_vertex_program +#define need_GL_ARB_vertex_shader +#define need_GL_ARB_vertex_buffer_object +#define need_GL_EXT_compiled_vertex_array +#define need_GL_EXT_fog_coord +#define need_GL_EXT_secondary_color +#define need_GL_EXT_framebuffer_object +#define need_GL_VERSION_2_0 +#define need_GL_VERSION_2_1 +#include "extension_helper.h" + +const struct dri_extension card_extensions[] = +{ + { "GL_ARB_multisample", GL_ARB_multisample_functions }, + { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, + { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, + { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, + { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, + { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, + { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, + { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, + { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, + { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, + { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, + { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, + { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, + { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, + { NULL, 0 } +}; + PUBLIC const char __driConfigOptions[] = DRI_CONF_BEGIN DRI_CONF_END; @@ -260,8 +299,7 @@ __driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, (nv_dri->bpp == 16) ? 0 : 8, 1); - driInitExtensions(NULL, common_extensions, GL_FALSE); - driInitExtensions(NULL, nv40_extensions, GL_FALSE); + driInitExtensions(NULL, card_extensions, GL_FALSE); return (void *)psp; } -- cgit v1.2.3 From af0b4a50e59435a782f59ccec7ad0552c0304016 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 15:49:49 +1100 Subject: nv40: make vbo state update static --- src/mesa/pipe/nv40/nv40_context.h | 1 - src/mesa/pipe/nv40/nv40_state_emit.c | 5 - src/mesa/pipe/nv40/nv40_vbo.c | 176 +++++++++++++++++++---------------- 3 files changed, 95 insertions(+), 87 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 83d3f7b676e..bf1534b69ba 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -98,7 +98,6 @@ extern boolean nv40_draw_elements(struct pipe_context *pipe, unsigned indexSize, unsigned mode, unsigned start, unsigned count); -extern void nv40_vbo_arrays_update(struct nv40_context *nv40); /* nv40_clear.c */ extern void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index fb2a3cb98e9..c53b5d852ae 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -25,10 +25,5 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40_vertprog_bind(nv40, nv40->vertprog.current); nv40->dirty &= ~NV40_NEW_VERTPROG; } - - if (nv40->dirty & NV40_NEW_ARRAYS) { - nv40_vbo_arrays_update(nv40); - nv40->dirty &= ~NV40_NEW_ARRAYS; - } } diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 19ae6ca3238..62c877b2dc6 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -6,6 +6,93 @@ #include "nv40_dma.h" #include "nv40_state.h" +static INLINE int +nv40_vbo_ncomp(uint format) +{ + int ncomp = 0; + + if (pf_size_x(format)) ncomp++; + if (pf_size_y(format)) ncomp++; + if (pf_size_z(format)) ncomp++; + if (pf_size_w(format)) ncomp++; + + return ncomp; +} + +static INLINE int +nv40_vbo_type(uint format) +{ + switch (pf_type(format)) { + case PIPE_FORMAT_TYPE_FLOAT: + return NV40TCL_VTXFMT_TYPE_FLOAT; + case PIPE_FORMAT_TYPE_UNORM: + return NV40TCL_VTXFMT_TYPE_UBYTE; + default: + assert(0); + } +} + +static void +nv40_vbo_arrays_update(struct nv40_context *nv40) +{ + struct nv40_vertex_program *vp = nv40->vertprog.active; + uint32_t inputs, vtxfmt[16]; + int hw, num_hw; + + inputs = vp->ir; + for (hw = 0; hw < 16 && inputs; hw++) { + if (inputs & (1 << hw)) { + num_hw = hw; + inputs &= ~(1 << hw); + } + } + num_hw++; + + inputs = vp->ir; + BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + for (hw = 0; hw < num_hw; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + + if (!(inputs & (1 << hw))) { + OUT_RING(0); + vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; + continue; + } + + ve = &nv40->vtxelt[hw]; + vb = &nv40->vtxbuf[ve->vertex_buffer_index]; + + OUT_RELOC(vb->buffer, vb->buffer_offset + ve->src_offset, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, + NV40TCL_VTXBUF_ADDRESS_DMA1); + vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | + (nv40_vbo_ncomp(ve->src_format) << + NV40TCL_VTXFMT_SIZE_SHIFT) | + nv40_vbo_type(ve->src_format)); + } + + BEGIN_RING(curie, 0x1710, 1); + OUT_RING (0); /* vtx cache flush */ + BEGIN_RING(curie, NV40TCL_VTXFMT(0), num_hw); + OUT_RINGp (vtxfmt, num_hw); +} + +static boolean +nv40_vbo_validate_state(struct nv40_context *nv40) +{ + if (nv40->dirty & ~NV40_NEW_ARRAYS) + nv40_emit_hw_state(nv40); + + if (nv40->dirty & NV40_NEW_ARRAYS) { + nv40_vbo_arrays_update(nv40); + nv40->dirty &= ~NV40_NEW_ARRAYS; + } + + return TRUE; +} + boolean nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) @@ -13,8 +100,7 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, struct nv40_context *nv40 = (struct nv40_context *)pipe; unsigned nr; - if (nv40->dirty) - nv40_emit_hw_state(nv40); + assert(nv40_vbo_validate_state(nv40)); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -26,13 +112,14 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, start += nr; } - /*XXX: large arrays (nr>2047) will blow up */ nr = count >> 8; - if (nr) { - assert (nr <= 2047); + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; - BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, nr); - while (nr--) { + BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, push); + while (push--) { OUT_RING(((0x100 - 1) << 24) | start); start += 0x100; } @@ -121,8 +208,7 @@ nv40_draw_elements(struct pipe_context *pipe, struct nv40_context *nv40 = (struct nv40_context *)pipe; void *ib; - if (nv40->dirty) - nv40_emit_hw_state(nv40); + assert(nv40_vbo_validate_state(nv40)); ib = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, PIPE_BUFFER_FLAG_READ); @@ -157,76 +243,4 @@ nv40_draw_elements(struct pipe_context *pipe, return TRUE; } -static INLINE int -nv40_vbo_ncomp(uint format) -{ - int ncomp = 0; - - if (pf_size_x(format)) ncomp++; - if (pf_size_y(format)) ncomp++; - if (pf_size_z(format)) ncomp++; - if (pf_size_w(format)) ncomp++; - - return ncomp; -} - -static INLINE int -nv40_vbo_type(uint format) -{ - switch (pf_type(format)) { - case PIPE_FORMAT_TYPE_FLOAT: - return NV40TCL_VTXFMT_TYPE_FLOAT; - case PIPE_FORMAT_TYPE_UNORM: - return NV40TCL_VTXFMT_TYPE_UBYTE; - default: - assert(0); - } -} - -void -nv40_vbo_arrays_update(struct nv40_context *nv40) -{ - struct nv40_vertex_program *vp = nv40->vertprog.active; - uint32_t inputs, vtxfmt[16]; - int hw, num_hw; - - inputs = vp->ir; - for (hw = 0; hw < 16 && inputs; hw++) { - if (inputs & (1 << hw)) { - num_hw = hw; - inputs &= ~(1 << hw); - } - } - num_hw++; - - inputs = vp->ir; - BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); - for (hw = 0; hw < num_hw; hw++) { - struct pipe_vertex_element *ve; - struct pipe_vertex_buffer *vb; - - if (!(inputs & (1 << hw))) { - OUT_RING(0); - vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; - continue; - } - - ve = &nv40->vtxelt[hw]; - vb = &nv40->vtxbuf[ve->vertex_buffer_index]; - - OUT_RELOC(vb->buffer, vb->buffer_offset + ve->src_offset, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV40TCL_VTXBUF_ADDRESS_DMA1); - vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_ncomp(ve->src_format) << - NV40TCL_VTXFMT_SIZE_SHIFT) | - nv40_vbo_type(ve->src_format)); - } - - BEGIN_RING(curie, 0x1710, 1); - OUT_RING (0); /* vtx cache flush */ - BEGIN_RING(curie, NV40TCL_VTXFMT(0), num_hw); - OUT_RINGp (vtxfmt, num_hw); -} -- cgit v1.2.3 From ab4c2e014d4117d6ef43685a57c0ea1b93ba5562 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 16:01:25 +1100 Subject: nv40: destroy programs --- src/mesa/pipe/nv40/nv40_context.h | 4 ++++ src/mesa/pipe/nv40/nv40_fragprog.c | 30 ++++++++++-------------------- src/mesa/pipe/nv40/nv40_state.c | 12 ++++++++++-- src/mesa/pipe/nv40/nv40_vertprog.c | 9 +++++++++ 4 files changed, 33 insertions(+), 22 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index bf1534b69ba..9a93ed35ace 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -79,12 +79,16 @@ extern void nv40_vertprog_translate(struct nv40_context *, struct nv40_vertex_program *); extern void nv40_vertprog_bind(struct nv40_context *, struct nv40_vertex_program *); +extern void nv40_vertprog_destroy(struct nv40_context *, + struct nv40_vertex_program *); /* nv40_fragprog.c */ extern void nv40_fragprog_translate(struct nv40_context *, struct nv40_fragment_program *); extern void nv40_fragprog_bind(struct nv40_context *, struct nv40_fragment_program *); +extern void nv40_fragprog_destroy(struct nv40_context *, + struct nv40_fragment_program *); /* nv40_state.c and friends */ extern void nv40_emit_hw_state(struct nv40_context *nv40); diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index a0d98ae4dcc..e801dae8e11 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -27,22 +27,6 @@ #define abs(s) nv40_sr_abs((s)) #define scale(s,v) nv40_sr_scale((s), NV40_FP_OP_DST_SCALE_##v) -static uint32_t -passthrough_fp_data[] = { - 0x01403e81, 0x1c9dc801, 0x0001c800, 0x3fe1c800 -}; - -static struct nv40_fragment_program -passthrough_fp = { - .pipe = NULL, - .translated = TRUE, - .insn = passthrough_fp_data, - .insn_len = sizeof(passthrough_fp_data) / sizeof(uint32_t), - .buffer = NULL, - .uses_kil = 0, - .num_regs = 2, -}; - struct nv40_fpc { struct nv40_fragment_program *fp; @@ -705,10 +689,8 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) if (!fp->translated) { nv40_fragprog_translate(nv40, fp); - if (!fp->translated) { - NOUVEAU_ERR("invalid, using passthrough shader\n"); - fp = &passthrough_fp; - } + if (!fp->translated) + assert(0); } if (fp->num_consts) { @@ -763,3 +745,11 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) nv40->fragprog.active = fp; } +void +nv40_fragprog_destroy(struct nv40_context *nv40, + struct nv40_fragment_program *fp) +{ + if (fp->insn_len) + free(fp->insn); +} + diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 5aeba684ee0..5b070a280d0 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -505,7 +505,11 @@ nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) static void nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_vertex_program *vp = hwcso; + + nv40_vertprog_destroy(nv40, vp); + free(vp); } static void * @@ -533,7 +537,11 @@ nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) static void nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_fragment_program *fp = hwcso; + + nv40_fragprog_destroy(nv40, fp); + free(fp); } static void diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index c9e1f251e88..981d70ec92b 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -721,3 +721,12 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) nv40->vertprog.active = vp; } +void +nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp) +{ + if (vp->nr_consts) + free(vp->consts); + if (vp->nr_insns) + free(vp->insns); +} + -- cgit v1.2.3 From 505e50de1019ca002408f0be6ded89ba714acfa7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 18:36:18 +1100 Subject: nv40: grow fragprog as needed --- src/mesa/pipe/nv40/nv40_fragprog.c | 62 ++++++++++++++++++++++++-------------- 1 file changed, 40 insertions(+), 22 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index e801dae8e11..f532eccd7b1 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -39,7 +39,9 @@ struct nv40_fpc { uint colour_id; boolean inst_has_const; - int inst_const_id; + int inst_const_id; + + unsigned inst_offset; }; static INLINE struct nv40_sreg @@ -60,8 +62,18 @@ temp(struct nv40_fpc *fpc) (d), (m), (s0), none, none) static void -emit_src(struct nv40_fpc *fpc, uint32_t *hw, int pos, struct nv40_sreg src) +grow_insns(struct nv40_fpc *fpc, int size) { + struct nv40_fragment_program *fp = fpc->fp; + + fp->insn_len += size; + fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len); +} + +static void +emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) +{ + uint32_t *hw = &fpc->fp->insn[fpc->inst_offset]; uint32_t sr = 0; switch (src.type) { @@ -103,9 +115,10 @@ emit_src(struct nv40_fpc *fpc, uint32_t *hw, int pos, struct nv40_sreg src) } static void -emit_dst(struct nv40_fpc *fpc, uint32_t *hw, struct nv40_sreg dst) +emit_dst(struct nv40_fpc *fpc, struct nv40_sreg dst) { struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; switch (dst.type) { case NV40SR_TEMP: @@ -135,7 +148,12 @@ nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) { struct nv40_fragment_program *fp = fpc->fp; - uint32_t *hw = &fp->insn[fp->insn_len]; + uint32_t *hw; + + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + memset(hw, 0, sizeof(uint32_t) * 4); fpc->inst_has_const = FALSE; @@ -156,17 +174,16 @@ nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, (dst.cc_swz[2] << NV40_FP_OP_COND_SWZ_Z_SHIFT) | (dst.cc_swz[3] << NV40_FP_OP_COND_SWZ_W_SHIFT)); - emit_dst(fpc, hw, dst); - emit_src(fpc, hw, 0, s0); - emit_src(fpc, hw, 1, s1); - emit_src(fpc, hw, 2, s2); + emit_dst(fpc, dst); + emit_src(fpc, 0, s0); + emit_src(fpc, 1, s1); + emit_src(fpc, 2, s2); - fp->insn_len += 4; if (fpc->inst_has_const) { + grow_insns(fpc, 4); fp->consts[fp->num_consts].pipe_id = fpc->inst_const_id; - fp->consts[fp->num_consts].hw_id = fp->insn_len; + fp->consts[fp->num_consts].hw_id = fpc->inst_offset + 4; fp->num_consts++; - fp->insn_len += 4; } } @@ -175,11 +192,8 @@ nv40_fp_tex(struct nv40_fpc *fpc, int sat, int op, int unit, struct nv40_sreg dst, int mask, struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) { - struct nv40_fragment_program *fp = fpc->fp; - uint32_t *hw = &fp->insn[fp->insn_len]; - nv40_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); - hw[0] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); + fpc->fp->insn[fpc->inst_offset] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); } static INLINE struct nv40_sreg @@ -622,7 +636,6 @@ nv40_fragprog_translate(struct nv40_context *nv40, fpc = calloc(1, sizeof(struct nv40_fpc)); if (!fpc) return; - fp->insn = calloc(1, 128*4*sizeof(uint32_t)); fpc->fp = fp; fpc->high_temp = -1; fp->num_regs = 2; @@ -667,12 +680,17 @@ nv40_fragprog_translate(struct nv40_context *nv40, } } - if (fpc->inst_has_const == FALSE) - fp->insn[fp->insn_len - 4] |= 0x00000001; - else - fp->insn[fp->insn_len - 8] |= 0x00000001; - fp->insn[fp->insn_len++] = 0x00000001; - + /* Terminate final instruction */ + fp->insn[fpc->inst_offset] |= 0x00000001; + + /* Append NOP + END instruction, may or may not be necessary. */ + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + fp->insn[fpc->inst_offset + 0] = 0x00000001; + fp->insn[fpc->inst_offset + 1] = 0x00000000; + fp->insn[fpc->inst_offset + 2] = 0x00000000; + fp->insn[fpc->inst_offset + 3] = 0x00000000; + fp->translated = TRUE; fp->on_hw = FALSE; out_err: -- cgit v1.2.3 From f287f687feebb6222137a297283d2dfa4ac11252 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 19:28:05 +1100 Subject: nv40: fp support for TGSI_FILE_IMMEDIATE. arbfslight works, nouveau's first GLSL prog :) --- src/mesa/pipe/nv40/nv40_fragprog.c | 97 +++++++++++++++++++++++++++----------- src/mesa/pipe/nv40/nv40_state.h | 12 +++-- 2 files changed, 77 insertions(+), 32 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index f532eccd7b1..8e2689e27d2 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -27,6 +27,8 @@ #define abs(s) nv40_sr_abs((s)) #define scale(s,v) nv40_sr_scale((s), NV40_FP_OP_DST_SCALE_##v) +#define MAX_CONSTS 128 +#define MAX_IMM 32 struct nv40_fpc { struct nv40_fragment_program *fp; @@ -38,10 +40,16 @@ struct nv40_fpc { uint depth_id; uint colour_id; - boolean inst_has_const; - int inst_const_id; - unsigned inst_offset; + + struct { + int pipe; + float vals[4]; + } consts[MAX_CONSTS]; + int nr_consts; + + struct nv40_sreg imm[MAX_IMM]; + unsigned nr_imm; }; static INLINE struct nv40_sreg @@ -54,6 +62,21 @@ temp(struct nv40_fpc *fpc) return nv40_sr(NV40SR_TEMP, idx); } +static INLINE struct nv40_sreg +constant(struct nv40_fpc *fpc, int pipe, float vals[4]) +{ + int idx; + + if (fpc->nr_consts == MAX_CONSTS) + assert(0); + idx = fpc->nr_consts++; + + fpc->consts[idx].pipe = pipe; + if (pipe == -1) + memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float)); + return nv40_sr(NV40SR_CONST, idx); +} + #define arith(cc,s,o,d,m,s0,s1,s2) \ nv40_fp_arith((cc), (s), NV40_FP_OP_OPCODE_##o, \ (d), (m), (s0), (s1), (s2)) @@ -73,7 +96,8 @@ grow_insns(struct nv40_fpc *fpc, int size) static void emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) { - uint32_t *hw = &fpc->fp->insn[fpc->inst_offset]; + struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; uint32_t sr = 0; switch (src.type) { @@ -89,9 +113,23 @@ emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) sr |= (src.index << NV40_FP_REG_SRC_SHIFT); break; case NV40SR_CONST: + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + if (fpc->consts[src.index].pipe >= 0) { + struct nv40_fragment_program_data *fpd; + + fp->consts = realloc(fp->consts, ++fp->nr_consts * + sizeof(*fpd)); + fpd = &fp->consts[fp->nr_consts - 1]; + fpd->offset = fpc->inst_offset + 4; + fpd->index = fpc->consts[src.index].pipe; + } else { + memcpy(&fp->insn[fpc->inst_offset + 4], + fpc->consts[src.index].vals, + sizeof(uint32_t) * 4); + } + sr |= (NV40_FP_REG_TYPE_CONST << NV40_FP_REG_TYPE_SHIFT); - fpc->inst_has_const = TRUE; - fpc->inst_const_id = src.index; break; case NV40SR_NONE: sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); @@ -155,8 +193,6 @@ nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, hw = &fp->insn[fpc->inst_offset]; memset(hw, 0, sizeof(uint32_t) * 4); - fpc->inst_has_const = FALSE; - if (op == NV40_FP_OP_OPCODE_KIL) fp->uses_kil = TRUE; hw[0] |= (op << NV40_FP_OP_OPCODE_SHIFT); @@ -178,13 +214,6 @@ nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, emit_src(fpc, 0, s0); emit_src(fpc, 1, s1); emit_src(fpc, 2, s2); - - if (fpc->inst_has_const) { - grow_insns(fpc, 4); - fp->consts[fp->num_consts].pipe_id = fpc->inst_const_id; - fp->consts[fp->num_consts].hw_id = fpc->inst_offset + 4; - fp->num_consts++; - } } static void @@ -207,7 +236,11 @@ tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) fpc->attrib_map[fsrc->SrcRegister.Index]); break; case TGSI_FILE_CONSTANT: - src = nv40_sr(NV40SR_CONST, fsrc->SrcRegister.Index); + src = constant(fpc, fsrc->SrcRegister.Index, NULL); + break; + case TGSI_FILE_IMMEDIATE: + assert(fsrc->SrcRegister.Index < fpc->nr_imm); + src = fpc->imm[fsrc->SrcRegister.Index]; break; case TGSI_FILE_TEMPORARY: src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index + 1); @@ -386,6 +419,7 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, } break; case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: if (ci == -1 || ci == fsrc->SrcRegister.Index) { ci = fsrc->SrcRegister.Index; src[i] = tgsi_src(fpc, fsrc); @@ -665,6 +699,19 @@ nv40_fragprog_translate(struct nv40_context *nv40, } break; case TGSI_TOKEN_TYPE_IMMEDIATE: + { + struct tgsi_full_immediate *imm; + float vals[4]; + int i; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); + assert(fpc->nr_imm < MAX_IMM); + + for (i = 0; i < imm->Immediate.Size; i++) + vals[i] = imm->u.ImmediateFloat32[i].Float; + fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); + } break; case TGSI_TOKEN_TYPE_INSTRUCTION: { @@ -711,21 +758,17 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) assert(0); } - if (fp->num_consts) { + if (fp->nr_consts) { float *map = ws->buffer_map(ws, nv40->fragprog.constant_buf, PIPE_BUFFER_FLAG_READ); - for (i = 0; i < fp->num_consts; i++) { - uint pid = fp->consts[i].pipe_id; + for (i = 0; i < fp->nr_consts; i++) { + struct nv40_fragment_program_data *fpd = &fp->consts[i]; + uint32_t *p = &fp->insn[fpd->offset]; + uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; - if (pid == -1) + if (!memcmp(p, cb, 4 * sizeof(float))) continue; - - if (!memcmp(&fp->insn[fp->consts[i].hw_id], &map[pid*4], - 4 * sizeof(float))) - continue; - - memcpy(&fp->insn[fp->consts[i].hw_id], &map[pid*4], - 4 * sizeof(float)); + memcpy(p, cb, 4 * sizeof(float)); fp->on_hw = 0; } ws->buffer_unmap(ws, nv40->fragprog.constant_buf); diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 8ab334d2675..cfc8b5637fb 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -84,6 +84,11 @@ struct nv40_vertex_program { uint32_t or; }; +struct nv40_fragment_program_data { + unsigned offset; + unsigned index; +}; + struct nv40_fragment_program { const struct pipe_shader_state *pipe; @@ -93,11 +98,8 @@ struct nv40_fragment_program { uint32_t *insn; int insn_len; - struct { - int pipe_id; - int hw_id; - } consts[256]; - int num_consts; + struct nv40_fragment_program_data *consts; + unsigned nr_consts; struct pipe_buffer_handle *buffer; -- cgit v1.2.3 From 19af6e9161c941d66eac4cd0c782e2a2caadce67 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 20:21:37 +1100 Subject: nv40: typo --- src/mesa/pipe/nv40/nv40_state.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 5b070a280d0..89d91bdb64d 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -309,7 +309,7 @@ nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, struct nv40_context *nv40 = (struct nv40_context *)pipe; nv40->tex_miptree[unit] = miptree; - nv40->tex_dirty |= unit; + nv40->tex_dirty |= (1 << unit); nv40->dirty |= NV40_NEW_TEXTURE; } -- cgit v1.2.3 From b337ecdd8e10b4fa224b2bfb7fdfeda5931d7a6e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 20:44:44 +1100 Subject: nv40: fix culling --- src/mesa/pipe/nv40/nv40_state.c | 24 ++++++++++++++++++------ 1 file changed, 18 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 89d91bdb64d..c1a882feffb 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -345,26 +345,38 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; if (cso->front_winding == PIPE_WINDING_CCW) { - rs->front_face = 0x0901; + rs->front_face = NV40TCL_FRONT_FACE_CCW; rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); } else { - rs->front_face = 0x0900; + rs->front_face = NV40TCL_FRONT_FACE_CW; rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); } - rs->cull_face_en = 0; - rs->cull_face = 0x0900; switch (cso->cull_mode) { case PIPE_WINDING_CCW: - rs->cull_face = 0x0901; - /* fall-through */ + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CCW) + rs->cull_face = NV40TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV40TCL_CULL_FACE_BACK; + break; case PIPE_WINDING_CW: rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CW) + rs->cull_face = NV40TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV40TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_BOTH: + rs->cull_face_en = 1; + rs->cull_face = NV40TCL_CULL_FACE_FRONT_AND_BACK; break; case PIPE_WINDING_NONE: default: + rs->cull_face_en = 0; + rs->cull_face = 0; break; } -- cgit v1.2.3 From 4c2a3356cf976e9e30f888bf57f6b140482a1e15 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 21:26:09 +1100 Subject: nouveau: some cleanups --- .../dri/nouveau_winsys/nouveau_winsys_softpipe.c | 5 +- src/mesa/pipe/nv40/nv40_fragprog.c | 22 +++---- src/mesa/pipe/nv40/nv40_state.h | 4 +- src/mesa/pipe/nv40/nv40_surface.c | 68 ---------------------- 4 files changed, 13 insertions(+), 86 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c index ebce4ddede1..3908c175088 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c @@ -31,8 +31,9 @@ #include "imports.h" -#include "pipe/softpipe/sp_winsys.h" #include "pipe/p_defines.h" +#include "pipe/p_format.h" +#include "pipe/softpipe/sp_winsys.h" #include "nouveau_context.h" #include "nouveau_winsys_pipe.h" @@ -51,7 +52,7 @@ nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_S8Z24_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: return TRUE; default: break; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 8e2689e27d2..9cc1267108d 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -36,6 +36,7 @@ struct nv40_fpc { int high_temp; int temp_temp_count; + int num_regs; uint depth_id; uint colour_id; @@ -160,12 +161,12 @@ emit_dst(struct nv40_fpc *fpc, struct nv40_sreg dst) switch (dst.type) { case NV40SR_TEMP: - if (fp->num_regs < (dst.index + 1)) - fp->num_regs = dst.index + 1; + if (fpc->num_regs < (dst.index + 1)) + fpc->num_regs = dst.index + 1; break; case NV40SR_OUTPUT: if (dst.index == 1) { - fp->writes_depth = 1; + fp->fp_control |= 0xe; } else { hw[0] |= NV40_FP_OP_OUT_REG_HALF; } @@ -194,7 +195,7 @@ nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, memset(hw, 0, sizeof(uint32_t) * 4); if (op == NV40_FP_OP_OPCODE_KIL) - fp->uses_kil = TRUE; + fp->fp_control |= NV40TCL_FP_CONTROL_KIL; hw[0] |= (op << NV40_FP_OP_OPCODE_SHIFT); hw[0] |= (mask << NV40_FP_OP_OUTMASK_SHIFT); hw[2] |= (dst.dst_scale << NV40_FP_OP_DST_SCALE_SHIFT); @@ -672,7 +673,7 @@ nv40_fragprog_translate(struct nv40_context *nv40, return; fpc->fp = fp; fpc->high_temp = -1; - fp->num_regs = 2; + fpc->num_regs = 2; tgsi_parse_init(&parse, fp->pipe->tokens); @@ -727,6 +728,8 @@ nv40_fragprog_translate(struct nv40_context *nv40, } } + fp->fp_control |= fpc->num_regs << NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT; + /* Terminate final instruction */ fp->insn[fpc->inst_offset] |= 0x00000001; @@ -749,7 +752,6 @@ void nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) { struct pipe_winsys *ws = nv40->pipe.winsys; - uint32_t fp_control; int i; if (!fp->translated) { @@ -790,18 +792,12 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->on_hw = TRUE; } - fp_control = fp->num_regs << NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT; - if (fp->uses_kil) - fp_control |= NV40TCL_FP_CONTROL_KIL; - if (fp->writes_depth) - fp_control |= 0xe; - BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); - OUT_RING (fp_control); + OUT_RING (fp->fp_control); nv40->fragprog.active = fp; } diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index cfc8b5637fb..eb8473842f6 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -103,9 +103,7 @@ struct nv40_fragment_program { struct pipe_buffer_handle *buffer; - boolean uses_kil; - boolean writes_depth; - int num_regs; + uint32_t fp_control; }; struct nv40_depth_push { diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index da9e18ac136..15dc7f23252 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -46,74 +46,6 @@ h = ps->height -y; \ } while(0) - -/** - * Note: this is exactly like a8r8g8b8_get_tile() in sp_surface.c - * Share it someday. - */ -static void -nv40_get_tile_rgba(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, float *p) -{ - const unsigned *src - = ((const unsigned *) (ps->map + ps->offset)) - + y * ps->pitch + x; - unsigned i, j; - unsigned w0 = w; - - CLIP_TILE; - - switch (ps->format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - for (i = 0; i < h; i++) { - float *pRow = p; - for (j = 0; j < w; j++) { - const unsigned pixel = src[j]; - pRow[0] = UBYTE_TO_FLOAT((pixel >> 16) & 0xff); - pRow[1] = UBYTE_TO_FLOAT((pixel >> 8) & 0xff); - pRow[2] = UBYTE_TO_FLOAT((pixel >> 0) & 0xff); - pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); - pRow += 4; - } - src += ps->pitch; - p += w0 * 4; - } - break; - case PIPE_FORMAT_Z24S8_UNORM: - { - const float scale = 1.0 / (float) 0xffffff; - for (i = 0; i < h; i++) { - float *pRow = p; - for (j = 0; j < w; j++) { - const unsigned pixel = src[j]; - pRow[0] = - pRow[1] = - pRow[2] = - pRow[3] = ((pixel & 0xffffff) >> 8) * scale; - pRow += 4; - } - src += ps->pitch; - p += w0 * 4; - } - } - break; - default: - assert(0); - } -} - - -static void -nv40_put_tile_rgba(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, const float *p) -{ - /* TODO */ - assert(0); -} - - /* * XXX note: same as code in sp_surface.c */ -- cgit v1.2.3 From f7e99bf22cd03a3d38b1d6263c96a0c2b85f29b7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Dec 2007 23:45:30 +1100 Subject: nv40: don't use vertex buffers for static vertices. --- src/mesa/pipe/nv40/nv40_vbo.c | 71 +++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 69 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 62c877b2dc6..4bd6c44e78b 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -32,6 +32,68 @@ nv40_vbo_type(uint format) } } +static boolean +nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, + struct pipe_vertex_element *ve, + struct pipe_vertex_buffer *vb) +{ + struct pipe_winsys *ws = nv40->pipe.winsys; + int type, ncomp; + void *map; + + type = nv40_vbo_type(ve->src_format); + ncomp = nv40_vbo_ncomp(ve->src_format); + + map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_FLAG_READ); + map += vb->buffer_offset + ve->src_offset; + + switch (type) { + case NV40TCL_VTXFMT_TYPE_FLOAT: + { + float *v = map; + + BEGIN_RING(curie, NV40TCL_VTX_ATTR_4F_X(attrib), 4); + switch (ncomp) { + case 4: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(v[3]); + break; + case 3: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(1.0); + break; + case 2: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + case 1: + OUT_RINGf(v[0]); + OUT_RINGf(0.0); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + } + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + + ws->buffer_unmap(ws, vb->buffer); + + return TRUE; +} + static void nv40_vbo_arrays_update(struct nv40_context *nv40) { @@ -49,13 +111,11 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) num_hw++; inputs = vp->ir; - BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); for (hw = 0; hw < num_hw; hw++) { struct pipe_vertex_element *ve; struct pipe_vertex_buffer *vb; if (!(inputs & (1 << hw))) { - OUT_RING(0); vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; continue; } @@ -63,6 +123,13 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) ve = &nv40->vtxelt[hw]; vb = &nv40->vtxbuf[ve->vertex_buffer_index]; + if (vb->pitch == 0) { + vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; + if (nv40_vbo_static_attrib(nv40, hw, ve, vb) == TRUE) + continue; + } + + BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(hw), 1); OUT_RELOC(vb->buffer, vb->buffer_offset + ve->src_offset, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, -- cgit v1.2.3 From 66013a252f556447ca02a257bf15909817022433 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Dec 2007 00:11:41 +1100 Subject: nv40: colormask thinko --- src/mesa/pipe/nv40/nv40_state.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index c1a882feffb..6a27854625c 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -93,14 +93,16 @@ nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) OUT_RING (cb->b_enable); OUT_RING (cb->b_srcfunc); OUT_RING (cb->b_dstfunc); - BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 2); + BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 1); OUT_RING (cb->b_eqn); - OUT_RING (cb->c_mask); } else { BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 1); OUT_RING (0); } + BEGIN_RING(curie, NV40TCL_COLOR_MASK, 1); + OUT_RING (cb->c_mask); + if (cb->l_enable) { BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); OUT_RING (cb->l_enable); -- cgit v1.2.3 From 7dcef36f4c5209bdd472b7f30cf9314344d2708e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Dec 2007 00:46:30 +1100 Subject: nouveau: bump pushbuf size a little, so a "full" packet can fit --- src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index a922300ff5e..36f201712aa 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -27,6 +27,8 @@ #include "nouveau_drmif.h" #include "nouveau_dma.h" +#define PB_RSVD_DWORDS 2 + int nouveau_pushbuf_init(struct nouveau_channel *chan) { @@ -72,7 +74,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) if (!nvpb) goto out_realloc; - if (nvpb->base.remaining == nvpb->res->size / 4) + if (nvpb->base.remaining == (nvpb->res->size / 4) - PB_RSVD_DWORDS) return 0; nvchan->pb_tail = NULL; @@ -154,13 +156,13 @@ out_realloc: if (!nvpb) return -ENOMEM; - while (nouveau_resource_alloc(nvchan->pb_heap, 0x2000, NULL, + while (nouveau_resource_alloc(nvchan->pb_heap, 0x2100, NULL, &nvpb->res)) { nouveau_fence_flush(chan); } nvpb->base.channel = chan; - nvpb->base.remaining = nvpb->res->size / 4; + nvpb->base.remaining = (nvpb->res->size / 4) - PB_RSVD_DWORDS; nvpb->base.cur = &nvchan->pushbuf[nvpb->res->start/4]; nvchan->pb_tail = &nvpb->base; -- cgit v1.2.3 From 8043d2d2618120828d1a94c861ac8fc16fee7af7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Dec 2007 21:45:13 +1100 Subject: nv40: clean state a bit --- src/mesa/pipe/nv40/nv40_state.c | 65 ++++++++++++++--------------------------- 1 file changed, 22 insertions(+), 43 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 6a27854625c..ceb6f4658bf 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -15,10 +15,8 @@ nv40_alpha_test_state_create(struct pipe_context *pipe, at = malloc(sizeof(struct nv40_alpha_test_state)); at->enabled = cso->enabled ? 1 : 0; - if (at->enabled) { - at->func = nvgl_comparison_op(cso->func); - at->ref = float_to_ubyte(cso->ref); - } + at->func = nvgl_comparison_op(cso->func); + at->ref = float_to_ubyte(cso->ref); return (void *)at; } @@ -29,15 +27,10 @@ nv40_alpha_test_state_bind(struct pipe_context *pipe, void *hwcso) struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nv40_alpha_test_state *at = hwcso; - if (at->enabled) { - BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); - OUT_RING (at->enabled); - OUT_RING (at->func); - OUT_RING (at->ref); - } else { - BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 1); - OUT_RING (0); - } + BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + OUT_RING (at->enabled); + OUT_RING (at->func); + OUT_RING (at->ref); } static void @@ -55,19 +48,15 @@ nv40_blend_state_create(struct pipe_context *pipe, cb = malloc(sizeof(struct nv40_blend_state)); cb->b_enable = cso->blend_enable ? 1 : 0; - if (cb->b_enable) { - cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | - (nvgl_blend_func(cso->rgb_src_factor))); - cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | - (nvgl_blend_func(cso->rgb_dst_factor))); - cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | - (nvgl_blend_eqn(cso->rgb_func))); - } + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | + (nvgl_blend_eqn(cso->rgb_func))); cb->l_enable = cso->logicop_enable ? 1 : 0; - if (cb->l_enable) { - cb->l_op = nvgl_logicop_func(cso->logicop_func); - } + cb->l_op = nvgl_logicop_func(cso->logicop_func); cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | @@ -88,29 +77,19 @@ nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) BEGIN_RING(curie, NV40TCL_DITHER_ENABLE, 1); OUT_RING (cb->d_enable); - if (cb->b_enable) { - BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 3); - OUT_RING (cb->b_enable); - OUT_RING (cb->b_srcfunc); - OUT_RING (cb->b_dstfunc); - BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 1); - OUT_RING (cb->b_eqn); - } else { - BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 1); - OUT_RING (0); - } + BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 3); + OUT_RING (cb->b_enable); + OUT_RING (cb->b_srcfunc); + OUT_RING (cb->b_dstfunc); + BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 1); + OUT_RING (cb->b_eqn); BEGIN_RING(curie, NV40TCL_COLOR_MASK, 1); OUT_RING (cb->c_mask); - if (cb->l_enable) { - BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (cb->l_enable); - OUT_RING (cb->l_op); - } else { - BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); - OUT_RING (0); - } + BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + OUT_RING (cb->l_enable); + OUT_RING (cb->l_op); } static void -- cgit v1.2.3 From c5ffd8e9afd3df1d4daf189344ec8cbf8a021143 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Dec 2007 21:49:16 +1100 Subject: nv40: fix valgrind complaints --- src/mesa/pipe/nv40/nv40_fragprog.c | 1 + src/mesa/pipe/nv40/nv40_miptree.c | 9 +++------ 2 files changed, 4 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 9cc1267108d..c95f31e1439 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -124,6 +124,7 @@ emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) fpd = &fp->consts[fp->nr_consts - 1]; fpd->offset = fpc->inst_offset + 4; fpd->index = fpc->consts[src.index].pipe; + memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4); } else { memcpy(&fp->insn[fpc->inst_offset + 4], fpc->consts[src.index].vals, diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 9d4fbb68264..625c2fc32d7 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -55,17 +55,14 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) static void nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) { - struct pipe_texture *mt = *pt; struct pipe_winsys *ws = pipe->winsys; struct nv40_miptree *nv40mt; - *pt = NULL; - - nv40mt = calloc(1, sizeof(struct nv40_miptree)); + nv40mt = realloc(*pt, sizeof(struct nv40_miptree)); if (!nv40mt) return; + *pt = NULL; - memcpy(&nv40mt->base, mt, sizeof(struct pipe_texture)); nv40_miptree_layout(nv40mt); nv40mt->buffer = ws->buffer_create(ws, 256, 0, 0); @@ -91,7 +88,7 @@ nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) int l; ws->buffer_reference(ws, &nv40mt->buffer, NULL); - for (l = 0; l < PIPE_MAX_TEXTURE_LEVELS; l++) { + for (l = mt->first_level; l <= mt->last_level; l++) { if (nv40mt->level[l].image_offset) free(nv40mt->level[l].image_offset); } -- cgit v1.2.3 From 3af35c6dfa4db1cbf830731882e61d57741f2898 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 18 Dec 2007 11:27:44 +1100 Subject: nouveau: match latest gallium changes --- src/mesa/pipe/nv40/nv40_state.c | 126 +++++++++++------------------- src/mesa/pipe/nv40/nv40_state.h | 29 ++++--- src/mesa/pipe/nv40/nv40_surface.c | 91 ++-------------------- src/mesa/pipe/nv50/nv50_state.c | 38 +++------ src/mesa/pipe/nv50/nv50_surface.c | 158 ++------------------------------------ 5 files changed, 79 insertions(+), 363 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index ceb6f4658bf..12b2c2a8e8f 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -6,39 +6,6 @@ #include "nv40_dma.h" #include "nv40_state.h" -static void * -nv40_alpha_test_state_create(struct pipe_context *pipe, - const struct pipe_alpha_test_state *cso) -{ - struct nv40_alpha_test_state *at; - - at = malloc(sizeof(struct nv40_alpha_test_state)); - - at->enabled = cso->enabled ? 1 : 0; - at->func = nvgl_comparison_op(cso->func); - at->ref = float_to_ubyte(cso->ref); - - return (void *)at; -} - -static void -nv40_alpha_test_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nv40_alpha_test_state *at = hwcso; - - BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); - OUT_RING (at->enabled); - OUT_RING (at->func); - OUT_RING (at->ref); -} - -static void -nv40_alpha_test_state_delete(struct pipe_context *pipe, void *hwcso) -{ - free(hwcso); -} - static void * nv40_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) @@ -414,61 +381,59 @@ nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) free(hwcso); } +static void +nv40_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, + unsigned idx, struct nv40_stencil_push *hw) +{ + hw->enable = cso->stencil[idx].enabled ? 1 : 0; + hw->wmask = cso->stencil[idx].write_mask; + hw->func = nvgl_comparison_op(cso->stencil[idx].func); + hw->ref = cso->stencil[idx].ref_value; + hw->vmask = cso->stencil[idx].value_mask; + hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); + hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); + hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); +} + static void * -nv40_depth_stencil_state_create(struct pipe_context *pipe, - const struct pipe_depth_stencil_state *cso) +nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) { - struct nv40_depth_stencil_state *zs; + struct nv40_depth_stencil_alpha_state *hw; - /*XXX: ignored: - * depth.occlusion_count - * depth.clear - * stencil.clear_value - */ - zs = malloc(sizeof(struct nv40_depth_stencil_state)); - - zs->depth.func = nvgl_comparison_op(cso->depth.func); - zs->depth.write_enable = cso->depth.writemask ? 1 : 0; - zs->depth.test_enable = cso->depth.enabled ? 1 : 0; - - zs->stencil.back.enable = cso->stencil.back_enabled ? 1 : 0; - zs->stencil.back.wmask = cso->stencil.write_mask[1]; - zs->stencil.back.func = - nvgl_comparison_op(cso->stencil.back_func); - zs->stencil.back.ref = cso->stencil.ref_value[1]; - zs->stencil.back.vmask = cso->stencil.value_mask[1]; - zs->stencil.back.fail = nvgl_stencil_op(cso->stencil.back_fail_op); - zs->stencil.back.zfail = nvgl_stencil_op(cso->stencil.back_zfail_op); - zs->stencil.back.zpass = nvgl_stencil_op(cso->stencil.back_zpass_op); - - zs->stencil.front.enable= cso->stencil.front_enabled ? 1 : 0; - zs->stencil.front.wmask = cso->stencil.write_mask[0]; - zs->stencil.front.func = - nvgl_comparison_op(cso->stencil.front_func); - zs->stencil.front.ref = cso->stencil.ref_value[0]; - zs->stencil.front.vmask = cso->stencil.value_mask[0]; - zs->stencil.front.fail = nvgl_stencil_op(cso->stencil.front_fail_op); - zs->stencil.front.zfail = nvgl_stencil_op(cso->stencil.front_zfail_op); - zs->stencil.front.zpass = nvgl_stencil_op(cso->stencil.front_zpass_op); - - return (void *)zs; + hw = malloc(sizeof(struct nv40_depth_stencil_alpha_state)); + + hw->depth.func = nvgl_comparison_op(cso->depth.func); + hw->depth.write_enable = cso->depth.writemask ? 1 : 0; + hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + + nv40_translate_stencil(cso, 0, &hw->stencil.front); + nv40_translate_stencil(cso, 1, &hw->stencil.back); + + hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; + hw->alpha.func = nvgl_comparison_op(cso->alpha.func); + hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + + return (void *)hw; } static void -nv40_depth_stencil_state_bind(struct pipe_context *pipe, void *hwcso) +nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nv40_depth_stencil_state *zs = hwcso; + struct nv40_depth_stencil_alpha_state *hw = hwcso; BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&zs->depth, 3); + OUT_RINGp ((uint32_t *)&hw->depth, 3); BEGIN_RING(curie, NV40TCL_STENCIL_BACK_ENABLE, 16); - OUT_RINGp ((uint32_t *)&zs->stencil.back, 8); - OUT_RINGp ((uint32_t *)&zs->stencil.front, 8); + OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); + OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); + BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); } static void -nv40_depth_stencil_state_delete(struct pipe_context *pipe, void *hwcso) +nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { free(hwcso); } @@ -785,10 +750,6 @@ nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, void nv40_init_state_functions(struct nv40_context *nv40) { - nv40->pipe.create_alpha_test_state = nv40_alpha_test_state_create; - nv40->pipe.bind_alpha_test_state = nv40_alpha_test_state_bind; - nv40->pipe.delete_alpha_test_state = nv40_alpha_test_state_delete; - nv40->pipe.create_blend_state = nv40_blend_state_create; nv40->pipe.bind_blend_state = nv40_blend_state_bind; nv40->pipe.delete_blend_state = nv40_blend_state_delete; @@ -802,9 +763,12 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind; nv40->pipe.delete_rasterizer_state = nv40_rasterizer_state_delete; - nv40->pipe.create_depth_stencil_state = nv40_depth_stencil_state_create; - nv40->pipe.bind_depth_stencil_state = nv40_depth_stencil_state_bind; - nv40->pipe.delete_depth_stencil_state = nv40_depth_stencil_state_delete; + nv40->pipe.create_depth_stencil_alpha_state = + nv40_depth_stencil_alpha_state_create; + nv40->pipe.bind_depth_stencil_alpha_state = + nv40_depth_stencil_alpha_state_bind; + nv40->pipe.delete_depth_stencil_alpha_state = + nv40_depth_stencil_alpha_state_delete; nv40->pipe.create_vs_state = nv40_vp_state_create; nv40->pipe.bind_vs_state = nv40_vp_state_bind; diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index eb8473842f6..e1a9d58525a 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -3,12 +3,6 @@ #include "pipe/p_state.h" -struct nv40_alpha_test_state { - uint32_t enabled; - uint32_t func; - uint32_t ref; -}; - struct nv40_blend_state { uint32_t b_enable; uint32_t b_srcfunc; @@ -106,12 +100,6 @@ struct nv40_fragment_program { uint32_t fp_control; }; -struct nv40_depth_push { - uint32_t func; - uint32_t write_enable; - uint32_t test_enable; -}; - struct nv40_stencil_push { uint32_t enable; uint32_t wmask; @@ -123,12 +111,23 @@ struct nv40_stencil_push { uint32_t zpass; }; -struct nv40_depth_stencil_state { - struct nv40_depth_push depth; - union { +struct nv40_depth_stencil_alpha_state { + struct { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; + } depth; + + struct { struct nv40_stencil_push back; struct nv40_stencil_push front; } stencil; + + struct { + uint32_t enabled; + uint32_t func; + uint32_t ref; + } alpha; }; struct nv40_miptree { diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index 15dc7f23252..6a16a280c24 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -31,88 +31,7 @@ #include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" -#include "pipe/softpipe/sp_rgba_tile.h" - - -#define CLIP_TILE \ - do { \ - if (x >= ps->width) \ - return; \ - if (y >= ps->height) \ - return; \ - if (x + w > ps->width) \ - w = ps->width - x; \ - if (y + h > ps->height) \ - h = ps->height -y; \ - } while(0) - -/* - * XXX note: same as code in sp_surface.c - */ -static void -nv40_get_tile(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, - void *p, int dst_stride) -{ - const uint cpp = ps->cpp; - const uint w0 = w; - const ubyte *pSrc; - ubyte *pDest; - uint i; - - assert(ps->map); - - CLIP_TILE; - - if (dst_stride == 0) { - dst_stride = w0 * cpp; - } - - pSrc = ps->map + ps->offset + (y * ps->pitch + x) * cpp; - pDest = (ubyte *) p; - - for (i = 0; i < h; i++) { - memcpy(pDest, pSrc, w0 * cpp); - pDest += dst_stride; - pSrc += ps->pitch * cpp; - } -} - - -/* - * XXX note: same as code in sp_surface.c - */ -static void -nv40_put_tile(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, - const void *p, int src_stride) -{ - const uint cpp = ps->cpp; - const uint w0 = w; - const ubyte *pSrc; - ubyte *pDest; - uint i; - - assert(ps->map); - - CLIP_TILE; - - if (src_stride == 0) { - src_stride = w0 * cpp; - } - - pSrc = (const ubyte *) p; - pDest = ps->map + ps->offset + (y * ps->pitch + x) * cpp; - - for (i = 0; i < h; i++) { - memcpy(pDest, pSrc, w0 * cpp); - pDest += ps->pitch * cpp; - pSrc += src_stride; - } -} - +#include "pipe/util/p_tile.h" static struct pipe_surface * nv40_get_tex_surface(struct pipe_context *pipe, @@ -185,10 +104,10 @@ void nv40_init_surface_functions(struct nv40_context *nv40) { nv40->pipe.get_tex_surface = nv40_get_tex_surface; - nv40->pipe.get_tile = nv40_get_tile; - nv40->pipe.put_tile = nv40_put_tile; - nv40->pipe.get_tile_rgba = softpipe_get_tile_rgba; - nv40->pipe.put_tile_rgba = softpipe_put_tile_rgba; + nv40->pipe.get_tile = pipe_get_tile_raw; + nv40->pipe.put_tile = pipe_put_tile_raw; + nv40->pipe.get_tile_rgba = pipe_get_tile_rgba; + nv40->pipe.put_tile_rgba = pipe_put_tile_rgba; nv40->pipe.surface_data = nv40_surface_data; nv40->pipe.surface_copy = nv40_surface_copy; nv40->pipe.surface_fill = nv40_surface_fill; diff --git a/src/mesa/pipe/nv50/nv50_state.c b/src/mesa/pipe/nv50/nv50_state.c index 0a27a4676eb..25eac41c2fc 100644 --- a/src/mesa/pipe/nv50/nv50_state.c +++ b/src/mesa/pipe/nv50/nv50_state.c @@ -6,23 +6,6 @@ #include "nv50_dma.h" #include "nv50_state.h" -static void * -nv50_alpha_test_state_create(struct pipe_context *pipe, - const struct pipe_alpha_test_state *cso) -{ - return NULL; -} - -static void -nv50_alpha_test_state_bind(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_alpha_test_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - static void * nv50_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) @@ -82,19 +65,19 @@ nv50_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) } static void * -nv50_depth_stencil_state_create(struct pipe_context *pipe, - const struct pipe_depth_stencil_state *cso) +nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) { return NULL; } static void -nv50_depth_stencil_state_bind(struct pipe_context *pipe, void *hwcso) +nv50_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { } static void -nv50_depth_stencil_state_delete(struct pipe_context *pipe, void *hwcso) +nv50_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { } @@ -187,10 +170,6 @@ nv50_set_vertex_element(struct pipe_context *pipe, unsigned index, void nv50_init_state_functions(struct nv50_context *nv50) { - nv50->pipe.create_alpha_test_state = nv50_alpha_test_state_create; - nv50->pipe.bind_alpha_test_state = nv50_alpha_test_state_bind; - nv50->pipe.delete_alpha_test_state = nv50_alpha_test_state_delete; - nv50->pipe.create_blend_state = nv50_blend_state_create; nv50->pipe.bind_blend_state = nv50_blend_state_bind; nv50->pipe.delete_blend_state = nv50_blend_state_delete; @@ -204,9 +183,12 @@ nv50_init_state_functions(struct nv50_context *nv50) nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; nv50->pipe.delete_rasterizer_state = nv50_rasterizer_state_delete; - nv50->pipe.create_depth_stencil_state = nv50_depth_stencil_state_create; - nv50->pipe.bind_depth_stencil_state = nv50_depth_stencil_state_bind; - nv50->pipe.delete_depth_stencil_state = nv50_depth_stencil_state_delete; + nv50->pipe.create_depth_stencil_alpha_state = + nv50_depth_stencil_alpha_state_create; + nv50->pipe.bind_depth_stencil_alpha_state = + nv50_depth_stencil_alpha_state_bind; + nv50->pipe.delete_depth_stencil_alpha_state = + nv50_depth_stencil_alpha_state_delete; nv50->pipe.create_vs_state = nv50_vp_state_create; nv50->pipe.bind_vs_state = nv50_vp_state_bind; diff --git a/src/mesa/pipe/nv50/nv50_surface.c b/src/mesa/pipe/nv50/nv50_surface.c index 7e294cdfdfc..44e6728b264 100644 --- a/src/mesa/pipe/nv50/nv50_surface.c +++ b/src/mesa/pipe/nv50/nv50_surface.c @@ -31,155 +31,7 @@ #include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" - - -#define CLIP_TILE \ - do { \ - if (x >= ps->width) \ - return; \ - if (y >= ps->height) \ - return; \ - if (x + w > ps->width) \ - w = ps->width - x; \ - if (y + h > ps->height) \ - h = ps->height -y; \ - } while(0) - - -/** - * Note: this is exactly like a8r8g8b8_get_tile() in sp_surface.c - * Share it someday. - */ -static void -nv50_get_tile_rgba(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, float *p) -{ - const unsigned *src - = ((const unsigned *) (ps->map + ps->offset)) - + y * ps->pitch + x; - unsigned i, j; - unsigned w0 = w; - - CLIP_TILE; - - switch (ps->format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - for (i = 0; i < h; i++) { - float *pRow = p; - for (j = 0; j < w; j++) { - const unsigned pixel = src[j]; - pRow[0] = UBYTE_TO_FLOAT((pixel >> 16) & 0xff); - pRow[1] = UBYTE_TO_FLOAT((pixel >> 8) & 0xff); - pRow[2] = UBYTE_TO_FLOAT((pixel >> 0) & 0xff); - pRow[3] = UBYTE_TO_FLOAT((pixel >> 24) & 0xff); - pRow += 4; - } - src += ps->pitch; - p += w0 * 4; - } - break; - case PIPE_FORMAT_Z24S8_UNORM: - { - const float scale = 1.0 / (float) 0xffffff; - for (i = 0; i < h; i++) { - float *pRow = p; - for (j = 0; j < w; j++) { - const unsigned pixel = src[j]; - pRow[0] = - pRow[1] = - pRow[2] = - pRow[3] = ((pixel & 0xffffff) >> 8) * scale; - pRow += 4; - } - src += ps->pitch; - p += w0 * 4; - } - } - break; - default: - assert(0); - } -} - - -static void -nv50_put_tile_rgba(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, const float *p) -{ - /* TODO */ - assert(0); -} - - -/* - * XXX note: same as code in sp_surface.c - */ -static void -nv50_get_tile(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, - void *p, int dst_stride) -{ - const uint cpp = ps->cpp; - const uint w0 = w; - const ubyte *pSrc; - ubyte *pDest; - uint i; - - assert(ps->map); - - CLIP_TILE; - - if (dst_stride == 0) { - dst_stride = w0 * cpp; - } - - pSrc = ps->map + ps->offset + (y * ps->pitch + x) * cpp; - pDest = (ubyte *) p; - - for (i = 0; i < h; i++) { - memcpy(pDest, pSrc, w0 * cpp); - pDest += dst_stride; - pSrc += ps->pitch * cpp; - } -} - - -/* - * XXX note: same as code in sp_surface.c - */ -static void -nv50_put_tile(struct pipe_context *pipe, - struct pipe_surface *ps, - uint x, uint y, uint w, uint h, - const void *p, int src_stride) -{ - const uint cpp = ps->cpp; - const uint w0 = w; - const ubyte *pSrc; - ubyte *pDest; - uint i; - - assert(ps->map); - - CLIP_TILE; - - if (src_stride == 0) { - src_stride = w0 * cpp; - } - - pSrc = (const ubyte *) p; - pDest = ps->map + ps->offset + (y * ps->pitch + x) * cpp; - - for (i = 0; i < h; i++) { - memcpy(pDest, pSrc, w0 * cpp); - pDest += ps->pitch * cpp; - pSrc += src_stride; - } -} - +#include "pipe/util/p_tile.h" static struct pipe_surface * nv50_get_tex_surface(struct pipe_context *pipe, @@ -230,10 +82,10 @@ void nv50_init_surface_functions(struct nv50_context *nv50) { nv50->pipe.get_tex_surface = nv50_get_tex_surface; - nv50->pipe.get_tile = nv50_get_tile; - nv50->pipe.put_tile = nv50_put_tile; - nv50->pipe.get_tile_rgba = nv50_get_tile_rgba; - nv50->pipe.put_tile_rgba = nv50_put_tile_rgba; + nv50->pipe.get_tile = pipe_get_tile_raw; + nv50->pipe.put_tile = pipe_put_tile_raw; + nv50->pipe.get_tile_rgba = pipe_get_tile_rgba; + nv50->pipe.put_tile_rgba = pipe_put_tile_rgba; nv50->pipe.surface_data = nv50_surface_data; nv50->pipe.surface_copy = nv50_surface_copy; nv50->pipe.surface_fill = nv50_surface_fill; -- cgit v1.2.3 From d67c2c1cd5d0bb77a08a46d2e9a2bdae6b2c8ff3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 19 Dec 2007 13:28:22 +1100 Subject: nouveau: remove sync hack, gallium got sort-of fixed. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 15 +++++---------- src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h | 1 - src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 10 ---------- 3 files changed, 5 insertions(+), 21 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index d31026b52d5..46df588b5ba 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -253,20 +253,15 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, if (!nvbo->drm.size) { nouveau_bo_realloc_gpu(nvbo, flags, nvbo->base.size); nouveau_bo_upload(nvbo); + if (!nvbo->user) { + free(nvbo->sysmem); + nvbo->sysmem = NULL; + } } else if (nvbo->user) { nouveau_bo_upload(nvbo); - } else - if (nvbo->base.map) { - nouveau_bo_upload(nvbo); - nvbo->sync_hack = 1; - } - - if (!nvbo->user && !nvbo->base.map) { - free(nvbo->sysmem); - nvbo->sysmem = NULL; } - + if (nvbo->fence) nouveau_fence_del(&nvbo->fence); nouveau_fence_ref(fence, &nvbo->fence); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 6d6633fac30..9978652dd6d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -252,7 +252,6 @@ struct nouveau_bo_priv { int user; int refcount; - int sync_hack; }; #define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 36f201712aa..981c4dd7dd7 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -68,7 +68,6 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); struct nouveau_pushbuf_bo *pbbo; struct nouveau_fence *fence = NULL; - int sync_hack = 0; int ret; if (!nvpb) @@ -90,9 +89,6 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) ret = nouveau_bo_validate(chan, bo, fence, pbbo->flags); assert (ret == 0); - sync_hack |= nouveau_bo(bo)->sync_hack; - nouveau_bo(bo)->sync_hack = 0; - while ((r = ptr_to_pbrel(pbbo->relocs))) { uint32_t push; @@ -144,12 +140,6 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) /* Kickoff */ FIRE_RING_CH(chan); - if (sync_hack) { - struct nouveau_fence *f = NULL; - nouveau_fence_ref(nvpb->fence, &f); - nouveau_fence_wait(&f); - } - /* Allocate space for next push buffer */ out_realloc: nvpb = calloc(1, sizeof(struct nouveau_pushbuf_priv)); -- cgit v1.2.3 From bf2410ca73d1eabc7134e51855d0b603897229e4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 19 Dec 2007 19:40:38 +1100 Subject: nouveau: nuke subchan auto-assign. It's annoying for several reasons, especially in its current form. May possibly be reincarnated later (DDX depends on it these days), but in not quite the same way. --- .../drivers/dri/nouveau_winsys/nouveau_context.h | 1 + src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 28 ---------------------- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h | 26 -------------------- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 6 ----- .../drivers/dri/nouveau_winsys/nouveau_fence.c | 11 +++++++-- .../drivers/dri/nouveau_winsys/nouveau_local.h | 11 ++++++++- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 17 ++++++++----- src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 4 ++++ src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c | 2 +- src/mesa/pipe/nouveau/nouveau_grobj.h | 6 ----- 10 files changed, 36 insertions(+), 76 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index cd59b7cc436..294b66e5188 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -54,6 +54,7 @@ struct nouveau_context { struct nouveau_grobj *NvM2MF; struct nouveau_grobj *Nv2D; uint32_t next_handle; + uint32_t next_subchannel; uint32_t next_sequence; /* pipe_surface accel */ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index b89f1b5cf5f..d00fddbff2f 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -126,34 +126,6 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size) return 0; } -#ifdef NOUVEAU_DMA_SUBCHAN_LRU -void -nouveau_dma_subc_bind(struct nouveau_grobj *grobj) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(grobj->channel); - int subc = -1, i; - - for (i = 0; i < 8; i++) { - if (nvchan->subchannel[i].grobj && - nvchan->subchannel[i].grobj->bound == - NOUVEAU_GROBJ_EXPLICIT_BIND) - continue; - if (nvchan->subchannel[i].seq < nvchan->subchannel[subc].seq) - subc = i; - } - assert(subc >= 0); - - if (nvchan->subchannel[subc].grobj) - nvchan->subchannel[subc].grobj->bound = 0; - nvchan->subchannel[subc].grobj = grobj; - grobj->subc = subc; - grobj->bound = NOUVEAU_GROBJ_BOUND; - - BEGIN_RING_CH(grobj->channel, grobj, 0, 1); - nouveau_dma_out (grobj->channel, grobj->handle); -} -#endif - #ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF static void nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h index d4dd3054f3d..c4667cc5c02 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h @@ -94,12 +94,6 @@ nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, struct nouveau_channel_priv *nvchan = nouveau_channel(chan); int push_size = size + 1; -#ifdef NOUVEAU_DMA_SUBCHAN_LRU - if (grobj->bound == NOUVEAU_GROBJ_UNBOUND) - nouveau_dma_subc_bind(grobj); - nvchan->subchannel[grobj->subc].seq = nvchan->subc_sequence++; -#endif - #ifdef NOUVEAU_DMA_TRACE NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, grobj->handle, grobj->subc, method, size); @@ -128,26 +122,6 @@ nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); } -static inline void -nouveau_dma_bind(struct nouveau_channel *chan, struct nouveau_grobj *grobj, - int subc) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - - if (nvchan->subchannel[subc].grobj == grobj) - return; - - if (nvchan->subchannel[subc].grobj) - nvchan->subchannel[subc].grobj->bound = NOUVEAU_GROBJ_UNBOUND; - nvchan->subchannel[subc].grobj = grobj; - grobj->subc = subc; - grobj->bound = NOUVEAU_GROBJ_EXPLICIT_BIND; - - nouveau_dma_begin(chan, grobj, 0x0000, 1, __FUNCTION__, __LINE__); - nouveau_dma_out (chan, grobj->handle); -} - -#define BIND_RING_CH(ch,gr,sc) nouveau_dma_bind((ch), (gr), (sc)) #define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) #define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) #define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 9978652dd6d..7e52de43235 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -161,12 +161,6 @@ struct nouveau_channel_priv { struct drm_nouveau_channel_alloc drm; - struct { - struct nouveau_grobj *grobj; - uint32_t seq; - } subchannel[8]; - uint32_t subc_sequence; - uint32_t *pushbuf; void *notifier_block; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c index 7861b6f84d3..c9f0903e272 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c @@ -108,8 +108,15 @@ nouveau_fence_emit(struct nouveau_fence *fence) if (nvfence->sequence == 0xffffffff) NOUVEAU_ERR("AII wrap unhandled\n"); - BEGIN_RING_CH(&nvchan->base, nvchan->subchannel[0].grobj, 0x50, 1); - OUT_RING_CH (&nvchan->base, nvfence->sequence); + /*XXX: assumes subc 0 is populated */ + if (nvchan->dma.free < 2) + WAIT_RING_CH(&nvchan->base, 2); + nvchan->dma.free -= 2; +#ifdef NOUVEAU_DMA_DEBUG + nvchan->dma.push_free += 2; +#endif + OUT_RING_CH(&nvchan->base, 0x00040050); + OUT_RING_CH(&nvchan->base, nvfence->sequence); if (nvchan->fence_tail) { nouveau_fence(nvchan->fence_tail)->next = fence; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index 1978edea02d..e7111de4dc1 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -20,7 +20,6 @@ //#define NOUVEAU_DMA_TRACE //#define NOUVEAU_DMA_DEBUG //#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF -#define NOUVEAU_DMA_SUBCHAN_LRU #define NOUVEAU_DMA_BARRIER #define NOUVEAU_DMA_TIMEOUT 2000 @@ -29,6 +28,10 @@ nv->pushbuf = nouveau_pipe_dma_beginp(nv->obj, (mthd), (size)); \ } while(0) +#define BEGIN_RING_GR(obj,mthd,size) do { \ + nv->pushbuf = nouveau_pipe_dma_beginp(obj, (mthd), (size)); \ +} while(0) + #define OUT_RING(data) do { \ (*nv->pushbuf++) = (data); \ } while(0) @@ -48,6 +51,12 @@ nouveau_pipe_dma_kickoff(nv->channel); \ } while(0) +#define BIND_RING(o,s) do { \ + nv->o->subc = (s); \ + BEGIN_RING(o, 0x0000, 1); \ + OUT_RING (nv->o->handle); \ +} while(0) + #define OUT_RELOC(bo,data,flags,vor,tor) do { \ nouveau_pushbuf_emit_reloc(nv->channel, nv->pushbuf, (void*)(bo), \ (data), (flags), (vor), (tor)); \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index e6481a20f2c..f588c956ca9 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -20,9 +20,18 @@ nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, struct nouveau_grobj **grobj) { struct nouveau_context *nv = nvws->nv; + int ret; - return nouveau_grobj_alloc(nv->channel, nv->next_handle++, - grclass, grobj); + ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, + grclass, grobj); + if (ret) + return ret; + + (*grobj)->subc = nv->next_subchannel++; + assert((*grobj)->subc <= 7); + BEGIN_RING_GR(*grobj, 0x0000, 1); + OUT_RING ((*grobj)->handle); + return 0; } uint32_t * @@ -34,10 +43,6 @@ nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) if (!nvchan->pb_tail || nvchan->pb_tail->remaining < (size + 1)) nouveau_pushbuf_flush(grobj->channel); - if (grobj->bound == NOUVEAU_GROBJ_UNBOUND) - nouveau_dma_subc_bind(grobj); - nvchan->subchannel[grobj->subc].seq = nvchan->subc_sequence++; - pushbuf = nvchan->pb_tail->cur; nvchan->pb_tail->cur += (size + 1); nvchan->pb_tail->remaining -= (size + 1); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c index 67e2514026b..b50e744d0d0 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c @@ -178,6 +178,7 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); return 1; } + BIND_RING (NvM2MF, nv->next_subchannel++); BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); @@ -188,6 +189,7 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); return 1; } + BIND_RING (NvCtxSurf2D, nv->next_subchannel++); BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RING (nv->channel->vram->handle); OUT_RING (nv->channel->vram->handle); @@ -199,6 +201,7 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) NOUVEAU_ERR("Error creating blit object: %d\n", ret); return 1; } + BIND_RING (NvImageBlit, nv->next_subchannel++); BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_SURFACE, 1); @@ -212,6 +215,7 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) NOUVEAU_ERR("Error creating rect object: %d\n", ret); return 1; } + BIND_RING (NvGdiRect, nv->next_subchannel++); BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c index 2a95b1fcd36..d00378d9045 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c @@ -151,7 +151,7 @@ nouveau_surface_init_nv50(struct nouveau_context *nv) &nv->Nv2D); if (ret) return ret; - + BIND_RING (Nv2D, 0); BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); diff --git a/src/mesa/pipe/nouveau/nouveau_grobj.h b/src/mesa/pipe/nouveau/nouveau_grobj.h index a36db7f3bc4..8f5abf90514 100644 --- a/src/mesa/pipe/nouveau/nouveau_grobj.h +++ b/src/mesa/pipe/nouveau/nouveau_grobj.h @@ -29,12 +29,6 @@ struct nouveau_grobj { struct nouveau_channel *channel; int grclass; uint32_t handle; - - enum { - NOUVEAU_GROBJ_UNBOUND = 0, - NOUVEAU_GROBJ_BOUND = 1, - NOUVEAU_GROBJ_EXPLICIT_BIND = 2, - } bound; int subc; }; -- cgit v1.2.3 From b4b002661e8a1f9a37201796c996503c33b8c96f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Dec 2007 16:05:02 +1100 Subject: nouveau: match gallium changes --- .../drivers/dri/nouveau_winsys/nouveau_screen.c | 5 +- src/mesa/pipe/nv40/nv40_fragprog.c | 5 +- src/mesa/pipe/nv40/nv40_vertprog.c | 65 +++++++++++++++++++++- 3 files changed, 69 insertions(+), 6 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index 6324a591311..784dfdd155e 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -141,8 +141,9 @@ nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, else stencil = PIPE_FORMAT_NONE; - nvfb->stfb = st_create_framebuffer(glVis, GL_TRUE, colour, depth, - stencil, (void*)nvfb); + nvfb->stfb = st_create_framebuffer(glVis, colour, depth, stencil, + driDrawPriv->w, driDrawPriv->h, + (void*)nvfb); if (!nvfb->stfb) { free(nvfb); return GL_FALSE; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index c95f31e1439..480c4c967b4 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -378,7 +378,7 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, int ai = -1, ci = -1; int i; - if (finst->Instruction.Opcode == TGSI_OPCODE_RET) + if (finst->Instruction.Opcode == TGSI_OPCODE_END) return TRUE; fpc->temp_temp_count = 0; @@ -537,6 +537,9 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, case TGSI_OPCODE_RCP: arith(fpc, sat, RCP, dst, mask, src[0], none, none); break; + case TGSI_OPCODE_RET: + assert(0); + break; case TGSI_OPCODE_RFL: tmp = temp(fpc); arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index 981d70ec92b..84618c1e1ba 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -47,9 +47,12 @@ struct nv40_vpc { int high_temp; int temp_temp_count; + + struct nv40_sreg *imm; + unsigned nr_imm; }; -static INLINE struct nv40_sreg +static struct nv40_sreg temp(struct nv40_vpc *vpc) { int idx; @@ -59,7 +62,7 @@ temp(struct nv40_vpc *vpc) return nv40_sr(NV40SR_TEMP, idx); } -static INLINE struct nv40_sreg +static struct nv40_sreg constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w) { struct nv40_vertex_program *vp = vpc->vp; @@ -251,6 +254,9 @@ tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { case TGSI_FILE_CONSTANT: src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); break; + case TGSI_FILE_IMMEDIATE: + src = vpc->imm[fsrc->SrcRegister.Index]; + break; case TGSI_FILE_TEMPORARY: if (vpc->high_temp < fsrc->SrcRegister.Index) vpc->high_temp = fsrc->SrcRegister.Index; @@ -315,7 +321,7 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, int ai = -1, ci = -1; int i; - if (finst->Instruction.Opcode == TGSI_OPCODE_RET) + if (finst->Instruction.Opcode == TGSI_OPCODE_END) return TRUE; vpc->temp_temp_count = 0; @@ -343,7 +349,11 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, tgsi_src(vpc, fsrc), none, none); } break; + /*XXX: index comparison is broken now that consts come from + * two different register files. + */ case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: if (ci == -1 || ci == fsrc->SrcRegister.Index) { ci = fsrc->SrcRegister.Index; src[i] = tgsi_src(vpc, fsrc); @@ -435,6 +445,8 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, case TGSI_OPCODE_RCP: arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); break; + case TGSI_OPCODE_RET: + break; case TGSI_OPCODE_RSQ: arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); break; @@ -518,6 +530,35 @@ nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, return TRUE; } +static boolean +nv40_vertprog_prepare(struct nv40_vpc *vpc) +{ + struct tgsi_parse_context p; + int nr_imm = 0; + + tgsi_parse_init(&p, vpc->vp->pipe->tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch(tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + nr_imm++; + break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (nr_imm) { + vpc->imm = calloc(nr_imm, sizeof(struct nv40_sreg)); + assert(vpc->imm); + } + + return TRUE; +} + void nv40_vertprog_translate(struct nv40_context *nv40, struct nv40_vertex_program *vp) @@ -531,6 +572,11 @@ nv40_vertprog_translate(struct nv40_context *nv40, vpc->vp = vp; vpc->high_temp = -1; + if (!nv40_vertprog_prepare(vpc)) { + free(vpc); + return; + } + tgsi_parse_init(&parse, vp->pipe->tokens); while (!tgsi_parse_end_of_tokens(&parse)) { @@ -552,6 +598,19 @@ nv40_vertprog_translate(struct nv40_context *nv40, } break; case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); +// assert(imm->Immediate.Size == 4); + vpc->imm[vpc->nr_imm++] = + constant(vpc, -1, + imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); + } break; case TGSI_TOKEN_TYPE_INSTRUCTION: { -- cgit v1.2.3 From fa605cf661c09c5866cd9aa316b6a5ce9eb65c24 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Dec 2007 16:06:24 +1100 Subject: nouveau: some cleanups --- .../drivers/dri/nouveau_winsys/nouveau_context.h | 13 ------ src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 5 +- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h | 30 +++++++----- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 8 +--- .../drivers/dri/nouveau_winsys/nouveau_fence.c | 11 ++--- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 53 ++++++++++++---------- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 12 ++--- src/mesa/pipe/nouveau/nouveau_channel.h | 2 + src/mesa/pipe/nouveau/nouveau_pushbuf.h | 32 +++++++++++++ 9 files changed, 95 insertions(+), 71 deletions(-) create mode 100644 src/mesa/pipe/nouveau/nouveau_pushbuf.h diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index 294b66e5188..f201001bd9b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -30,19 +30,6 @@ struct nouveau_context { struct nouveau_screen *nv_screen; struct pipe_surface *frontbuffer; - /* Bufmgr */ - struct { - struct nouveau_channel *channel; - struct nouveau_notifier *notify; - struct nouveau_grobj *m2mf; - uint32_t m2mf_src_ctxdma; - uint32_t m2mf_dst_ctxdma; - uint32_t next_sequence; - } bo; - - /* Relocations */ - struct nouveau_bo *reloc_head; - /* Hardware context */ uint32_t *pushbuf; struct nouveau_channel *channel; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index d00fddbff2f..3bb7c49762b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -59,12 +59,13 @@ nouveau_dma_channel_init(struct nouveau_channel *chan) int i; nvchan->dma.base = nvchan->drm.put_base; - nvchan->dma.cur = nvchan->dma.put = RING_SKIPS; + nvchan->dma.cur = nvchan->dma.put = 0; nvchan->dma.max = (nvchan->drm.cmdbuf_size >> 2) - 2; nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; + RING_SPACE_CH(chan, RING_SKIPS); for (i = 0; i < RING_SKIPS; i++) - nvchan->pushbuf[i] = 0x00000000; + OUT_RING_CH(chan, 0); } #define CHECK_TIMEOUT() do { \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h index c4667cc5c02..940a196bfe2 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h @@ -87,12 +87,27 @@ nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) #endif } +static inline void +nouveau_dma_space(struct nouveau_channel *chan, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + + if (nvchan->dma.free < size) { + if (nouveau_dma_wait(chan, size) && chan->hang_notify) + chan->hang_notify(chan); + } + nvchan->dma.free -= size; +#ifdef NOUVEAU_DMA_DEBUG + nvchan->dma.push_free = size; +#endif +} + static inline void nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, int method, int size, const char* file, int line) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - int push_size = size + 1; + (void)nvchan; #ifdef NOUVEAU_DMA_TRACE NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, @@ -108,20 +123,11 @@ nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, sprintf(faulty,"%s:%d",file,line); #endif - if (nvchan->dma.free < push_size) { - if (nouveau_dma_wait(chan, push_size) && - chan->hang_notify) { - chan->hang_notify(chan); - } - } - nvchan->dma.free -= push_size; -#ifdef NOUVEAU_DMA_DEBUG - nvchan->dma.push_free = push_size; -#endif - + nouveau_dma_space(chan, (size + 1)); nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); } +#define RING_SPACE_CH(ch,sz) nouveau_dma_space((ch), (sz)) #define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) #define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) #define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 7e52de43235..4dc40b7593d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -33,6 +33,7 @@ #include "pipe/nouveau/nouveau_notifier.h" #include "pipe/nouveau/nouveau_bo.h" #include "pipe/nouveau/nouveau_resource.h" +#include "pipe/nouveau/nouveau_pushbuf.h" struct nouveau_device_priv { struct nouveau_device base; @@ -121,12 +122,6 @@ struct nouveau_pushbuf_bo { int nr_relocs; }; -struct nouveau_pushbuf { - struct nouveau_channel *channel; - unsigned remaining; - uint32_t *cur; -}; - struct nouveau_pushbuf_priv { struct nouveau_pushbuf base; struct nouveau_pushbuf *next; @@ -138,6 +133,7 @@ struct nouveau_pushbuf_priv { int nr_buffers; }; #define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) + #define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o)) #define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h)) #define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c index c9f0903e272..3b35a11bcbf 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c @@ -109,14 +109,9 @@ nouveau_fence_emit(struct nouveau_fence *fence) NOUVEAU_ERR("AII wrap unhandled\n"); /*XXX: assumes subc 0 is populated */ - if (nvchan->dma.free < 2) - WAIT_RING_CH(&nvchan->base, 2); - nvchan->dma.free -= 2; -#ifdef NOUVEAU_DMA_DEBUG - nvchan->dma.push_free += 2; -#endif - OUT_RING_CH(&nvchan->base, 0x00040050); - OUT_RING_CH(&nvchan->base, nvfence->sequence); + RING_SPACE_CH(fence->channel, 2); + OUT_RING_CH (fence->channel, 0x00040050); + OUT_RING_CH (fence->channel, nvfence->sequence); if (nvchan->fence_tail) { nouveau_fence(nvchan->fence_tail)->next = fence; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 981c4dd7dd7..2179664838e 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -29,6 +29,30 @@ #define PB_RSVD_DWORDS 2 +static int +nouveau_pushbuf_space(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb; + + nvpb = calloc(1, sizeof(struct nouveau_pushbuf_priv)); + if (!nvpb) + return -ENOMEM; + + while (nouveau_resource_alloc(nvchan->pb_heap, 0x2100, NULL, + &nvpb->res)) { + nouveau_fence_flush(chan); + } + + nvpb->base.channel = chan; + nvpb->base.remaining = (nvpb->res->size / 4) - PB_RSVD_DWORDS; + nvpb->base.cur = &nvchan->pushbuf[nvpb->res->start/4]; + nvchan->pb_tail = &nvpb->base; + nvchan->base.pushbuf = nvchan->pb_tail; + + return 0; +} + int nouveau_pushbuf_init(struct nouveau_channel *chan) { @@ -47,6 +71,8 @@ nouveau_pushbuf_init(struct nouveau_channel *chan) nvchan->dma.max = (4096 / 4) - 2; nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; + assert(!nouveau_pushbuf_space(chan)); + return 0; } @@ -65,14 +91,11 @@ int nouveau_pushbuf_flush(struct nouveau_channel *chan) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); struct nouveau_pushbuf_bo *pbbo; struct nouveau_fence *fence = NULL; int ret; - if (!nvpb) - goto out_realloc; - if (nvpb->base.remaining == (nvpb->res->size / 4) - PB_RSVD_DWORDS) return 0; nvchan->pb_tail = NULL; @@ -119,12 +142,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) nvpb->nr_buffers = 0; /* Emit JMP to indirect pushbuf */ - if (nvchan->dma.free < 1) - WAIT_RING_CH(chan, 1); - nvchan->dma.free -= 1; -#ifdef NOUVEAU_DMA_DEBUG - nvchan->dma.push_free = 1; -#endif + RING_SPACE_CH(chan, 1); OUT_RING_CH(chan, 0x20000000 | nvpb->res->start); /* Add JMP back to master pushbuf from indirect pushbuf */ @@ -141,20 +159,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) FIRE_RING_CH(chan); /* Allocate space for next push buffer */ -out_realloc: - nvpb = calloc(1, sizeof(struct nouveau_pushbuf_priv)); - if (!nvpb) - return -ENOMEM; - - while (nouveau_resource_alloc(nvchan->pb_heap, 0x2100, NULL, - &nvpb->res)) { - nouveau_fence_flush(chan); - } - - nvpb->base.channel = chan; - nvpb->base.remaining = (nvpb->res->size / 4) - PB_RSVD_DWORDS; - nvpb->base.cur = &nvchan->pushbuf[nvpb->res->start/4]; - nvchan->pb_tail = &nvpb->base; + assert(!nouveau_pushbuf_space(chan)); return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index f588c956ca9..b31e0dcd1dd 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -37,15 +37,15 @@ nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, uint32_t * nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) { - struct nouveau_channel_priv *nvchan = nouveau_channel(grobj->channel); + struct nouveau_channel *chan = grobj->channel; uint32_t *pushbuf; - if (!nvchan->pb_tail || nvchan->pb_tail->remaining < (size + 1)) - nouveau_pushbuf_flush(grobj->channel); + if (chan->pushbuf->remaining < (size + 1)) + nouveau_pushbuf_flush(chan); - pushbuf = nvchan->pb_tail->cur; - nvchan->pb_tail->cur += (size + 1); - nvchan->pb_tail->remaining -= (size + 1); + pushbuf = chan->pushbuf->cur; + chan->pushbuf->cur += (size + 1); + chan->pushbuf->remaining -= (size + 1); (*pushbuf++) = ((grobj->subc << 13) | (size << 18) | mthd); return pushbuf; diff --git a/src/mesa/pipe/nouveau/nouveau_channel.h b/src/mesa/pipe/nouveau/nouveau_channel.h index eccc7b3bd4d..b99de9add86 100644 --- a/src/mesa/pipe/nouveau/nouveau_channel.h +++ b/src/mesa/pipe/nouveau/nouveau_channel.h @@ -27,6 +27,8 @@ struct nouveau_channel { struct nouveau_device *device; int id; + struct nouveau_pushbuf *pushbuf; + struct nouveau_grobj *vram; struct nouveau_grobj *gart; diff --git a/src/mesa/pipe/nouveau/nouveau_pushbuf.h b/src/mesa/pipe/nouveau/nouveau_pushbuf.h new file mode 100644 index 00000000000..19097650982 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_pushbuf.h @@ -0,0 +1,32 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_PUSHBUF_H__ +#define __NOUVEAU_PUSHBUF_H__ + +struct nouveau_pushbuf { + struct nouveau_channel *channel; + unsigned remaining; + uint32_t *cur; +}; + +#endif -- cgit v1.2.3 From 74757eb970984b0404c17255f3639d8a40591964 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Dec 2007 16:18:00 +1100 Subject: nouveau: don't wait when deleting fences unless needed --- .../drivers/dri/nouveau_winsys/nouveau_fence.c | 41 ++++++++++++++++++---- 1 file changed, 35 insertions(+), 6 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c index 3b35a11bcbf..53037d16823 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c @@ -22,6 +22,7 @@ #include #include +#include #include "nouveau_drmif.h" #include "nouveau_dma.h" @@ -59,6 +60,28 @@ nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) return 0; } +static void +nouveau_fence_del_unsignalled(struct nouveau_fence *fence) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); + struct nouveau_fence *le; + + if (nvchan->fence_head == fence) { + nvchan->fence_head = nouveau_fence(fence)->next; + if (nvchan->fence_head == NULL) + nvchan->fence_tail = NULL; + return; + } + + le = nvchan->fence_head; + while (le && nouveau_fence(le)->next != fence) + le = nouveau_fence(le)->next; + assert(le && nouveau_fence(le)->next == fence); + nouveau_fence(le)->next = nouveau_fence(fence)->next; + if (nvchan->fence_tail == fence) + nvchan->fence_tail = le; +} + void nouveau_fence_del(struct nouveau_fence **fence) { @@ -69,11 +92,19 @@ nouveau_fence_del(struct nouveau_fence **fence) nvfence = nouveau_fence(*fence); *fence = NULL; - if (--nvfence->refcount <= 0) { - if (nvfence->emitted && !nvfence->signalled) + if (--nvfence->refcount) + return; + + if (nvfence->emitted && !nvfence->signalled) { + if (nvfence->signal_cb) { + nvfence->refcount++; nouveau_fence_wait((void *)&nvfence); - free(nvfence); + return; + } + + nouveau_fence_del_unsignalled(&nvfence->base); } + free(nvfence); } int @@ -139,9 +170,7 @@ nouveau_fence_flush(struct nouveau_channel *chan) break; } - nvchan->fence_head = nvfence->next; - if (nvchan->fence_head == NULL) - nvchan->fence_tail = NULL; + nouveau_fence_del_unsignalled(&nvfence->base); nvfence->signalled = 1; while (nvfence->signal_cb) { -- cgit v1.2.3 From 7372a596a9b960b212c839ef4ee1a1358b224047 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Dec 2007 16:19:25 +1100 Subject: nv40: seems we have stencil faces around the wrong way. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 3 +++ src/mesa/pipe/nv40/nv40_state.c | 3 ++- 2 files changed, 5 insertions(+), 1 deletion(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 46df588b5ba..a86b7265440 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -222,6 +222,9 @@ nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) if (!nvbo) return -EINVAL; + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + if (nvbo->sysmem) bo->map = nvbo->sysmem; else diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 12b2c2a8e8f..8fa40ce8ef7 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -425,9 +425,10 @@ nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); OUT_RINGp ((uint32_t *)&hw->depth, 3); + /*XXX: fix in nouveau_class.h, faces backwards.. */ BEGIN_RING(curie, NV40TCL_STENCIL_BACK_ENABLE, 16); - OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); + OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); } -- cgit v1.2.3 From 5fcffcd31279d7777ad543809e5115f1e578022c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Dec 2007 17:00:08 +1100 Subject: nouveau: speed up user buffers. Try and fit user buffers into a small GART scratch area at validate time, instead of going to a lot of effort to fit these (mostly) use-once-and-discard objects into VRAM. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 205 +++++++++++++++++---- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 6 + .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 1 + 3 files changed, 173 insertions(+), 39 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index a86b7265440..09e66d1bdcd 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -23,41 +23,69 @@ #include #include #include +#include #include "nouveau_drmif.h" #include "nouveau_dma.h" #include "nouveau_local.h" -int -nouveau_bo_init(struct nouveau_device *dev) +static void +nouveau_mem_free(struct nouveau_device *dev, struct drm_nouveau_mem_alloc *ma, + void **map) { - return 0; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_mem_free mf; + + if (map && *map) { + drmUnmap(*map, ma->size); + *map = NULL; + } + + if (ma->size) { + mf.offset = ma->offset; + mf.flags = ma->flags; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, + &mf, sizeof(mf)); + ma->size = 0; + } } -void -nouveau_bo_takedown(struct nouveau_device *dev) +static int +nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, + uint32_t flags, struct drm_nouveau_mem_alloc *ma, void **map) { + struct nouveau_device_priv *nvdev = nouveau_device(dev); + int ret; + + ma->alignment = align; + ma->size = size; + ma->flags = flags; + if (map) + ma->flags |= NOUVEAU_MEM_MAPPED; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, ma, + sizeof(struct drm_nouveau_mem_alloc)); + if (ret) + return ret; + + if (map) { + ret = drmMap(nvdev->fd, ma->map_handle, ma->size, map); + if (ret) { + *map = NULL; + nouveau_mem_free(dev, ma, map); + return ret; + } + } + + return 0; } static int nouveau_bo_realloc_gpu(struct nouveau_bo_priv *nvbo, uint32_t flags, int size) { - struct nouveau_device_priv *nvdev = nouveau_device(nvbo->base.device); int ret; if (nvbo->drm.size && nvbo->drm.size != size) { - struct drm_nouveau_mem_free f; - - if (nvbo->map) { - drmUnmap(nvbo->map, nvbo->drm.size); - nvbo->map = NULL; - } - - f.flags = nvbo->drm.flags; - f.offset = nvbo->drm.offset; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, &f, sizeof(f)); - - nvbo->drm.size = 0; + nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); } if (size && !nvbo->drm.size) { @@ -71,27 +99,76 @@ nouveau_bo_realloc_gpu(struct nouveau_bo_priv *nvbo, uint32_t flags, int size) nvbo->drm.flags |= NOUVEAU_MEM_MAPPED; } - nvbo->drm.size = size; - - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, - &nvbo->drm, sizeof(nvbo->drm)); + ret = nouveau_mem_alloc(nvbo->base.device, size, + nvbo->drm.alignment, nvbo->drm.flags, + &nvbo->drm, &nvbo->map); if (ret) { - free(nvbo); - return ret; + assert(0); } + } - ret = drmMap(nvdev->fd, nvbo->drm.map_handle, nvbo->drm.size, - &nvbo->map); - if (ret) { - nvbo->map = NULL; - nouveau_bo_del((void *)&nvbo); - return ret; - } + return 0; +} + +static void +nouveau_bo_tmp_del(void *priv) +{ + struct nouveau_resource *r = priv; + + nouveau_fence_del((struct nouveau_fence **)&r->priv); + nouveau_resource_free(&r); +} + +static struct nouveau_resource * +nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, + struct nouveau_fence *fence) +{ + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_resource *r = NULL; + struct nouveau_fence *ref = NULL; + + if (fence) + nouveau_fence_ref(fence, &ref); + else + nouveau_fence_new(chan, &ref); + assert(ref); + + while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) { + nouveau_fence_flush(chan); + } + nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r); + + return r; +} + +int +nouveau_bo_init(struct nouveau_device *dev) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + int ret; + + ret = nouveau_mem_alloc(dev, 128*1024, 0, NOUVEAU_MEM_AGP | + NOUVEAU_MEM_PCI, &nvdev->sa, &nvdev->sa_map); + if (ret) + return ret; + + ret = nouveau_resource_init(&nvdev->sa_heap, 0, nvdev->sa.size); + if (ret) { + nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); + return ret; } return 0; } +void +nouveau_bo_takedown(struct nouveau_device *dev) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); +} + int nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, int size, struct nouveau_bo **bo) @@ -247,9 +324,36 @@ nouveau_bo_upload(struct nouveau_bo_priv *nvbo) return 0; } -int -nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) +static int +nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_resource *r; + + if (nvchan->user_charge + bo->size > nvdev->sa.size) + return 1; + nvchan->user_charge += bo->size; + + if (!(flags & NOUVEAU_BO_GART)) + return 1; + + r = nouveau_bo_tmp(chan, bo->size, fence); + if (!r) + return 1; + + memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); + + nvbo->base.offset = nvdev->sa.offset + r->start; + nvbo->base.flags = NOUVEAU_BO_GART; + return 0; +} + +static int +nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) { struct nouveau_bo_priv *nvbo = nouveau_bo(bo); @@ -264,13 +368,9 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, if (nvbo->user) { nouveau_bo_upload(nvbo); } - - if (nvbo->fence) - nouveau_fence_del(&nvbo->fence); - nouveau_fence_ref(fence, &nvbo->fence); nvbo->base.offset = nvbo->drm.offset; - if (nvbo->drm.flags & NOUVEAU_MEM_AGP) + if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) nvbo->base.flags = NOUVEAU_BO_GART; else nvbo->base.flags = NOUVEAU_BO_VRAM; @@ -278,3 +378,30 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, return 0; } +int +nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + int ret; + + if (nvbo->user) { + ret = nouveau_bo_validate_user(chan, bo, fence, flags); + if (ret) { + ret = nouveau_bo_validate_bo(chan, bo, fence, flags); + if (ret) + return ret; + } + } else { + ret = nouveau_bo_validate_bo(chan, bo, fence, flags); + if (ret) + return ret; + } + + if (nvbo->fence) + nouveau_fence_del(&nvbo->fence); + nouveau_fence_ref(fence, &nvbo->fence); + + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 4dc40b7593d..11a15d632ce 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -42,6 +42,10 @@ struct nouveau_device_priv { drm_context_t ctx; drmLock *lock; int needs_close; + + struct drm_nouveau_mem_alloc sa; + void *sa_map; + struct nouveau_resource *sa_heap; }; #define nouveau_device(n) ((struct nouveau_device_priv *)(n)) @@ -180,6 +184,8 @@ struct nouveau_channel_priv { struct nouveau_resource *pb_heap; struct nouveau_pushbuf *pb_head; struct nouveau_pushbuf *pb_tail; + + unsigned user_charge; }; #define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 2179664838e..11754417893 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -105,6 +105,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) return ret; /* Validate buffers + apply relocations */ + nvchan->user_charge = 0; while ((pbbo = ptr_to_pbbo(nvpb->buffers))) { struct nouveau_pushbuf_reloc *r; struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; -- cgit v1.2.3 From f9cfc323768b6926dcc731484cc1ac4ded000e5a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 24 Dec 2007 18:53:41 +1100 Subject: nv40: ensure all required buffers are accounted for during state validation --- src/mesa/pipe/nv40/nv40_context.h | 15 +++++++++ src/mesa/pipe/nv40/nv40_fragprog.c | 4 --- src/mesa/pipe/nv40/nv40_state.c | 30 +++++------------ src/mesa/pipe/nv40/nv40_state_emit.c | 65 ++++++++++++++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_state_tex.c | 11 +++--- src/mesa/pipe/nv40/nv40_vbo.c | 37 +++++++++++++++----- 6 files changed, 121 insertions(+), 41 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 9a93ed35ace..5d0bf89680d 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -44,6 +44,21 @@ struct nv40_context { struct pipe_texture *tex_miptree[PIPE_MAX_SAMPLERS]; uint32_t tex_dirty; + uint32_t rt_enable; + struct pipe_buffer_handle *rt[4]; + struct pipe_buffer_handle *zeta; + + struct { + struct pipe_buffer_handle *buffer; + uint32_t format; + } tex[16]; + + unsigned vb_enable; + struct { + struct pipe_buffer_handle *buffer; + unsigned delta; + } vb[16]; + struct { struct nouveau_resource *exec_heap; struct nouveau_resource *data_heap; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 480c4c967b4..6ee6a7daff7 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -796,10 +796,6 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->on_hw = TRUE; } - BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, - NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); OUT_RING (fp->fp_control); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 8fa40ce8ef7..30cd4290750 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -630,50 +630,36 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(rt[0]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); + BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 1); OUT_RING (rt[0]->pitch * rt[0]->cpp); - OUT_RELOCl(rt[0]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + nv40->rt[0] = rt[0]->buffer; } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(rt[1]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); - OUT_RELOCl(rt[1]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR1_PITCH, 2); OUT_RING (rt[1]->pitch * rt[1]->cpp); + nv40->rt[1] = rt[1]->buffer; } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(rt[2]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(rt[2]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); OUT_RING (rt[2]->pitch * rt[2]->cpp); + nv40->rt[2] = rt[2]->buffer; } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(rt[3]->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(rt[3]->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); OUT_RING (rt[3]->pitch * rt[3]->cpp); + nv40->rt[3] = rt[3]->buffer; } if (zeta_format) { - BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(zeta->buffer, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | NOUVEAU_BO_RD); - BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(zeta->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR | - NOUVEAU_BO_RD); BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); OUT_RING (zeta->pitch * zeta->cpp); + nv40->zeta = zeta->buffer; } + nv40->rt_enable = rt_enable; BEGIN_RING(curie, NV40TCL_RT_ENABLE, 1); OUT_RING (rt_enable); BEGIN_RING(curie, NV40TCL_RT_HORIZ, 3); diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index c53b5d852ae..fc8a0a952af 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -5,6 +5,8 @@ void nv40_emit_hw_state(struct nv40_context *nv40) { + int i; + if (nv40->dirty & NV40_NEW_FRAGPROG) { nv40_fragprog_bind(nv40, nv40->fragprog.current); /*XXX: clear NV40_NEW_FRAGPROG if no now program uploaded */ @@ -25,5 +27,68 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40_vertprog_bind(nv40, nv40->vertprog.current); nv40->dirty &= ~NV40_NEW_VERTPROG; } + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv40_vbo.c + */ + + /* Render targets */ + if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR0) { + BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR0_OFFSET, 1); + OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR1) { + BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 1); + OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR2) { + BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); + OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); + OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR3) { + BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); + OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); + OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv40->zeta) { + BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); + OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + /* Texture images */ + for (i = 0; i < 16; i++) { + if (!nv40->tex[i].buffer) + continue; + BEGIN_RING(curie, NV40TCL_TEX_OFFSET(i), 2); + OUT_RELOCl(nv40->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv40->tex[i].buffer, nv40->tex[i].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, + NV40TCL_TEX_FORMAT_DMA1); + } + + /* Fragment program */ + BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); + OUT_RELOC (nv40->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, + NV40TCL_FP_ADDRESS_DMA1); } diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 05c136fcc19..9fb274d6274 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -109,12 +109,10 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) if (pt->format == PIPE_FORMAT_U_A8_L8) txs |= (1<<16); /*nfi*/ - BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); - OUT_RELOCl(nv40mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD); - OUT_RELOCd(nv40mt->buffer, txf, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, NV40TCL_TEX_FORMAT_DMA0, - NV40TCL_TEX_FORMAT_DMA1); + nv40->tex[unit].buffer = nv40mt->buffer; + nv40->tex[unit].format = txf; + + BEGIN_RING(curie, NV40TCL_TEX_WRAP(unit), 6); OUT_RING (ps->wrap); OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | (0x00078000) /* mipmap related? */); @@ -135,6 +133,7 @@ nv40_state_tex_update(struct nv40_context *nv40) if (nv40->tex_miptree[unit]) { nv40_tex_unit_enable(nv40, unit); } else { + nv40->tex[unit].buffer = NULL; BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); OUT_RING (0); } diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 4bd6c44e78b..fda32cdb492 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -6,6 +6,9 @@ #include "nv40_dma.h" #include "nv40_state.h" +#include "pipe/nouveau/nouveau_channel.h" +#include "pipe/nouveau/nouveau_pushbuf.h" + static INLINE int nv40_vbo_ncomp(uint format) { @@ -101,6 +104,8 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) uint32_t inputs, vtxfmt[16]; int hw, num_hw; + nv40->vb_enable = 0; + inputs = vp->ir; for (hw = 0; hw < 16 && inputs; hw++) { if (inputs & (1 << hw)) { @@ -129,19 +134,16 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) continue; } - BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(hw), 1); - OUT_RELOC(vb->buffer, vb->buffer_offset + ve->src_offset, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV40TCL_VTXBUF_ADDRESS_DMA1); + nv40->vb_enable |= (1 << hw); + nv40->vb[hw].delta = vb->buffer_offset + ve->src_offset; + nv40->vb[hw].buffer = vb->buffer; + vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | (nv40_vbo_ncomp(ve->src_format) << NV40TCL_VTXFMT_SIZE_SHIFT) | nv40_vbo_type(ve->src_format)); } - BEGIN_RING(curie, 0x1710, 1); - OUT_RING (0); /* vtx cache flush */ BEGIN_RING(curie, NV40TCL_VTXFMT(0), num_hw); OUT_RINGp (vtxfmt, num_hw); } @@ -149,14 +151,31 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) static boolean nv40_vbo_validate_state(struct nv40_context *nv40) { - if (nv40->dirty & ~NV40_NEW_ARRAYS) - nv40_emit_hw_state(nv40); + unsigned inputs; + + nv40_emit_hw_state(nv40); if (nv40->dirty & NV40_NEW_ARRAYS) { nv40_vbo_arrays_update(nv40); nv40->dirty &= ~NV40_NEW_ARRAYS; } + inputs = nv40->vb_enable; + while (inputs) { + unsigned a = ffs(inputs) - 1; + + inputs &= ~(1 << a); + + BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(a), 1); + OUT_RELOC (nv40->vb[a].buffer, nv40->vb[a].delta, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, + NV40TCL_VTXBUF_ADDRESS_DMA1); + } + + BEGIN_RING(curie, 0x1710, 1); + OUT_RING (0); /* vtx cache flush */ + return TRUE; } -- cgit v1.2.3 From d732728590e93de54dd4f4576b394ca2442c3db8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 24 Dec 2007 19:21:17 +1100 Subject: nouveau: pushbuf code, now with 50% less suck! Far more efficient, if not a bit more complicated. Hopefully not too buggy still. This commit will potentially expose some unrelated bugs, fixes for them will follow "real soon now". --- src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 77 +++++++------ src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h | 41 ++++--- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 32 +++--- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 121 +++++++++++---------- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 7 +- 5 files changed, 155 insertions(+), 123 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c index 3bb7c49762b..f8a8ba04f6d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c @@ -28,23 +28,23 @@ #include "nouveau_dma.h" #include "nouveau_local.h" -static __inline__ uint32_t +static inline uint32_t READ_GET(struct nouveau_channel_priv *nvchan) { - return ((*nvchan->get - nvchan->dma.base) >> 2); + return *nvchan->get; } -static __inline__ void +static inline void WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) { - uint32_t put = ((val << 2) + nvchan->dma.base); + uint32_t put = ((val << 2) + nvchan->dma->base); volatile int dum; NOUVEAU_DMA_BARRIER; dum = READ_GET(nvchan); *nvchan->put = put; - nvchan->dma.put = val; + nvchan->dma->put = val; #ifdef NOUVEAU_DMA_TRACE NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put); #endif @@ -52,16 +52,30 @@ WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) NOUVEAU_DMA_BARRIER; } +static inline int +LOCAL_GET(struct nouveau_dma_priv *dma, uint32_t *val) +{ + uint32_t get = *val; + + if (get >= dma->base && get <= (dma->base + (dma->max << 2))) { + *val = (get - dma->base) >> 2; + return 1; + } + + return 0; +} + void nouveau_dma_channel_init(struct nouveau_channel *chan) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); int i; - nvchan->dma.base = nvchan->drm.put_base; - nvchan->dma.cur = nvchan->dma.put = 0; - nvchan->dma.max = (nvchan->drm.cmdbuf_size >> 2) - 2; - nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; + nvchan->dma = &nvchan->dma_master; + nvchan->dma->base = nvchan->drm.put_base; + nvchan->dma->cur = nvchan->dma->put = 0; + nvchan->dma->max = (nvchan->drm.cmdbuf_size >> 2) - 2; + nvchan->dma->free = nvchan->dma->max - nvchan->dma->cur; RING_SPACE_CH(chan, RING_SKIPS); for (i = 0; i < RING_SKIPS; i++) @@ -73,54 +87,51 @@ nouveau_dma_channel_init(struct nouveau_channel *chan) return - EBUSY; \ } while(0) -#define IN_MASTER_RING(chan, ptr) ((ptr) <= (chan)->dma.max) - int nouveau_dma_wait(struct nouveau_channel *chan, int size) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; uint32_t get, t_start; FIRE_RING_CH(chan); t_start = NOUVEAU_TIME_MSEC(); - while (nvchan->dma.free < size) { + while (dma->free < size) { CHECK_TIMEOUT(); get = READ_GET(nvchan); - if (!IN_MASTER_RING(nvchan, get)) + if (!LOCAL_GET(dma, &get)) continue; - if (nvchan->dma.put >= get) { - nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; + if (dma->put >= get) { + dma->free = dma->max - dma->cur; - if (nvchan->dma.free < size) { + if (dma->free < size) { #ifdef NOUVEAU_DMA_DEBUG - nvchan->dma.push_free = 1; + dma->push_free = 1; #endif - OUT_RING_CH(chan, - 0x20000000 | nvchan->dma.base); + OUT_RING_CH(chan, 0x20000000 | dma->base); if (get <= RING_SKIPS) { /*corner case - will be idle*/ - if (nvchan->dma.put <= RING_SKIPS) + if (dma->put <= RING_SKIPS) WRITE_PUT(nvchan, RING_SKIPS + 1); do { CHECK_TIMEOUT(); get = READ_GET(nvchan); - if (!IN_MASTER_RING(nvchan, - get)) - continue; + if (!LOCAL_GET(dma, &get)) + get = 0; } while (get <= RING_SKIPS); } WRITE_PUT(nvchan, RING_SKIPS); - nvchan->dma.cur = nvchan->dma.put = RING_SKIPS; - nvchan->dma.free = get - (RING_SKIPS + 1); + dma->cur = dma->put = RING_SKIPS; + dma->free = get - (RING_SKIPS + 1); } } else { - nvchan->dma.free = get - nvchan->dma.cur - 1; + dma->free = get - dma->cur - 1; } } @@ -135,7 +146,7 @@ nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) unsigned mthd_count = 0; while (get != put) { - uint32_t gpuget = (get << 2) + nvchan->dma.base; + uint32_t gpuget = (get << 2) + nvchan->drm.put_base; uint32_t data; if (get < 0 || get >= nvchan->drm.cmdbuf_size) { @@ -188,21 +199,21 @@ void nouveau_dma_kickoff(struct nouveau_channel *chan) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; - if (nvchan->dma.cur == nvchan->dma.put) + if (dma->cur == dma->put) return; #ifdef NOUVEAU_DMA_DEBUG - if (nvchan->dma.push_free) { - NOUVEAU_ERR("Packet incomplete: %d left\n", - nvchan->dma.push_free); + if (dma->push_free) { + NOUVEAU_ERR("Packet incomplete: %d left\n", dma->push_free); return; } #endif #ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF - nouveau_dma_parse_pushbuf(chan, nvchan->dma.put, nvchan->dma.cur); + nouveau_dma_parse_pushbuf(chan, dma->put, dma->cur); #endif - WRITE_PUT(nvchan, nvchan->dma.cur); + WRITE_PUT(nvchan, dma->cur); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h index 940a196bfe2..cfa6d26e828 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h @@ -42,34 +42,37 @@ static inline void nouveau_dma_out(struct nouveau_channel *chan, uint32_t data) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; #ifdef NOUVEAU_DMA_DEBUG - if (nvchan->dma.push_free == 0) { - NOUVEAU_ERR("No space left in packet. Error at %s\n",faulty); + if (dma->push_free == 0) { + NOUVEAU_ERR("No space left in packet at %s\n", faulty); return; } - nvchan->dma.push_free--; + dma->push_free--; #endif #ifdef NOUVEAU_DMA_TRACE { - uint32_t offset = (nvchan->dma.cur << 2) + nvchan->dma.base; + uint32_t offset = (dma->cur << 2) + dma->base; NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", nvchan->drm.channel, offset, data); } #endif - nvchan->pushbuf[nvchan->dma.cur++] = data; + nvchan->pushbuf[dma->cur + (dma->base - nvchan->drm.put_base)/4] = data; + dma->cur++; } static inline void nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - (void)chan; + struct nouveau_dma_priv *dma = nvchan->dma; + (void)dma; #ifdef NOUVEAU_DMA_DEBUG - if (nvchan->dma.push_free < size) { + if (dma->push_free < size) { NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", - nvchan->dma.push_free, size); + dma->push_free, size); return; } #endif @@ -79,11 +82,11 @@ nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) ptr++; } #else - memcpy(&nvchan->pushbuf[nvchan->dma.cur], ptr, size << 2); + memcpy(&nvchan->pushbuf[dma->cur], ptr, size << 2); #ifdef NOUVEAU_DMA_DEBUG - nvchan->dma.push_free -= size; + dma->push_free -= size; #endif - nvchan->dma.cur += size; + dma->cur += size; #endif } @@ -91,14 +94,15 @@ static inline void nouveau_dma_space(struct nouveau_channel *chan, int size) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; - if (nvchan->dma.free < size) { + if (dma->free < size) { if (nouveau_dma_wait(chan, size) && chan->hang_notify) chan->hang_notify(chan); } - nvchan->dma.free -= size; + dma->free -= size; #ifdef NOUVEAU_DMA_DEBUG - nvchan->dma.push_free = size; + dma->push_free = size; #endif } @@ -107,7 +111,8 @@ nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, int method, int size, const char* file, int line) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - (void)nvchan; + struct nouveau_dma_priv *dma = nvchan->dma; + (void)dma; #ifdef NOUVEAU_DMA_TRACE NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, @@ -115,9 +120,9 @@ nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, #endif #ifdef NOUVEAU_DMA_DEBUG - if (nvchan->dma.push_free) { - NOUVEAU_ERR("Previous packet incomplete: %d left. Error at %s\n", - nvchan->dma.push_free,faulty); + if (dma->push_free) { + NOUVEAU_ERR("Previous packet incomplete: %d left at %s\n", + dma->push_free, faulty); return; } sprintf(faulty,"%s:%d",file,line); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 11a15d632ce..724677961e6 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -128,10 +128,10 @@ struct nouveau_pushbuf_bo { struct nouveau_pushbuf_priv { struct nouveau_pushbuf base; - struct nouveau_pushbuf *next; - struct nouveau_resource *res; - struct nouveau_fence *fence; + unsigned nop_jump; + unsigned start; + unsigned size; uint64_t buffers; int nr_buffers; @@ -149,13 +149,23 @@ extern int nouveau_pushbuf_init(struct nouveau_channel *); extern int -nouveau_pushbuf_flush(struct nouveau_channel *); +nouveau_pushbuf_flush(struct nouveau_channel *, unsigned min); extern int nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr, struct nouveau_bo *, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor); +struct nouveau_dma_priv { + uint32_t base; + uint32_t max; + uint32_t cur; + uint32_t put; + uint32_t free; + + int push_free; +} dma; + struct nouveau_channel_priv { struct nouveau_channel base; @@ -169,21 +179,15 @@ struct nouveau_channel_priv { volatile uint32_t *get; volatile uint32_t *ref_cnt; - struct { - uint32_t base, max; - uint32_t cur, put; - uint32_t free; - - int push_free; - } dma; + struct nouveau_dma_priv dma_master; + struct nouveau_dma_priv dma_bufmgr; + struct nouveau_dma_priv *dma; struct nouveau_fence *fence_head; struct nouveau_fence *fence_tail; uint32_t fence_sequence; - struct nouveau_resource *pb_heap; - struct nouveau_pushbuf *pb_head; - struct nouveau_pushbuf *pb_tail; + struct nouveau_pushbuf_priv pb; unsigned user_charge; }; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 11754417893..148406b85a7 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -27,28 +27,34 @@ #include "nouveau_drmif.h" #include "nouveau_dma.h" -#define PB_RSVD_DWORDS 2 +#define PB_BUFMGR_DWORDS (4096 / 2) +#define PB_MIN_USER_DWORDS 2048 static int -nouveau_pushbuf_space(struct nouveau_channel *chan) +nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb; + struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; - nvpb = calloc(1, sizeof(struct nouveau_pushbuf_priv)); - if (!nvpb) - return -ENOMEM; + assert((min + 1) <= nvchan->dma->max); - while (nouveau_resource_alloc(nvchan->pb_heap, 0x2100, NULL, - &nvpb->res)) { - nouveau_fence_flush(chan); - } + /* Wait for enough space in push buffer */ + min = min < PB_MIN_USER_DWORDS ? PB_MIN_USER_DWORDS : min; + min += 1; /* a bit extra for the NOP */ + if (nvchan->dma->free < min) + WAIT_RING_CH(chan, min); + + /* Insert NOP, may turn into a jump later */ + RING_SPACE_CH(chan, 1); + nvpb->nop_jump = nvchan->dma->cur; + OUT_RING_CH(chan, 0); + /* Any remaining space is available to the user */ + nvpb->start = nvchan->dma->cur; + nvpb->size = nvchan->dma->free; nvpb->base.channel = chan; - nvpb->base.remaining = (nvpb->res->size / 4) - PB_RSVD_DWORDS; - nvpb->base.cur = &nvchan->pushbuf[nvpb->res->start/4]; - nvchan->pb_tail = &nvpb->base; - nvchan->base.pushbuf = nvchan->pb_tail; + nvpb->base.remaining = nvpb->size; + nvpb->base.cur = &nvchan->pushbuf[nvpb->start]; return 0; } @@ -57,53 +63,65 @@ int nouveau_pushbuf_init(struct nouveau_channel *chan) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *m = &nvchan->dma_master; + struct nouveau_dma_priv *b = &nvchan->dma_bufmgr; + int i; if (!nvchan) return -EINVAL; - /* Everything except first 4KiB of the push buffer is managed by us */ - if (nouveau_resource_init(&nvchan->pb_heap, 4096, - nvchan->drm.cmdbuf_size - 4096)) - return -EINVAL; - - /* Shrink master ring to 4KiB */ - assert(nvchan->dma.cur <= (4096/4)); - nvchan->dma.max = (4096 / 4) - 2; - nvchan->dma.free = nvchan->dma.max - nvchan->dma.cur; - - assert(!nouveau_pushbuf_space(chan)); + /* Reassign last bit of push buffer for a "separate" bufmgr + * ring buffer + */ + m->max -= PB_BUFMGR_DWORDS; + m->free -= PB_BUFMGR_DWORDS; + + b->base = m->base + ((m->max + 2) << 2); + b->max = PB_BUFMGR_DWORDS - 2; + b->cur = b->put = 0; + b->free = b->max - b->cur; + + /* Some NOPs just to be safe + *XXX: RING_SKIPS + */ + nvchan->dma = b; + RING_SPACE_CH(chan, 8); + for (i = 0; i < 8; i++) + OUT_RING_CH(chan, 0); + nvchan->dma = m; + + nouveau_pushbuf_space(chan, 0); + chan->pushbuf = &nvchan->pb.base; return 0; } -static void -nouveau_pushbuf_fence_signalled(void *priv) -{ - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(priv); - - nouveau_fence_del(&nvpb->fence); - nouveau_resource_free(&nvpb->res); - free(nvpb); -} - /* This would be our TTM "superioctl" */ int -nouveau_pushbuf_flush(struct nouveau_channel *chan) +nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); + struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; struct nouveau_pushbuf_bo *pbbo; struct nouveau_fence *fence = NULL; int ret; - if (nvpb->base.remaining == (nvpb->res->size / 4) - PB_RSVD_DWORDS) + if (nvpb->base.remaining == nvpb->size) return 0; - nvchan->pb_tail = NULL; + + nvpb->size -= nvpb->base.remaining; + nvchan->dma->cur += nvpb->size; + nvchan->dma->free -= nvpb->size; + assert(nvchan->dma->cur <= nvchan->dma->max); ret = nouveau_fence_new(chan, &fence); if (ret) return ret; + nvchan->dma = &nvchan->dma_bufmgr; + nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 | + (nvchan->dma->base + (nvchan->dma->cur << 2)); + /* Validate buffers + apply relocations */ nvchan->user_charge = 0; while ((pbbo = ptr_to_pbbo(nvpb->buffers))) { @@ -142,25 +160,19 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) } nvpb->nr_buffers = 0; - /* Emit JMP to indirect pushbuf */ + /* Switch back to user's ring */ RING_SPACE_CH(chan, 1); - OUT_RING_CH(chan, 0x20000000 | nvpb->res->start); + OUT_RING_CH(chan, 0x20000000 | ((nvpb->start << 2) + + nvchan->dma_master.base)); + nvchan->dma = &nvchan->dma_master; - /* Add JMP back to master pushbuf from indirect pushbuf */ - (*nvpb->base.cur++) = - 0x20000000 | ((nvchan->dma.cur << 2) + nvchan->dma.base); - - /* Fence */ - nvpb->fence = fence; - nouveau_fence_signal_cb(nvpb->fence, nouveau_pushbuf_fence_signalled, - nvpb); - nouveau_fence_emit(nvpb->fence); - - /* Kickoff */ + /* Fence + kickoff */ + nouveau_fence_emit(fence); FIRE_RING_CH(chan); + nouveau_fence_del(&fence); /* Allocate space for next push buffer */ - assert(!nouveau_pushbuf_space(chan)); + assert(!nouveau_pushbuf_space(chan, min)); return 0; } @@ -168,8 +180,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan) static struct nouveau_pushbuf_bo * nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) { - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(nvchan->pb_tail); + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); struct nouveau_pushbuf_bo *pbbo = ptr_to_pbbo(nvpb->buffers); while (pbbo) { diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index b31e0dcd1dd..403647ab17e 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -40,8 +40,9 @@ nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) struct nouveau_channel *chan = grobj->channel; uint32_t *pushbuf; - if (chan->pushbuf->remaining < (size + 1)) - nouveau_pushbuf_flush(chan); + if (chan->pushbuf->remaining < (size + 1)) { + nouveau_pushbuf_flush(chan, size + 1); + } pushbuf = chan->pushbuf->cur; chan->pushbuf->cur += (size + 1); @@ -54,7 +55,7 @@ nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) void nouveau_pipe_dma_kickoff(struct nouveau_channel *chan) { - nouveau_pushbuf_flush(chan); + nouveau_pushbuf_flush(chan, 0); } static int -- cgit v1.2.3 From e32e0e2b8ea81e7b49c1cca4a99ac9b43790ce44 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 26 Dec 2007 00:37:21 +1100 Subject: nv40: use index buffers rather than inline indices. We probably want to use inline indices in some situations still, but this commit's primary purpose is to workaround some mis-rendering caused by a more complicated problem that'll get fixed eventually. --- src/mesa/pipe/nv40/nv40_vbo.c | 113 ++++++++++++++++++++++++++++++++++-------- 1 file changed, 93 insertions(+), 20 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index fda32cdb492..7562781fcfe 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -149,7 +149,8 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) } static boolean -nv40_vbo_validate_state(struct nv40_context *nv40) +nv40_vbo_validate_state(struct nv40_context *nv40, + struct pipe_buffer_handle *ib, unsigned ib_format) { unsigned inputs; @@ -173,6 +174,14 @@ nv40_vbo_validate_state(struct nv40_context *nv40) NV40TCL_VTXBUF_ADDRESS_DMA1); } + if (ib) { + BEGIN_RING(curie, 0x181c, 2); + OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD); + OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_OR, 0, 1); + } + BEGIN_RING(curie, 0x1710, 1); OUT_RING (0); /* vtx cache flush */ @@ -186,7 +195,7 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, struct nv40_context *nv40 = (struct nv40_context *)pipe; unsigned nr; - assert(nv40_vbo_validate_state(nv40)); + assert(nv40_vbo_validate_state(nv40, NULL, 0)); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -286,45 +295,109 @@ nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, } } -boolean -nv40_draw_elements(struct pipe_context *pipe, - struct pipe_buffer_handle *indexBuffer, unsigned indexSize, - unsigned mode, unsigned start, unsigned count) +static boolean +nv40_draw_elements_inline(struct pipe_context *pipe, + struct pipe_buffer_handle *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - void *ib; + struct pipe_winsys *ws = pipe->winsys; + void *map; - assert(nv40_vbo_validate_state(nv40)); + assert(nv40_vbo_validate_state(nv40, NULL, 0)); - ib = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, - PIPE_BUFFER_FLAG_READ); - if (!ib) { - NOUVEAU_ERR("Couldn't map index buffer!!\n"); - return FALSE; - } + map = ws->buffer_map(ws, ib, PIPE_BUFFER_FLAG_READ); + if (!ib) + assert(0); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); - switch (indexSize) { + switch (ib_size) { case 1: - nv40_draw_elements_u08(nv40, ib, start, count); + nv40_draw_elements_u08(nv40, map, start, count); break; case 2: - nv40_draw_elements_u16(nv40, ib, start, count); + nv40_draw_elements_u16(nv40, map, start, count); break; case 4: - nv40_draw_elements_u32(nv40, ib, start, count); + nv40_draw_elements_u32(nv40, map, start, count); break; default: - NOUVEAU_ERR("unsupported elt size %d\n", indexSize); + assert(0); + break; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + ws->buffer_unmap(ws, ib); + + return TRUE; +} + +static boolean +nv40_draw_elements_vbo(struct pipe_context *pipe, + struct pipe_buffer_handle *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv40_context *nv40 = (struct nv40_context *)pipe; + unsigned nr; + + switch (ib_size) { + case 2: + assert(nv40_vbo_validate_state(nv40, ib, 0x00000010)); break; + case 4: + assert(nv40_vbo_validate_state(nv40, ib, 0x00000000)); + break; + default: + assert(0); + } + + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(curie, 0x1824, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(curie, 0x1824, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } } BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (0); - pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); + return TRUE; +} + +boolean +nv40_draw_elements(struct pipe_context *pipe, + struct pipe_buffer_handle *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ + if (indexSize != 1) { + nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, + mode, start, count); + } else { + nv40_draw_elements_inline(pipe, indexBuffer, indexSize, + mode, start, count); + } + pipe->flush(pipe, 0); return TRUE; } -- cgit v1.2.3 From f33467d72a0bcfc48a21c9d2bf158dabda926442 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 26 Dec 2007 00:56:00 +1100 Subject: nouveau: header update --- src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 17 +- src/mesa/pipe/nouveau/nouveau_class.h | 4922 +++++++++++--------- src/mesa/pipe/nv40/nv40_state.c | 3 +- src/mesa/pipe/nv40/nv40_vbo.c | 16 +- 4 files changed, 2625 insertions(+), 2333 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c index b50e744d0d0..baf4804dde0 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c @@ -40,7 +40,7 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, while (h) { int count = (h > 2047) ? 2047 : h; - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | @@ -81,7 +81,8 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, * to NV_MEMORY_TO_MEMORY_FORMAT in this case. */ if ((src->offset & 63) || (dst->offset & 63)) { - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + BEGIN_RING(NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | @@ -179,7 +180,7 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) return 1; } BIND_RING (NvM2MF, nv->next_subchannel++); - BEGIN_RING(NvM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); + BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : @@ -194,7 +195,7 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) OUT_RING (nv->channel->vram->handle); OUT_RING (nv->channel->vram->handle); - class = nv->chipset < 0x10 ? NV_IMAGE_BLIT : + class = nv->chipset < 0x10 ? NV04_IMAGE_BLIT : NV12_IMAGE_BLIT; if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, &nv->NvImageBlit))) { @@ -202,12 +203,12 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) return 1; } BIND_RING (NvImageBlit, nv->next_subchannel++); - BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_DMA_NOTIFY, 1); + BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_SURFACE, 1); + BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvImageBlit, NV_IMAGE_BLIT_OPERATION, 1); - OUT_RING (NV_IMAGE_BLIT_OPERATION_SRCCOPY); + BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); + OUT_RING (NV04_IMAGE_BLIT_OPERATION_SRCCOPY); class = NV04_GDI_RECTANGLE_TEXT; if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h index 22b526a54ea..56f4e9001ec 100644 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -595,25 +595,25 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK 0xffff0000 -#define NV_MEMORY_TO_MEMORY_FORMAT 0x00000039 - -#define NV_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 -#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 -#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 -#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 -#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 -#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c -#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 -#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 -#define NV_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 -#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c -#define NV_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 -#define NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 -#define NV_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 +#define NV04_MEMORY_TO_MEMORY_FORMAT 0x00000039 + +#define NV04_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 +#define NV04_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 +#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 +#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 +#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 +#define NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c +#define NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 +#define NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 +#define NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 +#define NV04_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c +#define NV04_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 +#define NV04_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 #define NV01_MEMORY_LOCAL_BANKED 0x0000003d @@ -1744,1159 +1744,1159 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 -#define NV10_TCL_PRIMITIVE_3D 0x00000056 - -#define NV10_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV10_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 -#define NV10_TCL_PRIMITIVE_3D_DISPLAY_LIST 0x0000018c -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 -#define NV10_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 -#define NV10_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 -#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00000218+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00000220+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT 7 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK 0x00000780 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8 0x00000080 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT 0x00000180 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_X8R8G8B8 0x00000380 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8 0x00000580 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1 0x00000600 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3 0x00000700 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5 0x00000780 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT 0x00000980 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT2 0x00000d80 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT 0x00000e80 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT 0x00001000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16 0x00001900 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16_RECT 0x00001a80 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00002500 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00002580 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00002600 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT (1 << 11) -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_REPEAT 0x01000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x02000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x03000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x04000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP 0x05000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_SHIFT 28 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MASK 0xf0000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_REPEAT 0x10000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x20000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x30000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x40000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP 0x50000000 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00000228+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_MASK 0x00000030 -#define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00000248+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x01000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x02000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 28 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0xf0000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x10000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x20000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x30000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x40000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x50000000 -#define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x60000000 -#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SHIFT 5 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_FOG 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_FOG 0x00030000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x000c0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x000d0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_FOG 0x03000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0c000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0d000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000268+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SHIFT 5 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_FOG 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_NV 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE1_NV 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_FOG 0x00030000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_NV 0x000c0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE1_NV 0x000d0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_FOG 0x03000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0c000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0d000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR(x) (0x00000270+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_B_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_R_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_COLOR_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000030 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000080 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000090 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x000000c0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x000000d0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM (1 << 14) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS (1 << 15) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT 17 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00020000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00040000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000280+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000030 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000080 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000090 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x000000c0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x000000d0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM (1 << 14) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS (1 << 15) -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT 17 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00020000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00040000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_SHIFT 27 -#define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_OPERATION_MASK 0x38000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_FOG 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SHIFT 5 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_MASK 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_FOG 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_NV 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE1_NV 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_FOG 0x00030000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_NV 0x000c0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE1_NV 0x000d0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_FOG 0x03000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_NV 0x0c000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE1_NV 0x0d000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_MASK 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_FOG 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_NV 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE1_NV 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SHIFT 13 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_MASK 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_MASK 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_FOG 0x00030000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_NV 0x000c0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE1_NV 0x000d0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SHIFT 21 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_MASK 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_MASK 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_FOG 0x03000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_NV 0x0c000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE1_NV 0x0d000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SHIFT 29 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_MASK 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL 0x00000294 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL (1 << 1) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE 0x00000298 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP 0x00000800 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP_2 0x00000802 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_EXP2 0x00000803 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR 0x00000804 -#define NV10_TCL_PRIMITIVE_3D_FOG_MODE_LINEAR_2 0x00002601 -#define NV10_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 -#define NV10_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c -#define NV10_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 -#define NV10_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 -#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 -#define NV10_TCL_PRIMITIVE_3D_POINT_SMOOTH_ENABLE 0x0000031c -#define NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_WEIGHT_ENABLE 0x00000328 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 -#define NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_B (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G (1 << 8) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R (1 << 16) -#define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A (1 << 24) -#define NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c -#define NV10_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c -#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV10_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV10_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 -#define NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV10_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 -#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV10_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_R 0x000003a8 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_G 0x000003ac -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_B 0x000003b0 -#define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_A 0x000003b4 -#define NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL 0x000003b8 -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT0 (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT1 (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT2 (1 << 4) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT3 (1 << 6) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT4 (1 << 8) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT5 (1 << 10) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT6 (1 << 12) -#define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT7 (1 << 14) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_FALSE 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 -#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE 0x000003e8 -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) -#define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) -#define NV10_TCL_PRIMITIVE_3D_POINT_SIZE 0x000003ec -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x00000540+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000580+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000600+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000604+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000608+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) -#define NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x00000680 -#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x00000684 -#define NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x00000688 -#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000006c4 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000006c8 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000006cc -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X 0x000006e8 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Y 0x000006ec -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Z 0x000006f0 -#define NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_W 0x000006f4 -#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x000006f8+((x)*4)) -#define NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) -#define NV10_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00000c00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00000c04 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00000c08 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00000c18 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x00000c1c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00000c20 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W 0x00000c24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00000c30 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00000c34 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00000c38 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000c40 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000c44 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R 0x00000c50 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G 0x00000c54 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B 0x00000c58 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A 0x00000c5c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R 0x00000c60 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G 0x00000c64 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B 0x00000c68 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x00000c6c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R 0x00000c80 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G 0x00000c84 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B 0x00000c88 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I 0x00000c8c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK 0x000000ff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK 0x00ff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00000c90 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00000c94 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00000c98 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x00000ca0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x00000ca4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x00000ca8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x00000cac -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x00000cb0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x00000cb4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x00000cb8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x00000cbc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x00000cc0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x00000cc8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x00000ccc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x00000cd0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x00000cd4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x00000cd8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x00000cdc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00000ce0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_WGH_1F 0x00000ce4 -#define NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x00000cec -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE 0x00000cf0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_POS 0x00000d00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS 0x00000d04 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL 0x00000d08 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL 0x00000d0c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00000dfc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX 0x00000e00 -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I0_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_DRAW_INDEX_I1_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END 0x000013fc -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 -#define NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001800 +#define NV10TCL 0x00000056 + +#define NV10TCL_NOP 0x00000100 +#define NV10TCL_NOTIFY 0x00000104 +#define NV10TCL_DMA_NOTIFY 0x00000180 +#define NV10TCL_DMA_IN_MEMORY0 0x00000184 +#define NV10TCL_DMA_IN_MEMORY1 0x00000188 +#define NV10TCL_DISPLAY_LIST 0x0000018c +#define NV10TCL_DMA_IN_MEMORY2 0x00000194 +#define NV10TCL_DMA_IN_MEMORY3 0x00000198 +#define NV10TCL_VIEWPORT_HORIZ 0x00000200 +#define NV10TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV10TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV10TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV10TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV10TCL_VIEWPORT_VERT 0x00000204 +#define NV10TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV10TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV10TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV10TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV10TCL_BUFFER_FORMAT 0x00000208 +#define NV10TCL_BUFFER_PITCH 0x0000020c +#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 +#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff +#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 +#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 +#define NV10TCL_COLOR_OFFSET 0x00000210 +#define NV10TCL_ZETA_OFFSET 0x00000214 +#define NV10TCL_TX_OFFSET(x) (0x00000218+((x)*4)) +#define NV10TCL_TX_OFFSET__SIZE 0x00000002 +#define NV10TCL_TX_FORMAT(x) (0x00000220+((x)*4)) +#define NV10TCL_TX_FORMAT__SIZE 0x00000002 +#define NV10TCL_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV10TCL_TX_FORMAT_FORMAT_SHIFT 7 +#define NV10TCL_TX_FORMAT_FORMAT_MASK 0x00000780 +#define NV10TCL_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV10TCL_TX_FORMAT_FORMAT_A8 0x00000080 +#define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000100 +#define NV10TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000180 +#define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000200 +#define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000300 +#define NV10TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000380 +#define NV10TCL_TX_FORMAT_FORMAT_INDEX8 0x00000580 +#define NV10TCL_TX_FORMAT_FORMAT_DXT1 0x00000600 +#define NV10TCL_TX_FORMAT_FORMAT_DXT3 0x00000700 +#define NV10TCL_TX_FORMAT_FORMAT_DXT5 0x00000780 +#define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00000800 +#define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00000900 +#define NV10TCL_TX_FORMAT_FORMAT_L8_RECT 0x00000980 +#define NV10TCL_TX_FORMAT_FORMAT_A8L8 0x00000d00 +#define NV10TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00000d80 +#define NV10TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 +#define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00000e80 +#define NV10TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00001000 +#define NV10TCL_TX_FORMAT_FORMAT_A16 0x00001900 +#define NV10TCL_TX_FORMAT_FORMAT_A16_RECT 0x00001a80 +#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00002500 +#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00002580 +#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00002600 +#define NV10TCL_TX_FORMAT_NPOT (1 << 11) +#define NV10TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV10TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV10TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV10TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV10TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV10TCL_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV10TCL_TX_FORMAT_WRAP_S_SHIFT 24 +#define NV10TCL_TX_FORMAT_WRAP_S_MASK 0x0f000000 +#define NV10TCL_TX_FORMAT_WRAP_S_REPEAT 0x01000000 +#define NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x02000000 +#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x03000000 +#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x04000000 +#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP 0x05000000 +#define NV10TCL_TX_FORMAT_WRAP_T_SHIFT 28 +#define NV10TCL_TX_FORMAT_WRAP_T_MASK 0xf0000000 +#define NV10TCL_TX_FORMAT_WRAP_T_REPEAT 0x10000000 +#define NV10TCL_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x20000000 +#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x30000000 +#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x40000000 +#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP 0x50000000 +#define NV10TCL_TX_ENABLE(x) (0x00000228+((x)*4)) +#define NV10TCL_TX_ENABLE__SIZE 0x00000002 +#define NV10TCL_TX_ENABLE_ANISOTROPY_SHIFT 4 +#define NV10TCL_TX_ENABLE_ANISOTROPY_MASK 0x00000030 +#define NV10TCL_TX_ENABLE_ENABLE (1 << 30) +#define NV10TCL_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) +#define NV10TCL_TX_NPOT_PITCH__SIZE 0x00000002 +#define NV10TCL_TX_NPOT_PITCH_PITCH_SHIFT 16 +#define NV10TCL_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 +#define NV10TCL_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) +#define NV10TCL_TX_NPOT_SIZE__SIZE 0x00000002 +#define NV10TCL_TX_NPOT_SIZE_H_SHIFT 0 +#define NV10TCL_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV10TCL_TX_NPOT_SIZE_W_SHIFT 16 +#define NV10TCL_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV10TCL_TX_FILTER(x) (0x00000248+((x)*4)) +#define NV10TCL_TX_FILTER__SIZE 0x00000002 +#define NV10TCL_TX_FILTER_MINIFY_SHIFT 24 +#define NV10TCL_TX_FILTER_MINIFY_MASK 0x0f000000 +#define NV10TCL_TX_FILTER_MINIFY_NEAREST 0x01000000 +#define NV10TCL_TX_FILTER_MINIFY_LINEAR 0x02000000 +#define NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV10TCL_TX_FILTER_MAGNIFY_SHIFT 28 +#define NV10TCL_TX_FILTER_MAGNIFY_MASK 0xf0000000 +#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST 0x10000000 +#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR 0x20000000 +#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x30000000 +#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x40000000 +#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x50000000 +#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x60000000 +#define NV10TCL_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) +#define NV10TCL_TX_PALETTE_OFFSET__SIZE 0x00000002 +#define NV10TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV10TCL_RC_IN_ALPHA__SIZE 0x00000002 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SHIFT 0 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f +#define NV10TCL_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) +#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SHIFT 5 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SHIFT 8 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_FOG 0x00000300 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SHIFT 13 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SHIFT 16 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_FOG 0x00030000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SHIFT 21 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SHIFT 24 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_FOG 0x03000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SHIFT 29 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_RC_IN_RGB(x) (0x00000268+((x)*4)) +#define NV10TCL_RC_IN_RGB__SIZE 0x00000002 +#define NV10TCL_RC_IN_RGB_D_INPUT_SHIFT 0 +#define NV10TCL_RC_IN_RGB_D_INPUT_MASK 0x0000000f +#define NV10TCL_RC_IN_RGB_D_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_IN_RGB_D_INPUT_FOG 0x00000003 +#define NV10TCL_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) +#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10TCL_RC_IN_RGB_D_MAPPING_SHIFT 5 +#define NV10TCL_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 +#define NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10TCL_RC_IN_RGB_C_INPUT_SHIFT 8 +#define NV10TCL_RC_IN_RGB_C_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_IN_RGB_C_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_IN_RGB_C_INPUT_FOG 0x00000300 +#define NV10TCL_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_SHIFT 13 +#define NV10TCL_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SHIFT 16 +#define NV10TCL_RC_IN_RGB_B_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_IN_RGB_B_INPUT_FOG 0x00030000 +#define NV10TCL_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_SHIFT 21 +#define NV10TCL_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SHIFT 24 +#define NV10TCL_RC_IN_RGB_A_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_FOG 0x03000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_SHIFT 29 +#define NV10TCL_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_RC_COLOR(x) (0x00000270+((x)*4)) +#define NV10TCL_RC_COLOR__SIZE 0x00000002 +#define NV10TCL_RC_COLOR_B_SHIFT 0 +#define NV10TCL_RC_COLOR_B_MASK 0x000000ff +#define NV10TCL_RC_COLOR_G_SHIFT 8 +#define NV10TCL_RC_COLOR_G_MASK 0x0000ff00 +#define NV10TCL_RC_COLOR_R_SHIFT 16 +#define NV10TCL_RC_COLOR_R_MASK 0x00ff0000 +#define NV10TCL_RC_COLOR_A_SHIFT 24 +#define NV10TCL_RC_COLOR_A_MASK 0xff000000 +#define NV10TCL_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) +#define NV10TCL_RC_OUT_ALPHA__SIZE 0x00000002 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000030 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000080 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000090 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x000000c0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x000000d0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000300 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) +#define NV10TCL_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) +#define NV10TCL_RC_OUT_ALPHA_MUX_SUM (1 << 14) +#define NV10TCL_RC_OUT_ALPHA_BIAS (1 << 15) +#define NV10TCL_RC_OUT_ALPHA_BIAS_NONE 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SHIFT 17 +#define NV10TCL_RC_OUT_ALPHA_SCALE_MASK 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_NONE 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00020000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00040000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 +#define NV10TCL_RC_OUT_RGB(x) (0x00000280+((x)*4)) +#define NV10TCL_RC_OUT_RGB__SIZE 0x00000002 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000030 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000080 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000090 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x000000c0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x000000d0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000300 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) +#define NV10TCL_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) +#define NV10TCL_RC_OUT_RGB_MUX_SUM (1 << 14) +#define NV10TCL_RC_OUT_RGB_BIAS (1 << 15) +#define NV10TCL_RC_OUT_RGB_BIAS_NONE 0x00000000 +#define NV10TCL_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 +#define NV10TCL_RC_OUT_RGB_SCALE_SHIFT 17 +#define NV10TCL_RC_OUT_RGB_SCALE_MASK 0x00000000 +#define NV10TCL_RC_OUT_RGB_SCALE_NONE 0x00000000 +#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00020000 +#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00040000 +#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 +#define NV10TCL_RC_OUT_RGB_OPERATION_SHIFT 27 +#define NV10TCL_RC_OUT_RGB_OPERATION_MASK 0x38000000 +#define NV10TCL_RC_FINAL0 0x00000288 +#define NV10TCL_RC_FINAL0_D_INPUT_SHIFT 0 +#define NV10TCL_RC_FINAL0_D_INPUT_MASK 0x0000000f +#define NV10TCL_RC_FINAL0_D_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_FINAL0_D_INPUT_FOG 0x00000003 +#define NV10TCL_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) +#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10TCL_RC_FINAL0_D_MAPPING_SHIFT 5 +#define NV10TCL_RC_FINAL0_D_MAPPING_MASK 0x000000e0 +#define NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10TCL_RC_FINAL0_C_INPUT_SHIFT 8 +#define NV10TCL_RC_FINAL0_C_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_FINAL0_C_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_FINAL0_C_INPUT_FOG 0x00000300 +#define NV10TCL_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_FINAL0_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_FINAL0_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_FINAL0_C_MAPPING_SHIFT 13 +#define NV10TCL_RC_FINAL0_C_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_FINAL0_B_INPUT_SHIFT 16 +#define NV10TCL_RC_FINAL0_B_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_FINAL0_B_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_FINAL0_B_INPUT_FOG 0x00030000 +#define NV10TCL_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_FINAL0_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_FINAL0_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_FINAL0_B_MAPPING_SHIFT 21 +#define NV10TCL_RC_FINAL0_B_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_FINAL0_A_INPUT_SHIFT 24 +#define NV10TCL_RC_FINAL0_A_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_FINAL0_A_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_FINAL0_A_INPUT_FOG 0x03000000 +#define NV10TCL_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_SHIFT 29 +#define NV10TCL_RC_FINAL0_A_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_RC_FINAL1 0x0000028c +#define NV10TCL_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) +#define NV10TCL_RC_FINAL1_G_INPUT_SHIFT 8 +#define NV10TCL_RC_FINAL1_G_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_FINAL1_G_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_FINAL1_G_INPUT_FOG 0x00000300 +#define NV10TCL_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_FINAL1_G_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_FINAL1_G_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_FINAL1_G_MAPPING_SHIFT 13 +#define NV10TCL_RC_FINAL1_G_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_FINAL1_F_INPUT_SHIFT 16 +#define NV10TCL_RC_FINAL1_F_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_FINAL1_F_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_FINAL1_F_INPUT_FOG 0x00030000 +#define NV10TCL_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_FINAL1_F_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_FINAL1_F_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_FINAL1_F_MAPPING_SHIFT 21 +#define NV10TCL_RC_FINAL1_F_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_FINAL1_E_INPUT_SHIFT 24 +#define NV10TCL_RC_FINAL1_E_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_FINAL1_E_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_FINAL1_E_INPUT_FOG 0x03000000 +#define NV10TCL_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_SHIFT 29 +#define NV10TCL_RC_FINAL1_E_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_LIGHT_MODEL 0x00000294 +#define NV10TCL_LIGHT_MODEL_COLOR_CONTROL (1 << 1) +#define NV10TCL_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) +#define NV10TCL_COLOR_MATERIAL_ENABLE 0x00000298 +#define NV10TCL_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) +#define NV10TCL_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) +#define NV10TCL_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) +#define NV10TCL_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) +#define NV10TCL_FOG_MODE 0x0000029c +#define NV10TCL_FOG_MODE_EXP 0x00000800 +#define NV10TCL_FOG_MODE_EXP_2 0x00000802 +#define NV10TCL_FOG_MODE_EXP2 0x00000803 +#define NV10TCL_FOG_MODE_LINEAR 0x00000804 +#define NV10TCL_FOG_MODE_LINEAR_2 0x00002601 +#define NV10TCL_FOG_COORD_DIST 0x000002a0 +#define NV10TCL_FOG_ENABLE 0x000002a4 +#define NV10TCL_FOG_COLOR 0x000002a8 +#define NV10TCL_FOG_COLOR_R_SHIFT 0 +#define NV10TCL_FOG_COLOR_R_MASK 0x000000ff +#define NV10TCL_FOG_COLOR_G_SHIFT 8 +#define NV10TCL_FOG_COLOR_G_MASK 0x0000ff00 +#define NV10TCL_FOG_COLOR_B_SHIFT 16 +#define NV10TCL_FOG_COLOR_B_MASK 0x00ff0000 +#define NV10TCL_FOG_COLOR_A_SHIFT 24 +#define NV10TCL_FOG_COLOR_A_MASK 0xff000000 +#define NV10TCL_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV10TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV10TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) +#define NV10TCL_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV10TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) +#define NV10TCL_ALPHA_FUNC_ENABLE 0x00000300 +#define NV10TCL_BLEND_FUNC_ENABLE 0x00000304 +#define NV10TCL_CULL_FACE_ENABLE 0x00000308 +#define NV10TCL_DEPTH_TEST_ENABLE 0x0000030c +#define NV10TCL_DITHER_ENABLE 0x00000310 +#define NV10TCL_LIGHTING_ENABLE 0x00000314 +#define NV10TCL_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV10TCL_POINT_SMOOTH_ENABLE 0x0000031c +#define NV10TCL_LINE_SMOOTH_ENABLE 0x00000320 +#define NV10TCL_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV10TCL_VERTEX_WEIGHT_ENABLE 0x00000328 +#define NV10TCL_STENCIL_ENABLE 0x0000032c +#define NV10TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV10TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV10TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV10TCL_ALPHA_FUNC_FUNC 0x0000033c +#define NV10TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV10TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV10TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV10TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10TCL_ALPHA_FUNC_REF 0x00000340 +#define NV10TCL_BLEND_FUNC_SRC 0x00000344 +#define NV10TCL_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV10TCL_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV10TCL_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10TCL_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10TCL_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV10TCL_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10TCL_BLEND_FUNC_DST 0x00000348 +#define NV10TCL_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV10TCL_BLEND_FUNC_DST_ONE 0x00000001 +#define NV10TCL_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10TCL_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10TCL_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10TCL_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV10TCL_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10TCL_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10TCL_BLEND_COLOR 0x0000034c +#define NV10TCL_BLEND_COLOR_B_SHIFT 0 +#define NV10TCL_BLEND_COLOR_B_MASK 0x000000ff +#define NV10TCL_BLEND_COLOR_G_SHIFT 8 +#define NV10TCL_BLEND_COLOR_G_MASK 0x0000ff00 +#define NV10TCL_BLEND_COLOR_R_SHIFT 16 +#define NV10TCL_BLEND_COLOR_R_MASK 0x00ff0000 +#define NV10TCL_BLEND_COLOR_A_SHIFT 24 +#define NV10TCL_BLEND_COLOR_A_MASK 0xff000000 +#define NV10TCL_BLEND_EQUATION 0x00000350 +#define NV10TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV10TCL_BLEND_EQUATION_MIN 0x00008007 +#define NV10TCL_BLEND_EQUATION_MAX 0x00008008 +#define NV10TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV10TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV10TCL_DEPTH_FUNC 0x00000354 +#define NV10TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV10TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV10TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV10TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV10TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV10TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV10TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV10TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV10TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV10TCL_COLOR_MASK 0x00000358 +#define NV10TCL_COLOR_MASK_B (1 << 0) +#define NV10TCL_COLOR_MASK_G (1 << 8) +#define NV10TCL_COLOR_MASK_R (1 << 16) +#define NV10TCL_COLOR_MASK_A (1 << 24) +#define NV10TCL_DEPTH_WRITE_ENABLE 0x0000035c +#define NV10TCL_STENCIL_MASK 0x00000360 +#define NV10TCL_STENCIL_FUNC_FUNC 0x00000364 +#define NV10TCL_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV10TCL_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV10TCL_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV10TCL_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10TCL_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10TCL_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10TCL_STENCIL_FUNC_REF 0x00000368 +#define NV10TCL_STENCIL_FUNC_MASK 0x0000036c +#define NV10TCL_STENCIL_OP_FAIL 0x00000370 +#define NV10TCL_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV10TCL_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV10TCL_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV10TCL_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV10TCL_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV10TCL_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV10TCL_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV10TCL_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV10TCL_STENCIL_OP_ZFAIL 0x00000374 +#define NV10TCL_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV10TCL_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV10TCL_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV10TCL_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV10TCL_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV10TCL_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV10TCL_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV10TCL_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV10TCL_STENCIL_OP_ZPASS 0x00000378 +#define NV10TCL_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV10TCL_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV10TCL_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV10TCL_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV10TCL_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV10TCL_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV10TCL_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV10TCL_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV10TCL_SHADE_MODEL 0x0000037c +#define NV10TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV10TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV10TCL_LINE_WIDTH 0x00000380 +#define NV10TCL_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV10TCL_POLYGON_OFFSET_UNITS 0x00000388 +#define NV10TCL_POLYGON_MODE_FRONT 0x0000038c +#define NV10TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV10TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV10TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV10TCL_POLYGON_MODE_BACK 0x00000390 +#define NV10TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV10TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV10TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV10TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV10TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV10TCL_CULL_FACE 0x0000039c +#define NV10TCL_CULL_FACE_FRONT 0x00000404 +#define NV10TCL_CULL_FACE_BACK 0x00000405 +#define NV10TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV10TCL_FRONT_FACE 0x000003a0 +#define NV10TCL_FRONT_FACE_CW 0x00000900 +#define NV10TCL_FRONT_FACE_CCW 0x00000901 +#define NV10TCL_NORMALIZE_ENABLE 0x000003a4 +#define NV10TCL_COLOR_MATERIAL_R 0x000003a8 +#define NV10TCL_COLOR_MATERIAL_G 0x000003ac +#define NV10TCL_COLOR_MATERIAL_B 0x000003b0 +#define NV10TCL_COLOR_MATERIAL_A 0x000003b4 +#define NV10TCL_COLOR_CONTROL 0x000003b8 +#define NV10TCL_ENABLED_LIGHTS 0x000003bc +#define NV10TCL_ENABLED_LIGHTS_LIGHT0 (1 << 0) +#define NV10TCL_ENABLED_LIGHTS_LIGHT1 (1 << 2) +#define NV10TCL_ENABLED_LIGHTS_LIGHT2 (1 << 4) +#define NV10TCL_ENABLED_LIGHTS_LIGHT3 (1 << 6) +#define NV10TCL_ENABLED_LIGHTS_LIGHT4 (1 << 8) +#define NV10TCL_ENABLED_LIGHTS_LIGHT5 (1 << 10) +#define NV10TCL_ENABLED_LIGHTS_LIGHT6 (1 << 12) +#define NV10TCL_ENABLED_LIGHTS_LIGHT7 (1 << 14) +#define NV10TCL_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV10TCL_CLIP_PLANE_ENABLE__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_ENABLE_FALSE 0x00000000 +#define NV10TCL_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 +#define NV10TCL_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 +#define NV10TCL_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) +#define NV10TCL_TX_MATRIX_ENABLE__SIZE 0x00000002 +#define NV10TCL_VIEW_MATRIX_ENABLE 0x000003e8 +#define NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) +#define NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) +#define NV10TCL_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) +#define NV10TCL_POINT_SIZE 0x000003ec +#define NV10TCL_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) +#define NV10TCL_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10TCL_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) +#define NV10TCL_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10TCL_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) +#define NV10TCL_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10TCL_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) +#define NV10TCL_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10TCL_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) +#define NV10TCL_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV10TCL_TX0_MATRIX(x) (0x00000540+((x)*4)) +#define NV10TCL_TX0_MATRIX__SIZE 0x00000010 +#define NV10TCL_TX1_MATRIX(x) (0x00000580+((x)*4)) +#define NV10TCL_TX1_MATRIX__SIZE 0x00000010 +#define NV10TCL_CLIP_PLANE_A(x) (0x00000600+((x)*16)) +#define NV10TCL_CLIP_PLANE_A__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_B(x) (0x00000604+((x)*16)) +#define NV10TCL_CLIP_PLANE_B__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_C(x) (0x00000608+((x)*16)) +#define NV10TCL_CLIP_PLANE_C__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) +#define NV10TCL_CLIP_PLANE_D__SIZE 0x00000008 +#define NV10TCL_FOG_EQUATION_CONSTANT 0x00000680 +#define NV10TCL_FOG_EQUATION_LINEAR 0x00000684 +#define NV10TCL_FOG_EQUATION_QUADRATIC 0x00000688 +#define NV10TCL_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) +#define NV10TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000006c4 +#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000006c8 +#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000006cc +#define NV10TCL_VIEWPORT_SCALE_X 0x000006e8 +#define NV10TCL_VIEWPORT_SCALE_Y 0x000006ec +#define NV10TCL_VIEWPORT_SCALE_Z 0x000006f0 +#define NV10TCL_VIEWPORT_SCALE_W 0x000006f4 +#define NV10TCL_POINT_PARAMETER(x) (0x000006f8+((x)*4)) +#define NV10TCL_POINT_PARAMETER__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) +#define NV10TCL_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) +#define NV10TCL_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) +#define NV10TCL_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) +#define NV10TCL_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) +#define NV10TCL_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) +#define NV10TCL_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) +#define NV10TCL_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) +#define NV10TCL_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) +#define NV10TCL_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV10TCL_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) +#define NV10TCL_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) +#define NV10TCL_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) +#define NV10TCL_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) +#define NV10TCL_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV10TCL_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) +#define NV10TCL_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV10TCL_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) +#define NV10TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV10TCL_VERTEX_POS_3F_X 0x00000c00 +#define NV10TCL_VERTEX_POS_3F_Y 0x00000c04 +#define NV10TCL_VERTEX_POS_3F_Z 0x00000c08 +#define NV10TCL_VERTEX_POS_4F_X 0x00000c18 +#define NV10TCL_VERTEX_POS_4F_Y 0x00000c1c +#define NV10TCL_VERTEX_POS_4F_Z 0x00000c20 +#define NV10TCL_VERTEX_POS_4F_W 0x00000c24 +#define NV10TCL_VERTEX_NOR_3F_X 0x00000c30 +#define NV10TCL_VERTEX_NOR_3F_Y 0x00000c34 +#define NV10TCL_VERTEX_NOR_3F_Z 0x00000c38 +#define NV10TCL_VERTEX_NOR_3I_XY 0x00000c40 +#define NV10TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV10TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV10TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV10TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV10TCL_VERTEX_NOR_3I_Z 0x00000c44 +#define NV10TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV10TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV10TCL_VERTEX_COL_4F_R 0x00000c50 +#define NV10TCL_VERTEX_COL_4F_G 0x00000c54 +#define NV10TCL_VERTEX_COL_4F_B 0x00000c58 +#define NV10TCL_VERTEX_COL_4F_A 0x00000c5c +#define NV10TCL_VERTEX_COL_3F_R 0x00000c60 +#define NV10TCL_VERTEX_COL_3F_G 0x00000c64 +#define NV10TCL_VERTEX_COL_3F_B 0x00000c68 +#define NV10TCL_VERTEX_COL_4I 0x00000c6c +#define NV10TCL_VERTEX_COL_4I_R_SHIFT 0 +#define NV10TCL_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV10TCL_VERTEX_COL_4I_G_SHIFT 8 +#define NV10TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV10TCL_VERTEX_COL_4I_B_SHIFT 16 +#define NV10TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV10TCL_VERTEX_COL_4I_A_SHIFT 24 +#define NV10TCL_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV10TCL_VERTEX_COL2_3F_R 0x00000c80 +#define NV10TCL_VERTEX_COL2_3F_G 0x00000c84 +#define NV10TCL_VERTEX_COL2_3F_B 0x00000c88 +#define NV10TCL_VERTEX_COL2_3I 0x00000c8c +#define NV10TCL_VERTEX_COL2_3I_R_SHIFT 0 +#define NV10TCL_VERTEX_COL2_3I_R_MASK 0x000000ff +#define NV10TCL_VERTEX_COL2_3I_G_SHIFT 8 +#define NV10TCL_VERTEX_COL2_3I_G_MASK 0x0000ff00 +#define NV10TCL_VERTEX_COL2_3I_B_SHIFT 16 +#define NV10TCL_VERTEX_COL2_3I_B_MASK 0x00ff0000 +#define NV10TCL_VERTEX_TX0_2F_S 0x00000c90 +#define NV10TCL_VERTEX_TX0_2F_T 0x00000c94 +#define NV10TCL_VERTEX_TX0_2I 0x00000c98 +#define NV10TCL_VERTEX_TX0_2I_S_SHIFT 0 +#define NV10TCL_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX0_2I_T_SHIFT 16 +#define NV10TCL_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX0_4F_S 0x00000ca0 +#define NV10TCL_VERTEX_TX0_4F_T 0x00000ca4 +#define NV10TCL_VERTEX_TX0_4F_R 0x00000ca8 +#define NV10TCL_VERTEX_TX0_4F_Q 0x00000cac +#define NV10TCL_VERTEX_TX0_4I_ST 0x00000cb0 +#define NV10TCL_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV10TCL_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV10TCL_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX0_4I_RQ 0x00000cb4 +#define NV10TCL_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV10TCL_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV10TCL_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX1_2F_S 0x00000cb8 +#define NV10TCL_VERTEX_TX1_2F_T 0x00000cbc +#define NV10TCL_VERTEX_TX1_2I 0x00000cc0 +#define NV10TCL_VERTEX_TX1_2I_S_SHIFT 0 +#define NV10TCL_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX1_2I_T_SHIFT 16 +#define NV10TCL_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX1_4F_S 0x00000cc8 +#define NV10TCL_VERTEX_TX1_4F_T 0x00000ccc +#define NV10TCL_VERTEX_TX1_4F_R 0x00000cd0 +#define NV10TCL_VERTEX_TX1_4F_Q 0x00000cd4 +#define NV10TCL_VERTEX_TX1_4I_ST 0x00000cd8 +#define NV10TCL_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV10TCL_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV10TCL_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX1_4I_RQ 0x00000cdc +#define NV10TCL_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV10TCL_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV10TCL_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV10TCL_VERTEX_FOG_1F 0x00000ce0 +#define NV10TCL_VERTEX_WGH_1F 0x00000ce4 +#define NV10TCL_EDGEFLAG_ENABLE 0x00000cec +#define NV10TCL_VERTEX_ARRAY_VALIDATE 0x00000cf0 +#define NV10TCL_VERTEX_ARRAY_OFFSET_POS 0x00000d00 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS 0x00000d04 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_COL 0x00000d08 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL 0x00000d0c +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_BEGIN_END 0x00000dfc +#define NV10TCL_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV10TCL_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV10TCL_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV10TCL_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10TCL_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10TCL_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV10TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10TCL_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10TCL_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV10TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV10TCL_DRAW_INDEX 0x00000e00 +#define NV10TCL_DRAW_INDEX_I0_SHIFT 0 +#define NV10TCL_DRAW_INDEX_I0_MASK 0x0000ffff +#define NV10TCL_DRAW_INDEX_I1_SHIFT 24 +#define NV10TCL_DRAW_INDEX_I1_MASK 0xff000000 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END 0x000013fc +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 +#define NV10TCL_VERTEX_ARRAY_DATA 0x00001800 #define NV04_CONTEXT_COLOR_KEY 0x00000057 @@ -2943,24 +2943,24 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV04_RENDER_SOLID_RECTANGLE_SURFACE 0x00000198 -#define NV_IMAGE_BLIT 0x0000005f +#define NV04_IMAGE_BLIT 0x0000005f -#define NV_IMAGE_BLIT_NOP 0x00000100 -#define NV_IMAGE_BLIT_NOTIFY 0x00000104 -#define NV_IMAGE_BLIT_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_BLIT_COLOR_KEY 0x00000184 -#define NV_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 -#define NV_IMAGE_BLIT_PATTERN 0x0000018c -#define NV_IMAGE_BLIT_ROP 0x00000190 -#define NV_IMAGE_BLIT_BETA4 0x00000198 -#define NV_IMAGE_BLIT_SURFACE 0x0000019c -#define NV_IMAGE_BLIT_OPERATION 0x000002fc -#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 -#define NV_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 -#define NV_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 -#define NV_IMAGE_BLIT_OPERATION_SRCCOPY 0x00000003 -#define NV_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000005 +#define NV04_IMAGE_BLIT_NOP 0x00000100 +#define NV04_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV04_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV04_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV04_IMAGE_BLIT_PATTERN 0x0000018c +#define NV04_IMAGE_BLIT_ROP 0x00000190 +#define NV04_IMAGE_BLIT_BETA4 0x00000198 +#define NV04_IMAGE_BLIT_SURFACE 0x0000019c +#define NV04_IMAGE_BLIT_OPERATION 0x000002fc +#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 +#define NV04_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 +#define NV04_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 +#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY 0x00000003 +#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV04_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000005 #define NV04_INDEXED_IMAGE_FROM_CPU 0x00000060 @@ -2979,10 +2979,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV04_INDEXED_IMAGE_FROM_CPU_COLOR 0x00000400 -#define NV_IMAGE_FROM_CPU 0x00000061 +#define NV04_IMAGE_FROM_CPU 0x00000061 -#define NV_IMAGE_FROM_CPU_BETA4 0x00000198 -#define NV_IMAGE_FROM_CPU_SURFACE 0x0000019c +#define NV04_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV04_IMAGE_FROM_CPU_SURFACE 0x0000019c #define NV10_CONTEXT_SURFACES_2D 0x00000062 @@ -3204,435 +3204,436 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. -#define NV11_TCL_PRIMITIVE_3D 0x00000096 - -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000d40 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000d44 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e -#define NV11_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f - - -#define NV20_TCL_PRIMITIVE_3D 0x00000097 - -#define NV20_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV20_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV20_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x00000194 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000198 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a4 -#define NV20_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001a8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT 0x00000208 -#define NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH 0x0000020c -#define NV20_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 -#define NV20_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 -#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000260+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_FINAL0 0x00000288 -#define NV20_TCL_PRIMITIVE_3D_RC_FINAL1 0x0000028c -#define NV20_TCL_PRIMITIVE_3D_LIGHT_CONTROL 0x00000294 -#define NV20_TCL_PRIMITIVE_3D_FOG_MODE 0x0000029c -#define NV20_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000002a0 -#define NV20_TCL_PRIMITIVE_3D_FOG_ENABLE 0x000002a4 -#define NV20_TCL_PRIMITIVE_3D_FOG_COLOR 0x000002a8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE 0x000002b4 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000300 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000304 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x00000308 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x0000030c -#define NV20_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000310 -#define NV20_TCL_PRIMITIVE_3D_LIGHTING_ENABLE 0x00000314 -#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00000318 -#define NV20_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x00000320 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00000324 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_ENABLE 0x0000032c -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000330 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000334 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000338 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x0000033c -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x00000340 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000344 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE 0x00000001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_DST_COLOR 0x00000306 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000348 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE 0x00000001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_COLOR 0x00000300 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_ALPHA 0x00000304 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_DST_COLOR 0x00000306 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 -#define NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV20_TCL_PRIMITIVE_3D_BLEND_COLOR 0x0000034c -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION 0x00000350 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_ADD 0x00008006 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MIN 0x00008007 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_MAX 0x00008008 -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000354 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV20_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000358 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x0000035c -#define NV20_TCL_PRIMITIVE_3D_STENCIL_MASK 0x00000360 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC 0x00000364 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NEVER 0x00000200 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LESS 0x00000201 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_EQUAL 0x00000202 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF 0x00000368 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK 0x0000036c -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL 0x00000370 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INVERT 0x0000150a -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_KEEP 0x00001e00 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_REPLACE 0x00001e01 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR 0x00001e02 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR 0x00001e03 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL 0x00000374 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INVERT 0x0000150a -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_KEEP 0x00001e00 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR 0x00001e02 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR 0x00001e03 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS 0x00000378 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_ZERO 0x00000000 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INVERT 0x0000150a -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_KEEP 0x00001e00 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_REPLACE 0x00001e01 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR 0x00001e02 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR 0x00001e03 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL 0x0000037c -#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV20_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV20_TCL_PRIMITIVE_3D_LINE_WIDTH 0x00000380 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000384 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000388 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x0000038c -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x00000390 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 -#define NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE 0x0000039c -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV20_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE 0x000003a0 -#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV20_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x000003a4 -#define NV20_TCL_PRIMITIVE_3D_SEPARATE_SPECULAR_ENABLE 0x000003b8 -#define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS 0x000003bc -#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_POINT_SIZE 0x0000043c -#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000009c0 -#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000009c4 -#define NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000009c8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c -#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER(x) (0x00000a30+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER__SIZE 0x00000007 -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000ac0+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000af0 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000af4 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000af8 -#define NV20_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000afc -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X 0x00001500 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y 0x00001504 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z 0x00001508 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X 0x00001518 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y 0x0000151c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z 0x00001520 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY 0x00001528 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_X_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z 0x0000152c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X 0x00001530 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y 0x00001534 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z 0x00001538 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00001540 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00001544 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_X 0x00001550 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Y 0x00001554 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_Z 0x00001558 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_W 0x0000155c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_X 0x00001560 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Y 0x00001564 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_Z 0x00001568 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I 0x0000156c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_X 0x00001580 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Y 0x00001584 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_Z 0x00001588 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I 0x0000158c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_R_MASK 0x000000ff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_SHIFT 8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_G_MASK 0x0000ff00 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_B_MASK 0x00ff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_SHIFT 24 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_4I_A_MASK 0xff000000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S 0x00001590 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T 0x00001594 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I 0x00001598 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S 0x000015a0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T 0x000015a4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R 0x000015a8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q 0x000015ac -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST 0x000015b0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ 0x000015b4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S 0x000015b8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T 0x000015bc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I 0x000015c0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S 0x000015c8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T 0x000015cc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R 0x000015d0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q 0x000015d4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST 0x000015d8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ 0x000015dc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S 0x000015e0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T 0x000015e4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I 0x000015e8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S 0x000015f0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T 0x000015f4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R 0x000015f8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q 0x000015fc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST 0x00001600 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ 0x00001604 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S 0x00001608 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T 0x0000160c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I 0x00001610 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S 0x00001620 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T 0x00001624 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R 0x00001628 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q 0x0000162c -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST 0x00001630 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ 0x00001634 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT 16 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 -#define NV20_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE 0x000016bc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR_OFFSET__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 -#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x000017bc -#define NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x000017c0 -#define NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 -#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE 0x000017f8 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x000017fc -#define NV20_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 -#define NV20_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA 0x00001818 -#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001b00+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001b04+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001b08+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001b0c+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001b14+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) -#define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 -#define NV20_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x00001c30 -#define NV20_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x00001c50 -#define NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE 0x00001d90 -#define NV20_TCL_PRIMITIVE_3D_CLEAR_BUFFERS 0x00001d94 -#define NV20_TCL_PRIMITIVE_3D_RC_COLOR0 0x00001e20 -#define NV20_TCL_PRIMITIVE_3D_RC_COLOR1 0x00001e24 -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00001e40+((x)*4)) -#define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 -#define NV20_TCL_PRIMITIVE_3D_RC_ENABLE 0x00001e60 -#define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP 0x00001e70 -#define NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID 0x00001ea4 - - -#define NV17_TCL_PRIMITIVE_3D 0x00000099 - -#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x000001ac -#define NV17_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001b0 -#define NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE 0x000002bc -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH 0x00000d5c -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE 0x00000d68 -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c -#define NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE 0x00001658 +#define NV11TCL 0x00000096 + +#define NV11TCL_COLOR_LOGIC_OP_ENABLE 0x00000d40 +#define NV11TCL_COLOR_LOGIC_OP_OP 0x00000d44 +#define NV11TCL_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV11TCL_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV11TCL_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV11TCL_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV11TCL_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV11TCL_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV11TCL_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV11TCL_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV11TCL_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV11TCL_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV11TCL_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV11TCL_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV11TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV11TCL_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV11TCL_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV11TCL_COLOR_LOGIC_OP_OP_SET 0x0000150f + + +#define NV20TCL 0x00000097 + +#define NV20TCL_NOP 0x00000100 +#define NV20TCL_NOTIFY 0x00000104 +#define NV20TCL_DMA_NOTIFY 0x00000180 +#define NV20TCL_DMA_IN_MEMORY0 0x00000184 +#define NV20TCL_DMA_IN_MEMORY1 0x00000188 +#define NV20TCL_DMA_IN_MEMORY2 0x00000194 +#define NV20TCL_DMA_IN_MEMORY3 0x00000198 +#define NV20TCL_DMA_IN_MEMORY6 0x000001a4 +#define NV20TCL_DMA_IN_MEMORY7 0x000001a8 +#define NV20TCL_VIEWPORT_HORIZ 0x00000200 +#define NV20TCL_VIEWPORT_VERT 0x00000204 +#define NV20TCL_BUFFER_FORMAT 0x00000208 +#define NV20TCL_BUFFER_PITCH 0x0000020c +#define NV20TCL_COLOR_OFFSET 0x00000210 +#define NV20TCL_ZETA_OFFSET 0x00000214 +#define NV20TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV20TCL_RC_IN_ALPHA__SIZE 0x00000008 +#define NV20TCL_RC_FINAL0 0x00000288 +#define NV20TCL_RC_FINAL1 0x0000028c +#define NV20TCL_LIGHT_CONTROL 0x00000294 +#define NV20TCL_FOG_MODE 0x0000029c +#define NV20TCL_FOG_COORD_DIST 0x000002a0 +#define NV20TCL_FOG_ENABLE 0x000002a4 +#define NV20TCL_FOG_COLOR 0x000002a8 +#define NV20TCL_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV20TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV20TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV20TCL_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV20TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV20TCL_ALPHA_FUNC_ENABLE 0x00000300 +#define NV20TCL_BLEND_FUNC_ENABLE 0x00000304 +#define NV20TCL_CULL_FACE_ENABLE 0x00000308 +#define NV20TCL_DEPTH_TEST_ENABLE 0x0000030c +#define NV20TCL_DITHER_ENABLE 0x00000310 +#define NV20TCL_LIGHTING_ENABLE 0x00000314 +#define NV20TCL_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV20TCL_LINE_SMOOTH_ENABLE 0x00000320 +#define NV20TCL_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV20TCL_STENCIL_ENABLE 0x0000032c +#define NV20TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV20TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV20TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV20TCL_ALPHA_FUNC_FUNC 0x0000033c +#define NV20TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV20TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV20TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV20TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20TCL_ALPHA_FUNC_REF 0x00000340 +#define NV20TCL_BLEND_FUNC_SRC 0x00000344 +#define NV20TCL_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV20TCL_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV20TCL_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20TCL_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20TCL_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV20TCL_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20TCL_BLEND_FUNC_DST 0x00000348 +#define NV20TCL_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV20TCL_BLEND_FUNC_DST_ONE 0x00000001 +#define NV20TCL_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20TCL_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20TCL_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20TCL_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV20TCL_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20TCL_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20TCL_BLEND_COLOR 0x0000034c +#define NV20TCL_BLEND_EQUATION 0x00000350 +#define NV20TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV20TCL_BLEND_EQUATION_MIN 0x00008007 +#define NV20TCL_BLEND_EQUATION_MAX 0x00008008 +#define NV20TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV20TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV20TCL_DEPTH_FUNC 0x00000354 +#define NV20TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV20TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV20TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV20TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV20TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV20TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV20TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV20TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV20TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV20TCL_COLOR_MASK 0x00000358 +#define NV20TCL_DEPTH_WRITE_ENABLE 0x0000035c +#define NV20TCL_STENCIL_MASK 0x00000360 +#define NV20TCL_STENCIL_FUNC_FUNC 0x00000364 +#define NV20TCL_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV20TCL_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV20TCL_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV20TCL_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20TCL_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20TCL_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20TCL_STENCIL_FUNC_REF 0x00000368 +#define NV20TCL_STENCIL_FUNC_MASK 0x0000036c +#define NV20TCL_STENCIL_OP_FAIL 0x00000370 +#define NV20TCL_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV20TCL_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV20TCL_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV20TCL_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV20TCL_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV20TCL_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV20TCL_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV20TCL_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV20TCL_STENCIL_OP_ZFAIL 0x00000374 +#define NV20TCL_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV20TCL_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV20TCL_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV20TCL_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV20TCL_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV20TCL_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV20TCL_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV20TCL_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV20TCL_STENCIL_OP_ZPASS 0x00000378 +#define NV20TCL_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV20TCL_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV20TCL_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV20TCL_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV20TCL_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV20TCL_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV20TCL_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV20TCL_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV20TCL_SHADE_MODEL 0x0000037c +#define NV20TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV20TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV20TCL_LINE_WIDTH 0x00000380 +#define NV20TCL_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV20TCL_POLYGON_OFFSET_UNITS 0x00000388 +#define NV20TCL_POLYGON_MODE_FRONT 0x0000038c +#define NV20TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV20TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV20TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV20TCL_POLYGON_MODE_BACK 0x00000390 +#define NV20TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV20TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV20TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV20TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV20TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV20TCL_CULL_FACE 0x0000039c +#define NV20TCL_CULL_FACE_FRONT 0x00000404 +#define NV20TCL_CULL_FACE_BACK 0x00000405 +#define NV20TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV20TCL_FRONT_FACE 0x000003a0 +#define NV20TCL_FRONT_FACE_CW 0x00000900 +#define NV20TCL_FRONT_FACE_CCW 0x00000901 +#define NV20TCL_NORMALIZE_ENABLE 0x000003a4 +#define NV20TCL_SEPARATE_SPECULAR_ENABLE 0x000003b8 +#define NV20TCL_ENABLED_LIGHTS 0x000003bc +#define NV20TCL_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV20TCL_CLIP_PLANE_ENABLE__SIZE 0x00000010 +#define NV20TCL_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) +#define NV20TCL_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV20TCL_POINT_SIZE 0x0000043c +#define NV20TCL_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV20TCL_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV20TCL_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV20TCL_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV20TCL_TX0_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV20TCL_TX1_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV20TCL_TX2_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV20TCL_TX3_MATRIX__SIZE 0x00000010 +#define NV20TCL_FOG_EQUATION_CONSTANT 0x000009c0 +#define NV20TCL_FOG_EQUATION_LINEAR 0x000009c4 +#define NV20TCL_FOG_EQUATION_QUADRATIC 0x000009c8 +#define NV20TCL_VIEWPORT_SCALE0_X 0x00000a20 +#define NV20TCL_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV20TCL_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV20TCL_VIEWPORT_SCALE0_W 0x00000a2c +#define NV20TCL_POINT_PARAMETER(x) (0x00000a30+((x)*4)) +#define NV20TCL_POINT_PARAMETER__SIZE 0x00000007 +#define NV20TCL_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) +#define NV20TCL_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV20TCL_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) +#define NV20TCL_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV20TCL_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) +#define NV20TCL_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV20TCL_RC_IN_RGB(x) (0x00000ac0+((x)*4)) +#define NV20TCL_RC_IN_RGB__SIZE 0x00000008 +#define NV20TCL_VIEWPORT_SCALE1_X 0x00000af0 +#define NV20TCL_VIEWPORT_SCALE1_Y 0x00000af4 +#define NV20TCL_VIEWPORT_SCALE1_Z 0x00000af8 +#define NV20TCL_VIEWPORT_SCALE1_W 0x00000afc +#define NV20TCL_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) +#define NV20TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV20TCL_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) +#define NV20TCL_VP_UPLOAD_CONST__SIZE 0x00000004 +#define NV20TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV20TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV20TCL_VERTEX_POS_3F_X 0x00001500 +#define NV20TCL_VERTEX_POS_3F_Y 0x00001504 +#define NV20TCL_VERTEX_POS_3F_Z 0x00001508 +#define NV20TCL_VERTEX_POS_4F_X 0x00001518 +#define NV20TCL_VERTEX_POS_4F_Y 0x0000151c +#define NV20TCL_VERTEX_POS_4F_Z 0x00001520 +#define NV20TCL_VERTEX_POS_3I_XY 0x00001528 +#define NV20TCL_VERTEX_POS_3I_XY_X_SHIFT 0 +#define NV20TCL_VERTEX_POS_3I_XY_X_MASK 0x0000ffff +#define NV20TCL_VERTEX_POS_3I_XY_Y_SHIFT 16 +#define NV20TCL_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 +#define NV20TCL_VERTEX_POS_3I_Z 0x0000152c +#define NV20TCL_VERTEX_POS_3I_Z_Z_SHIFT 0 +#define NV20TCL_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff +#define NV20TCL_VERTEX_NOR_3F_X 0x00001530 +#define NV20TCL_VERTEX_NOR_3F_Y 0x00001534 +#define NV20TCL_VERTEX_NOR_3F_Z 0x00001538 +#define NV20TCL_VERTEX_NOR_3I_XY 0x00001540 +#define NV20TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV20TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV20TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV20TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV20TCL_VERTEX_NOR_3I_Z 0x00001544 +#define NV20TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV20TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV20TCL_VERTEX_COL_4F_X 0x00001550 +#define NV20TCL_VERTEX_COL_4F_Y 0x00001554 +#define NV20TCL_VERTEX_COL_4F_Z 0x00001558 +#define NV20TCL_VERTEX_COL_4F_W 0x0000155c +#define NV20TCL_VERTEX_COL_3F_X 0x00001560 +#define NV20TCL_VERTEX_COL_3F_Y 0x00001564 +#define NV20TCL_VERTEX_COL_3F_Z 0x00001568 +#define NV20TCL_VERTEX_COL_4I 0x0000156c +#define NV20TCL_VERTEX_COL_4I_R_SHIFT 0 +#define NV20TCL_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV20TCL_VERTEX_COL_4I_G_SHIFT 8 +#define NV20TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV20TCL_VERTEX_COL_4I_B_SHIFT 16 +#define NV20TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV20TCL_VERTEX_COL_4I_A_SHIFT 24 +#define NV20TCL_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV20TCL_VERTEX_COL2_3F_X 0x00001580 +#define NV20TCL_VERTEX_COL2_3F_Y 0x00001584 +#define NV20TCL_VERTEX_COL2_3F_Z 0x00001588 +#define NV20TCL_VERTEX_COL2_4I 0x0000158c +#define NV20TCL_VERTEX_COL2_4I_R_SHIFT 0 +#define NV20TCL_VERTEX_COL2_4I_R_MASK 0x000000ff +#define NV20TCL_VERTEX_COL2_4I_G_SHIFT 8 +#define NV20TCL_VERTEX_COL2_4I_G_MASK 0x0000ff00 +#define NV20TCL_VERTEX_COL2_4I_B_SHIFT 16 +#define NV20TCL_VERTEX_COL2_4I_B_MASK 0x00ff0000 +#define NV20TCL_VERTEX_COL2_4I_A_SHIFT 24 +#define NV20TCL_VERTEX_COL2_4I_A_MASK 0xff000000 +#define NV20TCL_VERTEX_TX0_2F_S 0x00001590 +#define NV20TCL_VERTEX_TX0_2F_T 0x00001594 +#define NV20TCL_VERTEX_TX0_2I 0x00001598 +#define NV20TCL_VERTEX_TX0_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX0_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX0_4F_S 0x000015a0 +#define NV20TCL_VERTEX_TX0_4F_T 0x000015a4 +#define NV20TCL_VERTEX_TX0_4F_R 0x000015a8 +#define NV20TCL_VERTEX_TX0_4F_Q 0x000015ac +#define NV20TCL_VERTEX_TX0_4I_ST 0x000015b0 +#define NV20TCL_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX0_4I_RQ 0x000015b4 +#define NV20TCL_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX1_2F_S 0x000015b8 +#define NV20TCL_VERTEX_TX1_2F_T 0x000015bc +#define NV20TCL_VERTEX_TX1_2I 0x000015c0 +#define NV20TCL_VERTEX_TX1_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX1_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX1_4F_S 0x000015c8 +#define NV20TCL_VERTEX_TX1_4F_T 0x000015cc +#define NV20TCL_VERTEX_TX1_4F_R 0x000015d0 +#define NV20TCL_VERTEX_TX1_4F_Q 0x000015d4 +#define NV20TCL_VERTEX_TX1_4I_ST 0x000015d8 +#define NV20TCL_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX1_4I_RQ 0x000015dc +#define NV20TCL_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX2_2F_S 0x000015e0 +#define NV20TCL_VERTEX_TX2_2F_T 0x000015e4 +#define NV20TCL_VERTEX_TX2_2I 0x000015e8 +#define NV20TCL_VERTEX_TX2_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX2_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX2_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX2_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX2_4F_S 0x000015f0 +#define NV20TCL_VERTEX_TX2_4F_T 0x000015f4 +#define NV20TCL_VERTEX_TX2_4F_R 0x000015f8 +#define NV20TCL_VERTEX_TX2_4F_Q 0x000015fc +#define NV20TCL_VERTEX_TX2_4I_ST 0x00001600 +#define NV20TCL_VERTEX_TX2_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX2_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX2_4I_RQ 0x00001604 +#define NV20TCL_VERTEX_TX2_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX2_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX3_2F_S 0x00001608 +#define NV20TCL_VERTEX_TX3_2F_T 0x0000160c +#define NV20TCL_VERTEX_TX3_2I 0x00001610 +#define NV20TCL_VERTEX_TX3_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX3_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX3_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX3_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX3_4F_S 0x00001620 +#define NV20TCL_VERTEX_TX3_4F_T 0x00001624 +#define NV20TCL_VERTEX_TX3_4F_R 0x00001628 +#define NV20TCL_VERTEX_TX3_4F_Q 0x0000162c +#define NV20TCL_VERTEX_TX3_4I_ST 0x00001630 +#define NV20TCL_VERTEX_TX3_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX3_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX3_4I_RQ 0x00001634 +#define NV20TCL_VERTEX_TX3_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX3_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_EDGEFLAG_ENABLE 0x000016bc +#define NV20TCL_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) +#define NV20TCL_VERTEX_ATTR_OFFSET__SIZE 0x00000010 +#define NV20TCL_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) +#define NV20TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV20TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV20TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV20TCL_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 +#define NV20TCL_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 +#define NV20TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV20TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV20TCL_COLOR_LOGIC_OP_ENABLE 0x000017bc +#define NV20TCL_COLOR_LOGIC_OP_OP 0x000017c0 +#define NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 +#define NV20TCL_TX_SHADER_CULL_MODE 0x000017f8 +#define NV20TCL_VERTEX_BEGIN_END 0x000017fc +#define NV20TCL_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 +#define NV20TCL_VERTEX_ARRAY_DATA 0x00001818 +#define NV20TCL_TX_OFFSET(x) (0x00001b00+((x)*64)) +#define NV20TCL_TX_OFFSET__SIZE 0x00000004 +#define NV20TCL_TX_FORMAT(x) (0x00001b04+((x)*64)) +#define NV20TCL_TX_FORMAT__SIZE 0x00000004 +#define NV20TCL_TX_WRAP(x) (0x00001b08+((x)*64)) +#define NV20TCL_TX_WRAP__SIZE 0x00000004 +#define NV20TCL_TX_ENABLE(x) (0x00001b0c+((x)*64)) +#define NV20TCL_TX_ENABLE__SIZE 0x00000004 +#define NV20TCL_TX_FILTER(x) (0x00001b14+((x)*64)) +#define NV20TCL_TX_FILTER__SIZE 0x00000004 +#define NV20TCL_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) +#define NV20TCL_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV20TCL_SCISSOR_HORIZ 0x00001c30 +#define NV20TCL_SCISSOR_VERT 0x00001c50 +#define NV20TCL_MULTISAMPLE_CONTROL 0x00001d7c +#define NV20TCL_CLEAR_VALUE 0x00001d90 +#define NV20TCL_CLEAR_BUFFERS 0x00001d94 +#define NV20TCL_RC_COLOR0 0x00001e20 +#define NV20TCL_RC_COLOR1 0x00001e24 +#define NV20TCL_RC_OUT_RGB(x) (0x00001e40+((x)*4)) +#define NV20TCL_RC_OUT_RGB__SIZE 0x00000008 +#define NV20TCL_RC_ENABLE 0x00001e60 +#define NV20TCL_TX_SHADER_OP 0x00001e70 +#define NV20TCL_VP_UPLOAD_CONST_ID 0x00001ea4 + + +#define NV17TCL 0x00000099 + +#define NV17TCL_DMA_IN_MEMORY4 0x000001ac +#define NV17TCL_DMA_IN_MEMORY5 0x000001b0 +#define NV17TCL_COLOR_MASK_ENABLE 0x000002bc +#define NV17TCL_LMA_DEPTH_BUFFER_PITCH 0x00000d5c +#define NV17TCL_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 +#define NV17TCL_LMA_DEPTH_FILL_VALUE 0x00000d68 +#define NV17TCL_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c +#define NV17TCL_LMA_DEPTH_ENABLE 0x00001658 #define NV20_SWIZZLED_SURFACE 0x0000009e @@ -3663,7 +3664,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. -#define NV30_TCL_PRIMITIVE_3D 0x00000397 +#define NV30TCL 0x00000397 @@ -3671,649 +3672,883 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. -#define NV36_TCL_PRIMITIVE_3D 0x00000497 - - - -#define NV25_TCL_PRIMITIVE_3D 0x00000597 - -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x0000019c -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a0 -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001ac -#define NV25_TCL_PRIMITIVE_3D_DMA_IN_MEMORY9 0x000001b0 - - -#define NV34_TCL_PRIMITIVE_3D 0x00000697 - -#define NV34_TCL_PRIMITIVE_3D_NOP 0x00000100 -#define NV34_TCL_PRIMITIVE_3D_NOTIFY 0x00000104 -#define NV34_TCL_PRIMITIVE_3D_DMA_NOTIFY 0x00000180 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY0 0x00000184 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY1 0x00000188 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY2 0x0000018c -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY3 0x00000194 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY4 0x00000198 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY5 0x000001a4 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY6 0x000001a8 -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY7 0x000001ac -#define NV34_TCL_PRIMITIVE_3D_DMA_IN_MEMORY8 0x000001b0 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_VERT 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_FORMAT 0x00000208 -#define NV34_TCL_PRIMITIVE_3D_PITCH 0x0000020c -#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_PITCH_COLOR_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_PITCH_ZETA_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_COLOR_OFFSET 0x00000210 -#define NV34_TCL_PRIMITIVE_3D_ZETA_OFFSET 0x00000214 -#define NV34_TCL_PRIMITIVE_3D_COLOR1_OFFSET 0x00000218 -#define NV34_TCL_PRIMITIVE_3D_COLOR1_PITCH 0x0000021c -#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_PITCH 0x0000022c -#define NV34_TCL_PRIMITIVE_3D_LMA_DEPTH_OFFSET 0x00000230 -#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_UNK0250(x) (0x00000250+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_UNK0250__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_DITHER_ENABLE 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF 0x0000030c -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE 0x00000310 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC 0x00000314 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST 0x00000318 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_COLOR 0x0000031c -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION 0x00000320 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MIN 0x00008007 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_MAX 0x00008008 -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV34_TCL_PRIMITIVE_3D_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK 0x00000324 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK 0xff000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE 0x00000328 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK 0x0000032c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC 0x00000330 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF 0x00000334 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK 0x00000338 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL 0x0000033c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL 0x00000340 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS 0x00000344 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE 0x00000348 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK 0x0000034c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC 0x00000350 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF 0x00000354 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK 0x00000358 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL 0x0000035c -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL 0x00000360 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS 0x00000364 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV34_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL 0x00000368 -#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_FLAT 0x00001d00 -#define NV34_TCL_PRIMITIVE_3D_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV34_TCL_PRIMITIVE_3D_FOG_ENABLE 0x0000036c -#define NV34_TCL_PRIMITIVE_3D_FOG_COLOR 0x00000370 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE 0x00000374 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP 0x00000378 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND 0x00001501 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY 0x00001503 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOOP 0x00001505 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_XOR 0x00001506 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR 0x00001507 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NOR 0x00001508 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_INVERT 0x0000150a -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_NAND 0x0000150e -#define NV34_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP_SET 0x0000150f -#define NV34_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE 0x0000037c -#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR 0x00000394 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR 0x00000398 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R 0x000003a0 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G 0x000003a4 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B 0x000003a8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A 0x000003b4 -#define NV34_TCL_PRIMITIVE_3D_LINE_WIDTH 0x000003b8 -#define NV34_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE 0x000003bc -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX(x) (0x000006c0+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX0_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX(x) (0x00000700+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX1_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX(x) (0x00000740+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX2_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX(x) (0x00000780+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_TX3_MATRIX__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ 0x000008c0 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_HORIZ_W_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT 0x000008c4 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_Y_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_SCISSOR_VERT_H_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_FOG_COORD_DIST 0x000008c8 -#define NV34_TCL_PRIMITIVE_3D_FOG_MODE 0x000008cc -#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT 0x000008d0 -#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR 0x000008d4 -#define NV34_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC 0x000008d8 -#define NV34_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM 0x000008e4 -#define NV34_TCL_PRIMITIVE_3D_RC_COLOR0 0x000008ec -#define NV34_TCL_PRIMITIVE_3D_RC_COLOR1 0x000008f0 -#define NV34_TCL_PRIMITIVE_3D_RC_FINAL0 0x000008f4 -#define NV34_TCL_PRIMITIVE_3D_RC_FINAL1 0x000008f8 -#define NV34_TCL_PRIMITIVE_3D_RC_ENABLE 0x000008fc -#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA(x) (0x00000900+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_IN_ALPHA__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB(x) (0x00000904+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_IN_RGB__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_ALPHA__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB(x) (0x00000914+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_RC_OUT_RGB__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x00000a10 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x00000a14 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x00000a18 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_X 0x00000a20 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Y 0x00000a24 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_Z 0x00000a28 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE0_W 0x00000a2c -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_X 0x00000a30 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Y 0x00000a34 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_Z 0x00000a38 -#define NV34_TCL_PRIMITIVE_3D_VIEWPORT_SCALE1_W 0x00000a3c -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC 0x00000a6c -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NEVER 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LESS 0x00000201 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_EQUAL 0x00000202 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GREATER 0x00000204 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE 0x00000a70 -#define NV34_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE 0x00000a74 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR 0x00000a78 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS 0x00000a7c -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY 0x00000a90 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z 0x00000a94 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_A__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_B__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_C__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_CLIP_PLANE_D__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_X__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) -#define NV34_TCL_PRIMITIVE_3D_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_X__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Y__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_3F_Z__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT__SIZE 0x0000000c -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000017a0 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000017a4 -#define NV34_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000017a8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R 0x000017b0 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G 0x000017b4 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B 0x000017b8 -#define NV34_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A 0x000017c0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END 0x00001808 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_STOP 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POINTS 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINES 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLES 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUADS 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV34_TCL_PRIMITIVE_3D_VERTEX_DATA 0x00001818 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT 0x00001828 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK 0x0000182c -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE 0x00001830 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT 0x00000404 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_BACK 0x00000405 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE 0x00001834 -#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CW 0x00000900 -#define NV34_TCL_PRIMITIVE_3D_FRONT_FACE_CCW 0x00000901 -#define NV34_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE 0x00001838 -#define NV34_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE 0x0000183c -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_X__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2F_Y__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I(x) (0x00001900+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_Y_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_Y_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_2I_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I__SIZE 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY 0x00001980 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW 0x00001984 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_S_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_ST_T_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET(x) (0x00001a00+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_OFFSET__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT(x) (0x00001a04+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_SHIFT 4 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_COMPONENTS_MASK 0x000000f0 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8 0x00000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8 0x00000100 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4 0x00000400 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8 0x00000600 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_X8R8G8B8 0x00000700 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8 0x00000b00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1 0x00000c00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3 0x00000e00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5 0x00000f00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT 0x00001000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT 0x00001200 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT 0x00001300 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8 0x00001a00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT2 0x00001b00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT 0x00001d00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT 0x00002000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16 0x00003200 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A16_RECT 0x00003500 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00004a00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00004b00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00004c00 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT (1 << 12) -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_RECT (1 << 14) -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_SHIFT 20 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_SHIFT 28 -#define NV34_TCL_PRIMITIVE_3D_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP(x) (0x00001a08+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_REPEAT 0x00000001 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_S_CLAMP 0x00000005 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_REPEAT 0x00000100 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_MIRRORED_REPEAT 0x00000200 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_EDGE 0x00000300 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP_TO_BORDER 0x00000400 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_T_CLAMP 0x00000500 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_REPEAT 0x00010000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_MIRRORED_REPEAT 0x00020000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_EDGE 0x00030000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP_TO_BORDER 0x00040000 -#define NV34_TCL_PRIMITIVE_3D_TX_WRAP_R_CLAMP 0x00050000 -#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE(x) (0x00001a0c+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE (1 << 30) -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE(x) (0x00001a10+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_SWIZZLE_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_SWIZZLE_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_RECT_PITCH_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_SWIZZLE_RECT_PITCH_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER(x) (0x00001a14+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_MASK 0x000f0000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST 0x00010000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR 0x00020000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00030000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00040000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00050000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00060000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_MASK 0x0f000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST 0x01000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR 0x02000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x03000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x04000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x05000000 -#define NV34_TCL_PRIMITIVE_3D_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x06000000 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_H_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_W_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR__SIZE 0x00000004 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_MASK 0x000000ff -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_SHIFT 8 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_MASK 0x0000ff00 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_MASK 0x00ff0000 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_SHIFT 24 -#define NV34_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_MASK 0xff000000 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_X__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Y__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_Z__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) -#define NV34_TCL_PRIMITIVE_3D_VERTEX_ATTR_4F_W__SIZE 0x00000010 -#define NV34_TCL_PRIMITIVE_3D_CLEAR_DEPTH_VALUE 0x00001d8c -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE 0x00001db4 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN 0x00001db8 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 -#define NV34_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 -#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34_TCL_PRIMITIVE_3D_VERTEX_FOG_1F 0x00001e54 -#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS__SIZE 0x00000008 -#define NV34_TCL_PRIMITIVE_3D_POINT_SIZE 0x00001ee0 -#define NV34_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE 0x00001ee4 -#define NV34_TCL_PRIMITIVE_3D_POINT_SPRITE 0x00001ee8 -#define NV34_TCL_PRIMITIVE_3D_UNK1f80(x) (0x00001f80+((x)*4)) -#define NV34_TCL_PRIMITIVE_3D_UNK1f80__SIZE 0x00000010 +#define NV36TCL 0x00000497 + + + +#define NV25TCL 0x00000597 + +#define NV25TCL_DMA_IN_MEMORY4 0x0000019c +#define NV25TCL_DMA_IN_MEMORY5 0x000001a0 +#define NV25TCL_DMA_IN_MEMORY8 0x000001ac +#define NV25TCL_DMA_IN_MEMORY9 0x000001b0 + + +#define NV34TCL 0x00000697 + +#define NV34TCL_NOP 0x00000100 +#define NV34TCL_NOTIFY 0x00000104 +#define NV34TCL_DMA_NOTIFY 0x00000180 +#define NV34TCL_DMA_TEXTURE0 0x00000184 +#define NV34TCL_DMA_TEXTURE1 0x00000188 +#define NV34TCL_DMA_COLOR1 0x0000018c +#define NV34TCL_DMA_COLOR0 0x00000194 +#define NV34TCL_DMA_ZETA 0x00000198 +#define NV34TCL_DMA_VTXBUF0 0x0000019c +#define NV34TCL_DMA_VTXBUF1 0x000001a0 +#define NV34TCL_DMA_FENCE 0x000001a4 +#define NV34TCL_DMA_QUERY 0x000001a8 +#define NV34TCL_DMA_IN_MEMORY7 0x000001ac +#define NV34TCL_DMA_IN_MEMORY8 0x000001b0 +#define NV34TCL_DMA_COLOR2 0x000001b4 +#define NV34TCL_DMA_COLOR3 0x000001b8 +#define NV34TCL_VIEWPORT_HORIZ 0x00000200 +#define NV34TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV34TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV34TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_VERT 0x00000204 +#define NV34TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV34TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV34TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV34TCL_RT_FORMAT 0x00000208 +#define NV34TCL_RT_FORMAT_TYPE_SHIFT 8 +#define NV34TCL_RT_FORMAT_TYPE_MASK 0x00000f00 +#define NV34TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 +#define NV34TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 +#define NV34TCL_RT_FORMAT_ZETA_SHIFT 5 +#define NV34TCL_RT_FORMAT_ZETA_MASK 0x000000e0 +#define NV34TCL_RT_FORMAT_ZETA_Z16 0x00000020 +#define NV34TCL_RT_FORMAT_ZETA_Z24S8 0x00000040 +#define NV34TCL_RT_FORMAT_COLOR_SHIFT 0 +#define NV34TCL_RT_FORMAT_COLOR_MASK 0x0000001f +#define NV34TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV34TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 +#define NV34TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV34TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV34TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f +#define NV34TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 +#define NV34TCL_COLOR0_PITCH 0x0000020c +#define NV34TCL_COLOR0_PITCH_COLOR0_SHIFT 0 +#define NV34TCL_COLOR0_PITCH_COLOR0_MASK 0x0000ffff +#define NV34TCL_COLOR0_PITCH_ZETA_SHIFT 16 +#define NV34TCL_COLOR0_PITCH_ZETA_MASK 0xffff0000 +#define NV34TCL_COLOR0_OFFSET 0x00000210 +#define NV34TCL_ZETA_OFFSET 0x00000214 +#define NV34TCL_COLOR1_OFFSET 0x00000218 +#define NV34TCL_COLOR1_PITCH 0x0000021c +#define NV34TCL_RT_ENABLE 0x00000220 +#define NV34TCL_RT_ENABLE_MRT (1 << 4) +#define NV34TCL_RT_ENABLE_COLOR3 (1 << 3) +#define NV34TCL_RT_ENABLE_COLOR2 (1 << 2) +#define NV34TCL_RT_ENABLE_COLOR1 (1 << 1) +#define NV34TCL_RT_ENABLE_COLOR0 (1 << 0) +#define NV34TCL_ZETA_PITCH 0x0000022c +#define NV34TCL_LMA_DEPTH_OFFSET 0x00000230 +#define NV34TCL_TX_UNITS_ENABLE 0x0000023c +#define NV34TCL_TX_UNITS_ENABLE_TX0 (1 << 0) +#define NV34TCL_TX_UNITS_ENABLE_TX1 (1 << 1) +#define NV34TCL_TX_UNITS_ENABLE_TX2 (1 << 2) +#define NV34TCL_TX_UNITS_ENABLE_TX3 (1 << 3) +#define NV34TCL_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) +#define NV34TCL_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV34TCL_UNK0250(x) (0x00000250+((x)*4)) +#define NV34TCL_UNK0250__SIZE 0x00000004 +#define NV34TCL_VIEWPORT_TX_ORIGIN 0x000002b8 +#define NV34TCL_VIEWPORT_TX_ORIGIN_X_SHIFT 0 +#define NV34TCL_VIEWPORT_TX_ORIGIN_X_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_TX_ORIGIN_Y_SHIFT 16 +#define NV34TCL_VIEWPORT_TX_ORIGIN_Y_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV34TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV34TCL_VIEWPORT_CLIP_HORIZ_L_SHIFT 0 +#define NV34TCL_VIEWPORT_CLIP_HORIZ_L_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_CLIP_HORIZ_R_SHIFT 16 +#define NV34TCL_VIEWPORT_CLIP_HORIZ_R_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV34TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV34TCL_VIEWPORT_CLIP_VERT_T_SHIFT 0 +#define NV34TCL_VIEWPORT_CLIP_VERT_T_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_CLIP_VERT_D_SHIFT 16 +#define NV34TCL_VIEWPORT_CLIP_VERT_D_MASK 0xffff0000 +#define NV34TCL_COLOR2_PITCH 0x00000280 +#define NV34TCL_COLOR3_PITCH 0x00000284 +#define NV34TCL_COLOR2_OFFSET 0x00000288 +#define NV34TCL_COLOR3_OFFSET 0x0000028c +#define NV34TCL_DITHER_ENABLE 0x00000300 +#define NV34TCL_ALPHA_FUNC_ENABLE 0x00000304 +#define NV34TCL_ALPHA_FUNC_FUNC 0x00000308 +#define NV34TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV34TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV34TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV34TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34TCL_ALPHA_FUNC_REF 0x0000030c +#define NV34TCL_BLEND_FUNC_ENABLE 0x00000310 +#define NV34TCL_BLEND_FUNC_SRC 0x00000314 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV34TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV34TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV34TCL_BLEND_FUNC_DST 0x00000318 +#define NV34TCL_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV34TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV34TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV34TCL_BLEND_FUNC_COLOR 0x0000031c +#define NV34TCL_BLEND_FUNC_EQUATION 0x00000320 +#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 +#define NV34TCL_BLEND_FUNC_EQUATION_MIN 0x00008007 +#define NV34TCL_BLEND_FUNC_EQUATION_MAX 0x00008008 +#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV34TCL_COLOR_MASK 0x00000324 +#define NV34TCL_COLOR_MASK_B_SHIFT 0 +#define NV34TCL_COLOR_MASK_B_MASK 0x000000ff +#define NV34TCL_COLOR_MASK_G_SHIFT 8 +#define NV34TCL_COLOR_MASK_G_MASK 0x0000ff00 +#define NV34TCL_COLOR_MASK_R_SHIFT 16 +#define NV34TCL_COLOR_MASK_R_MASK 0x00ff0000 +#define NV34TCL_COLOR_MASK_A_SHIFT 24 +#define NV34TCL_COLOR_MASK_A_MASK 0xff000000 +#define NV34TCL_STENCIL_BACK_ENABLE 0x00000328 +#define NV34TCL_STENCIL_BACK_MASK 0x0000032c +#define NV34TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34TCL_STENCIL_BACK_FUNC_REF 0x00000334 +#define NV34TCL_STENCIL_BACK_FUNC_MASK 0x00000338 +#define NV34TCL_STENCIL_BACK_OP_FAIL 0x0000033c +#define NV34TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_BACK_OP_ZPASS 0x00000344 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV34TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_FRONT_ENABLE 0x00000348 +#define NV34TCL_STENCIL_FRONT_MASK 0x0000034c +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34TCL_STENCIL_FRONT_FUNC_REF 0x00000354 +#define NV34TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 +#define NV34TCL_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV34TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34TCL_SHADE_MODEL 0x00000368 +#define NV34TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV34TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV34TCL_FOG_ENABLE 0x0000036c +#define NV34TCL_FOG_COLOR 0x00000370 +#define NV34TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV34TCL_COLOR_LOGIC_OP_OP 0x00000378 +#define NV34TCL_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV34TCL_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV34TCL_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV34TCL_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV34TCL_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV34TCL_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV34TCL_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV34TCL_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV34TCL_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV34TCL_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV34TCL_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV34TCL_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV34TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV34TCL_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV34TCL_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV34TCL_COLOR_LOGIC_OP_OP_SET 0x0000150f +#define NV34TCL_NORMALIZE_ENABLE 0x0000037c +#define NV34TCL_COLOR_MATERIAL 0x00000390 +#define NV34TCL_COLOR_MATERIAL_FRONT_EMISSION_ENABLE (1 << 0) +#define NV34TCL_COLOR_MATERIAL_FRONT_AMBIENT_ENABLE (1 << 2) +#define NV34TCL_COLOR_MATERIAL_FRONT_DIFFUSE_ENABLE (1 << 4) +#define NV34TCL_COLOR_MATERIAL_FRONT_SPECULAR_ENABLE (1 << 6) +#define NV34TCL_COLOR_MATERIAL_BACK_EMISSION_ENABLE (1 << 8) +#define NV34TCL_COLOR_MATERIAL_BACK_AMBIENT_ENABLE (1 << 10) +#define NV34TCL_COLOR_MATERIAL_BACK_DIFFUSE_ENABLE (1 << 12) +#define NV34TCL_COLOR_MATERIAL_BACK_SPECULAR_ENABLE (1 << 14) +#define NV34TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV34TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV34TCL_COLOR_MATERIAL_FRONT_R 0x000003a0 +#define NV34TCL_COLOR_MATERIAL_FRONT_G 0x000003a4 +#define NV34TCL_COLOR_MATERIAL_FRONT_B 0x000003a8 +#define NV34TCL_COLOR_MATERIAL_FRONT_A 0x000003b4 +#define NV34TCL_LINE_WIDTH 0x000003b8 +#define NV34TCL_LINE_SMOOTH_ENABLE 0x000003bc +#define NV34TCL_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) +#define NV34TCL_CLIP_PLANE_ENABLE__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_ENABLE_FALSE 0x00000000 +#define NV34TCL_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 +#define NV34TCL_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 +#define NV34TCL_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV34TCL_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV34TCL_INVERSE_MODELVIEW_MATRIX(x) (0x00000580+((x)*4)) +#define NV34TCL_INVERSE_MODELVIEW_MATRIX__SIZE 0x0000000c +#define NV34TCL_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV34TCL_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV34TCL_TX0_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV34TCL_TX1_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV34TCL_TX2_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV34TCL_TX3_MATRIX__SIZE 0x00000010 +#define NV34TCL_SCISSOR_HORIZ 0x000008c0 +#define NV34TCL_SCISSOR_HORIZ_X_SHIFT 0 +#define NV34TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_SCISSOR_HORIZ_W_SHIFT 16 +#define NV34TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_SCISSOR_VERT 0x000008c4 +#define NV34TCL_SCISSOR_VERT_Y_SHIFT 0 +#define NV34TCL_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV34TCL_SCISSOR_VERT_H_SHIFT 16 +#define NV34TCL_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV34TCL_FOG_COORD_DIST 0x000008c8 +#define NV34TCL_FOG_MODE 0x000008cc +#define NV34TCL_FOG_MODE_EXP 0x00000800 +#define NV34TCL_FOG_MODE_EXP_2 0x00000802 +#define NV34TCL_FOG_MODE_EXP2 0x00000803 +#define NV34TCL_FOG_MODE_LINEAR 0x00000804 +#define NV34TCL_FOG_MODE_LINEAR_2 0x00002601 +#define NV34TCL_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV34TCL_FOG_EQUATION_LINEAR 0x000008d4 +#define NV34TCL_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV34TCL_FP_ACTIVE_PROGRAM 0x000008e4 +#define NV34TCL_RC_COLOR0 0x000008ec +#define NV34TCL_RC_COLOR0_B_SHIFT 0 +#define NV34TCL_RC_COLOR0_B_MASK 0x000000ff +#define NV34TCL_RC_COLOR0_G_SHIFT 8 +#define NV34TCL_RC_COLOR0_G_MASK 0x0000ff00 +#define NV34TCL_RC_COLOR0_R_SHIFT 16 +#define NV34TCL_RC_COLOR0_R_MASK 0x00ff0000 +#define NV34TCL_RC_COLOR0_A_SHIFT 24 +#define NV34TCL_RC_COLOR0_A_MASK 0xff000000 +#define NV34TCL_RC_COLOR1 0x000008f0 +#define NV34TCL_RC_COLOR1_B_SHIFT 0 +#define NV34TCL_RC_COLOR1_B_MASK 0x000000ff +#define NV34TCL_RC_COLOR1_G_SHIFT 8 +#define NV34TCL_RC_COLOR1_G_MASK 0x0000ff00 +#define NV34TCL_RC_COLOR1_R_SHIFT 16 +#define NV34TCL_RC_COLOR1_R_MASK 0x00ff0000 +#define NV34TCL_RC_COLOR1_A_SHIFT 24 +#define NV34TCL_RC_COLOR1_A_MASK 0xff000000 +#define NV34TCL_RC_FINAL0 0x000008f4 +#define NV34TCL_RC_FINAL1 0x000008f8 +#define NV34TCL_RC_ENABLE 0x000008fc +#define NV34TCL_RC_IN_ALPHA(x) (0x00000900+((x)*32)) +#define NV34TCL_RC_IN_ALPHA__SIZE 0x00000008 +#define NV34TCL_RC_IN_RGB(x) (0x00000904+((x)*32)) +#define NV34TCL_RC_IN_RGB__SIZE 0x00000008 +#define NV34TCL_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) +#define NV34TCL_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV34TCL_RC_CONSTANT_COLOR0_B_SHIFT 0 +#define NV34TCL_RC_CONSTANT_COLOR0_B_MASK 0x000000ff +#define NV34TCL_RC_CONSTANT_COLOR0_G_SHIFT 8 +#define NV34TCL_RC_CONSTANT_COLOR0_G_MASK 0x0000ff00 +#define NV34TCL_RC_CONSTANT_COLOR0_R_SHIFT 16 +#define NV34TCL_RC_CONSTANT_COLOR0_R_MASK 0x00ff0000 +#define NV34TCL_RC_CONSTANT_COLOR0_A_SHIFT 24 +#define NV34TCL_RC_CONSTANT_COLOR0_A_MASK 0xff000000 +#define NV34TCL_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) +#define NV34TCL_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV34TCL_RC_CONSTANT_COLOR1_B_SHIFT 0 +#define NV34TCL_RC_CONSTANT_COLOR1_B_MASK 0x000000ff +#define NV34TCL_RC_CONSTANT_COLOR1_G_SHIFT 8 +#define NV34TCL_RC_CONSTANT_COLOR1_G_MASK 0x0000ff00 +#define NV34TCL_RC_CONSTANT_COLOR1_R_SHIFT 16 +#define NV34TCL_RC_CONSTANT_COLOR1_R_MASK 0x00ff0000 +#define NV34TCL_RC_CONSTANT_COLOR1_A_SHIFT 24 +#define NV34TCL_RC_CONSTANT_COLOR1_A_MASK 0xff000000 +#define NV34TCL_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) +#define NV34TCL_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV34TCL_RC_OUT_RGB(x) (0x00000914+((x)*32)) +#define NV34TCL_RC_OUT_RGB__SIZE 0x00000008 +#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x00000a10 +#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x00000a14 +#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x00000a18 +#define NV34TCL_VIEWPORT_SCALE0_X 0x00000a20 +#define NV34TCL_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV34TCL_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV34TCL_VIEWPORT_SCALE0_W 0x00000a2c +#define NV34TCL_VIEWPORT_SCALE1_X 0x00000a30 +#define NV34TCL_VIEWPORT_SCALE1_Y 0x00000a34 +#define NV34TCL_VIEWPORT_SCALE1_Z 0x00000a38 +#define NV34TCL_VIEWPORT_SCALE1_W 0x00000a3c +#define NV34TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV34TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV34TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV34TCL_DEPTH_FUNC 0x00000a6c +#define NV34TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV34TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV34TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV34TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV34TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV34TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV34TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV34TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV34TCL_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV34TCL_DEPTH_TEST_ENABLE 0x00000a74 +#define NV34TCL_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV34TCL_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV34TCL_VERTEX_NOR_3I_XY 0x00000a90 +#define NV34TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV34TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV34TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV34TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV34TCL_VERTEX_NOR_3I_Z 0x00000a94 +#define NV34TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV34TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV34TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV34TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV34TCL_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) +#define NV34TCL_CLIP_PLANE_A__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) +#define NV34TCL_CLIP_PLANE_B__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) +#define NV34TCL_CLIP_PLANE_C__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) +#define NV34TCL_CLIP_PLANE_D__SIZE 0x00000020 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) +#define NV34TCL_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) +#define NV34TCL_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) +#define NV34TCL_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) +#define NV34TCL_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) +#define NV34TCL_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) +#define NV34TCL_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) +#define NV34TCL_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) +#define NV34TCL_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) +#define NV34TCL_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV34TCL_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) +#define NV34TCL_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) +#define NV34TCL_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) +#define NV34TCL_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) +#define NV34TCL_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV34TCL_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) +#define NV34TCL_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV34TCL_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) +#define NV34TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV34TCL_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) +#define NV34TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34TCL_FP_REG_CONTROL 0x00001450 +#define NV34TCL_FP_REG_CONTROL_USED_REGS_SHIFT 16 +#define NV34TCL_FP_REG_CONTROL_USED_REGS_MASK 0xffff0000 +#define NV34TCL_VP_CLIP_PLANES_ENABLE 0x00001478 +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE0 (1 << 1) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE1 (1 << 5) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE2 (1 << 9) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE3 (1 << 13) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE4 (1 << 17) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE5 (1 << 21) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE6 (1 << 25) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE7 (1 << 29) +#define NV34TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV34TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV34TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV34TCL_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV34TCL_VERTEX_ATTR_3F_X__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV34TCL_VERTEX_ATTR_3F_Y__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV34TCL_VERTEX_ATTR_3F_Z__SIZE 0x00000010 +#define NV34TCL_VP_CLIP_PLANE_A(x) (0x00001600+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_A__SIZE 0x00000006 +#define NV34TCL_VP_CLIP_PLANE_B(x) (0x00001604+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_B__SIZE 0x00000006 +#define NV34TCL_VP_CLIP_PLANE_C(x) (0x00001608+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_C__SIZE 0x00000006 +#define NV34TCL_VP_CLIP_PLANE_D(x) (0x0000160c+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_D__SIZE 0x00000006 +#define NV34TCL_VERTEX_BUFFER_ADDRESS(x) (0x00001680+((x)*4)) +#define NV34TCL_VERTEX_BUFFER_ADDRESS__SIZE 0x00000010 +#define NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1 (1 << 31) +#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_SHIFT 0 +#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_MASK 0x0fffffff +#define NV34TCL_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) +#define NV34TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT 0x00000002 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE 0x00000004 +#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT 4 +#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_MASK 0x000000f0 +#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000017a0 +#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000017a4 +#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000017a8 +#define NV34TCL_COLOR_MATERIAL_BACK_R 0x000017b0 +#define NV34TCL_COLOR_MATERIAL_BACK_G 0x000017b4 +#define NV34TCL_COLOR_MATERIAL_BACK_B 0x000017b8 +#define NV34TCL_COLOR_MATERIAL_BACK_A 0x000017c0 +#define NV34TCL_QUERY_RESET 0x000017c8 +#define NV34TCL_QUERY_UNK17CC 0x000017cc +#define NV34TCL_QUERY_GET 0x00001800 +#define NV34TCL_QUERY_GET_UNK24_SHIFT 24 +#define NV34TCL_QUERY_GET_UNK24_MASK 0xff000000 +#define NV34TCL_QUERY_GET_OFFSET_SHIFT 0 +#define NV34TCL_QUERY_GET_OFFSET_MASK 0x00ffffff +#define NV34TCL_VERTEX_BEGIN_END 0x00001808 +#define NV34TCL_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV34TCL_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV34TCL_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV34TCL_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV34TCL_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV34TCL_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV34TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV34TCL_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV34TCL_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV34TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV34TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV34TCL_VB_ELEMENT_U16 0x0000180c +#define NV34TCL_VB_ELEMENT_U16_I0_SHIFT 0 +#define NV34TCL_VB_ELEMENT_U16_I0_MASK 0x0000ffff +#define NV34TCL_VB_ELEMENT_U16_I1_SHIFT 16 +#define NV34TCL_VB_ELEMENT_U16_I1_MASK 0xffff0000 +#define NV34TCL_VB_ELEMENT_U32 0x00001810 +#define NV34TCL_VB_VERTEX_BATCH 0x00001814 +#define NV34TCL_VB_VERTEX_BATCH_OFFSET_SHIFT 0 +#define NV34TCL_VB_VERTEX_BATCH_OFFSET_MASK 0x00ffffff +#define NV34TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 +#define NV34TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 +#define NV34TCL_VERTEX_DATA 0x00001818 +#define NV34TCL_POLYGON_MODE_FRONT 0x00001828 +#define NV34TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV34TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV34TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV34TCL_POLYGON_MODE_BACK 0x0000182c +#define NV34TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV34TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV34TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV34TCL_CULL_FACE 0x00001830 +#define NV34TCL_CULL_FACE_FRONT 0x00000404 +#define NV34TCL_CULL_FACE_BACK 0x00000405 +#define NV34TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV34TCL_FRONT_FACE 0x00001834 +#define NV34TCL_FRONT_FACE_CW 0x00000900 +#define NV34TCL_FRONT_FACE_CCW 0x00000901 +#define NV34TCL_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV34TCL_CULL_FACE_ENABLE 0x0000183c +#define NV34TCL_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV34TCL_VERTEX_ATTR_2F_X__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV34TCL_VERTEX_ATTR_2F_Y__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV34TCL_VERTEX_ATTR_2I__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_2I_Y_SHIFT 16 +#define NV34TCL_VERTEX_ATTR_2I_Y_MASK 0xffff0000 +#define NV34TCL_VERTEX_ATTR_2I_X_SHIFT 0 +#define NV34TCL_VERTEX_ATTR_2I_X_MASK 0x0000ffff +#define NV34TCL_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) +#define NV34TCL_VERTEX_COL_4I__SIZE 0x00000002 +#define NV34TCL_VERTEX_COL_4I_R_SHIFT 0 +#define NV34TCL_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV34TCL_VERTEX_COL_4I_G_SHIFT 8 +#define NV34TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV34TCL_VERTEX_COL_4I_B_SHIFT 16 +#define NV34TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV34TCL_VERTEX_COL_4I_A_SHIFT 24 +#define NV34TCL_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV34TCL_VERTEX_POS_4I_XY 0x00001980 +#define NV34TCL_VERTEX_POS_4I_XY_X_SHIFT 0 +#define NV34TCL_VERTEX_POS_4I_XY_X_MASK 0x0000ffff +#define NV34TCL_VERTEX_POS_4I_XY_Y_SHIFT 16 +#define NV34TCL_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 +#define NV34TCL_VERTEX_POS_4I_ZW 0x00001984 +#define NV34TCL_VERTEX_POS_4I_ZW_Z_SHIFT 0 +#define NV34TCL_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff +#define NV34TCL_VERTEX_POS_4I_ZW_W_SHIFT 16 +#define NV34TCL_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 +#define NV34TCL_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) +#define NV34TCL_VERTEX_TX_4I_ST__SIZE 0x00000004 +#define NV34TCL_VERTEX_TX_4I_ST_S_SHIFT 0 +#define NV34TCL_VERTEX_TX_4I_ST_S_MASK 0x0000ffff +#define NV34TCL_VERTEX_TX_4I_ST_T_SHIFT 16 +#define NV34TCL_VERTEX_TX_4I_ST_T_MASK 0xffff0000 +#define NV34TCL_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) +#define NV34TCL_VERTEX_TX_4I_RQ__SIZE 0x00000004 +#define NV34TCL_VERTEX_TX_4I_RQ_R_SHIFT 0 +#define NV34TCL_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff +#define NV34TCL_VERTEX_TX_4I_RQ_Q_SHIFT 16 +#define NV34TCL_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 +#define NV34TCL_TX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV34TCL_TX_OFFSET__SIZE 0x00000004 +#define NV34TCL_TX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV34TCL_TX_FORMAT__SIZE 0x00000004 +#define NV34TCL_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV34TCL_TX_FORMAT_COMPONENTS_SHIFT 4 +#define NV34TCL_TX_FORMAT_COMPONENTS_MASK 0x000000f0 +#define NV34TCL_TX_FORMAT_FORMAT_SHIFT 8 +#define NV34TCL_TX_FORMAT_FORMAT_MASK 0x0000ff00 +#define NV34TCL_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV34TCL_TX_FORMAT_FORMAT_A8 0x00000100 +#define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000200 +#define NV34TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000300 +#define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000400 +#define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000600 +#define NV34TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000700 +#define NV34TCL_TX_FORMAT_FORMAT_INDEX8 0x00000b00 +#define NV34TCL_TX_FORMAT_FORMAT_DXT1 0x00000c00 +#define NV34TCL_TX_FORMAT_FORMAT_DXT3 0x00000e00 +#define NV34TCL_TX_FORMAT_FORMAT_DXT5 0x00000f00 +#define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00001000 +#define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00001200 +#define NV34TCL_TX_FORMAT_FORMAT_L8_RECT 0x00001300 +#define NV34TCL_TX_FORMAT_FORMAT_A8L8 0x00001a00 +#define NV34TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00001b00 +#define NV34TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 +#define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00001d00 +#define NV34TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00002000 +#define NV34TCL_TX_FORMAT_FORMAT_A16 0x00003200 +#define NV34TCL_TX_FORMAT_FORMAT_A16_RECT 0x00003500 +#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00004a00 +#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00004b00 +#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00004c00 +#define NV34TCL_TX_FORMAT_NPOT (1 << 12) +#define NV34TCL_TX_FORMAT_RECT (1 << 14) +#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 +#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 +#define NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 20 +#define NV34TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 +#define NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV34TCL_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 +#define NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT 28 +#define NV34TCL_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 +#define NV34TCL_TX_WRAP(x) (0x00001a08+((x)*32)) +#define NV34TCL_TX_WRAP__SIZE 0x00000004 +#define NV34TCL_TX_WRAP_S_SHIFT 0 +#define NV34TCL_TX_WRAP_S_MASK 0x000000ff +#define NV34TCL_TX_WRAP_S_REPEAT 0x00000001 +#define NV34TCL_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV34TCL_TX_WRAP_S_CLAMP 0x00000005 +#define NV34TCL_TX_WRAP_T_SHIFT 8 +#define NV34TCL_TX_WRAP_T_MASK 0x0000ff00 +#define NV34TCL_TX_WRAP_T_REPEAT 0x00000100 +#define NV34TCL_TX_WRAP_T_MIRRORED_REPEAT 0x00000200 +#define NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE 0x00000300 +#define NV34TCL_TX_WRAP_T_CLAMP_TO_BORDER 0x00000400 +#define NV34TCL_TX_WRAP_T_CLAMP 0x00000500 +#define NV34TCL_TX_WRAP_R_SHIFT 16 +#define NV34TCL_TX_WRAP_R_MASK 0x00ff0000 +#define NV34TCL_TX_WRAP_R_REPEAT 0x00010000 +#define NV34TCL_TX_WRAP_R_MIRRORED_REPEAT 0x00020000 +#define NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE 0x00030000 +#define NV34TCL_TX_WRAP_R_CLAMP_TO_BORDER 0x00040000 +#define NV34TCL_TX_WRAP_R_CLAMP 0x00050000 +#define NV34TCL_TX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV34TCL_TX_ENABLE__SIZE 0x00000004 +#define NV34TCL_TX_ENABLE_ENABLE (1 << 30) +#define NV34TCL_TX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV34TCL_TX_SWIZZLE__SIZE 0x00000004 +#define NV34TCL_TX_SWIZZLE_S0_X_SHIFT 14 +#define NV34TCL_TX_SWIZZLE_S0_X_MASK 0x0000c000 +#define NV34TCL_TX_SWIZZLE_S0_X_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_X_ONE 0x00004000 +#define NV34TCL_TX_SWIZZLE_S0_X_S1 0x00008000 +#define NV34TCL_TX_SWIZZLE_S0_Y_SHIFT 12 +#define NV34TCL_TX_SWIZZLE_S0_Y_MASK 0x00003000 +#define NV34TCL_TX_SWIZZLE_S0_Y_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_Y_ONE 0x00001000 +#define NV34TCL_TX_SWIZZLE_S0_Y_S1 0x00002000 +#define NV34TCL_TX_SWIZZLE_S0_Z_SHIFT 10 +#define NV34TCL_TX_SWIZZLE_S0_Z_MASK 0x00000c00 +#define NV34TCL_TX_SWIZZLE_S0_Z_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_Z_ONE 0x00000400 +#define NV34TCL_TX_SWIZZLE_S0_Z_S1 0x00000800 +#define NV34TCL_TX_SWIZZLE_S0_W_SHIFT 8 +#define NV34TCL_TX_SWIZZLE_S0_W_MASK 0x00000300 +#define NV34TCL_TX_SWIZZLE_S0_W_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_W_ONE 0x00000100 +#define NV34TCL_TX_SWIZZLE_S0_W_S1 0x00000200 +#define NV34TCL_TX_SWIZZLE_S1_X_SHIFT 6 +#define NV34TCL_TX_SWIZZLE_S1_X_MASK 0x000000c0 +#define NV34TCL_TX_SWIZZLE_S1_X_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_X_Z 0x00000040 +#define NV34TCL_TX_SWIZZLE_S1_X_Y 0x00000080 +#define NV34TCL_TX_SWIZZLE_S1_X_X 0x000000c0 +#define NV34TCL_TX_SWIZZLE_S1_Y_SHIFT 4 +#define NV34TCL_TX_SWIZZLE_S1_Y_MASK 0x00000030 +#define NV34TCL_TX_SWIZZLE_S1_Y_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_Y_Z 0x00000010 +#define NV34TCL_TX_SWIZZLE_S1_Y_Y 0x00000020 +#define NV34TCL_TX_SWIZZLE_S1_Y_X 0x00000030 +#define NV34TCL_TX_SWIZZLE_S1_Z_SHIFT 2 +#define NV34TCL_TX_SWIZZLE_S1_Z_MASK 0x0000000c +#define NV34TCL_TX_SWIZZLE_S1_Z_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_Z_Z 0x00000004 +#define NV34TCL_TX_SWIZZLE_S1_Z_Y 0x00000008 +#define NV34TCL_TX_SWIZZLE_S1_Z_X 0x0000000c +#define NV34TCL_TX_SWIZZLE_S1_W_SHIFT 0 +#define NV34TCL_TX_SWIZZLE_S1_W_MASK 0x00000003 +#define NV34TCL_TX_SWIZZLE_S1_W_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_W_Z 0x00000001 +#define NV34TCL_TX_SWIZZLE_S1_W_Y 0x00000002 +#define NV34TCL_TX_SWIZZLE_S1_W_X 0x00000003 +#define NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT 16 +#define NV34TCL_TX_SWIZZLE_RECT_PITCH_MASK 0xffff0000 +#define NV34TCL_TX_FILTER(x) (0x00001a14+((x)*32)) +#define NV34TCL_TX_FILTER__SIZE 0x00000004 +#define NV34TCL_TX_FILTER_MINIFY_SHIFT 16 +#define NV34TCL_TX_FILTER_MINIFY_MASK 0x000f0000 +#define NV34TCL_TX_FILTER_MINIFY_NEAREST 0x00010000 +#define NV34TCL_TX_FILTER_MINIFY_LINEAR 0x00020000 +#define NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00030000 +#define NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00040000 +#define NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00050000 +#define NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00060000 +#define NV34TCL_TX_FILTER_MAGNIFY_SHIFT 24 +#define NV34TCL_TX_FILTER_MAGNIFY_MASK 0x0f000000 +#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST 0x01000000 +#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR 0x02000000 +#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV34TCL_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) +#define NV34TCL_TX_NPOT_SIZE__SIZE 0x00000004 +#define NV34TCL_TX_NPOT_SIZE_H_SHIFT 0 +#define NV34TCL_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV34TCL_TX_NPOT_SIZE_W_SHIFT 16 +#define NV34TCL_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV34TCL_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV34TCL_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV34TCL_TX_BORDER_COLOR_B_SHIFT 0 +#define NV34TCL_TX_BORDER_COLOR_B_MASK 0x000000ff +#define NV34TCL_TX_BORDER_COLOR_G_SHIFT 8 +#define NV34TCL_TX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV34TCL_TX_BORDER_COLOR_R_SHIFT 16 +#define NV34TCL_TX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV34TCL_TX_BORDER_COLOR_A_SHIFT 24 +#define NV34TCL_TX_BORDER_COLOR_A_MASK 0xff000000 +#define NV34TCL_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_X__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_Y__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_Z__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_W__SIZE 0x00000010 +#define NV34TCL_FP_CONTROL 0x00001d60 +#define NV34TCL_FP_CONTROL_USES_KIL (1 << 7) +#define NV34TCL_MULTISAMPLE_CONTROL 0x00001d7c +#define NV34TCL_CLEAR_DEPTH_VALUE 0x00001d8c +#define NV34TCL_CLEAR_COLOR_VALUE 0x00001d90 +#define NV34TCL_CLEAR_COLOR_VALUE_B_SHIFT 0 +#define NV34TCL_CLEAR_COLOR_VALUE_B_MASK 0x000000ff +#define NV34TCL_CLEAR_COLOR_VALUE_G_SHIFT 8 +#define NV34TCL_CLEAR_COLOR_VALUE_G_MASK 0x0000ff00 +#define NV34TCL_CLEAR_COLOR_VALUE_R_SHIFT 16 +#define NV34TCL_CLEAR_COLOR_VALUE_R_MASK 0x00ff0000 +#define NV34TCL_CLEAR_COLOR_VALUE_A_SHIFT 24 +#define NV34TCL_CLEAR_COLOR_VALUE_A_MASK 0xff000000 +#define NV34TCL_CLEAR_BUFFERS 0x00001d94 +#define NV34TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) +#define NV34TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) +#define NV34TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) +#define NV34TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) +#define NV34TCL_CLEAR_BUFFERS_STENCIL (1 << 1) +#define NV34TCL_CLEAR_BUFFERS_DEPTH (1 << 0) +#define NV34TCL_DO_VERTICES 0x00001dac +#define NV34TCL_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV34TCL_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV34TCL_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) +#define NV34TCL_BACK_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34TCL_VERTEX_FOG_1F 0x00001e54 +#define NV34TCL_VP_UPLOAD_FROM_ID 0x00001e9c +#define NV34TCL_VP_START_FROM_ID 0x00001ea0 +#define NV34TCL_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) +#define NV34TCL_POINT_PARAMETERS__SIZE 0x00000008 +#define NV34TCL_POINT_SIZE 0x00001ee0 +#define NV34TCL_POINT_PARAMETERS_ENABLE 0x00001ee4 +#define NV34TCL_POINT_SPRITE 0x00001ee8 +#define NV34TCL_POINT_SPRITE_ENABLE (1 << 0) +#define NV34TCL_POINT_SPRITE_R_MODE_SHIFT 1 +#define NV34TCL_POINT_SPRITE_R_MODE_MASK 0x00000006 +#define NV34TCL_POINT_SPRITE_R_MODE_GL_ZERO 0x00000000 +#define NV34TCL_POINT_SPRITE_R_MODE_GL_R 0x00000002 +#define NV34TCL_POINT_SPRITE_R_MODE_GL_S 0x00000004 +#define NV34TCL_POINT_SPRITE_COORD_REPLACE (1 << 11) +#define NV34TCL_VP_UPLOAD_CONST_ID 0x00001efc +#define NV34TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 +#define NV34TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 +#define NV34TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 +#define NV34TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 +#define NV34TCL_UNK1f80(x) (0x00001f80+((x)*4)) +#define NV34TCL_UNK1f80__SIZE 0x00000010 +#define NV34TCL_VP_ATTRIB_EN 0x00001ff0 +#define NV34TCL_VP_RESULT_EN 0x00001ff4 #define NV40_CONTEXT_SURFACES_2D 0x00003062 @@ -4388,6 +4623,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 #define NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 #define NV40TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV40TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f +#define NV40TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 #define NV40TCL_COLOR0_PITCH 0x0000020c #define NV40TCL_COLOR0_OFFSET 0x00000210 #define NV40TCL_ZETA_OFFSET 0x00000214 @@ -4517,50 +4754,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_COLOR_MASK_BUFFER0_R_MASK 0x00ff0000 #define NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT 24 #define NV40TCL_COLOR_MASK_BUFFER0_A_MASK 0xff000000 -#define NV40TCL_STENCIL_BACK_ENABLE 0x00000328 -#define NV40TCL_STENCIL_BACK_MASK 0x0000032c -#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000334 -#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000338 -#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000033c -#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000344 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000348 -#define NV40TCL_STENCIL_FRONT_MASK 0x0000034c -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000328 +#define NV40TCL_STENCIL_FRONT_MASK 0x0000032c +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000330 #define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 #define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 #define NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 @@ -4570,9 +4766,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 #define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 #define NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000354 -#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 -#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000334 +#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000338 +#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000033c #define NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 #define NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a #define NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 @@ -4581,7 +4777,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 #define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 #define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000340 #define NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 #define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a #define NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 @@ -4590,7 +4786,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 #define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 #define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000344 #define NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 #define NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a #define NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 @@ -4599,6 +4795,47 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 #define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 #define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_ENABLE 0x00000348 +#define NV40TCL_STENCIL_BACK_MASK 0x0000034c +#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000350 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000354 +#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000358 +#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000035c +#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000360 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000364 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 #define NV40TCL_SHADE_MODEL 0x00000368 #define NV40TCL_SHADE_MODEL_FLAT 0x00001d00 #define NV40TCL_SHADE_MODEL_SMOOTH 0x00001d01 @@ -4764,6 +5001,18 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VB_VERTEX_BATCH_START_SHIFT 0 #define NV40TCL_VB_VERTEX_BATCH_START_MASK 0x00ffffff #define NV40TCL_VERTEX_DATA 0x00001818 +#define NV40TCL_IDXBUF_ADDRESS 0x0000181c +#define NV40TCL_IDXBUF_FORMAT 0x00001820 +#define NV40TCL_IDXBUF_FORMAT_TYPE_SHIFT 4 +#define NV40TCL_IDXBUF_FORMAT_TYPE_MASK 0x000000f0 +#define NV40TCL_IDXBUF_FORMAT_TYPE_U32 0x00000000 +#define NV40TCL_IDXBUF_FORMAT_TYPE_U16 0x00000010 +#define NV40TCL_IDXBUF_FORMAT_DMA1 (1 << 0) +#define NV40TCL_VB_INDEX_BATCH 0x00001824 +#define NV40TCL_VB_INDEX_BATCH_COUNT_SHIFT 24 +#define NV40TCL_VB_INDEX_BATCH_COUNT_MASK 0xff000000 +#define NV40TCL_VB_INDEX_BATCH_START_SHIFT 0 +#define NV40TCL_VB_INDEX_BATCH_START_MASK 0x00ffffff #define NV40TCL_POLYGON_MODE_FRONT 0x00001828 #define NV40TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 #define NV40TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 @@ -4964,8 +5213,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT 24 #define NV40TCL_FP_CONTROL_TEMP_COUNT_MASK 0xff000000 #define NV40TCL_FP_CONTROL_KIL (1 << 7) -#define NV40TCL_FENCE_OFFSET 0x00001d6c -#define NV40TCL_FENCE_VALUE 0x00001d70 +#define NV40TCL_MULTISAMPLE_CONTROL 0x00001d7c #define NV40TCL_CLEAR_VALUE_DEPTH 0x00001d8c #define NV40TCL_CLEAR_VALUE_COLOR 0x00001d90 #define NV40TCL_CLEAR_BUFFERS 0x00001d94 @@ -5075,6 +5323,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 +#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH 0x00000238 +#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT_HIGH 0x0000023c #define NV50TCL 0x00005097 @@ -5777,6 +6027,46 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50TCL_COLOR_MASK__SIZE 0x00000008 +#define NV50_COMPUTE 0x000050c0 + +#define NV50_COMPUTE_DMA_UNK0 0x000001a0 +#define NV50_COMPUTE_DMA_STATUS 0x000001a4 +#define NV50_COMPUTE_DMA_UNK1 0x000001b8 +#define NV50_COMPUTE_DMA_UNK2 0x000001bc +#define NV50_COMPUTE_DMA_UNK3 0x000001c0 +#define NV50_COMPUTE_UNK4_HIGH 0x00000210 +#define NV50_COMPUTE_UNK4_LOW 0x00000214 +#define NV50_COMPUTE_UNK5_HIGH 0x00000218 +#define NV50_COMPUTE_UNK5_LOW 0x0000021c +#define NV50_COMPUTE_UNK6_HIGH 0x00000294 +#define NV50_COMPUTE_UNK6_LOW 0x00000298 +#define NV50_COMPUTE_CONST_BASE_HIGH 0x000002a4 +#define NV50_COMPUTE_CONST_BASE_LO 0x000002a8 +#define NV50_COMPUTE_CONST_SIZE_SEG 0x000002ac +#define NV50_COMPUTE_REG_COUNT 0x000002c0 +#define NV50_COMPUTE_STATUS_HIGH 0x00000310 +#define NV50_COMPUTE_STATUS_LOW 0x00000314 +#define NV50_COMPUTE_EXECUTE 0x0000031c +#define NV50_COMPUTE_USER_PARAM_COUNT 0x00000374 +#define NV50_COMPUTE_GRIDDIM_YX 0x000003a4 +#define NV50_COMPUTE_SHARED_SIZE 0x000003a8 +#define NV50_COMPUTE_BLOCKDIM_YX 0x000003ac +#define NV50_COMPUTE_BLOCKDIM_Z 0x000003b0 +#define NV50_COMPUTE_CALL_ADDRESS 0x000003b4 +#define NV50_COMPUTE_GLOBAL_BASE_HIGH(x) (0x00000400+((x)*32)) +#define NV50_COMPUTE_GLOBAL_BASE_HIGH__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_BASE_LOW(x) (0x00000404+((x)*32)) +#define NV50_COMPUTE_GLOBAL_BASE_LOW__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_LIMIT_HIGH(x) (0x00000408+((x)*32)) +#define NV50_COMPUTE_GLOBAL_LIMIT_HIGH__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_LIMIT_LOW(x) (0x0000040c+((x)*32)) +#define NV50_COMPUTE_GLOBAL_LIMIT_LOW__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_UNK(x) (0x00000410+((x)*32)) +#define NV50_COMPUTE_GLOBAL_UNK__SIZE 0x00000010 +#define NV50_COMPUTE_USER_PARAM(x) (0x00000600+((x)*4)) +#define NV50_COMPUTE_USER_PARAM__SIZE 0x00000040 + + #define NV54TCL 0x00008297 diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 30cd4290750..42a158986f6 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -425,8 +425,7 @@ nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); OUT_RINGp ((uint32_t *)&hw->depth, 3); - /*XXX: fix in nouveau_class.h, faces backwards.. */ - BEGIN_RING(curie, NV40TCL_STENCIL_BACK_ENABLE, 16); + BEGIN_RING(curie, NV40TCL_STENCIL_FRONT_ENABLE, 16); OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 7562781fcfe..d462e984003 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -175,11 +175,12 @@ nv40_vbo_validate_state(struct nv40_context *nv40, } if (ib) { - BEGIN_RING(curie, 0x181c, 2); + BEGIN_RING(curie, NV40TCL_IDXBUF_ADDRESS, 2); OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_OR, 0, 1); + NOUVEAU_BO_RD | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); } BEGIN_RING(curie, 0x1710, 1); @@ -342,26 +343,27 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - unsigned nr; + unsigned nr, type; switch (ib_size) { case 2: - assert(nv40_vbo_validate_state(nv40, ib, 0x00000010)); + type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; break; case 4: - assert(nv40_vbo_validate_state(nv40, ib, 0x00000000)); + type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; break; default: assert(0); } + assert(nv40_vbo_validate_state(nv40, ib, type)); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); nr = (count & 0xff); if (nr) { - BEGIN_RING(curie, 0x1824, 1); + BEGIN_RING(curie, NV40TCL_VB_INDEX_BATCH, 1); OUT_RING (((nr - 1) << 24) | start); start += nr; } @@ -372,7 +374,7 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, nr -= push; - BEGIN_RING_NI(curie, 0x1824, push); + BEGIN_RING_NI(curie, NV40TCL_VB_INDEX_BATCH, push); while (push--) { OUT_RING(((0x100 - 1) << 24) | start); start += 0x100; -- cgit v1.2.3 From a794fcfbdbc0417ed9960bd92e0ac49367660635 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 27 Dec 2007 06:02:01 +1100 Subject: nouveau: remove fence_del(), _ref() can be used for the same purpose. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 5 +- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 4 +- .../drivers/dri/nouveau_winsys/nouveau_fence.c | 105 ++++++++++++--------- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 2 +- 4 files changed, 61 insertions(+), 55 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 09e66d1bdcd..ca1e89e959d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -115,7 +115,7 @@ nouveau_bo_tmp_del(void *priv) { struct nouveau_resource *r = priv; - nouveau_fence_del((struct nouveau_fence **)&r->priv); + nouveau_fence_ref(NULL, (struct nouveau_fence **)&r->priv); nouveau_resource_free(&r); } @@ -398,10 +398,7 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, return ret; } - if (nvbo->fence) - nouveau_fence_del(&nvbo->fence); nouveau_fence_ref(fence, &nvbo->fence); - return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 724677961e6..1ce7174e758 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -94,11 +94,9 @@ nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **); extern int nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); -extern void -nouveau_fence_del(struct nouveau_fence **); - extern int nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *); + extern void nouveau_fence_emit(struct nouveau_fence *); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c index 53037d16823..7714e6f2485 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c @@ -28,38 +28,6 @@ #include "nouveau_dma.h" #include "nouveau_local.h" -int -nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!chan || !fence || *fence) - return -EINVAL; - - nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); - if (!nvfence) - return -ENOMEM; - nvfence->base.channel = chan; - nvfence->refcount = 1; - - *fence = &nvfence->base; - return 0; -} - -int -nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!ref || !fence || *fence) - return -EINVAL; - nvfence = nouveau_fence(ref); - nvfence->refcount++; - - *fence = &nvfence->base; - return 0; -} - static void nouveau_fence_del_unsignalled(struct nouveau_fence *fence) { @@ -82,7 +50,7 @@ nouveau_fence_del_unsignalled(struct nouveau_fence *fence) nvchan->fence_tail = le; } -void +static void nouveau_fence_del(struct nouveau_fence **fence) { struct nouveau_fence_priv *nvfence; @@ -107,6 +75,46 @@ nouveau_fence_del(struct nouveau_fence **fence) free(nvfence); } +int +nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!chan || !fence || *fence) + return -EINVAL; + + nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); + if (!nvfence) + return -ENOMEM; + nvfence->base.channel = chan; + nvfence->refcount = 1; + + *fence = &nvfence->base; + return 0; +} + +int +nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence) + return -EINVAL; + + if (*fence) { + nouveau_fence_del(fence); + *fence = NULL; + } + + if (ref) { + nvfence = nouveau_fence(ref); + nvfence->refcount++; + *fence = &nvfence->base; + } + + return 0; +} + int nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *), void *priv) @@ -159,28 +167,31 @@ nouveau_fence_flush(struct nouveau_channel *chan) uint32_t sequence = *nvchan->ref_cnt; while (nvchan->fence_head) { - struct nouveau_fence *fence = NULL; struct nouveau_fence_priv *nvfence; - nouveau_fence_ref(nvchan->fence_head, &fence); nvfence = nouveau_fence(nvchan->fence_head); - - if (nvfence->sequence > sequence) { - nouveau_fence_del(&fence); + if (nvfence->sequence > sequence) break; - } nouveau_fence_del_unsignalled(&nvfence->base); nvfence->signalled = 1; - while (nvfence->signal_cb) { - struct nouveau_fence_cb *cb = nvfence->signal_cb; - nvfence->signal_cb = cb->next; - cb->func(cb->priv); - free(cb); - } + if (nvfence->signal_cb) { + struct nouveau_fence *fence = NULL; + + nouveau_fence_ref(nvchan->fence_head, &fence); - nouveau_fence_del(&fence); + while (nvfence->signal_cb) { + struct nouveau_fence_cb *cb; + + cb = nvfence->signal_cb; + nvfence->signal_cb = cb->next; + cb->func(cb->priv); + free(cb); + } + + nouveau_fence_ref(NULL, &fence); + } } } @@ -197,7 +208,7 @@ nouveau_fence_wait(struct nouveau_fence **fence) while (!nvfence->signalled) nouveau_fence_flush(nvfence->base.channel); } - nouveau_fence_del(fence); + nouveau_fence_ref(NULL, fence); return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 148406b85a7..020b1e08407 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -169,7 +169,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) /* Fence + kickoff */ nouveau_fence_emit(fence); FIRE_RING_CH(chan); - nouveau_fence_del(&fence); + nouveau_fence_ref(NULL, &fence); /* Allocate space for next push buffer */ assert(!nouveau_pushbuf_space(chan, min)); -- cgit v1.2.3 From 3b8efe1dfedec0b6598dc9cc3d6897fe2b3b788d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 27 Dec 2007 06:10:49 +1100 Subject: nouveau: allow unfenced read of buffers in a few situations --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 8 +++++++- src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h | 1 + 2 files changed, 8 insertions(+), 1 deletion(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index ca1e89e959d..08d38c58af3 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -299,8 +299,10 @@ nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) if (!nvbo) return -EINVAL; - if (nvbo->fence) + if (flags & NOUVEAU_BO_WR) nouveau_fence_wait(&nvbo->fence); + else + nouveau_fence_wait(&nvbo->wr_fence); if (nvbo->sysmem) bo->map = nvbo->sysmem; @@ -385,6 +387,8 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, struct nouveau_bo_priv *nvbo = nouveau_bo(bo); int ret; + assert(bo->map == NULL); + if (nvbo->user) { ret = nouveau_bo_validate_user(chan, bo, fence, flags); if (ret) { @@ -398,6 +402,8 @@ nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, return ret; } + if (flags & NOUVEAU_BO_WR) + nouveau_fence_ref(fence, &nvbo->wr_fence); nouveau_fence_ref(fence, &nvbo->fence); return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 1ce7174e758..4b42baf809a 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -242,6 +242,7 @@ struct nouveau_bo_priv { struct nouveau_bo base; struct nouveau_fence *fence; + struct nouveau_fence *wr_fence; struct drm_nouveau_mem_alloc drm; void *map; -- cgit v1.2.3 From dc6d73e1a47cfd7f4cb615e75d9f7f944efabf03 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 27 Dec 2007 06:21:13 +1100 Subject: nouveau: track last validated offsets, so we know when relocs can be avoided. --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 10 +++++----- src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 +++ src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 10 ++++++++++ 3 files changed, 18 insertions(+), 5 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 08d38c58af3..288674f231b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -348,8 +348,8 @@ nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); - nvbo->base.offset = nvdev->sa.offset + r->start; - nvbo->base.flags = NOUVEAU_BO_GART; + nvbo->offset = nvdev->sa.offset + r->start; + nvbo->flags = NOUVEAU_BO_GART; return 0; } @@ -371,11 +371,11 @@ nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, nouveau_bo_upload(nvbo); } - nvbo->base.offset = nvbo->drm.offset; + nvbo->offset = nvbo->drm.offset; if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) - nvbo->base.flags = NOUVEAU_BO_GART; + nvbo->flags = NOUVEAU_BO_GART; else - nvbo->base.flags = NOUVEAU_BO_VRAM; + nvbo->flags = NOUVEAU_BO_VRAM; return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 4b42baf809a..3e886869d85 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -251,6 +251,9 @@ struct nouveau_bo_priv { int user; int refcount; + + uint64_t offset; + uint64_t flags; }; #define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index 020b1e08407..a34a5c18662 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -131,6 +131,16 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) ret = nouveau_bo_validate(chan, bo, fence, pbbo->flags); assert (ret == 0); + if (bo->offset == nouveau_bo(bo)->offset && + bo->flags == nouveau_bo(bo)->flags) { + /*XXX: could avoid reloc in this case, except with the + * current design we'd confuse the GPU quite a bit + * if we did this. Will fix soon. + */ + } + bo->offset = nouveau_bo(bo)->offset; + bo->flags = nouveau_bo(bo)->flags; + while ((r = ptr_to_pbrel(pbbo->relocs))) { uint32_t push; -- cgit v1.2.3 From 8ed3a69fc8f811c2010615af94a6bb8003c8a296 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 27 Dec 2007 09:01:08 +1100 Subject: nouveau: remove legacy stuff --- .../drivers/dri/nouveau_winsys/nouveau_context.h | 1 - .../drivers/dri/nouveau_winsys/nouveau_local.h | 31 +++++++++++--------- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 29 ++---------------- src/mesa/pipe/nouveau/nouveau_winsys.h | 12 ++++---- src/mesa/pipe/nv40/nv40_context.h | 1 - src/mesa/pipe/nv40/nv40_dma.h | 34 ++++++++++++---------- src/mesa/pipe/nv50/nv50_context.h | 1 - src/mesa/pipe/nv50/nv50_dma.h | 34 ++++++++++++---------- 8 files changed, 62 insertions(+), 81 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index f201001bd9b..2fd3336065c 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -31,7 +31,6 @@ struct nouveau_context { struct pipe_surface *frontbuffer; /* Hardware context */ - uint32_t *pushbuf; struct nouveau_channel *channel; struct nouveau_notifier *sync_notifier; struct nouveau_grobj *NvNull; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index e7111de4dc1..a53b6992028 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -24,21 +24,13 @@ #define NOUVEAU_DMA_TIMEOUT 2000 /* Push buffer access macros */ -#define BEGIN_RING(obj,mthd,size) do { \ - nv->pushbuf = nouveau_pipe_dma_beginp(nv->obj, (mthd), (size)); \ -} while(0) - -#define BEGIN_RING_GR(obj,mthd,size) do { \ - nv->pushbuf = nouveau_pipe_dma_beginp(obj, (mthd), (size)); \ -} while(0) - #define OUT_RING(data) do { \ - (*nv->pushbuf++) = (data); \ + (*nv->channel->pushbuf->cur++) = (data); \ } while(0) #define OUT_RINGp(src,size) do { \ - memcpy(nv->pushbuf, (src), (size)<<2); \ - nv->pushbuf += (size); \ + memcpy(nv->channel->pushbuf->cur, (src), (size)<<2); \ + nv->channel->pushbuf->cur += (size); \ } while(0) #define OUT_RINGf(data) do { \ @@ -48,7 +40,18 @@ } while(0) #define FIRE_RING() do { \ - nouveau_pipe_dma_kickoff(nv->channel); \ + nouveau_pushbuf_flush(nv->channel, 0); \ +} while(0) + +#define BEGIN_RING_GR(obj,mthd,size) do { \ + if (nv->channel->pushbuf->remaining < ((size) + 1)) \ + nouveau_pushbuf_flush(nv->channel, ((size) + 1)); \ + OUT_RING(((obj)->subc << 13) | ((size) << 18) | (mthd)); \ + nv->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING(obj,mthd,size) do { \ + BEGIN_RING_GR(nv->obj, (mthd), (size)); \ } while(0) #define BIND_RING(o,s) do { \ @@ -58,8 +61,8 @@ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nouveau_pushbuf_emit_reloc(nv->channel, nv->pushbuf, (void*)(bo), \ - (data), (flags), (vor), (tor)); \ + nouveau_pushbuf_emit_reloc(nv->channel, nv->channel->pushbuf->cur, \ + (void*)(bo), (data), (flags), (vor), (tor));\ OUT_RING(0); \ } while(0) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index 403647ab17e..f5e45461000 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -34,30 +34,6 @@ nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, return 0; } -uint32_t * -nouveau_pipe_dma_beginp(struct nouveau_grobj *grobj, int mthd, int size) -{ - struct nouveau_channel *chan = grobj->channel; - uint32_t *pushbuf; - - if (chan->pushbuf->remaining < (size + 1)) { - nouveau_pushbuf_flush(chan, size + 1); - } - - pushbuf = chan->pushbuf->cur; - chan->pushbuf->cur += (size + 1); - chan->pushbuf->remaining -= (size + 1); - - (*pushbuf++) = ((grobj->subc << 13) | (size << 18) | mthd); - return pushbuf; -} - -void -nouveau_pipe_dma_kickoff(struct nouveau_channel *chan) -{ - nouveau_pushbuf_flush(chan, 0); -} - static int nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst, unsigned dx, unsigned dy, struct pipe_surface *src, @@ -126,9 +102,8 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->res_alloc = nouveau_resource_alloc; nvws->res_free = nouveau_resource_free; - nvws->begin_ring = nouveau_pipe_dma_beginp; - nvws->out_reloc = nouveau_pushbuf_emit_reloc; - nvws->fire_ring = nouveau_pipe_dma_kickoff; + nvws->push_reloc = nouveau_pushbuf_emit_reloc; + nvws->push_flush = nouveau_pushbuf_flush; nvws->grobj_alloc = nouveau_pipe_grobj_alloc; nvws->grobj_free = nouveau_grobj_free; diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 7b80027c3a6..27c991077e7 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -11,6 +11,7 @@ #include "pipe/nouveau/nouveau_grobj.h" #include "pipe/nouveau/nouveau_notifier.h" #include "pipe/nouveau/nouveau_resource.h" +#include "pipe/nouveau/nouveau_pushbuf.h" struct nouveau_winsys { struct nouveau_context *nv; @@ -23,13 +24,10 @@ struct nouveau_winsys { struct nouveau_resource **); void (*res_free)(struct nouveau_resource **); - /*XXX: this is crappy, and bound to be slow.. however, it's nice and - * simple, it'll do for the moment*/ - uint32_t *(*begin_ring)(struct nouveau_grobj *, int mthd, int size); - int (*out_reloc)(struct nouveau_channel *, void *ptr, - struct nouveau_bo *, uint32_t data, - uint32_t flags, uint32_t vor, uint32_t tor); - void (*fire_ring)(struct nouveau_channel *); + int (*push_reloc)(struct nouveau_channel *, void *ptr, + struct nouveau_bo *, uint32_t data, + uint32_t flags, uint32_t vor, uint32_t tor); + int (*push_flush)(struct nouveau_channel *, unsigned size); int (*grobj_alloc)(struct nouveau_winsys *, int grclass, struct nouveau_grobj **); diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 5d0bf89680d..934f68ef1aa 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -31,7 +31,6 @@ struct nv40_context { int chipset; struct nouveau_grobj *curie; struct nouveau_notifier *sync; - uint32_t *pushbuf; /* query objects */ struct nouveau_notifier *query; diff --git a/src/mesa/pipe/nv40/nv40_dma.h b/src/mesa/pipe/nv40/nv40_dma.h index 3775ce6e72d..1fb82677689 100644 --- a/src/mesa/pipe/nv40/nv40_dma.h +++ b/src/mesa/pipe/nv40/nv40_dma.h @@ -3,21 +3,13 @@ #include "pipe/nouveau/nouveau_winsys.h" -#define BEGIN_RING(obj,mthd,size) do { \ - nv40->pushbuf = nv40->nvws->begin_ring(nv40->obj, (mthd), (size)); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - #define OUT_RING(data) do { \ - (*nv40->pushbuf++) = (data); \ + (*nv40->nvws->channel->pushbuf->cur++) = (data); \ } while(0) #define OUT_RINGp(src,size) do { \ - memcpy(nv40->pushbuf, (src), (size) * 4); \ - nv40->pushbuf += (size); \ + memcpy(nv40->nvws->channel->pushbuf->cur, (src), (size) * 4); \ + nv40->nvws->channel->pushbuf->cur += (size); \ } while(0) #define OUT_RINGf(data) do { \ @@ -26,14 +18,26 @@ OUT_RING(c.u); \ } while(0) +#define BEGIN_RING(obj,mthd,size) do { \ + if (nv40->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + nv40->nvws->push_flush(nv40->nvws->channel, ((size) + 1)); \ + OUT_RING((nv40->obj->subc << 13) | ((size) << 18) | (mthd)); \ + nv40->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + #define FIRE_RING() do { \ - nv40->nvws->fire_ring(nv40->nvws->channel); \ + nv40->nvws->push_flush(nv40->nvws->channel, 0); \ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nv40->nvws->out_reloc(nv40->nvws->channel, nv40->pushbuf, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ + nv40->nvws->push_reloc(nv40->nvws->channel, \ + nv40->nvws->channel->pushbuf->cur, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ OUT_RING(0); \ } while(0) diff --git a/src/mesa/pipe/nv50/nv50_context.h b/src/mesa/pipe/nv50/nv50_context.h index 5be4e5cb1be..b1c4d5b8948 100644 --- a/src/mesa/pipe/nv50/nv50_context.h +++ b/src/mesa/pipe/nv50/nv50_context.h @@ -26,7 +26,6 @@ struct nv50_context { int chipset; struct nouveau_grobj *tesla; struct nouveau_notifier *sync; - uint32_t *pushbuf; }; diff --git a/src/mesa/pipe/nv50/nv50_dma.h b/src/mesa/pipe/nv50/nv50_dma.h index f8121b61cc6..366316df755 100644 --- a/src/mesa/pipe/nv50/nv50_dma.h +++ b/src/mesa/pipe/nv50/nv50_dma.h @@ -3,21 +3,13 @@ #include "pipe/nouveau/nouveau_winsys.h" -#define BEGIN_RING(obj,mthd,size) do { \ - nv50->pushbuf = nv50->nvws->begin_ring(nv50->obj, (mthd), (size)); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - #define OUT_RING(data) do { \ - (*nv50->pushbuf++) = (data); \ + (*nv50->nvws->channel->pushbuf->cur++) = (data); \ } while(0) #define OUT_RINGp(src,size) do { \ - memcpy(nv50->pushbuf, (src), (size) * 4); \ - nv50->pushbuf += (size); \ + memcpy(nv50->nvws->channel->pushbuf->cur, (src), (size) * 4); \ + nv50->nvws->channel->pushbuf->cur += (size); \ } while(0) #define OUT_RINGf(data) do { \ @@ -26,14 +18,26 @@ OUT_RING(c.u); \ } while(0) +#define BEGIN_RING(obj,mthd,size) do { \ + if (nv50->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + nv50->nvws->push_flush(nv50->nvws->channel, ((size) + 1)); \ + OUT_RING((nv50->obj->subc << 13) | ((size) << 18) | (mthd)); \ + nv50->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + #define FIRE_RING() do { \ - nv50->nvws->fire_ring(nv50->nvws->channel); \ + nv50->nvws->push_flush(nv50->nvws->channel, 0); \ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nv50->nvws->out_reloc(nv50->nvws->channel, nv50->pushbuf, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ + nv50->nvws->push_reloc(nv50->nvws->channel, \ + nv50->nvws->channel->pushbuf->cur, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ OUT_RING(0); \ } while(0) -- cgit v1.2.3 From 09612416d74d92c22bf757fc513a70bec30f15b5 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Dec 2007 00:37:58 +1100 Subject: nouveau: typo fix --- src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c index baf4804dde0..2288f9c71a8 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c @@ -31,7 +31,7 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, unsigned sx, unsigned sy, unsigned w, unsigned h) { struct pipe_surface *dst = nv->surf_dst; - struct pipe_surface *src = nv->surf_dst; + struct pipe_surface *src = nv->surf_src; unsigned dst_offset, src_offset; dst_offset = dst->offset + (dy * dst->pitch + dx) * dst->cpp; @@ -53,8 +53,8 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, OUT_RING (0); h -= count; - src_offset += src->pitch * count; - dst_offset += dst->pitch * count; + src_offset += src->pitch * src->cpp * count; + dst_offset += dst->pitch * dst->cpp * count; } } -- cgit v1.2.3 From 24cfb7d2e29aae5afa2529d3798fa94d813dc017 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Dec 2007 01:03:24 +1100 Subject: nv40: untested fragprog mods for big-endian chips --- src/mesa/pipe/nv40/nv40_fragprog.c | 26 +++++++++++++++++++++----- 1 file changed, 21 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 6ee6a7daff7..7d9676b8414 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -781,18 +781,34 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) } if (!fp->on_hw) { + const uint32_t le = 1; + uint32_t *map; + if (!fp->buffer) fp->buffer = ws->buffer_create(ws, 0x100, 0, 0); + ws->buffer_data(ws, fp->buffer, fp->insn_len * 4, NULL, 0); + + map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_FLAG_READ); #if 0 - for (i = 0; i < fp->insn_len; i++) + for (i = 0; i < fp->insn_len; i++) { NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + } #endif - nv40->pipe.winsys->buffer_data(nv40->pipe.winsys, fp->buffer, - fp->insn_len * sizeof(uint32_t), - fp->insn, - PIPE_BUFFER_USAGE_PIXEL); + if ((*(const uint8_t *)&le)) { + for (i = 0; i < fp->insn_len; i++) { + map[i] = fp->insn[i]; + } + } else { + /* Weird swapping for big-endian chips */ + for (i = 0; i < fp->insn_len; i++) { + map[i] = ((fp->insn[i] & 0xffff) << 16) | + ((fp->insn[i] >> 16) & 0xffff); + } + } + + ws->buffer_unmap(ws, fp->buffer); fp->on_hw = TRUE; } -- cgit v1.2.3 From 6af6bb6817bdf55a0e0f651070f0367621594c11 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Dec 2007 01:17:47 +1100 Subject: nouveau: header update --- src/mesa/pipe/nouveau/nouveau_class.h | 26 ++++++++++++++++++++------ src/mesa/pipe/nv40/nv40_state.c | 30 +++++++++++++++--------------- 2 files changed, 35 insertions(+), 21 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h index 56f4e9001ec..1f2c52891d9 100644 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -3700,8 +3700,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_DMA_QUERY 0x000001a8 #define NV34TCL_DMA_IN_MEMORY7 0x000001ac #define NV34TCL_DMA_IN_MEMORY8 0x000001b0 -#define NV34TCL_DMA_COLOR2 0x000001b4 -#define NV34TCL_DMA_COLOR3 0x000001b8 #define NV34TCL_VIEWPORT_HORIZ 0x00000200 #define NV34TCL_VIEWPORT_HORIZ_X_SHIFT 0 #define NV34TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff @@ -3772,10 +3770,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VIEWPORT_CLIP_VERT_T_MASK 0x0000ffff #define NV34TCL_VIEWPORT_CLIP_VERT_D_SHIFT 16 #define NV34TCL_VIEWPORT_CLIP_VERT_D_MASK 0xffff0000 -#define NV34TCL_COLOR2_PITCH 0x00000280 -#define NV34TCL_COLOR3_PITCH 0x00000284 -#define NV34TCL_COLOR2_OFFSET 0x00000288 -#define NV34TCL_COLOR3_OFFSET 0x0000028c #define NV34TCL_DITHER_ENABLE 0x00000300 #define NV34TCL_ALPHA_FUNC_ENABLE 0x00000304 #define NV34TCL_ALPHA_FUNC_FUNC 0x00000308 @@ -5118,9 +5112,29 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE 0x00060000 #define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER 0x00070000 #define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP 0x00080000 +#define NV40TCL_TEX_WRAP_RCOMP_SHIFT 28 +#define NV40TCL_TEX_WRAP_RCOMP_MASK 0xf0000000 +#define NV40TCL_TEX_WRAP_RCOMP_NEVER 0x00000000 +#define NV40TCL_TEX_WRAP_RCOMP_GREATER 0x10000000 +#define NV40TCL_TEX_WRAP_RCOMP_EQUAL 0x20000000 +#define NV40TCL_TEX_WRAP_RCOMP_GEQUAL 0x30000000 +#define NV40TCL_TEX_WRAP_RCOMP_LESS 0x40000000 +#define NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL 0x50000000 +#define NV40TCL_TEX_WRAP_RCOMP_LEQUAL 0x60000000 +#define NV40TCL_TEX_WRAP_RCOMP_ALWAYS 0x70000000 #define NV40TCL_TEX_ENABLE(x) (0x00001a0c+((x)*32)) #define NV40TCL_TEX_ENABLE__SIZE 0x00000010 #define NV40TCL_TEX_ENABLE_ENABLE (1 << 31) +#define NV40TCL_TEX_ENABLE_ANISO_SHIFT 4 +#define NV40TCL_TEX_ENABLE_ANISO_MASK 0x000000f0 +#define NV40TCL_TEX_ENABLE_ANISO_NONE 0x00000000 +#define NV40TCL_TEX_ENABLE_ANISO_2X 0x00000010 +#define NV40TCL_TEX_ENABLE_ANISO_4X 0x00000020 +#define NV40TCL_TEX_ENABLE_ANISO_6X 0x00000030 +#define NV40TCL_TEX_ENABLE_ANISO_8X 0x00000040 +#define NV40TCL_TEX_ENABLE_ANISO_10X 0x00000050 +#define NV40TCL_TEX_ENABLE_ANISO_12X 0x00000060 +#define NV40TCL_TEX_ENABLE_ANISO_16X 0x00000070 #define NV40TCL_TEX_SWIZZLE(x) (0x00001a10+((x)*32)) #define NV40TCL_TEX_SWIZZLE__SIZE 0x00000010 #define NV40TCL_TEX_SWIZZLE_S0_X_SHIFT 14 diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 42a158986f6..70abc0feba0 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -127,24 +127,24 @@ nv40_sampler_state_create(struct pipe_context *pipe, ps->wrap |= (1 << 5); if (cso->max_anisotropy >= 16.0) { - ps->en |= (7 << 4); + ps->en |= NV40TCL_TEX_ENABLE_ANISO_16X; } else if (cso->max_anisotropy >= 12.0) { - ps->en |= (6 << 4); + ps->en |= NV40TCL_TEX_ENABLE_ANISO_12X; } else if (cso->max_anisotropy >= 10.0) { - ps->en |= (5 << 4); + ps->en |= NV40TCL_TEX_ENABLE_ANISO_10X; } else if (cso->max_anisotropy >= 8.0) { - ps->en |= (4 << 4); + ps->en |= NV40TCL_TEX_ENABLE_ANISO_8X; } else if (cso->max_anisotropy >= 6.0) { - ps->en |= (3 << 4); + ps->en |= NV40TCL_TEX_ENABLE_ANISO_6X; } else if (cso->max_anisotropy >= 4.0) { - ps->en |= (2 << 4); + ps->en |= NV40TCL_TEX_ENABLE_ANISO_4X; } else { - ps->en |= (1 << 4); /* 2.0 */ + ps->en |= NV40TCL_TEX_ENABLE_ANISO_2X; } } @@ -195,28 +195,28 @@ nv40_sampler_state_create(struct pipe_context *pipe, if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { switch (cso->compare_func) { case PIPE_FUNC_NEVER: - ps->wrap |= (0x0 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NEVER; break; case PIPE_FUNC_GREATER: - ps->wrap |= (0x1 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GREATER; break; case PIPE_FUNC_EQUAL: - ps->wrap |= (0x2 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_EQUAL; break; case PIPE_FUNC_GEQUAL: - ps->wrap |= (0x3 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GEQUAL; break; case PIPE_FUNC_LESS: - ps->wrap |= (0x4 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LESS; break; case PIPE_FUNC_NOTEQUAL: - ps->wrap |= (0x5 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL; break; case PIPE_FUNC_LEQUAL: - ps->wrap |= (0x6 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LEQUAL; break; case PIPE_FUNC_ALWAYS: - ps->wrap |= (0x7 << 28); + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_ALWAYS; break; default: break; -- cgit v1.2.3 From 92ee96d83d26e56857b2999356d7c2a75e6eaf76 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Dec 2007 15:24:16 +1100 Subject: nv40: s/READ/WRITE/ --- src/mesa/pipe/nv40/nv40_fragprog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 7d9676b8414..d23cb5ee9d8 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -788,7 +788,7 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->buffer = ws->buffer_create(ws, 0x100, 0, 0); ws->buffer_data(ws, fp->buffer, fp->insn_len * 4, NULL, 0); - map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_FLAG_READ); + map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_FLAG_WRITE); #if 0 for (i = 0; i < fp->insn_len; i++) { -- cgit v1.2.3 From 732540f997ef0501ccbc1237148bc44aaba38d66 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 2 Jan 2008 23:44:24 +1100 Subject: nv40: some cleanups --- src/mesa/pipe/nv40/nv40_context.c | 107 +++++++++++++++----------------------- src/mesa/pipe/nv40/nv40_context.h | 3 +- src/mesa/pipe/nv40/nv40_query.c | 87 ++++++++++++++++--------------- src/mesa/pipe/nv40/nv40_surface.c | 60 +++++++++++++++++---- 4 files changed, 137 insertions(+), 120 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index e7d0a186378..442ff04e753 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -6,46 +6,6 @@ #include "nv40_context.h" #include "nv40_dma.h" -static boolean -nv40_is_format_supported(struct pipe_context *pipe, enum pipe_format format, - uint type) -{ - switch (type) { - case PIPE_SURFACE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return TRUE; - default: - break; - } - break; - case PIPE_TEXTURE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A1R5G5B5_UNORM: - case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: - case PIPE_FORMAT_Z16_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - } - break; - default: - assert(0); - }; - - return FALSE; -} - static const char * nv40_get_name(struct pipe_context *pipe) { @@ -149,8 +109,21 @@ static void nv40_destroy(struct pipe_context *pipe) { struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nouveau_winsys *nvws = nv40->nvws; + + if (nv40->draw) + draw_destroy(nv40->draw); + + nvws->res_free(&nv40->vertprog.exec_heap); + nvws->res_free(&nv40->vertprog.data_heap); + + nvws->res_free(&nv40->query_heap); + nvws->notifier_free(&nv40->query); + + nvws->notifier_free(&nv40->sync); + + nvws->grobj_free(&nv40->curie); - draw_destroy(nv40->draw); free(nv40); } @@ -160,14 +133,8 @@ nv40_init_hwctx(struct nv40_context *nv40, int curie_class) struct nouveau_winsys *nvws = nv40->nvws; int ret; - if ((ret = nvws->notifier_alloc(nvws, nv40->num_query_objects, - &nv40->query))) { - NOUVEAU_ERR("Error creating query notifier objects: %d\n", ret); - return FALSE; - } - - if ((ret = nvws->grobj_alloc(nvws, curie_class, - &nv40->curie))) { + ret = nvws->grobj_alloc(nvws, curie_class, &nv40->curie); + if (ret) { NOUVEAU_ERR("Error creating 3D object: %d\n", ret); return FALSE; } @@ -237,12 +204,12 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, } if (GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) { - curie_class = 0x4097; + curie_class = NV40TCL; } else if (GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) { - curie_class = 0x4497; + curie_class = NV44TCL; } else { - NOUVEAU_ERR("Unknown NV4X chipset: NV%02x\n", chipset); + NOUVEAU_ERR("Unknown NV4x chipset: NV%02x\n", chipset); return NULL; } @@ -252,37 +219,46 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, nv40->chipset = chipset; nv40->nvws = nvws; - if ((ret = nvws->notifier_alloc(nvws, 1, &nv40->sync))) { + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &nv40->sync); + if (ret) { NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - free(nv40); + nv40_destroy(&nv40->pipe); return NULL; } - nv40->num_query_objects = 32; - nv40->query_objects = calloc(nv40->num_query_objects, - sizeof(struct pipe_query_object *)); - if (!nv40->query_objects) { - free(nv40); + /* Query objects */ + ret = nvws->notifier_alloc(nvws, 32, &nv40->query); + if (ret) { + NOUVEAU_ERR("Error initialising query objects: %d\n", ret); + nv40_destroy(&nv40->pipe); return NULL; } + ret = nvws->res_init(&nv40->query_heap, 0, 32); + if (ret) { + NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); + nv40_destroy(&nv40->pipe); + return NULL; + } + + /* Vtxprog resources */ if (nvws->res_init(&nv40->vertprog.exec_heap, 0, 512) || nvws->res_init(&nv40->vertprog.data_heap, 0, 256)) { - nvws->res_free(&nv40->vertprog.exec_heap); - nvws->res_free(&nv40->vertprog.data_heap); - free(nv40); + nv40_destroy(&nv40->pipe); return NULL; } + /* Static curie initialisation */ if (!nv40_init_hwctx(nv40, curie_class)) { - free(nv40); + nv40_destroy(&nv40->pipe); return NULL; } + /* Pipe context setup */ nv40->pipe.winsys = pipe_winsys; nv40->pipe.destroy = nv40_destroy; - nv40->pipe.is_format_supported = nv40_is_format_supported; nv40->pipe.get_name = nv40_get_name; nv40->pipe.get_vendor = nv40_get_vendor; nv40->pipe.get_param = nv40_get_param; @@ -305,5 +281,4 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, return &nv40->pipe; } - - + diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 934f68ef1aa..975b1096ccf 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -34,8 +34,7 @@ struct nv40_context { /* query objects */ struct nouveau_notifier *query; - boolean *query_objects; - uint num_query_objects; + struct nouveau_resource *query_heap; uint32_t dirty; diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index f51b34b119d..6e5fcae8cad 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -3,94 +3,97 @@ #include "nv40_context.h" #include "nv40_dma.h" -/*XXX: Maybe go notifier per-query one day? not sure if PRAMIN space is - * plentiful enough however.. */ struct nv40_query { + struct nouveau_resource *object; unsigned type; - int id; + boolean ready; + uint64_t result; }; #define nv40_query(o) ((struct nv40_query *)(o)) static struct pipe_query * nv40_query_create(struct pipe_context *pipe, unsigned query_type) { - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *nv40query; - int id; - - for (id = 0; id < nv40->num_query_objects; id++) { - if (nv40->query_objects[id] == 0) - break; - } - - if (id == nv40->num_query_objects) - return NULL; - nv40->query_objects[id] = TRUE; + struct nv40_query *q; - nv40query = malloc(sizeof(struct nv40_query)); - nv40query->type = query_type; - nv40query->id = id; + q = calloc(1, sizeof(struct nv40_query)); + q->type = query_type; - return (struct pipe_query *)nv40query; + return (struct pipe_query *)q; } static void -nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *q) +nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *nv40query = nv40_query(q); + struct nv40_query *q = nv40_query(pq); - assert(nv40->query_objects[nv40query->id]); - nv40->query_objects[nv40query->id] = FALSE; - free(nv40query); + if (q->object) + nv40->nvws->res_free(&q->object); + free(q); } static void -nv40_query_begin(struct pipe_context *pipe, struct pipe_query *q) +nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *nv40query = nv40_query(q); + struct nv40_query *q = nv40_query(pq); - assert(nv40query->type == PIPE_QUERY_OCCLUSION_COUNTER); + assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - nv40->nvws->notifier_reset(nv40->query, nv40query->id); + if (nv40->nvws->res_alloc(nv40->query_heap, 1, NULL, &q->object)) + assert(0); + nv40->nvws->notifier_reset(nv40->query, q->object->start); BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); OUT_RING (1); BEGIN_RING(curie, NV40TCL_QUERY_UNK17CC, 1); OUT_RING (1); + + q->ready = FALSE; } static void -nv40_query_end(struct pipe_context *pipe, struct pipe_query *q) +nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nv40_query *nv40query = nv40_query(q); + struct nv40_query *q = nv40_query(pq); BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | - ((nv40query->id * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); + ((q->object->start * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); FIRE_RING(); } static boolean -nv40_query_result(struct pipe_context *pipe, struct pipe_query *q, +nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, boolean wait, uint64_t *result) { struct nv40_context *nv40 = (struct nv40_context *)pipe; - struct nv40_query *nv40query = nv40_query(q); + struct nv40_query *q = nv40_query(pq); struct nouveau_winsys *nvws = nv40->nvws; - boolean status; - - status = nvws->notifier_status(nv40->query, nv40query->id); - if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { - if (wait == FALSE) - return FALSE; - nvws->notifier_wait(nv40->query, nv40query->id, - NV_NOTIFY_STATE_STATUS_COMPLETED, 0); + + assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (!q->ready) { + unsigned status; + + status = nvws->notifier_status(nv40->query, q->object->start); + if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { + if (wait == FALSE) + return FALSE; + nvws->notifier_wait(nv40->query, q->object->start, + NV_NOTIFY_STATE_STATUS_COMPLETED, + 0); + } + + q->result = nvws->notifier_retval(nv40->query, + q->object->start); + q->ready = TRUE; + nvws->res_free(&q->object); } - *result = nvws->notifier_retval(nv40->query, nv40query->id); + *result = q->result; return TRUE; } diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index 6a16a280c24..ed144c636c7 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -33,9 +33,48 @@ #include "pipe/p_inlines.h" #include "pipe/util/p_tile.h" +static boolean +nv40_surface_format_supported(struct pipe_context *pipe, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + static struct pipe_surface * -nv40_get_tex_surface(struct pipe_context *pipe, - struct pipe_texture *pt, +nv40_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { struct pipe_winsys *ws = pipe->winsys; @@ -103,12 +142,13 @@ nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, void nv40_init_surface_functions(struct nv40_context *nv40) { - nv40->pipe.get_tex_surface = nv40_get_tex_surface; - nv40->pipe.get_tile = pipe_get_tile_raw; - nv40->pipe.put_tile = pipe_put_tile_raw; - nv40->pipe.get_tile_rgba = pipe_get_tile_rgba; - nv40->pipe.put_tile_rgba = pipe_put_tile_rgba; - nv40->pipe.surface_data = nv40_surface_data; - nv40->pipe.surface_copy = nv40_surface_copy; - nv40->pipe.surface_fill = nv40_surface_fill; + nv40->pipe.is_format_supported = nv40_surface_format_supported; + nv40->pipe.get_tex_surface = nv40_get_tex_surface; + nv40->pipe.get_tile = pipe_get_tile_raw; + nv40->pipe.put_tile = pipe_put_tile_raw; + nv40->pipe.get_tile_rgba = pipe_get_tile_rgba; + nv40->pipe.put_tile_rgba = pipe_put_tile_rgba; + nv40->pipe.surface_data = nv40_surface_data; + nv40->pipe.surface_copy = nv40_surface_copy; + nv40->pipe.surface_fill = nv40_surface_fill; } -- cgit v1.2.3 From df34b6b2bcf26392eeffc828c9e90ea68a908736 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 6 Jan 2008 21:52:24 +1100 Subject: nv40: bind textures based on fragprog usage. --- src/mesa/pipe/nv40/nv40_context.h | 7 ++++--- src/mesa/pipe/nv40/nv40_fragprog.c | 6 +++++- src/mesa/pipe/nv40/nv40_state.c | 12 ++++-------- src/mesa/pipe/nv40/nv40_state.h | 1 + src/mesa/pipe/nv40/nv40_state_emit.c | 11 ++++++----- src/mesa/pipe/nv40/nv40_state_tex.c | 36 ++++++++++++++++++++++-------------- 6 files changed, 42 insertions(+), 31 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 975b1096ccf..0a89ae8bede 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -17,7 +17,6 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV40_NEW_TEXTURE (1 << 0) #define NV40_NEW_VERTPROG (1 << 1) #define NV40_NEW_FRAGPROG (1 << 2) #define NV40_NEW_ARRAYS (1 << 3) @@ -39,8 +38,10 @@ struct nv40_context { uint32_t dirty; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct pipe_texture *tex_miptree[PIPE_MAX_SAMPLERS]; - uint32_t tex_dirty; + struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; uint32_t rt_enable; struct pipe_buffer_handle *rt[4]; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index d23cb5ee9d8..01bf5c3460b 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -223,8 +223,12 @@ nv40_fp_tex(struct nv40_fpc *fpc, int sat, int op, int unit, struct nv40_sreg dst, int mask, struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) { + struct nv40_fragment_program *fp = fpc->fp; + nv40_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); - fpc->fp->insn[fpc->inst_offset] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); + + fp->insn[fpc->inst_offset] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); + fp->samplers |= (1 << unit); } static INLINE struct nv40_sreg diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 70abc0feba0..a7a5fefc6d6 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -238,10 +238,8 @@ nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, struct nv40_context *nv40 = (struct nv40_context *)pipe; struct nv40_sampler_state *ps = hwcso; - nv40->tex_sampler[unit] = ps; - nv40->tex_dirty |= (1 << unit); - - nv40->dirty |= NV40_NEW_TEXTURE; + nv40->tex_sampler[unit] = ps; + nv40->dirty_samplers |= (1 << unit); } static void @@ -256,10 +254,8 @@ nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, { struct nv40_context *nv40 = (struct nv40_context *)pipe; - nv40->tex_miptree[unit] = miptree; - nv40->tex_dirty |= (1 << unit); - - nv40->dirty |= NV40_NEW_TEXTURE; + nv40->tex_miptree[unit] = (struct nv40_miptree *)miptree; + nv40->dirty_samplers |= (1 << unit); } static void * diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index e1a9d58525a..4c4c847c9a1 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -88,6 +88,7 @@ struct nv40_fragment_program { boolean translated; boolean on_hw; + unsigned samplers; uint32_t *insn; int insn_len; diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index fc8a0a952af..080ade3827f 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -9,18 +9,17 @@ nv40_emit_hw_state(struct nv40_context *nv40) if (nv40->dirty & NV40_NEW_FRAGPROG) { nv40_fragprog_bind(nv40, nv40->fragprog.current); - /*XXX: clear NV40_NEW_FRAGPROG if no now program uploaded */ + /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ } - if (nv40->dirty & NV40_NEW_TEXTURE) + if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { nv40_state_tex_update(nv40); - if (nv40->dirty & (NV40_NEW_TEXTURE | NV40_NEW_FRAGPROG)) { BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); OUT_RING (2); BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); OUT_RING (1); - nv40->dirty &= ~(NV40_NEW_TEXTURE | NV40_NEW_FRAGPROG); + nv40->dirty &= ~NV40_NEW_FRAGPROG; } if (nv40->dirty & NV40_NEW_VERTPROG) { @@ -28,6 +27,8 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->dirty &= ~NV40_NEW_VERTPROG; } + nv40->dirty_samplers = 0; + /* Emit relocs for every referenced buffer. * This is to ensure the bufmgr has an accurate idea of how * the buffer is used. This isn't very efficient, but we don't @@ -73,7 +74,7 @@ nv40_emit_hw_state(struct nv40_context *nv40) /* Texture images */ for (i = 0; i < 16; i++) { - if (!nv40->tex[i].buffer) + if (!(nv40->fp_samplers & (1 << i))) continue; BEGIN_RING(curie, NV40TCL_TEX_OFFSET(i), 2); OUT_RELOCl(nv40->tex[i].buffer, 0, NOUVEAU_BO_VRAM | diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 9fb274d6274..8e8609f5c95 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -60,8 +60,8 @@ static void nv40_tex_unit_enable(struct nv40_context *nv40, int unit) { struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; - struct pipe_texture *pt = nv40->tex_miptree[unit]; - struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; + struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; + struct pipe_texture *pt = &nv40mt->base; struct nv40_texture_format *tf; uint32_t txf, txs, txp; int swizzled = 0; /*XXX: implement in region code? */ @@ -127,18 +127,26 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) void nv40_state_tex_update(struct nv40_context *nv40) { - while (nv40->tex_dirty) { - int unit = ffs(nv40->tex_dirty) - 1; - - if (nv40->tex_miptree[unit]) { - nv40_tex_unit_enable(nv40, unit); - } else { - nv40->tex[unit].buffer = NULL; - BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); - OUT_RING (0); - } - - nv40->tex_dirty &= ~(1 << unit); + struct nv40_fragment_program *fp = nv40->fragprog.active; + unsigned samplers, unit; + + samplers = nv40->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv40->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv40_tex_unit_enable(nv40, unit); } + + nv40->fp_samplers = fp->samplers; } -- cgit v1.2.3 From 37e95871789780a8f5848c5e85ec123571e8dff7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 6 Jan 2008 22:14:22 +1100 Subject: nv40: do nothing for set_clip_state() There's no way possible for us to support clip planes with the current interfaces anyway. --- src/mesa/pipe/nv40/nv40_state.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index a7a5fefc6d6..14e12626d43 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -515,9 +515,6 @@ static void nv40_set_clip_state(struct pipe_context *pipe, const struct pipe_clip_state *clip) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; - - nv40->dirty |= NV40_NEW_VERTPROG; } static void -- cgit v1.2.3 From 9bac4c7268cdbc79a1ce2d05779ed05c32e1e57b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 7 Jan 2008 01:02:30 +1100 Subject: nouveau: share pushbuf interface code between pipe drivers --- src/mesa/pipe/nouveau/nouveau_push.h | 75 ++++++++++++++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_clear.c | 2 - src/mesa/pipe/nv40/nv40_context.c | 1 - src/mesa/pipe/nv40/nv40_context.h | 4 ++ src/mesa/pipe/nv40/nv40_fragprog.c | 1 - src/mesa/pipe/nv40/nv40_query.c | 1 - src/mesa/pipe/nv40/nv40_state.c | 1 - src/mesa/pipe/nv40/nv40_state_emit.c | 1 - src/mesa/pipe/nv40/nv40_state_tex.c | 1 - src/mesa/pipe/nv40/nv40_vbo.c | 1 - src/mesa/pipe/nv40/nv40_vertprog.c | 1 - src/mesa/pipe/nv50/nv50_clear.c | 2 - src/mesa/pipe/nv50/nv50_context.c | 1 - src/mesa/pipe/nv50/nv50_context.h | 4 ++ src/mesa/pipe/nv50/nv50_dma.h | 66 ------------------------------- src/mesa/pipe/nv50/nv50_query.c | 1 - src/mesa/pipe/nv50/nv50_state.c | 3 +- src/mesa/pipe/nv50/nv50_vbo.c | 1 - 18 files changed, 85 insertions(+), 82 deletions(-) create mode 100644 src/mesa/pipe/nouveau/nouveau_push.h delete mode 100644 src/mesa/pipe/nv50/nv50_dma.h diff --git a/src/mesa/pipe/nouveau/nouveau_push.h b/src/mesa/pipe/nouveau/nouveau_push.h new file mode 100644 index 00000000000..0fc972d2a22 --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_push.h @@ -0,0 +1,75 @@ +#ifndef __NOUVEAU_PUSH_H__ +#define __NOUVEAU_PUSH_H__ + +#include "pipe/nouveau/nouveau_winsys.h" + +#ifndef NOUVEAU_PUSH_CONTEXT +#error undefined push context +#endif + +#define OUT_RING(data) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + (*pc->nvws->channel->pushbuf->cur++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + memcpy(pc->nvws->channel->pushbuf->cur, (src), (size) * 4); \ + pc->nvws->channel->pushbuf->cur += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define BEGIN_RING(obj,mthd,size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + OUT_RING((pc->obj->subc << 13) | ((size) << 18) | (mthd)); \ + pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + +#define FIRE_RING() do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + pc->nvws->push_flush(pc->nvws->channel, 0); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + pc->nvws->push_reloc(pc->nvws->channel, \ + pc->nvws->channel->pushbuf->cur, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ + OUT_RING(0); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + pc->nvws->channel->vram->handle, \ + pc->nvws->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/mesa/pipe/nv40/nv40_clear.c b/src/mesa/pipe/nv40/nv40_clear.c index 380a2a642fc..2c4e8f01fda 100644 --- a/src/mesa/pipe/nv40/nv40_clear.c +++ b/src/mesa/pipe/nv40/nv40_clear.c @@ -3,8 +3,6 @@ #include "pipe/p_state.h" #include "nv40_context.h" -#include "nv40_dma.h" - void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 442ff04e753..334e50ef892 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -4,7 +4,6 @@ #include "pipe/p_util.h" #include "nv40_context.h" -#include "nv40_dma.h" static const char * nv40_get_name(struct pipe_context *pipe) diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 0a89ae8bede..86aebbb927a 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -10,6 +10,10 @@ #include "pipe/nouveau/nouveau_winsys.h" #include "pipe/nouveau/nouveau_gldefs.h" +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv40_context *ctx = nv40 +#include "pipe/nouveau/nouveau_push.h" + #include "nv40_state.h" #define NOUVEAU_ERR(fmt, args...) \ diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 01bf5c3460b..fded11d1703 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -7,7 +7,6 @@ #include "pipe/tgsi/util/tgsi_util.h" #include "nv40_context.h" -#include "nv40_dma.h" #define SWZ_X 0 #define SWZ_Y 1 diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index 6e5fcae8cad..ea30d3ede84 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -1,7 +1,6 @@ #include "pipe/p_context.h" #include "nv40_context.h" -#include "nv40_dma.h" struct nv40_query { struct nouveau_resource *object; diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 14e12626d43..41de5650aa8 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -3,7 +3,6 @@ #include "pipe/p_util.h" #include "nv40_context.h" -#include "nv40_dma.h" #include "nv40_state.h" static void * diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 080ade3827f..7bb69e617ee 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -1,5 +1,4 @@ #include "nv40_context.h" -#include "nv40_dma.h" #include "nv40_state.h" void diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 8e8609f5c95..7c353c138fc 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -1,5 +1,4 @@ #include "nv40_context.h" -#include "nv40_dma.h" #define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ { \ diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index d462e984003..7cf089e007a 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -3,7 +3,6 @@ #include "pipe/p_util.h" #include "nv40_context.h" -#include "nv40_dma.h" #include "nv40_state.h" #include "pipe/nouveau/nouveau_channel.h" diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index 84618c1e1ba..bd052fe8341 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -6,7 +6,6 @@ #include "pipe/tgsi/util/tgsi_parse.h" #include "nv40_context.h" -#include "nv40_dma.h" #include "nv40_state.h" /* TODO (at least...): diff --git a/src/mesa/pipe/nv50/nv50_clear.c b/src/mesa/pipe/nv50/nv50_clear.c index 2b453a49d15..552b92f72e2 100644 --- a/src/mesa/pipe/nv50/nv50_clear.c +++ b/src/mesa/pipe/nv50/nv50_clear.c @@ -3,8 +3,6 @@ #include "pipe/p_state.h" #include "nv50_context.h" -#include "nv50_dma.h" - void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, diff --git a/src/mesa/pipe/nv50/nv50_context.c b/src/mesa/pipe/nv50/nv50_context.c index ed12c6c0d93..33c8eebb0bc 100644 --- a/src/mesa/pipe/nv50/nv50_context.c +++ b/src/mesa/pipe/nv50/nv50_context.c @@ -4,7 +4,6 @@ #include "pipe/p_util.h" #include "nv50_context.h" -#include "nv50_dma.h" static boolean nv50_is_format_supported(struct pipe_context *pipe, enum pipe_format format, diff --git a/src/mesa/pipe/nv50/nv50_context.h b/src/mesa/pipe/nv50/nv50_context.h index b1c4d5b8948..c43202c3d57 100644 --- a/src/mesa/pipe/nv50/nv50_context.h +++ b/src/mesa/pipe/nv50/nv50_context.h @@ -10,6 +10,10 @@ #include "pipe/nouveau/nouveau_winsys.h" #include "pipe/nouveau/nouveau_gldefs.h" +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv50_context *ctx = nv50 +#include "pipe/nouveau/nouveau_push.h" + #include "nv50_state.h" #define NOUVEAU_ERR(fmt, args...) \ diff --git a/src/mesa/pipe/nv50/nv50_dma.h b/src/mesa/pipe/nv50/nv50_dma.h deleted file mode 100644 index 366316df755..00000000000 --- a/src/mesa/pipe/nv50/nv50_dma.h +++ /dev/null @@ -1,66 +0,0 @@ -#ifndef __NV50_DMA_H__ -#define __NV50_DMA_H__ - -#include "pipe/nouveau/nouveau_winsys.h" - -#define OUT_RING(data) do { \ - (*nv50->nvws->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - memcpy(nv50->nvws->channel->pushbuf->cur, (src), (size) * 4); \ - nv50->nvws->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - if (nv50->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - nv50->nvws->push_flush(nv50->nvws->channel, ((size) + 1)); \ - OUT_RING((nv50->obj->subc << 13) | ((size) << 18) | (mthd)); \ - nv50->nvws->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - -#define FIRE_RING() do { \ - nv50->nvws->push_flush(nv50->nvws->channel, 0); \ -} while(0) - -#define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nv50->nvws->push_reloc(nv50->nvws->channel, \ - nv50->nvws->channel->pushbuf->cur, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ -} while(0) - -/* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) - -/* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - nv50->nvws->channel->vram->handle, \ - nv50->nvws->channel->gart->handle); \ -} while(0) - -/* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) - -/* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) - -#endif diff --git a/src/mesa/pipe/nv50/nv50_query.c b/src/mesa/pipe/nv50/nv50_query.c index 69eb0e700b4..7e48e5121c2 100644 --- a/src/mesa/pipe/nv50/nv50_query.c +++ b/src/mesa/pipe/nv50/nv50_query.c @@ -1,7 +1,6 @@ #include "pipe/p_context.h" #include "nv50_context.h" -#include "nv50_dma.h" static struct pipe_query * nv50_query_create(struct pipe_context *pipe, unsigned type) diff --git a/src/mesa/pipe/nv50/nv50_state.c b/src/mesa/pipe/nv50/nv50_state.c index 25eac41c2fc..99dcab51b26 100644 --- a/src/mesa/pipe/nv50/nv50_state.c +++ b/src/mesa/pipe/nv50/nv50_state.c @@ -3,7 +3,6 @@ #include "pipe/p_util.h" #include "nv50_context.h" -#include "nv50_dma.h" #include "nv50_state.h" static void * @@ -85,6 +84,7 @@ static void * nv50_vp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { + return NULL; } static void @@ -101,6 +101,7 @@ static void * nv50_fp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { + return NULL; } static void diff --git a/src/mesa/pipe/nv50/nv50_vbo.c b/src/mesa/pipe/nv50/nv50_vbo.c index faae08f6014..0c9d2806d78 100644 --- a/src/mesa/pipe/nv50/nv50_vbo.c +++ b/src/mesa/pipe/nv50/nv50_vbo.c @@ -3,7 +3,6 @@ #include "pipe/p_util.h" #include "nv50_context.h" -#include "nv50_dma.h" #include "nv50_state.h" boolean -- cgit v1.2.3 From a50f224f41faaf4ba75902d2dd46f1fe595f098f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 7 Jan 2008 04:36:22 +1100 Subject: nv40: use a more "normal" 2x8bit format, the other is likely a HILO format. --- src/mesa/pipe/nv40/nv40_state_tex.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index 7c353c138fc..f54a65c7809 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -13,6 +13,7 @@ #define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x1200 #define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x1000 +#define NV40TCL_TEX_FORMAT_FORMAT_A8L8_2 0x0b00 struct nv40_texture_format { boolean defined; @@ -30,7 +31,7 @@ nv40_texture_formats[] = { _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, Z, W, X, Y), + _(U_A8_L8 , A8L8_2 , S1, S1, S1, S1, X, X, X, Y), _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), // _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), @@ -105,8 +106,6 @@ nv40_tex_unit_enable(struct nv40_context *nv40, int unit) } txs = tf->swizzle; - if (pt->format == PIPE_FORMAT_U_A8_L8) - txs |= (1<<16); /*nfi*/ nv40->tex[unit].buffer = nv40mt->buffer; nv40->tex[unit].format = txf; -- cgit v1.2.3 From 9de335f5807553bc9251931ba3e80c4b454b5818 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 8 Jan 2008 02:28:43 +1100 Subject: nouveau: object header update --- src/mesa/pipe/nouveau/nouveau_class.h | 29 +++++++++++++++++++++++++++-- src/mesa/pipe/nv40/nv40_state_tex.c | 6 +----- 2 files changed, 28 insertions(+), 7 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h index 1f2c52891d9..e3c284095de 100644 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -3672,7 +3672,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. -#define NV36TCL 0x00000497 +#define NV35TCL 0x00000497 @@ -5070,7 +5070,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_TEX_FORMAT_FORMAT_DXT1 0x00000600 #define NV40TCL_TEX_FORMAT_FORMAT_DXT3 0x00000700 #define NV40TCL_TEX_FORMAT_FORMAT_DXT5 0x00000800 -#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00001800 +#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00000b00 +#define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x00001000 +#define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x00001200 +#define NV40TCL_TEX_FORMAT_FORMAT_HILO8 0x00001800 #define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 #define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 #define NV40TCL_TEX_FORMAT_DIMS_1D 0x00000010 @@ -5183,6 +5186,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_TEX_SWIZZLE_S1_W_X 0x00000003 #define NV40TCL_TEX_FILTER(x) (0x00001a14+((x)*32)) #define NV40TCL_TEX_FILTER__SIZE 0x00000010 +#define NV40TCL_TEX_FILTER_SIGNED_ALPHA (1 << 31) +#define NV40TCL_TEX_FILTER_SIGNED_RED (1 << 30) +#define NV40TCL_TEX_FILTER_SIGNED_GREEN (1 << 29) +#define NV40TCL_TEX_FILTER_SIGNED_BLUE (1 << 28) #define NV40TCL_TEX_FILTER_MIN_SHIFT 16 #define NV40TCL_TEX_FILTER_MIN_MASK 0x000f0000 #define NV40TCL_TEX_FILTER_MIN_NEAREST 0x00010000 @@ -5333,6 +5340,24 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50_2D_BLIT_DST_H 0x000008bc #define NV50_2D_BLIT_SRC_X 0x000008d4 #define NV50_2D_BLIT_SRC_Y 0x000008dc +#define NV50_2D_SIFC_UNK0800 0x00000800 +#define NV50_2D_SIFC_FORMAT 0x00000804 +#define NV50_2D_SIFC_FORMAT_32BPP 0x000000cf +#define NV50_2D_SIFC_FORMAT_24BPP 0x000000e6 +#define NV50_2D_SIFC_FORMAT_16BPP 0x000000e8 +#define NV50_2D_SIFC_FORMAT_8BPP 0x000000f3 +#define NV50_2D_SIFC_FORMAT_15BPP 0x000000f8 +#define NV50_2D_SIFC_WIDTH 0x00000838 +#define NV50_2D_SIFC_HEIGHT 0x0000083c +#define NV50_2D_SIFC_SCALE_UNK0840 0x00000840 +#define NV50_2D_SIFC_SCALE_UNK0844 0x00000844 +#define NV50_2D_SIFC_SCALE_UNK0848 0x00000848 +#define NV50_2D_SIFC_SCALE_UNK084C 0x0000084c +#define NV50_2D_SIFC_UNK0850 0x00000850 +#define NV50_2D_SIFC_DST_X 0x00000854 +#define NV50_2D_SIFC_UNK0858 0x00000858 +#define NV50_2D_SIFC_DST_Y 0x0000085c +#define NV50_2D_SIFC_DATA 0x00000860 #define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c index f54a65c7809..f2303e34db3 100644 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ b/src/mesa/pipe/nv40/nv40_state_tex.c @@ -11,10 +11,6 @@ NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ } -#define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x1200 -#define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x1000 -#define NV40TCL_TEX_FORMAT_FORMAT_A8L8_2 0x0b00 - struct nv40_texture_format { boolean defined; uint pipe; @@ -31,7 +27,7 @@ nv40_texture_formats[] = { _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8_2 , S1, S1, S1, S1, X, X, X, Y), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), // _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -- cgit v1.2.3 From 545471018e8325bd4c3283728c596f307071a685 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 16 Jan 2008 13:18:46 +1100 Subject: nouveau: gallium changes.. --- src/mesa/pipe/nv40/nv40_surface.c | 2 -- src/mesa/pipe/nv50/nv50_surface.c | 2 -- 2 files changed, 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index ed144c636c7..91ac9e36948 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -146,8 +146,6 @@ nv40_init_surface_functions(struct nv40_context *nv40) nv40->pipe.get_tex_surface = nv40_get_tex_surface; nv40->pipe.get_tile = pipe_get_tile_raw; nv40->pipe.put_tile = pipe_put_tile_raw; - nv40->pipe.get_tile_rgba = pipe_get_tile_rgba; - nv40->pipe.put_tile_rgba = pipe_put_tile_rgba; nv40->pipe.surface_data = nv40_surface_data; nv40->pipe.surface_copy = nv40_surface_copy; nv40->pipe.surface_fill = nv40_surface_fill; diff --git a/src/mesa/pipe/nv50/nv50_surface.c b/src/mesa/pipe/nv50/nv50_surface.c index 44e6728b264..d2079946889 100644 --- a/src/mesa/pipe/nv50/nv50_surface.c +++ b/src/mesa/pipe/nv50/nv50_surface.c @@ -84,8 +84,6 @@ nv50_init_surface_functions(struct nv50_context *nv50) nv50->pipe.get_tex_surface = nv50_get_tex_surface; nv50->pipe.get_tile = pipe_get_tile_raw; nv50->pipe.put_tile = pipe_put_tile_raw; - nv50->pipe.get_tile_rgba = pipe_get_tile_rgba; - nv50->pipe.put_tile_rgba = pipe_put_tile_rgba; nv50->pipe.surface_data = nv50_surface_data; nv50->pipe.surface_copy = nv50_surface_copy; nv50->pipe.surface_fill = nv50_surface_fill; -- cgit v1.2.3 From bc1ca3793d661de343b956d66ef70605d7de2dbb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 16 Jan 2008 13:37:13 +1100 Subject: nouveau: don't use GL headers at all --- src/mesa/pipe/nouveau/nouveau_gldefs.h | 174 ++++++++++++++++----------------- 1 file changed, 86 insertions(+), 88 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_gldefs.h b/src/mesa/pipe/nouveau/nouveau_gldefs.h index 15ff318023d..556a1ec58b4 100644 --- a/src/mesa/pipe/nouveau/nouveau_gldefs.h +++ b/src/mesa/pipe/nouveau/nouveau_gldefs.h @@ -1,44 +1,42 @@ -#ifndef __NVGL_PIPE_H__ -#define __NVGL_PIPE_H__ - -#include +#ifndef __NOUVEAU_GLDEFS_H__ +#define __NOUVEAU_GLDEFS_H__ static INLINE unsigned nvgl_blend_func(unsigned factor) { switch (factor) { + case PIPE_BLENDFACTOR_ZERO: + return 0x0000; case PIPE_BLENDFACTOR_ONE: - return GL_ONE; + return 0x0001; case PIPE_BLENDFACTOR_SRC_COLOR: - return GL_SRC_COLOR; + return 0x0300; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + return 0x0301; case PIPE_BLENDFACTOR_SRC_ALPHA: - return GL_SRC_ALPHA; + return 0x0302; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + return 0x0303; case PIPE_BLENDFACTOR_DST_ALPHA: - return GL_DST_ALPHA; + return 0x0304; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + return 0x0305; case PIPE_BLENDFACTOR_DST_COLOR: - return GL_DST_COLOR; + return 0x0306; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + return 0x0307; case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - return GL_SRC_ALPHA_SATURATE; + return 0x0308; case PIPE_BLENDFACTOR_CONST_COLOR: - return GL_CONSTANT_COLOR; - case PIPE_BLENDFACTOR_CONST_ALPHA: - return GL_CONSTANT_ALPHA; - case PIPE_BLENDFACTOR_ZERO: - return GL_ZERO; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - return GL_ONE_MINUS_SRC_COLOR; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - return GL_ONE_MINUS_SRC_ALPHA; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - return GL_ONE_MINUS_DST_ALPHA; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - return GL_ONE_MINUS_DST_COLOR; + return 0x8001; case PIPE_BLENDFACTOR_INV_CONST_COLOR: - return GL_ONE_MINUS_CONSTANT_COLOR; + return 0x8002; + case PIPE_BLENDFACTOR_CONST_ALPHA: + return 0x8003; case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - return GL_ONE_MINUS_CONSTANT_ALPHA; + return 0x8004; default: - return GL_ONE; + assert(0); } } @@ -47,17 +45,17 @@ nvgl_blend_eqn(unsigned func) { switch (func) { case PIPE_BLEND_ADD: - return GL_FUNC_ADD; - case PIPE_BLEND_SUBTRACT: - return GL_FUNC_SUBTRACT; - case PIPE_BLEND_REVERSE_SUBTRACT: - return GL_FUNC_REVERSE_SUBTRACT; + return 0x8006; case PIPE_BLEND_MIN: - return GL_MIN; + return 0x8007; case PIPE_BLEND_MAX: - return GL_MAX; + return 0x8008; + case PIPE_BLEND_SUBTRACT: + return 0x800a; + case PIPE_BLEND_REVERSE_SUBTRACT: + return 0x800b; default: - return GL_FUNC_ADD; + assert(0); } } @@ -66,39 +64,39 @@ nvgl_logicop_func(unsigned func) { switch (func) { case PIPE_LOGICOP_CLEAR: - return GL_CLEAR; + return 0x1500; case PIPE_LOGICOP_NOR: - return GL_NOR; + return 0x1508; case PIPE_LOGICOP_AND_INVERTED: - return GL_AND_INVERTED; + return 0x1504; case PIPE_LOGICOP_COPY_INVERTED: - return GL_COPY_INVERTED; + return 0x150c; case PIPE_LOGICOP_AND_REVERSE: - return GL_AND_REVERSE; + return 0x1502; case PIPE_LOGICOP_INVERT: - return GL_INVERT; + return 0x150a; case PIPE_LOGICOP_XOR: - return GL_XOR; + return 0x1506; case PIPE_LOGICOP_NAND: - return GL_NAND; + return 0x150e; case PIPE_LOGICOP_AND: - return GL_AND; + return 0x1501; case PIPE_LOGICOP_EQUIV: - return GL_EQUIV; + return 0x1509; case PIPE_LOGICOP_NOOP: - return GL_NOOP; + return 0x1505; case PIPE_LOGICOP_OR_INVERTED: - return GL_OR_INVERTED; + return 0x150d; case PIPE_LOGICOP_COPY: - return GL_COPY; + return 0x1503; case PIPE_LOGICOP_OR_REVERSE: - return GL_OR_REVERSE; + return 0x150b; case PIPE_LOGICOP_OR: - return GL_OR; + return 0x1507; case PIPE_LOGICOP_SET: - return GL_SET; + return 0x150f; default: - return GL_CLEAR; + assert(0); } } @@ -107,23 +105,23 @@ nvgl_comparison_op(unsigned op) { switch (op) { case PIPE_FUNC_NEVER: - return GL_NEVER; + return 0x0200; case PIPE_FUNC_LESS: - return GL_LESS; + return 0x0201; case PIPE_FUNC_EQUAL: - return GL_EQUAL; + return 0x0202; case PIPE_FUNC_LEQUAL: - return GL_LEQUAL; + return 0x0203; case PIPE_FUNC_GREATER: - return GL_GREATER; + return 0x0204; case PIPE_FUNC_NOTEQUAL: - return GL_NOTEQUAL; + return 0x0205; case PIPE_FUNC_GEQUAL: - return GL_GEQUAL; + return 0x0206; case PIPE_FUNC_ALWAYS: - return GL_ALWAYS; + return 0x0207; default: - return GL_NEVER; + assert(0); } } @@ -131,14 +129,14 @@ static INLINE unsigned nvgl_polygon_mode(unsigned mode) { switch (mode) { - case PIPE_POLYGON_MODE_FILL: - return GL_FILL; - case PIPE_POLYGON_MODE_LINE: - return GL_LINE; case PIPE_POLYGON_MODE_POINT: - return GL_POINT; + return 0x1b00; + case PIPE_POLYGON_MODE_LINE: + return 0x1b01; + case PIPE_POLYGON_MODE_FILL: + return 0x1b02; default: - return GL_FILL; + assert(0); } } @@ -146,24 +144,24 @@ static INLINE unsigned nvgl_stencil_op(unsigned op) { switch (op) { - case PIPE_STENCIL_OP_KEEP: - return GL_KEEP; case PIPE_STENCIL_OP_ZERO: - return GL_ZERO; + return 0x0000; + case PIPE_STENCIL_OP_INVERT: + return 0x150a; + case PIPE_STENCIL_OP_KEEP: + return 0x1e00; case PIPE_STENCIL_OP_REPLACE: - return GL_REPLACE; + return 0x1e01; case PIPE_STENCIL_OP_INCR: - return GL_INCR; + return 0x1e02; case PIPE_STENCIL_OP_DECR: - return GL_DECR; + return 0x1e03; case PIPE_STENCIL_OP_INCR_WRAP: - return GL_INCR_WRAP; + return 0x8507; case PIPE_STENCIL_OP_DECR_WRAP: - return GL_DECR_WRAP; - case PIPE_STENCIL_OP_INVERT: - return GL_INVERT; + return 0x8508; default: - return GL_KEEP; + assert(0); } } @@ -171,27 +169,27 @@ static INLINE unsigned nvgl_primitive(unsigned prim) { switch (prim) { case PIPE_PRIM_POINTS: - return GL_POINTS + 1; + return 0x0001; case PIPE_PRIM_LINES: - return GL_LINES + 1; + return 0x0002; case PIPE_PRIM_LINE_LOOP: - return GL_LINE_LOOP + 1; + return 0x0003; case PIPE_PRIM_LINE_STRIP: - return GL_LINE_STRIP + 1; + return 0x0004; case PIPE_PRIM_TRIANGLES: - return GL_TRIANGLES + 1; + return 0x0005; case PIPE_PRIM_TRIANGLE_STRIP: - return GL_TRIANGLE_STRIP + 1; + return 0x0006; case PIPE_PRIM_TRIANGLE_FAN: - return GL_TRIANGLE_FAN + 1; + return 0x0007; case PIPE_PRIM_QUADS: - return GL_QUADS + 1; + return 0x0008; case PIPE_PRIM_QUAD_STRIP: - return GL_QUAD_STRIP + 1; + return 0x0009; case PIPE_PRIM_POLYGON: - return GL_POLYGON + 1; + return 0x000a; default: - return GL_POINTS + 1; + assert(0); } } -- cgit v1.2.3 From e0e9326bdadb268dbb5eeedb14404ac2f48c103e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 16 Jan 2008 13:41:50 +1100 Subject: nv40: stronger type safety --- src/mesa/pipe/nv40/nv40_context.c | 6 +++--- src/mesa/pipe/nv40/nv40_context.h | 7 ++++++- src/mesa/pipe/nv40/nv40_query.c | 13 +++++++++---- src/mesa/pipe/nv40/nv40_state.c | 34 +++++++++++++++++----------------- src/mesa/pipe/nv40/nv40_surface.c | 6 +++--- src/mesa/pipe/nv40/nv40_vbo.c | 6 +++--- 6 files changed, 41 insertions(+), 31 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 334e50ef892..f6fdd6b4376 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -8,7 +8,7 @@ static const char * nv40_get_name(struct pipe_context *pipe) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); static char buffer[128]; snprintf(buffer, sizeof(buffer), "NV%02X", nv40->chipset); @@ -80,7 +80,7 @@ nv40_get_paramf(struct pipe_context *pipe, int param) static void nv40_flush(struct pipe_context *pipe, unsigned flags) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_winsys *nvws = nv40->nvws; if (flags & PIPE_FLUSH_TEXTURE_CACHE) { @@ -107,7 +107,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) static void nv40_destroy(struct pipe_context *pipe) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_winsys *nvws = nv40->nvws; if (nv40->draw) diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 86aebbb927a..b505efab54e 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -82,7 +82,12 @@ struct nv40_context { struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; -#define nv40_context(ctx) ((struct nv40_context *)(ctx)) + +static inline struct nv40_context * +nv40_context(struct pipe_context *pipe) +{ + return (struct nv40_context *)pipe; +} extern void nv40_init_state_functions(struct nv40_context *nv40); extern void nv40_init_surface_functions(struct nv40_context *nv40); diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index ea30d3ede84..06f41fe84f7 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -8,7 +8,12 @@ struct nv40_query { boolean ready; uint64_t result; }; -#define nv40_query(o) ((struct nv40_query *)(o)) + +static inline struct nv40_query * +nv40_query(struct pipe_query *pipe) +{ + return (struct nv40_query *)pipe; +} static struct pipe_query * nv40_query_create(struct pipe_context *pipe, unsigned query_type) @@ -55,7 +60,7 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) static void nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_query *q = nv40_query(pq); BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); @@ -66,9 +71,9 @@ nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq) static boolean nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, - boolean wait, uint64_t *result) + boolean wait, uint64 *result) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_query *q = nv40_query(pq); struct nouveau_winsys *nvws = nv40->nvws; diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 41de5650aa8..03a9e735c07 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -37,7 +37,7 @@ nv40_blend_state_create(struct pipe_context *pipe, static void nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_blend_state *cb = hwcso; BEGIN_RING(curie, NV40TCL_DITHER_ENABLE, 1); @@ -234,7 +234,7 @@ static void nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_sampler_state *ps = hwcso; nv40->tex_sampler[unit] = ps; @@ -251,7 +251,7 @@ static void nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, struct pipe_texture *miptree) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); nv40->tex_miptree[unit] = (struct nv40_miptree *)miptree; nv40->dirty_samplers |= (1 << unit); @@ -339,7 +339,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_rasterizer_state *rs = hwcso; BEGIN_RING(curie, NV40TCL_SHADE_MODEL, 1); @@ -415,7 +415,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, static void nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_depth_stencil_alpha_state *hw = hwcso; BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); @@ -448,7 +448,7 @@ nv40_vp_state_create(struct pipe_context *pipe, static void nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_vertex_program *vp = hwcso; nv40->vertprog.current = vp; @@ -458,7 +458,7 @@ nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) static void nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_vertex_program *vp = hwcso; nv40_vertprog_destroy(nv40, vp); @@ -480,7 +480,7 @@ nv40_fp_state_create(struct pipe_context *pipe, static void nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_fragment_program *fp = hwcso; nv40->fragprog.current = fp; @@ -490,7 +490,7 @@ nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) static void nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_fragment_program *fp = hwcso; nv40_fragprog_destroy(nv40, fp); @@ -501,7 +501,7 @@ static void nv40_set_blend_color(struct pipe_context *pipe, const struct pipe_blend_color *bcol) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); BEGIN_RING(curie, NV40TCL_BLEND_COLOR, 1); OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | @@ -520,7 +520,7 @@ static void nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, const struct pipe_constant_buffer *buf ) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); if (shader == PIPE_SHADER_VERTEX) { nv40->vertprog.constant_buf = buf->buffer; @@ -536,7 +536,7 @@ static void nv40_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct pipe_surface *rt[4], *zeta; uint32_t rt_enable, rt_format, w, h; int i, colour_format = 0, zeta_format = 0; @@ -669,7 +669,7 @@ static void nv40_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); OUT_RINGp ((uint32_t *)stipple->stipple, 32); @@ -679,7 +679,7 @@ static void nv40_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); BEGIN_RING(curie, NV40TCL_SCISSOR_HORIZ, 2); OUT_RING (((s->maxx - s->minx) << 16) | s->minx); @@ -690,7 +690,7 @@ static void nv40_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); BEGIN_RING(curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); OUT_RINGf (vpt->translate[0]); @@ -707,7 +707,7 @@ static void nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, const struct pipe_vertex_buffer *vb) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); nv40->vtxbuf[index] = *vb; @@ -718,7 +718,7 @@ static void nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, const struct pipe_vertex_element *ve) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); nv40->vtxelt[index] = *ve; diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index 91ac9e36948..159837388f3 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -109,7 +109,7 @@ nv40_surface_data(struct pipe_context *pipe, struct pipe_surface *dest, unsigned src_stride, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_winsys *nvws = nv40->nvws; nvws->surface_data(nvws, dest, destx, desty, src, src_stride, @@ -121,7 +121,7 @@ nv40_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, unsigned destx, unsigned desty, struct pipe_surface *src, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_winsys *nvws = nv40->nvws; nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, @@ -133,7 +133,7 @@ nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, unsigned destx, unsigned desty, unsigned width, unsigned height, unsigned value) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_winsys *nvws = nv40->nvws; nvws->surface_fill(nvws, dest, destx, desty, width, height, value); diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 7cf089e007a..3de87ecab5e 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -192,7 +192,7 @@ boolean nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); unsigned nr; assert(nv40_vbo_validate_state(nv40, NULL, 0)); @@ -300,7 +300,7 @@ nv40_draw_elements_inline(struct pipe_context *pipe, struct pipe_buffer_handle *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); struct pipe_winsys *ws = pipe->winsys; void *map; @@ -341,7 +341,7 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, struct pipe_buffer_handle *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { - struct nv40_context *nv40 = (struct nv40_context *)pipe; + struct nv40_context *nv40 = nv40_context(pipe); unsigned nr, type; switch (ib_size) { -- cgit v1.2.3 From 0feec292ddc279998a1e25c10ea70d211f7b4b62 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 16 Jan 2008 13:50:00 +1100 Subject: nouveau: remove assertions from PIPE_ -> GL_ It seems we need to keep some sane defaults around for a bit, at least until the pipe drivers learn to avoid translating things like alpha_src_factor when blend_enable is false. --- src/mesa/pipe/nouveau/nouveau_gldefs.h | 12 ++++++------ src/mesa/pipe/nv50/nv50_query.c | 2 +- 2 files changed, 7 insertions(+), 7 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_gldefs.h b/src/mesa/pipe/nouveau/nouveau_gldefs.h index 556a1ec58b4..8ba3bdef649 100644 --- a/src/mesa/pipe/nouveau/nouveau_gldefs.h +++ b/src/mesa/pipe/nouveau/nouveau_gldefs.h @@ -36,7 +36,7 @@ nvgl_blend_func(unsigned factor) case PIPE_BLENDFACTOR_INV_CONST_ALPHA: return 0x8004; default: - assert(0); + return 0x0000; } } @@ -55,7 +55,7 @@ nvgl_blend_eqn(unsigned func) case PIPE_BLEND_REVERSE_SUBTRACT: return 0x800b; default: - assert(0); + return 0x8006; } } @@ -96,7 +96,7 @@ nvgl_logicop_func(unsigned func) case PIPE_LOGICOP_SET: return 0x150f; default: - assert(0); + return 0x1505; } } @@ -121,7 +121,7 @@ nvgl_comparison_op(unsigned op) case PIPE_FUNC_ALWAYS: return 0x0207; default: - assert(0); + return 0x0207; } } @@ -136,7 +136,7 @@ nvgl_polygon_mode(unsigned mode) case PIPE_POLYGON_MODE_FILL: return 0x1b02; default: - assert(0); + return 0x1b02; } } @@ -161,7 +161,7 @@ nvgl_stencil_op(unsigned op) case PIPE_STENCIL_OP_DECR_WRAP: return 0x8508; default: - assert(0); + return 0x1e00; } } diff --git a/src/mesa/pipe/nv50/nv50_query.c b/src/mesa/pipe/nv50/nv50_query.c index 7e48e5121c2..d8c3491c2c5 100644 --- a/src/mesa/pipe/nv50/nv50_query.c +++ b/src/mesa/pipe/nv50/nv50_query.c @@ -29,7 +29,7 @@ nv50_query_end(struct pipe_context *pipe, struct pipe_query *q) static boolean nv50_query_result(struct pipe_context *pipe, struct pipe_query *q, - boolean wait, uint64_t *result) + boolean wait, uint64 *result) { NOUVEAU_ERR("unimplemented\n"); *result = 0xdeadcafe; -- cgit v1.2.3 From b56b3035f1446ef06bb7d2fc435344a573cdb3c9 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 20 Jan 2008 11:13:08 +1100 Subject: nv40: clarify that tex code is fp-only, we'll do vertex textures eventually. --- src/mesa/pipe/nv40/Makefile | 2 +- src/mesa/pipe/nv40/nv40_context.h | 3 + src/mesa/pipe/nv40/nv40_fragtex.c | 144 ++++++++++++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_state_emit.c | 2 +- src/mesa/pipe/nv40/nv40_state_tex.c | 146 ----------------------------------- 5 files changed, 149 insertions(+), 148 deletions(-) create mode 100644 src/mesa/pipe/nv40/nv40_fragtex.c delete mode 100644 src/mesa/pipe/nv40/nv40_state_tex.c diff --git a/src/mesa/pipe/nv40/Makefile b/src/mesa/pipe/nv40/Makefile index 9818889ee32..c9c3a8032e5 100644 --- a/src/mesa/pipe/nv40/Makefile +++ b/src/mesa/pipe/nv40/Makefile @@ -8,11 +8,11 @@ DRIVER_SOURCES = \ nv40_context.c \ nv40_draw.c \ nv40_fragprog.c \ + nv40_fragtex.c \ nv40_miptree.c \ nv40_query.c \ nv40_state.c \ nv40_state_emit.c \ - nv40_state_tex.c \ nv40_surface.c \ nv40_vbo.c \ nv40_vertprog.c diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index b505efab54e..df8489f3a6b 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -113,6 +113,9 @@ extern void nv40_fragprog_bind(struct nv40_context *, extern void nv40_fragprog_destroy(struct nv40_context *, struct nv40_fragment_program *); +/* nv40_fragtex.c */ +extern void nv40_fragtex_bind(struct nv40_context *); + /* nv40_state.c and friends */ extern void nv40_emit_hw_state(struct nv40_context *nv40); extern void nv40_state_tex_update(struct nv40_context *nv40); diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c new file mode 100644 index 00000000000..48d6eb629f5 --- /dev/null +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -0,0 +1,144 @@ +#include "nv40_context.h" + +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV40TCL_TEX_FORMAT_FORMAT_##tf, \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x | NV40TCL_TEX_SWIZZLE_S0_Y_##ts0y | \ + NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w | \ + NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y | \ + NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ +} + +struct nv40_texture_format { + boolean defined; + uint pipe; + int format; + int swizzle; +}; + +static struct nv40_texture_format +nv40_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), + _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), + _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), + _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), + _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), +// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + {}, +}; + +static struct nv40_texture_format * +nv40_fragtex_format(uint pipe_format) +{ + struct nv40_texture_format *tf = nv40_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv40_fragtex_build(struct nv40_context *nv40, int unit) +{ + struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; + struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; + struct pipe_texture *pt = &nv40mt->base; + struct nv40_texture_format *tf; + uint32_t txf, txs, txp; + int swizzled = 0; /*XXX: implement in region code? */ + + tf = nv40_fragtex_format(pt->format); + if (!tf) + assert(0); + + txf = ps->fmt; + txf |= tf->format | 0x8000; + txf |= ((pt->last_level - pt->first_level + 1) << + NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); + + if (1) /* XXX */ + txf |= NV40TCL_TEX_FORMAT_NO_BORDER; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE: + txf |= NV40TCL_TEX_FORMAT_CUBIC; + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= NV40TCL_TEX_FORMAT_DIMS_2D; + break; + case PIPE_TEXTURE_3D: + txf |= NV40TCL_TEX_FORMAT_DIMS_3D; + break; + case PIPE_TEXTURE_1D: + txf |= NV40TCL_TEX_FORMAT_DIMS_1D; + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + if (swizzled) { + txp = 0; + } else { + txp = nv40mt->level[0].pitch; + txf |= NV40TCL_TEX_FORMAT_LINEAR; + } + + txs = tf->swizzle; + + nv40->tex[unit].buffer = nv40mt->buffer; + nv40->tex[unit].format = txf; + + BEGIN_RING(curie, NV40TCL_TEX_WRAP(unit), 6); + OUT_RING (ps->wrap); + OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | + (0x00078000) /* mipmap related? */); + OUT_RING (txs); + OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); + OUT_RING ((pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); + OUT_RING (ps->bcol); + BEGIN_RING(curie, NV40TCL_TEX_SIZE1(unit), 1); + OUT_RING ((pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); +} + +void +nv40_fragtex_bind(struct nv40_context *nv40) +{ + struct nv40_fragment_program *fp = nv40->fragprog.active; + unsigned samplers, unit; + + samplers = nv40->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv40->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv40_fragtex_build(nv40, unit); + } + + nv40->fp_samplers = fp->samplers; +} + diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 7bb69e617ee..c9a7a2e364b 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -12,7 +12,7 @@ nv40_emit_hw_state(struct nv40_context *nv40) } if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { - nv40_state_tex_update(nv40); + nv40_fragtex_bind(nv40); BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); OUT_RING (2); diff --git a/src/mesa/pipe/nv40/nv40_state_tex.c b/src/mesa/pipe/nv40/nv40_state_tex.c deleted file mode 100644 index f2303e34db3..00000000000 --- a/src/mesa/pipe/nv40/nv40_state_tex.c +++ /dev/null @@ -1,146 +0,0 @@ -#include "nv40_context.h" - -#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ -{ \ - TRUE, \ - PIPE_FORMAT_##m, \ - NV40TCL_TEX_FORMAT_FORMAT_##tf, \ - (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x | NV40TCL_TEX_SWIZZLE_S0_Y_##ts0y | \ - NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w | \ - NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y | \ - NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ -} - -struct nv40_texture_format { - boolean defined; - uint pipe; - int format; - int swizzle; -}; - -static struct nv40_texture_format -nv40_texture_formats[] = { - _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), - _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), - _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), - _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), - _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), - _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), - _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), -// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), - {}, -}; - -static struct nv40_texture_format * -nv40_tex_format(uint pipe_format) -{ - struct nv40_texture_format *tf = nv40_texture_formats; - - while (tf->defined) { - if (tf->pipe == pipe_format) - return tf; - tf++; - } - - return NULL; -} - - -static void -nv40_tex_unit_enable(struct nv40_context *nv40, int unit) -{ - struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; - struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; - struct pipe_texture *pt = &nv40mt->base; - struct nv40_texture_format *tf; - uint32_t txf, txs, txp; - int swizzled = 0; /*XXX: implement in region code? */ - - tf = nv40_tex_format(pt->format); - if (!tf || !tf->defined) { - NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); - return; - } - - txf = ps->fmt; - txf |= tf->format | 0x8000; - txf |= ((pt->last_level - pt->first_level + 1) << - NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); - - if (1) /* XXX */ - txf |= NV40TCL_TEX_FORMAT_NO_BORDER; - - switch (pt->target) { - case PIPE_TEXTURE_CUBE: - txf |= NV40TCL_TEX_FORMAT_CUBIC; - /* fall-through */ - case PIPE_TEXTURE_2D: - txf |= NV40TCL_TEX_FORMAT_DIMS_2D; - break; - case PIPE_TEXTURE_3D: - txf |= NV40TCL_TEX_FORMAT_DIMS_3D; - break; - case PIPE_TEXTURE_1D: - txf |= NV40TCL_TEX_FORMAT_DIMS_1D; - break; - default: - NOUVEAU_ERR("Unknown target %d\n", pt->target); - return; - } - - if (swizzled) { - txp = 0; - } else { - txp = nv40mt->level[0].pitch; - txf |= NV40TCL_TEX_FORMAT_LINEAR; - } - - txs = tf->swizzle; - - nv40->tex[unit].buffer = nv40mt->buffer; - nv40->tex[unit].format = txf; - - BEGIN_RING(curie, NV40TCL_TEX_WRAP(unit), 6); - OUT_RING (ps->wrap); - OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | - (0x00078000) /* mipmap related? */); - OUT_RING (txs); - OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); - OUT_RING ((pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); - OUT_RING (ps->bcol); - BEGIN_RING(curie, NV40TCL_TEX_SIZE1(unit), 1); - OUT_RING ((pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); -} - -void -nv40_state_tex_update(struct nv40_context *nv40) -{ - struct nv40_fragment_program *fp = nv40->fragprog.active; - unsigned samplers, unit; - - samplers = nv40->fp_samplers & ~fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); - OUT_RING (0); - } - - samplers = nv40->dirty_samplers & fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - nv40_tex_unit_enable(nv40, unit); - } - - nv40->fp_samplers = fp->samplers; -} - -- cgit v1.2.3 From 4481c6c0b3e90450b20f1e7b4a0edc3ce346a4bb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 20 Jan 2008 23:20:42 +1100 Subject: nouveau: small oops --- src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index 784dfdd155e..c58af22c98c 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -158,7 +158,7 @@ nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) { struct nouveau_framebuffer *nvfb; - nvfb = (struct nouveau_framebuffer *)driDrawPriv; + nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate; st_unreference_framebuffer(&nvfb->stfb); free(nvfb); } -- cgit v1.2.3 From 3e39bc3d87a82d81ffaf383303ada32e530b879e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 28 Jan 2008 18:32:46 +1100 Subject: nouveau: quicky-port to gallium changes --- .../drivers/dri/nouveau_winsys/nouveau_context.c | 6 +- .../drivers/dri/nouveau_winsys/nouveau_context.h | 3 - .../drivers/dri/nouveau_winsys/nouveau_local.h | 6 +- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 18 +-- .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 152 +++++++-------------- .../dri/nouveau_winsys/nouveau_winsys_pipe.h | 11 ++ src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 10 -- src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c | 10 -- src/mesa/pipe/nouveau/nouveau_push.h | 3 +- src/mesa/pipe/nouveau/nouveau_winsys.h | 5 +- src/mesa/pipe/nv40/nv40_context.h | 14 +- src/mesa/pipe/nv40/nv40_draw.c | 18 +-- src/mesa/pipe/nv40/nv40_fragprog.c | 10 +- src/mesa/pipe/nv40/nv40_miptree.c | 8 +- src/mesa/pipe/nv40/nv40_state.c | 32 +---- src/mesa/pipe/nv40/nv40_state.h | 4 +- src/mesa/pipe/nv40/nv40_surface.c | 18 +-- src/mesa/pipe/nv40/nv40_vbo.c | 12 +- src/mesa/pipe/nv40/nv40_vertprog.c | 2 +- src/mesa/pipe/nv50/nv50_context.h | 2 +- src/mesa/pipe/nv50/nv50_draw.c | 18 +-- src/mesa/pipe/nv50/nv50_surface.c | 16 --- src/mesa/pipe/nv50/nv50_vbo.c | 2 +- 23 files changed, 123 insertions(+), 257 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c index 9c524fa0017..01fada5b89b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c @@ -73,6 +73,7 @@ nouveau_context_create(const __GLcontextModes *glVis, */ { struct pipe_surface *fb_surf; + struct nouveau_pipe_buffer *fb_buf; struct nouveau_bo_priv *fb_bo; fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); @@ -87,12 +88,15 @@ nouveau_context_create(const __GLcontextModes *glVis, fb_bo->base.size = fb_bo->drm.size; fb_bo->base.device = nv_screen->device; + fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer)); + fb_buf->bo = &fb_bo->base; + fb_surf = calloc(1, sizeof(struct pipe_surface)); fb_surf->cpp = nv_screen->front_cpp; fb_surf->pitch = nv_screen->front_pitch / fb_surf->cpp; fb_surf->height = nv_screen->front_height; fb_surf->refcount = 1; - fb_surf->buffer = (void *)fb_bo; + fb_surf->buffer = &fb_buf->base; nv->frontbuffer = fb_surf; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h index 2fd3336065c..7a74f7deecd 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h @@ -54,9 +54,6 @@ struct nouveau_context { void (*surface_copy_done)(struct nouveau_context *); int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, unsigned, unsigned, unsigned, unsigned, unsigned); - int (*surface_data)(struct nouveau_context *, struct pipe_surface *, - unsigned, unsigned, const void *, unsigned, - unsigned, unsigned, unsigned, unsigned); }; extern GLboolean nouveau_context_create(const __GLcontextModes *, diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index a53b6992028..7a539c81a94 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -60,9 +60,9 @@ OUT_RING (nv->o->handle); \ } while(0) -#define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nouveau_pushbuf_emit_reloc(nv->channel, nv->channel->pushbuf->cur, \ - (void*)(bo), (data), (flags), (vor), (tor));\ +#define OUT_RELOC(buf,data,flags,vor,tor) do { \ + nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur, \ + buf, (data), (flags), (vor), (tor)); \ OUT_RING(0); \ } while(0) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index f5e45461000..ddf3cba5f02 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -59,16 +59,13 @@ nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, return 0; } -static int -nouveau_pipe_surface_data(struct nouveau_winsys *nvws, struct pipe_surface *dst, - unsigned dx, unsigned dy, const void *src, - unsigned src_pitch, unsigned sx, unsigned sy, - unsigned w, unsigned h) +int +nouveau_pipe_emit_reloc(struct nouveau_channel *chan, void *ptr, + struct pipe_buffer *buf, uint32_t data, + uint32_t flags, uint32_t vor, uint32_t tor) { - if (nvws->nv->surface_data(nvws->nv, dst, dx, dy, src, src_pitch, sx, - sy, w, h)) - return 1; - return 0; + return nouveau_pushbuf_emit_reloc(chan, ptr, nouveau_buffer(buf)->bo, + data, flags, vor, tor); } struct pipe_context * @@ -102,7 +99,7 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->res_alloc = nouveau_resource_alloc; nvws->res_free = nouveau_resource_free; - nvws->push_reloc = nouveau_pushbuf_emit_reloc; + nvws->push_reloc = nouveau_pipe_emit_reloc; nvws->push_flush = nouveau_pushbuf_flush; nvws->grobj_alloc = nouveau_pipe_grobj_alloc; @@ -117,7 +114,6 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->surface_copy = nouveau_pipe_surface_copy; nvws->surface_fill = nouveau_pipe_surface_fill; - nvws->surface_data = nouveau_pipe_surface_data; return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index 1e160f04edd..f2087aaf9a2 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -1,6 +1,7 @@ #include "pipe/p_winsys.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" +#include "pipe/p_inlines.h" #include "nouveau_context.h" #include "nouveau_device.h" @@ -54,7 +55,6 @@ nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, enum pipe_format format, unsigned flags) { unsigned pitch = ((width * pf_get_size(format)) + 63) & ~63; - int ret; surf->format = format; surf->width = width; @@ -62,16 +62,11 @@ nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, surf->cpp = pf_get_size(format); surf->pitch = pitch / surf->cpp; - surf->buffer = ws->buffer_create(ws, 256, 0, 0); + surf->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + pitch * height); if (!surf->buffer) return 1; - ret = ws->buffer_data(ws, surf->buffer, pitch * height, NULL, 0); - if (ret) { - ws->buffer_reference(ws, &surf->buffer, NULL); - return ret; - } - return 0; } @@ -83,131 +78,87 @@ nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) *s = NULL; if (--surf->refcount <= 0) { if (surf->buffer) - ws->buffer_reference(ws, &surf->buffer, NULL); + pipe_buffer_reference(ws, &surf->buffer, NULL); free(surf); } } -static struct pipe_buffer_handle * +static struct pipe_buffer * nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, - unsigned flags, unsigned hint) + unsigned usage, unsigned size) { struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; struct nouveau_device *dev = nvpws->nv->nv_screen->device; - struct nouveau_bo *nvbo = NULL; + struct nouveau_pipe_buffer *nvbuf; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.alignment = alignment; + nvbuf->base.usage = usage; + nvbuf->base.size = size; - if (nouveau_bo_new(dev, NOUVEAU_BO_LOCAL, alignment, 0, &nvbo)) + if (nouveau_bo_new(dev, NOUVEAU_BO_LOCAL, alignment, size, &nvbuf->bo)) { + free(nvbuf); return NULL; - return (struct pipe_buffer_handle *)nvbo; + } + + return &nvbuf->base; } -static struct pipe_buffer_handle * +static struct pipe_buffer * nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) { struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; struct nouveau_device *dev = nvpws->nv->nv_screen->device; - struct nouveau_bo *nvbo = NULL; + struct nouveau_pipe_buffer *nvbuf; - if (nouveau_bo_user(dev, ptr, bytes, &nvbo)) + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) return NULL; - return (struct pipe_buffer_handle *)nvbo; -} - -static void * -nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, - unsigned flags) -{ - struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; - uint32_t map_flags = 0; + nvbuf->base.refcount = 1; + nvbuf->base.size = bytes; - if (flags & PIPE_BUFFER_FLAG_READ) - map_flags |= NOUVEAU_BO_RD; - if (flags & PIPE_BUFFER_FLAG_WRITE) - map_flags |= NOUVEAU_BO_WR; - - if (nouveau_bo_map(nvbo, map_flags)) + if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) { + free(nvbuf); return NULL; - return nvbo->map; -} - -static void -nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer_handle *bo) -{ - struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; - - nouveau_bo_unmap(nvbo); -} - -static void -nouveau_pipe_bo_reference(struct pipe_winsys *pws, - struct pipe_buffer_handle **ptr, - struct pipe_buffer_handle *bo) -{ - struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; - struct nouveau_context *nv = nvpws->nv; - struct nouveau_device *dev = nv->nv_screen->device; - - if (*ptr) { - struct nouveau_bo *nvbo = (struct nouveau_bo *)*ptr; - FIRE_RING(); - nouveau_bo_del(&nvbo); - *ptr = NULL; } - if (bo) { - struct nouveau_bo *nvbo = (struct nouveau_bo *)bo, *new = NULL; - nouveau_bo_ref(dev, nvbo->handle, &new); - *ptr = bo; - } + return &nvbuf->base; } -static int -nouveau_pipe_bo_data(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, - unsigned size, const void *data, unsigned usage) +static void +nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) { - struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; - - if (nvbo->size != size) - nouveau_bo_resize(nvbo, size); + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - if (data) { - if (nouveau_bo_map(nvbo, NOUVEAU_BO_WR)) - return 1; - memcpy(nvbo->map, data, size); - nouveau_bo_unmap(nvbo); - } - - return 0; + nouveau_bo_del(&nvbuf->bo); } -static int -nouveau_pipe_bo_subdata(struct pipe_winsys *pws, struct pipe_buffer_handle *bo, - unsigned long offset, unsigned long size, - const void *data) +static void * +nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf, + unsigned flags) { - struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + uint32_t map_flags = 0; - if (nouveau_bo_map(nvbo, NOUVEAU_BO_WR)) - return 1; - memcpy(nvbo->map + offset, data, size); - nouveau_bo_unmap(nvbo); + if (flags & PIPE_BUFFER_USAGE_CPU_READ) + map_flags |= NOUVEAU_BO_RD; + if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) + map_flags |= NOUVEAU_BO_WR; - return 0; + if (nouveau_bo_map(nvbuf->bo, map_flags)) + return NULL; + return nvbuf->bo->map; } -static int -nouveau_pipe_bo_get_subdata(struct pipe_winsys *pws, - struct pipe_buffer_handle *bo, unsigned long offset, - unsigned long size, void *data) +static void +nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) { - struct nouveau_bo *nvbo = (struct nouveau_bo *)bo; + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - if (nouveau_bo_map(nvbo, NOUVEAU_BO_RD)) - return 1; - memcpy(data, nvbo->map + offset, size); - nouveau_bo_unmap(nvbo); - - return 0; + nouveau_bo_unmap(nvbuf->bo); } struct pipe_winsys * @@ -230,13 +181,10 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws->surface_release = nouveau_surface_release; pws->buffer_create = nouveau_pipe_bo_create; + pws->buffer_destroy = nouveau_pipe_bo_del; pws->user_buffer_create = nouveau_pipe_bo_user_create; pws->buffer_map = nouveau_pipe_bo_map; pws->buffer_unmap = nouveau_pipe_bo_unmap; - pws->buffer_reference = nouveau_pipe_bo_reference; - pws->buffer_data = nouveau_pipe_bo_data; - pws->buffer_subdata = nouveau_pipe_bo_subdata; - pws->buffer_get_subdata= nouveau_pipe_bo_get_subdata; pws->get_name = nouveau_get_name; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h index bcac60851cb..6a03ac0d773 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h @@ -5,6 +5,17 @@ #include "pipe/p_winsys.h" #include "nouveau_context.h" +struct nouveau_pipe_buffer { + struct pipe_buffer base; + struct nouveau_bo *bo; +}; + +static inline struct nouveau_pipe_buffer * +nouveau_buffer(struct pipe_buffer *buf) +{ + return (struct nouveau_pipe_buffer *)buf; +} + struct nouveau_pipe_winsys { struct pipe_winsys pws; diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c index 2288f9c71a8..fe1ea4ed70f 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c @@ -159,15 +159,6 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, return 0; } -static int -nv04_surface_data(struct nouveau_context *nv, struct pipe_surface *dst, - unsigned dx, unsigned dy, const void *src, unsigned src_pitch, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - NOUVEAU_ERR("unimplemented!!\n"); - return 0; -} - int nouveau_surface_init_nv04(struct nouveau_context *nv) { @@ -230,7 +221,6 @@ nouveau_surface_init_nv04(struct nouveau_context *nv) nv->surface_copy = nv04_surface_copy_blit; nv->surface_copy_done = nv04_surface_copy_done; nv->surface_fill = nv04_surface_fill; - nv->surface_data = nv04_surface_data; return 0; } diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c index d00378d9045..15a10028613 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c @@ -133,15 +133,6 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, return 0; } -static int -nv50_surface_data(struct nouveau_context *nv, struct pipe_surface *dst, - unsigned dx, unsigned dy, const void *src, unsigned src_pitch, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - NOUVEAU_ERR("unimplemented!!\n"); - return 0; -} - int nouveau_surface_init_nv50(struct nouveau_context *nv) { @@ -164,7 +155,6 @@ nouveau_surface_init_nv50(struct nouveau_context *nv) nv->surface_copy = nv50_surface_copy; nv->surface_copy_done = nv50_surface_copy_done; nv->surface_fill = nv50_surface_fill; - nv->surface_data = nv50_surface_data; return 0; } diff --git a/src/mesa/pipe/nouveau/nouveau_push.h b/src/mesa/pipe/nouveau/nouveau_push.h index 0fc972d2a22..117e3535cf7 100644 --- a/src/mesa/pipe/nouveau/nouveau_push.h +++ b/src/mesa/pipe/nouveau/nouveau_push.h @@ -45,8 +45,7 @@ NOUVEAU_PUSH_CONTEXT(pc); \ pc->nvws->push_reloc(pc->nvws->channel, \ pc->nvws->channel->pushbuf->cur, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ + (bo), (data), (flags), (vor), (tor)); \ OUT_RING(0); \ } while(0) diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 27c991077e7..0b394cfce6b 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -25,7 +25,7 @@ struct nouveau_winsys { void (*res_free)(struct nouveau_resource **); int (*push_reloc)(struct nouveau_channel *, void *ptr, - struct nouveau_bo *, uint32_t data, + struct pipe_buffer *, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor); int (*push_flush)(struct nouveau_channel *, unsigned size); @@ -47,9 +47,6 @@ struct nouveau_winsys { unsigned, unsigned, unsigned, unsigned); int (*surface_fill)(struct nouveau_winsys *, struct pipe_surface *, unsigned, unsigned, unsigned, unsigned, unsigned); - int (*surface_data)(struct nouveau_winsys *, struct pipe_surface *, - unsigned, unsigned, const void *, unsigned, - unsigned, unsigned, unsigned, unsigned); }; extern struct pipe_context * diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index df8489f3a6b..2d069619d7c 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -48,17 +48,17 @@ struct nv40_context { unsigned vp_samplers; uint32_t rt_enable; - struct pipe_buffer_handle *rt[4]; - struct pipe_buffer_handle *zeta; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; struct { - struct pipe_buffer_handle *buffer; + struct pipe_buffer *buffer; uint32_t format; } tex[16]; unsigned vb_enable; struct { - struct pipe_buffer_handle *buffer; + struct pipe_buffer *buffer; unsigned delta; } vb[16]; @@ -69,14 +69,14 @@ struct nv40_context { struct nv40_vertex_program *active; struct nv40_vertex_program *current; - struct pipe_buffer_handle *constant_buf; + struct pipe_buffer *constant_buf; } vertprog; struct { struct nv40_fragment_program *active; struct nv40_fragment_program *current; - struct pipe_buffer_handle *constant_buf; + struct pipe_buffer *constant_buf; } fragprog; struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; @@ -124,7 +124,7 @@ extern void nv40_state_tex_update(struct nv40_context *nv40); extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, unsigned start, unsigned count); extern boolean nv40_draw_elements(struct pipe_context *pipe, - struct pipe_buffer_handle *indexBuffer, + struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count); diff --git a/src/mesa/pipe/nv40/nv40_draw.c b/src/mesa/pipe/nv40/nv40_draw.c index 6ad2b66cce8..d361d5f07d9 100644 --- a/src/mesa/pipe/nv40/nv40_draw.c +++ b/src/mesa/pipe/nv40/nv40_draw.c @@ -9,33 +9,26 @@ struct nv40_draw_stage { }; static void -nv40_draw_begin(struct draw_stage *draw) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv40_draw_end(struct draw_stage *draw) +nv40_draw_point(struct draw_stage *draw, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv40_draw_point(struct draw_stage *draw, struct prim_header *prim) +nv40_draw_line(struct draw_stage *draw, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv40_draw_line(struct draw_stage *draw, struct prim_header *prim) +nv40_draw_tri(struct draw_stage *draw, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv40_draw_tri(struct draw_stage *draw, struct prim_header *prim) +nv40_draw_flush(struct draw_stage *draw, unsigned flags) { - NOUVEAU_ERR("\n"); } static void @@ -57,11 +50,10 @@ nv40_draw_render_stage(struct nv40_context *nv40) nv40draw->nv40 = nv40; nv40draw->draw.draw = nv40->draw; - nv40draw->draw.begin = nv40_draw_begin; nv40draw->draw.point = nv40_draw_point; nv40draw->draw.line = nv40_draw_line; nv40draw->draw.tri = nv40_draw_tri; - nv40draw->draw.end = nv40_draw_end; + nv40draw->draw.flush = nv40_draw_flush; nv40draw->draw.reset_stipple_counter = nv40_draw_reset_stipple_counter; nv40draw->draw.destroy = nv40_draw_destroy; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index fded11d1703..714634396d2 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -769,7 +769,7 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) if (fp->nr_consts) { float *map = ws->buffer_map(ws, nv40->fragprog.constant_buf, - PIPE_BUFFER_FLAG_READ); + PIPE_BUFFER_USAGE_CPU_READ); for (i = 0; i < fp->nr_consts; i++) { struct nv40_fragment_program_data *fpd = &fp->consts[i]; uint32_t *p = &fp->insn[fpd->offset]; @@ -788,10 +788,10 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) uint32_t *map; if (!fp->buffer) - fp->buffer = ws->buffer_create(ws, 0x100, 0, 0); - ws->buffer_data(ws, fp->buffer, fp->insn_len * 4, NULL, 0); - - map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_FLAG_WRITE); + fp->buffer = ws->buffer_create(ws, 0x100, 0, + fp->insn_len * 4); + map = ws->buffer_map(ws, fp->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); #if 0 for (i = 0; i < fp->insn_len; i++) { diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 625c2fc32d7..48062d1ae8a 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -1,6 +1,7 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" +#include "pipe/p_inlines.h" #include "nv40_context.h" @@ -65,14 +66,13 @@ nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) nv40_miptree_layout(nv40mt); - nv40mt->buffer = ws->buffer_create(ws, 256, 0, 0); + nv40mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + nv40mt->total_size); if (!nv40mt->buffer) { free(nv40mt); return; } - ws->buffer_data(ws, nv40mt->buffer, nv40mt->total_size, NULL, - PIPE_BUFFER_USAGE_PIXEL); *pt = &nv40mt->base; } @@ -87,7 +87,7 @@ nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) struct nv40_miptree *nv40mt = (struct nv40_miptree *)mt; int l; - ws->buffer_reference(ws, &nv40mt->buffer, NULL); + pipe_buffer_reference(ws, &nv40mt->buffer, NULL); for (l = mt->first_level; l <= mt->last_level; l++) { if (nv40mt->level[l].image_offset) free(nv40mt->level[l].image_offset); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 03a9e735c07..c619948b55a 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -563,35 +563,17 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, NV40TCL_RT_ENABLE_COLOR3)) rt_enable |= NV40TCL_RT_ENABLE_MRT; - if (fb->zbuf) { + if (fb->zsbuf) { if (colour_format) { - assert(w == fb->zbuf->width); - assert(h == fb->zbuf->height); + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); } else { - w = fb->zbuf->width; - h = fb->zbuf->height; + w = fb->zsbuf->width; + h = fb->zsbuf->height; } - zeta_format = fb->zbuf->format; - zeta = fb->zbuf; - } - - if (fb->sbuf) { - if (colour_format) { - assert(w == fb->sbuf->width); - assert(h == fb->sbuf->height); - } else { - w = fb->zbuf->width; - h = fb->zbuf->height; - } - - if (zeta_format) { - assert(fb->sbuf->format == zeta_format); - assert(fb->sbuf == zeta); - } else { - zeta_format = fb->sbuf->format; - zeta = fb->sbuf; - } + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; } rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 4c4c847c9a1..24eea36db3a 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -96,7 +96,7 @@ struct nv40_fragment_program { struct nv40_fragment_program_data *consts; unsigned nr_consts; - struct pipe_buffer_handle *buffer; + struct pipe_buffer *buffer; uint32_t fp_control; }; @@ -134,7 +134,7 @@ struct nv40_depth_stencil_alpha_state { struct nv40_miptree { struct pipe_texture base; - struct pipe_buffer_handle *buffer; + struct pipe_buffer *buffer; uint total_size; struct { diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c index 159837388f3..d8f87d9adc2 100644 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ b/src/mesa/pipe/nv40/nv40_surface.c @@ -84,7 +84,7 @@ nv40_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, ps = ws->surface_alloc(ws); if (!ps) return NULL; - ws->buffer_reference(ws, &ps->buffer, nv40mt->buffer); + pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); ps->format = pt->format; ps->cpp = pt->cpp; ps->width = pt->width[level]; @@ -103,19 +103,6 @@ nv40_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, return ps; } -static void -nv40_surface_data(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, const void *src, - unsigned src_stride, unsigned srcx, unsigned srcy, - unsigned width, unsigned height) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; - - nvws->surface_data(nvws, dest, destx, desty, src, src_stride, - srcx, srcy, width, height); -} - static void nv40_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, unsigned destx, unsigned desty, struct pipe_surface *src, @@ -144,9 +131,6 @@ nv40_init_surface_functions(struct nv40_context *nv40) { nv40->pipe.is_format_supported = nv40_surface_format_supported; nv40->pipe.get_tex_surface = nv40_get_tex_surface; - nv40->pipe.get_tile = pipe_get_tile_raw; - nv40->pipe.put_tile = pipe_put_tile_raw; - nv40->pipe.surface_data = nv40_surface_data; nv40->pipe.surface_copy = nv40_surface_copy; nv40->pipe.surface_fill = nv40_surface_fill; } diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 3de87ecab5e..3fa8ddcb81b 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -46,7 +46,7 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, type = nv40_vbo_type(ve->src_format); ncomp = nv40_vbo_ncomp(ve->src_format); - map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_FLAG_READ); + map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); map += vb->buffer_offset + ve->src_offset; switch (type) { @@ -149,7 +149,7 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) static boolean nv40_vbo_validate_state(struct nv40_context *nv40, - struct pipe_buffer_handle *ib, unsigned ib_format) + struct pipe_buffer *ib, unsigned ib_format) { unsigned inputs; @@ -297,7 +297,7 @@ nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, static boolean nv40_draw_elements_inline(struct pipe_context *pipe, - struct pipe_buffer_handle *ib, unsigned ib_size, + struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); @@ -306,7 +306,7 @@ nv40_draw_elements_inline(struct pipe_context *pipe, assert(nv40_vbo_validate_state(nv40, NULL, 0)); - map = ws->buffer_map(ws, ib, PIPE_BUFFER_FLAG_READ); + map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); if (!ib) assert(0); @@ -338,7 +338,7 @@ nv40_draw_elements_inline(struct pipe_context *pipe, static boolean nv40_draw_elements_vbo(struct pipe_context *pipe, - struct pipe_buffer_handle *ib, unsigned ib_size, + struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); @@ -388,7 +388,7 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, boolean nv40_draw_elements(struct pipe_context *pipe, - struct pipe_buffer_handle *indexBuffer, unsigned indexSize, + struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { if (indexSize != 1) { diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index bd052fe8341..e15ddbbcde5 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -727,7 +727,7 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) if (nv40->vertprog.constant_buf) { map = ws->buffer_map(ws, nv40->vertprog.constant_buf, - PIPE_BUFFER_FLAG_READ); + PIPE_BUFFER_USAGE_CPU_READ); } for (i = 0; i < vp->nr_consts; i++) { diff --git a/src/mesa/pipe/nv50/nv50_context.h b/src/mesa/pipe/nv50/nv50_context.h index c43202c3d57..5491c0cbb5d 100644 --- a/src/mesa/pipe/nv50/nv50_context.h +++ b/src/mesa/pipe/nv50/nv50_context.h @@ -45,7 +45,7 @@ extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50); extern boolean nv50_draw_arrays(struct pipe_context *, unsigned mode, unsigned start, unsigned count); extern boolean nv50_draw_elements(struct pipe_context *pipe, - struct pipe_buffer_handle *indexBuffer, + struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count); diff --git a/src/mesa/pipe/nv50/nv50_draw.c b/src/mesa/pipe/nv50/nv50_draw.c index 97088920284..85d347f5e00 100644 --- a/src/mesa/pipe/nv50/nv50_draw.c +++ b/src/mesa/pipe/nv50/nv50_draw.c @@ -9,33 +9,26 @@ struct nv50_draw_stage { }; static void -nv50_draw_begin(struct draw_stage *draw) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv50_draw_end(struct draw_stage *draw) +nv50_draw_point(struct draw_stage *draw, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv50_draw_point(struct draw_stage *draw, struct prim_header *prim) +nv50_draw_line(struct draw_stage *draw, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv50_draw_line(struct draw_stage *draw, struct prim_header *prim) +nv50_draw_tri(struct draw_stage *draw, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv50_draw_tri(struct draw_stage *draw, struct prim_header *prim) +nv50_draw_flush(struct draw_stage *draw, unsigned flags) { - NOUVEAU_ERR("\n"); } static void @@ -51,11 +44,10 @@ nv50_draw_render_stage(struct nv50_context *nv50) nv50draw->nv50 = nv50; nv50draw->draw.draw = nv50->draw; - nv50draw->draw.begin = nv50_draw_begin; nv50draw->draw.point = nv50_draw_point; nv50draw->draw.line = nv50_draw_line; nv50draw->draw.tri = nv50_draw_tri; - nv50draw->draw.end = nv50_draw_end; + nv50draw->draw.flush = nv50_draw_flush; nv50draw->draw.reset_stipple_counter = nv50_draw_reset_stipple_counter; return &nv50draw->draw; diff --git a/src/mesa/pipe/nv50/nv50_surface.c b/src/mesa/pipe/nv50/nv50_surface.c index d2079946889..cfb370da77e 100644 --- a/src/mesa/pipe/nv50/nv50_surface.c +++ b/src/mesa/pipe/nv50/nv50_surface.c @@ -42,19 +42,6 @@ nv50_get_tex_surface(struct pipe_context *pipe, return NULL; } -static void -nv50_surface_data(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, const void *src, - unsigned src_stride, unsigned srcx, unsigned srcy, - unsigned width, unsigned height) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - nvws->surface_data(nvws, dest, destx, desty, src, src_stride, - srcx, srcy, width, height); -} - static void nv50_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, unsigned destx, unsigned desty, struct pipe_surface *src, @@ -82,9 +69,6 @@ void nv50_init_surface_functions(struct nv50_context *nv50) { nv50->pipe.get_tex_surface = nv50_get_tex_surface; - nv50->pipe.get_tile = pipe_get_tile_raw; - nv50->pipe.put_tile = pipe_put_tile_raw; - nv50->pipe.surface_data = nv50_surface_data; nv50->pipe.surface_copy = nv50_surface_copy; nv50->pipe.surface_fill = nv50_surface_fill; } diff --git a/src/mesa/pipe/nv50/nv50_vbo.c b/src/mesa/pipe/nv50/nv50_vbo.c index 0c9d2806d78..6c0dc23a439 100644 --- a/src/mesa/pipe/nv50/nv50_vbo.c +++ b/src/mesa/pipe/nv50/nv50_vbo.c @@ -15,7 +15,7 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, boolean nv50_draw_elements(struct pipe_context *pipe, - struct pipe_buffer_handle *indexBuffer, unsigned indexSize, + struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { NOUVEAU_ERR("unimplemented\n"); -- cgit v1.2.3 From d342a6eeccb829dfbdb95f259bc9a272a16a8b4b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 28 Jan 2008 18:40:11 +1100 Subject: nouveau: ddx versioning changed to match nouveau.ko versioning --- src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c index c58af22c98c..f06e1784831 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c @@ -264,7 +264,8 @@ __driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, __GLcontextModes ** driver_modes) { __DRIscreenPrivate *psp; - static const __DRIversion ddx_expected = { 1, 2, 0 }; + static const __DRIversion ddx_expected = + { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; -- cgit v1.2.3 From ae7dcae81bc54bc16f16127f2fa1d01d9aa2332c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 30 Jan 2008 00:05:50 +1100 Subject: nouveau: support nv67 --- .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 1 + src/mesa/pipe/nv40/nv40_context.c | 34 +++++++++++++--------- 2 files changed, 22 insertions(+), 13 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index ddf3cba5f02..8ffe89feecb 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -81,6 +81,7 @@ nouveau_pipe_create(struct nouveau_context *nv) switch (nv->chipset & 0xf0) { case 0x40: + case 0x60: hw_create = nv40_create; break; case 0x50: diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index f6fdd6b4376..d0d9583e32e 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -188,26 +188,34 @@ nv40_init_hwctx(struct nv40_context *nv40, int curie_class) return TRUE; } -#define GRCLASS4097_CHIPSETS 0x00000baf -#define GRCLASS4497_CHIPSETS 0x00005450 +#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf +#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 +#define NV6X_GRCLASS4497_CHIPSETS 0x00000080 + struct pipe_context * nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, unsigned chipset) { struct nv40_context *nv40; - int curie_class, ret; - - if ((chipset & 0xf0) != 0x40) { - NOUVEAU_ERR("Not a NV4X chipset\n"); - return NULL; + int curie_class = 0, ret; + + switch (chipset & 0xf0) { + case 0x40: + if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV40TCL; + else + if (NV4X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV44TCL; + break; + case 0x60: + if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV44TCL; + break; + default: + break; } - if (GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) { - curie_class = NV40TCL; - } else - if (GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) { - curie_class = NV44TCL; - } else { + if (!curie_class) { NOUVEAU_ERR("Unknown NV4x chipset: NV%02x\n", chipset); return NULL; } -- cgit v1.2.3 From 1bcd6b8ff9c25a0d8c2d15c365000a50e643b303 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 30 Jan 2008 15:34:02 +1100 Subject: nouveau: 0xdeadc0de --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 23 ---------------------- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 --- 2 files changed, 26 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 288674f231b..4c235845b7a 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -246,29 +246,6 @@ nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, return 0; } -int -nouveau_bo_resize(struct nouveau_bo *bo, int size) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - int ret; - - if (!nvbo || nvbo->user) - return -EINVAL; - - if (nvbo->sysmem) { - nvbo->sysmem = realloc(nvbo->sysmem, size); - if (!nvbo->sysmem) - return -ENOMEM; - } else { - ret = nouveau_bo_realloc_gpu(nvbo, 0, size); - if (ret) - return ret; - } - - nvbo->base.size = size; - return 0; -} - void nouveau_bo_del(struct nouveau_bo **bo) { diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 3e886869d85..7ea4c654654 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -274,9 +274,6 @@ nouveau_bo_user(struct nouveau_device *, void *ptr, int size, extern int nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); -extern int -nouveau_bo_resize(struct nouveau_bo *, int size); - extern void nouveau_bo_del(struct nouveau_bo **); -- cgit v1.2.3 From edbeec2b9177fef27b5a04b7b9b975b8abf84367 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 2 Feb 2008 14:21:06 +1100 Subject: nouveau: implement a bo_set_status() --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 133 ++++++++++++--------- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 + .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 4 +- 3 files changed, 80 insertions(+), 60 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 4c235845b7a..6887ffa6886 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -79,37 +79,6 @@ nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, return 0; } -static int -nouveau_bo_realloc_gpu(struct nouveau_bo_priv *nvbo, uint32_t flags, int size) -{ - int ret; - - if (nvbo->drm.size && nvbo->drm.size != size) { - nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); - } - - if (size && !nvbo->drm.size) { - if (flags) { - nvbo->drm.flags = 0; - if (flags & NOUVEAU_BO_VRAM) - nvbo->drm.flags |= NOUVEAU_MEM_FB; - if (flags & NOUVEAU_BO_GART) - nvbo->drm.flags |= (NOUVEAU_MEM_AGP | - NOUVEAU_MEM_PCI); - nvbo->drm.flags |= NOUVEAU_MEM_MAPPED; - } - - ret = nouveau_mem_alloc(nvbo->base.device, size, - nvbo->drm.alignment, nvbo->drm.flags, - &nvbo->drm, &nvbo->map); - if (ret) { - assert(0); - } - } - - return 0; -} - static void nouveau_bo_tmp_del(void *priv) { @@ -183,26 +152,17 @@ nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, if (!nvbo) return -ENOMEM; nvbo->base.device = dev; + nvbo->base.size = size; + nvbo->base.handle = bo_to_ptr(nvbo); nvbo->drm.alignment = align; + nvbo->refcount = 1; - if (flags & NOUVEAU_BO_PIN) { - ret = nouveau_bo_realloc_gpu(nvbo, flags, size); - if (ret) { - free(nvbo); - return ret; - } - } else { - nvbo->sysmem = malloc(size); - if (!nvbo->sysmem) { - free(nvbo); - return -ENOMEM; - } + ret = nouveau_bo_set_status(&nvbo->base, flags); + if (ret) { + free(nvbo); + return ret; } - nvbo->base.size = size; - nvbo->base.offset = nvbo->drm.offset; - nvbo->base.handle = bo_to_ptr(nvbo); - nvbo->refcount = 1; *bo = &nvbo->base; return 0; } @@ -261,8 +221,7 @@ nouveau_bo_del(struct nouveau_bo **bo) if (nvbo->fence) nouveau_fence_wait(&nvbo->fence); - - nouveau_bo_realloc_gpu(nvbo, 0, 0); + nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); if (nvbo->sysmem && !nvbo->user) free(nvbo->sysmem); free(nvbo); @@ -303,6 +262,66 @@ nouveau_bo_upload(struct nouveau_bo_priv *nvbo) return 0; } +int +nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct drm_nouveau_mem_alloc new; + void *new_map = NULL, *new_sysmem = NULL; + unsigned new_flags = 0, ret; + + assert(!bo->map); + + /* Check current memtype vs requested, if they match do nothing */ + if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) + return 0; + if ((nvbo->drm.flags & NOUVEAU_MEM_AGP) && (flags & NOUVEAU_BO_GART)) + return 0; + if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) + return 0; + + memset(&new, 0x00, sizeof(new)); + + /* Allocate new memory */ + if (flags & NOUVEAU_BO_VRAM) + new_flags |= NOUVEAU_MEM_FB; + else + if (flags & NOUVEAU_BO_GART) + new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); + + if (new_flags) { + ret = nouveau_mem_alloc(bo->device, bo->size, + nvbo->drm.alignment, new_flags, + &new, &new_map); + if (ret) + return ret; + } else { + new_sysmem = malloc(bo->size); + } + + /* Copy old -> new */ + /*XXX: use M2MF */ + if (nvbo->sysmem || nvbo->map) { + nouveau_bo_map(bo, NOUVEAU_BO_RD); + memcpy(new_map, bo->map, bo->size); + nouveau_bo_unmap(bo); + } + + /* Free old memory */ + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem) + free(nvbo->sysmem); + + nvbo->drm = new; + nvbo->map = new_map; + nvbo->sysmem = new_sysmem; + bo->flags = flags; + bo->offset = nvbo->drm.offset; + return 0; +} + static int nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, struct nouveau_fence *fence, uint32_t flags) @@ -335,18 +354,14 @@ nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, struct nouveau_fence *fence, uint32_t flags) { struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + int ret; - if (!nvbo->drm.size) { - nouveau_bo_realloc_gpu(nvbo, flags, nvbo->base.size); - nouveau_bo_upload(nvbo); - if (!nvbo->user) { - free(nvbo->sysmem); - nvbo->sysmem = NULL; - } - } else - if (nvbo->user) { + ret = nouveau_bo_set_status(bo, flags); + if (ret) + return ret; + + if (nvbo->user) nouveau_bo_upload(nvbo); - } nvbo->offset = nvbo->drm.offset; if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 7ea4c654654..67e19f1cfe8 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -274,6 +274,9 @@ nouveau_bo_user(struct nouveau_device *, void *ptr, int size, extern int nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); +extern int +nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags); + extern void nouveau_bo_del(struct nouveau_bo **); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index f2087aaf9a2..7d7fefa8016 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -90,6 +90,7 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; struct nouveau_device *dev = nvpws->nv->nv_screen->device; struct nouveau_pipe_buffer *nvbuf; + uint32_t flags = 0; nvbuf = calloc(1, sizeof(*nvbuf)); if (!nvbuf) @@ -99,7 +100,8 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, nvbuf->base.usage = usage; nvbuf->base.size = size; - if (nouveau_bo_new(dev, NOUVEAU_BO_LOCAL, alignment, size, &nvbuf->bo)) { + flags = NOUVEAU_BO_LOCAL; + if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { free(nvbuf); return NULL; } -- cgit v1.2.3 From f87a8cace2ce946d969eb6d93def2d65cb541bda Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 3 Feb 2008 12:08:31 +1100 Subject: nouveau: avoid relocations where possible. Potential relocations are emitted as NOPs where they're needed. In the event a buffer moves, the pushbuf code will emit the relevant state changes into the NOPs. Just a start, more work is needed to get this looking how I want it to. --- .../drivers/dri/nouveau_winsys/nouveau_local.h | 3 +- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 61 +++++---- src/mesa/pipe/nouveau/nouveau_bo.h | 1 + src/mesa/pipe/nouveau/nouveau_push.h | 13 +- src/mesa/pipe/nv40/nv40_fragprog.c | 5 + src/mesa/pipe/nv40/nv40_fragtex.c | 8 +- src/mesa/pipe/nv40/nv40_state.c | 32 ++++- src/mesa/pipe/nv40/nv40_state_emit.c | 144 ++++++++++++--------- 8 files changed, 170 insertions(+), 97 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index 7a539c81a94..59febca2929 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -61,9 +61,8 @@ } while(0) #define OUT_RELOC(buf,data,flags,vor,tor) do { \ - nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur, \ + nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur++, \ buf, (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ } while(0) /* Raw data + flags depending on FB/TT buffer */ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index a34a5c18662..7d5eddb92ff 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -96,6 +96,31 @@ nouveau_pushbuf_init(struct nouveau_channel *chan) return 0; } +static uint32_t +nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo, + struct nouveau_pushbuf_reloc *r) +{ + uint32_t push; + + if (r->flags & NOUVEAU_BO_LOW) { + push = bo->offset + r->data; + } else + if (r->flags & NOUVEAU_BO_HIGH) { + push = (bo->offset + r->data) >> 32; + } else { + push = r->data; + } + + if (r->flags & NOUVEAU_BO_OR) { + if (bo->flags & NOUVEAU_BO_VRAM) + push |= r->vor; + else + push |= r->tor; + } + + return push; +} + /* This would be our TTM "superioctl" */ int nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) @@ -133,34 +158,20 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) if (bo->offset == nouveau_bo(bo)->offset && bo->flags == nouveau_bo(bo)->flags) { - /*XXX: could avoid reloc in this case, except with the - * current design we'd confuse the GPU quite a bit - * if we did this. Will fix soon. - */ + while ((r = ptr_to_pbrel(pbbo->relocs))) { + pbbo->relocs = r->next; + free(r); + } + + nvpb->buffers = pbbo->next; + free(pbbo); + continue; } bo->offset = nouveau_bo(bo)->offset; bo->flags = nouveau_bo(bo)->flags; while ((r = ptr_to_pbrel(pbbo->relocs))) { - uint32_t push; - - if (r->flags & NOUVEAU_BO_LOW) { - push = bo->offset + r->data; - } else - if (r->flags & NOUVEAU_BO_HIGH) { - push = (bo->offset + r->data) >> 32; - } else { - push = r->data; - } - - if (r->flags & NOUVEAU_BO_OR) { - if (bo->flags & NOUVEAU_BO_VRAM) - push |= r->vor; - else - push |= r->tor; - } - - *r->ptr = push; + *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); pbbo->relocs = r->next; free(r); } @@ -241,6 +252,10 @@ nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, r->vor = vor; r->tor = tor; + if (flags & NOUVEAU_BO_DUMMY) + *(uint32_t *)ptr = 0; + else + *(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r); return 0; } diff --git a/src/mesa/pipe/nouveau/nouveau_bo.h b/src/mesa/pipe/nouveau/nouveau_bo.h index 2b57ee92633..18020e9c652 100644 --- a/src/mesa/pipe/nouveau/nouveau_bo.h +++ b/src/mesa/pipe/nouveau/nouveau_bo.h @@ -35,6 +35,7 @@ #define NOUVEAU_BO_HIGH (1 << 7) #define NOUVEAU_BO_OR (1 << 8) #define NOUVEAU_BO_LOCAL (1 << 9) +#define NOUVEAU_BO_DUMMY (1 << 31) struct nouveau_bo { struct nouveau_device *device; diff --git a/src/mesa/pipe/nouveau/nouveau_push.h b/src/mesa/pipe/nouveau/nouveau_push.h index 117e3535cf7..679472669b9 100644 --- a/src/mesa/pipe/nouveau/nouveau_push.h +++ b/src/mesa/pipe/nouveau/nouveau_push.h @@ -44,9 +44,8 @@ #define OUT_RELOC(bo,data,flags,vor,tor) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ pc->nvws->push_reloc(pc->nvws->channel, \ - pc->nvws->channel->pushbuf->cur, \ + pc->nvws->channel->pushbuf->cur++, \ (bo), (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ } while(0) /* Raw data + flags depending on FB/TT buffer */ @@ -71,4 +70,14 @@ OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ } while(0) +/* A reloc which'll recombine into a NV_DMA_METHOD packet header */ +#define OUT_RELOCm(bo, flags, obj, mthd, size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + OUT_RELOCd((bo), (pc->obj->subc << 13) | ((size) << 18) | (mthd), \ + (flags), 0, 0); \ + pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + #endif diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 714634396d2..ef320393d94 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -815,6 +815,11 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->on_hw = TRUE; } + BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); + OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, + NV40TCL_FP_ADDRESS_DMA1); BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); OUT_RING (fp->fp_control); diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 48d6eb629f5..7c5ecd5c566 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -104,7 +104,13 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) nv40->tex[unit].buffer = nv40mt->buffer; nv40->tex[unit].format = txf; - BEGIN_RING(curie, NV40TCL_TEX_WRAP(unit), 6); + BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); + OUT_RELOCl(nv40->tex[unit].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv40->tex[unit].buffer, nv40->tex[unit].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, + NV40TCL_TEX_FORMAT_DMA1); OUT_RING (ps->wrap); OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | (0x00078000) /* mipmap related? */); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index c619948b55a..bb435b106b4 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -603,33 +603,51 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 1); - OUT_RING (rt[0]->pitch * rt[0]->cpp); nv40->rt[0] = rt[0]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); + OUT_RING (rt[0]->pitch * rt[0]->cpp); + OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(curie, NV40TCL_COLOR1_PITCH, 2); - OUT_RING (rt[1]->pitch * rt[1]->cpp); nv40->rt[1] = rt[1]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); + OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { + nv40->rt[2] = rt[2]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); + OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); + OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); OUT_RING (rt[2]->pitch * rt[2]->cpp); - nv40->rt[2] = rt[2]->buffer; } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { + nv40->rt[3] = rt[3]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); + OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); + OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); OUT_RING (rt[3]->pitch * rt[3]->cpp); - nv40->rt[3] = rt[3]->buffer; } if (zeta_format) { + nv40->zeta = zeta->buffer; + BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); + OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); OUT_RING (zeta->pitch * zeta->cpp); - nv40->zeta = zeta->buffer; } nv40->rt_enable = rt_enable; diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index c9a7a2e364b..66b98d5fab3 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -1,94 +1,114 @@ #include "nv40_context.h" #include "nv40_state.h" -void -nv40_emit_hw_state(struct nv40_context *nv40) +/* Emit relocs for every referenced buffer. + * + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. These relocs appear in the push buffer as + * NOPs, and will only be turned into state changes if a buffer + * actually moves. + */ +static void +nv40_state_emit_dummy_relocs(struct nv40_context *nv40) { - int i; - - if (nv40->dirty & NV40_NEW_FRAGPROG) { - nv40_fragprog_bind(nv40, nv40->fragprog.current); - /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ - } - - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { - nv40_fragtex_bind(nv40); - - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (1); - nv40->dirty &= ~NV40_NEW_FRAGPROG; - } - - if (nv40->dirty & NV40_NEW_VERTPROG) { - nv40_vertprog_bind(nv40, nv40->vertprog.current); - nv40->dirty &= ~NV40_NEW_VERTPROG; - } - - nv40->dirty_samplers = 0; - - /* Emit relocs for every referenced buffer. - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. This isn't very efficient, but we don't - * seem to take a significant performance hit. Will be improved - * at some point. Vertex arrays are emitted by nv40_vbo.c - */ + unsigned rt_flags, tx_flags, fp_flags; + int i; + + rt_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR | NOUVEAU_BO_DUMMY; + tx_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_DUMMY; + fp_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_DUMMY; /* Render targets */ if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR0_OFFSET, 1); - OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[0], rt_flags, + curie, NV40TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv40->rt[0], rt_flags); + OUT_RELOCm(nv40->rt[0], rt_flags, + curie, NV40TCL_COLOR0_OFFSET, 1); + OUT_RELOCl(nv40->rt[0], 0, rt_flags); } if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 1); - OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[1], rt_flags, + curie, NV40TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv40->rt[1], rt_flags); + OUT_RELOCm(nv40->rt[1], rt_flags, + curie, NV40TCL_COLOR1_OFFSET, 1); + OUT_RELOCl(nv40->rt[1], 0, rt_flags); } if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[2], rt_flags, + curie, NV40TCL_DMA_COLOR2, 1); + OUT_RELOCo(nv40->rt[2], rt_flags); + OUT_RELOCm(nv40->rt[2], rt_flags, + curie, NV40TCL_COLOR2_OFFSET, 1); + OUT_RELOCl(nv40->rt[2], 0, rt_flags); } if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[3], rt_flags, + curie, NV40TCL_DMA_COLOR3, 1); + OUT_RELOCo(nv40->rt[3], rt_flags); + OUT_RELOCm(nv40->rt[3], rt_flags, + curie, NV40TCL_COLOR3_OFFSET, 1); + OUT_RELOCl(nv40->rt[3], 0, rt_flags); } if (nv40->zeta) { - BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_DMA_ZETA, 1); + OUT_RELOCo(nv40->zeta, rt_flags); + OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv40->zeta, 0, rt_flags); } /* Texture images */ for (i = 0; i < 16; i++) { if (!(nv40->fp_samplers & (1 << i))) continue; - BEGIN_RING(curie, NV40TCL_TEX_OFFSET(i), 2); - OUT_RELOCl(nv40->tex[i].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCm(nv40->tex[i].buffer, tx_flags, + curie, NV40TCL_TEX_OFFSET(i), 2); + OUT_RELOCl(nv40->tex[i].buffer, 0, tx_flags); OUT_RELOCd(nv40->tex[i].buffer, nv40->tex[i].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, + tx_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); } /* Fragment program */ - BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (nv40->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, - NV40TCL_FP_ADDRESS_DMA1); + OUT_RELOCm(nv40->fragprog.active->buffer, fp_flags, + curie, NV40TCL_FP_ADDRESS, 1); + OUT_RELOC (nv40->fragprog.active->buffer, 0, + fp_flags | NOUVEAU_BO_OR | NOUVEAU_BO_LOW, + NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); +} + +void +nv40_emit_hw_state(struct nv40_context *nv40) +{ + if (nv40->dirty & NV40_NEW_FRAGPROG) { + nv40_fragprog_bind(nv40, nv40->fragprog.current); + /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ + } + + if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { + nv40_fragtex_bind(nv40); + + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (1); + nv40->dirty &= ~NV40_NEW_FRAGPROG; + } + + if (nv40->dirty & NV40_NEW_VERTPROG) { + nv40_vertprog_bind(nv40, nv40->vertprog.current); + nv40->dirty &= ~NV40_NEW_VERTPROG; + } + + nv40->dirty_samplers = 0; + + nv40_state_emit_dummy_relocs(nv40); } -- cgit v1.2.3 From 63923b8c9568520981ecdb46569f5ce626b584e6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 4 Feb 2008 01:37:07 +1100 Subject: nv40: the 0x4497 version of curie doesn't support index buffers. --- src/mesa/pipe/nv40/nv40_vbo.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 3fa8ddcb81b..0bb54c7610e 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -391,12 +391,17 @@ nv40_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { - if (indexSize != 1) { - nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, - mode, start, count); - } else { + struct nv40_context *nv40 = nv40_context(pipe); + + /* 0x4497 doesn't support real index buffers, and there doesn't appear + * to be support on any chipset for 8-bit indices. + */ + if (nv40->curie->grclass == NV44TCL || indexSize == 1) { nv40_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); + } else { + nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, + mode, start, count); } pipe->flush(pipe, 0); -- cgit v1.2.3 From 5351e7270f71dbcea9f790d4904327da87f931ba Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sun, 3 Feb 2008 23:08:06 +0100 Subject: Add NV63. --- src/mesa/pipe/nv40/nv40_context.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index d0d9583e32e..1351a79fe0a 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -190,7 +190,7 @@ nv40_init_hwctx(struct nv40_context *nv40, int curie_class) #define NV4X_GRCLASS4097_CHIPSETS 0x00000baf #define NV4X_GRCLASS4497_CHIPSETS 0x00005450 -#define NV6X_GRCLASS4497_CHIPSETS 0x00000080 +#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 struct pipe_context * nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, -- cgit v1.2.3 From f71400876b1469ce7b080cbddb5dde4f61ed78b7 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Mon, 4 Feb 2008 17:31:45 +0100 Subject: nouveau: update to latest header. --- src/mesa/pipe/nouveau/nouveau_class.h | 53 +++++++++++++++++++++-------------- 1 file changed, 32 insertions(+), 21 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h index e3c284095de..95f646a9915 100644 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -15,10 +15,10 @@ Stephane Marchesin, Serge Martin, Sylvain Munaut, + Simon Raffeiner, Ben Skeggs, Erik Waling, koala_br, - sturmflut. All Rights Reserved. @@ -1390,10 +1390,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 #define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST 0x10000000 #define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR 0x20000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x30000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x40000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x50000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x60000000 #define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) #define NV04_DX5_TEXTURED_TRIANGLE_BLEND 0x00000310 #define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT 0 @@ -1853,10 +1849,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_TX_FILTER_MAGNIFY_MASK 0xf0000000 #define NV10TCL_TX_FILTER_MAGNIFY_NEAREST 0x10000000 #define NV10TCL_TX_FILTER_MAGNIFY_LINEAR 0x20000000 -#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x30000000 -#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x40000000 -#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x50000000 -#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x60000000 #define NV10TCL_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) #define NV10TCL_TX_PALETTE_OFFSET__SIZE 0x00000002 #define NV10TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) @@ -4033,6 +4025,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_FOG_EQUATION_LINEAR 0x000008d4 #define NV34TCL_FOG_EQUATION_QUADRATIC 0x000008d8 #define NV34TCL_FP_ACTIVE_PROGRAM 0x000008e4 +#define NV34TCL_FP_ACTIVE_PROGRAM_DMA0 (1 << 0) +#define NV34TCL_FP_ACTIVE_PROGRAM_DMA1 (1 << 1) +#define NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_SHIFT 2 +#define NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_MASK 0xfffffffc #define NV34TCL_RC_COLOR0 0x000008ec #define NV34TCL_RC_COLOR0_B_SHIFT 0 #define NV34TCL_RC_COLOR0_B_MASK 0x000000ff @@ -4187,8 +4183,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) #define NV34TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 #define NV34TCL_FP_REG_CONTROL 0x00001450 -#define NV34TCL_FP_REG_CONTROL_USED_REGS_SHIFT 16 -#define NV34TCL_FP_REG_CONTROL_USED_REGS_MASK 0xffff0000 +#define NV34TCL_FP_REG_CONTROL_UNK1_SHIFT 16 +#define NV34TCL_FP_REG_CONTROL_UNK1_MASK 0xffff0000 +#define NV34TCL_FP_REG_CONTROL_UNK0_SHIFT 0 +#define NV34TCL_FP_REG_CONTROL_UNK0_MASK 0x0000ffff #define NV34TCL_VP_CLIP_PLANES_ENABLE 0x00001478 #define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE0 (1 << 1) #define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE1 (1 << 5) @@ -4331,9 +4329,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_OFFSET__SIZE 0x00000004 #define NV34TCL_TX_FORMAT(x) (0x00001a04+((x)*32)) #define NV34TCL_TX_FORMAT__SIZE 0x00000004 +#define NV34TCL_TX_FORMAT_DMA0 (1 << 0) +#define NV34TCL_TX_FORMAT_DMA1 (1 << 1) #define NV34TCL_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV34TCL_TX_FORMAT_COMPONENTS_SHIFT 4 -#define NV34TCL_TX_FORMAT_COMPONENTS_MASK 0x000000f0 +#define NV34TCL_TX_FORMAT_DIMS_SHIFT 4 +#define NV34TCL_TX_FORMAT_DIMS_MASK 0x000000f0 +#define NV34TCL_TX_FORMAT_DIMS_1D 0x00000010 +#define NV34TCL_TX_FORMAT_DIMS_2D 0x00000020 +#define NV34TCL_TX_FORMAT_DIMS_3D 0x00000030 #define NV34TCL_TX_FORMAT_FORMAT_SHIFT 8 #define NV34TCL_TX_FORMAT_FORMAT_MASK 0x0000ff00 #define NV34TCL_TX_FORMAT_FORMAT_L8 0x00000000 @@ -4458,10 +4461,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_FILTER_MAGNIFY_MASK 0x0f000000 #define NV34TCL_TX_FILTER_MAGNIFY_NEAREST 0x01000000 #define NV34TCL_TX_FILTER_MAGNIFY_LINEAR 0x02000000 -#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_NEAREST 0x03000000 -#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_NEAREST 0x04000000 -#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST_MIPMAP_LINEAR 0x05000000 -#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV34TCL_TX_FILTER_SIGNED_BLUE (1 << 28) +#define NV34TCL_TX_FILTER_SIGNED_GREEN (1 << 29) +#define NV34TCL_TX_FILTER_SIGNED_RED (1 << 30) +#define NV34TCL_TX_FILTER_SIGNED_ALPHA (1 << 31) #define NV34TCL_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) #define NV34TCL_TX_NPOT_SIZE__SIZE 0x00000004 #define NV34TCL_TX_NPOT_SIZE_H_SHIFT 0 @@ -4488,6 +4491,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VERTEX_ATTR_4F_W__SIZE 0x00000010 #define NV34TCL_FP_CONTROL 0x00001d60 #define NV34TCL_FP_CONTROL_USES_KIL (1 << 7) +#define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_SHIFT 0 +#define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_MASK 0x0000000f #define NV34TCL_MULTISAMPLE_CONTROL 0x00001d7c #define NV34TCL_CLEAR_DEPTH_VALUE 0x00001d8c #define NV34TCL_CLEAR_COLOR_VALUE 0x00001d90 @@ -5074,6 +5079,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x00001000 #define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x00001200 #define NV40TCL_TEX_FORMAT_FORMAT_HILO8 0x00001800 +#define NV40TCL_TEX_FORMAT_FORMAT_RGBA16F 0x00001a00 +#define NV40TCL_TEX_FORMAT_FORMAT_RGBA32F 0x00001b00 #define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 #define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 #define NV40TCL_TEX_FORMAT_DIMS_1D 0x00000010 @@ -5202,10 +5209,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_TEX_FILTER_MAG_MASK 0x0f000000 #define NV40TCL_TEX_FILTER_MAG_NEAREST 0x01000000 #define NV40TCL_TEX_FILTER_MAG_LINEAR 0x02000000 -#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_NEAREST 0x03000000 -#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_NEAREST 0x04000000 -#define NV40TCL_TEX_FILTER_MAG_NEAREST_MIPMAP_LINEAR 0x05000000 -#define NV40TCL_TEX_FILTER_MAG_LINEAR_MIPMAP_LINEAR 0x06000000 #define NV40TCL_TEX_SIZE0(x) (0x00001a18+((x)*32)) #define NV40TCL_TEX_SIZE0__SIZE 0x00000010 #define NV40TCL_TEX_SIZE0_H_SHIFT 0 @@ -6064,6 +6067,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50TCL_CLEAR_BUFFERS 0x000019d0 #define NV50TCL_COLOR_MASK(x) (0x00001a00+((x)*4)) #define NV50TCL_COLOR_MASK__SIZE 0x00000008 +#define NV50TCL_COLOR_MASK_R_SHIFT 0 +#define NV50TCL_COLOR_MASK_R_MASK 0x0000000f +#define NV50TCL_COLOR_MASK_G_SHIFT 4 +#define NV50TCL_COLOR_MASK_G_MASK 0x000000f0 +#define NV50TCL_COLOR_MASK_B_SHIFT 8 +#define NV50TCL_COLOR_MASK_B_MASK 0x00000f00 +#define NV50TCL_COLOR_MASK_A_SHIFT 12 +#define NV50TCL_COLOR_MASK_A_MASK 0x0000f000 #define NV50_COMPUTE 0x000050c0 -- cgit v1.2.3 From 9bd17ea3a3755d77ff0d1225052dbb824309fa3d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 6 Feb 2008 00:26:49 +1100 Subject: nv40: cleanup state handling a bit --- src/mesa/pipe/nouveau/nouveau_stateobj.h | 139 +++++++++ src/mesa/pipe/nv40/nv40_context.h | 18 +- src/mesa/pipe/nv40/nv40_fragprog.c | 18 +- src/mesa/pipe/nv40/nv40_fragtex.c | 38 +-- src/mesa/pipe/nv40/nv40_state.c | 464 ++++++++++++++++--------------- src/mesa/pipe/nv40/nv40_state.h | 68 +---- src/mesa/pipe/nv40/nv40_state_emit.c | 70 +---- src/mesa/pipe/nv40/nv40_vbo.c | 81 +++--- 8 files changed, 462 insertions(+), 434 deletions(-) create mode 100644 src/mesa/pipe/nouveau/nouveau_stateobj.h diff --git a/src/mesa/pipe/nouveau/nouveau_stateobj.h b/src/mesa/pipe/nouveau/nouveau_stateobj.h new file mode 100644 index 00000000000..8dfc0e9e9ac --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_stateobj.h @@ -0,0 +1,139 @@ +#ifndef __NOUVEAU_STATEOBJ_H__ +#define __NOUVEAU_STATEOBJ_H__ + +struct nouveau_stateobj_reloc { + struct pipe_buffer *bo; + + unsigned offset; + unsigned packet; + + unsigned data; + unsigned flags; + unsigned vor; + unsigned tor; +}; + +struct nouveau_stateobj { + int refcount; + + unsigned *push; + struct nouveau_stateobj_reloc *reloc; + + unsigned *cur; + unsigned cur_packet; + unsigned cur_reloc; +}; + +static inline struct nouveau_stateobj * +so_new(unsigned push, unsigned reloc) +{ + struct nouveau_stateobj *so; + + so = malloc(sizeof(struct nouveau_stateobj)); + so->refcount = 0; + so->push = malloc(sizeof(unsigned) * push); + so->reloc = malloc(sizeof(struct nouveau_stateobj_reloc) * reloc); + + so->cur = so->push; + so->cur_reloc = so->cur_packet = 0; + + return so; +} + +static inline void +so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) +{ + struct nouveau_stateobj *so; + + so = *pso; + if (so) { + if (--so->refcount <= 0) { + free(so->push); + free(so->reloc); + free(so); + } + *pso = NULL; + } + + if (ref) { + ref->refcount++; + *pso = ref; + } +} + +static inline void +so_data(struct nouveau_stateobj *so, unsigned data) +{ + (*so->cur++) = (data); + so->cur_packet += 4; +} + +static inline void +so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, + unsigned mthd, unsigned size) +{ + so->cur_packet = (gr->subc << 13) | (1 << 18) | (mthd - 4); + so_data(so, (gr->subc << 13) | (size << 18) | mthd); +} + +static inline void +so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + struct nouveau_stateobj_reloc *r = &so->reloc[so->cur_reloc++]; + + r->bo = bo; + r->offset = so->cur - so->push; + r->packet = so->cur_packet; + r->data = data; + r->flags = flags; + r->vor = vor; + r->tor = tor; + so_data(so, data); +} + +static inline void +so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) +{ + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + unsigned nr, i; + + nr = so->cur - so->push; + if (pb->remaining < nr) + nvws->push_flush(nvws->channel, nr); + pb->remaining -= nr; + + memcpy(pb->cur, so->push, nr * 4); + for (i = 0; i < so->cur_reloc; i++) { + struct nouveau_stateobj_reloc *r = &so->reloc[i]; + + nvws->push_reloc(nvws->channel, pb->cur + r->offset, r->bo, + r->data, r->flags, r->vor, r->tor); + } + pb->cur += nr; +} + +static inline void +so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) +{ + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + unsigned i; + + i = so->cur_reloc << 1; + if (nvws->channel->pushbuf->remaining < i) + nvws->push_flush(nvws->channel, i); + nvws->channel->pushbuf->remaining -= i; + + for (i = 0; i < so->cur_reloc; i++) { + struct nouveau_stateobj_reloc *r = &so->reloc[i]; + + nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->packet, + (r->flags & + (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART)) | + NOUVEAU_BO_DUMMY, 0, 0); + nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->data, + r->flags | NOUVEAU_BO_DUMMY, r->vor, r->tor); + } +} + +#endif diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 2d069619d7c..1a31f00ad6a 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -13,6 +13,7 @@ #define NOUVEAU_PUSH_CONTEXT(ctx) \ struct nv40_context *ctx = nv40 #include "pipe/nouveau/nouveau_push.h" +#include "pipe/nouveau/nouveau_stateobj.h" #include "nv40_state.h" @@ -47,20 +48,9 @@ struct nv40_context { unsigned fp_samplers; unsigned vp_samplers; - uint32_t rt_enable; - struct pipe_buffer *rt[4]; - struct pipe_buffer *zeta; - - struct { - struct pipe_buffer *buffer; - uint32_t format; - } tex[16]; - - unsigned vb_enable; - struct { - struct pipe_buffer *buffer; - unsigned delta; - } vb[16]; + struct nouveau_stateobj *so_framebuffer; + struct nouveau_stateobj *so_fragtex[16]; + struct nouveau_stateobj *so_vtxbuf; struct { struct nouveau_resource *exec_heap; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index ef320393d94..b351c853866 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -759,6 +759,7 @@ void nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) { struct pipe_winsys *ws = nv40->pipe.winsys; + struct nouveau_stateobj *so; int i; if (!fp->translated) { @@ -815,13 +816,16 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->on_hw = TRUE; } - BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, - NV40TCL_FP_ADDRESS_DMA1); - BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); - OUT_RING (fp->fp_control); + so = so_new(4, 1); + so_method(so, nv40->curie, NV40TCL_FP_ADDRESS, 1); + so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); + so_method(so, nv40->curie, NV40TCL_FP_CONTROL, 1); + so_data (so, fp->fp_control); + + so_emit(nv40->nvws, so); + so_ref(so, &fp->so); nv40->fragprog.active = fp; } diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 7c5ecd5c566..283d49704af 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -59,8 +59,10 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; struct pipe_texture *pt = &nv40mt->base; struct nv40_texture_format *tf; + struct nouveau_stateobj *so; uint32_t txf, txs, txp; int swizzled = 0; /*XXX: implement in region code? */ + unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; tf = nv40_fragtex_format(pt->format); if (!tf) @@ -101,25 +103,24 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txs = tf->swizzle; - nv40->tex[unit].buffer = nv40mt->buffer; - nv40->tex[unit].format = txf; - - BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); - OUT_RELOCl(nv40->tex[unit].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv40->tex[unit].buffer, nv40->tex[unit].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, - NV40TCL_TEX_FORMAT_DMA1); - OUT_RING (ps->wrap); - OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | + so = so_new(16, 2); + so_method(so, nv40->curie, NV40TCL_TEX_OFFSET(unit), 8); + so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, + NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); + so_data (so, ps->wrap); + so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en | (0x00078000) /* mipmap related? */); - OUT_RING (txs); - OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); - OUT_RING ((pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); - OUT_RING (ps->bcol); - BEGIN_RING(curie, NV40TCL_TEX_SIZE1(unit), 1); - OUT_RING ((pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); + so_data (so, txs); + so_data (so, ps->filt | 0x3fd6 /*voodoo*/); + so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | + pt->height[0]); + so_data (so, ps->bcol); + so_method(so, nv40->curie, NV40TCL_TEX_SIZE1(unit), 1); + so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); + + so_emit(nv40->nvws, so); + so_ref (so, &nv40->so_fragtex[unit]); } void @@ -133,6 +134,7 @@ nv40_fragtex_bind(struct nv40_context *nv40) unit = ffs(samplers) - 1; samplers &= ~(1 << unit); + so_ref(NULL, &nv40->so_fragtex[unit]); BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); OUT_RING (0); } diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index bb435b106b4..125134afdcd 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -9,59 +9,59 @@ static void * nv40_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { - struct nv40_blend_state *cb; - - cb = malloc(sizeof(struct nv40_blend_state)); - - cb->b_enable = cso->blend_enable ? 1 : 0; - cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | - (nvgl_blend_func(cso->rgb_src_factor))); - cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | - (nvgl_blend_func(cso->rgb_dst_factor))); - cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | - (nvgl_blend_eqn(cso->rgb_func))); + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(16, 0); + + if (cso->blend_enable) { + so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 3); + so_data (so, 1); + so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | + nvgl_blend_func(cso->rgb_src_factor)); + so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | + nvgl_blend_func(cso->rgb_dst_factor)); + so_method(so, nv40->curie, NV40TCL_BLEND_EQUATION, 1); + so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | + nvgl_blend_eqn(cso->rgb_func)); + } else { + so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 1); + so_data (so, 0); + } - cb->l_enable = cso->logicop_enable ? 1 : 0; - cb->l_op = nvgl_logicop_func(cso->logicop_func); + so_method(so, nv40->curie, NV40TCL_COLOR_MASK, 1); + so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); - cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | - ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | - ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | - ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + if (cso->logicop_enable) { + so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + so_data (so, 1); + so_data (so, nvgl_logicop_func(cso->logicop_func)); + } else { + so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); + so_data (so, 0); + } - cb->d_enable = cso->dither ? 1 : 0; + so_method(so, nv40->curie, NV40TCL_DITHER_ENABLE, 1); + so_data (so, cso->dither ? 1 : 0); - return (void *)cb; + return (void *)so; } static void nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_blend_state *cb = hwcso; - BEGIN_RING(curie, NV40TCL_DITHER_ENABLE, 1); - OUT_RING (cb->d_enable); - - BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 3); - OUT_RING (cb->b_enable); - OUT_RING (cb->b_srcfunc); - OUT_RING (cb->b_dstfunc); - BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 1); - OUT_RING (cb->b_eqn); - - BEGIN_RING(curie, NV40TCL_COLOR_MASK, 1); - OUT_RING (cb->c_mask); - - BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (cb->l_enable); - OUT_RING (cb->l_op); + so_emit(nv40->nvws, hwcso); } static void nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); } @@ -261,8 +261,8 @@ static void * nv40_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { - struct nv40_rasterizer_state *rs; - int i; + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(32, 0); /*XXX: ignored: * light_twoside @@ -272,165 +272,163 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, * multisample * offset_units / offset_scale */ - rs = malloc(sizeof(struct nv40_rasterizer_state)); - - rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; - rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; - rs->line_smooth_en = cso->line_smooth ? 1 : 0; - rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; - rs->line_stipple = (cso->line_stipple_pattern << 16) | - cso->line_stipple_factor; + so_method(so, nv40->curie, NV40TCL_SHADE_MODEL, 1); + so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : + NV40TCL_SHADE_MODEL_SMOOTH); - rs->point_size = *(uint32_t*)&cso->point_size; + so_method(so, nv40->curie, NV40TCL_LINE_WIDTH, 2); + so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); + so_data (so, cso->line_smooth ? 1 : 0); + so_method(so, nv40->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + so_data (so, cso->line_stipple_enable ? 1 : 0); + so_data (so, (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor); - rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; - rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; + so_method(so, nv40->curie, NV40TCL_POINT_SIZE, 1); + so_data (so, fui(cso->point_size)); + so_method(so, nv40->curie, NV40TCL_POLYGON_MODE_FRONT, 6); if (cso->front_winding == PIPE_WINDING_CCW) { - rs->front_face = NV40TCL_FRONT_FACE_CCW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV40TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_CW: + so_data(so, NV40TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, 0); + break; + } + so_data(so, NV40TCL_FRONT_FACE_CCW); } else { - rs->front_face = NV40TCL_FRONT_FACE_CW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV40TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_CW: + so_data(so, NV40TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, 0); + break; + } + so_data(so, NV40TCL_FRONT_FACE_CW); } + so_data(so, cso->poly_smooth ? 1 : 0); + so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CCW) - rs->cull_face = NV40TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV40TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_CW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CW) - rs->cull_face = NV40TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV40TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_BOTH: - rs->cull_face_en = 1; - rs->cull_face = NV40TCL_CULL_FACE_FRONT_AND_BACK; - break; - case PIPE_WINDING_NONE: - default: - rs->cull_face_en = 0; - rs->cull_face = 0; - break; - } + so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, cso->poly_stipple_enable ? 1 : 0); + so_method(so, nv40->curie, NV40TCL_POINT_SPRITE, 1); if (cso->point_sprite) { - rs->point_sprite = (1 << 0); + unsigned psctl = (1 << 0), i; + for (i = 0; i < 8; i++) { if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) - rs->point_sprite |= (1 << (8 + i)); + psctl |= (1 << (8 + i)); } + + so_data(so, psctl); } else { - rs->point_sprite = 0; + so_data(so, 0); } - return (void *)rs; + return (void *)so; } static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_rasterizer_state *rs = hwcso; - - BEGIN_RING(curie, NV40TCL_SHADE_MODEL, 1); - OUT_RING (rs->shade_model); - - BEGIN_RING(curie, NV40TCL_LINE_WIDTH, 2); - OUT_RING (rs->line_width); - OUT_RING (rs->line_smooth_en); - BEGIN_RING(curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); - OUT_RING (rs->line_stipple_en); - OUT_RING (rs->line_stipple); - - BEGIN_RING(curie, NV40TCL_POINT_SIZE, 1); - OUT_RING (rs->point_size); - - BEGIN_RING(curie, NV40TCL_POLYGON_MODE_FRONT, 6); - OUT_RING (rs->poly_mode_front); - OUT_RING (rs->poly_mode_back); - OUT_RING (rs->cull_face); - OUT_RING (rs->front_face); - OUT_RING (rs->poly_smooth_en); - OUT_RING (rs->cull_face_en); - - BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); - OUT_RING (rs->poly_stipple_en); - - BEGIN_RING(curie, NV40TCL_POINT_SPRITE, 1); - OUT_RING (rs->point_sprite); + + so_emit(nv40->nvws, hwcso); } static void nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); -} + struct nouveau_stateobj *so = hwcso; -static void -nv40_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, - unsigned idx, struct nv40_stencil_push *hw) -{ - hw->enable = cso->stencil[idx].enabled ? 1 : 0; - hw->wmask = cso->stencil[idx].write_mask; - hw->func = nvgl_comparison_op(cso->stencil[idx].func); - hw->ref = cso->stencil[idx].ref_value; - hw->vmask = cso->stencil[idx].value_mask; - hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); - hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); - hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); + so_ref(NULL, &so); } static void * nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { - struct nv40_depth_stencil_alpha_state *hw; - - hw = malloc(sizeof(struct nv40_depth_stencil_alpha_state)); - - hw->depth.func = nvgl_comparison_op(cso->depth.func); - hw->depth.write_enable = cso->depth.writemask ? 1 : 0; - hw->depth.test_enable = cso->depth.enabled ? 1 : 0; - - nv40_translate_stencil(cso, 0, &hw->stencil.front); - nv40_translate_stencil(cso, 1, &hw->stencil.back); + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(32, 0); + + so_method(so, nv40->curie, NV40TCL_DEPTH_FUNC, 3); + so_data (so, nvgl_comparison_op(cso->depth.func)); + so_data (so, cso->depth.writemask ? 1 : 0); + so_data (so, cso->depth.enabled ? 1 : 0); + + so_method(so, nv40->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_data (so, cso->alpha.enabled ? 1 : 0); + so_data (so, nvgl_comparison_op(cso->alpha.func)); + so_data (so, float_to_ubyte(cso->alpha.ref)); + + if (cso->stencil[0].enabled) { + so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_data (so, cso->stencil[0].enabled ? 1 : 0); + so_data (so, cso->stencil[0].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[0].func)); + so_data (so, cso->stencil[0].ref_value); + so_data (so, cso->stencil[0].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); + } else { + so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_data (so, 0); + } - hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; - hw->alpha.func = nvgl_comparison_op(cso->alpha.func); - hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + if (cso->stencil[1].enabled) { + so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_data (so, cso->stencil[1].enabled ? 1 : 0); + so_data (so, cso->stencil[1].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[1].func)); + so_data (so, cso->stencil[1].ref_value); + so_data (so, cso->stencil[1].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); + } else { + so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_data (so, 0); + } - return (void *)hw; + return (void *)so; } static void nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_depth_stencil_alpha_state *hw = hwcso; - - BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&hw->depth, 3); - BEGIN_RING(curie, NV40TCL_STENCIL_FRONT_ENABLE, 16); - OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); - OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); - BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); - OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); + + so_emit(nv40->nvws, hwcso); } static void nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); } static void * @@ -502,12 +500,16 @@ nv40_set_blend_color(struct pipe_context *pipe, const struct pipe_blend_color *bcol) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(2, 0); + + so_method(so, nv40->curie, NV40TCL_BLEND_COLOR, 1); + so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0))); - BEGIN_RING(curie, NV40TCL_BLEND_COLOR, 1); - OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0)); + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void @@ -540,6 +542,8 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, struct pipe_surface *rt[4], *zeta; uint32_t rt_enable, rt_format, w, h; int i, colour_format = 0, zeta_format = 0; + struct nouveau_stateobj *so = so_new(64, 10); + unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; rt_enable = 0; for (i = 0; i < 4; i++) { @@ -603,66 +607,78 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - nv40->rt[0] = rt[0]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); - OUT_RING (rt[0]->pitch * rt[0]->cpp); - OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR0, 1); + so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR0_PITCH, 2); + so_data (so, rt[0]->pitch * rt[0]->cpp); + so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - nv40->rt[1] = rt[1]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); - OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RING (rt[1]->pitch * rt[1]->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR1, 1); + so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR1_OFFSET, 2); + so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - nv40->rt[2] = rt[2]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); - OUT_RING (rt[2]->pitch * rt[2]->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR2, 1); + so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR2_OFFSET, 1); + so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->curie, NV40TCL_COLOR2_PITCH, 1); + so_data (so, rt[2]->pitch * rt[2]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - nv40->rt[3] = rt[3]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); - OUT_RING (rt[3]->pitch * rt[3]->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR3, 1); + so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR3_OFFSET, 1); + so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->curie, NV40TCL_COLOR3_PITCH, 1); + so_data (so, rt[3]->pitch * rt[3]->cpp); } if (zeta_format) { - nv40->zeta = zeta->buffer; - BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); - OUT_RING (zeta->pitch * zeta->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_ZETA, 1); + so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_ZETA_OFFSET, 1); + so_reloc (so, zeta->buffer, zeta->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->curie, NV40TCL_ZETA_PITCH, 1); + so_data (so, zeta->pitch * zeta->cpp); } - nv40->rt_enable = rt_enable; - BEGIN_RING(curie, NV40TCL_RT_ENABLE, 1); - OUT_RING (rt_enable); - BEGIN_RING(curie, NV40TCL_RT_HORIZ, 3); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - OUT_RING (rt_format); - BEGIN_RING(curie, NV40TCL_VIEWPORT_HORIZ, 2); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - BEGIN_RING(curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0); - OUT_RING (((h - 1) << 16) | 0); + so_method(so, nv40->curie, NV40TCL_RT_ENABLE, 1); + so_data (so, rt_enable); + so_method(so, nv40->curie, NV40TCL_RT_HORIZ, 3); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_data (so, rt_format); + so_method(so, nv40->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_method(so, nv40->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_data (so, ((w - 1) << 16) | 0); + so_data (so, ((h - 1) << 16) | 0); + + so_emit(nv40->nvws, so); + so_ref (so, &nv40->so_framebuffer); } static void @@ -670,9 +686,15 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(33, 0); + unsigned i; - BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); - OUT_RINGp ((uint32_t *)stipple->stipple, 32); + so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, stipple->stipple[i]); + + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void @@ -680,10 +702,14 @@ nv40_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(3, 0); + + so_method(so, nv40->curie, NV40TCL_SCISSOR_HORIZ, 2); + so_data (so, ((s->maxx - s->minx) << 16) | s->minx); + so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - BEGIN_RING(curie, NV40TCL_SCISSOR_HORIZ, 2); - OUT_RING (((s->maxx - s->minx) << 16) | s->minx); - OUT_RING (((s->maxy - s->miny) << 16) | s->miny); + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void @@ -691,16 +717,20 @@ nv40_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) { struct nv40_context *nv40 = nv40_context(pipe); - - BEGIN_RING(curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); - OUT_RINGf (vpt->translate[0]); - OUT_RINGf (vpt->translate[1]); - OUT_RINGf (vpt->translate[2]); - OUT_RINGf (vpt->translate[3]); - OUT_RINGf (vpt->scale[0]); - OUT_RINGf (vpt->scale[1]); - OUT_RINGf (vpt->scale[2]); - OUT_RINGf (vpt->scale[3]); + struct nouveau_stateobj *so = so_new(9, 0); + + so_method(so, nv40->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(vpt->translate[0])); + so_data (so, fui(vpt->translate[1])); + so_data (so, fui(vpt->translate[2])); + so_data (so, fui(vpt->translate[3])); + so_data (so, fui(vpt->scale[0])); + so_data (so, fui(vpt->scale[1])); + so_data (so, fui(vpt->scale[2])); + so_data (so, fui(vpt->scale[3])); + + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 24eea36db3a..e82ab9de98a 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -3,20 +3,6 @@ #include "pipe/p_state.h" -struct nv40_blend_state { - uint32_t b_enable; - uint32_t b_srcfunc; - uint32_t b_dstfunc; - uint32_t b_eqn; - - uint32_t l_enable; - uint32_t l_op; - - uint32_t c_mask; - - uint32_t d_enable; -}; - struct nv40_sampler_state { uint32_t fmt; uint32_t wrap; @@ -25,29 +11,6 @@ struct nv40_sampler_state { uint32_t bcol; }; -struct nv40_rasterizer_state { - uint32_t shade_model; - - uint32_t line_width; - uint32_t line_smooth_en; - uint32_t line_stipple_en; - uint32_t line_stipple; - - uint32_t point_size; - - uint32_t poly_smooth_en; - uint32_t poly_stipple_en; - - uint32_t poly_mode_front; - uint32_t poly_mode_back; - - uint32_t front_face; - uint32_t cull_face; - uint32_t cull_face_en; - - uint32_t point_sprite; -}; - struct nv40_vertex_program_exec { uint32_t data[4]; boolean has_branch_offset; @@ -99,36 +62,7 @@ struct nv40_fragment_program { struct pipe_buffer *buffer; uint32_t fp_control; -}; - -struct nv40_stencil_push { - uint32_t enable; - uint32_t wmask; - uint32_t func; - uint32_t ref; - uint32_t vmask; - uint32_t fail; - uint32_t zfail; - uint32_t zpass; -}; - -struct nv40_depth_stencil_alpha_state { - struct { - uint32_t func; - uint32_t write_enable; - uint32_t test_enable; - } depth; - - struct { - struct nv40_stencil_push back; - struct nv40_stencil_push front; - } stencil; - - struct { - uint32_t enabled; - uint32_t func; - uint32_t ref; - } alpha; + struct nouveau_stateobj *so; }; struct nv40_miptree { diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 66b98d5fab3..3a22cd4bd56 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -11,77 +11,15 @@ static void nv40_state_emit_dummy_relocs(struct nv40_context *nv40) { - unsigned rt_flags, tx_flags, fp_flags; - int i; + unsigned i; - rt_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR | NOUVEAU_BO_DUMMY; - tx_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_DUMMY; - fp_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_DUMMY; - - /* Render targets */ - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - OUT_RELOCm(nv40->rt[0], rt_flags, - curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv40->rt[0], rt_flags); - OUT_RELOCm(nv40->rt[0], rt_flags, - curie, NV40TCL_COLOR0_OFFSET, 1); - OUT_RELOCl(nv40->rt[0], 0, rt_flags); - } - - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - OUT_RELOCm(nv40->rt[1], rt_flags, - curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv40->rt[1], rt_flags); - OUT_RELOCm(nv40->rt[1], rt_flags, - curie, NV40TCL_COLOR1_OFFSET, 1); - OUT_RELOCl(nv40->rt[1], 0, rt_flags); - } - - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - OUT_RELOCm(nv40->rt[2], rt_flags, - curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(nv40->rt[2], rt_flags); - OUT_RELOCm(nv40->rt[2], rt_flags, - curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(nv40->rt[2], 0, rt_flags); - } - - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - OUT_RELOCm(nv40->rt[3], rt_flags, - curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(nv40->rt[3], rt_flags); - OUT_RELOCm(nv40->rt[3], rt_flags, - curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(nv40->rt[3], 0, rt_flags); - } - - if (nv40->zeta) { - OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(nv40->zeta, rt_flags); - OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv40->zeta, 0, rt_flags); - } - - /* Texture images */ + so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); for (i = 0; i < 16; i++) { if (!(nv40->fp_samplers & (1 << i))) continue; - OUT_RELOCm(nv40->tex[i].buffer, tx_flags, - curie, NV40TCL_TEX_OFFSET(i), 2); - OUT_RELOCl(nv40->tex[i].buffer, 0, tx_flags); - OUT_RELOCd(nv40->tex[i].buffer, nv40->tex[i].format, - tx_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, - NV40TCL_TEX_FORMAT_DMA1); + so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); } - - /* Fragment program */ - OUT_RELOCm(nv40->fragprog.active->buffer, fp_flags, - curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (nv40->fragprog.active->buffer, 0, - fp_flags | NOUVEAU_BO_OR | NOUVEAU_BO_LOW, - NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); + so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); } void diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 0bb54c7610e..e2cb3fda8f2 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -97,13 +97,13 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, } static void -nv40_vbo_arrays_update(struct nv40_context *nv40) +nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, + unsigned ib_format) { struct nv40_vertex_program *vp = nv40->vertprog.active; - uint32_t inputs, vtxfmt[16]; - int hw, num_hw; - - nv40->vb_enable = 0; + struct nouveau_stateobj *vtxbuf, *vtxfmt; + unsigned inputs, hw, num_hw; + unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; inputs = vp->ir; for (hw = 0; hw < 16 && inputs; hw++) { @@ -114,73 +114,64 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) } num_hw++; + vtxbuf = so_new(20, 18); + so_method(vtxbuf, nv40->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + vtxfmt = so_new(17, 0); + so_method(vtxfmt, nv40->curie, NV40TCL_VTXFMT(0), num_hw); + inputs = vp->ir; for (hw = 0; hw < num_hw; hw++) { struct pipe_vertex_element *ve; struct pipe_vertex_buffer *vb; if (!(inputs & (1 << hw))) { - vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); continue; } ve = &nv40->vtxelt[hw]; vb = &nv40->vtxbuf[ve->vertex_buffer_index]; - if (vb->pitch == 0) { - vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; - if (nv40_vbo_static_attrib(nv40, hw, ve, vb) == TRUE) - continue; + if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; } - nv40->vb_enable |= (1 << hw); - nv40->vb[hw].delta = vb->buffer_offset + ve->src_offset; - nv40->vb[hw].buffer = vb->buffer; + so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, + vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + 0, NV40TCL_VTXBUF_ADDRESS_DMA1); + so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | + (nv40_vbo_ncomp(ve->src_format) << + NV40TCL_VTXFMT_SIZE_SHIFT) | + nv40_vbo_type(ve->src_format))); + } - vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_ncomp(ve->src_format) << - NV40TCL_VTXFMT_SIZE_SHIFT) | - nv40_vbo_type(ve->src_format)); + if (ib) { + so_method(vtxbuf, nv40->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); } - BEGIN_RING(curie, NV40TCL_VTXFMT(0), num_hw); - OUT_RINGp (vtxfmt, num_hw); + so_emit(nv40->nvws, vtxfmt); + so_emit(nv40->nvws, vtxbuf); + so_ref (vtxbuf, &nv40->so_vtxbuf); + so_ref (NULL, &vtxfmt); } static boolean nv40_vbo_validate_state(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned ib_format) { - unsigned inputs; - nv40_emit_hw_state(nv40); - - if (nv40->dirty & NV40_NEW_ARRAYS) { - nv40_vbo_arrays_update(nv40); + if (nv40->dirty & NV40_NEW_ARRAYS || ib) { + nv40_vbo_arrays_update(nv40, ib, ib_format); nv40->dirty &= ~NV40_NEW_ARRAYS; } - inputs = nv40->vb_enable; - while (inputs) { - unsigned a = ffs(inputs) - 1; - - inputs &= ~(1 << a); - - BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(a), 1); - OUT_RELOC (nv40->vb[a].buffer, nv40->vb[a].delta, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV40TCL_VTXBUF_ADDRESS_DMA1); - } - - if (ib) { - BEGIN_RING(curie, NV40TCL_IDXBUF_ADDRESS, 2); - OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD); - OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_OR, - 0, NV40TCL_IDXBUF_FORMAT_DMA1); - } + so_emit_reloc_markers(nv40->nvws, nv40->so_vtxbuf); BEGIN_RING(curie, 0x1710, 1); OUT_RING (0); /* vtx cache flush */ -- cgit v1.2.3 From ae905056d4feb5a39d956a38ab377f4d78bf4065 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 9 Feb 2008 14:08:54 +1100 Subject: nv40: delay all state emit until before draw --- src/mesa/pipe/nouveau/nouveau_stateobj.h | 2 +- src/mesa/pipe/nv40/nv40_context.h | 22 +++++++++++++++++++--- src/mesa/pipe/nv40/nv40_state.c | 26 +++++++++++++++++--------- src/mesa/pipe/nv40/nv40_state_emit.c | 25 +++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_vbo.c | 5 ++++- 5 files changed, 66 insertions(+), 14 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_stateobj.h b/src/mesa/pipe/nouveau/nouveau_stateobj.h index 8dfc0e9e9ac..58167a24de7 100644 --- a/src/mesa/pipe/nouveau/nouveau_stateobj.h +++ b/src/mesa/pipe/nouveau/nouveau_stateobj.h @@ -30,7 +30,7 @@ so_new(unsigned push, unsigned reloc) struct nouveau_stateobj *so; so = malloc(sizeof(struct nouveau_stateobj)); - so->refcount = 0; + so->refcount = 1; so->push = malloc(sizeof(unsigned) * push); so->reloc = malloc(sizeof(struct nouveau_stateobj_reloc) * reloc); diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 1a31f00ad6a..4aa34847e85 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -22,9 +22,18 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV40_NEW_VERTPROG (1 << 1) -#define NV40_NEW_FRAGPROG (1 << 2) -#define NV40_NEW_ARRAYS (1 << 3) +#define NV40_NEW_BLEND (1 << 0) +#define NV40_NEW_RAST (1 << 1) +#define NV40_NEW_ZSA (1 << 2) +#define NV40_NEW_SAMPLER (1 << 3) +#define NV40_NEW_FB (1 << 4) +#define NV40_NEW_STIPPLE (1 << 5) +#define NV40_NEW_SCISSOR (1 << 6) +#define NV40_NEW_VIEWPORT (1 << 7) +#define NV40_NEW_BCOL (1 << 8) +#define NV40_NEW_VERTPROG (1 << 9) +#define NV40_NEW_FRAGPROG (1 << 10) +#define NV40_NEW_ARRAYS (1 << 11) struct nv40_context { struct pipe_context pipe; @@ -51,6 +60,13 @@ struct nv40_context { struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; struct nouveau_stateobj *so_vtxbuf; + struct nouveau_stateobj *so_blend; + struct nouveau_stateobj *so_rast; + struct nouveau_stateobj *so_zsa; + struct nouveau_stateobj *so_bcol; + struct nouveau_stateobj *so_scissor; + struct nouveau_stateobj *so_viewport; + struct nouveau_stateobj *so_stipple; struct { struct nouveau_resource *exec_heap; diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 125134afdcd..ab53b03cb08 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -53,7 +53,8 @@ nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_emit(nv40->nvws, hwcso); + so_ref(hwcso, &nv40->so_blend); + nv40->dirty |= NV40_NEW_BLEND; } static void @@ -354,7 +355,8 @@ nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_emit(nv40->nvws, hwcso); + so_ref(hwcso, &nv40->so_rast); + nv40->dirty |= NV40_NEW_RAST; } static void @@ -420,7 +422,8 @@ nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_emit(nv40->nvws, hwcso); + so_ref(hwcso, &nv40->so_zsa); + nv40->dirty |= NV40_NEW_ZSA; } static void @@ -508,8 +511,9 @@ nv40_set_blend_color(struct pipe_context *pipe, (float_to_ubyte(bcol->color[1]) << 8) | (float_to_ubyte(bcol->color[2]) << 0))); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_bcol); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_BCOL; } static void @@ -677,8 +681,9 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, so_data (so, ((w - 1) << 16) | 0); so_data (so, ((h - 1) << 16) | 0); - so_emit(nv40->nvws, so); - so_ref (so, &nv40->so_framebuffer); + so_ref(so, &nv40->so_framebuffer); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_FB; } static void @@ -693,8 +698,9 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, for (i = 0; i < 32; i++) so_data(so, stipple->stipple[i]); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_stipple); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_STIPPLE; } static void @@ -708,8 +714,9 @@ nv40_set_scissor_state(struct pipe_context *pipe, so_data (so, ((s->maxx - s->minx) << 16) | s->minx); so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_scissor); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_SCISSOR; } static void @@ -729,8 +736,9 @@ nv40_set_viewport_state(struct pipe_context *pipe, so_data (so, fui(vpt->scale[2])); so_data (so, fui(vpt->scale[3])); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_viewport); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_VIEWPORT; } static void diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 3a22cd4bd56..a10c9955480 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -25,6 +25,30 @@ nv40_state_emit_dummy_relocs(struct nv40_context *nv40) void nv40_emit_hw_state(struct nv40_context *nv40) { + if (nv40->dirty & NV40_NEW_FB) + so_emit(nv40->nvws, nv40->so_framebuffer); + + if (nv40->dirty & NV40_NEW_BLEND) + so_emit(nv40->nvws, nv40->so_blend); + + if (nv40->dirty & NV40_NEW_RAST) + so_emit(nv40->nvws, nv40->so_rast); + + if (nv40->dirty & NV40_NEW_ZSA) + so_emit(nv40->nvws, nv40->so_zsa); + + if (nv40->dirty & NV40_NEW_BCOL) + so_emit(nv40->nvws, nv40->so_bcol); + + if (nv40->dirty & NV40_NEW_SCISSOR) + so_emit(nv40->nvws, nv40->so_scissor); + + if (nv40->dirty & NV40_NEW_VIEWPORT) + so_emit(nv40->nvws, nv40->so_viewport); + + if (nv40->dirty & NV40_NEW_STIPPLE) + so_emit(nv40->nvws, nv40->so_stipple); + if (nv40->dirty & NV40_NEW_FRAGPROG) { nv40_fragprog_bind(nv40, nv40->fragprog.current); /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ @@ -46,6 +70,7 @@ nv40_emit_hw_state(struct nv40_context *nv40) } nv40->dirty_samplers = 0; + nv40->dirty = 0; nv40_state_emit_dummy_relocs(nv40); } diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index e2cb3fda8f2..fd1d884193b 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -158,6 +158,7 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, so_emit(nv40->nvws, vtxfmt); so_emit(nv40->nvws, vtxbuf); so_ref (vtxbuf, &nv40->so_vtxbuf); + so_ref (NULL, &vtxbuf); so_ref (NULL, &vtxfmt); } @@ -165,8 +166,10 @@ static boolean nv40_vbo_validate_state(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned ib_format) { + unsigned vdn = nv40->dirty & NV40_NEW_ARRAYS; + nv40_emit_hw_state(nv40); - if (nv40->dirty & NV40_NEW_ARRAYS || ib) { + if (vdn || ib) { nv40_vbo_arrays_update(nv40, ib, ib_format); nv40->dirty &= ~NV40_NEW_ARRAYS; } -- cgit v1.2.3 From 8973ca12a58a5bd1db434f63f7e8772a090f285b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 9 Feb 2008 16:25:29 +1100 Subject: nouveau: interface updates --- src/mesa/pipe/nouveau/nouveau_gldefs.h | 2 +- src/mesa/pipe/nv40/nv40_context.c | 2 ++ src/mesa/pipe/nv40/nv40_fragprog.c | 1 + src/mesa/pipe/nv40/nv40_fragtex.c | 1 + src/mesa/pipe/nv40/nv40_miptree.c | 35 +++++++++++++++++----------------- src/mesa/pipe/nv50/nv50_miptree.c | 5 +++-- 6 files changed, 25 insertions(+), 21 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_gldefs.h b/src/mesa/pipe/nouveau/nouveau_gldefs.h index 8ba3bdef649..e1015c93a27 100644 --- a/src/mesa/pipe/nouveau/nouveau_gldefs.h +++ b/src/mesa/pipe/nouveau/nouveau_gldefs.h @@ -189,7 +189,7 @@ nvgl_primitive(unsigned prim) { case PIPE_PRIM_POLYGON: return 0x000a; default: - assert(0); + return 0x0001; } } diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index d0d9583e32e..2754429c5b9 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -71,6 +71,8 @@ nv40_get_paramf(struct pipe_context *pipe, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index b351c853866..86cf74835e1 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -826,6 +826,7 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) so_emit(nv40->nvws, so); so_ref(so, &fp->so); + so_ref(NULL, &so); nv40->fragprog.active = fp; } diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 283d49704af..c87e3618317 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -121,6 +121,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_emit(nv40->nvws, so); so_ref (so, &nv40->so_fragtex[unit]); + so_ref (NULL, &so); } void diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 48062d1ae8a..df70feaa057 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -53,27 +53,26 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) nv40mt->total_size = offset; } -static void -nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +static struct pipe_texture * +nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) { struct pipe_winsys *ws = pipe->winsys; - struct nv40_miptree *nv40mt; - - nv40mt = realloc(*pt, sizeof(struct nv40_miptree)); - if (!nv40mt) - return; - *pt = NULL; - - nv40_miptree_layout(nv40mt); - - nv40mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - nv40mt->total_size); - if (!nv40mt->buffer) { - free(nv40mt); - return; + struct nv40_miptree *mt; + + mt = malloc(sizeof(struct nv40_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + nv40_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + free(mt); + return NULL; } - - *pt = &nv40mt->base; + + return &mt->base; } static void diff --git a/src/mesa/pipe/nv50/nv50_miptree.c b/src/mesa/pipe/nv50/nv50_miptree.c index 51442d64f3e..0c034ed4387 100644 --- a/src/mesa/pipe/nv50/nv50_miptree.c +++ b/src/mesa/pipe/nv50/nv50_miptree.c @@ -4,10 +4,11 @@ #include "nv50_context.h" -static void -nv50_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +static struct pipe_texture * +nv50_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) { NOUVEAU_ERR("unimplemented\n"); + return NULL; } static void -- cgit v1.2.3 From 0ac4e8cd496dc0b74e71df5f6906af8bf7a74ccb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 10 Feb 2008 15:16:50 +1100 Subject: nv40: prep for multiple pipe contexts on a single hw channel --- src/mesa/pipe/nv40/nv40_context.c | 237 ++++++++++++++++++++----------------- src/mesa/pipe/nv40/nv40_context.h | 29 +++-- src/mesa/pipe/nv40/nv40_fragprog.c | 4 +- src/mesa/pipe/nv40/nv40_fragtex.c | 4 +- src/mesa/pipe/nv40/nv40_query.c | 11 +- src/mesa/pipe/nv40/nv40_state.c | 83 ++++++------- src/mesa/pipe/nv40/nv40_vbo.c | 8 +- src/mesa/pipe/nv40/nv40_vertprog.c | 4 +- 8 files changed, 206 insertions(+), 174 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 2754429c5b9..4343d493cc5 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -5,6 +5,10 @@ #include "nv40_context.h" +#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf +#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 +#define NV6X_GRCLASS4497_CHIPSETS 0x00000080 + static const char * nv40_get_name(struct pipe_context *pipe) { @@ -93,7 +97,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) } if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv40->sync, 0); + nvws->notifier_reset(nv40->hw->sync, 0); BEGIN_RING(curie, 0x104, 1); OUT_RING (0); BEGIN_RING(curie, 0x100, 1); @@ -103,104 +107,32 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) FIRE_RING(); if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv40->sync, 0, 0, 2000); + nvws->notifier_wait(nv40->hw->sync, 0, 0, 2000); } static void -nv40_destroy(struct pipe_context *pipe) +nv40_channel_takedown(struct nv40_channel_context *cnv40) { - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; - - if (nv40->draw) - draw_destroy(nv40->draw); - - nvws->res_free(&nv40->vertprog.exec_heap); - nvws->res_free(&nv40->vertprog.data_heap); - - nvws->res_free(&nv40->query_heap); - nvws->notifier_free(&nv40->query); - - nvws->notifier_free(&nv40->sync); - - nvws->grobj_free(&nv40->curie); - - free(nv40); + struct nouveau_winsys *nvws = cnv40->nvws; + + nvws->res_free(&cnv40->vp_exec_heap); + nvws->res_free(&cnv40->vp_data_heap); + nvws->res_free(&cnv40->query_heap); + nvws->notifier_free(&cnv40->query); + nvws->notifier_free(&cnv40->sync); + nvws->grobj_free(&cnv40->curie); + free(cnv40); } -static boolean -nv40_init_hwctx(struct nv40_context *nv40, int curie_class) +static struct nv40_channel_context * +nv40_channel_init(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) { - struct nouveau_winsys *nvws = nv40->nvws; + struct nv40_channel_context *cnv40 = NULL; + struct nouveau_stateobj *so; + unsigned curie_class = 0; int ret; - ret = nvws->grobj_alloc(nvws, curie_class, &nv40->curie); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - BEGIN_RING(curie, NV40TCL_DMA_NOTIFY, 1); - OUT_RING (nv40->sync->handle); - BEGIN_RING(curie, NV40TCL_DMA_TEXTURE0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(curie, NV40TCL_DMA_VTXBUF0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(curie, NV40TCL_DMA_FENCE, 2); - OUT_RING (0); - OUT_RING (nv40->query->handle); - BEGIN_RING(curie, NV40TCL_DMA_UNK01AC, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - - BEGIN_RING(curie, 0x1ea4, 3); - OUT_RING (0x00000010); - OUT_RING (0x01000100); - OUT_RING (0xff800006); - - /* vtxprog output routing */ - BEGIN_RING(curie, 0x1fc4, 1); - OUT_RING (0x06144321); - BEGIN_RING(curie, 0x1fc8, 2); - OUT_RING (0xedcba987); - OUT_RING (0x00000021); - BEGIN_RING(curie, 0x1fd0, 1); - OUT_RING (0x00171615); - BEGIN_RING(curie, 0x1fd4, 1); - OUT_RING (0x001b1a19); - - BEGIN_RING(curie, 0x1ef8, 1); - OUT_RING (0x0020ffff); - BEGIN_RING(curie, 0x1d64, 1); - OUT_RING (0x00d30000); - BEGIN_RING(curie, 0x1e94, 1); - OUT_RING (0x00000001); - - FIRE_RING (); - return TRUE; -} - -#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf -#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 -#define NV6X_GRCLASS4497_CHIPSETS 0x00000080 - -struct pipe_context * -nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv40_context *nv40; - int curie_class = 0, ret; - switch (chipset & 0xf0) { case 0x40: if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) @@ -218,65 +150,152 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, } if (!curie_class) { - NOUVEAU_ERR("Unknown NV4x chipset: NV%02x\n", chipset); + NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset); return NULL; } - nv40 = CALLOC_STRUCT(nv40_context); - if (!nv40) + cnv40 = calloc(1, sizeof(struct nv40_channel_context)); + if (!cnv40) return NULL; - nv40->chipset = chipset; - nv40->nvws = nvws; + cnv40->chipset = chipset; + cnv40->nvws = nvws; /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &nv40->sync); + ret = nvws->notifier_alloc(nvws, 1, &cnv40->sync); if (ret) { NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv40_destroy(&nv40->pipe); + nv40_channel_takedown(cnv40); return NULL; } /* Query objects */ - ret = nvws->notifier_alloc(nvws, 32, &nv40->query); + ret = nvws->notifier_alloc(nvws, 32, &cnv40->query); if (ret) { NOUVEAU_ERR("Error initialising query objects: %d\n", ret); - nv40_destroy(&nv40->pipe); + nv40_channel_takedown(cnv40); return NULL; } - ret = nvws->res_init(&nv40->query_heap, 0, 32); + ret = nvws->res_init(&cnv40->query_heap, 0, 32); if (ret) { NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); - nv40_destroy(&nv40->pipe); + nv40_channel_takedown(cnv40); return NULL; } /* Vtxprog resources */ - if (nvws->res_init(&nv40->vertprog.exec_heap, 0, 512) || - nvws->res_init(&nv40->vertprog.data_heap, 0, 256)) { - nv40_destroy(&nv40->pipe); + if (nvws->res_init(&cnv40->vp_exec_heap, 0, 512) || + nvws->res_init(&cnv40->vp_data_heap, 0, 256)) { + nv40_channel_takedown(cnv40); return NULL; } + /* 3D object */ + ret = nvws->grobj_alloc(nvws, curie_class, &cnv40->curie); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + /* Static curie initialisation */ - if (!nv40_init_hwctx(nv40, curie_class)) { + so = so_new(128, 0); + so_method(so, cnv40->curie, NV40TCL_DMA_NOTIFY, 1); + so_data (so, cnv40->sync->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_TEXTURE0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR1, 1); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_VTXBUF0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_FENCE, 2); + so_data (so, 0); + so_data (so, cnv40->query->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_UNK01AC, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR2, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + + so_method(so, cnv40->curie, 0x1ea4, 3); + so_data (so, 0x00000010); + so_data (so, 0x01000100); + so_data (so, 0xff800006); + + /* vtxprog output routing */ + so_method(so, cnv40->curie, 0x1fc4, 1); + so_data (so, 0x06144321); + so_method(so, cnv40->curie, 0x1fc8, 2); + so_data (so, 0xedcba987); + so_data (so, 0x00000021); + so_method(so, cnv40->curie, 0x1fd0, 1); + so_data (so, 0x00171615); + so_method(so, cnv40->curie, 0x1fd4, 1); + so_data (so, 0x001b1a19); + + so_method(so, cnv40->curie, 0x1ef8, 1); + so_data (so, 0x0020ffff); + so_method(so, cnv40->curie, 0x1d64, 1); + so_data (so, 0x00d30000); + so_method(so, cnv40->curie, 0x1e94, 1); + so_data (so, 0x00000001); + + so_emit(nvws, so); + so_ref(NULL, &so); + nvws->push_flush(nvws->channel, 0); + + return cnv40; +} + +static void +nv40_destroy(struct pipe_context *pipe) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + if (nv40->draw) + draw_destroy(nv40->draw); + + if (nv40->hw) { + if (--nv40->hw->refcount == 0) + nv40_channel_takedown(nv40->hw); + } + + free(nv40); +} + +struct pipe_context * +nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv40_context *nv40; + + nv40 = calloc(1, sizeof(struct nv40_context)); + if (!nv40) + return NULL; + + nv40->hw = nv40_channel_init(ws, nvws, chipset); + if (!nv40->hw) { nv40_destroy(&nv40->pipe); return NULL; } - /* Pipe context setup */ - nv40->pipe.winsys = pipe_winsys; + nv40->chipset = chipset; + nv40->nvws = nvws; + nv40->pipe.winsys = ws; nv40->pipe.destroy = nv40_destroy; nv40->pipe.get_name = nv40_get_name; nv40->pipe.get_vendor = nv40_get_vendor; nv40->pipe.get_param = nv40_get_param; nv40->pipe.get_paramf = nv40_get_paramf; - nv40->pipe.draw_arrays = nv40_draw_arrays; nv40->pipe.draw_elements = nv40_draw_elements; nv40->pipe.clear = nv40_clear; - nv40->pipe.flush = nv40_flush; nv40_init_query_functions(nv40); diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 4aa34847e85..d7c9ee7851e 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -11,7 +11,7 @@ #include "pipe/nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv40_context *ctx = nv40 + struct nv40_channel_context *ctx = nv40->hw #include "pipe/nouveau/nouveau_push.h" #include "pipe/nouveau/nouveau_stateobj.h" @@ -35,20 +35,34 @@ #define NV40_NEW_FRAGPROG (1 << 10) #define NV40_NEW_ARRAYS (1 << 11) -struct nv40_context { - struct pipe_context pipe; +struct nv40_channel_context { struct nouveau_winsys *nvws; + unsigned refcount; - struct draw_context *draw; + unsigned chipset; - int chipset; + /* HW graphics objects */ struct nouveau_grobj *curie; struct nouveau_notifier *sync; - /* query objects */ + /* Query object resources */ struct nouveau_notifier *query; struct nouveau_resource *query_heap; + /* Vtxprog resources */ + struct nouveau_resource *vp_exec_heap; + struct nouveau_resource *vp_data_heap; +}; + +struct nv40_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct nv40_channel_context *hw; + struct draw_context *draw; + + int chipset; + uint32_t dirty; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; @@ -69,9 +83,6 @@ struct nv40_context { struct nouveau_stateobj *so_stipple; struct { - struct nouveau_resource *exec_heap; - struct nouveau_resource *data_heap; - struct nv40_vertex_program *active; struct nv40_vertex_program *current; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 86cf74835e1..26a196531be 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -817,11 +817,11 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) } so = so_new(4, 1); - so_method(so, nv40->curie, NV40TCL_FP_ADDRESS, 1); + so_method(so, nv40->hw->curie, NV40TCL_FP_ADDRESS, 1); so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); - so_method(so, nv40->curie, NV40TCL_FP_CONTROL, 1); + so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); so_data (so, fp->fp_control); so_emit(nv40->nvws, so); diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index c87e3618317..d278ce18975 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -104,7 +104,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txs = tf->swizzle; so = so_new(16, 2); - so_method(so, nv40->curie, NV40TCL_TEX_OFFSET(unit), 8); + so_method(so, nv40->hw->curie, NV40TCL_TEX_OFFSET(unit), 8); so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); @@ -116,7 +116,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); so_data (so, ps->bcol); - so_method(so, nv40->curie, NV40TCL_TEX_SIZE1(unit), 1); + so_method(so, nv40->hw->curie, NV40TCL_TEX_SIZE1(unit), 1); so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); so_emit(nv40->nvws, so); diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index 06f41fe84f7..eb305e64446 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -45,9 +45,9 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - if (nv40->nvws->res_alloc(nv40->query_heap, 1, NULL, &q->object)) + if (nv40->nvws->res_alloc(nv40->hw->query_heap, 1, NULL, &q->object)) assert(0); - nv40->nvws->notifier_reset(nv40->query, q->object->start); + nv40->nvws->notifier_reset(nv40->hw->query, q->object->start); BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); OUT_RING (1); @@ -82,16 +82,17 @@ nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, if (!q->ready) { unsigned status; - status = nvws->notifier_status(nv40->query, q->object->start); + status = nvws->notifier_status(nv40->hw->query, + q->object->start); if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { if (wait == FALSE) return FALSE; - nvws->notifier_wait(nv40->query, q->object->start, + nvws->notifier_wait(nv40->hw->query, q->object->start, NV_NOTIFY_STATE_STATUS_COMPLETED, 0); } - q->result = nvws->notifier_retval(nv40->query, + q->result = nvws->notifier_retval(nv40->hw->query, q->object->start); q->ready = TRUE; nvws->res_free(&q->object); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index ab53b03cb08..80e94737efb 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -10,39 +10,40 @@ nv40_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_grobj *curie = nv40->hw->curie; struct nouveau_stateobj *so = so_new(16, 0); if (cso->blend_enable) { - so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 3); + so_method(so, curie, NV40TCL_BLEND_ENABLE, 3); so_data (so, 1); so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | nvgl_blend_func(cso->rgb_src_factor)); so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | nvgl_blend_func(cso->rgb_dst_factor)); - so_method(so, nv40->curie, NV40TCL_BLEND_EQUATION, 1); + so_method(so, curie, NV40TCL_BLEND_EQUATION, 1); so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | nvgl_blend_eqn(cso->rgb_func)); } else { - so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 1); + so_method(so, curie, NV40TCL_BLEND_ENABLE, 1); so_data (so, 0); } - so_method(so, nv40->curie, NV40TCL_COLOR_MASK, 1); + so_method(so, curie, NV40TCL_COLOR_MASK, 1); so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); if (cso->logicop_enable) { - so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); so_data (so, 1); so_data (so, nvgl_logicop_func(cso->logicop_func)); } else { - so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); + so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); so_data (so, 0); } - so_method(so, nv40->curie, NV40TCL_DITHER_ENABLE, 1); + so_method(so, curie, NV40TCL_DITHER_ENABLE, 1); so_data (so, cso->dither ? 1 : 0); return (void *)so; @@ -274,22 +275,22 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, * offset_units / offset_scale */ - so_method(so, nv40->curie, NV40TCL_SHADE_MODEL, 1); + so_method(so, nv40->hw->curie, NV40TCL_SHADE_MODEL, 1); so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : NV40TCL_SHADE_MODEL_SMOOTH); - so_method(so, nv40->curie, NV40TCL_LINE_WIDTH, 2); + so_method(so, nv40->hw->curie, NV40TCL_LINE_WIDTH, 2); so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); so_data (so, cso->line_smooth ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + so_method(so, nv40->hw->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); so_data (so, cso->line_stipple_enable ? 1 : 0); so_data (so, (cso->line_stipple_pattern << 16) | cso->line_stipple_factor); - so_method(so, nv40->curie, NV40TCL_POINT_SIZE, 1); + so_method(so, nv40->hw->curie, NV40TCL_POINT_SIZE, 1); so_data (so, fui(cso->point_size)); - so_method(so, nv40->curie, NV40TCL_POLYGON_MODE_FRONT, 6); + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_MODE_FRONT, 6); if (cso->front_winding == PIPE_WINDING_CCW) { so_data(so, nvgl_polygon_mode(cso->fill_ccw)); so_data(so, nvgl_polygon_mode(cso->fill_cw)); @@ -330,10 +331,10 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, cso->poly_smooth ? 1 : 0); so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, cso->poly_stipple_enable ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_POINT_SPRITE, 1); + so_method(so, nv40->hw->curie, NV40TCL_POINT_SPRITE, 1); if (cso->point_sprite) { unsigned psctl = (1 << 0), i; @@ -374,18 +375,18 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(32, 0); - so_method(so, nv40->curie, NV40TCL_DEPTH_FUNC, 3); + so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); so_data (so, nvgl_comparison_op(cso->depth.func)); so_data (so, cso->depth.writemask ? 1 : 0); so_data (so, cso->depth.enabled ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_method(so, nv40->hw->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); so_data (so, cso->alpha.enabled ? 1 : 0); so_data (so, nvgl_comparison_op(cso->alpha.func)); so_data (so, float_to_ubyte(cso->alpha.ref)); if (cso->stencil[0].enabled) { - so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); so_data (so, cso->stencil[0].enabled ? 1 : 0); so_data (so, cso->stencil[0].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[0].func)); @@ -395,12 +396,12 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); } else { - so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); so_data (so, 0); } if (cso->stencil[1].enabled) { - so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); so_data (so, cso->stencil[1].enabled ? 1 : 0); so_data (so, cso->stencil[1].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[1].func)); @@ -410,7 +411,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); } else { - so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); so_data (so, 0); } @@ -505,7 +506,7 @@ nv40_set_blend_color(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(2, 0); - so_method(so, nv40->curie, NV40TCL_BLEND_COLOR, 1); + so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | (float_to_ubyte(bcol->color[0]) << 16) | (float_to_ubyte(bcol->color[1]) << 8) | @@ -611,73 +612,73 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR0, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR0_PITCH, 2); + so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); so_data (so, rt[0]->pitch * rt[0]->cpp); so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR1, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR1_OFFSET, 2); + so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); so_data (so, rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR2, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR2_OFFSET, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->curie, NV40TCL_COLOR2_PITCH, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); so_data (so, rt[2]->pitch * rt[2]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR3, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR3_OFFSET, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->curie, NV40TCL_COLOR3_PITCH, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); so_data (so, rt[3]->pitch * rt[3]->cpp); } if (zeta_format) { - so_method(so, nv40->curie, NV40TCL_DMA_ZETA, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_ZETA_OFFSET, 1); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); so_reloc (so, zeta->buffer, zeta->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->curie, NV40TCL_ZETA_PITCH, 1); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); so_data (so, zeta->pitch * zeta->cpp); } - so_method(so, nv40->curie, NV40TCL_RT_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); so_data (so, rt_enable); - so_method(so, nv40->curie, NV40TCL_RT_HORIZ, 3); + so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); so_data (so, (w << 16) | 0); so_data (so, (h << 16) | 0); so_data (so, rt_format); - so_method(so, nv40->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); so_data (so, (w << 16) | 0); so_data (so, (h << 16) | 0); - so_method(so, nv40->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); so_data (so, ((w - 1) << 16) | 0); so_data (so, ((h - 1) << 16) | 0); @@ -694,7 +695,7 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, struct nouveau_stateobj *so = so_new(33, 0); unsigned i; - so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); for (i = 0; i < 32; i++) so_data(so, stipple->stipple[i]); @@ -710,7 +711,7 @@ nv40_set_scissor_state(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(3, 0); - so_method(so, nv40->curie, NV40TCL_SCISSOR_HORIZ, 2); + so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); so_data (so, ((s->maxx - s->minx) << 16) | s->minx); so_data (so, ((s->maxy - s->miny) << 16) | s->miny); @@ -726,7 +727,7 @@ nv40_set_viewport_state(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(9, 0); - so_method(so, nv40->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); so_data (so, fui(vpt->translate[0])); so_data (so, fui(vpt->translate[1])); so_data (so, fui(vpt->translate[2])); diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index fd1d884193b..552058b3ae7 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -115,9 +115,9 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, num_hw++; vtxbuf = so_new(20, 18); - so_method(vtxbuf, nv40->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); vtxfmt = so_new(17, 0); - so_method(vtxfmt, nv40->curie, NV40TCL_VTXFMT(0), num_hw); + so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); inputs = vp->ir; for (hw = 0; hw < num_hw; hw++) { @@ -149,7 +149,7 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, } if (ib) { - so_method(vtxbuf, nv40->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, 0, NV40TCL_IDXBUF_FORMAT_DMA1); @@ -390,7 +390,7 @@ nv40_draw_elements(struct pipe_context *pipe, /* 0x4497 doesn't support real index buffers, and there doesn't appear * to be support on any chipset for 8-bit indices. */ - if (nv40->curie->grclass == NV44TCL || indexSize == 1) { + if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { nv40_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); } else { diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index e15ddbbcde5..415b3c70c79 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -648,7 +648,7 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) /* Allocate hw vtxprog exec slots */ if (!vp->exec) { - struct nouveau_resource *heap = nv40->vertprog.exec_heap; + struct nouveau_resource *heap = nv40->hw->vp_exec_heap; uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { @@ -668,7 +668,7 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) /* Allocate hw vtxprog const slots */ if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv40->vertprog.data_heap; + struct nouveau_resource *heap = nv40->hw->vp_data_heap; if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { while (heap->next && heap->size < vp->nr_consts) { -- cgit v1.2.3 From af344660d7b77b8adad17cd8018253c8b08045cf Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 10 Feb 2008 15:19:18 +1100 Subject: nouveau: memory leak --- src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index 7d7fefa8016..e1a9271395b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -136,6 +136,7 @@ nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); nouveau_bo_del(&nvbuf->bo); + free(nvbuf); } static void * -- cgit v1.2.3 From 5ba3dbe2cc8a9af5cae01f45eaf497f834400170 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 12 Feb 2008 15:13:37 +1100 Subject: nv40: fix inline u08/u16 indices --- src/mesa/pipe/nv40/nv40_vbo.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 552058b3ae7..4e9cdb4585a 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -235,9 +235,9 @@ nv40_draw_elements_u08(struct nv40_context *nv40, void *ib, } while (count) { - push = MIN2(count, 2046); + push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push); + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); @@ -260,9 +260,9 @@ nv40_draw_elements_u16(struct nv40_context *nv40, void *ib, } while (count) { - push = MIN2(count, 2046); + push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push); + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); -- cgit v1.2.3 From 6be095ae15d411c82f7417c98a87867dbdf1947b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 13 Feb 2008 14:27:32 +1100 Subject: nv40: remove use of pt->first_level --- src/mesa/pipe/nv40/nv40_fragtex.c | 3 +-- src/mesa/pipe/nv40/nv40_miptree.c | 6 +++--- 2 files changed, 4 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index d278ce18975..69c0fcb5e05 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -70,8 +70,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txf = ps->fmt; txf |= tf->format | 0x8000; - txf |= ((pt->last_level - pt->first_level + 1) << - NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); + txf |= ((pt->last_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); if (1) /* XXX */ txf |= NV40TCL_TEX_FORMAT_NO_BORDER; diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index df70feaa057..1b3c27dc45b 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -23,7 +23,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) nr_faces = 1; } - for (l = pt->first_level; l <= pt->last_level; l++) { + for (l = 0; l <= pt->last_level; l++) { pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; @@ -44,7 +44,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) } for (f = 0; f < nr_faces; f++) { - for (l = pt->first_level; l <= pt->last_level; l++) { + for (l = 0; l <= pt->last_level; l++) { nv40mt->level[l].image_offset[f] = offset; offset += nv40mt->level[l].pitch * pt->height[l]; } @@ -87,7 +87,7 @@ nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) int l; pipe_buffer_reference(ws, &nv40mt->buffer, NULL); - for (l = mt->first_level; l <= mt->last_level; l++) { + for (l = 0; l <= mt->last_level; l++) { if (nv40mt->level[l].image_offset) free(nv40mt->level[l].image_offset); } -- cgit v1.2.3 From 263de4b814eb78cc971ad28f173985b61b77465c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 13 Feb 2008 19:33:41 +1100 Subject: nv40: attempt at obeying sampler min_lod/max_lod/lod_bias --- src/mesa/pipe/nv40/nv40_context.c | 2 +- src/mesa/pipe/nv40/nv40_fragtex.c | 5 ++--- src/mesa/pipe/nv40/nv40_state.c | 14 ++++++++++++++ 3 files changed, 17 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 4343d493cc5..a6e5f332e63 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -74,7 +74,7 @@ nv40_get_paramf(struct pipe_context *pipe, int param) case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; + return 16.0; case PIPE_CAP_BITMAP_TEXCOORD_BIAS: return 0.0; default: diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 69c0fcb5e05..5af5fbe7465 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -108,10 +108,9 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); so_data (so, ps->wrap); - so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en | - (0x00078000) /* mipmap related? */); + so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en); so_data (so, txs); - so_data (so, ps->filt | 0x3fd6 /*voodoo*/); + so_data (so, ps->filt | 0x2000 /*voodoo*/); so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); so_data (so, ps->bcol); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 80e94737efb..bcd244528d5 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -193,6 +193,20 @@ nv40_sampler_state_create(struct pipe_context *pipe, ps->filt = filter; + { + float limit; + + limit = CLAMP(cso->lod_bias, -16.0, 15.0); + ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff; + + limit = CLAMP(cso->max_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 7; + + limit = CLAMP(cso->min_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 19; + } + + if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { switch (cso->compare_func) { case PIPE_FUNC_NEVER: -- cgit v1.2.3 From 2effa9b36cc47ca3fc0acc21908f5bb132eca3e7 Mon Sep 17 00:00:00 2001 From: Maarten Maathuis Date: Wed, 13 Feb 2008 22:57:09 +0100 Subject: nv40: Avoid a nasty array overflow leading to a corrupt memory pointer. --- src/mesa/pipe/nv40/nv40_fragprog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 714634396d2..14897f97982 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -716,7 +716,7 @@ nv40_fragprog_translate(struct nv40_context *nv40, assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); assert(fpc->nr_imm < MAX_IMM); - for (i = 0; i < imm->Immediate.Size; i++) + for (i = 0; i < (imm->Immediate.Size - 1); i++) vals[i] = imm->u.ImmediateFloat32[i].Float; fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); } -- cgit v1.2.3 From 73b3a29b16fe5d798026db4eeabb8d33bb6c2cb0 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sun, 10 Feb 2008 19:14:02 +0100 Subject: Hook nv30 into the build. --- src/mesa/pipe/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/Makefile b/src/mesa/pipe/Makefile index 6012b2bcead..2b03b5c2446 100644 --- a/src/mesa/pipe/Makefile +++ b/src/mesa/pipe/Makefile @@ -10,7 +10,7 @@ ifeq ($(CONFIG_NAME), linux-llvm) LLVM_DIR = llvm endif -SUBDIRS = softpipe i915simple i965simple nv40 nv50 failover pipebuffer \ +SUBDIRS = softpipe i915simple i965simple nv30 nv40 nv50 failover pipebuffer \ $(CELL_DIR) $(LLVM_DIR) -- cgit v1.2.3 From e713cb26c9adeff4e35a9b2cac35e7c025ef72b4 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 15 Feb 2008 02:23:56 +0100 Subject: nouveau: Update to latest header. --- src/mesa/pipe/nouveau/nouveau_class.h | 46 +++++++++++++++++++++-------------- 1 file changed, 28 insertions(+), 18 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h index 95f646a9915..5998945677b 100644 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ b/src/mesa/pipe/nouveau/nouveau_class.h @@ -3692,16 +3692,16 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_DMA_QUERY 0x000001a8 #define NV34TCL_DMA_IN_MEMORY7 0x000001ac #define NV34TCL_DMA_IN_MEMORY8 0x000001b0 -#define NV34TCL_VIEWPORT_HORIZ 0x00000200 -#define NV34TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV34TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV34TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV34TCL_VIEWPORT_VERT 0x00000204 -#define NV34TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV34TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV34TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV34TCL_RT_HORIZ 0x00000200 +#define NV34TCL_RT_HORIZ_X_SHIFT 0 +#define NV34TCL_RT_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_RT_HORIZ_W_SHIFT 16 +#define NV34TCL_RT_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_RT_VERT 0x00000204 +#define NV34TCL_RT_VERT_Y_SHIFT 0 +#define NV34TCL_RT_VERT_Y_MASK 0x0000ffff +#define NV34TCL_RT_VERT_H_SHIFT 16 +#define NV34TCL_RT_VERT_H_MASK 0xffff0000 #define NV34TCL_RT_FORMAT 0x00000208 #define NV34TCL_RT_FORMAT_TYPE_SHIFT 8 #define NV34TCL_RT_FORMAT_TYPE_MASK 0x00000f00 @@ -4078,17 +4078,27 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_RC_OUT_ALPHA__SIZE 0x00000008 #define NV34TCL_RC_OUT_RGB(x) (0x00000914+((x)*32)) #define NV34TCL_RC_OUT_RGB__SIZE 0x00000008 +#define NV34TCL_VIEWPORT_HORIZ 0x00000a00 +#define NV34TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV34TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV34TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_VERT 0x00000a04 +#define NV34TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV34TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV34TCL_VIEWPORT_VERT_H_MASK 0xffff0000 #define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x00000a10 #define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x00000a14 #define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x00000a18 -#define NV34TCL_VIEWPORT_SCALE0_X 0x00000a20 -#define NV34TCL_VIEWPORT_SCALE0_Y 0x00000a24 -#define NV34TCL_VIEWPORT_SCALE0_Z 0x00000a28 -#define NV34TCL_VIEWPORT_SCALE0_W 0x00000a2c -#define NV34TCL_VIEWPORT_SCALE1_X 0x00000a30 -#define NV34TCL_VIEWPORT_SCALE1_Y 0x00000a34 -#define NV34TCL_VIEWPORT_SCALE1_Z 0x00000a38 -#define NV34TCL_VIEWPORT_SCALE1_W 0x00000a3c +#define NV34TCL_VIEWPORT_TRANSLATE_X 0x00000a20 +#define NV34TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 +#define NV34TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 +#define NV34TCL_VIEWPORT_TRANSLATE_W 0x00000a2c +#define NV34TCL_VIEWPORT_SCALE_X 0x00000a30 +#define NV34TCL_VIEWPORT_SCALE_Y 0x00000a34 +#define NV34TCL_VIEWPORT_SCALE_Z 0x00000a38 +#define NV34TCL_VIEWPORT_SCALE_W 0x00000a3c #define NV34TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 #define NV34TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 #define NV34TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 -- cgit v1.2.3 From e538dc52c13eb9225afd8cb6c1099e97e723a558 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 15 Feb 2008 02:25:17 +0100 Subject: nouveau: hook in nv30. --- src/mesa/drivers/dri/nouveau_winsys/Makefile | 1 + src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c | 3 +++ src/mesa/pipe/nouveau/nouveau_winsys.h | 3 +++ 3 files changed, 7 insertions(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index f547ec43766..59ba561eb93 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -8,6 +8,7 @@ MINIGLX_SOURCES = PIPE_DRIVERS = \ $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a \ + $(TOP)/src/mesa/pipe/nv40/libnv30.a \ $(TOP)/src/mesa/pipe/nv40/libnv40.a \ $(TOP)/src/mesa/pipe/nv50/libnv50.a diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c index 8ffe89feecb..1494bd48dd4 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c @@ -80,6 +80,9 @@ nouveau_pipe_create(struct nouveau_context *nv) return NULL; switch (nv->chipset & 0xf0) { + case 0x30: + hw_create = nv30_create; + break; case 0x40: case 0x60: hw_create = nv40_create; diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h index 0b394cfce6b..818ae9afae2 100644 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ b/src/mesa/pipe/nouveau/nouveau_winsys.h @@ -49,6 +49,9 @@ struct nouveau_winsys { unsigned, unsigned, unsigned, unsigned, unsigned); }; +extern struct pipe_context * +nv30_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); + extern struct pipe_context * nv40_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); -- cgit v1.2.3 From 583f424d61d8080079a55f3c962f647b795d9337 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 15 Feb 2008 02:36:28 +0100 Subject: nouveau: add nv30. --- src/mesa/pipe/nv30/Makefile | 29 ++ src/mesa/pipe/nv30/nv30_clear.c | 12 + src/mesa/pipe/nv30/nv30_context.c | 429 ++++++++++++++++++ src/mesa/pipe/nv30/nv30_context.h | 136 ++++++ src/mesa/pipe/nv30/nv30_dma.h | 66 +++ src/mesa/pipe/nv30/nv30_draw.c | 62 +++ src/mesa/pipe/nv30/nv30_fragprog.c | 834 +++++++++++++++++++++++++++++++++++ src/mesa/pipe/nv30/nv30_fragtex.c | 160 +++++++ src/mesa/pipe/nv30/nv30_miptree.c | 105 +++++ src/mesa/pipe/nv30/nv30_query.c | 112 +++++ src/mesa/pipe/nv30/nv30_shader.h | 490 ++++++++++++++++++++ src/mesa/pipe/nv30/nv30_state.c | 740 +++++++++++++++++++++++++++++++ src/mesa/pipe/nv30/nv30_state.h | 147 ++++++ src/mesa/pipe/nv30/nv30_state_emit.c | 83 ++++ src/mesa/pipe/nv30/nv30_surface.c | 136 ++++++ src/mesa/pipe/nv30/nv30_vbo.c | 406 +++++++++++++++++ src/mesa/pipe/nv30/nv30_vertprog.c | 778 ++++++++++++++++++++++++++++++++ 17 files changed, 4725 insertions(+) create mode 100644 src/mesa/pipe/nv30/Makefile create mode 100644 src/mesa/pipe/nv30/nv30_clear.c create mode 100644 src/mesa/pipe/nv30/nv30_context.c create mode 100644 src/mesa/pipe/nv30/nv30_context.h create mode 100644 src/mesa/pipe/nv30/nv30_dma.h create mode 100644 src/mesa/pipe/nv30/nv30_draw.c create mode 100644 src/mesa/pipe/nv30/nv30_fragprog.c create mode 100644 src/mesa/pipe/nv30/nv30_fragtex.c create mode 100644 src/mesa/pipe/nv30/nv30_miptree.c create mode 100644 src/mesa/pipe/nv30/nv30_query.c create mode 100644 src/mesa/pipe/nv30/nv30_shader.h create mode 100644 src/mesa/pipe/nv30/nv30_state.c create mode 100644 src/mesa/pipe/nv30/nv30_state.h create mode 100644 src/mesa/pipe/nv30/nv30_state_emit.c create mode 100644 src/mesa/pipe/nv30/nv30_surface.c create mode 100644 src/mesa/pipe/nv30/nv30_vbo.c create mode 100644 src/mesa/pipe/nv30/nv30_vertprog.c diff --git a/src/mesa/pipe/nv30/Makefile b/src/mesa/pipe/nv30/Makefile new file mode 100644 index 00000000000..dd4b7e73cda --- /dev/null +++ b/src/mesa/pipe/nv30/Makefile @@ -0,0 +1,29 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv30 + +DRIVER_SOURCES = \ + nv30_clear.c \ + nv30_context.c \ + nv30_draw.c \ + nv30_fragprog.c \ + nv30_fragtex.c \ + nv30_miptree.c \ + nv30_query.c \ + nv30_state.c \ + nv30_state_emit.c \ + nv30_surface.c \ + nv30_vbo.c \ + nv30_vertprog.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../Makefile.template + +symlinks: + diff --git a/src/mesa/pipe/nv30/nv30_clear.c b/src/mesa/pipe/nv30/nv30_clear.c new file mode 100644 index 00000000000..71f413588ee --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv30_context.h" + +void +nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/mesa/pipe/nv30/nv30_context.c b/src/mesa/pipe/nv30/nv30_context.c new file mode 100644 index 00000000000..c56f918ad9d --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_context.c @@ -0,0 +1,429 @@ +#include "pipe/draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" + +static const char * +nv30_get_name(struct pipe_context *pipe) +{ + struct nv30_context *nv30 = nv30_context(pipe); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv30->chipset); + return buffer; +} + +static const char * +nv30_get_vendor(struct pipe_context *pipe) +{ + return "nouveau"; +} + +static int +nv30_get_param(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 16; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 1; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 2; + case PIPE_CAP_OCCLUSION_QUERY: + return 1; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 1; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv30_get_paramf(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv30_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + if (flags & PIPE_FLUSH_TEXTURE_CACHE) { + BEGIN_RING(rankine, 0x1fd8, 1); + OUT_RING (2); + BEGIN_RING(rankine, 0x1fd8, 1); + OUT_RING (1); + } + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv30->sync, 0); + BEGIN_RING(rankine, 0x104, 1); + OUT_RING (0); + BEGIN_RING(rankine, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv30->sync, 0, 0, 2000); +} + +static void +nv30_destroy(struct pipe_context *pipe) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + if (nv30->draw) + draw_destroy(nv30->draw); + + nvws->res_free(&nv30->vertprog.exec_heap); + nvws->res_free(&nv30->vertprog.data_heap); + + nvws->res_free(&nv30->query_heap); + nvws->notifier_free(&nv30->query); + + nvws->notifier_free(&nv30->sync); + + nvws->grobj_free(&nv30->rankine); + + free(nv30); +} + +static boolean +nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) +{ + struct nouveau_winsys *nvws = nv30->nvws; + int ret; + int i; + + ret = nvws->grobj_alloc(nvws, rankine_class, &nv30->rankine); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + BEGIN_RING(rankine, NV34TCL_DMA_NOTIFY, 1); + OUT_RING (nv30->sync->handle); + BEGIN_RING(rankine, NV34TCL_DMA_TEXTURE0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(rankine, NV34TCL_DMA_VTXBUF0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); +/* BEGIN_RING(rankine, NV34TCL_DMA_FENCE, 2); + OUT_RING (0); + OUT_RING (nv30->query->handle);*/ + BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY7, 1); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY8, 1); + OUT_RING (nvws->channel->vram->handle); + + for (i=1; i<8; i++) { + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(i), 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_VERT(i), 1); + OUT_RING (0); + } + + BEGIN_RING(rankine, 0x220, 1); + OUT_RING (1); + + BEGIN_RING(rankine, 0x03b0, 1); + OUT_RING (0x00100000); + BEGIN_RING(rankine, 0x1454, 1); + OUT_RING (0); + BEGIN_RING(rankine, 0x1d80, 1); + OUT_RING (3); + BEGIN_RING(rankine, 0x1450, 1); + OUT_RING (0x00030004); + + /* NEW */ + BEGIN_RING(rankine, 0x1e98, 1); + OUT_RING (0); + BEGIN_RING(rankine, 0x17e0, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x3f800000); + BEGIN_RING(rankine, 0x1f80, 16); + OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); + OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); + OUT_RING (0x0000ffff); + OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); + OUT_RING (0); OUT_RING (0); OUT_RING (0); + + BEGIN_RING(rankine, 0x120, 3); + OUT_RING (0); + OUT_RING (1); + OUT_RING (2); + + BEGIN_RING(rankine, 0x1d88, 1); + OUT_RING (0x00001200); + + BEGIN_RING(rankine, NV34TCL_RC_ENABLE, 1); + OUT_RING (0); + + /* Attempt to setup a known state.. Probably missing a heap of + * stuff here.. + */ + BEGIN_RING(rankine, NV34TCL_STENCIL_FRONT_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_DEPTH_WRITE_ENABLE, 2); + OUT_RING (0); /* wr disable */ + OUT_RING (0); /* test disable */ + BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); + OUT_RING (0x01010101); /* TR,TR,TR,TR */ + BEGIN_RING(rankine, NV34TCL_CULL_FACE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 5); + OUT_RING (0); /* Blend enable */ + OUT_RING (0); /* Blend src */ + OUT_RING (0); /* Blend dst */ + OUT_RING (0x00000000); /* Blend colour */ + OUT_RING (0x8006); /* FUNC_ADD */ + BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); + OUT_RING (0); + OUT_RING (0x1503 /*GL_COPY*/); + BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); + OUT_RING (1); + BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); + OUT_RING (0x1d01 /*GL_SMOOTH*/); + BEGIN_RING(rankine, NV34TCL_POLYGON_OFFSET_FACTOR,2); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (0x1b02 /*GL_FILL*/); + OUT_RING (0x1b02 /*GL_FILL*/); + /* - Disable texture units + * - Set fragprog to MOVR result.color, fragment.color */ + for (i=0;i<16;i++) { + BEGIN_RING(rankine, + NV34TCL_TX_ENABLE(i), 1); + OUT_RING (0); + } + /* Polygon stipple */ + BEGIN_RING(rankine, + NV34TCL_POLYGON_STIPPLE_PATTERN(0), 0x20); + for (i=0;i<0x20;i++) + OUT_RING (0xFFFFFFFF); + + int w=4096; + int h=4096; + int pitch=4096*4; + BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 5); + OUT_RING (w<<16); + OUT_RING (h<<16); + OUT_RING (0x148); /* format */ + OUT_RING (pitch << 16 | pitch); + OUT_RING (0x0); + BEGIN_RING(rankine, 0x0a00, 2); + OUT_RING ((w<<16) | 0); + OUT_RING ((h<<16) | 0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING ((w-1)<<16); + OUT_RING ((h-1)<<16); + BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); + OUT_RING (w<<16); + OUT_RING (h<<16); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); + OUT_RING (w<<16); + OUT_RING (h<<16); + + BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + + BEGIN_RING(rankine, NV34TCL_MODELVIEW_MATRIX(0), 16); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + + BEGIN_RING(rankine, NV34TCL_PROJECTION_MATRIX(0), 16); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + + BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); + OUT_RING (4096<<16); + OUT_RING (4096<<16); + + BEGIN_RING(rankine, NV34TCL_MULTISAMPLE_CONTROL, 1); + OUT_RING (0xffff0000); + + FIRE_RING (); + return TRUE; +} + +#define NV30TCL_CHIPSET_3X_MASK 0x00000003 +#define NV34TCL_CHIPSET_3X_MASK 0x00000010 +#define NV35TCL_CHIPSET_3X_MASK 0x000001e0 + +struct pipe_context * +nv30_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv30_context *nv30; + int rankine_class = 0, ret; + + if ((chipset & 0xf0) != 0x30) { + NOUVEAU_ERR("Not a NV3X chipset\n"); + return NULL; + } + + if (NV30TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { + rankine_class = 0x0397; + } else if (NV34TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { + rankine_class = 0x0697; + } else if (NV35TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { + rankine_class = 0x0497; + } else { + NOUVEAU_ERR("Unknown NV3X chipset: NV%02x\n", chipset); + return NULL; + } + + nv30 = CALLOC_STRUCT(nv30_context); + if (!nv30) + return NULL; + nv30->chipset = chipset; + nv30->nvws = nvws; + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &nv30->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Query objects */ + ret = nvws->notifier_alloc(nvws, 32, &nv30->query); + if (ret) { + NOUVEAU_ERR("Error initialising query objects: %d\n", ret); + nv30_destroy(&nv30->pipe); + return NULL; + } + + ret = nvws->res_init(&nv30->query_heap, 0, 32); + if (ret) { + NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Vtxprog resources */ + if (nvws->res_init(&nv30->vertprog.exec_heap, 0, 512) || + nvws->res_init(&nv30->vertprog.data_heap, 0, 256)) { + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Static rankine initialisation */ + if (!nv30_init_hwctx(nv30, rankine_class)) { + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Pipe context setup */ + nv30->pipe.winsys = pipe_winsys; + + nv30->pipe.destroy = nv30_destroy; + nv30->pipe.get_name = nv30_get_name; + nv30->pipe.get_vendor = nv30_get_vendor; + nv30->pipe.get_param = nv30_get_param; + nv30->pipe.get_paramf = nv30_get_paramf; + + nv30->pipe.draw_arrays = nv30_draw_arrays; + nv30->pipe.draw_elements = nv30_draw_elements; + nv30->pipe.clear = nv30_clear; + + nv30->pipe.flush = nv30_flush; + + nv30_init_query_functions(nv30); + nv30_init_surface_functions(nv30); + nv30_init_state_functions(nv30); + nv30_init_miptree_functions(nv30); + + nv30->draw = draw_create(); + assert(nv30->draw); + draw_set_rasterize_stage(nv30->draw, nv30_draw_render_stage(nv30)); + + return &nv30->pipe; +} + diff --git a/src/mesa/pipe/nv30/nv30_context.h b/src/mesa/pipe/nv30/nv30_context.h new file mode 100644 index 00000000000..d2262c5065f --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_context.h @@ -0,0 +1,136 @@ +#ifndef __NV30_CONTEXT_H__ +#define __NV30_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/draw/draw_vertex.h" + +#include "pipe/nouveau/nouveau_winsys.h" +#include "pipe/nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv30_context *ctx = nv30 +#include "pipe/nouveau/nouveau_push.h" + +#include "nv30_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV30_NEW_VERTPROG (1 << 1) +#define NV30_NEW_FRAGPROG (1 << 2) +#define NV30_NEW_ARRAYS (1 << 3) + +struct nv30_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct draw_context *draw; + + int chipset; + struct nouveau_grobj *rankine; + struct nouveau_notifier *sync; + + /* query objects */ + struct nouveau_notifier *query; + struct nouveau_resource *query_heap; + + uint32_t dirty; + + struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + uint32_t rt_enable; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; + + struct { + struct pipe_buffer *buffer; + uint32_t format; + } tex[16]; + + unsigned vb_enable; + struct { + struct pipe_buffer *buffer; + unsigned delta; + } vb[16]; + + struct { + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + + struct nv30_vertex_program *active; + + struct nv30_vertex_program *current; + struct pipe_buffer *constant_buf; + } vertprog; + + struct { + struct nv30_fragment_program *active; + + struct nv30_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; +}; + +static inline struct nv30_context * +nv30_context(struct pipe_context *pipe) +{ + return (struct nv30_context *)pipe; +} + +extern void nv30_init_state_functions(struct nv30_context *nv30); +extern void nv30_init_surface_functions(struct nv30_context *nv30); +extern void nv30_init_miptree_functions(struct nv30_context *nv30); +extern void nv30_init_query_functions(struct nv30_context *nv30); + +/* nv30_draw.c */ +extern struct draw_stage *nv30_draw_render_stage(struct nv30_context *nv30); + +/* nv30_vertprog.c */ +extern void nv30_vertprog_translate(struct nv30_context *, + struct nv30_vertex_program *); +extern void nv30_vertprog_bind(struct nv30_context *, + struct nv30_vertex_program *); +extern void nv30_vertprog_destroy(struct nv30_context *, + struct nv30_vertex_program *); + +/* nv30_fragprog.c */ +extern void nv30_fragprog_translate(struct nv30_context *, + struct nv30_fragment_program *); +extern void nv30_fragprog_bind(struct nv30_context *, + struct nv30_fragment_program *); +extern void nv30_fragprog_destroy(struct nv30_context *, + struct nv30_fragment_program *); + +/* nv30_fragtex.c */ +extern void nv30_fragtex_bind(struct nv30_context *); + +/* nv30_state.c and friends */ +extern void nv30_emit_hw_state(struct nv30_context *nv30); +extern void nv30_state_tex_update(struct nv30_context *nv30); + +/* nv30_vbo.c */ +extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv30_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned mode, unsigned start, + unsigned count); + +/* nv30_clear.c */ +extern void nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +#endif diff --git a/src/mesa/pipe/nv30/nv30_dma.h b/src/mesa/pipe/nv30/nv30_dma.h new file mode 100644 index 00000000000..6eff6b42903 --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_dma.h @@ -0,0 +1,66 @@ +#ifndef __NV30_DMA_H__ +#define __NV30_DMA_H__ + +#include "pipe/nouveau/nouveau_winsys.h" + +#define OUT_RING(data) do { \ + (*nv30->nvws->channel->pushbuf->cur++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + memcpy(nv30->nvws->channel->pushbuf->cur, (src), (size) * 4); \ + nv30->nvws->channel->pushbuf->cur += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define BEGIN_RING(obj,mthd,size) do { \ + if (nv30->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + nv30->nvws->push_flush(nv30->nvws->channel, ((size) + 1)); \ + OUT_RING((nv30->obj->subc << 13) | ((size) << 18) | (mthd)); \ + nv30->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + +#define FIRE_RING() do { \ + nv30->nvws->push_flush(nv30->nvws->channel, 0); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + nv30->nvws->push_reloc(nv30->nvws->channel, \ + nv30->nvws->channel->pushbuf->cur, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ + OUT_RING(0); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + nv30->nvws->channel->vram->handle, \ + nv30->nvws->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/mesa/pipe/nv30/nv30_draw.c b/src/mesa/pipe/nv30/nv30_draw.c new file mode 100644 index 00000000000..bdeb975ca1e --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_draw.c @@ -0,0 +1,62 @@ +#include "pipe/draw/draw_private.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" + +struct nv30_draw_stage { + struct draw_stage draw; + struct nv30_context *nv30; +}; + +static void +nv30_draw_point(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_line(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_tri(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_flush(struct draw_stage *draw, unsigned flags) +{ +} + +static void +nv30_draw_reset_stipple_counter(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_destroy(struct draw_stage *draw) +{ + free(draw); +} + +struct draw_stage * +nv30_draw_render_stage(struct nv30_context *nv30) +{ + struct nv30_draw_stage *nv30draw = CALLOC_STRUCT(nv30_draw_stage); + + nv30draw->nv30 = nv30; + nv30draw->draw.draw = nv30->draw; + nv30draw->draw.point = nv30_draw_point; + nv30draw->draw.line = nv30_draw_line; + nv30draw->draw.tri = nv30_draw_tri; + nv30draw->draw.flush = nv30_draw_flush; + nv30draw->draw.reset_stipple_counter = nv30_draw_reset_stipple_counter; + nv30draw->draw.destroy = nv30_draw_destroy; + + return &nv30draw->draw; +} + diff --git a/src/mesa/pipe/nv30/nv30_fragprog.c b/src/mesa/pipe/nv30/nv30_fragprog.c new file mode 100644 index 00000000000..0233873d925 --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_fragprog.c @@ -0,0 +1,834 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "pipe/tgsi/util/tgsi_parse.h" +#include "pipe/tgsi/util/tgsi_util.h" + +#include "nv30_context.h" + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 1 +#define MASK_Y 2 +#define MASK_Z 4 +#define MASK_W 8 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE NV30_FP_OP_DST_SCALE_1X +#define DEF_CTEST NV30_FP_OP_COND_TR +#include "nv30_shader.h" + +#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv30_sr_neg((s)) +#define abs(s) nv30_sr_abs((s)) +#define scale(s,v) nv30_sr_scale((s), NV30_FP_OP_DST_SCALE_##v) + +#define MAX_CONSTS 128 +#define MAX_IMM 32 +struct nv30_fpc { + struct nv30_fragment_program *fp; + + uint attrib_map[PIPE_MAX_SHADER_INPUTS]; + + int high_temp; + int temp_temp_count; + int num_regs; + + uint depth_id; + uint colour_id; + + unsigned inst_offset; + + struct { + int pipe; + float vals[4]; + } consts[MAX_CONSTS]; + int nr_consts; + + struct nv30_sreg imm[MAX_IMM]; + unsigned nr_imm; +}; + +static INLINE struct nv30_sreg +temp(struct nv30_fpc *fpc) +{ + int idx; + + idx = fpc->temp_temp_count++; + idx += fpc->high_temp + 1; + return nv30_sr(NV30SR_TEMP, idx); +} + +static INLINE struct nv30_sreg +constant(struct nv30_fpc *fpc, int pipe, float vals[4]) +{ + int idx; + + if (fpc->nr_consts == MAX_CONSTS) + assert(0); + idx = fpc->nr_consts++; + + fpc->consts[idx].pipe = pipe; + if (pipe == -1) + memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float)); + return nv30_sr(NV30SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv30_fp_arith((cc), (s), NV30_FP_OP_OPCODE_##o, \ + (d), (m), (s0), (s1), (s2)) +#define tex(cc,s,o,u,d,m,s0,s1,s2) \ + nv30_fp_tex((cc), (s), NV30_FP_OP_OPCODE_##o, (u), \ + (d), (m), (s0), none, none) + +static void +grow_insns(struct nv30_fpc *fpc, int size) +{ + struct nv30_fragment_program *fp = fpc->fp; + + fp->insn_len += size; + fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len); +} + +static void +emit_src(struct nv30_fpc *fpc, int pos, struct nv30_sreg src) +{ + struct nv30_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; + uint32_t sr = 0; + + switch (src.type) { + case NV30SR_INPUT: + sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); + hw[0] |= (src.index << NV30_FP_OP_INPUT_SRC_SHIFT); + break; + case NV30SR_OUTPUT: + sr |= NV30_FP_REG_SRC_HALF; + /* fall-through */ + case NV30SR_TEMP: + sr |= (NV30_FP_REG_TYPE_TEMP << NV30_FP_REG_TYPE_SHIFT); + sr |= (src.index << NV30_FP_REG_SRC_SHIFT); + break; + case NV30SR_CONST: + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + if (fpc->consts[src.index].pipe >= 0) { + struct nv30_fragment_program_data *fpd; + + fp->consts = realloc(fp->consts, ++fp->nr_consts * + sizeof(*fpd)); + fpd = &fp->consts[fp->nr_consts - 1]; + fpd->offset = fpc->inst_offset + 4; + fpd->index = fpc->consts[src.index].pipe; + memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4); + } else { + memcpy(&fp->insn[fpc->inst_offset + 4], + fpc->consts[src.index].vals, + sizeof(uint32_t) * 4); + } + + sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT); + break; + case NV30SR_NONE: + sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV30_FP_REG_NEGATE; + + if (src.abs) + hw[1] |= (1 << (29 + pos)); + + sr |= ((src.swz[0] << NV30_FP_REG_SWZ_X_SHIFT) | + (src.swz[1] << NV30_FP_REG_SWZ_Y_SHIFT) | + (src.swz[2] << NV30_FP_REG_SWZ_Z_SHIFT) | + (src.swz[3] << NV30_FP_REG_SWZ_W_SHIFT)); + + hw[pos + 1] |= sr; +} + +static void +emit_dst(struct nv30_fpc *fpc, struct nv30_sreg dst) +{ + struct nv30_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; + + switch (dst.type) { + case NV30SR_TEMP: + if (fpc->num_regs < (dst.index + 1)) + fpc->num_regs = dst.index + 1; + break; + case NV30SR_OUTPUT: + if (dst.index == 1) { + fp->fp_control |= 0xe; + } else { + hw[0] |= NV30_FP_OP_OUT_REG_HALF; + } + break; + case NV30SR_NONE: + hw[0] |= (1 << 30); + break; + default: + assert(0); + } + + hw[0] |= (dst.index << NV30_FP_OP_OUT_REG_SHIFT); +} + +static void +nv30_fp_arith(struct nv30_fpc *fpc, int sat, int op, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2) +{ + struct nv30_fragment_program *fp = fpc->fp; + uint32_t *hw; + + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + memset(hw, 0, sizeof(uint32_t) * 4); + + if (op == NV30_FP_OP_OPCODE_KIL) + fp->fp_control |= NV34TCL_FP_CONTROL_USES_KIL; + hw[0] |= (op << NV30_FP_OP_OPCODE_SHIFT); + hw[0] |= (mask << NV30_FP_OP_OUTMASK_SHIFT); + hw[2] |= (dst.dst_scale << NV30_FP_OP_DST_SCALE_SHIFT); + + if (sat) + hw[0] |= NV30_FP_OP_OUT_SAT; + + if (dst.cc_update) + hw[0] |= NV30_FP_OP_COND_WRITE_ENABLE; + hw[1] |= (dst.cc_test << NV30_FP_OP_COND_SHIFT); + hw[1] |= ((dst.cc_swz[0] << NV30_FP_OP_COND_SWZ_X_SHIFT) | + (dst.cc_swz[1] << NV30_FP_OP_COND_SWZ_Y_SHIFT) | + (dst.cc_swz[2] << NV30_FP_OP_COND_SWZ_Z_SHIFT) | + (dst.cc_swz[3] << NV30_FP_OP_COND_SWZ_W_SHIFT)); + + emit_dst(fpc, dst); + emit_src(fpc, 0, s0); + emit_src(fpc, 1, s1); + emit_src(fpc, 2, s2); +} + +static void +nv30_fp_tex(struct nv30_fpc *fpc, int sat, int op, int unit, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2) +{ + struct nv30_fragment_program *fp = fpc->fp; + + nv30_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); + + fp->insn[fpc->inst_offset] |= (unit << NV30_FP_OP_TEX_UNIT_SHIFT); + fp->samplers |= (1 << unit); +} + +static INLINE struct nv30_sreg +tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc) +{ + struct nv30_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv30_sr(NV30SR_INPUT, + fpc->attrib_map[fsrc->SrcRegister.Index]); + break; + case TGSI_FILE_CONSTANT: + src = constant(fpc, fsrc->SrcRegister.Index, NULL); + break; + case TGSI_FILE_IMMEDIATE: + assert(fsrc->SrcRegister.Index < fpc->nr_imm); + src = fpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index + 1); + if (fpc->high_temp < src.index) + fpc->high_temp = src.index; + break; + /* This is clearly insane, but gallium hands us shaders like this. + * Luckily fragprog results are just temp regs.. + */ + case TGSI_FILE_OUTPUT: + if (fsrc->SrcRegister.Index == fpc->colour_id) + return nv30_sr(NV30SR_OUTPUT, 0); + else + return nv30_sr(NV30SR_OUTPUT, 1); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv30_sreg +tgsi_dst(struct nv30_fpc *fpc, const struct tgsi_full_dst_register *fdst) { + int idx; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + if (fdst->DstRegister.Index == fpc->colour_id) + return nv30_sr(NV30SR_OUTPUT, 0); + else + return nv30_sr(NV30SR_OUTPUT, 1); + break; + case TGSI_FILE_TEMPORARY: + idx = fdst->DstRegister.Index + 1; + if (fpc->high_temp < idx) + fpc->high_temp = idx; + return nv30_sr(NV30SR_TEMP, idx); + case TGSI_FILE_NULL: + return nv30_sr(NV30SR_NONE, 0); + default: + NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); + return nv30_sr(NV30SR_NONE, 0); + } +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +src_native_swz(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc, + struct nv30_sreg *src) +{ + const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + struct nv30_sreg tgsi = tgsi_src(fpc, fsrc); + uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; + uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, + fsrc->SrcRegisterExtSwz.NegateY, + fsrc->SrcRegisterExtSwz.NegateZ, + fsrc->SrcRegisterExtSwz.NegateW }; + uint c; + + for (c = 0; c < 4; c++) { + switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + mask |= (1 << c); + break; + case TGSI_EXTSWIZZLE_ZERO: + zero_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + case TGSI_EXTSWIZZLE_ONE: + one_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + default: + assert(0); + } + + if (!tgsi.negate && neg[c]) + neg_mask |= (1 << c); + } + + if (mask == MASK_ALL && !neg_mask) + return TRUE; + + *src = temp(fpc); + + if (mask) + arith(fpc, 0, MOV, *src, mask, tgsi, none, none); + + if (zero_mask) + arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); + + if (one_mask) + arith(fpc, 0, STR, *src, one_mask, *src, none, none); + + if (neg_mask) { + struct nv30_sreg one = temp(fpc); + arith(fpc, 0, STR, one, neg_mask, one, none, none); + arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none); + } + + return FALSE; +} + +static boolean +nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, + const struct tgsi_full_instruction *finst) +{ + const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + struct nv30_sreg src[3], dst, tmp; + int mask, sat, unit; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + fpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(fpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + case TGSI_FILE_CONSTANT: + case TGSI_FILE_TEMPORARY: + if (!src_native_swz(fpc, fsrc, &src[i])) + continue; + break; + default: + break; + } + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + NOUVEAU_MSG("extra src attr %d\n", + fsrc->SrcRegister.Index); + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + case TGSI_FILE_SAMPLER: + unit = fsrc->SrcRegister.Index; + break; + case TGSI_FILE_OUTPUT: + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_CMP: + tmp = temp(fpc); + arith(fpc, sat, MOV, dst, mask, src[2], none, none); + tmp.cc_update = 1; + arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); + dst.cc_test = NV30_VP_INST_COND_LT; + arith(fpc, sat, MOV, dst, mask, src[1], none, none); + break; + case TGSI_OPCODE_COS: + arith(fpc, sat, COS, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(fpc, sat, DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none); + arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X), + swz(src[1], W, W, W, W), none); + break; + case TGSI_OPCODE_DST: + arith(fpc, sat, DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(fpc, sat, EX2, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FLR: + arith(fpc, sat, FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(fpc, sat, FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_KIL: + arith(fpc, 0, KIL, none, 0, none, none, none); + break; + case TGSI_OPCODE_KILP: + dst = nv30_sr(NV30SR_NONE, 0); + dst.cc_update = 1; + arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); + dst.cc_update = 0; dst.cc_test = NV30_FP_OP_COND_LT; + arith(fpc, 0, KIL, dst, 0, none, none, none); + break; + case TGSI_OPCODE_LG2: + arith(fpc, sat, LG2, dst, mask, src[0], none, none); + break; +// case TGSI_OPCODE_LIT: + case TGSI_OPCODE_LRP: + tmp = temp(fpc); + arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); + arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); + break; + case TGSI_OPCODE_MAD: + arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(fpc, sat, MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(fpc, sat, MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(fpc, sat, MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(fpc); + arith(fpc, 0, LG2, tmp, MASK_X, + swz(src[0], X, X, X, X), none, none); + arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(fpc, sat, EX2, dst, mask, + swz(tmp, X, X, X, X), none, none); + break; + case TGSI_OPCODE_RCP: + arith(fpc, sat, RCP, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_RET: + assert(0); + break; + case TGSI_OPCODE_RFL: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); + arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); + arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, + swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); + arith(fpc, sat, MAD, dst, mask, + swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); + break; + case TGSI_OPCODE_RSQ: + tmp = temp(fpc); + arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, + abs(swz(src[0], X, X, X, X)), none, none); + arith(fpc, sat, EX2, dst, mask, + neg(swz(tmp, X, X, X, X)), none, none); + break; + case TGSI_OPCODE_SCS: + if (mask & MASK_X) { + arith(fpc, sat, COS, dst, MASK_X, + swz(src[0], X, X, X, X), none, none); + } + if (mask & MASK_Y) { + arith(fpc, sat, SIN, dst, MASK_Y, + swz(src[0], X, X, X, X), none, none); + } + break; + case TGSI_OPCODE_SIN: + arith(fpc, sat, SIN, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_SGE: + arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); + break; + case TGSI_OPCODE_TEX: + if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == + TGSI_EXTSWIZZLE_W) { + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + } else + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_TXB: + tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_XPD: + tmp = temp(fpc); + arith(fpc, 0, MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(fpc, sat, MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV30_FP_OP_INPUT_SRC_POSITION; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_FP_OP_INPUT_SRC_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_FP_OP_INPUT_SRC_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV30_FP_OP_INPUT_SRC_FOGC; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV30_FP_OP_INPUT_SRC_TC(fdec->Semantic. + SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad input semantic\n"); + return FALSE; + } + + fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + fpc->depth_id = fdec->u.DeclarationRange.First; + break; + case TGSI_SEMANTIC_COLOR: + fpc->colour_id = fdec->u.DeclarationRange.First; + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + return TRUE; +} + +void +nv30_fragprog_translate(struct nv30_context *nv30, + struct nv30_fragment_program *fp) +{ + struct tgsi_parse_context parse; + struct nv30_fpc *fpc = NULL; + + fpc = calloc(1, sizeof(struct nv30_fpc)); + if (!fpc) + return; + fpc->fp = fp; + fpc->high_temp = -1; + fpc->num_regs = 2; + + tgsi_parse_init(&parse, fp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_INPUT: + if (!nv30_fragprog_parse_decl_attrib(fpc, fdec)) + goto out_err; + break; + case TGSI_FILE_OUTPUT: + if (!nv30_fragprog_parse_decl_output(fpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + struct tgsi_full_immediate *imm; + float vals[4]; + int i; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); + assert(fpc->nr_imm < MAX_IMM); + + for (i = 0; i < imm->Immediate.Size; i++) + vals[i] = imm->u.ImmediateFloat32[i].Float; + fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + + finst = &parse.FullToken.FullInstruction; + if (!nv30_fragprog_parse_instruction(fpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + fp->fp_control |= (fpc->num_regs-1)/2; + fp->fp_reg_control = (1<<16)|0x4; + + /* Terminate final instruction */ + fp->insn[fpc->inst_offset] |= 0x00000001; + + /* Append NOP + END instruction, may or may not be necessary. */ + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + fp->insn[fpc->inst_offset + 0] = 0x00000001; + fp->insn[fpc->inst_offset + 1] = 0x00000000; + fp->insn[fpc->inst_offset + 2] = 0x00000000; + fp->insn[fpc->inst_offset + 3] = 0x00000000; + + fp->translated = TRUE; + fp->on_hw = FALSE; +out_err: + tgsi_parse_free(&parse); + free(fpc); +} + +void +nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) +{ + struct pipe_winsys *ws = nv30->pipe.winsys; + int i; + + if (!fp->translated) { + nv30_fragprog_translate(nv30, fp); + if (!fp->translated) + assert(0); + } + + if (fp->nr_consts) { + float *map = ws->buffer_map(ws, nv30->fragprog.constant_buf, + PIPE_BUFFER_USAGE_CPU_READ); + for (i = 0; i < fp->nr_consts; i++) { + struct nv30_fragment_program_data *fpd = &fp->consts[i]; + uint32_t *p = &fp->insn[fpd->offset]; + uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; + + if (!memcmp(p, cb, 4 * sizeof(float))) + continue; + memcpy(p, cb, 4 * sizeof(float)); + fp->on_hw = 0; + } + ws->buffer_unmap(ws, nv30->fragprog.constant_buf); + } + + if (!fp->on_hw) { + const uint32_t le = 1; + uint32_t *map; + + if (!fp->buffer) + fp->buffer = ws->buffer_create(ws, 0x100, 0, + fp->insn_len * 4); + map = ws->buffer_map(ws, fp->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); + +#if 0 + for (i = 0; i < fp->insn_len; i++) { + NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + } +#endif + + if ((*(const uint8_t *)&le)) { + for (i = 0; i < fp->insn_len; i++) { + map[i] = fp->insn[i]; + } + } else { + /* Weird swapping for big-endian chips */ + for (i = 0; i < fp->insn_len; i++) { + map[i] = ((fp->insn[i] & 0xffff) << 16) | + ((fp->insn[i] >> 16) & 0xffff); + } + } + + ws->buffer_unmap(ws, fp->buffer); + fp->on_hw = TRUE; + } + + BEGIN_RING(rankine, NV34TCL_FP_CONTROL, 1); + OUT_RING (fp->fp_control); + BEGIN_RING(rankine, NV34TCL_FP_REG_CONTROL, 1); + OUT_RING (fp->fp_reg_control); + + nv30->fragprog.active = fp; +} + +void +nv30_fragprog_destroy(struct nv30_context *nv30, + struct nv30_fragment_program *fp) +{ + if (fp->insn_len) + free(fp->insn); +} + diff --git a/src/mesa/pipe/nv30/nv30_fragtex.c b/src/mesa/pipe/nv30/nv30_fragtex.c new file mode 100644 index 00000000000..e75b1f7f28b --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_fragtex.c @@ -0,0 +1,160 @@ +#include "nv30_context.h" + +static inline int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV34TCL_TX_FORMAT_FORMAT_##tf, \ + (NV34TCL_TX_SWIZZLE_S0_X_##ts0x | NV34TCL_TX_SWIZZLE_S0_Y_##ts0y | \ + NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w | \ + NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y | \ + NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w), \ +} + +struct nv30_texture_format { + boolean defined; + uint pipe; + int format; + int swizzle; +}; + +static struct nv30_texture_format +nv30_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), +// _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), + _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), +// _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), +// _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), +// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + {}, +}; + +static struct nv30_texture_format * +nv30_fragtex_format(uint pipe_format) +{ + struct nv30_texture_format *tf = nv30_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv30_fragtex_build(struct nv30_context *nv30, int unit) +{ + struct nv30_sampler_state *ps = nv30->tex_sampler[unit]; + struct nv30_miptree *nv30mt = nv30->tex_miptree[unit]; + struct pipe_texture *pt = &nv30mt->base; + struct nv30_texture_format *tf; + uint32_t txf, txs, txp; + int swizzled = 0; /*XXX: implement in region code? */ + + tf = nv30_fragtex_format(pt->format); + if (!tf || !tf->defined) { + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); + return; + } + + txf = tf->format << 8; + txf |= (pt->last_level - pt->first_level + 1) << 16; + txf |= log2i(pt->width[0]) << 20; + txf |= log2i(pt->height[0]) << 24; + txf |= log2i(pt->depth[0]) << 28; + txf |= 8; + + switch (pt->target) { +/* case PIPE_TEXTURE_CUBE: + txf |= NV34TCL_TEX_FORMAT_CUBIC;*/ + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= (2<<4); + break; + case PIPE_TEXTURE_3D: + txf |= (3<<4); + break; + case PIPE_TEXTURE_1D: + txf |= (1<<4); + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + txs = tf->swizzle; + + BEGIN_RING(rankine, NV34TCL_TX_OFFSET(unit), 8); + OUT_RELOCl(nv30mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv30mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RING (ps->wrap); + OUT_RING (0x40000000); /* enable */ + OUT_RING (txs); + OUT_RING (ps->filt | 0x2000 /* magic */); + OUT_RING ((pt->width[0] << 16) | pt->height[0]); + OUT_RING (ps->bcol); +} + +void +nv30_fragtex_bind(struct nv30_context *nv30) +{ + struct nv30_fragment_program *fp = nv30->fragprog.active; + unsigned samplers, unit; + + samplers = nv30->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(rankine, NV34TCL_TX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv30->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv30_fragtex_build(nv30, unit); + } + + nv30->fp_samplers = fp->samplers; +} + diff --git a/src/mesa/pipe/nv30/nv30_miptree.c b/src/mesa/pipe/nv30/nv30_miptree.c new file mode 100644 index 00000000000..75e9b993c16 --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_miptree.c @@ -0,0 +1,105 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nv30_context.h" + +static void +nv30_miptree_layout(struct nv30_miptree *nv30mt) +{ + struct pipe_texture *pt = &nv30mt->base; + boolean swizzled = FALSE; + uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint offset = 0; + int nr_faces, l, f; + + if (pt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else + if (pt->target == PIPE_TEXTURE_3D) { + nr_faces = pt->depth[0]; + } else { + nr_faces = 1; + } + + for (l = pt->first_level; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->depth[l] = depth; + + if (swizzled) + nv30mt->level[l].pitch = pt->width[l] * pt->cpp; + else + nv30mt->level[l].pitch = pt->width[0] * pt->cpp; + nv30mt->level[l].pitch = (nv30mt->level[l].pitch + 63) & ~63; + + nv30mt->level[l].image_offset = + calloc(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + depth = MAX2(1, depth >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = pt->first_level; l <= pt->last_level; l++) { + nv30mt->level[l].image_offset[f] = offset; + offset += nv30mt->level[l].pitch * pt->height[l]; + } + } + + nv30mt->total_size = offset; +} + +static void +nv30_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv30_miptree *nv30mt; + + nv30mt = realloc(*pt, sizeof(struct nv30_miptree)); + if (!nv30mt) + return; + *pt = NULL; + + nv30_miptree_layout(nv30mt); + + nv30mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + nv30mt->total_size); + if (!nv30mt->buffer) { + free(nv30mt); + return; + } + + *pt = &nv30mt->base; +} + +static void +nv30_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv30_miptree *nv30mt = (struct nv30_miptree *)mt; + int l; + + pipe_buffer_reference(ws, &nv30mt->buffer, NULL); + for (l = mt->first_level; l <= mt->last_level; l++) { + if (nv30mt->level[l].image_offset) + free(nv30mt->level[l].image_offset); + } + free(nv30mt); + } +} + +void +nv30_init_miptree_functions(struct nv30_context *nv30) +{ + nv30->pipe.texture_create = nv30_miptree_create; + nv30->pipe.texture_release = nv30_miptree_release; +} + diff --git a/src/mesa/pipe/nv30/nv30_query.c b/src/mesa/pipe/nv30/nv30_query.c new file mode 100644 index 00000000000..ea74c0f5f1f --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_query.c @@ -0,0 +1,112 @@ +#include "pipe/p_context.h" + +#include "nv30_context.h" + +struct nv30_query { + struct nouveau_resource *object; + unsigned type; + boolean ready; + uint64_t result; +}; + +static inline struct nv30_query * +nv30_query(struct pipe_query *pipe) +{ + return (struct nv30_query *)pipe; +} + +static struct pipe_query * +nv30_query_create(struct pipe_context *pipe, unsigned query_type) +{ + struct nv30_query *q; + + q = calloc(1, sizeof(struct nv30_query)); + q->type = query_type; + + return (struct pipe_query *)q; +} + +static void +nv30_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + + if (q->object) + nv30->nvws->res_free(&q->object); + free(q); +} + +static void +nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + + assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (nv30->nvws->res_alloc(nv30->query_heap, 1, NULL, &q->object)) + assert(0); + nv30->nvws->notifier_reset(nv30->query, q->object->start); + + BEGIN_RING(rankine, NV34TCL_QUERY_RESET, 1); + OUT_RING (1); + BEGIN_RING(rankine, NV34TCL_QUERY_UNK17CC, 1); + OUT_RING (1); + + q->ready = FALSE; +} + +static void +nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + + BEGIN_RING(rankine, NV34TCL_QUERY_GET, 1); + OUT_RING ((0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) | + ((q->object->start * 32) << NV34TCL_QUERY_GET_OFFSET_SHIFT)); + FIRE_RING(); +} + +static boolean +nv30_query_result(struct pipe_context *pipe, struct pipe_query *pq, + boolean wait, uint64 *result) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + struct nouveau_winsys *nvws = nv30->nvws; + + assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (!q->ready) { + unsigned status; + + status = nvws->notifier_status(nv30->query, q->object->start); + if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { + if (wait == FALSE) + return FALSE; + nvws->notifier_wait(nv30->query, q->object->start, + NV_NOTIFY_STATE_STATUS_COMPLETED, + 0); + } + + q->result = nvws->notifier_retval(nv30->query, + q->object->start); + q->ready = TRUE; + nvws->res_free(&q->object); + } + + *result = q->result; + return TRUE; +} + +void +nv30_init_query_functions(struct nv30_context *nv30) +{ + nv30->pipe.create_query = nv30_query_create; + nv30->pipe.destroy_query = nv30_query_destroy; + nv30->pipe.begin_query = nv30_query_begin; + nv30->pipe.end_query = nv30_query_end; + nv30->pipe.get_query_result = nv30_query_result; +} diff --git a/src/mesa/pipe/nv30/nv30_shader.h b/src/mesa/pipe/nv30/nv30_shader.h new file mode 100644 index 00000000000..dd3a36f78f3 --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_shader.h @@ -0,0 +1,490 @@ +#ifndef __NV30_SHADER_H__ +#define __NV30_SHADER_H__ + +/* Vertex programs instruction set + * + * 128bit opcodes, split into 4 32-bit ones for ease of use. + * + * Non-native instructions + * ABS - MOV + NV40_VP_INST0_DEST_ABS + * POW - EX2 + MUL + LG2 + * SUB - ADD, second source negated + * SWZ - MOV + * XPD - + * + * Register access + * - Only one INPUT can be accessed per-instruction (move extras into TEMPs) + * - Only one CONST can be accessed per-instruction (move extras into TEMPs) + * + * Relative Addressing + * According to the value returned for + * MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB + * + * there are only two address registers available. The destination in the + * ARL instruction is set to TEMP (The temp isn't actually written). + * + * When using vanilla ARB_v_p, the proprietary driver will squish both the + * available ADDRESS regs into the first hardware reg in the X and Y + * components. + * + * To use an address reg as an index into consts, the CONST_SRC is set to + * (const_base + offset) and INDEX_CONST is set. + * + * To access the second address reg use ADDR_REG_SELECT_1. A particular + * component of the address regs is selected with ADDR_SWZ. + * + * Only one address register can be accessed per instruction. + * + * Conditional execution (see NV_vertex_program{2,3} for details) Conditional + * execution of an instruction is enabled by setting COND_TEST_ENABLE, and + * selecting the condition which will allow the test to pass with + * COND_{FL,LT,...}. It is possible to swizzle the values in the condition + * register, which allows for testing against an individual component. + * + * Branching: + * + * The BRA/CAL instructions seem to follow a slightly different opcode + * layout. The destination instruction ID (IADDR) overlaps a source field. + * Instruction ID's seem to be numbered based on the UPLOAD_FROM_ID FIFO + * command, and is incremented automatically on each UPLOAD_INST FIFO + * command. + * + * Conditional branching is achieved by using the condition tests described + * above. There doesn't appear to be dedicated looping instructions, but + * this can be done using a temp reg + conditional branching. + * + * Subroutines may be uploaded before the main program itself, but the first + * executed instruction is determined by the PROGRAM_START_ID FIFO command. + * + */ + +/* DWORD 0 */ + +#define NV30_VP_INST_ADDR_REG_SELECT_1 (1 << 24) +#define NV30_VP_INST_SRC2_ABS (1 << 23) /* guess */ +#define NV30_VP_INST_SRC1_ABS (1 << 22) /* guess */ +#define NV30_VP_INST_SRC0_ABS (1 << 21) /* guess */ +#define NV30_VP_INST_VEC_RESULT (1 << 20) +#define NV30_VP_INST_DEST_TEMP_ID_SHIFT 16 +#define NV30_VP_INST_DEST_TEMP_ID_MASK (0x0F << 16) +#define NV30_VP_INST_COND_UPDATE_ENABLE (1<<15) +#define NV30_VP_INST_VEC_DEST_TEMP_MASK (0xF << 16) +#define NV30_VP_INST_COND_TEST_ENABLE (1<<14) +#define NV30_VP_INST_COND_SHIFT 11 +#define NV30_VP_INST_COND_MASK (0x07 << 11) +# define NV30_VP_INST_COND_FL 0 /* guess */ +# define NV30_VP_INST_COND_LT 1 +# define NV30_VP_INST_COND_EQ 2 +# define NV30_VP_INST_COND_LE 3 +# define NV30_VP_INST_COND_GT 4 +# define NV30_VP_INST_COND_NE 5 +# define NV30_VP_INST_COND_GE 6 +# define NV30_VP_INST_COND_TR 7 /* guess */ +#define NV30_VP_INST_COND_SWZ_X_SHIFT 9 +#define NV30_VP_INST_COND_SWZ_X_MASK (0x03 << 9) +#define NV30_VP_INST_COND_SWZ_Y_SHIFT 7 +#define NV30_VP_INST_COND_SWZ_Y_MASK (0x03 << 7) +#define NV30_VP_INST_COND_SWZ_Z_SHIFT 5 +#define NV30_VP_INST_COND_SWZ_Z_MASK (0x03 << 5) +#define NV30_VP_INST_COND_SWZ_W_SHIFT 3 +#define NV30_VP_INST_COND_SWZ_W_MASK (0x03 << 3) +#define NV30_VP_INST_COND_SWZ_ALL_SHIFT 3 +#define NV30_VP_INST_COND_SWZ_ALL_MASK (0xFF << 3) +#define NV30_VP_INST_ADDR_SWZ_SHIFT 1 +#define NV30_VP_INST_ADDR_SWZ_MASK (0x03 << 1) +#define NV30_VP_INST_SCA_OPCODEH_SHIFT 0 +#define NV30_VP_INST_SCA_OPCODEH_MASK (0x01 << 0) + +/* DWORD 1 */ +#define NV30_VP_INST_SCA_OPCODEL_SHIFT 28 +#define NV30_VP_INST_SCA_OPCODEL_MASK (0x0F << 28) +# define NV30_VP_INST_OP_NOP 0x00 +# define NV30_VP_INST_OP_RCP 0x02 +# define NV30_VP_INST_OP_RCC 0x03 +# define NV30_VP_INST_OP_RSQ 0x04 +# define NV30_VP_INST_OP_EXP 0x05 +# define NV30_VP_INST_OP_LOG 0x06 +# define NV30_VP_INST_OP_LIT 0x07 +# define NV30_VP_INST_OP_BRA 0x09 +# define NV30_VP_INST_OP_CAL 0x0B +# define NV30_VP_INST_OP_RET 0x0C +# define NV30_VP_INST_OP_LG2 0x0D +# define NV30_VP_INST_OP_EX2 0x0E +# define NV30_VP_INST_OP_SIN 0x0F +# define NV30_VP_INST_OP_COS 0x10 +#define NV30_VP_INST_VEC_OPCODE_SHIFT 23 +#define NV30_VP_INST_VEC_OPCODE_MASK (0x1F << 23) +# define NV30_VP_INST_OP_NOPV 0x00 +# define NV30_VP_INST_OP_MOV 0x01 +# define NV30_VP_INST_OP_MUL 0x02 +# define NV30_VP_INST_OP_ADD 0x03 +# define NV30_VP_INST_OP_MAD 0x04 +# define NV30_VP_INST_OP_DP3 0x05 +# define NV30_VP_INST_OP_DP4 0x07 +# define NV30_VP_INST_OP_DPH 0x06 +# define NV30_VP_INST_OP_DST 0x08 +# define NV30_VP_INST_OP_MIN 0x09 +# define NV30_VP_INST_OP_MAX 0x0A +# define NV30_VP_INST_OP_SLT 0x0B +# define NV30_VP_INST_OP_SGE 0x0C +# define NV30_VP_INST_OP_ARL 0x0D +# define NV30_VP_INST_OP_FRC 0x0E +# define NV30_VP_INST_OP_FLR 0x0F +# define NV30_VP_INST_OP_SEQ 0x10 +# define NV30_VP_INST_OP_SFL 0x11 +# define NV30_VP_INST_OP_SGT 0x12 +# define NV30_VP_INST_OP_SLE 0x13 +# define NV30_VP_INST_OP_SNE 0x14 +# define NV30_VP_INST_OP_STR 0x15 +# define NV30_VP_INST_OP_SSG 0x16 +# define NV30_VP_INST_OP_ARR 0x17 +# define NV30_VP_INST_OP_ARA 0x18 +#define NV30_VP_INST_CONST_SRC_SHIFT 14 +#define NV30_VP_INST_CONST_SRC_MASK (0xFF << 14) +#define NV30_VP_INST_INPUT_SRC_SHIFT 9 /*NV20*/ +#define NV30_VP_INST_INPUT_SRC_MASK (0x0F << 9) /*NV20*/ +# define NV30_VP_INST_IN_POS 0 /* These seem to match the bindings specified in */ +# define NV30_VP_INST_IN_WEIGHT 1 /* the ARB_v_p spec (2.14.3.1) */ +# define NV30_VP_INST_IN_NORMAL 2 +# define NV30_VP_INST_IN_COL0 3 /* Should probably confirm them all though */ +# define NV30_VP_INST_IN_COL1 4 +# define NV30_VP_INST_IN_FOGC 5 +# define NV30_VP_INST_IN_TC0 8 +# define NV30_VP_INST_IN_TC(n) (8+n) +#define NV30_VP_INST_SRC0H_SHIFT 0 /*NV20*/ +#define NV30_VP_INST_SRC0H_MASK (0x1FF << 0) /*NV20*/ + +/* Please note: the IADDR fields overlap other fields because they are used + * only for branch instructions. See Branching: label above + * + * DWORD 2 + */ +#define NV30_VP_INST_SRC0L_SHIFT 26 /*NV20*/ +#define NV30_VP_INST_SRC0L_MASK (0x3F <<26) /* NV30_VP_SRC0_LOW_MASK << 26 */ +#define NV30_VP_INST_SRC1_SHIFT 11 /*NV20*/ +#define NV30_VP_INST_SRC1_MASK (0x7FFF<<11) /*NV20*/ +#define NV30_VP_INST_SRC2H_SHIFT 0 /*NV20*/ +#define NV30_VP_INST_SRC2H_MASK (0x7FF << 0) /* NV30_VP_SRC2_HIGH_MASK >> 4*/ +#define NV30_VP_INST_IADDR_SHIFT 2 +#define NV30_VP_INST_IADDR_MASK (0xF << 28) /* NV30_VP_SRC2_LOW_MASK << 28 */ + +/* DWORD 3 */ +#define NV30_VP_INST_SRC2L_SHIFT 28 /*NV20*/ +#define NV30_VP_INST_SRC2L_MASK (0x0F <<28) /*NV20*/ +#define NV30_VP_INST_STEMP_WRITEMASK_SHIFT 24 +#define NV30_VP_INST_STEMP_WRITEMASK_MASK (0x0F << 24) +#define NV30_VP_INST_VTEMP_WRITEMASK_SHIFT 20 +#define NV30_VP_INST_VTEMP_WRITEMASK_MASK (0x0F << 20) +#define NV30_VP_INST_SDEST_WRITEMASK_SHIFT 16 +#define NV30_VP_INST_SDEST_WRITEMASK_MASK (0x0F << 16) +#define NV30_VP_INST_VDEST_WRITEMASK_SHIFT 12 /*NV20*/ +#define NV30_VP_INST_VDEST_WRITEMASK_MASK (0x0F << 12) /*NV20*/ +#define NV30_VP_INST_DEST_SHIFT 2 +#define NV30_VP_INST_DEST_MASK (0x0F << 2) +# define NV30_VP_INST_DEST_POS 0 +# define NV30_VP_INST_DEST_BFC0 1 +# define NV30_VP_INST_DEST_BFC1 2 +# define NV30_VP_INST_DEST_COL0 3 +# define NV30_VP_INST_DEST_COL1 4 +# define NV30_VP_INST_DEST_FOGC 5 +# define NV30_VP_INST_DEST_PSZ 6 +# define NV30_VP_INST_DEST_TC(n) (8+n) + +#define NV30_VP_INST_LAST (1 << 0) + +/* Useful to split the source selection regs into their pieces */ +#define NV30_VP_SRC0_HIGH_SHIFT 6 +#define NV30_VP_SRC0_HIGH_MASK 0x00007FC0 +#define NV30_VP_SRC0_LOW_MASK 0x0000003F +#define NV30_VP_SRC2_HIGH_SHIFT 4 +#define NV30_VP_SRC2_HIGH_MASK 0x00007FF0 +#define NV30_VP_SRC2_LOW_MASK 0x0000000F + + +/* Source-register definition - matches NV20 exactly */ +#define NV30_VP_SRC_NEGATE (1<<14) +#define NV30_VP_SRC_SWZ_X_SHIFT 12 +#define NV30_VP_SRC_REG_SWZ_X_MASK (0x03 <<12) +#define NV30_VP_SRC_SWZ_Y_SHIFT 10 +#define NV30_VP_SRC_REG_SWZ_Y_MASK (0x03 <<10) +#define NV30_VP_SRC_SWZ_Z_SHIFT 8 +#define NV30_VP_SRC_REG_SWZ_Z_MASK (0x03 << 8) +#define NV30_VP_SRC_SWZ_W_SHIFT 6 +#define NV30_VP_SRC_REG_SWZ_W_MASK (0x03 << 6) +#define NV30_VP_SRC_REG_SWZ_ALL_SHIFT 6 +#define NV30_VP_SRC_REG_SWZ_ALL_MASK (0xFF << 6) +#define NV30_VP_SRC_TEMP_SRC_SHIFT 2 +#define NV30_VP_SRC_REG_TEMP_ID_MASK (0x0F << 0) +#define NV30_VP_SRC_REG_TYPE_SHIFT 0 +#define NV30_VP_SRC_REG_TYPE_MASK (0x03 << 0) +#define NV30_VP_SRC_REG_TYPE_TEMP 1 +#define NV30_VP_SRC_REG_TYPE_INPUT 2 +#define NV30_VP_SRC_REG_TYPE_CONST 3 /* guess */ + +/* + * Each fragment program opcode appears to be comprised of 4 32-bit values. + * + * 0 - Opcode, output reg/mask, ATTRIB source + * 1 - Source 0 + * 2 - Source 1 + * 3 - Source 2 + * + * There appears to be no special difference between result regs and temp regs. + * result.color == R0.xyzw + * result.depth == R1.z + * When the fragprog contains instructions to write depth, NV30_TCL_PRIMITIVE_3D_UNK1D78=0 + * otherwise it is set to 1. + * + * Constants are inserted directly after the instruction that uses them. + * + * It appears that it's not possible to use two input registers in one + * instruction as the input sourcing is done in the instruction dword + * and not the source selection dwords. As such instructions such as: + * + * ADD result.color, fragment.color, fragment.texcoord[0]; + * + * must be split into two MOV's and then an ADD (nvidia does this) but + * I'm not sure why it's not just one MOV and then source the second input + * in the ADD instruction.. + * + * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary + * negation requires multiplication with a const. + * + * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO/SWIZZLE_ONE + * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as SWIZZLE_ZERO + * is implemented simply by not writing to the relevant components of the destination. + * + * Conditional execution + * TODO + * + * Non-native instructions: + * LIT + * LRP - MAD+MAD + * SUB - ADD, negate second source + * RSQ - LG2 + EX2 + * POW - LG2 + MUL + EX2 + * SCS - COS + SIN + * XPD + */ + +//== Opcode / Destination selection == +#define NV30_FP_OP_PROGRAM_END (1 << 0) +#define NV30_FP_OP_OUT_REG_SHIFT 1 +#define NV30_FP_OP_OUT_REG_MASK (31 << 1) /* uncertain */ +/* Needs to be set when writing outputs to get expected result.. */ +#define NV30_FP_OP_OUT_REG_HALF (1 << 7) +#define NV30_FP_OP_COND_WRITE_ENABLE (1 << 8) +#define NV30_FP_OP_OUTMASK_SHIFT 9 +#define NV30_FP_OP_OUTMASK_MASK (0xF << 9) +# define NV30_FP_OP_OUT_X (1<<9) +# define NV30_FP_OP_OUT_Y (1<<10) +# define NV30_FP_OP_OUT_Z (1<<11) +# define NV30_FP_OP_OUT_W (1<<12) +/* Uncertain about these, especially the input_src values.. it's possible that + * they can be dynamically changed. + */ +#define NV30_FP_OP_INPUT_SRC_SHIFT 13 +#define NV30_FP_OP_INPUT_SRC_MASK (15 << 13) +# define NV30_FP_OP_INPUT_SRC_POSITION 0x0 +# define NV30_FP_OP_INPUT_SRC_COL0 0x1 +# define NV30_FP_OP_INPUT_SRC_COL1 0x2 +# define NV30_FP_OP_INPUT_SRC_FOGC 0x3 +# define NV30_FP_OP_INPUT_SRC_TC0 0x4 +# define NV30_FP_OP_INPUT_SRC_TC(n) (0x4 + n) +#define NV30_FP_OP_TEX_UNIT_SHIFT 17 +#define NV30_FP_OP_TEX_UNIT_MASK (0xF << 17) /* guess */ +#define NV30_FP_OP_PRECISION_SHIFT 22 +#define NV30_FP_OP_PRECISION_MASK (3 << 22) +# define NV30_FP_PRECISION_FP32 0 +# define NV30_FP_PRECISION_FP16 1 +# define NV30_FP_PRECISION_FX12 2 +#define NV30_FP_OP_OPCODE_SHIFT 24 +#define NV30_FP_OP_OPCODE_MASK (0x3F << 24) +# define NV30_FP_OP_OPCODE_NOP 0x00 +# define NV30_FP_OP_OPCODE_MOV 0x01 +# define NV30_FP_OP_OPCODE_MUL 0x02 +# define NV30_FP_OP_OPCODE_ADD 0x03 +# define NV30_FP_OP_OPCODE_MAD 0x04 +# define NV30_FP_OP_OPCODE_DP3 0x05 +# define NV30_FP_OP_OPCODE_DP4 0x06 +# define NV30_FP_OP_OPCODE_DST 0x07 +# define NV30_FP_OP_OPCODE_MIN 0x08 +# define NV30_FP_OP_OPCODE_MAX 0x09 +# define NV30_FP_OP_OPCODE_SLT 0x0A +# define NV30_FP_OP_OPCODE_SGE 0x0B +# define NV30_FP_OP_OPCODE_SLE 0x0C +# define NV30_FP_OP_OPCODE_SGT 0x0D +# define NV30_FP_OP_OPCODE_SNE 0x0E +# define NV30_FP_OP_OPCODE_SEQ 0x0F +# define NV30_FP_OP_OPCODE_FRC 0x10 +# define NV30_FP_OP_OPCODE_FLR 0x11 +# define NV30_FP_OP_OPCODE_KIL 0x12 +# define NV30_FP_OP_OPCODE_PK4B 0x13 +# define NV30_FP_OP_OPCODE_UP4B 0x14 +# define NV30_FP_OP_OPCODE_DDX 0x15 /* can only write XY */ +# define NV30_FP_OP_OPCODE_DDY 0x16 /* can only write XY */ +# define NV30_FP_OP_OPCODE_TEX 0x17 +# define NV30_FP_OP_OPCODE_TXP 0x18 +# define NV30_FP_OP_OPCODE_TXD 0x19 +# define NV30_FP_OP_OPCODE_RCP 0x1A +# define NV30_FP_OP_OPCODE_RSQ 0x1B +# define NV30_FP_OP_OPCODE_EX2 0x1C +# define NV30_FP_OP_OPCODE_LG2 0x1D +# define NV30_FP_OP_OPCODE_LIT 0x1E +# define NV30_FP_OP_OPCODE_LRP 0x1F +# define NV30_FP_OP_OPCODE_STR 0x20 +# define NV30_FP_OP_OPCODE_SFL 0x21 +# define NV30_FP_OP_OPCODE_COS 0x22 +# define NV30_FP_OP_OPCODE_SIN 0x23 +# define NV30_FP_OP_OPCODE_PK2H 0x24 +# define NV30_FP_OP_OPCODE_UP2H 0x25 +# define NV30_FP_OP_OPCODE_POW 0x26 +# define NV30_FP_OP_OPCODE_PK4UB 0x27 +# define NV30_FP_OP_OPCODE_UP4UB 0x28 +# define NV30_FP_OP_OPCODE_PK2US 0x29 +# define NV30_FP_OP_OPCODE_UP2US 0x2A +# define NV30_FP_OP_OPCODE_DP2A 0x2E +# define NV30_FP_OP_OPCODE_TXB 0x31 +# define NV30_FP_OP_OPCODE_RFL 0x36 +# define NV30_FP_OP_OPCODE_DIV 0x3A +#define NV30_FP_OP_OUT_SAT (1 << 31) + +/* high order bits of SRC0 */ +#define NV30_FP_OP_OUT_ABS (1 << 29) +#define NV30_FP_OP_COND_SWZ_W_SHIFT 27 +#define NV30_FP_OP_COND_SWZ_W_MASK (3 << 27) +#define NV30_FP_OP_COND_SWZ_Z_SHIFT 25 +#define NV30_FP_OP_COND_SWZ_Z_MASK (3 << 25) +#define NV30_FP_OP_COND_SWZ_Y_SHIFT 23 +#define NV30_FP_OP_COND_SWZ_Y_MASK (3 << 23) +#define NV30_FP_OP_COND_SWZ_X_SHIFT 21 +#define NV30_FP_OP_COND_SWZ_X_MASK (3 << 21) +#define NV30_FP_OP_COND_SWZ_ALL_SHIFT 21 +#define NV30_FP_OP_COND_SWZ_ALL_MASK (0xFF << 21) +#define NV30_FP_OP_COND_SHIFT 18 +#define NV30_FP_OP_COND_MASK (0x07 << 18) +# define NV30_FP_OP_COND_FL 0 +# define NV30_FP_OP_COND_LT 1 +# define NV30_FP_OP_COND_EQ 2 +# define NV30_FP_OP_COND_LE 3 +# define NV30_FP_OP_COND_GT 4 +# define NV30_FP_OP_COND_NE 5 +# define NV30_FP_OP_COND_GE 6 +# define NV30_FP_OP_COND_TR 7 + +/* high order bits of SRC1 */ +#define NV30_FP_OP_DST_SCALE_SHIFT 28 +#define NV30_FP_OP_DST_SCALE_MASK (3 << 28) +#define NV30_FP_OP_DST_SCALE_1X 0 +#define NV30_FP_OP_DST_SCALE_2X 1 +#define NV30_FP_OP_DST_SCALE_4X 2 +#define NV30_FP_OP_DST_SCALE_8X 3 +#define NV30_FP_OP_DST_SCALE_INV_2X 5 +#define NV30_FP_OP_DST_SCALE_INV_4X 6 +#define NV30_FP_OP_DST_SCALE_INV_8X 7 + + +/* high order bits of SRC2 */ +#define NV30_FP_OP_INDEX_INPUT (1 << 30) + +//== Register selection == +#define NV30_FP_REG_TYPE_SHIFT 0 +#define NV30_FP_REG_TYPE_MASK (3 << 0) +# define NV30_FP_REG_TYPE_TEMP 0 +# define NV30_FP_REG_TYPE_INPUT 1 +# define NV30_FP_REG_TYPE_CONST 2 +#define NV30_FP_REG_SRC_SHIFT 2 /* uncertain */ +#define NV30_FP_REG_SRC_MASK (31 << 2) +#define NV30_FP_REG_SRC_HALF (1 << 8) +#define NV30_FP_REG_SWZ_ALL_SHIFT 9 +#define NV30_FP_REG_SWZ_ALL_MASK (255 << 9) +#define NV30_FP_REG_SWZ_X_SHIFT 9 +#define NV30_FP_REG_SWZ_X_MASK (3 << 9) +#define NV30_FP_REG_SWZ_Y_SHIFT 11 +#define NV30_FP_REG_SWZ_Y_MASK (3 << 11) +#define NV30_FP_REG_SWZ_Z_SHIFT 13 +#define NV30_FP_REG_SWZ_Z_MASK (3 << 13) +#define NV30_FP_REG_SWZ_W_SHIFT 15 +#define NV30_FP_REG_SWZ_W_MASK (3 << 15) +# define NV30_FP_SWIZZLE_X 0 +# define NV30_FP_SWIZZLE_Y 1 +# define NV30_FP_SWIZZLE_Z 2 +# define NV30_FP_SWIZZLE_W 3 +#define NV30_FP_REG_NEGATE (1 << 17) + +#define NV30SR_NONE 0 +#define NV30SR_OUTPUT 1 +#define NV30SR_INPUT 2 +#define NV30SR_TEMP 3 +#define NV30SR_CONST 4 + +struct nv30_sreg { + int type; + int index; + + int dst_scale; + + int negate; + int abs; + int swz[4]; + + int cc_update; + int cc_update_reg; + int cc_test; + int cc_test_reg; + int cc_swz[4]; +}; + +static INLINE struct nv30_sreg +nv30_sr(int type, int index) +{ + struct nv30_sreg temp = { + .type = type, + .index = index, + .dst_scale = DEF_SCALE, + .abs = 0, + .negate = 0, + .swz = { 0, 1, 2, 3 }, + .cc_update = 0, + .cc_update_reg = 0, + .cc_test = DEF_CTEST, + .cc_test_reg = 0, + .cc_swz = { 0, 1, 2, 3 }, + }; + return temp; +} + +static INLINE struct nv30_sreg +nv30_sr_swz(struct nv30_sreg src, int x, int y, int z, int w) +{ + struct nv30_sreg dst = src; + + dst.swz[SWZ_X] = src.swz[x]; + dst.swz[SWZ_Y] = src.swz[y]; + dst.swz[SWZ_Z] = src.swz[z]; + dst.swz[SWZ_W] = src.swz[w]; + return dst; +} + +static INLINE struct nv30_sreg +nv30_sr_neg(struct nv30_sreg src) +{ + src.negate = !src.negate; + return src; +} + +static INLINE struct nv30_sreg +nv30_sr_abs(struct nv30_sreg src) +{ + src.abs = 1; + return src; +} + +static INLINE struct nv30_sreg +nv30_sr_scale(struct nv30_sreg src, int scale) +{ + src.dst_scale = scale; + return src; +} + +#endif diff --git a/src/mesa/pipe/nv30/nv30_state.c b/src/mesa/pipe/nv30/nv30_state.c new file mode 100644 index 00000000000..c29a6448093 --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_state.c @@ -0,0 +1,740 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +static void * +nv30_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv30_blend_state *cb; + + cb = malloc(sizeof(struct nv30_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | + (nvgl_blend_eqn(cso->rgb_func))); + + cb->l_enable = cso->logicop_enable ? 1 : 0; + cb->l_op = nvgl_logicop_func(cso->logicop_func); + + cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + + cb->d_enable = cso->dither ? 1 : 0; + + return (void *)cb; +} + +static void +nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_blend_state *cb = hwcso; + + BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); + OUT_RING (cb->d_enable); + + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 3); + OUT_RING (cb->b_enable); + OUT_RING (cb->b_srcfunc); + OUT_RING (cb->b_dstfunc); + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_EQUATION, 1); + OUT_RING (cb->b_eqn); + + BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); + OUT_RING (cb->c_mask); + + BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); + OUT_RING (cb->l_enable); + OUT_RING (cb->l_op); +} + +static void +nv30_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV34TCL_TX_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV34TCL_TX_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV34TCL_TX_WRAP_S_CLAMP; + break; +/* case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP; + break;*/ + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV34TCL_TX_WRAP_S_REPEAT; + break; + } + + return ret >> NV34TCL_TX_WRAP_S_SHIFT; +} + +static void * +nv30_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv30_sampler_state *ps; + uint32_t filter = 0; + + ps = malloc(sizeof(struct nv30_sampler_state)); + + ps->fmt = 0; + if (!cso->normalized_coords) + ps->fmt |= NV34TCL_TX_FORMAT_RECT; + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV34TCL_TX_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV34TCL_TX_WRAP_T_SHIFT) | + (wrap_mode(cso->wrap_r) << NV34TCL_TX_WRAP_R_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy >= 2.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + +/* if (cso->max_anisotropy >= 16.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_16X; + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_12X; + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_10X; + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_6X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_4X; + } else { + ps->en |= NV34TCL_TX_ENABLE_ANISO_2X; + }*/ + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV34TCL_TX_FILTER_MAGNIFY_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV34TCL_TX_FILTER_MAGNIFY_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST; + break; + } + break; + } + + ps->filt = filter; + +/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NEVER; + break; + case PIPE_FUNC_GREATER: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GREATER; + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_EQUAL; + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GEQUAL; + break; + case PIPE_FUNC_LESS: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LESS; + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NOTEQUAL; + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LEQUAL; + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_ALWAYS; + break; + default: + break; + } + }*/ + + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv30_sampler_state_bind(struct pipe_context *pipe, unsigned unit, + void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_sampler_state *ps = hwcso; + + nv30->tex_sampler[unit] = ps; + nv30->dirty_samplers |= (1 << unit); +} + +static void +nv30_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv30_set_sampler_texture(struct pipe_context *pipe, unsigned unit, + struct pipe_texture *miptree) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->tex_miptree[unit] = (struct nv30_miptree *)miptree; + nv30->dirty_samplers |= (1 << unit); +} + +static void * +nv30_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv30_rasterizer_state *rs; + int i; + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + rs = malloc(sizeof(struct nv30_rasterizer_state)); + + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; + + rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; + rs->line_smooth_en = cso->line_smooth ? 1 : 0; + rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; + rs->line_stipple = (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor; + + rs->point_size = *(uint32_t*)&cso->point_size; + + rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; + rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; + + if (cso->front_winding == PIPE_WINDING_CCW) { + rs->front_face = NV34TCL_FRONT_FACE_CCW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + } else { + rs->front_face = NV34TCL_FRONT_FACE_CW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + } + + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CCW) + rs->cull_face = NV34TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV34TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_CW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CW) + rs->cull_face = NV34TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV34TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_BOTH: + rs->cull_face_en = 1; + rs->cull_face = NV34TCL_CULL_FACE_FRONT_AND_BACK; + break; + case PIPE_WINDING_NONE: + default: + rs->cull_face_en = 0; + rs->cull_face = 0; + break; + } + + if (cso->point_sprite) { + rs->point_sprite = (1 << 0); + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + rs->point_sprite |= (1 << (8 + i)); + } + } else { + rs->point_sprite = 0; + } + + return (void *)rs; +} + +static void +nv30_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_rasterizer_state *rs = hwcso; + + BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); + OUT_RING (rs->shade_model); + + BEGIN_RING(rankine, NV34TCL_LINE_WIDTH, 2); + OUT_RING (rs->line_width); + OUT_RING (rs->line_smooth_en); + BEGIN_RING(rankine, NV34TCL_LINE_STIPPLE_ENABLE, 2); + OUT_RING (rs->line_stipple_en); + OUT_RING (rs->line_stipple); + + BEGIN_RING(rankine, NV34TCL_POINT_SIZE, 1); + OUT_RING (rs->point_size); + + BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 6); + OUT_RING (rs->poly_mode_front); + OUT_RING (rs->poly_mode_back); + OUT_RING (rs->cull_face); + OUT_RING (rs->front_face); + OUT_RING (rs->poly_smooth_en); + OUT_RING (rs->cull_face_en); + + BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); + OUT_RING (rs->poly_stipple_en); + + BEGIN_RING(rankine, NV34TCL_POINT_SPRITE, 1); + OUT_RING (rs->point_sprite); +} + +static void +nv30_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv30_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, + unsigned idx, struct nv30_stencil_push *hw) +{ + hw->enable = cso->stencil[idx].enabled ? 1 : 0; + hw->wmask = cso->stencil[idx].write_mask; + hw->func = nvgl_comparison_op(cso->stencil[idx].func); + hw->ref = cso->stencil[idx].ref_value; + hw->vmask = cso->stencil[idx].value_mask; + hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); + hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); + hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); +} + +static void * +nv30_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv30_depth_stencil_alpha_state *hw; + + hw = malloc(sizeof(struct nv30_depth_stencil_alpha_state)); + + hw->depth.func = nvgl_comparison_op(cso->depth.func); + hw->depth.write_enable = cso->depth.writemask ? 1 : 0; + hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + + nv30_translate_stencil(cso, 0, &hw->stencil.front); + nv30_translate_stencil(cso, 1, &hw->stencil.back); + + hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; + hw->alpha.func = nvgl_comparison_op(cso->alpha.func); + hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + + return (void *)hw; +} + +static void +nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_depth_stencil_alpha_state *hw = hwcso; + + BEGIN_RING(rankine, NV34TCL_DEPTH_FUNC, 3); + OUT_RINGp ((uint32_t *)&hw->depth, 3); + BEGIN_RING(rankine, NV34TCL_STENCIL_FRONT_ENABLE, 16); + OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); + OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); + BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3); + OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); +} + +static void +nv30_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv30_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv30_vertex_program *vp; + + vp = calloc(1, sizeof(struct nv30_vertex_program)); + vp->pipe = cso; + + return (void *)vp; +} + +static void +nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_vertex_program *vp = hwcso; + + nv30->vertprog.current = vp; + nv30->dirty |= NV30_NEW_VERTPROG; +} + +static void +nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_vertex_program *vp = hwcso; + + nv30_vertprog_destroy(nv30, vp); + free(vp); +} + +static void * +nv30_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv30_fragment_program *fp; + + fp = calloc(1, sizeof(struct nv30_fragment_program)); + fp->pipe = cso; + + return (void *)fp; +} + +static void +nv30_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_fragment_program *fp = hwcso; + + nv30->fragprog.current = fp; + nv30->dirty |= NV30_NEW_FRAGPROG; +} + +static void +nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_fragment_program *fp = hwcso; + + nv30_fragprog_destroy(nv30, fp); + free(fp); +} + +static void +nv30_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_COLOR, 1); + OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0)); +} + +static void +nv30_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv30_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + if (shader == PIPE_SHADER_VERTEX) { + nv30->vertprog.constant_buf = buf->buffer; + nv30->dirty |= NV30_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv30->fragprog.constant_buf = buf->buffer; + nv30->dirty |= NV30_NEW_FRAGPROG; + } +} + +static void +nv30_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct pipe_surface *rt[4], *zeta; + uint32_t rt_enable, rt_format, w, h; + int i, colour_format = 0, zeta_format = 0; + + rt_enable = 0; + for (i = 0; i < 4; i++) { + if (!fb->cbufs[i]) + continue; + + if (colour_format) { + assert(w == fb->cbufs[i]->width); + assert(h == fb->cbufs[i]->height); + assert(colour_format == fb->cbufs[i]->format); + } else { + w = fb->cbufs[i]->width; + h = fb->cbufs[i]->height; + colour_format = fb->cbufs[i]->format; + rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i); + rt[i] = fb->cbufs[i]; + } + } + + if (rt_enable & (NV34TCL_RT_ENABLE_COLOR1 | NV34TCL_RT_ENABLE_COLOR2 | + NV34TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV34TCL_RT_ENABLE_MRT; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV34TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + switch (zeta_format) { + case PIPE_FORMAT_Z16_UNORM: + rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16; + break; + case PIPE_FORMAT_Z24S8_UNORM: + case 0: + rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8; + break; + default: + assert(0); + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { + BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); + OUT_RING (rt[0]->pitch * rt[0]->cpp); + nv30->rt[0] = rt[0]->buffer; + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { + BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 2); + OUT_RING (rt[1]->pitch * rt[1]->cpp); + nv30->rt[1] = rt[1]->buffer; + } + + if (zeta_format) { + BEGIN_RING(rankine, NV34TCL_ZETA_PITCH, 1); + OUT_RING (zeta->pitch * zeta->cpp); + nv30->zeta = zeta->buffer; + } + + nv30->rt_enable = rt_enable; + BEGIN_RING(rankine, NV34TCL_RT_ENABLE, 1); + OUT_RING (rt_enable); + BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 3); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((w - 1) << 16) | 0); + OUT_RING (((h - 1) << 16) | 0); +} + +static void +nv30_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32); + OUT_RINGp ((uint32_t *)stipple->stipple, 32); +} + +static void +nv30_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny); +} + +static void +nv30_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); + OUT_RINGf (vpt->translate[0]); + OUT_RINGf (vpt->translate[1]); + OUT_RINGf (vpt->translate[2]); + OUT_RINGf (vpt->translate[3]); + OUT_RINGf (vpt->scale[0]); + OUT_RINGf (vpt->scale[1]); + OUT_RINGf (vpt->scale[2]); + OUT_RINGf (vpt->scale[3]); +} + +static void +nv30_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->vtxbuf[index] = *vb; + + nv30->dirty |= NV30_NEW_ARRAYS; +} + +static void +nv30_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->vtxelt[index] = *ve; + + nv30->dirty |= NV30_NEW_ARRAYS; +} + +void +nv30_init_state_functions(struct nv30_context *nv30) +{ + nv30->pipe.create_blend_state = nv30_blend_state_create; + nv30->pipe.bind_blend_state = nv30_blend_state_bind; + nv30->pipe.delete_blend_state = nv30_blend_state_delete; + + nv30->pipe.create_sampler_state = nv30_sampler_state_create; + nv30->pipe.bind_sampler_state = nv30_sampler_state_bind; + nv30->pipe.delete_sampler_state = nv30_sampler_state_delete; + nv30->pipe.set_sampler_texture = nv30_set_sampler_texture; + + nv30->pipe.create_rasterizer_state = nv30_rasterizer_state_create; + nv30->pipe.bind_rasterizer_state = nv30_rasterizer_state_bind; + nv30->pipe.delete_rasterizer_state = nv30_rasterizer_state_delete; + + nv30->pipe.create_depth_stencil_alpha_state = + nv30_depth_stencil_alpha_state_create; + nv30->pipe.bind_depth_stencil_alpha_state = + nv30_depth_stencil_alpha_state_bind; + nv30->pipe.delete_depth_stencil_alpha_state = + nv30_depth_stencil_alpha_state_delete; + + nv30->pipe.create_vs_state = nv30_vp_state_create; + nv30->pipe.bind_vs_state = nv30_vp_state_bind; + nv30->pipe.delete_vs_state = nv30_vp_state_delete; + + nv30->pipe.create_fs_state = nv30_fp_state_create; + nv30->pipe.bind_fs_state = nv30_fp_state_bind; + nv30->pipe.delete_fs_state = nv30_fp_state_delete; + + nv30->pipe.set_blend_color = nv30_set_blend_color; + nv30->pipe.set_clip_state = nv30_set_clip_state; + nv30->pipe.set_constant_buffer = nv30_set_constant_buffer; + nv30->pipe.set_framebuffer_state = nv30_set_framebuffer_state; + nv30->pipe.set_polygon_stipple = nv30_set_polygon_stipple; + nv30->pipe.set_scissor_state = nv30_set_scissor_state; + nv30->pipe.set_viewport_state = nv30_set_viewport_state; + + nv30->pipe.set_vertex_buffer = nv30_set_vertex_buffer; + nv30->pipe.set_vertex_element = nv30_set_vertex_element; +} + diff --git a/src/mesa/pipe/nv30/nv30_state.h b/src/mesa/pipe/nv30/nv30_state.h new file mode 100644 index 00000000000..233600f69ab --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_state.h @@ -0,0 +1,147 @@ +#ifndef __NV30_STATE_H__ +#define __NV30_STATE_H__ + +#include "pipe/p_state.h" + +struct nv30_blend_state { + uint32_t b_enable; + uint32_t b_srcfunc; + uint32_t b_dstfunc; + uint32_t b_eqn; + + uint32_t l_enable; + uint32_t l_op; + + uint32_t c_mask; + + uint32_t d_enable; +}; + +struct nv30_sampler_state { + uint32_t fmt; + uint32_t wrap; + uint32_t en; + uint32_t filt; + uint32_t bcol; +}; + +struct nv30_rasterizer_state { + uint32_t shade_model; + + uint32_t line_width; + uint32_t line_smooth_en; + uint32_t line_stipple_en; + uint32_t line_stipple; + + uint32_t point_size; + + uint32_t poly_smooth_en; + uint32_t poly_stipple_en; + + uint32_t poly_mode_front; + uint32_t poly_mode_back; + + uint32_t front_face; + uint32_t cull_face; + uint32_t cull_face_en; + + uint32_t point_sprite; +}; + +struct nv30_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv30_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + +struct nv30_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + struct nv30_vertex_program_exec *insns; + unsigned nr_insns; + struct nv30_vertex_program_data *consts; + unsigned nr_consts; + + struct nouveau_resource *exec; + unsigned exec_start; + struct nouveau_resource *data; + unsigned data_start; + unsigned data_start_min; + + uint32_t ir; + uint32_t or; +}; + +struct nv30_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv30_fragment_program { + const struct pipe_shader_state *pipe; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv30_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + uint32_t fp_reg_control; +}; + +struct nv30_stencil_push { + uint32_t enable; + uint32_t wmask; + uint32_t func; + uint32_t ref; + uint32_t vmask; + uint32_t fail; + uint32_t zfail; + uint32_t zpass; +}; + +struct nv30_depth_stencil_alpha_state { + struct { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; + } depth; + + struct { + struct nv30_stencil_push back; + struct nv30_stencil_push front; + } stencil; + + struct { + uint32_t enabled; + uint32_t func; + uint32_t ref; + } alpha; +}; + +struct nv30_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +#endif diff --git a/src/mesa/pipe/nv30/nv30_state_emit.c b/src/mesa/pipe/nv30/nv30_state_emit.c new file mode 100644 index 00000000000..70b98836f0f --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_state_emit.c @@ -0,0 +1,83 @@ +#include "nv30_context.h" +#include "nv30_state.h" + +void +nv30_emit_hw_state(struct nv30_context *nv30) +{ + int i; + + if (nv30->dirty & NV30_NEW_FRAGPROG) { + nv30_fragprog_bind(nv30, nv30->fragprog.current); + /*XXX: clear NV30_NEW_FRAGPROG if no new program uploaded */ + } + + if (nv30->dirty_samplers || (nv30->dirty & NV30_NEW_FRAGPROG)) { + nv30_fragtex_bind(nv30); +/* + BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); + OUT_RING (1);*/ + nv30->dirty &= ~NV30_NEW_FRAGPROG; + } + + if (nv30->dirty & NV30_NEW_VERTPROG) { + nv30_vertprog_bind(nv30, nv30->vertprog.current); + nv30->dirty &= ~NV30_NEW_VERTPROG; + } + + nv30->dirty_samplers = 0; + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv30_vbo.c + */ + + /* Render targets */ + if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR0) { + BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv30->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(rankine, NV34TCL_COLOR0_OFFSET, 1); + OUT_RELOCl(nv30->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR1) { + BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv30->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(rankine, NV34TCL_COLOR1_OFFSET, 1); + OUT_RELOCl(nv30->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv30->zeta) { + BEGIN_RING(rankine, NV34TCL_DMA_ZETA, 1); + OUT_RELOCo(nv30->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(rankine, NV34TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv30->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + /* XXX allocate LMA */ +/* BEGIN_RING(rankine, NV34TCL_LMA_DEPTH_OFFSET, 1); + OUT_RING(0);*/ + } + + /* Texture images */ + for (i = 0; i < 16; i++) { + if (!(nv30->fp_samplers & (1 << i))) + continue; + BEGIN_RING(rankine, NV34TCL_TX_OFFSET(i), 2); + OUT_RELOCl(nv30->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv30->tex[i].buffer, nv30->tex[i].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV34TCL_TX_FORMAT_DMA0, + NV34TCL_TX_FORMAT_DMA1); + } + + /* Fragment program */ + BEGIN_RING(rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); + OUT_RELOC (nv30->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR, NV34TCL_FP_ACTIVE_PROGRAM_DMA0, + NV34TCL_FP_ACTIVE_PROGRAM_DMA1); +} + diff --git a/src/mesa/pipe/nv30/nv30_surface.c b/src/mesa/pipe/nv30/nv30_surface.c new file mode 100644 index 00000000000..31745e3d6e8 --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_surface.c @@ -0,0 +1,136 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv30_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "pipe/util/p_tile.h" + +static boolean +nv30_surface_format_supported(struct pipe_context *pipe, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static struct pipe_surface * +nv30_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv30mt->level[level].pitch / ps->cpp; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv30mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv30mt->level[level].image_offset[zslice]; + } else { + ps->offset = nv30mt->level[level].image_offset[0]; + } + + return ps; +} + +static void +nv30_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, struct pipe_surface *src, + unsigned srcx, unsigned srcy, unsigned width, unsigned height) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv30_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv30_init_surface_functions(struct nv30_context *nv30) +{ + nv30->pipe.is_format_supported = nv30_surface_format_supported; + nv30->pipe.get_tex_surface = nv30_get_tex_surface; + nv30->pipe.surface_copy = nv30_surface_copy; + nv30->pipe.surface_fill = nv30_surface_fill; +} diff --git a/src/mesa/pipe/nv30/nv30_vbo.c b/src/mesa/pipe/nv30/nv30_vbo.c new file mode 100644 index 00000000000..e6c50d3820c --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_vbo.c @@ -0,0 +1,406 @@ +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +#include "pipe/nouveau/nouveau_channel.h" +#include "pipe/nouveau/nouveau_pushbuf.h" + +static INLINE int +nv30_vbo_ncomp(uint format) +{ + int ncomp = 0; + + if (pf_size_x(format)) ncomp++; + if (pf_size_y(format)) ncomp++; + if (pf_size_z(format)) ncomp++; + if (pf_size_w(format)) ncomp++; + + return ncomp; +} + +static INLINE int +nv30_vbo_type(uint format) +{ + switch (pf_type(format)) { + case PIPE_FORMAT_TYPE_FLOAT: + return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + case PIPE_FORMAT_TYPE_UNORM: + return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE; + default: + assert(0); + } +} + +static boolean +nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, + struct pipe_vertex_element *ve, + struct pipe_vertex_buffer *vb) +{ + struct pipe_winsys *ws = nv30->pipe.winsys; + int type, ncomp; + void *map; + + type = nv30_vbo_type(ve->src_format); + ncomp = nv30_vbo_ncomp(ve->src_format); + + map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); + map += vb->buffer_offset + ve->src_offset; + + switch (type) { + case NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT: + { + float *v = map; + + BEGIN_RING(rankine, NV34TCL_VERTEX_ATTR_4F_X(attrib), 4); + switch (ncomp) { + case 4: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(v[3]); + break; + case 3: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(1.0); + break; + case 2: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + case 1: + OUT_RINGf(v[0]); + OUT_RINGf(0.0); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + } + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + + ws->buffer_unmap(ws, vb->buffer); + + return TRUE; +} + +static void +nv30_vbo_arrays_update(struct nv30_context *nv30) +{ + struct nv30_vertex_program *vp = nv30->vertprog.active; + uint32_t inputs, vtxfmt[16]; + int hw, num_hw; + + nv30->vb_enable = 0; + + inputs = vp->ir; + for (hw = 0; hw < 16 && inputs; hw++) { + if (inputs & (1 << hw)) { + num_hw = hw; + inputs &= ~(1 << hw); + } + } + num_hw++; + + inputs = vp->ir; + for (hw = 0; hw < num_hw; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + + if (!(inputs & (1 << hw))) { + vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + continue; + } + + ve = &nv30->vtxelt[hw]; + vb = &nv30->vtxbuf[ve->vertex_buffer_index]; + + if (vb->pitch == 0) { + vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + if (nv30_vbo_static_attrib(nv30, hw, ve, vb) == TRUE) + continue; + } + + nv30->vb_enable |= (1 << hw); + nv30->vb[hw].delta = vb->buffer_offset + ve->src_offset; + nv30->vb[hw].buffer = vb->buffer; + + vtxfmt[hw] = ((vb->pitch << NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT) | + (nv30_vbo_ncomp(ve->src_format) << + NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT) | + nv30_vbo_type(ve->src_format)); + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_ARRAY_FORMAT(0), num_hw); + OUT_RINGp (vtxfmt, num_hw); +} + +static boolean +nv30_vbo_validate_state(struct nv30_context *nv30, + struct pipe_buffer *ib, unsigned ib_format) +{ + unsigned inputs; + + nv30_emit_hw_state(nv30); + + if (nv30->dirty & NV30_NEW_ARRAYS) { + nv30_vbo_arrays_update(nv30); + nv30->dirty &= ~NV30_NEW_ARRAYS; + } + + inputs = nv30->vb_enable; + while (inputs) { + unsigned a = ffs(inputs) - 1; + + inputs &= ~(1 << a); + + BEGIN_RING(rankine, NV34TCL_VERTEX_BUFFER_ADDRESS(a), 1); + OUT_RELOC (nv30->vb[a].buffer, nv30->vb[a].delta, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, + NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1); + } + + if (ib) { + BEGIN_RING(rankine, NV40TCL_IDXBUF_ADDRESS, 2); + OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD); + OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); + } + + BEGIN_RING(rankine, 0x1710, 1); + OUT_RING (0); /* vtx cache flush */ + + return TRUE; +} + +boolean +nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, + unsigned count) +{ + struct nv30_context *nv30 = nv30_context(pipe); + unsigned nr; + + assert(nv30_vbo_validate_state(nv30, NULL, 0)); + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + pipe->flush(pipe, 0); + return TRUE; +} + +static INLINE void +nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, + unsigned start, unsigned count) +{ + uint8_t *elts = (uint8_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(rankine, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2046); + + BEGIN_RING_NI(rankine, NV40TCL_VB_ELEMENT_U16, push); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, + unsigned start, unsigned count) +{ + uint16_t *elts = (uint16_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2046); + + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, + unsigned start, unsigned count) +{ + uint32_t *elts = (uint32_t *)ib + start; + int push; + + while (count) { + push = MIN2(count, 2047); + + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push); + OUT_RINGp (elts, push); + + count -= push; + elts += push; + } +} + +static boolean +nv30_draw_elements_inline(struct pipe_context *pipe, + struct pipe_buffer *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + void *map; + + assert(nv30_vbo_validate_state(nv30, NULL, 0)); + + map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); + if (!ib) + assert(0); + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + switch (ib_size) { + case 1: + nv30_draw_elements_u08(nv30, map, start, count); + break; + case 2: + nv30_draw_elements_u16(nv30, map, start, count); + break; + case 4: + nv30_draw_elements_u32(nv30, map, start, count); + break; + default: + assert(0); + break; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + ws->buffer_unmap(ws, ib); + + return TRUE; +} + +static boolean +nv30_draw_elements_vbo(struct pipe_context *pipe, + struct pipe_buffer *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv30_context *nv30 = nv30_context(pipe); + unsigned nr, type; + + switch (ib_size) { + case 2: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; + break; + case 4: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; + break; + default: + assert(0); + } + + assert(nv30_vbo_validate_state(nv30, ib, type)); + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(rankine, NV40TCL_VB_INDEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(rankine, NV40TCL_VB_INDEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + return TRUE; +} + +boolean +nv30_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ + if (indexSize != 1) { + nv30_draw_elements_vbo(pipe, indexBuffer, indexSize, + mode, start, count); + } else { + nv30_draw_elements_inline(pipe, indexBuffer, indexSize, + mode, start, count); + } + + pipe->flush(pipe, 0); + return TRUE; +} + + diff --git a/src/mesa/pipe/nv30/nv30_vertprog.c b/src/mesa/pipe/nv30/nv30_vertprog.c new file mode 100644 index 00000000000..b712049fa7c --- /dev/null +++ b/src/mesa/pipe/nv30/nv30_vertprog.c @@ -0,0 +1,778 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "pipe/tgsi/util/tgsi_parse.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +/* TODO (at least...): + * 1. Indexed consts + ARL + * 2. Arb. swz/negation + * 3. NV_vp11, NV_vp2, NV_vp3 features + * - extra arith opcodes + * - branching + * - texture sampling + * - indexed attribs + * - indexed results + * 4. bugs + */ + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 8 +#define MASK_Y 4 +#define MASK_Z 2 +#define MASK_W 1 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE 0 +#define DEF_CTEST 0 +#include "nv30_shader.h" + +#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv30_sr_neg((s)) +#define abs(s) nv30_sr_abs((s)) + +struct nv30_vpc { + struct nv30_vertex_program *vp; + + struct nv30_vertex_program_exec *vpi; + + unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; + + int high_temp; + int temp_temp_count; + + struct nv30_sreg *imm; + unsigned nr_imm; +}; + +static struct nv30_sreg +temp(struct nv30_vpc *vpc) +{ + int idx; + + idx = vpc->temp_temp_count++; + idx += vpc->high_temp + 1; + return nv30_sr(NV30SR_TEMP, idx); +} + +static struct nv30_sreg +constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) +{ + struct nv30_vertex_program *vp = vpc->vp; + struct nv30_vertex_program_data *vpd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < vp->nr_consts; idx++) { + if (vp->consts[idx].index == pipe) + return nv30_sr(NV30SR_CONST, idx); + } + } + + idx = vp->nr_consts++; + vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); + vpd = &vp->consts[idx]; + + vpd->index = pipe; + vpd->value[0] = x; + vpd->value[1] = y; + vpd->value[2] = z; + vpd->value[3] = w; + return nv30_sr(NV30SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) + +static void +emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t sr = 0; + + switch (src.type) { + case NV30SR_TEMP: + sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT); + sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV30SR_INPUT: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + vp->ir |= (1 << src.index); + hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT); + break; + case NV30SR_CONST: + sr |= (NV30_VP_SRC_REG_TYPE_CONST << + NV30_VP_SRC_REG_TYPE_SHIFT); + assert(vpc->vpi->const_index == -1 || + vpc->vpi->const_index == src.index); + vpc->vpi->const_index = src.index; + break; + case NV30SR_NONE: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV30_VP_SRC_NEGATE; + + if (src.abs) + hw[0] |= (1 << (21 + pos)); + + sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) | + (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) | + (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) | + (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT)); + +/* + * |VVV| + * d°.°b + * \u/ + * + */ + + switch (pos) { + case 0: + hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >> + NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT; + hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) << + NV30_VP_INST_SRC0L_SHIFT; + break; + case 1: + hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT; + break; + case 2: + hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >> + NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT; + hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) << + NV30_VP_INST_SRC2L_SHIFT; + break; + default: + assert(0); + } +} + +static void +emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) +{ + struct nv30_vertex_program *vp = vpc->vp; + + switch (dst.type) { + case NV30SR_TEMP: + hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT); + break; + case NV30SR_OUTPUT: + switch (dst.index) { + case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; + case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; + case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; + case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; + case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; + case NV30_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; + case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; + case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; + case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; + case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; + case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; + case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; + case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; + case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + default: + break; + } + + hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); + hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); + break; + default: + assert(0); + } +} + +static void +nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, + struct nv30_sreg s2) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t *hw; + + vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); + vpc->vpi = &vp->insns[vp->nr_insns - 1]; + memset(vpc->vpi, 0, sizeof(*vpc->vpi)); + vpc->vpi->const_index = -1; + + hw = vpc->vpi->data; + + hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT); + hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) | + (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) | + (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) | + (3 << NV30_VP_INST_COND_SWZ_W_SHIFT)); + + hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT); +// hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; +// hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); + + emit_dst(vpc, hw, slot, dst); + emit_src(vpc, hw, 0, s0); + emit_src(vpc, hw, 1, s1); + emit_src(vpc, hw, 2, s2); +} + +static INLINE struct nv30_sreg +tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv30_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); + break; + case TGSI_FILE_CONSTANT: + src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + break; + case TGSI_FILE_IMMEDIATE: + src = vpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + if (vpc->high_temp < fsrc->SrcRegister.Index) + vpc->high_temp = fsrc->SrcRegister.Index; + src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv30_sreg +tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + struct nv30_sreg dst; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + dst = nv30_sr(NV30SR_OUTPUT, + vpc->output_map[fdst->DstRegister.Index]); + + break; + case TGSI_FILE_TEMPORARY: + dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); + if (vpc->high_temp < dst.index) + vpc->high_temp = dst.index; + break; + default: + NOUVEAU_ERR("bad dst file\n"); + break; + } + + return dst; +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, + const struct tgsi_full_instruction *finst) +{ + struct nv30_sreg src[3], dst, tmp; + struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + int mask; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + vpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(vpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + /*XXX: index comparison is broken now that consts come from + * two different register files. + */ + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); + break; + case TGSI_OPCODE_ARL: + arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DST: + arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_EXP: + arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_FLR: + arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_LG2: + arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LIT: + arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LOG: + arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_MAD: + arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(vpc); + arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, + swz(src[0], X, X, X, X)); + arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(vpc, 1, OP_EX2, dst, mask, none, none, + swz(tmp, X, X, X, X)); + break; + case TGSI_OPCODE_RCP: + arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_RET: + break; + case TGSI_OPCODE_RSQ: + arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_SGE: + arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); + break; + case TGSI_OPCODE_XPD: + tmp = temp(vpc); + arith(vpc, 0, OP_MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV30_VP_INST_DEST_POS; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_BCOLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_BFC0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_BFC1; + } else { + NOUVEAU_ERR("bad bcolour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV30_VP_INST_DEST_FOGC; + break; + case TGSI_SEMANTIC_PSIZE: + hw = NV30_VP_INST_DEST_PSZ; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + vpc->output_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv30_vertprog_prepare(struct nv30_vpc *vpc) +{ + struct tgsi_parse_context p; + int nr_imm = 0; + + tgsi_parse_init(&p, vpc->vp->pipe->tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch(tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + nr_imm++; + break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (nr_imm) { + vpc->imm = calloc(nr_imm, sizeof(struct nv30_sreg)); + assert(vpc->imm); + } + + return TRUE; +} + +void +nv30_vertprog_translate(struct nv30_context *nv30, + struct nv30_vertex_program *vp) +{ + struct tgsi_parse_context parse; + struct nv30_vpc *vpc = NULL; + + vpc = calloc(1, sizeof(struct nv30_vpc)); + if (!vpc) + return; + vpc->vp = vp; + vpc->high_temp = -1; + + if (!nv30_vertprog_prepare(vpc)) { + free(vpc); + return; + } + + tgsi_parse_init(&parse, vp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_OUTPUT: + if (!nv30_vertprog_parse_decl_output(vpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); +// assert(imm->Immediate.Size == 4); + vpc->imm[vpc->nr_imm++] = + constant(vpc, -1, + imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + finst = &parse.FullToken.FullInstruction; + if (!nv30_vertprog_parse_instruction(vpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST; + vp->translated = TRUE; +out_err: + tgsi_parse_free(&parse); + free(vpc); +} + +void +nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) +{ + struct nouveau_winsys *nvws = nv30->nvws; + struct pipe_winsys *ws = nv30->pipe.winsys; + boolean upload_code = FALSE, upload_data = FALSE; + int i; + + /* Translate TGSI shader into hw bytecode */ + if (!vp->translated) { + nv30_vertprog_translate(nv30, vp); + if (!vp->translated) + assert(0); + } + + /* Allocate hw vtxprog exec slots */ + if (!vp->exec) { + struct nouveau_resource *heap = nv30->vertprog.exec_heap; + uint vplen = vp->nr_insns; + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { + while (heap->next && heap->size < vplen) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->exec); + } + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) + assert(0); + } + + upload_code = TRUE; + } + + /* Allocate hw vtxprog const slots */ + if (vp->nr_consts && !vp->data) { + struct nouveau_resource *heap = nv30->vertprog.data_heap; + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { + while (heap->next && heap->size < vp->nr_consts) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) + assert(0); + } + + /*XXX: handle this some day */ + assert(vp->data->start >= vp->data_start_min); + + upload_data = TRUE; + if (vp->data_start != vp->data->start) + upload_code = TRUE; + } + + /* If exec or data segments moved we need to patch the program to + * fixup offsets and register IDs. + */ + if (vp->exec_start != vp->exec->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->has_branch_offset) { + assert(0); + } + } + + vp->exec_start = vp->exec->start; + } + + if (vp->nr_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->const_index >= 0) { + vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; + vpi->data[1] |= + (vpi->const_index + vp->data->start) << + NV30_VP_INST_CONST_SRC_SHIFT; + + } + } + + vp->data_start = vp->data->start; + } + + /* Update + Upload constant values */ + if (vp->nr_consts) { + float *map = NULL; + + if (nv30->vertprog.constant_buf) { + map = ws->buffer_map(ws, nv30->vertprog.constant_buf, + PIPE_BUFFER_USAGE_CPU_READ); + } + + for (i = 0; i < vp->nr_consts; i++) { + struct nv30_vertex_program_data *vpd = &vp->consts[i]; + + if (vpd->index >= 0) { + if (!upload_data && + !memcmp(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float))) + continue; + memcpy(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float)); + } + + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (i + vp->data->start); + OUT_RINGp ((uint32_t *)vpd->value, 4); + } + + if (map) { + ws->buffer_unmap(ws, nv30->vertprog.constant_buf); + } + } + + /* Upload vtxprog */ + if (upload_code) { +#if 0 + for (i = 0; i < vp->nr_insns; i++) { + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); + } +#endif + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (vp->exec->start); + for (i = 0; i < vp->nr_insns; i++) { + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (vp->insns[i].data, 4); + } + } + + BEGIN_RING(rankine, NV34TCL_VP_START_FROM_ID, 1); + OUT_RING (vp->exec->start); + BEGIN_RING(rankine, NV34TCL_VP_ATTRIB_EN, 2); + OUT_RING (vp->ir); + OUT_RING (vp->or); + + nv30->vertprog.active = vp; +} + +void +nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) +{ + if (vp->nr_consts) + free(vp->consts); + if (vp->nr_insns) + free(vp->insns); +} + -- cgit v1.2.3 From 903521a6c031757d63b48129d08ba043d183dbdc Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 15 Feb 2008 02:41:34 +0100 Subject: nouveau: oops and make nouveau winsys build by default --- configs/linux-dri | 2 +- src/mesa/drivers/dri/nouveau_winsys/Makefile | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/configs/linux-dri b/configs/linux-dri index 936fce99829..494b0aab8e9 100644 --- a/configs/linux-dri +++ b/configs/linux-dri @@ -66,4 +66,4 @@ WINDOW_SYSTEM=dri # gamma are missing because they have not been converted to use the new # interface. -DRI_DIRS = intel_winsys +DRI_DIRS = intel_winsys nouveau_winsys diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile index 59ba561eb93..98ec5a79f5d 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ b/src/mesa/drivers/dri/nouveau_winsys/Makefile @@ -8,7 +8,7 @@ MINIGLX_SOURCES = PIPE_DRIVERS = \ $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a \ - $(TOP)/src/mesa/pipe/nv40/libnv30.a \ + $(TOP)/src/mesa/pipe/nv30/libnv30.a \ $(TOP)/src/mesa/pipe/nv40/libnv40.a \ $(TOP)/src/mesa/pipe/nv50/libnv50.a -- cgit v1.2.3 From a7872d4c41a692a81cc54cb5eaaab04308604bdf Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Mon, 28 Jan 2008 15:48:51 +0900 Subject: Clone vf module. --- src/mesa/pipe/draw/draw_vf.c | 374 +++++++++++++ src/mesa/pipe/draw/draw_vf.h | 249 +++++++++ src/mesa/pipe/draw/draw_vf_generic.c | 983 +++++++++++++++++++++++++++++++++++ src/mesa/pipe/draw/draw_vf_sse.c | 664 +++++++++++++++++++++++ src/mesa/sources | 3 + 5 files changed, 2273 insertions(+) create mode 100644 src/mesa/pipe/draw/draw_vf.c create mode 100644 src/mesa/pipe/draw/draw_vf.h create mode 100644 src/mesa/pipe/draw/draw_vf_generic.c create mode 100644 src/mesa/pipe/draw/draw_vf_sse.c diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c new file mode 100644 index 00000000000..f758460b5f6 --- /dev/null +++ b/src/mesa/pipe/draw/draw_vf.c @@ -0,0 +1,374 @@ +/* + * Copyright 2003 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + */ + +#include "glheader.h" +#include "context.h" +#include "colormac.h" + +#include "pipe/p_compiler.h" + +#include "draw_vf.h" + +#define DBG 0 + + + +static boolean match_fastpath( struct draw_vertex_fetch *vf, + const struct draw_vf_fastpath *fp) +{ + unsigned j; + + if (vf->attr_count != fp->attr_count) + return FALSE; + + for (j = 0; j < vf->attr_count; j++) + if (vf->attr[j].format != fp->attr[j].format || + vf->attr[j].inputsize != fp->attr[j].size || + vf->attr[j].vertoffset != fp->attr[j].offset) + return FALSE; + + if (fp->match_strides) { + if (vf->vertex_stride != fp->vertex_stride) + return FALSE; + + for (j = 0; j < vf->attr_count; j++) + if (vf->attr[j].inputstride != fp->attr[j].stride) + return FALSE; + } + + return TRUE; +} + +static boolean search_fastpath_emit( struct draw_vertex_fetch *vf ) +{ + struct draw_vf_fastpath *fp = vf->fastpath; + + for ( ; fp ; fp = fp->next) { + if (match_fastpath(vf, fp)) { + vf->emit = fp->func; + return TRUE; + } + } + + return FALSE; +} + +void draw_vf_register_fastpath( struct draw_vertex_fetch *vf, + boolean match_strides ) +{ + struct draw_vf_fastpath *fastpath = CALLOC_STRUCT(draw_vf_fastpath); + unsigned i; + + fastpath->vertex_stride = vf->vertex_stride; + fastpath->attr_count = vf->attr_count; + fastpath->match_strides = match_strides; + fastpath->func = vf->emit; + fastpath->attr = (struct draw_vf_attr_type *) + _mesa_malloc(vf->attr_count * sizeof(fastpath->attr[0])); + + for (i = 0; i < vf->attr_count; i++) { + fastpath->attr[i].format = vf->attr[i].format; + fastpath->attr[i].stride = vf->attr[i].inputstride; + fastpath->attr[i].size = vf->attr[i].inputsize; + fastpath->attr[i].offset = vf->attr[i].vertoffset; + } + + fastpath->next = vf->fastpath; + vf->fastpath = fastpath; +} + + + + +/*********************************************************************** + * Build codegen functions or return generic ones: + */ +static void choose_emit_func( struct draw_vertex_fetch *vf, + unsigned count, + uint8_t *dest) +{ + vf->emit = NULL; + + /* Does this match an existing (hardwired, codegen or known-bad) + * fastpath? + */ + if (search_fastpath_emit(vf)) { + /* Use this result. If it is null, then it is already known + * that the current state will fail for codegen and there is no + * point trying again. + */ + } + else if (vf->codegen_emit) { + vf->codegen_emit( vf ); + } + + if (!vf->emit) { + draw_vf_generate_hardwired_emit(vf); + } + + /* Otherwise use the generic version: + */ + if (!vf->emit) + vf->emit = draw_vf_generic_emit; + + vf->emit( vf, count, dest ); +} + + + + + +/*********************************************************************** + * Public entrypoints, mostly dispatch to the above: + */ + + + +unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, + const struct draw_vf_attr_map *map, + unsigned nr, + unsigned vertex_stride ) +{ + unsigned offset = 0; + unsigned i, j; + + assert(nr < DRAW_VF_ATTRIB_MAX); + + memset(vf->lookup, 0, sizeof(vf->lookup)); + + for (j = 0, i = 0; i < nr; i++) { + const unsigned format = map[i].format; + if (format == EMIT_PAD) { + if (DBG) + _mesa_printf("%d: pad %d, offset %d\n", i, + map[i].offset, offset); + + offset += map[i].offset; + + } + else { + assert(vf->lookup[map[i].attrib] == 0); + vf->lookup[map[i].attrib] = &vf->attr[j]; + + vf->attr[j].attrib = map[i].attrib; + vf->attr[j].format = format; + vf->attr[j].insert = draw_vf_format_info[format].insert; + vf->attr[j].extract = draw_vf_format_info[format].extract; + vf->attr[j].vertattrsize = draw_vf_format_info[format].attrsize; + vf->attr[j].vertoffset = offset; + + if (DBG) + _mesa_printf("%d: %s, offset %d\n", i, + draw_vf_format_info[format].name, + vf->attr[j].vertoffset); + + offset += draw_vf_format_info[format].attrsize; + j++; + } + } + + vf->attr_count = j; + vf->vertex_stride = vertex_stride ? vertex_stride : offset; + vf->emit = choose_emit_func; + + assert(vf->vertex_stride >= offset); + return vf->vertex_stride; +} + + + +void draw_vf_set_vp_matrix( struct draw_vertex_fetch *vf, + const float *viewport ) +{ + assert(vf->allow_viewport_emits); + + /* scale */ + vf->vp[0] = viewport[MAT_SX]; + vf->vp[1] = viewport[MAT_SY]; + vf->vp[2] = viewport[MAT_SZ]; + vf->vp[3] = 1.0; + + /* translate */ + vf->vp[4] = viewport[MAT_TX]; + vf->vp[5] = viewport[MAT_TY]; + vf->vp[6] = viewport[MAT_TZ]; + vf->vp[7] = 0.0; +} + +void draw_vf_set_vp_scale_translate( struct draw_vertex_fetch *vf, + const float *scale, + const float *translate ) +{ + assert(vf->allow_viewport_emits); + + vf->vp[0] = scale[0]; + vf->vp[1] = scale[1]; + vf->vp[2] = scale[2]; + vf->vp[3] = scale[3]; + + vf->vp[4] = translate[0]; + vf->vp[5] = translate[1]; + vf->vp[6] = translate[2]; + vf->vp[7] = translate[3]; +} + + +/* Set attribute pointers, adjusted for start position: + */ +void draw_vf_set_sources( struct draw_vertex_fetch *vf, + GLvector4f * const sources[], + unsigned start ) +{ + struct draw_vf_attr *a = vf->attr; + unsigned j; + + for (j = 0; j < vf->attr_count; j++) { + const GLvector4f *vptr = sources[a[j].attrib]; + + if ((a[j].inputstride != vptr->stride) || + (a[j].inputsize != vptr->size)) + vf->emit = choose_emit_func; + + a[j].inputstride = vptr->stride; + a[j].inputsize = vptr->size; + a[j].do_insert = a[j].insert[vptr->size - 1]; + a[j].inputptr = ((uint8_t *)vptr->data) + start * vptr->stride; + } +} + + + +/* Emit count VB vertices to dest. + */ +void draw_vf_emit_vertices( struct draw_vertex_fetch *vf, + unsigned count, + void *dest ) +{ + vf->emit( vf, count, (uint8_t*) dest ); +} + + +/* Extract a named attribute from a hardware vertex. Will have to + * reverse any viewport transformation, swizzling or other conversions + * which may have been applied. + * + * This is mainly required for on-the-fly vertex translations to + * swrast format. + */ +void draw_vf_get_attr( struct draw_vertex_fetch *vf, + const void *vertex, + GLenum attr, + const float *dflt, + float *dest ) +{ + const struct draw_vf_attr *a = vf->attr; + const unsigned attr_count = vf->attr_count; + unsigned j; + + for (j = 0; j < attr_count; j++) { + if (a[j].attrib == attr) { + a[j].extract( &a[j], dest, (uint8_t *)vertex + a[j].vertoffset ); + return; + } + } + + /* Else return the value from ctx->Current. + */ + _mesa_memcpy( dest, dflt, 4*sizeof(float)); +} + + + + +struct draw_vertex_fetch *draw_vf_create( boolean allow_viewport_emits ) +{ + struct draw_vertex_fetch *vf = CALLOC_STRUCT(draw_vertex_fetch); + unsigned i; + + for (i = 0; i < DRAW_VF_ATTRIB_MAX; i++) + vf->attr[i].vf = vf; + + vf->allow_viewport_emits = allow_viewport_emits; + + switch(CHAN_TYPE) { + case GL_UNSIGNED_BYTE: + vf->chan_scale[0] = 255.0; + vf->chan_scale[1] = 255.0; + vf->chan_scale[2] = 255.0; + vf->chan_scale[3] = 255.0; + break; + case GL_UNSIGNED_SHORT: + vf->chan_scale[0] = 65535.0; + vf->chan_scale[1] = 65535.0; + vf->chan_scale[2] = 65535.0; + vf->chan_scale[3] = 65535.0; + break; + default: + vf->chan_scale[0] = 1.0; + vf->chan_scale[1] = 1.0; + vf->chan_scale[2] = 1.0; + vf->chan_scale[3] = 1.0; + break; + } + + vf->identity[0] = 0.0; + vf->identity[1] = 0.0; + vf->identity[2] = 0.0; + vf->identity[3] = 1.0; + + vf->codegen_emit = NULL; + +#ifdef USE_SSE_ASM + if (!_mesa_getenv("MESA_NO_CODEGEN")) + vf->codegen_emit = draw_vf_generate_sse_emit; +#endif + + return vf; +} + + +void draw_vf_destroy( struct draw_vertex_fetch *vf ) +{ + struct draw_vf_fastpath *fp, *tmp; + + for (fp = vf->fastpath ; fp ; fp = tmp) { + tmp = fp->next; + FREE(fp->attr); + + /* KW: At the moment, fp->func is constrained to be allocated by + * _mesa_exec_alloc(), as the hardwired fastpaths in + * t_vertex_generic.c are handled specially. It would be nice + * to unify them, but this probably won't change until this + * module gets another overhaul. + */ + _mesa_exec_free((void *) fp->func); + FREE(fp); + } + + vf->fastpath = NULL; + FREE(vf); +} diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h new file mode 100644 index 00000000000..279570aad5f --- /dev/null +++ b/src/mesa/pipe/draw/draw_vf.h @@ -0,0 +1,249 @@ +/* + * Copyright 2003 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + */ + +#ifndef DRAW_VF_H +#define DRAW_VF_H + + +#include "pipe/p_compiler.h" +#include "math/m_vector.h" + + +enum { + DRAW_VF_ATTRIB_POS = 0, + DRAW_VF_ATTRIB_WEIGHT = 1, + DRAW_VF_ATTRIB_NORMAL = 2, + DRAW_VF_ATTRIB_COLOR0 = 3, + DRAW_VF_ATTRIB_COLOR1 = 4, + DRAW_VF_ATTRIB_FOG = 5, + DRAW_VF_ATTRIB_COLOR_INDEX = 6, + DRAW_VF_ATTRIB_EDGEFLAG = 7, + DRAW_VF_ATTRIB_TEX0 = 8, + DRAW_VF_ATTRIB_TEX1 = 9, + DRAW_VF_ATTRIB_TEX2 = 10, + DRAW_VF_ATTRIB_TEX3 = 11, + DRAW_VF_ATTRIB_TEX4 = 12, + DRAW_VF_ATTRIB_TEX5 = 13, + DRAW_VF_ATTRIB_TEX6 = 14, + DRAW_VF_ATTRIB_TEX7 = 15, + DRAW_VF_ATTRIB_VAR0 = 16, + DRAW_VF_ATTRIB_VAR1 = 17, + DRAW_VF_ATTRIB_VAR2 = 18, + DRAW_VF_ATTRIB_VAR3 = 19, + DRAW_VF_ATTRIB_VAR4 = 20, + DRAW_VF_ATTRIB_VAR5 = 21, + DRAW_VF_ATTRIB_VAR6 = 22, + DRAW_VF_ATTRIB_VAR7 = 23, + DRAW_VF_ATTRIB_POINTSIZE = 24, + DRAW_VF_ATTRIB_BFC0 = 25, + DRAW_VF_ATTRIB_BFC1 = 26, + DRAW_VF_ATTRIB_CLIP_POS = 27, + DRAW_VF_ATTRIB_VERTEX_HEADER = 28, + DRAW_VF_ATTRIB_MAX = 29 +}; + +enum draw_vf_attr_format { + EMIT_1F, + EMIT_2F, + EMIT_3F, + EMIT_4F, + EMIT_2F_VIEWPORT, /**< do viewport transform and emit */ + EMIT_3F_VIEWPORT, /**< do viewport transform and emit */ + EMIT_4F_VIEWPORT, /**< do viewport transform and emit */ + EMIT_3F_XYW, /**< for projective texture */ + EMIT_1UB_1F, /**< for fog coordinate */ + EMIT_3UB_3F_RGB, /**< for specular color */ + EMIT_3UB_3F_BGR, /**< for specular color */ + EMIT_4UB_4F_RGBA, /**< for color */ + EMIT_4UB_4F_BGRA, /**< for color */ + EMIT_4UB_4F_ARGB, /**< for color */ + EMIT_4UB_4F_ABGR, /**< for color */ + EMIT_4CHAN_4F_RGBA, /**< for swrast color */ + EMIT_PAD, /**< leave a hole of 'offset' bytes */ + EMIT_MAX +}; + +struct draw_vf_attr_map { + unsigned attrib; + enum draw_vf_attr_format format; + unsigned offset; +}; + +struct draw_vertex_fetch; + + +void +draw_vf_set_vp_matrix( struct draw_vertex_fetch *vf, + const float *viewport ); + +void +draw_vf_set_vp_scale_translate( struct draw_vertex_fetch *vf, + const float *scale, + const float *translate ); + +unsigned +draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, + const struct draw_vf_attr_map *map, + unsigned nr, + unsigned vertex_stride ); + +void +draw_vf_set_sources( struct draw_vertex_fetch *vf, + GLvector4f * const attrib[], + unsigned start ); + +void +draw_vf_emit_vertices( struct draw_vertex_fetch *vf, + unsigned count, + void *dest ); + +void +draw_vf_get_attr( struct draw_vertex_fetch *vf, + const void *vertex, + GLenum attr, + const float *dflt, + float *dest ); + +struct draw_vertex_fetch * +draw_vf_create( boolean allow_viewport_emits ); + +void +draw_vf_destroy( struct draw_vertex_fetch *vf ); + + + +/*********************************************************************** + * Internal functions and structs: + */ + +struct draw_vf_attr; + +typedef void (*draw_vf_extract_func)( const struct draw_vf_attr *a, + float *out, + const uint8_t *v ); + +typedef void (*draw_vf_insert_func)( const struct draw_vf_attr *a, + uint8_t *v, + const float *in ); + +typedef void (*draw_vf_emit_func)( struct draw_vertex_fetch *vf, + unsigned count, + uint8_t *dest ); + + + +/** + * Describes how to convert/move a vertex attribute from a vertex + * array to a vertex structure. + */ +struct draw_vf_attr +{ + struct draw_vertex_fetch *vf; + + unsigned format; + unsigned inputsize; + unsigned inputstride; + unsigned vertoffset; /**< position of the attrib in the vertex struct */ + + unsigned attrib; /**< which vertex attrib (0=position, etc) */ + unsigned vertattrsize; /**< size of the attribute in bytes */ + + uint8_t *inputptr; + const draw_vf_insert_func *insert; + draw_vf_insert_func do_insert; + draw_vf_extract_func extract; +}; + +struct draw_vertex_fetch +{ + struct draw_vf_attr attr[DRAW_VF_ATTRIB_MAX]; + unsigned attr_count; + unsigned vertex_stride; + + struct draw_vf_attr *lookup[DRAW_VF_ATTRIB_MAX]; + + draw_vf_emit_func emit; + + /* Parameters and constants for codegen: + */ + boolean allow_viewport_emits; + float vp[8]; + float chan_scale[4]; + float identity[4]; + + struct draw_vf_fastpath *fastpath; + + void (*codegen_emit)( struct draw_vertex_fetch *vf ); +}; + + +struct draw_vf_attr_type { + unsigned format; + unsigned size; + unsigned stride; + unsigned offset; +}; + +struct draw_vf_fastpath { + unsigned vertex_stride; + unsigned attr_count; + boolean match_strides; + + struct draw_vf_attr_type *attr; + + draw_vf_emit_func func; + struct draw_vf_fastpath *next; +}; + + +void +draw_vf_register_fastpath( struct draw_vertex_fetch *vtx, + boolean match_strides ); + +void +draw_vf_generic_emit( struct draw_vertex_fetch *vf, + unsigned count, + uint8_t *v ); + +void +draw_vf_generate_hardwired_emit( struct draw_vertex_fetch *vf ); + +void +draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ); + + +struct draw_vf_format_info { + const char *name; + draw_vf_extract_func extract; + draw_vf_insert_func insert[4]; + const unsigned attrsize; +}; + +const struct draw_vf_format_info draw_vf_format_info[EMIT_MAX]; + + +#endif diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c new file mode 100644 index 00000000000..19e6c587e5e --- /dev/null +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -0,0 +1,983 @@ + +/* + * Copyright 2003 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + */ + +#include "glheader.h" +#include "context.h" +#include "colormac.h" +#include "simple_list.h" + +#include "pipe/p_compiler.h" + +#include "draw_vf.h" + + +/* + * These functions take the NDC coordinates pointed to by 'in', apply the + * NDC->Viewport mapping and store the results at 'v'. + */ + +static INLINE void insert_4f_viewport_4( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = scale[1] * in[1] + trans[1]; + out[2] = scale[2] * in[2] + trans[2]; + out[3] = in[3]; +} + +static INLINE void insert_4f_viewport_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = scale[1] * in[1] + trans[1]; + out[2] = scale[2] * in[2] + trans[2]; + out[3] = 1; +} + +static INLINE void insert_4f_viewport_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = scale[1] * in[1] + trans[1]; + out[2] = trans[2]; + out[3] = 1; +} + +static INLINE void insert_4f_viewport_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = trans[1]; + out[2] = trans[2]; + out[3] = 1; +} + +static INLINE void insert_3f_viewport_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = scale[1] * in[1] + trans[1]; + out[2] = scale[2] * in[2] + trans[2]; +} + +static INLINE void insert_3f_viewport_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = scale[1] * in[1] + trans[1]; + out[2] = scale[2] * in[2] + trans[2]; +} + +static INLINE void insert_3f_viewport_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = trans[1]; + out[2] = trans[2]; +} + +static INLINE void insert_2f_viewport_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = scale[1] * in[1] + trans[1]; +} + +static INLINE void insert_2f_viewport_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + float *out = (float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = scale[0] * in[0] + trans[0]; + out[1] = trans[1]; +} + + +/* + * These functions do the same as above, except for the viewport mapping. + */ + +static INLINE void insert_4f_4( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = in[2]; + out[3] = in[3]; +} + +static INLINE void insert_4f_3( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = in[2]; + out[3] = 1; +} + +static INLINE void insert_4f_2( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = 0; + out[3] = 1; +} + +static INLINE void insert_4f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = 0; + out[2] = 0; + out[3] = 1; +} + +static INLINE void insert_3f_xyw_4( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = in[3]; +} + +static INLINE void insert_3f_xyw_err( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + (void) a; (void) v; (void) in; + _mesa_exit(1); +} + +static INLINE void insert_3f_3( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = in[2]; +} + +static INLINE void insert_3f_2( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = 0; +} + +static INLINE void insert_3f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = 0; + out[2] = 0; +} + + +static INLINE void insert_2f_2( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = in[1]; +} + +static INLINE void insert_2f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; + out[1] = 0; +} + +static INLINE void insert_1f_1( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + float *out = (float *)(v); + (void) a; + + out[0] = in[0]; +} + +static INLINE void insert_null( const struct draw_vf_attr *a, uint8_t *v, const float *in ) +{ + (void) a; (void) v; (void) in; +} + +static INLINE void insert_4chan_4f_rgba_4( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + GLchan *c = (GLchan *)v; + (void) a; + UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); + UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); + UNCLAMPED_FLOAT_TO_CHAN(c[2], in[2]); + UNCLAMPED_FLOAT_TO_CHAN(c[3], in[3]); +} + +static INLINE void insert_4chan_4f_rgba_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + GLchan *c = (GLchan *)v; + (void) a; + UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); + UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); + UNCLAMPED_FLOAT_TO_CHAN(c[2], in[2]); + c[3] = CHAN_MAX; +} + +static INLINE void insert_4chan_4f_rgba_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + GLchan *c = (GLchan *)v; + (void) a; + UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); + UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); + c[2] = 0; + c[3] = CHAN_MAX; +} + +static INLINE void insert_4chan_4f_rgba_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + GLchan *c = (GLchan *)v; + (void) a; + UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); + c[1] = 0; + c[2] = 0; + c[3] = CHAN_MAX; +} + +static INLINE void insert_4ub_4f_rgba_4( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[2]); + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[3]); +} + +static INLINE void insert_4ub_4f_rgba_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[2]); + v[3] = 0xff; +} + +static INLINE void insert_4ub_4f_rgba_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + v[2] = 0; + v[3] = 0xff; +} + +static INLINE void insert_4ub_4f_rgba_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + v[1] = 0; + v[2] = 0; + v[3] = 0xff; +} + +static INLINE void insert_4ub_4f_bgra_4( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[2]); + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[3]); +} + +static INLINE void insert_4ub_4f_bgra_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[2]); + v[3] = 0xff; +} + +static INLINE void insert_4ub_4f_bgra_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + v[0] = 0; + v[3] = 0xff; +} + +static INLINE void insert_4ub_4f_bgra_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + v[1] = 0; + v[0] = 0; + v[3] = 0xff; +} + +static INLINE void insert_4ub_4f_argb_4( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[2]); + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[3]); +} + +static INLINE void insert_4ub_4f_argb_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[2]); + v[0] = 0xff; +} + +static INLINE void insert_4ub_4f_argb_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); + v[3] = 0x00; + v[0] = 0xff; +} + +static INLINE void insert_4ub_4f_argb_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]); + v[2] = 0x00; + v[3] = 0x00; + v[0] = 0xff; +} + +static INLINE void insert_4ub_4f_abgr_4( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[2]); + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[3]); +} + +static INLINE void insert_4ub_4f_abgr_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[2]); + v[0] = 0xff; +} + +static INLINE void insert_4ub_4f_abgr_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]); + v[1] = 0x00; + v[0] = 0xff; +} + +static INLINE void insert_4ub_4f_abgr_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]); + v[2] = 0x00; + v[1] = 0x00; + v[0] = 0xff; +} + +static INLINE void insert_3ub_3f_rgb_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[2]); +} + +static INLINE void insert_3ub_3f_rgb_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + v[2] = 0; +} + +static INLINE void insert_3ub_3f_rgb_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); + v[1] = 0; + v[2] = 0; +} + +static INLINE void insert_3ub_3f_bgr_3( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[2]); +} + +static INLINE void insert_3ub_3f_bgr_2( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]); + v[0] = 0; +} + +static INLINE void insert_3ub_3f_bgr_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]); + v[1] = 0; + v[0] = 0; +} + + +static INLINE void insert_1ub_1f_1( const struct draw_vf_attr *a, uint8_t *v, + const float *in ) +{ + (void) a; + UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]); +} + + +/*********************************************************************** + * Functions to perform the reverse operations to the above, for + * swrast translation and clip-interpolation. + * + * Currently always extracts a full 4 floats. + */ + +static void extract_4f_viewport( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + const float *in = (const float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + /* Although included for completeness, the position coordinate is + * usually handled differently during clipping. + */ + out[0] = (in[0] - trans[0]) / scale[0]; + out[1] = (in[1] - trans[1]) / scale[1]; + out[2] = (in[2] - trans[2]) / scale[2]; + out[3] = in[3]; +} + +static void extract_3f_viewport( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + const float *in = (const float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = (in[0] - trans[0]) / scale[0]; + out[1] = (in[1] - trans[1]) / scale[1]; + out[2] = (in[2] - trans[2]) / scale[2]; + out[3] = 1; +} + + +static void extract_2f_viewport( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + const float *in = (const float *)v; + const float *scale = a->vf->vp; + const float *trans = a->vf->vp + 4; + + out[0] = (in[0] - trans[0]) / scale[0]; + out[1] = (in[1] - trans[1]) / scale[1]; + out[2] = 0; + out[3] = 1; +} + + +static void extract_4f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) +{ + const float *in = (const float *)v; + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = in[2]; + out[3] = in[3]; +} + +static void extract_3f_xyw( const struct draw_vf_attr *a, float *out, const uint8_t *v ) +{ + const float *in = (const float *)v; + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = 0; + out[3] = in[2]; +} + + +static void extract_3f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) +{ + const float *in = (const float *)v; + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = in[2]; + out[3] = 1; +} + + +static void extract_2f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) +{ + const float *in = (const float *)v; + (void) a; + + out[0] = in[0]; + out[1] = in[1]; + out[2] = 0; + out[3] = 1; +} + +static void extract_1f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) +{ + const float *in = (const float *)v; + (void) a; + + out[0] = in[0]; + out[1] = 0; + out[2] = 0; + out[3] = 1; +} + +static void extract_4chan_4f_rgba( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + GLchan *c = (GLchan *)v; + (void) a; + + out[0] = CHAN_TO_FLOAT(c[0]); + out[1] = CHAN_TO_FLOAT(c[1]); + out[2] = CHAN_TO_FLOAT(c[2]); + out[3] = CHAN_TO_FLOAT(c[3]); +} + +static void extract_4ub_4f_rgba( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + (void) a; + out[0] = UBYTE_TO_FLOAT(v[0]); + out[1] = UBYTE_TO_FLOAT(v[1]); + out[2] = UBYTE_TO_FLOAT(v[2]); + out[3] = UBYTE_TO_FLOAT(v[3]); +} + +static void extract_4ub_4f_bgra( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + (void) a; + out[2] = UBYTE_TO_FLOAT(v[0]); + out[1] = UBYTE_TO_FLOAT(v[1]); + out[0] = UBYTE_TO_FLOAT(v[2]); + out[3] = UBYTE_TO_FLOAT(v[3]); +} + +static void extract_4ub_4f_argb( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + (void) a; + out[3] = UBYTE_TO_FLOAT(v[0]); + out[0] = UBYTE_TO_FLOAT(v[1]); + out[1] = UBYTE_TO_FLOAT(v[2]); + out[2] = UBYTE_TO_FLOAT(v[3]); +} + +static void extract_4ub_4f_abgr( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + (void) a; + out[3] = UBYTE_TO_FLOAT(v[0]); + out[2] = UBYTE_TO_FLOAT(v[1]); + out[1] = UBYTE_TO_FLOAT(v[2]); + out[0] = UBYTE_TO_FLOAT(v[3]); +} + +static void extract_3ub_3f_rgb( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + (void) a; + out[0] = UBYTE_TO_FLOAT(v[0]); + out[1] = UBYTE_TO_FLOAT(v[1]); + out[2] = UBYTE_TO_FLOAT(v[2]); + out[3] = 1; +} + +static void extract_3ub_3f_bgr( const struct draw_vf_attr *a, float *out, + const uint8_t *v ) +{ + (void) a; + out[2] = UBYTE_TO_FLOAT(v[0]); + out[1] = UBYTE_TO_FLOAT(v[1]); + out[0] = UBYTE_TO_FLOAT(v[2]); + out[3] = 1; +} + +static void extract_1ub_1f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) +{ + (void) a; + out[0] = UBYTE_TO_FLOAT(v[0]); + out[1] = 0; + out[2] = 0; + out[3] = 1; +} + + +const struct draw_vf_format_info draw_vf_format_info[EMIT_MAX] = +{ + { "1f", + extract_1f, + { insert_1f_1, insert_1f_1, insert_1f_1, insert_1f_1 }, + sizeof(float) }, + + { "2f", + extract_2f, + { insert_2f_1, insert_2f_2, insert_2f_2, insert_2f_2 }, + 2 * sizeof(float) }, + + { "3f", + extract_3f, + { insert_3f_1, insert_3f_2, insert_3f_3, insert_3f_3 }, + 3 * sizeof(float) }, + + { "4f", + extract_4f, + { insert_4f_1, insert_4f_2, insert_4f_3, insert_4f_4 }, + 4 * sizeof(float) }, + + { "2f_viewport", + extract_2f_viewport, + { insert_2f_viewport_1, insert_2f_viewport_2, insert_2f_viewport_2, + insert_2f_viewport_2 }, + 2 * sizeof(float) }, + + { "3f_viewport", + extract_3f_viewport, + { insert_3f_viewport_1, insert_3f_viewport_2, insert_3f_viewport_3, + insert_3f_viewport_3 }, + 3 * sizeof(float) }, + + { "4f_viewport", + extract_4f_viewport, + { insert_4f_viewport_1, insert_4f_viewport_2, insert_4f_viewport_3, + insert_4f_viewport_4 }, + 4 * sizeof(float) }, + + { "3f_xyw", + extract_3f_xyw, + { insert_3f_xyw_err, insert_3f_xyw_err, insert_3f_xyw_err, + insert_3f_xyw_4 }, + 3 * sizeof(float) }, + + { "1ub_1f", + extract_1ub_1f, + { insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1 }, + sizeof(uint8_t) }, + + { "3ub_3f_rgb", + extract_3ub_3f_rgb, + { insert_3ub_3f_rgb_1, insert_3ub_3f_rgb_2, insert_3ub_3f_rgb_3, + insert_3ub_3f_rgb_3 }, + 3 * sizeof(uint8_t) }, + + { "3ub_3f_bgr", + extract_3ub_3f_bgr, + { insert_3ub_3f_bgr_1, insert_3ub_3f_bgr_2, insert_3ub_3f_bgr_3, + insert_3ub_3f_bgr_3 }, + 3 * sizeof(uint8_t) }, + + { "4ub_4f_rgba", + extract_4ub_4f_rgba, + { insert_4ub_4f_rgba_1, insert_4ub_4f_rgba_2, insert_4ub_4f_rgba_3, + insert_4ub_4f_rgba_4 }, + 4 * sizeof(uint8_t) }, + + { "4ub_4f_bgra", + extract_4ub_4f_bgra, + { insert_4ub_4f_bgra_1, insert_4ub_4f_bgra_2, insert_4ub_4f_bgra_3, + insert_4ub_4f_bgra_4 }, + 4 * sizeof(uint8_t) }, + + { "4ub_4f_argb", + extract_4ub_4f_argb, + { insert_4ub_4f_argb_1, insert_4ub_4f_argb_2, insert_4ub_4f_argb_3, + insert_4ub_4f_argb_4 }, + 4 * sizeof(uint8_t) }, + + { "4ub_4f_abgr", + extract_4ub_4f_abgr, + { insert_4ub_4f_abgr_1, insert_4ub_4f_abgr_2, insert_4ub_4f_abgr_3, + insert_4ub_4f_abgr_4 }, + 4 * sizeof(uint8_t) }, + + { "4chan_4f_rgba", + extract_4chan_4f_rgba, + { insert_4chan_4f_rgba_1, insert_4chan_4f_rgba_2, insert_4chan_4f_rgba_3, + insert_4chan_4f_rgba_4 }, + 4 * sizeof(GLchan) }, + + { "pad", + NULL, + { NULL, NULL, NULL, NULL }, + 0 } + +}; + + + + +/*********************************************************************** + * Hardwired fastpaths for emitting whole vertices or groups of + * vertices + */ +#define EMIT5(NR, F0, F1, F2, F3, F4, NAME) \ +static void NAME( struct draw_vertex_fetch *vf, \ + unsigned count, \ + uint8_t *v ) \ +{ \ + struct draw_vf_attr *a = vf->attr; \ + unsigned i; \ + \ + for (i = 0 ; i < count ; i++, v += vf->vertex_stride) { \ + if (NR > 0) { \ + F0( &a[0], v + a[0].vertoffset, (float *)a[0].inputptr ); \ + a[0].inputptr += a[0].inputstride; \ + } \ + \ + if (NR > 1) { \ + F1( &a[1], v + a[1].vertoffset, (float *)a[1].inputptr ); \ + a[1].inputptr += a[1].inputstride; \ + } \ + \ + if (NR > 2) { \ + F2( &a[2], v + a[2].vertoffset, (float *)a[2].inputptr ); \ + a[2].inputptr += a[2].inputstride; \ + } \ + \ + if (NR > 3) { \ + F3( &a[3], v + a[3].vertoffset, (float *)a[3].inputptr ); \ + a[3].inputptr += a[3].inputstride; \ + } \ + \ + if (NR > 4) { \ + F4( &a[4], v + a[4].vertoffset, (float *)a[4].inputptr ); \ + a[4].inputptr += a[4].inputstride; \ + } \ + } \ +} + + +#define EMIT2(F0, F1, NAME) EMIT5(2, F0, F1, insert_null, \ + insert_null, insert_null, NAME) + +#define EMIT3(F0, F1, F2, NAME) EMIT5(3, F0, F1, F2, insert_null, \ + insert_null, NAME) + +#define EMIT4(F0, F1, F2, F3, NAME) EMIT5(4, F0, F1, F2, F3, \ + insert_null, NAME) + + +EMIT2(insert_3f_viewport_3, insert_4ub_4f_rgba_4, emit_viewport3_rgba4) +EMIT2(insert_3f_viewport_3, insert_4ub_4f_bgra_4, emit_viewport3_bgra4) +EMIT2(insert_3f_3, insert_4ub_4f_rgba_4, emit_xyz3_rgba4) + +EMIT3(insert_4f_viewport_4, insert_4ub_4f_rgba_4, insert_2f_2, emit_viewport4_rgba4_st2) +EMIT3(insert_4f_viewport_4, insert_4ub_4f_bgra_4, insert_2f_2, emit_viewport4_bgra4_st2) +EMIT3(insert_4f_4, insert_4ub_4f_rgba_4, insert_2f_2, emit_xyzw4_rgba4_st2) + +EMIT4(insert_4f_viewport_4, insert_4ub_4f_rgba_4, insert_2f_2, insert_2f_2, emit_viewport4_rgba4_st2_st2) +EMIT4(insert_4f_viewport_4, insert_4ub_4f_bgra_4, insert_2f_2, insert_2f_2, emit_viewport4_bgra4_st2_st2) +EMIT4(insert_4f_4, insert_4ub_4f_rgba_4, insert_2f_2, insert_2f_2, emit_xyzw4_rgba4_st2_st2) + + +/* Use the codegen paths to select one of a number of hardwired + * fastpaths. + */ +void draw_vf_generate_hardwired_emit( struct draw_vertex_fetch *vf ) +{ + draw_vf_emit_func func = NULL; + + /* Does it fit a hardwired fastpath? Help! this is growing out of + * control! + */ + switch (vf->attr_count) { + case 2: + if (vf->attr[0].do_insert == insert_3f_viewport_3) { + if (vf->attr[1].do_insert == insert_4ub_4f_bgra_4) + func = emit_viewport3_bgra4; + else if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) + func = emit_viewport3_rgba4; + } + else if (vf->attr[0].do_insert == insert_3f_3 && + vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { + func = emit_xyz3_rgba4; + } + break; + case 3: + if (vf->attr[2].do_insert == insert_2f_2) { + if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { + if (vf->attr[0].do_insert == insert_4f_viewport_4) + func = emit_viewport4_rgba4_st2; + else if (vf->attr[0].do_insert == insert_4f_4) + func = emit_xyzw4_rgba4_st2; + } + else if (vf->attr[1].do_insert == insert_4ub_4f_bgra_4 && + vf->attr[0].do_insert == insert_4f_viewport_4) + func = emit_viewport4_bgra4_st2; + } + break; + case 4: + if (vf->attr[2].do_insert == insert_2f_2 && + vf->attr[3].do_insert == insert_2f_2) { + if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { + if (vf->attr[0].do_insert == insert_4f_viewport_4) + func = emit_viewport4_rgba4_st2_st2; + else if (vf->attr[0].do_insert == insert_4f_4) + func = emit_xyzw4_rgba4_st2_st2; + } + else if (vf->attr[1].do_insert == insert_4ub_4f_bgra_4 && + vf->attr[0].do_insert == insert_4f_viewport_4) + func = emit_viewport4_bgra4_st2_st2; + } + break; + } + + vf->emit = func; +} + +/*********************************************************************** + * Generic (non-codegen) functions for whole vertices or groups of + * vertices + */ + +void draw_vf_generic_emit( struct draw_vertex_fetch *vf, + unsigned count, + uint8_t *v ) +{ + struct draw_vf_attr *a = vf->attr; + const unsigned attr_count = vf->attr_count; + const unsigned stride = vf->vertex_stride; + unsigned i, j; + + for (i = 0 ; i < count ; i++, v += stride) { + for (j = 0; j < attr_count; j++) { + float *in = (float *)a[j].inputptr; + a[j].inputptr += a[j].inputstride; + a[j].do_insert( &a[j], v + a[j].vertoffset, in ); + } + } +} + + diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c new file mode 100644 index 00000000000..2cf3a45ff95 --- /dev/null +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -0,0 +1,664 @@ +/* + * Copyright 2003 Tungsten Graphics, inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + */ + +#include "glheader.h" +#include "colormac.h" +#include "simple_list.h" +#include "enums.h" + +#include "pipe/p_compiler.h" + +#include "draw_vf.h" + +#if defined(USE_SSE_ASM) + +#include "x86/rtasm/x86sse.h" +#include "x86/common_x86_asm.h" + + +#define X 0 +#define Y 1 +#define Z 2 +#define W 3 + + +struct x86_program { + struct x86_function func; + + struct draw_vertex_fetch *vf; + boolean inputs_safe; + boolean outputs_safe; + boolean have_sse2; + + struct x86_reg identity; + struct x86_reg chan0; +}; + + +static struct x86_reg get_identity( struct x86_program *p ) +{ + return p->identity; +} + +static void emit_load4f_4( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + sse_movups(&p->func, dest, arg0); +} + +static void emit_load4f_3( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + /* Have to jump through some hoops: + * + * c 0 0 0 + * c 0 0 1 + * 0 0 c 1 + * a b c 1 + */ + sse_movss(&p->func, dest, x86_make_disp(arg0, 8)); + sse_shufps(&p->func, dest, get_identity(p), SHUF(X,Y,Z,W) ); + sse_shufps(&p->func, dest, dest, SHUF(Y,Z,X,W) ); + sse_movlps(&p->func, dest, arg0); +} + +static void emit_load4f_2( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + /* Initialize from identity, then pull in low two words: + */ + sse_movups(&p->func, dest, get_identity(p)); + sse_movlps(&p->func, dest, arg0); +} + +static void emit_load4f_1( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + /* Pull in low word, then swizzle in identity */ + sse_movss(&p->func, dest, arg0); + sse_shufps(&p->func, dest, get_identity(p), SHUF(X,Y,Z,W) ); +} + + + +static void emit_load3f_3( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + /* Over-reads by 1 dword - potential SEGV if input is a vertex + * array. + */ + if (p->inputs_safe) { + sse_movups(&p->func, dest, arg0); + } + else { + /* c 0 0 0 + * c c c c + * a b c c + */ + sse_movss(&p->func, dest, x86_make_disp(arg0, 8)); + sse_shufps(&p->func, dest, dest, SHUF(X,X,X,X)); + sse_movlps(&p->func, dest, arg0); + } +} + +static void emit_load3f_2( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + emit_load4f_2(p, dest, arg0); +} + +static void emit_load3f_1( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + emit_load4f_1(p, dest, arg0); +} + +static void emit_load2f_2( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + sse_movlps(&p->func, dest, arg0); +} + +static void emit_load2f_1( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + emit_load4f_1(p, dest, arg0); +} + +static void emit_load1f_1( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + sse_movss(&p->func, dest, arg0); +} + +static void (*load[4][4])( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) = { + { emit_load1f_1, + emit_load1f_1, + emit_load1f_1, + emit_load1f_1 }, + + { emit_load2f_1, + emit_load2f_2, + emit_load2f_2, + emit_load2f_2 }, + + { emit_load3f_1, + emit_load3f_2, + emit_load3f_3, + emit_load3f_3 }, + + { emit_load4f_1, + emit_load4f_2, + emit_load4f_3, + emit_load4f_4 } +}; + +static void emit_load( struct x86_program *p, + struct x86_reg dest, + unsigned sz, + struct x86_reg src, + unsigned src_sz) +{ + load[sz-1][src_sz-1](p, dest, src); +} + +static void emit_store4f( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + sse_movups(&p->func, dest, arg0); +} + +static void emit_store3f( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + if (p->outputs_safe) { + /* Emit the extra dword anyway. This may hurt writecombining, + * may cause other problems. + */ + sse_movups(&p->func, dest, arg0); + } + else { + /* Alternate strategy - emit two, shuffle, emit one. + */ + sse_movlps(&p->func, dest, arg0); + sse_shufps(&p->func, arg0, arg0, SHUF(Z,Z,Z,Z) ); /* NOTE! destructive */ + sse_movss(&p->func, x86_make_disp(dest,8), arg0); + } +} + +static void emit_store2f( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + sse_movlps(&p->func, dest, arg0); +} + +static void emit_store1f( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) +{ + sse_movss(&p->func, dest, arg0); +} + + +static void (*store[4])( struct x86_program *p, + struct x86_reg dest, + struct x86_reg arg0 ) = +{ + emit_store1f, + emit_store2f, + emit_store3f, + emit_store4f +}; + +static void emit_store( struct x86_program *p, + struct x86_reg dest, + unsigned sz, + struct x86_reg temp ) + +{ + store[sz-1](p, dest, temp); +} + +static void emit_pack_store_4ub( struct x86_program *p, + struct x86_reg dest, + struct x86_reg temp ) +{ + /* Scale by 255.0 + */ + sse_mulps(&p->func, temp, p->chan0); + + if (p->have_sse2) { + sse2_cvtps2dq(&p->func, temp, temp); + sse2_packssdw(&p->func, temp, temp); + sse2_packuswb(&p->func, temp, temp); + sse_movss(&p->func, dest, temp); + } + else { + struct x86_reg mmx0 = x86_make_reg(file_MMX, 0); + struct x86_reg mmx1 = x86_make_reg(file_MMX, 1); + sse_cvtps2pi(&p->func, mmx0, temp); + sse_movhlps(&p->func, temp, temp); + sse_cvtps2pi(&p->func, mmx1, temp); + mmx_packssdw(&p->func, mmx0, mmx1); + mmx_packuswb(&p->func, mmx0, mmx0); + mmx_movd(&p->func, dest, mmx0); + } +} + +static int get_offset( const void *a, const void *b ) +{ + return (const char *)b - (const char *)a; +} + +/* Not much happens here. Eventually use this function to try and + * avoid saving/reloading the source pointers each vertex (if some of + * them can fit in registers). + */ +static void get_src_ptr( struct x86_program *p, + struct x86_reg srcREG, + struct x86_reg vfREG, + struct draw_vf_attr *a ) +{ + struct draw_vertex_fetch *vf = p->vf; + struct x86_reg ptr_to_src = x86_make_disp(vfREG, get_offset(vf, &a->inputptr)); + + /* Load current a[j].inputptr + */ + x86_mov(&p->func, srcREG, ptr_to_src); +} + +static void update_src_ptr( struct x86_program *p, + struct x86_reg srcREG, + struct x86_reg vfREG, + struct draw_vf_attr *a ) +{ + if (a->inputstride) { + struct draw_vertex_fetch *vf = p->vf; + struct x86_reg ptr_to_src = x86_make_disp(vfREG, get_offset(vf, &a->inputptr)); + + /* add a[j].inputstride (hardcoded value - could just as easily + * pull the stride value from memory each time). + */ + x86_lea(&p->func, srcREG, x86_make_disp(srcREG, a->inputstride)); + + /* save new value of a[j].inputptr + */ + x86_mov(&p->func, ptr_to_src, srcREG); + } +} + + +/* Lots of hardcoding + * + * EAX -- pointer to current output vertex + * ECX -- pointer to current attribute + * + */ +static boolean build_vertex_emit( struct x86_program *p ) +{ + struct draw_vertex_fetch *vf = p->vf; + unsigned j = 0; + + struct x86_reg vertexEAX = x86_make_reg(file_REG32, reg_AX); + struct x86_reg srcECX = x86_make_reg(file_REG32, reg_CX); + struct x86_reg countEBP = x86_make_reg(file_REG32, reg_BP); + struct x86_reg vfESI = x86_make_reg(file_REG32, reg_SI); + struct x86_reg temp = x86_make_reg(file_XMM, 0); + struct x86_reg vp0 = x86_make_reg(file_XMM, 1); + struct x86_reg vp1 = x86_make_reg(file_XMM, 2); + uint8_t *fixup, *label; + + /* Push a few regs? + */ + x86_push(&p->func, countEBP); + x86_push(&p->func, vfESI); + + + /* Get vertex count, compare to zero + */ + x86_xor(&p->func, srcECX, srcECX); + x86_mov(&p->func, countEBP, x86_fn_arg(&p->func, 2)); + x86_cmp(&p->func, countEBP, srcECX); + fixup = x86_jcc_forward(&p->func, cc_E); + + /* Initialize destination register. + */ + x86_mov(&p->func, vertexEAX, x86_fn_arg(&p->func, 3)); + + /* Move argument 1 (vf) into a reg: + */ + x86_mov(&p->func, vfESI, x86_fn_arg(&p->func, 1)); + + + /* Possibly load vp0, vp1 for viewport calcs: + */ + if (vf->allow_viewport_emits) { + sse_movups(&p->func, vp0, x86_make_disp(vfESI, get_offset(vf, &vf->vp[0]))); + sse_movups(&p->func, vp1, x86_make_disp(vfESI, get_offset(vf, &vf->vp[4]))); + } + + /* always load, needed or not: + */ + sse_movups(&p->func, p->chan0, x86_make_disp(vfESI, get_offset(vf, &vf->chan_scale[0]))); + sse_movups(&p->func, p->identity, x86_make_disp(vfESI, get_offset(vf, &vf->identity[0]))); + + /* Note address for loop jump */ + label = x86_get_label(&p->func); + + /* Emit code for each of the attributes. Currently routes + * everything through SSE registers, even when it might be more + * efficient to stick with regular old x86. No optimization or + * other tricks - enough new ground to cover here just getting + * things working. + */ + while (j < vf->attr_count) { + struct draw_vf_attr *a = &vf->attr[j]; + struct x86_reg dest = x86_make_disp(vertexEAX, a->vertoffset); + + /* Now, load an XMM reg from src, perhaps transform, then save. + * Could be shortcircuited in specific cases: + */ + switch (a->format) { + case EMIT_1F: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 1, x86_deref(srcECX), a->inputsize); + emit_store(p, dest, 1, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_2F: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); + emit_store(p, dest, 2, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_3F: + /* Potentially the worst case - hardcode 2+1 copying: + */ + if (0) { + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); + emit_store(p, dest, 3, temp); + update_src_ptr(p, srcECX, vfESI, a); + } + else { + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); + emit_store(p, dest, 2, temp); + if (a->inputsize > 2) { + emit_load(p, temp, 1, x86_make_disp(srcECX, 8), 1); + emit_store(p, x86_make_disp(dest,8), 1, temp); + } + else { + sse_movss(&p->func, x86_make_disp(dest,8), get_identity(p)); + } + update_src_ptr(p, srcECX, vfESI, a); + } + break; + case EMIT_4F: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + emit_store(p, dest, 4, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_2F_VIEWPORT: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); + sse_mulps(&p->func, temp, vp0); + sse_addps(&p->func, temp, vp1); + emit_store(p, dest, 2, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_3F_VIEWPORT: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); + sse_mulps(&p->func, temp, vp0); + sse_addps(&p->func, temp, vp1); + emit_store(p, dest, 3, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_4F_VIEWPORT: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + sse_mulps(&p->func, temp, vp0); + sse_addps(&p->func, temp, vp1); + emit_store(p, dest, 4, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_3F_XYW: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + sse_shufps(&p->func, temp, temp, SHUF(X,Y,W,Z)); + emit_store(p, dest, 3, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + + case EMIT_1UB_1F: + /* Test for PAD3 + 1UB: + */ + if (j > 0 && + a[-1].vertoffset + a[-1].vertattrsize <= a->vertoffset - 3) + { + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 1, x86_deref(srcECX), a->inputsize); + sse_shufps(&p->func, temp, temp, SHUF(X,X,X,X)); + emit_pack_store_4ub(p, x86_make_disp(dest, -3), temp); /* overkill! */ + update_src_ptr(p, srcECX, vfESI, a); + } + else { + _mesa_printf("Can't emit 1ub %x %x %d\n", a->vertoffset, a[-1].vertoffset, a[-1].vertattrsize ); + return FALSE; + } + break; + case EMIT_3UB_3F_RGB: + case EMIT_3UB_3F_BGR: + /* Test for 3UB + PAD1: + */ + if (j == vf->attr_count - 1 || + a[1].vertoffset >= a->vertoffset + 4) { + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); + if (a->format == EMIT_3UB_3F_BGR) + sse_shufps(&p->func, temp, temp, SHUF(Z,Y,X,W)); + emit_pack_store_4ub(p, dest, temp); + update_src_ptr(p, srcECX, vfESI, a); + } + /* Test for 3UB + 1UB: + */ + else if (j < vf->attr_count - 1 && + a[1].format == EMIT_1UB_1F && + a[1].vertoffset == a->vertoffset + 3) { + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); + update_src_ptr(p, srcECX, vfESI, a); + + /* Make room for incoming value: + */ + sse_shufps(&p->func, temp, temp, SHUF(W,X,Y,Z)); + + get_src_ptr(p, srcECX, vfESI, &a[1]); + emit_load(p, temp, 1, x86_deref(srcECX), a[1].inputsize); + update_src_ptr(p, srcECX, vfESI, &a[1]); + + /* Rearrange and possibly do BGR conversion: + */ + if (a->format == EMIT_3UB_3F_BGR) + sse_shufps(&p->func, temp, temp, SHUF(W,Z,Y,X)); + else + sse_shufps(&p->func, temp, temp, SHUF(Y,Z,W,X)); + + emit_pack_store_4ub(p, dest, temp); + j++; /* NOTE: two attrs consumed */ + } + else { + _mesa_printf("Can't emit 3ub\n"); + } + return FALSE; /* add this later */ + break; + + case EMIT_4UB_4F_RGBA: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + emit_pack_store_4ub(p, dest, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_4UB_4F_BGRA: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + sse_shufps(&p->func, temp, temp, SHUF(Z,Y,X,W)); + emit_pack_store_4ub(p, dest, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_4UB_4F_ARGB: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + sse_shufps(&p->func, temp, temp, SHUF(W,X,Y,Z)); + emit_pack_store_4ub(p, dest, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_4UB_4F_ABGR: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + sse_shufps(&p->func, temp, temp, SHUF(W,Z,Y,X)); + emit_pack_store_4ub(p, dest, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case EMIT_4CHAN_4F_RGBA: + switch (CHAN_TYPE) { + case GL_UNSIGNED_BYTE: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + emit_pack_store_4ub(p, dest, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case GL_FLOAT: + get_src_ptr(p, srcECX, vfESI, a); + emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); + emit_store(p, dest, 4, temp); + update_src_ptr(p, srcECX, vfESI, a); + break; + case GL_UNSIGNED_SHORT: + default: + _mesa_printf("unknown CHAN_TYPE %s\n", _mesa_lookup_enum_by_nr(CHAN_TYPE)); + return FALSE; + } + break; + default: + _mesa_printf("unknown a[%d].format %d\n", j, a->format); + return FALSE; /* catch any new opcodes */ + } + + /* Increment j by at least 1 - may have been incremented above also: + */ + j++; + } + + /* Next vertex: + */ + x86_lea(&p->func, vertexEAX, x86_make_disp(vertexEAX, vf->vertex_stride)); + + /* decr count, loop if not zero + */ + x86_dec(&p->func, countEBP); + x86_test(&p->func, countEBP, countEBP); + x86_jcc(&p->func, cc_NZ, label); + + /* Exit mmx state? + */ + if (p->func.need_emms) + mmx_emms(&p->func); + + /* Land forward jump here: + */ + x86_fixup_fwd_jump(&p->func, fixup); + + /* Pop regs and return + */ + x86_pop(&p->func, x86_get_base_reg(vfESI)); + x86_pop(&p->func, countEBP); + x86_ret(&p->func); + + vf->emit = (draw_vf_emit_func)x86_get_func(&p->func); + return TRUE; +} + + + +void draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ) +{ + struct x86_program p; + + if (!cpu_has_xmm) { + vf->codegen_emit = NULL; + return; + } + + _mesa_memset(&p, 0, sizeof(p)); + + p.vf = vf; + p.inputs_safe = 0; /* for now */ + p.outputs_safe = 1; /* for now */ + p.have_sse2 = cpu_has_xmm2; + p.identity = x86_make_reg(file_XMM, 6); + p.chan0 = x86_make_reg(file_XMM, 7); + + x86_init_func(&p.func); + + if (build_vertex_emit(&p)) { + draw_vf_register_fastpath( vf, TRUE ); + } + else { + /* Note the failure so that we don't keep trying to codegen an + * impossible state: + */ + draw_vf_register_fastpath( vf, FALSE ); + x86_release_func(&p.func); + } +} + +#else + +void draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ) +{ + /* Dummy version for when USE_SSE_ASM not defined */ +} + +#endif diff --git a/src/mesa/sources b/src/mesa/sources index 97ef7e1936d..e31d8cc4667 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -175,6 +175,9 @@ DRAW_SOURCES = \ pipe/draw/draw_vertex_fetch.c \ pipe/draw/draw_vertex_shader.c \ pipe/draw/draw_vertex_shader_llvm.c \ + pipe/draw/draw_vf.c \ + pipe/draw/draw_vf_generic.c \ + pipe/draw/draw_vf_sse.c \ pipe/draw/draw_wide_prims.c TGSIEXEC_SOURCES = \ -- cgit v1.2.3 From 169a74196fdca320cabd5cde33fda17683cc823d Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Mon, 28 Jan 2008 18:46:21 +0900 Subject: First stab at hooking draw_vbuf & vf. Emit disabled for now. Tested with softpipe. Only one vertex at a time for now (slow). --- src/mesa/pipe/draw/draw_vbuf.c | 183 ++++++++++++++++++++++++++++++++++- src/mesa/pipe/draw/draw_vf.c | 18 +++- src/mesa/pipe/draw/draw_vf.h | 46 +++++---- src/mesa/pipe/draw/draw_vf_generic.c | 2 +- src/mesa/pipe/draw/draw_vf_sse.c | 38 ++++---- 5 files changed, 241 insertions(+), 46 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vbuf.c b/src/mesa/pipe/draw/draw_vbuf.c index 1e260c61568..a3d0b5bca31 100644 --- a/src/mesa/pipe/draw/draw_vbuf.c +++ b/src/mesa/pipe/draw/draw_vbuf.c @@ -35,12 +35,15 @@ #include +#include -#include "pipe/draw/draw_vbuf.h" -#include "pipe/draw/draw_private.h" -#include "pipe/draw/draw_vertex.h" #include "pipe/p_util.h" +#include "draw_vbuf.h" +#include "draw_private.h" +#include "draw_vertex.h" +#include "draw_vf.h" + /** * Vertex buffer emit stage. @@ -55,6 +58,8 @@ struct vbuf_stage { /** Vertex size in bytes */ unsigned vertex_size; + struct draw_vertex_fetch *vf; + /* FIXME: we have no guarantee that 'unsigned' is 32bit */ /** Vertices in hardware format */ @@ -121,6 +126,7 @@ static INLINE void emit_vertex( struct vbuf_stage *vbuf, struct vertex_header *vertex ) { +#if 0 const struct vertex_info *vinfo = vbuf->vinfo; uint i; @@ -151,9 +157,11 @@ emit_vertex( struct vbuf_stage *vbuf, case EMIT_ALL: /* just copy the whole vertex as-is to the vbuf */ assert(i == 0); + assert(j == 0); memcpy(vbuf->vertex_ptr, vertex, vinfo->size * 4); vbuf->vertex_ptr += vinfo->size; - return; + count += vinfo->size; + break; case EMIT_1F: *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); count++; @@ -192,6 +200,156 @@ emit_vertex( struct vbuf_stage *vbuf, } } assert(count == vinfo->size); +#else + if(vertex->vertex_id != UNDEFINED_VERTEX_ID) { + if(vertex->vertex_id < vbuf->nr_vertices) + return; + else + fprintf(stderr, "Bad vertex id 0x%04x (>= 0x%04x)\n", + vertex->vertex_id, vbuf->nr_vertices); + return; + } + + vertex->vertex_id = vbuf->nr_vertices++; + + draw_vf_set_data(vbuf->vf, vertex->data); + draw_vf_emit_vertices(vbuf->vf, 1, vbuf->vertex_ptr); + + vbuf->vertex_ptr += vbuf->vertex_size/4; +#endif +} + + +static void +vbuf_set_vf_attributes(struct vbuf_stage *vbuf ) +{ + const struct vertex_info *vinfo = vbuf->vinfo; + struct draw_vf_attr_map attrs[PIPE_MAX_SHADER_INPUTS]; + uint i; + uint count = 0; /* for debug/sanity */ + unsigned nr_attrs = 0; + +// fprintf(stderr, "emit vertex %d to %p\n", +// vbuf->nr_vertices, vbuf->vertex_ptr); + +#if 0 + if(vertex->vertex_id != UNDEFINED_VERTEX_ID) { + if(vertex->vertex_id < vbuf->nr_vertices) + return; + else + fprintf(stderr, "Bad vertex id 0x%04x (>= 0x%04x)\n", + vertex->vertex_id, vbuf->nr_vertices); + return; + } +#endif + + for (i = 0; i < vinfo->num_attribs; i++) { + uint j = vinfo->src_index[i]; + switch (vinfo->emit[i]) { + case EMIT_OMIT: + /* no-op */ + break; + case EMIT_ALL: { + /* just copy the whole vertex as-is to the vbuf */ + unsigned k, s = vinfo->size; + assert(i == 0); + assert(j == 0); + /* copy the vertex header */ + /* XXX: we actually don't copy the header, just pad it */ + attrs[nr_attrs].attrib = 0; + attrs[nr_attrs].format = DRAW_EMIT_PAD; + attrs[nr_attrs].offset = offsetof(struct vertex_header, data); + s -= offsetof(struct vertex_header, data)/4; + count += offsetof(struct vertex_header, data)/4; + nr_attrs++; + /* copy the vertex data */ + for(k = 0; k < (s & ~0x3); k += 4) { + attrs[nr_attrs].attrib = k/4; + attrs[nr_attrs].format = DRAW_EMIT_4F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 4; + } + /* tail */ + /* XXX: actually, this shouldn't be needed */ + attrs[nr_attrs].attrib = k/4; + attrs[nr_attrs].offset = 0; + switch(s & 0x3) { + case 0: + break; + case 1: + attrs[nr_attrs].format = DRAW_EMIT_1F; + nr_attrs++; + count += 1; + break; + case 2: + attrs[nr_attrs].format = DRAW_EMIT_2F; + nr_attrs++; + count += 2; + break; + case 3: + attrs[nr_attrs].format = DRAW_EMIT_3F; + nr_attrs++; + count += 3; + break; + } + break; + } + case EMIT_1F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_1F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count++; + break; + case EMIT_1F_PSIZE: + /* FIXME */ + assert(0); + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_PAD; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count++; + break; + case EMIT_2F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_2F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 2; + break; + case EMIT_3F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_3F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 3; + break; + case EMIT_4F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_4F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 4; + break; + case EMIT_4UB: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_4UB_4F_BGRA; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 1; + break; + default: + assert(0); + } + } + + assert(count == vinfo->size); + + draw_vf_set_vertex_attributes(vbuf->vf, + attrs, + nr_attrs, + vbuf->vertex_size); } @@ -269,6 +427,7 @@ vbuf_set_prim( struct vbuf_stage *vbuf, uint newprim ) vbuf->vinfo = vinfo; vbuf->vertex_size = vertex_size; + vbuf_set_vf_attributes(vbuf); if (!vbuf->vertices) vbuf_alloc_vertices(vbuf); @@ -423,7 +582,12 @@ static void vbuf_destroy( struct draw_stage *stage ) { struct vbuf_stage *vbuf = vbuf_stage( stage ); - align_free( vbuf->indices ); + if(vbuf->indices) + align_free( vbuf->indices ); + + if(vbuf->vf) + draw_vf_destroy( vbuf->vf ); + FREE( stage ); } @@ -436,6 +600,9 @@ struct draw_stage *draw_vbuf_stage( struct draw_context *draw, { struct vbuf_stage *vbuf = CALLOC_STRUCT(vbuf_stage); + if(!vbuf) + return NULL; + vbuf->stage.draw = draw; vbuf->stage.point = vbuf_first_point; vbuf->stage.line = vbuf_first_line; @@ -450,11 +617,17 @@ struct draw_stage *draw_vbuf_stage( struct draw_context *draw, vbuf->max_indices = render->max_indices; vbuf->indices = (ushort *) align_malloc( vbuf->max_indices * sizeof(vbuf->indices[0]), 16 ); + if(!vbuf->indices) + vbuf_destroy(&vbuf->stage); vbuf->vertices = NULL; vbuf->vertex_ptr = vbuf->vertices; vbuf->prim = ~0; + vbuf->vf = draw_vf_create(FALSE); + if(!vbuf->vf) + vbuf_destroy(&vbuf->stage); + return &vbuf->stage; } diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index f758460b5f6..675974c6bc2 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -162,7 +162,7 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, for (j = 0, i = 0; i < nr; i++) { const unsigned format = map[i].format; - if (format == EMIT_PAD) { + if (format == DRAW_EMIT_PAD) { if (DBG) _mesa_printf("%d: pad %d, offset %d\n", i, map[i].offset, offset); @@ -261,6 +261,22 @@ void draw_vf_set_sources( struct draw_vertex_fetch *vf, } +/* Set attribute pointers, adjusted for start position: + */ +void draw_vf_set_data( struct draw_vertex_fetch *vf, + float data[][4]) +{ + struct draw_vf_attr *a = vf->attr; + unsigned j; + + for (j = 0; j < vf->attr_count; j++) { + a[j].inputstride = 0; /* XXX: one-vertex-max ATM */ + a[j].inputsize = 4; + a[j].do_insert = a[j].insert[4 - 1]; + a[j].inputptr = (uint8_t *)&data[a[j].attrib][0]; + } +} + /* Emit count VB vertices to dest. */ diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index 279570aad5f..7619c0ee271 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -29,9 +29,11 @@ #define DRAW_VF_H -#include "pipe/p_compiler.h" #include "math/m_vector.h" +#include "pipe/p_compiler.h" +#include "draw_vertex.h" + enum { DRAW_VF_ATTRIB_POS = 0, @@ -67,24 +69,24 @@ enum { }; enum draw_vf_attr_format { - EMIT_1F, - EMIT_2F, - EMIT_3F, - EMIT_4F, - EMIT_2F_VIEWPORT, /**< do viewport transform and emit */ - EMIT_3F_VIEWPORT, /**< do viewport transform and emit */ - EMIT_4F_VIEWPORT, /**< do viewport transform and emit */ - EMIT_3F_XYW, /**< for projective texture */ - EMIT_1UB_1F, /**< for fog coordinate */ - EMIT_3UB_3F_RGB, /**< for specular color */ - EMIT_3UB_3F_BGR, /**< for specular color */ - EMIT_4UB_4F_RGBA, /**< for color */ - EMIT_4UB_4F_BGRA, /**< for color */ - EMIT_4UB_4F_ARGB, /**< for color */ - EMIT_4UB_4F_ABGR, /**< for color */ - EMIT_4CHAN_4F_RGBA, /**< for swrast color */ - EMIT_PAD, /**< leave a hole of 'offset' bytes */ - EMIT_MAX + DRAW_EMIT_1F, + DRAW_EMIT_2F, + DRAW_EMIT_3F, + DRAW_EMIT_4F, + DRAW_EMIT_2F_VIEWPORT, /**< do viewport transform and emit */ + DRAW_EMIT_3F_VIEWPORT, /**< do viewport transform and emit */ + DRAW_EMIT_4F_VIEWPORT, /**< do viewport transform and emit */ + DRAW_EMIT_3F_XYW, /**< for projective texture */ + DRAW_EMIT_1UB_1F, /**< for fog coordinate */ + DRAW_EMIT_3UB_3F_RGB, /**< for specular color */ + DRAW_EMIT_3UB_3F_BGR, /**< for specular color */ + DRAW_EMIT_4UB_4F_RGBA, /**< for color */ + DRAW_EMIT_4UB_4F_BGRA, /**< for color */ + DRAW_EMIT_4UB_4F_ARGB, /**< for color */ + DRAW_EMIT_4UB_4F_ABGR, /**< for color */ + DRAW_EMIT_4CHAN_4F_RGBA, /**< for swrast color */ + DRAW_EMIT_PAD, /**< leave a hole of 'offset' bytes */ + DRAW_EMIT_MAX }; struct draw_vf_attr_map { @@ -116,6 +118,10 @@ draw_vf_set_sources( struct draw_vertex_fetch *vf, GLvector4f * const attrib[], unsigned start ); +void +draw_vf_set_data( struct draw_vertex_fetch *vf, + float data[][4]); + void draw_vf_emit_vertices( struct draw_vertex_fetch *vf, unsigned count, @@ -243,7 +249,7 @@ struct draw_vf_format_info { const unsigned attrsize; }; -const struct draw_vf_format_info draw_vf_format_info[EMIT_MAX]; +const struct draw_vf_format_info draw_vf_format_info[DRAW_EMIT_MAX]; #endif diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c index 19e6c587e5e..42effc0c658 100644 --- a/src/mesa/pipe/draw/draw_vf_generic.c +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -735,7 +735,7 @@ static void extract_1ub_1f( const struct draw_vf_attr *a, float *out, const uint } -const struct draw_vf_format_info draw_vf_format_info[EMIT_MAX] = +const struct draw_vf_format_info draw_vf_format_info[DRAW_EMIT_MAX] = { { "1f", extract_1f, diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c index 2cf3a45ff95..a7019a47e64 100644 --- a/src/mesa/pipe/draw/draw_vf_sse.c +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -398,19 +398,19 @@ static boolean build_vertex_emit( struct x86_program *p ) * Could be shortcircuited in specific cases: */ switch (a->format) { - case EMIT_1F: + case DRAW_EMIT_1F: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 1, x86_deref(srcECX), a->inputsize); emit_store(p, dest, 1, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_2F: + case DRAW_EMIT_2F: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); emit_store(p, dest, 2, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_3F: + case DRAW_EMIT_3F: /* Potentially the worst case - hardcode 2+1 copying: */ if (0) { @@ -433,13 +433,13 @@ static boolean build_vertex_emit( struct x86_program *p ) update_src_ptr(p, srcECX, vfESI, a); } break; - case EMIT_4F: + case DRAW_EMIT_4F: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); emit_store(p, dest, 4, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_2F_VIEWPORT: + case DRAW_EMIT_2F_VIEWPORT: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); sse_mulps(&p->func, temp, vp0); @@ -447,7 +447,7 @@ static boolean build_vertex_emit( struct x86_program *p ) emit_store(p, dest, 2, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_3F_VIEWPORT: + case DRAW_EMIT_3F_VIEWPORT: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); sse_mulps(&p->func, temp, vp0); @@ -455,7 +455,7 @@ static boolean build_vertex_emit( struct x86_program *p ) emit_store(p, dest, 3, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_4F_VIEWPORT: + case DRAW_EMIT_4F_VIEWPORT: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); sse_mulps(&p->func, temp, vp0); @@ -463,7 +463,7 @@ static boolean build_vertex_emit( struct x86_program *p ) emit_store(p, dest, 4, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_3F_XYW: + case DRAW_EMIT_3F_XYW: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); sse_shufps(&p->func, temp, temp, SHUF(X,Y,W,Z)); @@ -471,7 +471,7 @@ static boolean build_vertex_emit( struct x86_program *p ) update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_1UB_1F: + case DRAW_EMIT_1UB_1F: /* Test for PAD3 + 1UB: */ if (j > 0 && @@ -488,15 +488,15 @@ static boolean build_vertex_emit( struct x86_program *p ) return FALSE; } break; - case EMIT_3UB_3F_RGB: - case EMIT_3UB_3F_BGR: + case DRAW_EMIT_3UB_3F_RGB: + case DRAW_EMIT_3UB_3F_BGR: /* Test for 3UB + PAD1: */ if (j == vf->attr_count - 1 || a[1].vertoffset >= a->vertoffset + 4) { get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); - if (a->format == EMIT_3UB_3F_BGR) + if (a->format == DRAW_EMIT_3UB_3F_BGR) sse_shufps(&p->func, temp, temp, SHUF(Z,Y,X,W)); emit_pack_store_4ub(p, dest, temp); update_src_ptr(p, srcECX, vfESI, a); @@ -504,7 +504,7 @@ static boolean build_vertex_emit( struct x86_program *p ) /* Test for 3UB + 1UB: */ else if (j < vf->attr_count - 1 && - a[1].format == EMIT_1UB_1F && + a[1].format == DRAW_EMIT_1UB_1F && a[1].vertoffset == a->vertoffset + 3) { get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); @@ -520,7 +520,7 @@ static boolean build_vertex_emit( struct x86_program *p ) /* Rearrange and possibly do BGR conversion: */ - if (a->format == EMIT_3UB_3F_BGR) + if (a->format == DRAW_EMIT_3UB_3F_BGR) sse_shufps(&p->func, temp, temp, SHUF(W,Z,Y,X)); else sse_shufps(&p->func, temp, temp, SHUF(Y,Z,W,X)); @@ -534,34 +534,34 @@ static boolean build_vertex_emit( struct x86_program *p ) return FALSE; /* add this later */ break; - case EMIT_4UB_4F_RGBA: + case DRAW_EMIT_4UB_4F_RGBA: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); emit_pack_store_4ub(p, dest, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_4UB_4F_BGRA: + case DRAW_EMIT_4UB_4F_BGRA: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); sse_shufps(&p->func, temp, temp, SHUF(Z,Y,X,W)); emit_pack_store_4ub(p, dest, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_4UB_4F_ARGB: + case DRAW_EMIT_4UB_4F_ARGB: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); sse_shufps(&p->func, temp, temp, SHUF(W,X,Y,Z)); emit_pack_store_4ub(p, dest, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_4UB_4F_ABGR: + case DRAW_EMIT_4UB_4F_ABGR: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); sse_shufps(&p->func, temp, temp, SHUF(W,Z,Y,X)); emit_pack_store_4ub(p, dest, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case EMIT_4CHAN_4F_RGBA: + case DRAW_EMIT_4CHAN_4F_RGBA: switch (CHAN_TYPE) { case GL_UNSIGNED_BYTE: get_src_ptr(p, srcECX, vfESI, a); -- cgit v1.2.3 From 25d2ffc6697fcd60edc9596f778d8901083f7755 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 10:14:13 +0000 Subject: gallium: remove dead code from draw_vf* --- src/mesa/pipe/draw/Makefile | 2 + src/mesa/pipe/draw/draw_vbuf.c | 2 +- src/mesa/pipe/draw/draw_vf.c | 90 +------- src/mesa/pipe/draw/draw_vf.h | 18 +- src/mesa/pipe/draw/draw_vf_generic.c | 420 +---------------------------------- src/mesa/pipe/draw/draw_vf_sse.c | 51 ----- 6 files changed, 9 insertions(+), 574 deletions(-) create mode 100644 src/mesa/pipe/draw/Makefile diff --git a/src/mesa/pipe/draw/Makefile b/src/mesa/pipe/draw/Makefile new file mode 100644 index 00000000000..451911a3545 --- /dev/null +++ b/src/mesa/pipe/draw/Makefile @@ -0,0 +1,2 @@ +default: + cd .. ; make diff --git a/src/mesa/pipe/draw/draw_vbuf.c b/src/mesa/pipe/draw/draw_vbuf.c index a3d0b5bca31..8ca225c65aa 100644 --- a/src/mesa/pipe/draw/draw_vbuf.c +++ b/src/mesa/pipe/draw/draw_vbuf.c @@ -625,7 +625,7 @@ struct draw_stage *draw_vbuf_stage( struct draw_context *draw, vbuf->prim = ~0; - vbuf->vf = draw_vf_create(FALSE); + vbuf->vf = draw_vf_create(); if(!vbuf->vf) vbuf_destroy(&vbuf->stage); diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 675974c6bc2..deedfc7bc7e 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -177,7 +177,6 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, vf->attr[j].attrib = map[i].attrib; vf->attr[j].format = format; vf->attr[j].insert = draw_vf_format_info[format].insert; - vf->attr[j].extract = draw_vf_format_info[format].extract; vf->attr[j].vertattrsize = draw_vf_format_info[format].attrsize; vf->attr[j].vertoffset = offset; @@ -201,41 +200,6 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, -void draw_vf_set_vp_matrix( struct draw_vertex_fetch *vf, - const float *viewport ) -{ - assert(vf->allow_viewport_emits); - - /* scale */ - vf->vp[0] = viewport[MAT_SX]; - vf->vp[1] = viewport[MAT_SY]; - vf->vp[2] = viewport[MAT_SZ]; - vf->vp[3] = 1.0; - - /* translate */ - vf->vp[4] = viewport[MAT_TX]; - vf->vp[5] = viewport[MAT_TY]; - vf->vp[6] = viewport[MAT_TZ]; - vf->vp[7] = 0.0; -} - -void draw_vf_set_vp_scale_translate( struct draw_vertex_fetch *vf, - const float *scale, - const float *translate ) -{ - assert(vf->allow_viewport_emits); - - vf->vp[0] = scale[0]; - vf->vp[1] = scale[1]; - vf->vp[2] = scale[2]; - vf->vp[3] = scale[3]; - - vf->vp[4] = translate[0]; - vf->vp[5] = translate[1]; - vf->vp[6] = translate[2]; - vf->vp[7] = translate[3]; -} - /* Set attribute pointers, adjusted for start position: */ @@ -288,39 +252,10 @@ void draw_vf_emit_vertices( struct draw_vertex_fetch *vf, } -/* Extract a named attribute from a hardware vertex. Will have to - * reverse any viewport transformation, swizzling or other conversions - * which may have been applied. - * - * This is mainly required for on-the-fly vertex translations to - * swrast format. - */ -void draw_vf_get_attr( struct draw_vertex_fetch *vf, - const void *vertex, - GLenum attr, - const float *dflt, - float *dest ) -{ - const struct draw_vf_attr *a = vf->attr; - const unsigned attr_count = vf->attr_count; - unsigned j; - - for (j = 0; j < attr_count; j++) { - if (a[j].attrib == attr) { - a[j].extract( &a[j], dest, (uint8_t *)vertex + a[j].vertoffset ); - return; - } - } - - /* Else return the value from ctx->Current. - */ - _mesa_memcpy( dest, dflt, 4*sizeof(float)); -} - -struct draw_vertex_fetch *draw_vf_create( boolean allow_viewport_emits ) +struct draw_vertex_fetch *draw_vf_create( void ) { struct draw_vertex_fetch *vf = CALLOC_STRUCT(draw_vertex_fetch); unsigned i; @@ -328,29 +263,6 @@ struct draw_vertex_fetch *draw_vf_create( boolean allow_viewport_emits ) for (i = 0; i < DRAW_VF_ATTRIB_MAX; i++) vf->attr[i].vf = vf; - vf->allow_viewport_emits = allow_viewport_emits; - - switch(CHAN_TYPE) { - case GL_UNSIGNED_BYTE: - vf->chan_scale[0] = 255.0; - vf->chan_scale[1] = 255.0; - vf->chan_scale[2] = 255.0; - vf->chan_scale[3] = 255.0; - break; - case GL_UNSIGNED_SHORT: - vf->chan_scale[0] = 65535.0; - vf->chan_scale[1] = 65535.0; - vf->chan_scale[2] = 65535.0; - vf->chan_scale[3] = 65535.0; - break; - default: - vf->chan_scale[0] = 1.0; - vf->chan_scale[1] = 1.0; - vf->chan_scale[2] = 1.0; - vf->chan_scale[3] = 1.0; - break; - } - vf->identity[0] = 0.0; vf->identity[1] = 0.0; vf->identity[2] = 0.0; diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index 7619c0ee271..c6a8fe0d534 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -73,9 +73,6 @@ enum draw_vf_attr_format { DRAW_EMIT_2F, DRAW_EMIT_3F, DRAW_EMIT_4F, - DRAW_EMIT_2F_VIEWPORT, /**< do viewport transform and emit */ - DRAW_EMIT_3F_VIEWPORT, /**< do viewport transform and emit */ - DRAW_EMIT_4F_VIEWPORT, /**< do viewport transform and emit */ DRAW_EMIT_3F_XYW, /**< for projective texture */ DRAW_EMIT_1UB_1F, /**< for fog coordinate */ DRAW_EMIT_3UB_3F_RGB, /**< for specular color */ @@ -84,7 +81,6 @@ enum draw_vf_attr_format { DRAW_EMIT_4UB_4F_BGRA, /**< for color */ DRAW_EMIT_4UB_4F_ARGB, /**< for color */ DRAW_EMIT_4UB_4F_ABGR, /**< for color */ - DRAW_EMIT_4CHAN_4F_RGBA, /**< for swrast color */ DRAW_EMIT_PAD, /**< leave a hole of 'offset' bytes */ DRAW_EMIT_MAX }; @@ -98,14 +94,6 @@ struct draw_vf_attr_map { struct draw_vertex_fetch; -void -draw_vf_set_vp_matrix( struct draw_vertex_fetch *vf, - const float *viewport ); - -void -draw_vf_set_vp_scale_translate( struct draw_vertex_fetch *vf, - const float *scale, - const float *translate ); unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, @@ -135,7 +123,7 @@ draw_vf_get_attr( struct draw_vertex_fetch *vf, float *dest ); struct draw_vertex_fetch * -draw_vf_create( boolean allow_viewport_emits ); +draw_vf_create( void ); void draw_vf_destroy( struct draw_vertex_fetch *vf ); @@ -196,9 +184,6 @@ struct draw_vertex_fetch /* Parameters and constants for codegen: */ - boolean allow_viewport_emits; - float vp[8]; - float chan_scale[4]; float identity[4]; struct draw_vf_fastpath *fastpath; @@ -244,7 +229,6 @@ draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ); struct draw_vf_format_info { const char *name; - draw_vf_extract_func extract; draw_vf_insert_func insert[4]; const unsigned attrsize; }; diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c index 42effc0c658..343428d26c7 100644 --- a/src/mesa/pipe/draw/draw_vf_generic.c +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -36,125 +36,6 @@ #include "draw_vf.h" -/* - * These functions take the NDC coordinates pointed to by 'in', apply the - * NDC->Viewport mapping and store the results at 'v'. - */ - -static INLINE void insert_4f_viewport_4( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = scale[1] * in[1] + trans[1]; - out[2] = scale[2] * in[2] + trans[2]; - out[3] = in[3]; -} - -static INLINE void insert_4f_viewport_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = scale[1] * in[1] + trans[1]; - out[2] = scale[2] * in[2] + trans[2]; - out[3] = 1; -} - -static INLINE void insert_4f_viewport_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = scale[1] * in[1] + trans[1]; - out[2] = trans[2]; - out[3] = 1; -} - -static INLINE void insert_4f_viewport_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = trans[1]; - out[2] = trans[2]; - out[3] = 1; -} - -static INLINE void insert_3f_viewport_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = scale[1] * in[1] + trans[1]; - out[2] = scale[2] * in[2] + trans[2]; -} - -static INLINE void insert_3f_viewport_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = scale[1] * in[1] + trans[1]; - out[2] = scale[2] * in[2] + trans[2]; -} - -static INLINE void insert_3f_viewport_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = trans[1]; - out[2] = trans[2]; -} - -static INLINE void insert_2f_viewport_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = scale[1] * in[1] + trans[1]; -} - -static INLINE void insert_2f_viewport_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - float *out = (float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = scale[0] * in[0] + trans[0]; - out[1] = trans[1]; -} - - -/* - * These functions do the same as above, except for the viewport mapping. - */ static INLINE void insert_4f_4( const struct draw_vf_attr *a, uint8_t *v, const float *in ) { @@ -278,50 +159,6 @@ static INLINE void insert_null( const struct draw_vf_attr *a, uint8_t *v, const (void) a; (void) v; (void) in; } -static INLINE void insert_4chan_4f_rgba_4( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - GLchan *c = (GLchan *)v; - (void) a; - UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); - UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); - UNCLAMPED_FLOAT_TO_CHAN(c[2], in[2]); - UNCLAMPED_FLOAT_TO_CHAN(c[3], in[3]); -} - -static INLINE void insert_4chan_4f_rgba_3( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - GLchan *c = (GLchan *)v; - (void) a; - UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); - UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); - UNCLAMPED_FLOAT_TO_CHAN(c[2], in[2]); - c[3] = CHAN_MAX; -} - -static INLINE void insert_4chan_4f_rgba_2( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - GLchan *c = (GLchan *)v; - (void) a; - UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); - UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); - c[2] = 0; - c[3] = CHAN_MAX; -} - -static INLINE void insert_4chan_4f_rgba_1( const struct draw_vf_attr *a, uint8_t *v, - const float *in ) -{ - GLchan *c = (GLchan *)v; - (void) a; - UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); - c[1] = 0; - c[2] = 0; - c[3] = CHAN_MAX; -} - static INLINE void insert_4ub_4f_rgba_4( const struct draw_vf_attr *a, uint8_t *v, const float *in ) { @@ -545,291 +382,64 @@ static INLINE void insert_1ub_1f_1( const struct draw_vf_attr *a, uint8_t *v, } -/*********************************************************************** - * Functions to perform the reverse operations to the above, for - * swrast translation and clip-interpolation. - * - * Currently always extracts a full 4 floats. - */ - -static void extract_4f_viewport( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - const float *in = (const float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - /* Although included for completeness, the position coordinate is - * usually handled differently during clipping. - */ - out[0] = (in[0] - trans[0]) / scale[0]; - out[1] = (in[1] - trans[1]) / scale[1]; - out[2] = (in[2] - trans[2]) / scale[2]; - out[3] = in[3]; -} - -static void extract_3f_viewport( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - const float *in = (const float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = (in[0] - trans[0]) / scale[0]; - out[1] = (in[1] - trans[1]) / scale[1]; - out[2] = (in[2] - trans[2]) / scale[2]; - out[3] = 1; -} - - -static void extract_2f_viewport( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - const float *in = (const float *)v; - const float *scale = a->vf->vp; - const float *trans = a->vf->vp + 4; - - out[0] = (in[0] - trans[0]) / scale[0]; - out[1] = (in[1] - trans[1]) / scale[1]; - out[2] = 0; - out[3] = 1; -} - - -static void extract_4f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) -{ - const float *in = (const float *)v; - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = in[2]; - out[3] = in[3]; -} - -static void extract_3f_xyw( const struct draw_vf_attr *a, float *out, const uint8_t *v ) -{ - const float *in = (const float *)v; - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = 0; - out[3] = in[2]; -} - - -static void extract_3f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) -{ - const float *in = (const float *)v; - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = in[2]; - out[3] = 1; -} - - -static void extract_2f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) -{ - const float *in = (const float *)v; - (void) a; - - out[0] = in[0]; - out[1] = in[1]; - out[2] = 0; - out[3] = 1; -} - -static void extract_1f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) -{ - const float *in = (const float *)v; - (void) a; - - out[0] = in[0]; - out[1] = 0; - out[2] = 0; - out[3] = 1; -} - -static void extract_4chan_4f_rgba( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - GLchan *c = (GLchan *)v; - (void) a; - - out[0] = CHAN_TO_FLOAT(c[0]); - out[1] = CHAN_TO_FLOAT(c[1]); - out[2] = CHAN_TO_FLOAT(c[2]); - out[3] = CHAN_TO_FLOAT(c[3]); -} - -static void extract_4ub_4f_rgba( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - (void) a; - out[0] = UBYTE_TO_FLOAT(v[0]); - out[1] = UBYTE_TO_FLOAT(v[1]); - out[2] = UBYTE_TO_FLOAT(v[2]); - out[3] = UBYTE_TO_FLOAT(v[3]); -} - -static void extract_4ub_4f_bgra( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - (void) a; - out[2] = UBYTE_TO_FLOAT(v[0]); - out[1] = UBYTE_TO_FLOAT(v[1]); - out[0] = UBYTE_TO_FLOAT(v[2]); - out[3] = UBYTE_TO_FLOAT(v[3]); -} - -static void extract_4ub_4f_argb( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - (void) a; - out[3] = UBYTE_TO_FLOAT(v[0]); - out[0] = UBYTE_TO_FLOAT(v[1]); - out[1] = UBYTE_TO_FLOAT(v[2]); - out[2] = UBYTE_TO_FLOAT(v[3]); -} - -static void extract_4ub_4f_abgr( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - (void) a; - out[3] = UBYTE_TO_FLOAT(v[0]); - out[2] = UBYTE_TO_FLOAT(v[1]); - out[1] = UBYTE_TO_FLOAT(v[2]); - out[0] = UBYTE_TO_FLOAT(v[3]); -} - -static void extract_3ub_3f_rgb( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - (void) a; - out[0] = UBYTE_TO_FLOAT(v[0]); - out[1] = UBYTE_TO_FLOAT(v[1]); - out[2] = UBYTE_TO_FLOAT(v[2]); - out[3] = 1; -} - -static void extract_3ub_3f_bgr( const struct draw_vf_attr *a, float *out, - const uint8_t *v ) -{ - (void) a; - out[2] = UBYTE_TO_FLOAT(v[0]); - out[1] = UBYTE_TO_FLOAT(v[1]); - out[0] = UBYTE_TO_FLOAT(v[2]); - out[3] = 1; -} - -static void extract_1ub_1f( const struct draw_vf_attr *a, float *out, const uint8_t *v ) -{ - (void) a; - out[0] = UBYTE_TO_FLOAT(v[0]); - out[1] = 0; - out[2] = 0; - out[3] = 1; -} - - const struct draw_vf_format_info draw_vf_format_info[DRAW_EMIT_MAX] = { { "1f", - extract_1f, { insert_1f_1, insert_1f_1, insert_1f_1, insert_1f_1 }, sizeof(float) }, { "2f", - extract_2f, { insert_2f_1, insert_2f_2, insert_2f_2, insert_2f_2 }, 2 * sizeof(float) }, { "3f", - extract_3f, { insert_3f_1, insert_3f_2, insert_3f_3, insert_3f_3 }, 3 * sizeof(float) }, { "4f", - extract_4f, { insert_4f_1, insert_4f_2, insert_4f_3, insert_4f_4 }, 4 * sizeof(float) }, - { "2f_viewport", - extract_2f_viewport, - { insert_2f_viewport_1, insert_2f_viewport_2, insert_2f_viewport_2, - insert_2f_viewport_2 }, - 2 * sizeof(float) }, - - { "3f_viewport", - extract_3f_viewport, - { insert_3f_viewport_1, insert_3f_viewport_2, insert_3f_viewport_3, - insert_3f_viewport_3 }, - 3 * sizeof(float) }, - - { "4f_viewport", - extract_4f_viewport, - { insert_4f_viewport_1, insert_4f_viewport_2, insert_4f_viewport_3, - insert_4f_viewport_4 }, - 4 * sizeof(float) }, - { "3f_xyw", - extract_3f_xyw, { insert_3f_xyw_err, insert_3f_xyw_err, insert_3f_xyw_err, insert_3f_xyw_4 }, 3 * sizeof(float) }, { "1ub_1f", - extract_1ub_1f, { insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1 }, sizeof(uint8_t) }, { "3ub_3f_rgb", - extract_3ub_3f_rgb, { insert_3ub_3f_rgb_1, insert_3ub_3f_rgb_2, insert_3ub_3f_rgb_3, insert_3ub_3f_rgb_3 }, 3 * sizeof(uint8_t) }, { "3ub_3f_bgr", - extract_3ub_3f_bgr, { insert_3ub_3f_bgr_1, insert_3ub_3f_bgr_2, insert_3ub_3f_bgr_3, insert_3ub_3f_bgr_3 }, 3 * sizeof(uint8_t) }, { "4ub_4f_rgba", - extract_4ub_4f_rgba, { insert_4ub_4f_rgba_1, insert_4ub_4f_rgba_2, insert_4ub_4f_rgba_3, insert_4ub_4f_rgba_4 }, 4 * sizeof(uint8_t) }, { "4ub_4f_bgra", - extract_4ub_4f_bgra, { insert_4ub_4f_bgra_1, insert_4ub_4f_bgra_2, insert_4ub_4f_bgra_3, insert_4ub_4f_bgra_4 }, 4 * sizeof(uint8_t) }, { "4ub_4f_argb", - extract_4ub_4f_argb, { insert_4ub_4f_argb_1, insert_4ub_4f_argb_2, insert_4ub_4f_argb_3, insert_4ub_4f_argb_4 }, 4 * sizeof(uint8_t) }, { "4ub_4f_abgr", - extract_4ub_4f_abgr, { insert_4ub_4f_abgr_1, insert_4ub_4f_abgr_2, insert_4ub_4f_abgr_3, insert_4ub_4f_abgr_4 }, 4 * sizeof(uint8_t) }, - { "4chan_4f_rgba", - extract_4chan_4f_rgba, - { insert_4chan_4f_rgba_1, insert_4chan_4f_rgba_2, insert_4chan_4f_rgba_3, - insert_4chan_4f_rgba_4 }, - 4 * sizeof(GLchan) }, - { "pad", - NULL, { NULL, NULL, NULL, NULL }, 0 } @@ -889,16 +499,10 @@ static void NAME( struct draw_vertex_fetch *vf, \ insert_null, NAME) -EMIT2(insert_3f_viewport_3, insert_4ub_4f_rgba_4, emit_viewport3_rgba4) -EMIT2(insert_3f_viewport_3, insert_4ub_4f_bgra_4, emit_viewport3_bgra4) EMIT2(insert_3f_3, insert_4ub_4f_rgba_4, emit_xyz3_rgba4) -EMIT3(insert_4f_viewport_4, insert_4ub_4f_rgba_4, insert_2f_2, emit_viewport4_rgba4_st2) -EMIT3(insert_4f_viewport_4, insert_4ub_4f_bgra_4, insert_2f_2, emit_viewport4_bgra4_st2) EMIT3(insert_4f_4, insert_4ub_4f_rgba_4, insert_2f_2, emit_xyzw4_rgba4_st2) -EMIT4(insert_4f_viewport_4, insert_4ub_4f_rgba_4, insert_2f_2, insert_2f_2, emit_viewport4_rgba4_st2_st2) -EMIT4(insert_4f_viewport_4, insert_4ub_4f_bgra_4, insert_2f_2, insert_2f_2, emit_viewport4_bgra4_st2_st2) EMIT4(insert_4f_4, insert_4ub_4f_rgba_4, insert_2f_2, insert_2f_2, emit_xyzw4_rgba4_st2_st2) @@ -914,42 +518,26 @@ void draw_vf_generate_hardwired_emit( struct draw_vertex_fetch *vf ) */ switch (vf->attr_count) { case 2: - if (vf->attr[0].do_insert == insert_3f_viewport_3) { - if (vf->attr[1].do_insert == insert_4ub_4f_bgra_4) - func = emit_viewport3_bgra4; - else if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) - func = emit_viewport3_rgba4; - } - else if (vf->attr[0].do_insert == insert_3f_3 && - vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { + if (vf->attr[0].do_insert == insert_3f_3 && + vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { func = emit_xyz3_rgba4; } break; case 3: if (vf->attr[2].do_insert == insert_2f_2) { if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { - if (vf->attr[0].do_insert == insert_4f_viewport_4) - func = emit_viewport4_rgba4_st2; - else if (vf->attr[0].do_insert == insert_4f_4) + if (vf->attr[0].do_insert == insert_4f_4) func = emit_xyzw4_rgba4_st2; } - else if (vf->attr[1].do_insert == insert_4ub_4f_bgra_4 && - vf->attr[0].do_insert == insert_4f_viewport_4) - func = emit_viewport4_bgra4_st2; } break; case 4: if (vf->attr[2].do_insert == insert_2f_2 && vf->attr[3].do_insert == insert_2f_2) { if (vf->attr[1].do_insert == insert_4ub_4f_rgba_4) { - if (vf->attr[0].do_insert == insert_4f_viewport_4) - func = emit_viewport4_rgba4_st2_st2; - else if (vf->attr[0].do_insert == insert_4f_4) + if (vf->attr[0].do_insert == insert_4f_4) func = emit_xyzw4_rgba4_st2_st2; } - else if (vf->attr[1].do_insert == insert_4ub_4f_bgra_4 && - vf->attr[0].do_insert == insert_4f_viewport_4) - func = emit_viewport4_bgra4_st2_st2; } break; } diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c index a7019a47e64..b238b542e76 100644 --- a/src/mesa/pipe/draw/draw_vf_sse.c +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -369,13 +369,6 @@ static boolean build_vertex_emit( struct x86_program *p ) x86_mov(&p->func, vfESI, x86_fn_arg(&p->func, 1)); - /* Possibly load vp0, vp1 for viewport calcs: - */ - if (vf->allow_viewport_emits) { - sse_movups(&p->func, vp0, x86_make_disp(vfESI, get_offset(vf, &vf->vp[0]))); - sse_movups(&p->func, vp1, x86_make_disp(vfESI, get_offset(vf, &vf->vp[4]))); - } - /* always load, needed or not: */ sse_movups(&p->func, p->chan0, x86_make_disp(vfESI, get_offset(vf, &vf->chan_scale[0]))); @@ -439,30 +432,6 @@ static boolean build_vertex_emit( struct x86_program *p ) emit_store(p, dest, 4, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case DRAW_EMIT_2F_VIEWPORT: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); - sse_mulps(&p->func, temp, vp0); - sse_addps(&p->func, temp, vp1); - emit_store(p, dest, 2, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_3F_VIEWPORT: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 3, x86_deref(srcECX), a->inputsize); - sse_mulps(&p->func, temp, vp0); - sse_addps(&p->func, temp, vp1); - emit_store(p, dest, 3, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case DRAW_EMIT_4F_VIEWPORT: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - sse_mulps(&p->func, temp, vp0); - sse_addps(&p->func, temp, vp1); - emit_store(p, dest, 4, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; case DRAW_EMIT_3F_XYW: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); @@ -561,26 +530,6 @@ static boolean build_vertex_emit( struct x86_program *p ) emit_pack_store_4ub(p, dest, temp); update_src_ptr(p, srcECX, vfESI, a); break; - case DRAW_EMIT_4CHAN_4F_RGBA: - switch (CHAN_TYPE) { - case GL_UNSIGNED_BYTE: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - emit_pack_store_4ub(p, dest, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case GL_FLOAT: - get_src_ptr(p, srcECX, vfESI, a); - emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); - emit_store(p, dest, 4, temp); - update_src_ptr(p, srcECX, vfESI, a); - break; - case GL_UNSIGNED_SHORT: - default: - _mesa_printf("unknown CHAN_TYPE %s\n", _mesa_lookup_enum_by_nr(CHAN_TYPE)); - return FALSE; - } - break; default: _mesa_printf("unknown a[%d].format %d\n", j, a->format); return FALSE; /* catch any new opcodes */ -- cgit v1.2.3 From 01ab6472cce1a5ff0186eb606ed3077d9008a53f Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 10:46:52 +0000 Subject: gallium: fill in missing formats for vertex_fetch --- src/mesa/pipe/draw/draw_vertex_fetch.c | 220 ++++++++++++++++++++++++++++++--- 1 file changed, 203 insertions(+), 17 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index fb64723a193..0789dc8e8c7 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -62,50 +62,236 @@ fetch_##NAME(const void *ptr, float *attrib) \ } \ } +#define CVT_64_FLOAT ((double *) ptr)[i] #define CVT_32_FLOAT ((float *) ptr)[i] + +#define CVT_8_USCALED (float) ((unsigned char *) ptr)[i] +#define CVT_16_USCALED (float) ((unsigned short *) ptr)[i] +#define CVT_32_USCALED (float) ((unsigned int *) ptr)[i] + +#define CVT_8_SSCALED (float) ((char *) ptr)[i] +#define CVT_16_SSCALED (float) ((short *) ptr)[i] #define CVT_32_SSCALED (float) ((int *) ptr)[i] + #define CVT_8_UNORM (float) ((unsigned char *) ptr)[i] / 255.0f +#define CVT_16_UNORM (float) ((unsigned short *) ptr)[i] / 65535.0f +#define CVT_32_UNORM (float) ((unsigned int *) ptr)[i] / 4294967295.0f + +#define CVT_8_SNORM (float) ((char *) ptr)[i] / 127.0f +#define CVT_16_SNORM (float) ((short *) ptr)[i] / 32767.0f +#define CVT_32_SNORM (float) ((int *) ptr)[i] / 2147483647.0f + +FETCH_ATTRIB( R64G64B64A64_FLOAT, 4, CVT_64_FLOAT ) +FETCH_ATTRIB( R64G64B64_FLOAT, 3, CVT_64_FLOAT ) +FETCH_ATTRIB( R64G64_FLOAT, 2, CVT_64_FLOAT ) +FETCH_ATTRIB( R64_FLOAT, 1, CVT_64_FLOAT ) FETCH_ATTRIB( R32G32B32A32_FLOAT, 4, CVT_32_FLOAT ) FETCH_ATTRIB( R32G32B32_FLOAT, 3, CVT_32_FLOAT ) FETCH_ATTRIB( R32G32_FLOAT, 2, CVT_32_FLOAT ) FETCH_ATTRIB( R32_FLOAT, 1, CVT_32_FLOAT ) + +FETCH_ATTRIB( R32G32B32A32_USCALED, 4, CVT_32_USCALED ) +FETCH_ATTRIB( R32G32B32_USCALED, 3, CVT_32_USCALED ) +FETCH_ATTRIB( R32G32_USCALED, 2, CVT_32_USCALED ) +FETCH_ATTRIB( R32_USCALED, 1, CVT_32_USCALED ) + FETCH_ATTRIB( R32G32B32A32_SSCALED, 4, CVT_32_SSCALED ) FETCH_ATTRIB( R32G32B32_SSCALED, 3, CVT_32_SSCALED ) FETCH_ATTRIB( R32G32_SSCALED, 2, CVT_32_SSCALED ) FETCH_ATTRIB( R32_SSCALED, 1, CVT_32_SSCALED ) + +FETCH_ATTRIB( R32G32B32A32_UNORM, 4, CVT_32_UNORM ) +FETCH_ATTRIB( R32G32B32_UNORM, 3, CVT_32_UNORM ) +FETCH_ATTRIB( R32G32_UNORM, 2, CVT_32_UNORM ) +FETCH_ATTRIB( R32_UNORM, 1, CVT_32_UNORM ) + +FETCH_ATTRIB( R32G32B32A32_SNORM, 4, CVT_32_SNORM ) +FETCH_ATTRIB( R32G32B32_SNORM, 3, CVT_32_SNORM ) +FETCH_ATTRIB( R32G32_SNORM, 2, CVT_32_SNORM ) +FETCH_ATTRIB( R32_SNORM, 1, CVT_32_SNORM ) + +FETCH_ATTRIB( R16G16B16A16_USCALED, 4, CVT_16_USCALED ) +FETCH_ATTRIB( R16G16B16_USCALED, 3, CVT_16_USCALED ) +FETCH_ATTRIB( R16G16_USCALED, 2, CVT_16_USCALED ) +FETCH_ATTRIB( R16_USCALED, 1, CVT_16_USCALED ) + +FETCH_ATTRIB( R16G16B16A16_SSCALED, 4, CVT_16_SSCALED ) +FETCH_ATTRIB( R16G16B16_SSCALED, 3, CVT_16_SSCALED ) +FETCH_ATTRIB( R16G16_SSCALED, 2, CVT_16_SSCALED ) +FETCH_ATTRIB( R16_SSCALED, 1, CVT_16_SSCALED ) + +FETCH_ATTRIB( R16G16B16A16_UNORM, 4, CVT_16_UNORM ) +FETCH_ATTRIB( R16G16B16_UNORM, 3, CVT_16_UNORM ) +FETCH_ATTRIB( R16G16_UNORM, 2, CVT_16_UNORM ) +FETCH_ATTRIB( R16_UNORM, 1, CVT_16_UNORM ) + +FETCH_ATTRIB( R16G16B16A16_SNORM, 4, CVT_16_SNORM ) +FETCH_ATTRIB( R16G16B16_SNORM, 3, CVT_16_SNORM ) +FETCH_ATTRIB( R16G16_SNORM, 2, CVT_16_SNORM ) +FETCH_ATTRIB( R16_SNORM, 1, CVT_16_SNORM ) + +FETCH_ATTRIB( R8G8B8A8_USCALED, 4, CVT_8_USCALED ) +FETCH_ATTRIB( R8G8B8_USCALED, 3, CVT_8_USCALED ) +FETCH_ATTRIB( R8G8_USCALED, 2, CVT_8_USCALED ) +FETCH_ATTRIB( R8_USCALED, 1, CVT_8_USCALED ) + +FETCH_ATTRIB( R8G8B8A8_SSCALED, 4, CVT_8_SSCALED ) +FETCH_ATTRIB( R8G8B8_SSCALED, 3, CVT_8_SSCALED ) +FETCH_ATTRIB( R8G8_SSCALED, 2, CVT_8_SSCALED ) +FETCH_ATTRIB( R8_SSCALED, 1, CVT_8_SSCALED ) + +FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) +FETCH_ATTRIB( R8G8B8_UNORM, 3, CVT_8_UNORM ) +FETCH_ATTRIB( R8G8_UNORM, 2, CVT_8_UNORM ) +FETCH_ATTRIB( R8_UNORM, 1, CVT_8_UNORM ) + +FETCH_ATTRIB( R8G8B8A8_SNORM, 4, CVT_8_SNORM ) +FETCH_ATTRIB( R8G8B8_SNORM, 3, CVT_8_SNORM ) +FETCH_ATTRIB( R8G8_SNORM, 2, CVT_8_SNORM ) +FETCH_ATTRIB( R8_SNORM, 1, CVT_8_SNORM ) + FETCH_ATTRIB( A8R8G8B8_UNORM, 4, CVT_8_UNORM ) -FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) +//FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) static fetch_func get_fetch_func( enum pipe_format format ) { switch (format) { - case PIPE_FORMAT_R32G32B32A32_FLOAT: - return fetch_R32G32B32A32_FLOAT; - case PIPE_FORMAT_R32G32B32_FLOAT: - return fetch_R32G32B32_FLOAT; - case PIPE_FORMAT_R32G32_FLOAT: - return fetch_R32G32_FLOAT; + case PIPE_FORMAT_R64_FLOAT: + return fetch_R64_FLOAT; + case PIPE_FORMAT_R64G64_FLOAT: + return fetch_R64G64_FLOAT; + case PIPE_FORMAT_R64G64B64_FLOAT: + return fetch_R64G64B64_FLOAT; + case PIPE_FORMAT_R64G64B64A64_FLOAT: + return fetch_R64G64B64A64_FLOAT; + case PIPE_FORMAT_R32_FLOAT: return fetch_R32_FLOAT; - case PIPE_FORMAT_R32G32B32A32_SSCALED: - return fetch_R32G32B32A32_SSCALED; - case PIPE_FORMAT_R32G32B32_SSCALED: - return fetch_R32G32B32_SSCALED; - case PIPE_FORMAT_R32G32_SSCALED: - return fetch_R32G32_SSCALED; + case PIPE_FORMAT_R32G32_FLOAT: + return fetch_R32G32_FLOAT; + case PIPE_FORMAT_R32G32B32_FLOAT: + return fetch_R32G32B32_FLOAT; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + return fetch_R32G32B32A32_FLOAT; + + case PIPE_FORMAT_R32_UNORM: + return fetch_R32_UNORM; + case PIPE_FORMAT_R32G32_UNORM: + return fetch_R32G32_UNORM; + case PIPE_FORMAT_R32G32B32_UNORM: + return fetch_R32G32B32_UNORM; + case PIPE_FORMAT_R32G32B32A32_UNORM: + return fetch_R32G32B32A32_UNORM; + + case PIPE_FORMAT_R32_USCALED: + return fetch_R32_USCALED; + case PIPE_FORMAT_R32G32_USCALED: + return fetch_R32G32_USCALED; + case PIPE_FORMAT_R32G32B32_USCALED: + return fetch_R32G32B32_USCALED; + case PIPE_FORMAT_R32G32B32A32_USCALED: + return fetch_R32G32B32A32_USCALED; + + case PIPE_FORMAT_R32_SNORM: + return fetch_R32_SNORM; + case PIPE_FORMAT_R32G32_SNORM: + return fetch_R32G32_SNORM; + case PIPE_FORMAT_R32G32B32_SNORM: + return fetch_R32G32B32_SNORM; + case PIPE_FORMAT_R32G32B32A32_SNORM: + return fetch_R32G32B32A32_SNORM; + case PIPE_FORMAT_R32_SSCALED: return fetch_R32_SSCALED; - case PIPE_FORMAT_A8R8G8B8_UNORM: - return fetch_A8R8G8B8_UNORM; + case PIPE_FORMAT_R32G32_SSCALED: + return fetch_R32G32_SSCALED; + case PIPE_FORMAT_R32G32B32_SSCALED: + return fetch_R32G32B32_SSCALED; + case PIPE_FORMAT_R32G32B32A32_SSCALED: + return fetch_R32G32B32A32_SSCALED; + + case PIPE_FORMAT_R16_UNORM: + return fetch_R16_UNORM; + case PIPE_FORMAT_R16G16_UNORM: + return fetch_R16G16_UNORM; + case PIPE_FORMAT_R16G16B16_UNORM: + return fetch_R16G16B16_UNORM; + case PIPE_FORMAT_R16G16B16A16_UNORM: + return fetch_R16G16B16A16_UNORM; + + case PIPE_FORMAT_R16_USCALED: + return fetch_R16_USCALED; + case PIPE_FORMAT_R16G16_USCALED: + return fetch_R16G16_USCALED; + case PIPE_FORMAT_R16G16B16_USCALED: + return fetch_R16G16B16_USCALED; + case PIPE_FORMAT_R16G16B16A16_USCALED: + return fetch_R16G16B16A16_USCALED; + + case PIPE_FORMAT_R16_SNORM: + return fetch_R16_SNORM; + case PIPE_FORMAT_R16G16_SNORM: + return fetch_R16G16_SNORM; + case PIPE_FORMAT_R16G16B16_SNORM: + return fetch_R16G16B16_SNORM; + case PIPE_FORMAT_R16G16B16A16_SNORM: + return fetch_R16G16B16A16_SNORM; + + case PIPE_FORMAT_R16_SSCALED: + return fetch_R16_SSCALED; + case PIPE_FORMAT_R16G16_SSCALED: + return fetch_R16G16_SSCALED; + case PIPE_FORMAT_R16G16B16_SSCALED: + return fetch_R16G16B16_SSCALED; + case PIPE_FORMAT_R16G16B16A16_SSCALED: + return fetch_R16G16B16A16_SSCALED; + + case PIPE_FORMAT_R8_UNORM: + return fetch_R8_UNORM; + case PIPE_FORMAT_R8G8_UNORM: + return fetch_R8G8_UNORM; + case PIPE_FORMAT_R8G8B8_UNORM: + return fetch_R8G8B8_UNORM; case PIPE_FORMAT_R8G8B8A8_UNORM: return fetch_R8G8B8A8_UNORM; + + case PIPE_FORMAT_R8_USCALED: + return fetch_R8_USCALED; + case PIPE_FORMAT_R8G8_USCALED: + return fetch_R8G8_USCALED; + case PIPE_FORMAT_R8G8B8_USCALED: + return fetch_R8G8B8_USCALED; + case PIPE_FORMAT_R8G8B8A8_USCALED: + return fetch_R8G8B8A8_USCALED; + + case PIPE_FORMAT_R8_SNORM: + return fetch_R8_SNORM; + case PIPE_FORMAT_R8G8_SNORM: + return fetch_R8G8_SNORM; + case PIPE_FORMAT_R8G8B8_SNORM: + return fetch_R8G8B8_SNORM; + case PIPE_FORMAT_R8G8B8A8_SNORM: + return fetch_R8G8B8A8_SNORM; + + case PIPE_FORMAT_R8_SSCALED: + return fetch_R8_SSCALED; + case PIPE_FORMAT_R8G8_SSCALED: + return fetch_R8G8_SSCALED; + case PIPE_FORMAT_R8G8B8_SSCALED: + return fetch_R8G8B8_SSCALED; + case PIPE_FORMAT_R8G8B8A8_SSCALED: + return fetch_R8G8B8A8_SSCALED; + + case PIPE_FORMAT_A8R8G8B8_UNORM: + return fetch_A8R8G8B8_UNORM; + case 0: - return NULL; + return NULL; /* not sure why this is needed */ + default: - /* Lots of missing cases! */ assert(0); return NULL; } -- cgit v1.2.3 From 85d7e7ceeecde86621e3d999c475c1e9d97091f0 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 10:48:22 +0000 Subject: gallium: explictly cast double to float in vertex fetch --- src/mesa/pipe/draw/draw_vertex_fetch.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index 0789dc8e8c7..af3983b7f03 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -62,7 +62,7 @@ fetch_##NAME(const void *ptr, float *attrib) \ } \ } -#define CVT_64_FLOAT ((double *) ptr)[i] +#define CVT_64_FLOAT (float) ((double *) ptr)[i] #define CVT_32_FLOAT ((float *) ptr)[i] #define CVT_8_USCALED (float) ((unsigned char *) ptr)[i] -- cgit v1.2.3 From f7e64c323fe6a646ee60c55ba2552923a7670c53 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 11:12:11 +0000 Subject: gallium: only call vertex/prim queue flush when there is something to flush --- src/mesa/pipe/draw/draw_prim.c | 9 +++++---- src/mesa/pipe/draw/draw_vertex_fetch.c | 10 ++++++++++ src/mesa/pipe/draw/draw_vertex_shader.c | 2 ++ 3 files changed, 17 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/draw/draw_prim.c b/src/mesa/pipe/draw/draw_prim.c index 243381aec0c..2a612a16733 100644 --- a/src/mesa/pipe/draw/draw_prim.c +++ b/src/mesa/pipe/draw/draw_prim.c @@ -63,8 +63,7 @@ static void draw_prim_queue_flush( struct draw_context *draw ) fprintf(stdout,"Flushing with %d prims, %d verts\n", draw->pq.queue_nr, draw->vs.queue_nr); - if (draw->pq.queue_nr == 0) - return; + assert (draw->pq.queue_nr != 0); /* NOTE: we cannot save draw->pipeline->first in a local var because * draw->pipeline->first is often changed by the first call to tri(), @@ -109,10 +108,12 @@ void draw_do_flush( struct draw_context *draw, unsigned flags ) if (flags >= DRAW_FLUSH_SHADER_QUEUE) { - draw_vertex_shader_queue_flush(draw); + if (draw->vs.queue_nr) + draw_vertex_shader_queue_flush(draw); if (flags >= DRAW_FLUSH_PRIM_QUEUE) { - draw_prim_queue_flush(draw); + if (draw->pq.queue_nr) + draw_prim_queue_flush(draw); if (flags >= DRAW_FLUSH_VERTEX_CACHE) { draw_vertex_cache_invalidate(draw); diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index af3983b7f03..143acdd3b49 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -158,6 +158,14 @@ FETCH_ATTRIB( A8R8G8B8_UNORM, 4, CVT_8_UNORM ) static fetch_func get_fetch_func( enum pipe_format format ) { +#if 0 + { + char tmp[80]; + pf_sprint_name(tmp, format); + _mesa_printf("%s: %s\n", __FUNCTION__, tmp); + } +#endif + switch (format) { case PIPE_FORMAT_R64_FLOAT: return fetch_R64_FLOAT; @@ -317,6 +325,8 @@ void draw_update_vertex_fetch( struct draw_context *draw ) { unsigned nr_attrs, i; +// _mesa_printf("%s\n", __FUNCTION__); + /* this may happend during context init */ if (!draw->vertex_shader) return; diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index 3041974b9a4..289c35c7ae4 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -201,6 +201,8 @@ draw_vertex_shader_queue_flush(struct draw_context *draw) { unsigned i, j; + assert(draw->vs.queue_nr != 0); + /* XXX: do this on statechange: */ draw_update_vertex_fetch( draw ); -- cgit v1.2.3 From bb37e7f5917dba3f2ad84ecf0b6c95bf58205faf Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 12:40:29 +0000 Subject: gallium: add a couple of hardwired vertex fetch functions --- src/mesa/pipe/draw/draw_private.h | 9 +- src/mesa/pipe/draw/draw_vertex_fetch.c | 150 +++++++++++++++++++++++++++----- src/mesa/pipe/draw/draw_vertex_shader.c | 12 ++- 3 files changed, 141 insertions(+), 30 deletions(-) diff --git a/src/mesa/pipe/draw/draw_private.h b/src/mesa/pipe/draw/draw_private.h index 1e59f5bd8d9..21de4006769 100644 --- a/src/mesa/pipe/draw/draw_private.h +++ b/src/mesa/pipe/draw/draw_private.h @@ -141,6 +141,10 @@ struct draw_vertex_shader { /* Internal function for vertex fetch. */ typedef void (*fetch_func)(const void *ptr, float *attrib); +typedef void (*full_fetch_func)( struct draw_context *draw, + struct tgsi_exec_machine *machine, + const unsigned *elts, + unsigned count ); @@ -210,6 +214,7 @@ struct draw_context unsigned pitch[PIPE_ATTRIB_MAX]; fetch_func fetch[PIPE_ATTRIB_MAX]; unsigned nr_attrs; + full_fetch_func fetch_func; } vertex_fetch; /* Post-tnl vertex cache: @@ -287,10 +292,6 @@ extern void draw_vertex_shader_queue_flush_llvm( struct draw_context *draw ); struct tgsi_exec_machine; extern void draw_update_vertex_fetch( struct draw_context *draw ); -extern void draw_vertex_fetch( struct draw_context *draw, - struct tgsi_exec_machine *machine, - const unsigned *elts, - unsigned count ); #define DRAW_FLUSH_SHADER_QUEUE 0x1 /* sized not to overflow, never raised */ diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index 143acdd3b49..afdf1971d24 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -320,42 +320,101 @@ transpose_4x4( float *out, const float *in ) } - -void draw_update_vertex_fetch( struct draw_context *draw ) + +static void fetch_xyz_rgb( struct draw_context *draw, + struct tgsi_exec_machine *machine, + const unsigned *elts, + unsigned count ) { - unsigned nr_attrs, i; + assert(count <= 4); // _mesa_printf("%s\n", __FUNCTION__); - - /* this may happend during context init */ - if (!draw->vertex_shader) - return; - nr_attrs = draw->vertex_shader->state->num_inputs; + /* loop over vertex attributes (vertex shader inputs) + */ - for (i = 0; i < nr_attrs; i++) { - unsigned buf = draw->vertex_element[i].vertex_buffer_index; - enum pipe_format format = draw->vertex_element[i].src_format; + const unsigned *pitch = draw->vertex_fetch.pitch; + const ubyte **src = draw->vertex_fetch.src_ptr; + int i; + + for (i = 0; i < 4; i++) { + { + const float *in = (const float *)(src[0] + elts[i] * pitch[0]); + float *out = &machine->Inputs[0].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + + { + const float *in = (const float *)(src[1] + elts[i] * pitch[1]); + float *out = &machine->Inputs[1].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + } +} - draw->vertex_fetch.src_ptr[i] = (const ubyte *) draw->user.vbuffer[buf] + - draw->vertex_buffer[buf].buffer_offset + - draw->vertex_element[i].src_offset; - draw->vertex_fetch.pitch[i] = draw->vertex_buffer[buf].pitch; - draw->vertex_fetch.fetch[i] = get_fetch_func( format ); - } - draw->vertex_fetch.nr_attrs = nr_attrs; + +static void fetch_xyz_rgb_st( struct draw_context *draw, + struct tgsi_exec_machine *machine, + const unsigned *elts, + unsigned count ) +{ + assert(count <= 4); + + /* loop over vertex attributes (vertex shader inputs) + */ + + const unsigned *pitch = draw->vertex_fetch.pitch; + const ubyte **src = draw->vertex_fetch.src_ptr; + int i; + + for (i = 0; i < 4; i++) { + { + const float *in = (const float *)(src[0] + elts[i] * pitch[0]); + float *out = &machine->Inputs[0].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + + { + const float *in = (const float *)(src[1] + elts[i] * pitch[1]); + float *out = &machine->Inputs[1].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + + { + const float *in = (const float *)(src[2] + elts[i] * pitch[2]); + float *out = &machine->Inputs[1].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = 0.0f; + out[12] = 1.0f; + } + } } + + /** * Fetch vertex attributes for 'count' vertices. */ -void draw_vertex_fetch( struct draw_context *draw, - struct tgsi_exec_machine *machine, - const unsigned *elts, - unsigned count ) +static void generic_vertex_fetch( struct draw_context *draw, + struct tgsi_exec_machine *machine, + const unsigned *elts, + unsigned count ) { unsigned nr_attrs = draw->vertex_fetch.nr_attrs; unsigned attr; @@ -402,3 +461,50 @@ void draw_vertex_fetch( struct draw_context *draw, } } + + +void draw_update_vertex_fetch( struct draw_context *draw ) +{ + unsigned nr_attrs, i; + +// _mesa_printf("%s\n", __FUNCTION__); + + /* this may happend during context init */ + if (!draw->vertex_shader) + return; + + nr_attrs = draw->vertex_shader->state->num_inputs; + + for (i = 0; i < nr_attrs; i++) { + unsigned buf = draw->vertex_element[i].vertex_buffer_index; + enum pipe_format format = draw->vertex_element[i].src_format; + + draw->vertex_fetch.src_ptr[i] = (const ubyte *) draw->user.vbuffer[buf] + + draw->vertex_buffer[buf].buffer_offset + + draw->vertex_element[i].src_offset; + + draw->vertex_fetch.pitch[i] = draw->vertex_buffer[buf].pitch; + draw->vertex_fetch.fetch[i] = get_fetch_func( format ); + } + + draw->vertex_fetch.nr_attrs = nr_attrs; + + draw->vertex_fetch.fetch_func = generic_vertex_fetch; + + switch (nr_attrs) { + case 2: + if (draw->vertex_element[0].src_format == PIPE_FORMAT_R32G32B32_FLOAT && + draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32B32_FLOAT) + draw->vertex_fetch.fetch_func = fetch_xyz_rgb; + break; + case 3: + if (draw->vertex_element[0].src_format == PIPE_FORMAT_R32G32B32_FLOAT && + draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32B32_FLOAT && + draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32_FLOAT) + draw->vertex_fetch.fetch_func = fetch_xyz_rgb_st; + break; + default: + break; + } + +} diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index 289c35c7ae4..0806e23d6c4 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -110,7 +110,7 @@ run_vertex_program(struct draw_context *draw, machine->Inputs = ALIGN16_ASSIGN(inputs); machine->Outputs = ALIGN16_ASSIGN(outputs); - draw_vertex_fetch( draw, machine, elts, count ); + draw->vertex_fetch.fetch_func( draw, machine, elts, count ); /* run shader */ #if defined(__i386__) || defined(__386__) @@ -219,14 +219,18 @@ draw_vertex_shader_queue_flush(struct draw_context *draw) for (i = 0; i < draw->vs.queue_nr; i += 4) { struct vertex_header *dests[4]; unsigned elts[4]; - int n; + int n = MIN2(4, draw->vs.queue_nr - i); - for (j = 0; j < 4; j++) { + for (j = 0; j < n; j++) { elts[j] = draw->vs.queue[i + j].elt; dests[j] = draw->vs.queue[i + j].dest; } - n = MIN2(4, draw->vs.queue_nr - i); + for ( ; j < 4; j++) { + elts[j] = elts[0]; + dests[j] = dests[0]; + } + assert(n > 0); assert(n <= 4); -- cgit v1.2.3 From af2ccd4c0c58e6565c2c6c6f9464db2cf4e0baab Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 15:50:02 +0000 Subject: gallium: handle flatshading explicitly in clipper stage We can do a better job in the clip stage than just relying on the brute-force approach of copying colors to all incoming vertices applied in the flatshade stage. At very least, it is only necessary to do this in the clipper when a primitive is actually being clipped. --- src/mesa/pipe/draw/draw_clip.c | 136 +++++++++++++++++++++++++------------ src/mesa/pipe/draw/draw_validate.c | 10 +-- 2 files changed, 99 insertions(+), 47 deletions(-) diff --git a/src/mesa/pipe/draw/draw_clip.c b/src/mesa/pipe/draw/draw_clip.c index 2d410e32447..da20028904a 100644 --- a/src/mesa/pipe/draw/draw_clip.c +++ b/src/mesa/pipe/draw/draw_clip.c @@ -33,6 +33,8 @@ #include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" + #include "draw_context.h" #include "draw_private.h" @@ -54,6 +56,12 @@ struct clipper { struct draw_stage stage; /**< base class */ + /* Basically duplicate some of the flatshading logic here: + */ + boolean flat; + uint num_color_attribs; + uint color_attribs[4]; /* front/back primary/secondary colors */ + float (*plane)[4]; }; @@ -82,6 +90,17 @@ static void interp_attr( float *fdst, fdst[3] = LINTERP( t, fout[3], fin[3] ); } +static void copy_colors( struct draw_stage *stage, + struct vertex_header *dst, + const struct vertex_header *src ) +{ + const struct clipper *clipper = clipper_stage(stage); + uint i; + for (i = 0; i < clipper->num_color_attribs; i++) { + const uint attr = clipper->color_attribs[i]; + COPY_4FV(dst->data[attr], src->data[attr]); + } +} @@ -134,27 +153,11 @@ static void interp( const struct clipper *clip, } } -#if 0 -static INLINE void do_tri( struct draw_stage *next, - struct prim_header *header ) -{ - unsigned i; - for (i = 0; i < 3; i++) { - float *ndc = header->v[i]->data[0]; - _mesa_printf("ndc %f %f %f\n", ndc[0], ndc[1], ndc[2]); - assert(ndc[0] >= -1 && ndc[0] <= 641); - assert(ndc[1] >= 30 && ndc[1] <= 481); - } - _mesa_printf("\n"); - next->tri(next, header); -} -#endif - static void emit_poly( struct draw_stage *stage, struct vertex_header **inlist, unsigned n, - const struct prim_header *origPrim) + const struct prim_header *origPrim) { struct prim_header header; unsigned i; @@ -163,16 +166,16 @@ static void emit_poly( struct draw_stage *stage, header.det = origPrim->det; for (i = 2; i < n; i++) { - header.v[0] = inlist[0]; - header.v[1] = inlist[i-1]; - header.v[2] = inlist[i]; + header.v[0] = inlist[i-1]; + header.v[1] = inlist[i]; + header.v[2] = inlist[0]; /* keep in v[2] for flatshading */ { - unsigned tmp0 = header.v[0]->edgeflag; + unsigned tmp1 = header.v[1]->edgeflag; unsigned tmp2 = header.v[2]->edgeflag; - if (i != 2) header.v[0]->edgeflag = 0; - if (i != n-1) header.v[2]->edgeflag = 0; + if (i != n-1) header.v[1]->edgeflag = 0; + if (i != 2) header.v[2]->edgeflag = 0; header.edgeflags = ((header.v[0]->edgeflag << 0) | (header.v[1]->edgeflag << 1) | @@ -180,27 +183,13 @@ static void emit_poly( struct draw_stage *stage, stage->next->tri( stage->next, &header ); - header.v[0]->edgeflag = tmp0; + header.v[1]->edgeflag = tmp1; header.v[2]->edgeflag = tmp2; } } } -#if 0 -static void emit_poly( struct draw_stage *stage ) -{ - unsigned i; - - for (i = 2; i < n; i++) { - header->v[0] = inlist[0]; - header->v[1] = inlist[i-1]; - header->v[2] = inlist[i]; - - stage->next->tri( stage->next, header ); - } -} -#endif /* Clip a triangle against the viewport and user clip planes. @@ -281,6 +270,18 @@ do_clip_tri( struct draw_stage *stage, } } + /* If flat-shading, copy color to new provoking vertex. + */ + if (clipper->flat && inlist[0] != header->v[2]) { + if (1) { + inlist[0] = dup_vert(stage, inlist[0], tmpnr++); + } + + copy_colors(stage, inlist[0], header->v[2]); + } + + + /* Emit the polygon as triangles to the setup stage: */ if (n >= 3) @@ -328,6 +329,10 @@ do_clip_line( struct draw_stage *stage, if (v0->clipmask) { interp( clipper, stage->tmp[0], t0, v0, v1 ); + + if (clipper->flat) + copy_colors(stage, stage->tmp[0], v0); + newprim.v[0] = stage->tmp[0]; } else { @@ -393,8 +398,55 @@ clip_tri( struct draw_stage *stage, } } -static void clip_flush( struct draw_stage *stage, unsigned flags ) +/* Update state. Could further delay this until we hit the first + * primitive that really requires clipping. + */ +static void +clip_init_state( struct draw_stage *stage ) +{ + struct clipper *clipper = clipper_stage( stage ); + + clipper->flat = stage->draw->rasterizer->flatshade; + + if (clipper->flat) { + const struct pipe_shader_state *vs = stage->draw->vertex_shader->state; + uint i; + + clipper->num_color_attribs = 0; + for (i = 0; i < vs->num_outputs; i++) { + if (vs->output_semantic_name[i] == TGSI_SEMANTIC_COLOR || + vs->output_semantic_name[i] == TGSI_SEMANTIC_BCOLOR) { + clipper->color_attribs[clipper->num_color_attribs++] = i; + } + } + } + + stage->tri = clip_tri; + stage->line = clip_line; +} + + + +static void clip_first_tri( struct draw_stage *stage, + struct prim_header *header ) +{ + clip_init_state( stage ); + stage->tri( stage, header ); +} + +static void clip_first_line( struct draw_stage *stage, + struct prim_header *header ) +{ + clip_init_state( stage ); + stage->line( stage, header ); +} + + +static void clip_flush( struct draw_stage *stage, + unsigned flags ) { + stage->tri = clip_first_tri; + stage->line = clip_first_line; stage->next->flush( stage->next, flags ); } @@ -420,12 +472,12 @@ struct draw_stage *draw_clip_stage( struct draw_context *draw ) { struct clipper *clipper = CALLOC_STRUCT(clipper); - draw_alloc_tmps( &clipper->stage, MAX_CLIPPED_VERTICES ); + draw_alloc_tmps( &clipper->stage, MAX_CLIPPED_VERTICES+1 ); clipper->stage.draw = draw; clipper->stage.point = clip_point; - clipper->stage.line = clip_line; - clipper->stage.tri = clip_tri; + clipper->stage.line = clip_first_line; + clipper->stage.tri = clip_first_tri; clipper->stage.flush = clip_flush; clipper->stage.reset_stipple_counter = clip_reset_stipple_counter; clipper->stage.destroy = clip_destroy; diff --git a/src/mesa/pipe/draw/draw_validate.c b/src/mesa/pipe/draw/draw_validate.c index 86d5a5f8142..4375ebabbc4 100644 --- a/src/mesa/pipe/draw/draw_validate.c +++ b/src/mesa/pipe/draw/draw_validate.c @@ -78,6 +78,11 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage ) precalc_flat = 1; /* only needed for triangles really */ need_det = 1; } + + if (draw->rasterizer->flatshade && precalc_flat) { + draw->pipeline.flatshade->next = next; + next = draw->pipeline.flatshade; + } if (draw->rasterizer->offset_cw || draw->rasterizer->offset_ccw) { @@ -110,13 +115,8 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage ) { draw->pipeline.clip->next = next; next = draw->pipeline.clip; - precalc_flat = 1; /* XXX: FIX ME! Only needed for clipped prims */ } - if (draw->rasterizer->flatshade && precalc_flat) { - draw->pipeline.flatshade->next = next; - next = draw->pipeline.flatshade; - } draw->pipeline.first = next; return next; -- cgit v1.2.3 From 72b671bd3986cc655fbe5df76349bc0989b1c083 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 09:23:29 -0700 Subject: gallium: check if surface has defined status in check_clear_depth_with_quad() This was part of Keith's patch from Friday. --- src/mesa/state_tracker/st_cb_clear.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 758d4a40868..0cd469c156a 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -408,7 +408,9 @@ check_clear_depth_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb) const struct st_renderbuffer *strb = st_renderbuffer(rb); const GLboolean isDS = is_depth_stencil_format(strb->surface->format); return ctx->Scissor.Enabled - || (isDS && ctx->DrawBuffer->Visual.stencilBits > 0); + || (isDS && + strb->surface->status == PIPE_SURFACE_STATUS_DEFINED && + ctx->DrawBuffer->Visual.stencilBits > 0); } -- cgit v1.2.3 From 5c7c0675a70b32f159e3a972279535554aa7f4d9 Mon Sep 17 00:00:00 2001 From: Brian Date: Sun, 27 Jan 2008 12:01:47 -0700 Subject: Cell: generalize the batch buffer code for vertex buffers... --- src/mesa/pipe/cell/common.h | 8 ++-- src/mesa/pipe/cell/ppu/cell_batch.c | 84 ++++++++++++++++++++--------------- src/mesa/pipe/cell/ppu/cell_batch.h | 3 ++ src/mesa/pipe/cell/ppu/cell_context.c | 5 ++- src/mesa/pipe/cell/ppu/cell_context.h | 10 +++-- src/mesa/pipe/cell/ppu/cell_spu.c | 4 +- src/mesa/pipe/cell/spu/spu_main.c | 22 ++++----- 7 files changed, 79 insertions(+), 57 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 0b63ed39be9..ce9c3819077 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -81,8 +81,8 @@ #define CELL_CMD_STATE_VERTEX_INFO 13 -#define CELL_NUM_BATCH_BUFFERS 3 -#define CELL_BATCH_BUFFER_SIZE 1024 /**< 16KB would be the max */ +#define CELL_NUM_BUFFERS 4 +#define CELL_BUFFER_SIZE (4*1024) /**< 16KB would be the max */ #define CELL_BUFFER_STATUS_FREE 10 #define CELL_BUFFER_STATUS_USED 20 @@ -147,7 +147,9 @@ struct cell_init_info unsigned id; unsigned num_spus; struct cell_command *cmd; - ubyte *batch_buffers[CELL_NUM_BATCH_BUFFERS]; + + /** Buffers for command batches, vertex/index data */ + ubyte *buffers[CELL_NUM_BUFFERS]; uint *buffer_status; /**< points at cell_context->buffer_status */ } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/ppu/cell_batch.c b/src/mesa/pipe/cell/ppu/cell_batch.c index c894ef86089..178caa74e1b 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.c +++ b/src/mesa/pipe/cell/ppu/cell_batch.c @@ -31,12 +31,46 @@ #include "cell_spu.h" + +uint +cell_get_empty_buffer(struct cell_context *cell) +{ + uint buf = 0; + + /* Find a buffer that's marked as free by all SPUs */ + while (1) { + uint spu, num_free = 0; + + for (spu = 0; spu < cell->num_spus; spu++) { + if (cell->buffer_status[spu][buf][0] == CELL_BUFFER_STATUS_FREE) { + num_free++; + + if (num_free == cell->num_spus) { + /* found a free buffer, now mark status as used */ + for (spu = 0; spu < cell->num_spus; spu++) { + cell->buffer_status[spu][buf][0] = CELL_BUFFER_STATUS_USED; + } + return buf; + } + } + else { + break; + } + } + + /* try next buf */ + buf = (buf + 1) % CELL_NUM_BUFFERS; + } +} + + + void cell_batch_flush(struct cell_context *cell) { static boolean flushing = FALSE; uint batch = cell->cur_batch; - const uint size = cell->batch_buffer_size[batch]; + const uint size = cell->buffer_size[batch]; uint spu, cmd_word; assert(!flushing); @@ -46,7 +80,7 @@ cell_batch_flush(struct cell_context *cell) flushing = TRUE; - assert(batch < CELL_NUM_BATCH_BUFFERS); + assert(batch < CELL_NUM_BUFFERS); /* printf("cell_batch_dispatch: buf %u at %p, size %u\n", @@ -68,28 +102,9 @@ cell_batch_flush(struct cell_context *cell) * array indicating that the PPU can re-use the buffer. */ + batch = cell_get_empty_buffer(cell); - /* Find a buffer that's marked as free by all SPUs */ - while (1) { - uint num_free = 0; - - batch = (batch + 1) % CELL_NUM_BATCH_BUFFERS; - - for (spu = 0; spu < cell->num_spus; spu++) { - if (cell->buffer_status[spu][batch][0] == CELL_BUFFER_STATUS_FREE) - num_free++; - } - - if (num_free == cell->num_spus) { - /* found a free buffer, now mark status as used */ - for (spu = 0; spu < cell->num_spus; spu++) { - cell->buffer_status[spu][batch][0] = CELL_BUFFER_STATUS_USED; - } - break; - } - } - - cell->batch_buffer_size[batch] = 0; /* empty */ + cell->buffer_size[batch] = 0; /* empty */ cell->cur_batch = batch; flushing = FALSE; @@ -99,8 +114,7 @@ cell_batch_flush(struct cell_context *cell) uint cell_batch_free_space(const struct cell_context *cell) { - uint free = CELL_BATCH_BUFFER_SIZE - - cell->batch_buffer_size[cell->cur_batch]; + uint free = CELL_BUFFER_SIZE - cell->buffer_size[cell->cur_batch]; return free; } @@ -117,18 +131,18 @@ cell_batch_append(struct cell_context *cell, const void *cmd, uint length) assert(length % 4 == 0); assert(cell->cur_batch >= 0); - size = cell->batch_buffer_size[cell->cur_batch]; + size = cell->buffer_size[cell->cur_batch]; - if (size + length > CELL_BATCH_BUFFER_SIZE) { + if (size + length > CELL_BUFFER_SIZE) { cell_batch_flush(cell); size = 0; } - assert(size + length <= CELL_BATCH_BUFFER_SIZE); + assert(size + length <= CELL_BUFFER_SIZE); - memcpy(cell->batch_buffer[cell->cur_batch] + size, cmd, length); + memcpy(cell->buffer[cell->cur_batch] + size, cmd, length); - cell->batch_buffer_size[cell->cur_batch] = size + length; + cell->buffer_size[cell->cur_batch] = size + length; } @@ -142,18 +156,18 @@ cell_batch_alloc(struct cell_context *cell, uint bytes) assert(cell->cur_batch >= 0); - size = cell->batch_buffer_size[cell->cur_batch]; + size = cell->buffer_size[cell->cur_batch]; - if (size + bytes > CELL_BATCH_BUFFER_SIZE) { + if (size + bytes > CELL_BUFFER_SIZE) { cell_batch_flush(cell); size = 0; } - assert(size + bytes <= CELL_BATCH_BUFFER_SIZE); + assert(size + bytes <= CELL_BUFFER_SIZE); - pos = (void *) (cell->batch_buffer[cell->cur_batch] + size); + pos = (void *) (cell->buffer[cell->cur_batch] + size); - cell->batch_buffer_size[cell->cur_batch] = size + bytes; + cell->buffer_size[cell->cur_batch] = size + bytes; return pos; } diff --git a/src/mesa/pipe/cell/ppu/cell_batch.h b/src/mesa/pipe/cell/ppu/cell_batch.h index c4ba7feb3da..b4c96f465a8 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.h +++ b/src/mesa/pipe/cell/ppu/cell_batch.h @@ -35,6 +35,9 @@ struct cell_context; +extern uint +cell_get_empty_buffer(struct cell_context *cell); + extern void cell_batch_flush(struct cell_context *cell); diff --git a/src/mesa/pipe/cell/ppu/cell_context.c b/src/mesa/pipe/cell/ppu/cell_context.c index 8cb0c48f407..e8020a49bc3 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.c +++ b/src/mesa/pipe/cell/ppu/cell_context.c @@ -254,8 +254,9 @@ cell_create_context(struct pipe_winsys *winsys, struct cell_winsys *cws) cell_start_spus(cell); - for (buf = 0; buf < CELL_NUM_BATCH_BUFFERS; buf++) { - cell->batch_buffer_size[buf] = 0; + /* init command, vertex/index buffer info */ + for (buf = 0; buf < CELL_NUM_BUFFERS; buf++) { + cell->buffer_size[buf] = 0; /* init batch buffer status values, * mark 0th buffer as used, rest as free. diff --git a/src/mesa/pipe/cell/ppu/cell_context.h b/src/mesa/pipe/cell/ppu/cell_context.h index 3bd88bfd5b8..de65fb5e9a8 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.h +++ b/src/mesa/pipe/cell/ppu/cell_context.h @@ -102,12 +102,14 @@ struct cell_context uint num_spus; - uint batch_buffer_size[CELL_NUM_BATCH_BUFFERS]; - ubyte batch_buffer[CELL_NUM_BATCH_BUFFERS][CELL_BATCH_BUFFER_SIZE] ALIGN16_ATTRIB; - int cur_batch; /**< which batch buffer is being filled */ + /** Buffers for command batches, vertex/index data */ + uint buffer_size[CELL_NUM_BUFFERS]; + ubyte buffer[CELL_NUM_BUFFERS][CELL_BUFFER_SIZE] ALIGN16_ATTRIB; + + int cur_batch; /**< which buffer is being filled w/ commands */ /** [4] to ensure 16-byte alignment for each status word */ - uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BATCH_BUFFERS][4] ALIGN16_ATTRIB; + uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4] ALIGN16_ATTRIB; }; diff --git a/src/mesa/pipe/cell/ppu/cell_spu.c b/src/mesa/pipe/cell/ppu/cell_spu.c index 4627bc8d1f4..7c83a47e574 100644 --- a/src/mesa/pipe/cell/ppu/cell_spu.c +++ b/src/mesa/pipe/cell/ppu/cell_spu.c @@ -111,8 +111,8 @@ cell_start_spus(struct cell_context *cell) cell_global.inits[i].id = i; cell_global.inits[i].num_spus = cell->num_spus; cell_global.inits[i].cmd = &cell_global.command[i]; - for (j = 0; j < CELL_NUM_BATCH_BUFFERS; j++) { - cell_global.inits[i].batch_buffers[j] = cell->batch_buffer[j]; + for (j = 0; j < CELL_NUM_BUFFERS; j++) { + cell_global.inits[i].buffers[j] = cell->buffer[j]; } cell_global.inits[i].buffer_status = &cell->buffer_status[0][0][0]; diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 0c83900a187..2097683b825 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -473,22 +473,22 @@ cmd_finish(void) /** - * Tell the PPU that this SPU has finished copying a batch buffer to + * Tell the PPU that this SPU has finished copying a buffer to * local store and that it may be reused by the PPU. * This is done by writting a 16-byte batch-buffer-status block back into - * main memory (in cell_contex->buffer_status[]). + * main memory (in cell_context->buffer_status[]). */ static void -release_batch_buffer(uint buffer) +release_buffer(uint buffer) { /* Evidently, using less than a 16-byte status doesn't work reliably */ static const uint status[4] ALIGN16_ATTRIB = {CELL_BUFFER_STATUS_FREE, 0, 0, 0}; - const uint index = 4 * (spu.init.id * CELL_NUM_BATCH_BUFFERS + buffer); + const uint index = 4 * (spu.init.id * CELL_NUM_BUFFERS + buffer); uint *dst = spu.init.buffer_status + index; - ASSERT(buffer < CELL_NUM_BATCH_BUFFERS); + ASSERT(buffer < CELL_NUM_BUFFERS); /* printf("SPU %u: Set batch status buf=%u, index %u, at %p to FREE\n", @@ -513,24 +513,24 @@ cmd_batch(uint opcode) { const uint buf = (opcode >> 8) & 0xff; uint size = (opcode >> 16); - uint buffer[CELL_BATCH_BUFFER_SIZE / 4] ALIGN16_ATTRIB; + uint buffer[CELL_BUFFER_SIZE / 4] ALIGN16_ATTRIB; const uint usize = size / sizeof(uint); uint pos; if (Debug) printf("SPU %u: BATCH buffer %u, len %u, from %p\n", - spu.init.id, buf, size, spu.init.batch_buffers[buf]); + spu.init.id, buf, size, spu.init.buffers[buf]); ASSERT((opcode & CELL_CMD_OPCODE_MASK) == CELL_CMD_BATCH); - ASSERT_ALIGN16(spu.init.batch_buffers[buf]); + ASSERT_ALIGN16(spu.init.buffers[buf]); size = ROUNDUP16(size); - ASSERT_ALIGN16(spu.init.batch_buffers[buf]); + ASSERT_ALIGN16(spu.init.buffers[buf]); mfc_get(buffer, /* dest */ - (unsigned int) spu.init.batch_buffers[buf], /* src */ + (unsigned int) spu.init.buffers[buf], /* src */ size, TAG_BATCH_BUFFER, 0, /* tid */ @@ -538,7 +538,7 @@ cmd_batch(uint opcode) wait_on_mask(1 << TAG_BATCH_BUFFER); /* Tell PPU we're done copying the buffer to local store */ - release_batch_buffer(buf); + release_buffer(buf); for (pos = 0; pos < usize; /* no incr */) { switch (buffer[pos]) { -- cgit v1.2.3 From 87c8f9c5834b7345615257d0faf5200f191e8eca Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 09:57:13 -0700 Subject: Cell: additional debug code, misc clean-up --- src/mesa/pipe/cell/ppu/cell_batch.c | 52 +++++++++++++++++++++++++++++-------- src/mesa/pipe/cell/ppu/cell_batch.h | 2 +- 2 files changed, 42 insertions(+), 12 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_batch.c b/src/mesa/pipe/cell/ppu/cell_batch.c index 178caa74e1b..2d032fc9026 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.c +++ b/src/mesa/pipe/cell/ppu/cell_batch.c @@ -35,7 +35,7 @@ uint cell_get_empty_buffer(struct cell_context *cell) { - uint buf = 0; + uint buf = 0, tries = 0; /* Find a buffer that's marked as free by all SPUs */ while (1) { @@ -50,6 +50,9 @@ cell_get_empty_buffer(struct cell_context *cell) for (spu = 0; spu < cell->num_spus; spu++) { cell->buffer_status[spu][buf][0] = CELL_BUFFER_STATUS_USED; } + /* + printf("PPU: ALLOC BUFFER %u\n", buf); + */ return buf; } } @@ -60,11 +63,17 @@ cell_get_empty_buffer(struct cell_context *cell) /* try next buf */ buf = (buf + 1) % CELL_NUM_BUFFERS; + + tries++; + if (tries == 100) { + /* + printf("PPU WAITING for buffer...\n"); + */ + } } } - void cell_batch_flush(struct cell_context *cell) { @@ -120,29 +129,39 @@ cell_batch_free_space(const struct cell_context *cell) /** - * \param cmd command to append - * \param length command size in bytes + * Append data to current batch. */ void -cell_batch_append(struct cell_context *cell, const void *cmd, uint length) +cell_batch_append(struct cell_context *cell, const void *data, uint bytes) { uint size; - assert(length % 4 == 0); - assert(cell->cur_batch >= 0); + ASSERT(bytes % 4 == 0); + ASSERT(bytes <= CELL_BUFFER_SIZE); + ASSERT(cell->cur_batch >= 0); + +#ifdef ASSERT + { + uint spu; + for (spu = 0; spu < cell->num_spus; spu++) { + ASSERT(cell->buffer_status[spu][cell->cur_batch][0] + == CELL_BUFFER_STATUS_USED); + } + } +#endif size = cell->buffer_size[cell->cur_batch]; - if (size + length > CELL_BUFFER_SIZE) { + if (size + bytes > CELL_BUFFER_SIZE) { cell_batch_flush(cell); size = 0; } - assert(size + length <= CELL_BUFFER_SIZE); + assert(size + bytes <= CELL_BUFFER_SIZE); - memcpy(cell->buffer[cell->cur_batch] + size, cmd, length); + memcpy(cell->buffer[cell->cur_batch] + size, data, bytes); - cell->buffer_size[cell->cur_batch] = size + length; + cell->buffer_size[cell->cur_batch] = size + bytes; } @@ -153,9 +172,20 @@ cell_batch_alloc(struct cell_context *cell, uint bytes) uint size; ASSERT(bytes % 4 == 0); + ASSERT(bytes <= CELL_BUFFER_SIZE); assert(cell->cur_batch >= 0); +#ifdef ASSERT + { + uint spu; + for (spu = 0; spu < cell->num_spus; spu++) { + ASSERT(cell->buffer_status[spu][cell->cur_batch][0] + == CELL_BUFFER_STATUS_USED); + } + } +#endif + size = cell->buffer_size[cell->cur_batch]; if (size + bytes > CELL_BUFFER_SIZE) { diff --git a/src/mesa/pipe/cell/ppu/cell_batch.h b/src/mesa/pipe/cell/ppu/cell_batch.h index b4c96f465a8..f4f37314a48 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.h +++ b/src/mesa/pipe/cell/ppu/cell_batch.h @@ -45,7 +45,7 @@ extern uint cell_batch_free_space(const struct cell_context *cell); extern void -cell_batch_append(struct cell_context *cell, const void *cmd, uint length); +cell_batch_append(struct cell_context *cell, const void *data, uint bytes); extern void * cell_batch_alloc(struct cell_context *cell, uint bytes); -- cgit v1.2.3 From 4f0906a18a0067d7e16c4fc7602dfb280e60f420 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 09:57:51 -0700 Subject: Cell: If flushing for swapbuffers, wait for frame completion --- src/mesa/pipe/cell/ppu/cell_flush.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/mesa/pipe/cell/ppu/cell_flush.c b/src/mesa/pipe/cell/ppu/cell_flush.c index b98bb566b12..cf4e676645b 100644 --- a/src/mesa/pipe/cell/ppu/cell_flush.c +++ b/src/mesa/pipe/cell/ppu/cell_flush.c @@ -39,6 +39,9 @@ cell_flush(struct pipe_context *pipe, unsigned flags) { struct cell_context *cell = cell_context(pipe); + if (flags & PIPE_FLUSH_SWAPBUFFERS) + flags |= PIPE_FLUSH_WAIT; + draw_flush( cell->draw ); cell_flush_int(pipe, flags); } -- cgit v1.2.3 From 9abbaacea6f340486a3b2bf68fbd4efa0d15a5d3 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 10:00:27 -0700 Subject: Cell: checkpoint commit: always inline prim indexes into batch buffer Also, explicit release-vertex-buffer command. Lots of debug/stale code still in place... --- src/mesa/pipe/cell/common.h | 12 ++++ src/mesa/pipe/cell/ppu/cell_vbuf.c | 113 ++++++++++++++++++++++++++----------- src/mesa/pipe/cell/spu/spu_main.c | 110 +++++++++++++++++++++++++----------- src/mesa/pipe/cell/spu/spu_main.h | 2 + 4 files changed, 171 insertions(+), 66 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index ce9c3819077..31637ed1cc8 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -75,6 +75,7 @@ #define CELL_CMD_FINISH 3 #define CELL_CMD_RENDER 4 #define CELL_CMD_BATCH 5 +#define CELL_CMD_RELEASE_VERTS 6 #define CELL_CMD_STATE_FRAMEBUFFER 10 #define CELL_CMD_STATE_DEPTH_STENCIL 11 #define CELL_CMD_STATE_SAMPLER 12 @@ -124,7 +125,11 @@ struct cell_command_render uint vertex_size; /**< bytes per vertex */ uint dummy; /* XXX this dummy field works around a compiler bug */ uint num_indexes; +#if 0 const void *vertex_data; +#else + uint vertex_buf; /**< which cell->buffer[] contains the vertex data */ +#endif const ushort *index_data; float xmin, ymin, xmax, ymax; boolean inline_indexes; @@ -132,6 +137,13 @@ struct cell_command_render } ALIGN16_ATTRIB; +struct cell_command_release_verts +{ + int opcode; /**< CELL_CMD_RELEASE_VERTS */ + uint vertex_buf; /**< in [0, CELL_NUM_BUFFERS-1] */ +}; + + /** XXX unions don't seem to work */ struct cell_command { diff --git a/src/mesa/pipe/cell/ppu/cell_vbuf.c b/src/mesa/pipe/cell/ppu/cell_vbuf.c index ee572b3a51b..6e12e16fe0a 100644 --- a/src/mesa/pipe/cell/ppu/cell_vbuf.c +++ b/src/mesa/pipe/cell/ppu/cell_vbuf.c @@ -40,8 +40,8 @@ /** Allow prim indexes, verts to be inlined after RENDER command */ -#define ALLOW_INLINE_INDEXES 1 -#define ALLOW_INLINE_VERTS 1 +#define ALLOW_INLINE_INDEXES 01 +#define ALLOW_INLINE_VERTS 0 /** @@ -55,6 +55,9 @@ struct cell_vbuf_render uint prim; uint vertex_size; void *vertex_buffer; +#if 1 + uint vertex_buf; +#endif }; @@ -81,13 +84,52 @@ cell_vbuf_allocate_vertices(struct vbuf_render *vbr, { struct cell_vbuf_render *cvbr = cell_vbuf_render(vbr); /*printf("Alloc verts %u * %u\n", vertex_size, nr_vertices);*/ +#if 0 assert(!cvbr->vertex_buffer); cvbr->vertex_buffer = align_malloc(vertex_size * nr_vertices, 16); +#else + assert(cvbr->vertex_buf == ~0); + cvbr->vertex_buf = cell_get_empty_buffer(cvbr->cell); + cvbr->vertex_buffer = cvbr->cell->buffer[cvbr->vertex_buf]; + printf("%s vertex_buf = %u\n", __FUNCTION__, cvbr->vertex_buf); +#endif cvbr->vertex_size = vertex_size; return cvbr->vertex_buffer; } +static void +cell_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices, + unsigned vertex_size, unsigned vertices_used) +{ + struct cell_vbuf_render *cvbr = cell_vbuf_render(vbr); + struct cell_context *cell = cvbr->cell; + + /*printf("Free verts %u * %u\n", vertex_size, vertices_used);*/ +#if 0 + align_free(vertices); +#else + printf("%s vertex_buf = %u count = %u\n", + __FUNCTION__, cvbr->vertex_buf, vertices_used); + + { + struct cell_command_release_verts *release + = (struct cell_command_release_verts *) + cell_batch_alloc(cell, sizeof(struct cell_command_release_verts)); + release->opcode = CELL_CMD_RELEASE_VERTS; + release->vertex_buf = cvbr->vertex_buf; + } + + cvbr->vertex_buf = ~0; + cell_flush_int(&cell->pipe, 0x0);/*NEW*/ +#endif + + assert(vertices == cvbr->vertex_buffer); + cvbr->vertex_buffer = NULL; +} + + + static void cell_vbuf_set_primitive(struct vbuf_render *vbr, unsigned prim) { @@ -124,7 +166,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, printf("%u %u %u, ", indices[i+0], indices[i+1], indices[i+2]); } printf("\n"); -#elif 0 +#elif 01 printf("cell_vbuf_draw() nr_indices = %u nr_verts = %u indexes = [%u %u %u ...]\n", nr_indices, nr_vertices, indices[0], indices[1], indices[2]); @@ -157,28 +199,26 @@ cell_vbuf_draw(struct vbuf_render *vbr, const uint index_bytes = ROUNDUP4(nr_indices * 2); const uint vertex_bytes = nr_vertices * 4 * cell->vertex_info.size; + const uint batch_size = sizeof(struct cell_command_render) + + index_bytes; + struct cell_command_render *render = (struct cell_command_render *) - cell_batch_alloc(cell, sizeof(*render)); + cell_batch_alloc(cell, batch_size); + render->opcode = CELL_CMD_RENDER; render->prim_type = cvbr->prim; render->num_indexes = nr_indices; - if (ALLOW_INLINE_INDEXES && - index_bytes <= cell_batch_free_space(cell)) { - /* indices inlined, right after render cmd */ - void *dst = cell_batch_alloc(cell, index_bytes); - memcpy(dst, indices, nr_indices * 2); - render->inline_indexes = TRUE; - render->index_data = NULL; - } - else { - /* indices in separate buffer */ - render->inline_indexes = FALSE; - render->index_data = indices; - ASSERT_ALIGN16(render->index_data); - } + /* append indices after render command */ + memcpy(render + 1, indices, nr_indices * 2); + render->inline_indexes = TRUE; + render->index_data = NULL; + + /* if there's room, append vertices after the indices, else leave + * vertices in the original/separate buffer. + */ render->vertex_size = 4 * cell->vertex_info.size; render->num_verts = nr_vertices; if (ALLOW_INLINE_VERTS && @@ -188,12 +228,21 @@ cell_vbuf_draw(struct vbuf_render *vbr, void *dst = cell_batch_alloc(cell, vertex_bytes); memcpy(dst, vertices, vertex_bytes); render->inline_verts = TRUE; +#if 0 render->vertex_data = NULL; +#else + render->vertex_buf = ~0; +#endif } else { render->inline_verts = FALSE; +#if 0 render->vertex_data = vertices; ASSERT_ALIGN16(render->vertex_data); +#else + ASSERT(cvbr->vertex_buf >= 0); + render->vertex_buf = cvbr->vertex_buf; +#endif } @@ -203,27 +252,13 @@ cell_vbuf_draw(struct vbuf_render *vbr, render->ymax = ymax; } -#if 01 +#if 0 /* XXX this is temporary */ cell_flush_int(&cell->pipe, PIPE_FLUSH_WAIT); #endif } -static void -cell_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices, - unsigned vertex_size, unsigned vertices_used) -{ - struct cell_vbuf_render *cvbr = cell_vbuf_render(vbr); - - /*printf("Free verts %u * %u\n", vertex_size, vertices_used);*/ - align_free(vertices); - - assert(vertices == cvbr->vertex_buffer); - cvbr->vertex_buffer = NULL; -} - - static void cell_vbuf_destroy(struct vbuf_render *vbr) { @@ -244,8 +279,17 @@ cell_init_vbuf(struct cell_context *cell) cell->vbuf_render = CALLOC_STRUCT(cell_vbuf_render); +#if 0 cell->vbuf_render->base.max_indices = CELL_MAX_VBUF_INDEXES; cell->vbuf_render->base.max_vertex_buffer_bytes = CELL_MAX_VBUF_SIZE; +#else + cell->vbuf_render->base.max_indices + = (CELL_BUFFER_SIZE + - sizeof(struct cell_command_render) + - sizeof(struct cell_command_release_verts)) + / sizeof(ushort); + cell->vbuf_render->base.max_vertex_buffer_bytes = CELL_BUFFER_SIZE; +#endif cell->vbuf_render->base.get_vertex_info = cell_vbuf_get_vertex_info; cell->vbuf_render->base.allocate_vertices = cell_vbuf_allocate_vertices; @@ -255,6 +299,9 @@ cell_init_vbuf(struct cell_context *cell) cell->vbuf_render->base.destroy = cell_vbuf_destroy; cell->vbuf_render->cell = cell; +#if 1 + cell->vbuf_render->vertex_buf = ~0; +#endif cell->vbuf = draw_vbuf_stage(cell->draw, &cell->vbuf_render->base); } diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 2097683b825..eb979718f85 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -69,6 +69,32 @@ wait_on_mask_all(unsigned tagMask) } +/** + * Tell the PPU that this SPU has finished copying a buffer to + * local store and that it may be reused by the PPU. + * This is done by writting a 16-byte batch-buffer-status block back into + * main memory (in cell_context->buffer_status[]). + */ +static void +release_buffer(uint buffer) +{ + /* Evidently, using less than a 16-byte status doesn't work reliably */ + static const uint status[4] ALIGN16_ATTRIB + = {CELL_BUFFER_STATUS_FREE, 0, 0, 0}; + + const uint index = 4 * (spu.init.id * CELL_NUM_BUFFERS + buffer); + uint *dst = spu.init.buffer_status + index; + + ASSERT(buffer < CELL_NUM_BUFFERS); + + mfc_put((void *) &status, /* src in local memory */ + (unsigned int) dst, /* dst in main memory */ + sizeof(status), /* size */ + TAG_MISC, /* tag is unimportant */ + 0, /* tid */ + 0 /* rid */); +} + /** * For tiles whose status is TILE_STATUS_CLEAR, write solid-filled @@ -237,13 +263,18 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) printf(" bound: %g, %g .. %g, %g\n", render->xmin, render->ymin, render->xmax, render->ymax); */ + /* printf("SPU %u: indices at %p vertices at %p\n", spu.init.id, render->index_data, render->vertex_data); + */ } ASSERT(sizeof(*render) % 4 == 0); +#if 0 ASSERT_ALIGN16(render->vertex_data); +#else +#endif ASSERT_ALIGN16(render->index_data); @@ -251,10 +282,18 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) ** Get vertex, index buffers if not inlined **/ if (!render->inline_verts) { + void *src; ASSERT(total_vertex_bytes % 16 == 0); +#if 0 + src = render->vertex_data; +#else + spu.cur_vertex_buf = render->vertex_buf; + src = spu.init.buffers[render->vertex_buf]; +#endif + mfc_get(vertex_data, /* dest */ - (unsigned int) render->vertex_data, /* src */ + (unsigned int) src, total_vertex_bytes, /* size */ TAG_VERTEX_BUFFER, 0, /* tid */ @@ -298,6 +337,7 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) /* vertices are after indexes, if inlined */ vertices = (const ubyte *) (render + 1) + *pos_incr * 4; *pos_incr = *pos_incr + total_vertex_bytes / 4; + spu.cur_vertex_buf = ~0; } } @@ -310,6 +350,12 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) mask |= (1 << TAG_INDEX_BUFFER); wait_on_mask_all(mask); +#if 0 + if (!render->inline_verts) { + printf("SPU %u: release vbuf %u\n", spu.init.id, render->vertex_buf); + release_buffer(render->vertex_buf); + } +#endif /** ** find tiles which intersect the prim bounding box @@ -359,6 +405,14 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) for (j = 0; j < render->num_indexes; j += 3) { const float *v0, *v1, *v2; + if (indexes[j] == 0xffff) { + printf("index[%u] = 0xffff\n", j); + } + + ASSERT(indexes[j] != 0xffff); + ASSERT(indexes[j+1] != 0xffff); + ASSERT(indexes[j+2] != 0xffff); + v0 = (const float *) (vertices + indexes[j+0] * vertex_size); v1 = (const float *) (vertices + indexes[j+1] * vertex_size); v2 = (const float *) (vertices + indexes[j+2] * vertex_size); @@ -391,6 +445,17 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) } +static void +cmd_release_verts(const struct cell_command_release_verts *release) +{ + if (Debug) + printf("SPU %u: RELEASE VERTS %u\n", + spu.init.id, spu.cur_vertex_buf); + ASSERT(spu.cur_vertex_buf == release->vertex_buf); + release_buffer(release->vertex_buf); +} + + static void cmd_state_framebuffer(const struct cell_command_framebuffer *cmd) { @@ -472,38 +537,6 @@ cmd_finish(void) } -/** - * Tell the PPU that this SPU has finished copying a buffer to - * local store and that it may be reused by the PPU. - * This is done by writting a 16-byte batch-buffer-status block back into - * main memory (in cell_context->buffer_status[]). - */ -static void -release_buffer(uint buffer) -{ - /* Evidently, using less than a 16-byte status doesn't work reliably */ - static const uint status[4] ALIGN16_ATTRIB - = {CELL_BUFFER_STATUS_FREE, 0, 0, 0}; - - const uint index = 4 * (spu.init.id * CELL_NUM_BUFFERS + buffer); - uint *dst = spu.init.buffer_status + index; - - ASSERT(buffer < CELL_NUM_BUFFERS); - - /* - printf("SPU %u: Set batch status buf=%u, index %u, at %p to FREE\n", - spu.init.id, buffer, index, dst); - */ - - mfc_put((void *) &status, /* src in local memory */ - (unsigned int) dst, /* dst in main memory */ - sizeof(status), /* size */ - TAG_MISC, /* tag is unimportant */ - 0, /* tid */ - 0 /* rid */); -} - - /** * Execute a batch of commands * The opcode param encodes the location of the buffer and its size. @@ -538,6 +571,8 @@ cmd_batch(uint opcode) wait_on_mask(1 << TAG_BATCH_BUFFER); /* Tell PPU we're done copying the buffer to local store */ + if (Debug) + printf("SPU %u: release batch buf %u\n", spu.init.id, buf); release_buffer(buf); for (pos = 0; pos < usize; /* no incr */) { @@ -567,6 +602,15 @@ cmd_batch(uint opcode) pos += sizeof(*render) / 4 + pos_incr; } break; + case CELL_CMD_RELEASE_VERTS: + { + struct cell_command_release_verts *release + = (struct cell_command_release_verts *) &buffer[pos]; + cmd_release_verts(release); + ASSERT(sizeof(*release) == 8); + pos += sizeof(*release) / 4; + } + break; case CELL_CMD_FINISH: cmd_finish(); pos += 1; diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 5bc5d9fa99d..68c7263b7f9 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -65,6 +65,8 @@ struct spu_global /* XXX more state to come */ + uint cur_vertex_buf; + } ALIGN16_ATTRIB; -- cgit v1.2.3 From c4ef36dec0aa5b8cd0293a6b12689bb68ad67ac5 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 10:41:27 -0700 Subject: Cell: clean-up of render path Finally removed a number of unneeded flush commands. Vertex buffers are allocated from the general buffer pool, freed by SPUs when done. Still an occasional failed assertion (invalid batch buffer command)... --- src/mesa/pipe/cell/common.h | 12 +--- src/mesa/pipe/cell/ppu/cell_vbuf.c | 60 ++++++-------------- src/mesa/pipe/cell/spu/spu_main.c | 112 +++++++------------------------------ src/mesa/pipe/cell/spu/spu_main.h | 2 - 4 files changed, 38 insertions(+), 148 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 31637ed1cc8..d6e1dd4f7d5 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -68,7 +68,7 @@ * The low byte of a mailbox word contains the command opcode. * Remaining higher bytes are command specific. */ -#define CELL_CMD_OPCODE_MASK 0xf +#define CELL_CMD_OPCODE_MASK 0xff #define CELL_CMD_EXIT 1 #define CELL_CMD_CLEAR_SURFACE 2 @@ -113,10 +113,6 @@ struct cell_command_clear_surface } ALIGN16_ATTRIB; -#define CELL_MAX_VBUF_SIZE (16 * 1024) -#define CELL_MAX_VBUF_INDEXES 1024 - - struct cell_command_render { uint opcode; /**< CELL_CMD_RENDER */ @@ -125,14 +121,8 @@ struct cell_command_render uint vertex_size; /**< bytes per vertex */ uint dummy; /* XXX this dummy field works around a compiler bug */ uint num_indexes; -#if 0 - const void *vertex_data; -#else uint vertex_buf; /**< which cell->buffer[] contains the vertex data */ -#endif - const ushort *index_data; float xmin, ymin, xmax, ymax; - boolean inline_indexes; boolean inline_verts; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/ppu/cell_vbuf.c b/src/mesa/pipe/cell/ppu/cell_vbuf.c index 6e12e16fe0a..b2a25d767b7 100644 --- a/src/mesa/pipe/cell/ppu/cell_vbuf.c +++ b/src/mesa/pipe/cell/ppu/cell_vbuf.c @@ -39,9 +39,8 @@ #include "pipe/draw/draw_vbuf.h" -/** Allow prim indexes, verts to be inlined after RENDER command */ -#define ALLOW_INLINE_INDEXES 01 -#define ALLOW_INLINE_VERTS 0 +/** Allow vertex data to be inlined after RENDER command */ +#define ALLOW_INLINE_VERTS 1 /** @@ -52,12 +51,10 @@ struct cell_vbuf_render { struct vbuf_render base; struct cell_context *cell; - uint prim; - uint vertex_size; - void *vertex_buffer; -#if 1 - uint vertex_buf; -#endif + uint prim; /**< PIPE_PRIM_x */ + uint vertex_size; /**< in bytes */ + void *vertex_buffer; /**< just for debug, really */ + uint vertex_buf; /**< in [0, CELL_NUM_BUFFERS-1] */ }; @@ -84,15 +81,10 @@ cell_vbuf_allocate_vertices(struct vbuf_render *vbr, { struct cell_vbuf_render *cvbr = cell_vbuf_render(vbr); /*printf("Alloc verts %u * %u\n", vertex_size, nr_vertices);*/ -#if 0 - assert(!cvbr->vertex_buffer); - cvbr->vertex_buffer = align_malloc(vertex_size * nr_vertices, 16); -#else + assert(cvbr->vertex_buf == ~0); cvbr->vertex_buf = cell_get_empty_buffer(cvbr->cell); cvbr->vertex_buffer = cvbr->cell->buffer[cvbr->vertex_buf]; - printf("%s vertex_buf = %u\n", __FUNCTION__, cvbr->vertex_buf); -#endif cvbr->vertex_size = vertex_size; return cvbr->vertex_buffer; } @@ -105,14 +97,13 @@ cell_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices, struct cell_vbuf_render *cvbr = cell_vbuf_render(vbr); struct cell_context *cell = cvbr->cell; - /*printf("Free verts %u * %u\n", vertex_size, vertices_used);*/ -#if 0 - align_free(vertices); -#else + /* printf("%s vertex_buf = %u count = %u\n", __FUNCTION__, cvbr->vertex_buf, vertices_used); + */ - { + /* Tell SPUs they can release the vert buf */ + if (cvbr->vertex_buf != ~0U) { struct cell_command_release_verts *release = (struct cell_command_release_verts *) cell_batch_alloc(cell, sizeof(struct cell_command_release_verts)); @@ -121,8 +112,7 @@ cell_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices, } cvbr->vertex_buf = ~0; - cell_flush_int(&cell->pipe, 0x0);/*NEW*/ -#endif + cell_flush_int(&cell->pipe, 0x0); assert(vertices == cvbr->vertex_buffer); cvbr->vertex_buffer = NULL; @@ -166,7 +156,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, printf("%u %u %u, ", indices[i+0], indices[i+1], indices[i+2]); } printf("\n"); -#elif 01 +#elif 0 printf("cell_vbuf_draw() nr_indices = %u nr_verts = %u indexes = [%u %u %u ...]\n", nr_indices, nr_vertices, indices[0], indices[1], indices[2]); @@ -213,8 +203,6 @@ cell_vbuf_draw(struct vbuf_render *vbr, /* append indices after render command */ memcpy(render + 1, indices, nr_indices * 2); - render->inline_indexes = TRUE; - render->index_data = NULL; /* if there's room, append vertices after the indices, else leave * vertices in the original/separate buffer. @@ -222,30 +210,20 @@ cell_vbuf_draw(struct vbuf_render *vbr, render->vertex_size = 4 * cell->vertex_info.size; render->num_verts = nr_vertices; if (ALLOW_INLINE_VERTS && - render->inline_indexes && vertex_bytes <= cell_batch_free_space(cell)) { /* vertex data inlined, after indices */ void *dst = cell_batch_alloc(cell, vertex_bytes); memcpy(dst, vertices, vertex_bytes); render->inline_verts = TRUE; -#if 0 - render->vertex_data = NULL; -#else render->vertex_buf = ~0; -#endif } else { + /* vertex data in separate buffer */ render->inline_verts = FALSE; -#if 0 - render->vertex_data = vertices; - ASSERT_ALIGN16(render->vertex_data); -#else ASSERT(cvbr->vertex_buf >= 0); render->vertex_buf = cvbr->vertex_buf; -#endif } - render->xmin = xmin; render->ymin = ymin; render->xmax = xmax; @@ -253,7 +231,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, } #if 0 - /* XXX this is temporary */ + /* helpful for debug */ cell_flush_int(&cell->pipe, PIPE_FLUSH_WAIT); #endif } @@ -279,17 +257,15 @@ cell_init_vbuf(struct cell_context *cell) cell->vbuf_render = CALLOC_STRUCT(cell_vbuf_render); -#if 0 - cell->vbuf_render->base.max_indices = CELL_MAX_VBUF_INDEXES; - cell->vbuf_render->base.max_vertex_buffer_bytes = CELL_MAX_VBUF_SIZE; -#else + /* The max number of indexes is what can fix into a batch buffer, + * minus the render and release-verts commands. + */ cell->vbuf_render->base.max_indices = (CELL_BUFFER_SIZE - sizeof(struct cell_command_render) - sizeof(struct cell_command_release_verts)) / sizeof(ushort); cell->vbuf_render->base.max_vertex_buffer_bytes = CELL_BUFFER_SIZE; -#endif cell->vbuf_render->base.get_vertex_info = cell_vbuf_get_vertex_info; cell->vbuf_render->base.allocate_vertices = cell_vbuf_allocate_vertices; diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index eb979718f85..5b50ec6953b 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -239,59 +239,45 @@ static void cmd_render(const struct cell_command_render *render, uint *pos_incr) { /* we'll DMA into these buffers */ - ubyte vertex_data[CELL_MAX_VBUF_SIZE] ALIGN16_ATTRIB; - ushort index_data[CELL_MAX_VBUF_INDEXES] ALIGN16_ATTRIB; + ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB; const uint vertex_size = render->vertex_size; /* in bytes */ const uint total_vertex_bytes = render->num_verts * vertex_size; const ubyte *vertices; const ushort *indexes; - uint mask; uint i, j; if (Debug) { printf("SPU %u: RENDER prim %u, num_vert=%u num_ind=%u " - "inline_vert=%u inline_ind=%u\n", + "inline_vert=%u\n", spu.init.id, render->prim_type, render->num_verts, render->num_indexes, - render->inline_verts, - render->inline_indexes); + render->inline_verts); /* printf(" bound: %g, %g .. %g, %g\n", render->xmin, render->ymin, render->xmax, render->ymax); */ - /* - printf("SPU %u: indices at %p vertices at %p\n", - spu.init.id, - render->index_data, render->vertex_data); - */ } ASSERT(sizeof(*render) % 4 == 0); -#if 0 - ASSERT_ALIGN16(render->vertex_data); -#else -#endif - ASSERT_ALIGN16(render->index_data); + ASSERT(total_vertex_bytes % 16 == 0); + /* indexes are right after the render command in the batch buffer */ + indexes = (const ushort *) (render + 1); + *pos_incr = (render->num_indexes * 2 + 3) / 4; - /** - ** Get vertex, index buffers if not inlined - **/ - if (!render->inline_verts) { - void *src; - ASSERT(total_vertex_bytes % 16 == 0); - -#if 0 - src = render->vertex_data; -#else - spu.cur_vertex_buf = render->vertex_buf; - src = spu.init.buffers[render->vertex_buf]; -#endif + if (render->inline_verts) { + /* Vertices are right after indexes in batch buffer */ + vertices = (const ubyte *) (render + 1) + *pos_incr * 4; + *pos_incr = *pos_incr + total_vertex_bytes / 4; + } + else { + /* Begin DMA fetch of vertex buffer */ + void *src = spu.init.buffers[render->vertex_buf]; mfc_get(vertex_data, /* dest */ (unsigned int) src, total_vertex_bytes, /* size */ @@ -300,63 +286,11 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) 0 /* rid */); vertices = vertex_data; - } - - if (!render->inline_indexes) { - uint total_index_bytes; - - *pos_incr = 0; - - total_index_bytes = render->num_indexes * sizeof(ushort); - if (total_index_bytes < 16) - total_index_bytes = 16; - else - total_index_bytes = ROUNDUP16(total_index_bytes); - indexes = index_data; - - /* get index data from main memory */ - mfc_get(index_data, /* dest */ - (unsigned int) render->index_data, /* src */ - total_index_bytes, - TAG_INDEX_BUFFER, - 0, /* tid */ - 0 /* rid */); - } - - - /** - ** Get pointers to inlined indexes, verts, if present - **/ - if (render->inline_indexes) { - /* indexes are right after the render command in the batch buffer */ - indexes = (ushort *) (render + 1); - *pos_incr = (render->num_indexes * 2 + 3) / 4; - - if (render->inline_verts) { - /* vertices are after indexes, if inlined */ - vertices = (const ubyte *) (render + 1) + *pos_incr * 4; - *pos_incr = *pos_incr + total_vertex_bytes / 4; - spu.cur_vertex_buf = ~0; - } + wait_on_mask(1 << TAG_VERTEX_BUFFER); } - /* wait for vertex and/or index buffers if not inlined */ - mask = 0x0; - if (!render->inline_verts) - mask |= (1 << TAG_VERTEX_BUFFER); - if (!render->inline_indexes) - mask |= (1 << TAG_INDEX_BUFFER); - wait_on_mask_all(mask); - -#if 0 - if (!render->inline_verts) { - printf("SPU %u: release vbuf %u\n", spu.init.id, render->vertex_buf); - release_buffer(render->vertex_buf); - } -#endif - /** ** find tiles which intersect the prim bounding box **/ @@ -372,7 +306,7 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) #endif /* make sure any pending clears have completed */ - wait_on_mask(1 << TAG_SURFACE_CLEAR); + wait_on_mask(1 << TAG_SURFACE_CLEAR); /* XXX temporary */ /** @@ -405,14 +339,6 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) for (j = 0; j < render->num_indexes; j += 3) { const float *v0, *v1, *v2; - if (indexes[j] == 0xffff) { - printf("index[%u] = 0xffff\n", j); - } - - ASSERT(indexes[j] != 0xffff); - ASSERT(indexes[j+1] != 0xffff); - ASSERT(indexes[j+2] != 0xffff); - v0 = (const float *) (vertices + indexes[j+0] * vertex_size); v1 = (const float *) (vertices + indexes[j+1] * vertex_size); v2 = (const float *) (vertices + indexes[j+2] * vertex_size); @@ -450,8 +376,8 @@ cmd_release_verts(const struct cell_command_release_verts *release) { if (Debug) printf("SPU %u: RELEASE VERTS %u\n", - spu.init.id, spu.cur_vertex_buf); - ASSERT(spu.cur_vertex_buf == release->vertex_buf); + spu.init.id, release->vertex_buf); + ASSERT(release->vertex_buf != ~0U); release_buffer(release->vertex_buf); } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 68c7263b7f9..5bc5d9fa99d 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -65,8 +65,6 @@ struct spu_global /* XXX more state to come */ - uint cur_vertex_buf; - } ALIGN16_ATTRIB; -- cgit v1.2.3 From 1cbe803922e1129d1077bcc1eb0640bf9204641d Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 10:41:51 -0700 Subject: Cell: remove unneeded flush(), dead code --- src/mesa/pipe/cell/ppu/cell_clear.c | 30 +----------------------------- 1 file changed, 1 insertion(+), 29 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_clear.c b/src/mesa/pipe/cell/ppu/cell_clear.c index e01640b9944..e61bfd9b0fd 100644 --- a/src/mesa/pipe/cell/ppu/cell_clear.c +++ b/src/mesa/pipe/cell/ppu/cell_clear.c @@ -48,7 +48,6 @@ cell_clear_surface(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { struct cell_context *cell = cell_context(pipe); - /*uint i;*/ uint surfIndex; if (!cell->cbuf_map[0]) @@ -61,29 +60,7 @@ cell_clear_surface(struct pipe_context *pipe, struct pipe_surface *ps, surfIndex = 0; } -#if 0 - for (i = 0; i < cell->num_spus; i++) { -#if 1 - uint clr = clearValue; - if (surfIndex == 0) { - /* XXX debug: clear color varied per-SPU to visualize tiles */ - if ((clr & 0xff) == 0) - clr |= 64 + i * 8; - if ((clr & 0xff00) == 0) - clr |= (64 + i * 8) << 8; - if ((clr & 0xff0000) == 0) - clr |= (64 + i * 8) << 16; - if ((clr & 0xff000000) == 0) - clr |= (64 + i * 8) << 24; - } - cell_global.command[i].clear.value = clr; -#else - cell_global.command[i].clear.value = clearValue; -#endif - cell_global.command[i].clear.surface = surfIndex; - send_mbox_message(cell_global.spe_contexts[i], CELL_CMD_CLEAR_SURFACE); - } -#else + { struct cell_command_clear_surface *clr = (struct cell_command_clear_surface *) @@ -92,9 +69,4 @@ cell_clear_surface(struct pipe_context *pipe, struct pipe_surface *ps, clr->surface = surfIndex; clr->value = clearValue; } -#endif - - /* XXX temporary */ - cell_flush(&cell->pipe, 0x0); - } -- cgit v1.2.3 From ca85eed771c1fb1c662fb8eea2535601e73c437d Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 11:20:47 -0700 Subject: Cell: make sure state commands aren't split across batches --- src/mesa/pipe/cell/ppu/cell_state_emit.c | 29 ++++++++++++++++++----------- 1 file changed, 18 insertions(+), 11 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_state_emit.c b/src/mesa/pipe/cell/ppu/cell_state_emit.c index dbca900c35f..6776ec88c7b 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_emit.c +++ b/src/mesa/pipe/cell/ppu/cell_state_emit.c @@ -33,6 +33,17 @@ +static void +emit_state_cmd(struct cell_context *cell, uint cmd, + const void *state, uint state_size) +{ + uint *dst = (uint *) cell_batch_alloc(cell, sizeof(uint) + state_size); + *dst = cmd; + memcpy(dst + 1, state, state_size); +} + + + void cell_emit_state(struct cell_context *cell) { @@ -51,22 +62,18 @@ cell_emit_state(struct cell_context *cell) } if (cell->dirty & CELL_NEW_DEPTH_STENCIL) { - uint cmd = CELL_CMD_STATE_DEPTH_STENCIL; - cell_batch_append(cell, &cmd, 4); - cell_batch_append(cell, cell->depth_stencil, - sizeof(struct pipe_depth_stencil_alpha_state)); + emit_state_cmd(cell, CELL_CMD_STATE_DEPTH_STENCIL, + cell->depth_stencil, + sizeof(struct pipe_depth_stencil_alpha_state)); } if (cell->dirty & CELL_NEW_SAMPLER) { - uint cmd = CELL_CMD_STATE_SAMPLER; - cell_batch_append(cell, &cmd, 4); - cell_batch_append(cell, cell->sampler[0], - sizeof(struct pipe_sampler_state)); + emit_state_cmd(cell, CELL_CMD_STATE_SAMPLER, + cell->sampler[0], sizeof(struct pipe_sampler_state)); } if (cell->dirty & CELL_NEW_VERTEX_INFO) { - uint cmd = CELL_CMD_STATE_VERTEX_INFO; - cell_batch_append(cell, &cmd, 4); - cell_batch_append(cell, &cell->vertex_info, sizeof(struct vertex_info)); + emit_state_cmd(cell, CELL_CMD_STATE_VERTEX_INFO, + &cell->vertex_info, sizeof(struct vertex_info)); } } -- cgit v1.2.3 From 5b4d14bf1c6363f82660a53ca9505e55696084b0 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 11:21:25 -0700 Subject: Cell: additional assertions --- src/mesa/pipe/cell/spu/spu_main.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 5b50ec6953b..62f6a357ba3 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -441,9 +441,12 @@ cmd_state_sampler(const struct pipe_sampler_state *state) static void cmd_state_vertex_info(const struct vertex_info *vinfo) { - if (Debug) + if (Debug) { printf("SPU %u: VERTEX_INFO num_attribs=%u\n", spu.init.id, vinfo->num_attribs); + } + ASSERT(vinfo->num_attribs >= 1); + ASSERT(vinfo->num_attribs <= 8); memcpy(&spu.vertex_info, vinfo, sizeof(*vinfo)); } -- cgit v1.2.3 From ab36a9346cb1263fdeac492c0df986ab8cfb38b3 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 12:35:20 -0700 Subject: Added d/D keys to change viewing distance, 'a' to toggle animation --- progs/demos/gears.c | 35 +++++++++++++++++++++++++++++------ 1 file changed, 29 insertions(+), 6 deletions(-) diff --git a/progs/demos/gears.c b/progs/demos/gears.c index ab9bc007421..7abbd3670bd 100644 --- a/progs/demos/gears.c +++ b/progs/demos/gears.c @@ -27,6 +27,9 @@ static GLint T0 = 0; static GLint Frames = 0; static GLint autoexit = 0; static GLint win = 0; +static GLboolean Visible = GL_TRUE; +static GLboolean Animate = GL_TRUE; +static GLfloat viewDist = 40.0; /** @@ -179,6 +182,9 @@ draw(void) glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); glPushMatrix(); + + glTranslatef(0.0, 0.0, -viewDist); + glRotatef(view_rotx, 1.0, 0.0, 0.0); glRotatef(view_roty, 0.0, 1.0, 0.0); glRotatef(view_rotz, 0.0, 0.0, 1.0); @@ -240,6 +246,15 @@ idle(void) glutPostRedisplay(); } +static void +update_idle_func(void) +{ + if (Visible && Animate) + glutIdleFunc(idle); + else + glutIdleFunc(NULL); +} + /* change view angle, exit upon ESC */ /* ARGSUSED1 */ static void @@ -252,6 +267,16 @@ key(unsigned char k, int x, int y) case 'Z': view_rotz -= 5.0; break; + case 'd': + viewDist += 1.0; + break; + case 'D': + viewDist -= 1.0; + break; + case 'a': + Animate = !Animate; + update_idle_func(); + break; case 27: /* Escape */ cleanup(); exit(0); @@ -297,8 +322,6 @@ reshape(int width, int height) glLoadIdentity(); glFrustum(-1.0, 1.0, -h, h, 5.0, 60.0); glMatrixMode(GL_MODELVIEW); - glLoadIdentity(); - glTranslatef(0.0, 0.0, -40.0); } static void @@ -351,13 +374,12 @@ init(int argc, char *argv[]) } } + static void visible(int vis) { - if (vis == GLUT_VISIBLE) - glutIdleFunc(idle); - else - glutIdleFunc(NULL); + Visible = vis; + update_idle_func(); } int main(int argc, char *argv[]) @@ -375,6 +397,7 @@ int main(int argc, char *argv[]) glutKeyboardFunc(key); glutSpecialFunc(special); glutVisibilityFunc(visible); + update_idle_func(); glutMainLoop(); return 0; /* ANSI C requires main to return int. */ -- cgit v1.2.3 From 0abef84995acd01d7155f4fc851c9c528a1cfa73 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 12:41:47 -0700 Subject: push out far clip plane to 200 --- progs/demos/gears.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/progs/demos/gears.c b/progs/demos/gears.c index 7abbd3670bd..2a9fefefb53 100644 --- a/progs/demos/gears.c +++ b/progs/demos/gears.c @@ -320,7 +320,7 @@ reshape(int width, int height) glViewport(0, 0, (GLint) width, (GLint) height); glMatrixMode(GL_PROJECTION); glLoadIdentity(); - glFrustum(-1.0, 1.0, -h, h, 5.0, 60.0); + glFrustum(-1.0, 1.0, -h, h, 5.0, 200.0); glMatrixMode(GL_MODELVIEW); } -- cgit v1.2.3 From 3c9e26e0fa6d97f50b74745ffad9215f68607fe3 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 20:57:16 +0000 Subject: gallium: fix typos in hardwired fetch path --- src/mesa/pipe/draw/draw_vertex_fetch.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index afdf1971d24..89e4c256a7d 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -396,7 +396,7 @@ static void fetch_xyz_rgb_st( struct draw_context *draw, { const float *in = (const float *)(src[2] + elts[i] * pitch[2]); - float *out = &machine->Inputs[1].xyzw[0].f[i]; + float *out = &machine->Inputs[2].xyzw[0].f[i]; out[0] = in[0]; out[4] = in[1]; out[8] = 0.0f; @@ -500,7 +500,7 @@ void draw_update_vertex_fetch( struct draw_context *draw ) case 3: if (draw->vertex_element[0].src_format == PIPE_FORMAT_R32G32B32_FLOAT && draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_element[1].src_format == PIPE_FORMAT_R32G32_FLOAT) + draw->vertex_element[2].src_format == PIPE_FORMAT_R32G32_FLOAT) draw->vertex_fetch.fetch_func = fetch_xyz_rgb_st; break; default: -- cgit v1.2.3 From eb085014035a55e2be5e582dcc1e0bfcae771ba5 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Mon, 28 Jan 2008 20:57:58 +0000 Subject: gallium: remove dead vars, code --- src/mesa/pipe/draw/draw_vf_sse.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c index b238b542e76..066d6c0b7b4 100644 --- a/src/mesa/pipe/draw/draw_vf_sse.c +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -343,8 +343,6 @@ static boolean build_vertex_emit( struct x86_program *p ) struct x86_reg countEBP = x86_make_reg(file_REG32, reg_BP); struct x86_reg vfESI = x86_make_reg(file_REG32, reg_SI); struct x86_reg temp = x86_make_reg(file_XMM, 0); - struct x86_reg vp0 = x86_make_reg(file_XMM, 1); - struct x86_reg vp1 = x86_make_reg(file_XMM, 2); uint8_t *fixup, *label; /* Push a few regs? @@ -371,7 +369,6 @@ static boolean build_vertex_emit( struct x86_program *p ) /* always load, needed or not: */ - sse_movups(&p->func, p->chan0, x86_make_disp(vfESI, get_offset(vf, &vf->chan_scale[0]))); sse_movups(&p->func, p->identity, x86_make_disp(vfESI, get_offset(vf, &vf->identity[0]))); /* Note address for loop jump */ -- cgit v1.2.3 From 5ec70aa03b59c514cba1fe8dae09118250fb15d6 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 09:34:09 +0900 Subject: gallium: Remove direct dependencies to mesa internals. _mesa_exec_free is still being called. More invasive refactoring is necessary to clean it out. --- src/mesa/pipe/draw/draw_vf.c | 20 ++++++++------ src/mesa/pipe/draw/draw_vf.h | 52 +++++------------------------------- src/mesa/pipe/draw/draw_vf_generic.c | 9 ++++--- src/mesa/pipe/draw/draw_vf_sse.c | 14 +++++----- 4 files changed, 31 insertions(+), 64 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index deedfc7bc7e..d36f6293b1f 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -25,17 +25,20 @@ * Keith Whitwell */ -#include "glheader.h" -#include "context.h" -#include "colormac.h" #include "pipe/p_compiler.h" +#include "pipe/p_util.h" #include "draw_vf.h" + #define DBG 0 +/* TODO: remove this */ +extern void +_mesa_exec_free( void *addr ); + static boolean match_fastpath( struct draw_vertex_fetch *vf, const struct draw_vf_fastpath *fp) @@ -88,7 +91,7 @@ void draw_vf_register_fastpath( struct draw_vertex_fetch *vf, fastpath->match_strides = match_strides; fastpath->func = vf->emit; fastpath->attr = (struct draw_vf_attr_type *) - _mesa_malloc(vf->attr_count * sizeof(fastpath->attr[0])); + MALLOC(vf->attr_count * sizeof(fastpath->attr[0])); for (i = 0; i < vf->attr_count; i++) { fastpath->attr[i].format = vf->attr[i].format; @@ -156,7 +159,7 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, unsigned offset = 0; unsigned i, j; - assert(nr < DRAW_VF_ATTRIB_MAX); + assert(nr < PIPE_ATTRIB_MAX); memset(vf->lookup, 0, sizeof(vf->lookup)); @@ -200,7 +203,7 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, - +#if 0 /* Set attribute pointers, adjusted for start position: */ void draw_vf_set_sources( struct draw_vertex_fetch *vf, @@ -223,6 +226,7 @@ void draw_vf_set_sources( struct draw_vertex_fetch *vf, a[j].inputptr = ((uint8_t *)vptr->data) + start * vptr->stride; } } +#endif /* Set attribute pointers, adjusted for start position: @@ -260,7 +264,7 @@ struct draw_vertex_fetch *draw_vf_create( void ) struct draw_vertex_fetch *vf = CALLOC_STRUCT(draw_vertex_fetch); unsigned i; - for (i = 0; i < DRAW_VF_ATTRIB_MAX; i++) + for (i = 0; i < PIPE_ATTRIB_MAX; i++) vf->attr[i].vf = vf; vf->identity[0] = 0.0; @@ -271,7 +275,7 @@ struct draw_vertex_fetch *draw_vf_create( void ) vf->codegen_emit = NULL; #ifdef USE_SSE_ASM - if (!_mesa_getenv("MESA_NO_CODEGEN")) + if (!GETENV("MESA_NO_CODEGEN")) vf->codegen_emit = draw_vf_generate_sse_emit; #endif diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index c6a8fe0d534..7d90f35b0f4 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -29,44 +29,11 @@ #define DRAW_VF_H -#include "math/m_vector.h" - #include "pipe/p_compiler.h" -#include "draw_vertex.h" +#include "pipe/p_state.h" +#include "draw_vertex.h" -enum { - DRAW_VF_ATTRIB_POS = 0, - DRAW_VF_ATTRIB_WEIGHT = 1, - DRAW_VF_ATTRIB_NORMAL = 2, - DRAW_VF_ATTRIB_COLOR0 = 3, - DRAW_VF_ATTRIB_COLOR1 = 4, - DRAW_VF_ATTRIB_FOG = 5, - DRAW_VF_ATTRIB_COLOR_INDEX = 6, - DRAW_VF_ATTRIB_EDGEFLAG = 7, - DRAW_VF_ATTRIB_TEX0 = 8, - DRAW_VF_ATTRIB_TEX1 = 9, - DRAW_VF_ATTRIB_TEX2 = 10, - DRAW_VF_ATTRIB_TEX3 = 11, - DRAW_VF_ATTRIB_TEX4 = 12, - DRAW_VF_ATTRIB_TEX5 = 13, - DRAW_VF_ATTRIB_TEX6 = 14, - DRAW_VF_ATTRIB_TEX7 = 15, - DRAW_VF_ATTRIB_VAR0 = 16, - DRAW_VF_ATTRIB_VAR1 = 17, - DRAW_VF_ATTRIB_VAR2 = 18, - DRAW_VF_ATTRIB_VAR3 = 19, - DRAW_VF_ATTRIB_VAR4 = 20, - DRAW_VF_ATTRIB_VAR5 = 21, - DRAW_VF_ATTRIB_VAR6 = 22, - DRAW_VF_ATTRIB_VAR7 = 23, - DRAW_VF_ATTRIB_POINTSIZE = 24, - DRAW_VF_ATTRIB_BFC0 = 25, - DRAW_VF_ATTRIB_BFC1 = 26, - DRAW_VF_ATTRIB_CLIP_POS = 27, - DRAW_VF_ATTRIB_VERTEX_HEADER = 28, - DRAW_VF_ATTRIB_MAX = 29 -}; enum draw_vf_attr_format { DRAW_EMIT_1F, @@ -101,10 +68,12 @@ draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, unsigned nr, unsigned vertex_stride ); +#if 0 void draw_vf_set_sources( struct draw_vertex_fetch *vf, GLvector4f * const attrib[], - unsigned start ); + unsigned start ); +#endif void draw_vf_set_data( struct draw_vertex_fetch *vf, @@ -115,13 +84,6 @@ draw_vf_emit_vertices( struct draw_vertex_fetch *vf, unsigned count, void *dest ); -void -draw_vf_get_attr( struct draw_vertex_fetch *vf, - const void *vertex, - GLenum attr, - const float *dflt, - float *dest ); - struct draw_vertex_fetch * draw_vf_create( void ); @@ -174,11 +136,11 @@ struct draw_vf_attr struct draw_vertex_fetch { - struct draw_vf_attr attr[DRAW_VF_ATTRIB_MAX]; + struct draw_vf_attr attr[PIPE_ATTRIB_MAX]; unsigned attr_count; unsigned vertex_stride; - struct draw_vf_attr *lookup[DRAW_VF_ATTRIB_MAX]; + struct draw_vf_attr *lookup[PIPE_ATTRIB_MAX]; draw_vf_emit_func emit; diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c index 343428d26c7..a16eb456b74 100644 --- a/src/mesa/pipe/draw/draw_vf_generic.c +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -26,12 +26,13 @@ * Keith Whitwell */ -#include "glheader.h" -#include "context.h" -#include "colormac.h" + +#include + #include "simple_list.h" #include "pipe/p_compiler.h" +#include "pipe/p_util.h" #include "draw_vf.h" @@ -94,7 +95,7 @@ static INLINE void insert_3f_xyw_4( const struct draw_vf_attr *a, uint8_t *v, co static INLINE void insert_3f_xyw_err( const struct draw_vf_attr *a, uint8_t *v, const float *in ) { (void) a; (void) v; (void) in; - _mesa_exit(1); + assert(0); } static INLINE void insert_3f_3( const struct draw_vf_attr *a, uint8_t *v, const float *in ) diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c index 066d6c0b7b4..4036ded1d84 100644 --- a/src/mesa/pipe/draw/draw_vf_sse.c +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -25,15 +25,14 @@ * Keith Whitwell */ -#include "glheader.h" -#include "colormac.h" + #include "simple_list.h" -#include "enums.h" #include "pipe/p_compiler.h" #include "draw_vf.h" + #if defined(USE_SSE_ASM) #include "x86/rtasm/x86sse.h" @@ -450,7 +449,8 @@ static boolean build_vertex_emit( struct x86_program *p ) update_src_ptr(p, srcECX, vfESI, a); } else { - _mesa_printf("Can't emit 1ub %x %x %d\n", a->vertoffset, a[-1].vertoffset, a[-1].vertattrsize ); + fprintf(stderr, "Can't emit 1ub %x %x %d\n", + a->vertoffset, a[-1].vertoffset, a[-1].vertattrsize ); return FALSE; } break; @@ -495,7 +495,7 @@ static boolean build_vertex_emit( struct x86_program *p ) j++; /* NOTE: two attrs consumed */ } else { - _mesa_printf("Can't emit 3ub\n"); + fprintf(stderr, "Can't emit 3ub\n"); } return FALSE; /* add this later */ break; @@ -528,7 +528,7 @@ static boolean build_vertex_emit( struct x86_program *p ) update_src_ptr(p, srcECX, vfESI, a); break; default: - _mesa_printf("unknown a[%d].format %d\n", j, a->format); + fprintf(stderr, "unknown a[%d].format %d\n", j, a->format); return FALSE; /* catch any new opcodes */ } @@ -577,7 +577,7 @@ void draw_vf_generate_sse_emit( struct draw_vertex_fetch *vf ) return; } - _mesa_memset(&p, 0, sizeof(p)); + memset(&p, 0, sizeof(p)); p.vf = vf; p.inputs_safe = 0; /* for now */ -- cgit v1.2.3 From d5dd52aea826c4b6a417d102ecdeae8c713e81f6 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 09:41:21 +0900 Subject: gallium: Use CALLOC for pb_buffer to ensure that all fields of pipe_buffer are initiallized. --- src/mesa/pipe/pipebuffer/pb_buffer_malloc.c | 3 +-- src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c | 2 ++ src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c | 2 +- 3 files changed, 4 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c b/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c index fc83a00f367..2151f1d691e 100644 --- a/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c +++ b/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c @@ -107,10 +107,9 @@ pb_malloc_buffer_create(size_t size, { struct malloc_buffer *buf; - /* TODO: accept an alignment parameter */ /* TODO: do a single allocation */ - buf = (struct malloc_buffer *)MALLOC(sizeof(struct malloc_buffer)); + buf = CALLOC_STRUCT(malloc_buffer); if(!buf) return NULL; diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c index 2694f57bca0..a2657dac592 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c @@ -399,6 +399,8 @@ mm_buffer_destroy(struct pb_buffer *buf) struct mm_buffer *mm_buf = mm_buffer(buf); struct mm_pb_manager *mm = mm_buf->mgr; + assert(buf->base.refcount == 0); + _glthread_LOCK_MUTEX(mm->mutex); mmFreeMem(mm_buf->block); FREE(buf); diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c index 7c299541122..f80c7e34c07 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c @@ -258,7 +258,7 @@ pool_bufmgr_create(struct pb_manager *provider, if(!pool->map) goto failure; - pool->bufs = (struct pool_buffer *) MALLOC(numBufs * sizeof(*pool->bufs)); + pool->bufs = (struct pool_buffer *)CALLOC(numBufs, sizeof(*pool->bufs)); if (!pool->bufs) goto failure; -- cgit v1.2.3 From d6667171dca5999bef0693963634ecda74c32d5a Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 09:42:03 +0900 Subject: gallium: Use GALLIUM_ prefix for env vars. --- src/mesa/pipe/draw/draw_vf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index d36f6293b1f..4fc2312ad1e 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -275,7 +275,7 @@ struct draw_vertex_fetch *draw_vf_create( void ) vf->codegen_emit = NULL; #ifdef USE_SSE_ASM - if (!GETENV("MESA_NO_CODEGEN")) + if (!GETENV("GALLIUM_NO_CODEGEN")) vf->codegen_emit = draw_vf_generate_sse_emit; #endif -- cgit v1.2.3 From deaa895fe241cfeab6f390791d462390ff1d1560 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 12:46:05 -0700 Subject: Cell: re-enable bounding boxes The geometry bounding box is used to restrict rasterization to just those tiles that are relevant. Note another dummy field had to be added to the cell_command_render struct. Apparently, every 4th word in a struct is susceptible to corruption in some circumstances. Might be a compiler bug. --- src/mesa/pipe/cell/common.h | 2 +- src/mesa/pipe/cell/ppu/cell_vbuf.c | 4 ++++ src/mesa/pipe/cell/spu/spu_main.c | 30 +++++++++++++++++++----------- 3 files changed, 24 insertions(+), 12 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index d6e1dd4f7d5..5e32b209e6a 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -122,7 +122,7 @@ struct cell_command_render uint dummy; /* XXX this dummy field works around a compiler bug */ uint num_indexes; uint vertex_buf; /**< which cell->buffer[] contains the vertex data */ - float xmin, ymin, xmax, ymax; + float xmin, dummy2, ymin, xmax, ymax; /* XXX another dummy field */ boolean inline_verts; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/ppu/cell_vbuf.c b/src/mesa/pipe/cell/ppu/cell_vbuf.c index b2a25d767b7..9f737287adb 100644 --- a/src/mesa/pipe/cell/ppu/cell_vbuf.c +++ b/src/mesa/pipe/cell/ppu/cell_vbuf.c @@ -180,6 +180,10 @@ cell_vbuf_draw(struct vbuf_render *vbr, if (v[1] > ymax) ymax = v[1]; } +#if 0 + printf("PPU Bounds %g, %g .. %g, %g\n", xmin, ymin, xmax, ymax); + fflush(stdout); +#endif if (cvbr->prim != PIPE_PRIM_TRIANGLES) return; /* only render tris for now */ diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 62f6a357ba3..c2b05ed5a2b 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -200,7 +200,7 @@ tile_bounding_box(const struct cell_command_render *render, uint *txmin, uint *tymin, uint *box_num_tiles, uint *box_width_tiles) { -#if 1 +#if 0 /* Debug: full-window bounding box */ uint txmax = spu.fb.width_tiles - 1; uint tymax = spu.fb.height_tiles - 1; @@ -223,13 +223,24 @@ tile_bounding_box(const struct cell_command_render *render, *box_num_tiles = *box_width_tiles * box_height_tiles; #endif #if 0 - printf("Render bounds: %g, %g ... %g, %g\n", + printf("SPU %u: bounds: %g, %g ... %g, %g\n", spu.init.id, render->xmin, render->ymin, render->xmax, render->ymax); - printf("Render tiles: %u, %u .. %u, %u\n", *txmin, *tymin, txmax, tymax); + printf("SPU %u: tiles: %u, %u .. %u, %u\n", + spu.init.id, *txmin, *tymin, txmax, tymax); + ASSERT(render->xmin <= render->xmax); + ASSERT(render->ymin <= render->ymax); #endif } +/** Check if the tile at (tx,ty) belongs to this SPU */ +static INLINE boolean +my_tile(uint tx, uint ty) +{ + return (spu.fb.width_tiles * ty + tx) % spu.init.num_spus == spu.init.id; +} + + /** * Render primitives * \param pos_incr returns value indicating how may words to skip after @@ -295,15 +306,9 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) ** find tiles which intersect the prim bounding box **/ uint txmin, tymin, box_width_tiles, box_num_tiles; -#if 0 tile_bounding_box(render, &txmin, &tymin, &box_num_tiles, &box_width_tiles); -#else - txmin = 0; - tymin = 0; - box_num_tiles = spu.fb.width_tiles * spu.fb.height_tiles; - box_width_tiles = spu.fb.width_tiles; -#endif + /* make sure any pending clears have completed */ wait_on_mask(1 << TAG_SURFACE_CLEAR); /* XXX temporary */ @@ -312,13 +317,16 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) /** ** loop over tiles, rendering tris **/ - for (i = spu.init.id; i < box_num_tiles; i += spu.init.num_spus) { + for (i = 0; i < box_num_tiles; i++) { const uint tx = txmin + i % box_width_tiles; const uint ty = tymin + i / box_width_tiles; ASSERT(tx < spu.fb.width_tiles); ASSERT(ty < spu.fb.height_tiles); + if (!my_tile(tx, ty)) + continue; + /* Start fetching color/z tiles. We'll wait for completion when * we need read/write to them later in triangle rasterization. */ -- cgit v1.2.3 From 7c596b80110da42435f8f0714d6f21b760f11c4f Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 13:02:11 -0700 Subject: Cell: emit state in cell_clear_surface() if dirty. Without this a program that does nothing but glClear() doesn't work. We need the framebuffer state. --- src/mesa/pipe/cell/ppu/cell_clear.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/mesa/pipe/cell/ppu/cell_clear.c b/src/mesa/pipe/cell/ppu/cell_clear.c index e61bfd9b0fd..07b908eec59 100644 --- a/src/mesa/pipe/cell/ppu/cell_clear.c +++ b/src/mesa/pipe/cell/ppu/cell_clear.c @@ -50,6 +50,10 @@ cell_clear_surface(struct pipe_context *pipe, struct pipe_surface *ps, struct cell_context *cell = cell_context(pipe); uint surfIndex; + if (cell->dirty) + cell_update_derived(cell); + + if (!cell->cbuf_map[0]) cell->cbuf_map[0] = pipe_surface_map(ps); -- cgit v1.2.3 From 2f868411a209d909f3ea8f29a317b7327fe6f88a Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 17:22:12 -0700 Subject: Cell: initial texture cache/sampling code --- src/mesa/pipe/cell/spu/spu_texture.c | 139 +++++++++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_texture.h | 43 +++++++++++ 2 files changed, 182 insertions(+) create mode 100644 src/mesa/pipe/cell/spu/spu_texture.c create mode 100644 src/mesa/pipe/cell/spu/spu_texture.h diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c new file mode 100644 index 00000000000..6d566a5006d --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -0,0 +1,139 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include "pipe/p_compiler.h" +#include "spu_main.h" +#include "spu_texture.h" +#include "spu_tile.h" + + +/** + * Number of texture tiles to cache. + * Note that this will probably be the largest consumer of SPU local store/ + * memory for this driver! + */ +#define CACHE_SIZE 16 + +static tile_t tex_tiles[CACHE_SIZE] ALIGN16_ATTRIB; + +static int tex_tile_x[CACHE_SIZE], tex_tile_y[CACHE_SIZE]; + + + +/** + * Mark all tex cache entries as invalid. + */ +void +invalidate_tex_cache(void) +{ + /* XXX memset? */ + uint i; + for (i = 0; i < CACHE_SIZE; i++) + tex_tile_x[i] = tex_tile_y[i] = -1; +} + + +/** + * Return the cache pos/index which corresponds to texel (i,j) + */ +static INLINE uint +cache_pos(uint i, uint j) +{ + uint tx = i / TILE_SIZE; + uint ty = j / TILE_SIZE; + uint pos = (tx + ty * 4) % CACHE_SIZE; + return pos; +} + + +/** + * Make sure the tile for texel (i,j) is present, return its position/index + * in the cache. + */ +static uint +get_tex_tile(uint i, uint j) +{ + const int tx = i / TILE_SIZE; + const int ty = j / TILE_SIZE; + const uint pos = cache_pos(i, j); + + if (tex_tile_x[pos] != tx || tex_tile_y[pos] != ty) { + /* texture cache miss, fetch tile from main memory */ + const uint tiles_per_row = spu.texture.width / TILE_SIZE; + const uint bytes_per_tile = sizeof(tile_t); + const void *src = (const ubyte *) spu.texture.start + + (ty * tiles_per_row + tx) * bytes_per_tile; + + printf("SPU %u: tex cache miss at %d, %d pos=%u old=%d,%d\n", + spu.init.id, tx, ty, pos, + tex_tile_x[pos], tex_tile_y[pos]); +#if 0 + printf("SPU %u: get tex tile from %p to %p\n", + spu.init.id, src, tex_tiles[pos].t32); +#endif + + ASSERT_ALIGN16(tex_tiles[pos].t32); + ASSERT_ALIGN16(src); + + mfc_get(tex_tiles[pos].t32, /* dest */ + (unsigned int) src, + bytes_per_tile, /* size */ + TAG_TEXTURE_TILE, + 0, /* tid */ + 0 /* rid */); + + wait_on_mask(1 << TAG_TEXTURE_TILE); + + tex_tile_x[pos] = tx; + tex_tile_y[pos] = ty; + } + else { +#if 0 + printf("SPU %u: tex cache HIT at %d, %d\n", + spu.init.id, tx, ty); +#endif + } + + return pos; +} + + +/** + * Get texture sample at texcoord. + * XXX this is extremely primitive for now. + */ +uint +sample_texture(const float *texcoord) +{ + /* wrap/repeat */ + uint i = (uint) (texcoord[0] * spu.texture.width) % spu.texture.width; + uint j = (uint) (texcoord[1] * spu.texture.height) % spu.texture.height; + uint pos = get_tex_tile(i, j); + uint texel = tex_tiles[pos].t32[j % TILE_SIZE][i % TILE_SIZE]; + return texel; +} diff --git a/src/mesa/pipe/cell/spu/spu_texture.h b/src/mesa/pipe/cell/spu/spu_texture.h new file mode 100644 index 00000000000..b75b7ac44fd --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_texture.h @@ -0,0 +1,43 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef SPU_TEXTURE_H +#define SPU_TEXTURE_H + + +#include "pipe/p_compiler.h" + + +extern void +invalidate_tex_cache(void); + + +extern uint +sample_texture(const float *texcoord); + + +#endif /* SPU_TEXTURE_H */ -- cgit v1.2.3 From 948dc8ad24d554ab23bea97aa3e405c4f6ad47c6 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 17:23:44 -0700 Subject: Cell: basic texture mapping Texture images are tiled in PPU code. SPUs use a texture cache for getting texels from textures. This is very rough code, but demos/texcyl.c works. --- src/mesa/pipe/cell/common.h | 10 +++- src/mesa/pipe/cell/ppu/cell_context.h | 5 +- src/mesa/pipe/cell/ppu/cell_state_emit.c | 12 +++- src/mesa/pipe/cell/ppu/cell_state_sampler.c | 10 +++- src/mesa/pipe/cell/ppu/cell_texture.c | 87 +++++++++++++++++++++++++++++ src/mesa/pipe/cell/ppu/cell_texture.h | 6 ++ src/mesa/pipe/cell/spu/Makefile | 1 + src/mesa/pipe/cell/spu/spu_main.c | 17 ++++++ src/mesa/pipe/cell/spu/spu_main.h | 3 + src/mesa/pipe/cell/spu/spu_tri.c | 60 ++++++++++++-------- 10 files changed, 183 insertions(+), 28 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 5e32b209e6a..f0d48ff403b 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -79,7 +79,8 @@ #define CELL_CMD_STATE_FRAMEBUFFER 10 #define CELL_CMD_STATE_DEPTH_STENCIL 11 #define CELL_CMD_STATE_SAMPLER 12 -#define CELL_CMD_STATE_VERTEX_INFO 13 +#define CELL_CMD_STATE_TEXTURE 13 +#define CELL_CMD_STATE_VERTEX_INFO 14 #define CELL_NUM_BUFFERS 4 @@ -134,6 +135,13 @@ struct cell_command_release_verts }; +struct cell_command_texture +{ + void *start; /**< Address in main memory */ + uint width, height; +}; + + /** XXX unions don't seem to work */ struct cell_command { diff --git a/src/mesa/pipe/cell/ppu/cell_context.h b/src/mesa/pipe/cell/ppu/cell_context.h index de65fb5e9a8..7d234f3e452 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.h +++ b/src/mesa/pipe/cell/ppu/cell_context.h @@ -76,7 +76,7 @@ struct cell_context struct pipe_framebuffer_state framebuffer; struct pipe_poly_stipple poly_stipple; struct pipe_scissor_state scissor; - struct pipe_texture *texture[PIPE_MAX_SAMPLERS]; + struct cell_texture *texture[PIPE_MAX_SAMPLERS]; struct pipe_viewport_state viewport; struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX]; @@ -84,6 +84,9 @@ struct cell_context ubyte *cbuf_map[PIPE_MAX_COLOR_BUFS]; ubyte *zsbuf_map; + struct pipe_surface *tex_surf; + uint *tex_map; + uint dirty; /** The primitive drawing context */ diff --git a/src/mesa/pipe/cell/ppu/cell_state_emit.c b/src/mesa/pipe/cell/ppu/cell_state_emit.c index 6776ec88c7b..391ff454ac2 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_emit.c +++ b/src/mesa/pipe/cell/ppu/cell_state_emit.c @@ -30,7 +30,7 @@ #include "cell_state.h" #include "cell_state_emit.h" #include "cell_batch.h" - +#include "cell_texture.h" static void @@ -72,6 +72,16 @@ cell_emit_state(struct cell_context *cell) cell->sampler[0], sizeof(struct pipe_sampler_state)); } + if (cell->dirty & CELL_NEW_TEXTURE) { + struct cell_command_texture texture; + texture.start = cell->texture[0]->tiled_data; + texture.width = cell->texture[0]->base.width[0]; + texture.height = cell->texture[0]->base.height[0]; + + emit_state_cmd(cell, CELL_CMD_STATE_TEXTURE, + &texture, sizeof(struct cell_command_texture)); + } + if (cell->dirty & CELL_NEW_VERTEX_INFO) { emit_state_cmd(cell, CELL_CMD_STATE_VERTEX_INFO, &cell->vertex_info, sizeof(struct vertex_info)); diff --git a/src/mesa/pipe/cell/ppu/cell_state_sampler.c b/src/mesa/pipe/cell/ppu/cell_state_sampler.c index ae1eeb46203..317f7603bb7 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_sampler.c +++ b/src/mesa/pipe/cell/ppu/cell_state_sampler.c @@ -30,12 +30,10 @@ */ #include "pipe/p_util.h" +#include "pipe/draw/draw_context.h" #include "cell_context.h" #include "cell_state.h" -#if 0 #include "cell_texture.h" -#include "cell_tile_cache.h" -#endif void * @@ -53,6 +51,8 @@ cell_bind_sampler_state(struct pipe_context *pipe, { struct cell_context *cell = cell_context(pipe); + draw_flush(cell->draw); + assert(unit < PIPE_MAX_SAMPLERS); cell->sampler[unit] = (struct pipe_sampler_state *)sampler; @@ -76,7 +76,11 @@ cell_set_sampler_texture(struct pipe_context *pipe, { struct cell_context *cell = cell_context(pipe); + draw_flush(cell->draw); + cell->texture[sampler] = texture; + cell_update_texture_mapping(cell); + cell->dirty |= CELL_NEW_TEXTURE; } diff --git a/src/mesa/pipe/cell/ppu/cell_texture.c b/src/mesa/pipe/cell/ppu/cell_texture.c index 0a8190d983c..acbe4c79f05 100644 --- a/src/mesa/pipe/cell/ppu/cell_texture.c +++ b/src/mesa/pipe/cell/ppu/cell_texture.c @@ -163,3 +163,90 @@ cell_get_tex_surface(struct pipe_context *pipe, } return ps; } + + + +static void +tile_copy_data(uint w, uint h, uint tile_size, uint *dst, const uint *src) +{ + const uint tile_size2 = tile_size * tile_size; + const uint h_t = h / tile_size, w_t = w / tile_size; + + uint it, jt; /* tile counters */ + uint i, j; /* intra-tile counters */ + + for (it = 0; it < h_t; it++) { + for (jt = 0; jt < w_t; jt++) { + /* fill in tile (i, j) */ + uint *tdst = dst + (it * w_t + jt) * tile_size2; + for (i = 0; i < tile_size; i++) { + for (j = 0; j < tile_size; j++) { + const uint srci = it * tile_size + i; + const uint srcj = jt * tile_size + j; + *tdst++ = src[srci * h + srcj]; + } + } + } + } +} + + + +/** + * Convert linear texture image data to tiled format for SPU usage. + */ +static void +cell_tile_texture(struct cell_context *cell, + struct cell_texture *texture) +{ + uint face = 0, level = 0, zslice = 0; + struct pipe_surface *surf; + const uint w = texture->base.width[0], h = texture->base.height[0]; + const uint *src; + + /* temporary restrictions: */ + assert(w >= TILE_SIZE); + assert(h >= TILE_SIZE); + assert(w % TILE_SIZE == 0); + assert(h % TILE_SIZE == 0); + + surf = cell_get_tex_surface(&cell->pipe, &texture->base, face, level, zslice); + ASSERT(surf); + + src = (const uint *) pipe_surface_map(surf); + + if (texture->tiled_data) { + align_free(texture->tiled_data); + } + texture->tiled_data = align_malloc(w * h * 4, 16); + + tile_copy_data(w, h, TILE_SIZE, texture->tiled_data, src); + + pipe_surface_unmap(surf); + + pipe_surface_reference(&surf, NULL); +} + + + +void +cell_update_texture_mapping(struct cell_context *cell) +{ + uint face = 0, level = 0, zslice = 0; + + cell_tile_texture(cell, cell->texture[0]); +#if 0 + if (cell->tex_surf && cell->tex_map) { + pipe_surface_unmap(cell->tex_surf); + cell->tex_map = NULL; + } + + /* XXX free old surface */ + + cell->tex_surf = cell_get_tex_surface(&cell->pipe, + &cell->texture[0]->base, + face, level, zslice); + + cell->tex_map = pipe_surface_map(cell->tex_surf); +#endif +} diff --git a/src/mesa/pipe/cell/ppu/cell_texture.h b/src/mesa/pipe/cell/ppu/cell_texture.h index ef5808c0862..bd434c87762 100644 --- a/src/mesa/pipe/cell/ppu/cell_texture.h +++ b/src/mesa/pipe/cell/ppu/cell_texture.h @@ -46,6 +46,8 @@ struct cell_texture */ struct pipe_buffer *buffer; unsigned long buffer_size; + + void *tiled_data; /* XXX this may be temporary */ /*ALIGN16*/ }; @@ -70,4 +72,8 @@ cell_get_tex_surface(struct pipe_context *pipe, unsigned face, unsigned level, unsigned zslice); +extern void +cell_update_texture_mapping(struct cell_context *cell); + + #endif /* CELL_TEXTURE */ diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index 417ae1b072a..011fdcefe3e 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -17,6 +17,7 @@ PROG_SPU_EMBED_O = $(PROG)_spu-embed.o SOURCES = \ spu_main.c \ + spu_texture.c \ spu_tile.c \ spu_tri.c diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index c2b05ed5a2b..5a5b17dd896 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -34,6 +34,7 @@ #include #include "spu_main.h" +#include "spu_texture.h" #include "spu_tri.h" #include "spu_tile.h" #include "pipe/cell/common.h" @@ -446,6 +447,17 @@ cmd_state_sampler(const struct pipe_sampler_state *state) } +static void +cmd_state_texture(const struct cell_command_texture *texture) +{ + if (Debug) + printf("SPU %u: TEXTURE at %p size %u x %u\n", + spu.init.id, texture->start, texture->width, texture->height); + + memcpy(&spu.texture, texture, sizeof(*texture)); +} + + static void cmd_state_vertex_info(const struct vertex_info *vinfo) { @@ -561,6 +573,10 @@ cmd_batch(uint opcode) cmd_state_sampler((struct pipe_sampler_state *) &buffer[pos+1]); pos += (1 + sizeof(struct pipe_sampler_state) / 4); break; + case CELL_CMD_STATE_TEXTURE: + cmd_state_texture((struct cell_command_texture *) &buffer[pos+1]); + pos += (1 + sizeof(struct cell_command_texture) / 4); + break; case CELL_CMD_STATE_VERTEX_INFO: cmd_state_vertex_info((struct vertex_info *) &buffer[pos+1]); pos += (1 + sizeof(struct vertex_info) / 4); @@ -656,6 +672,7 @@ one_time_init(void) { memset(tile_status, TILE_STATUS_DEFINED, sizeof(tile_status)); memset(tile_status_z, TILE_STATUS_DEFINED, sizeof(tile_status_z)); + invalidate_tex_cache(); } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 5bc5d9fa99d..480c54ebd08 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -60,6 +60,7 @@ struct spu_global struct pipe_depth_stencil_alpha_state depth_stencil; struct pipe_blend_state blend; struct pipe_sampler_state sampler[PIPE_MAX_SAMPLERS]; + struct cell_command_texture texture; struct vertex_info vertex_info; @@ -84,6 +85,8 @@ extern struct spu_global spu; #define TAG_INDEX_BUFFER 16 #define TAG_BATCH_BUFFER 17 #define TAG_MISC 18 +#define TAG_TEXTURE_TILE 19 + extern void diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 3d0d106c10e..aad28f1036f 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -33,6 +33,7 @@ #include "pipe/p_format.h" #include "pipe/p_util.h" #include "spu_main.h" +#include "spu_texture.h" #include "spu_tile.h" #include "spu_tri.h" @@ -362,9 +363,24 @@ emit_quad( struct setup_stage *setup, int x, int y, unsigned mask ) /* Cell: "write" quad fragments to the tile by setting prim color */ const int ix = x - setup->cliprect_minx; const int iy = y - setup->cliprect_miny; - float colors[4][4]; - - eval_coeff(setup, 1, (float) x, (float) y, colors); + uint colors[4]; /* indexed by QUAD_x */ + + if (spu.texture.start) { + float texcoords[4][4]; + uint i; + eval_coeff(setup, 2, (float) x, (float) y, texcoords); + for (i = 0; i < 4; i++) { + colors[i] = sample_texture(texcoords[i]); + } + } + else { + float fcolors[4][4]; + eval_coeff(setup, 1, (float) x, (float) y, fcolors); + colors[QUAD_TOP_LEFT] = pack_color(fcolors[QUAD_TOP_LEFT]); + colors[QUAD_TOP_RIGHT] = pack_color(fcolors[QUAD_TOP_RIGHT]); + colors[QUAD_BOTTOM_LEFT] = pack_color(fcolors[QUAD_BOTTOM_LEFT]); + colors[QUAD_BOTTOM_RIGHT] = pack_color(fcolors[QUAD_BOTTOM_RIGHT]); + } if (spu.depth_stencil.depth.enabled) { mask &= do_depth_test(setup, x, y, mask); @@ -382,13 +398,13 @@ emit_quad( struct setup_stage *setup, int x, int y, unsigned mask ) tile_status[setup->ty][setup->tx] = TILE_STATUS_DIRTY; if (mask & MASK_TOP_LEFT) - ctile.t32[iy][ix] = pack_color(colors[QUAD_TOP_LEFT]); + ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; if (mask & MASK_TOP_RIGHT) - ctile.t32[iy][ix+1] = pack_color(colors[QUAD_TOP_RIGHT]); + ctile.t32[iy][ix+1] = colors[QUAD_TOP_RIGHT]; if (mask & MASK_BOTTOM_LEFT) - ctile.t32[iy+1][ix] = pack_color(colors[QUAD_BOTTOM_LEFT]); + ctile.t32[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; if (mask & MASK_BOTTOM_RIGHT) - ctile.t32[iy+1][ix+1] = pack_color(colors[QUAD_BOTTOM_RIGHT]); + ctile.t32[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; } #endif } @@ -606,7 +622,6 @@ static boolean setup_sort_vertices( struct setup_stage *setup, } -#if 0 /** * Compute a0 for a constant-valued coefficient (GL_FLAT shading). * The value value comes from vertex->data[slot][i]. @@ -614,21 +629,20 @@ static boolean setup_sort_vertices( struct setup_stage *setup, * \param slot which attribute slot * \param i which component of the slot (0..3) */ -static void const_coeff( struct setup_stage *setup, - unsigned slot, - unsigned i ) +static void const_coeff(struct setup_stage *setup, uint slot) { - assert(slot < PIPE_MAX_SHADER_INPUTS); - assert(i <= 3); + uint i; + ASSERT(slot < PIPE_MAX_SHADER_INPUTS); - setup->coef[slot].dadx[i] = 0; - setup->coef[slot].dady[i] = 0; + for (i = 0; i < 4; i++) { + setup->coef[slot].dadx[i] = 0; + setup->coef[slot].dady[i] = 0; - /* need provoking vertex info! - */ - setup->coef[slot].a0[i] = setup->vprovoke->data[slot][i]; + /* need provoking vertex info! + */ + setup->coef[slot].a0[i] = setup->vprovoke->data[slot][i]; + } } -#endif /** @@ -735,15 +749,17 @@ static void setup_tri_coefficients( struct setup_stage *setup ) case INTERP_NONE: break; case INTERP_POS: - tri_linear_coeff(setup, i, 2, 3); /* slot 0, z */ + tri_linear_coeff(setup, i, 2, 3); /* XXX interp W if PERSPECTIVE... */ break; case INTERP_CONSTANT: - /* fall-through */ + const_coeff(setup, i); + break; case INTERP_LINEAR: - tri_linear_coeff(setup, i, 0, 4); /* slot 1, color */ + tri_linear_coeff(setup, i, 0, 4); break; case INTERP_PERSPECTIVE: + tri_linear_coeff(setup, i, 0, 4); /* XXX temporary */ break; default: ASSERT(0); -- cgit v1.2.3 From 5f54cfaba16c2ac268472e148f1e788a9d7b2a6a Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 17:32:23 -0700 Subject: Cell: minor optimization for flat shading --- src/mesa/pipe/cell/spu/spu_tri.c | 37 ++++++++++++++++++++++++++++--------- 1 file changed, 28 insertions(+), 9 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index aad28f1036f..19a231d9c4d 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -200,16 +200,35 @@ static INLINE void eval_coeff( struct setup_stage *setup, uint slot, float x, float y, float result[4][4]) { - uint i; - const float *dadx = setup->coef[slot].dadx; - const float *dady = setup->coef[slot].dady; + switch (spu.vertex_info.interp_mode[slot]) { + case INTERP_CONSTANT: + { + uint i; + for (i = 0; i < 4; i++) { + result[QUAD_TOP_LEFT][i] = + result[QUAD_TOP_RIGHT][i] = + result[QUAD_BOTTOM_LEFT][i] = + result[QUAD_BOTTOM_RIGHT][i] = setup->coef[slot].a0[i]; + } + } + break; - /* loop over XYZW comps */ - for (i = 0; i < 4; i++) { - result[QUAD_TOP_LEFT][i] = setup->coef[slot].a0[i] + x * dadx[i] + y * dady[i]; - result[QUAD_TOP_RIGHT][i] = result[0][i] + dadx[i]; - result[QUAD_BOTTOM_LEFT][i] = result[0][i] + dady[i]; - result[QUAD_BOTTOM_RIGHT][i] = result[0][i] + dadx[i] + dady[i]; + case INTERP_LINEAR: + /* fall-through, for now */ + default: + { + uint i; + const float *dadx = setup->coef[slot].dadx; + const float *dady = setup->coef[slot].dady; + + /* loop over XYZW comps */ + for (i = 0; i < 4; i++) { + result[QUAD_TOP_LEFT][i] = setup->coef[slot].a0[i] + x * dadx[i] + y * dady[i]; + result[QUAD_TOP_RIGHT][i] = result[0][i] + dadx[i]; + result[QUAD_BOTTOM_LEFT][i] = result[0][i] + dady[i]; + result[QUAD_BOTTOM_RIGHT][i] = result[0][i] + dadx[i] + dady[i]; + } + } } } -- cgit v1.2.3 From 7012dd9b76328b4b1f54404df1948e50f23c1fe3 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 18:03:45 -0700 Subject: Cell: compute min index referenced in draw command, use it to reduce size of vertex data payload --- src/mesa/pipe/cell/common.h | 2 ++ src/mesa/pipe/cell/ppu/cell_vbuf.c | 13 +++++++++++-- src/mesa/pipe/cell/spu/spu_main.c | 20 ++++++++++++++++---- 3 files changed, 29 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index f0d48ff403b..90aa46a5343 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -124,6 +124,8 @@ struct cell_command_render uint num_indexes; uint vertex_buf; /**< which cell->buffer[] contains the vertex data */ float xmin, dummy2, ymin, xmax, ymax; /* XXX another dummy field */ + uint dummy3; + uint min_index; boolean inline_verts; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/ppu/cell_vbuf.c b/src/mesa/pipe/cell/ppu/cell_vbuf.c index 9f737287adb..e63b34cf525 100644 --- a/src/mesa/pipe/cell/ppu/cell_vbuf.c +++ b/src/mesa/pipe/cell/ppu/cell_vbuf.c @@ -138,16 +138,23 @@ cell_vbuf_draw(struct vbuf_render *vbr, struct cell_context *cell = cvbr->cell; float xmin, ymin, xmax, ymax; uint i; - uint nr_vertices = 0; + uint nr_vertices = 0, min_index = ~0; const void *vertices = cvbr->vertex_buffer; const uint vertex_size = cvbr->vertex_size; for (i = 0; i < nr_indices; i++) { if (indices[i] > nr_vertices) nr_vertices = indices[i]; + if (indices[i] < min_index) + min_index = indices[i]; } nr_vertices++; +#if 0 + /*if (min_index > 0)*/ + printf("%s min_index = %u\n", __FUNCTION__, min_index); +#endif + #if 0 printf("cell_vbuf_draw() nr_indices = %u nr_verts = %u\n", nr_indices, nr_vertices); @@ -169,7 +176,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, /* compute x/y bounding box */ xmin = ymin = 1e50; xmax = ymax = -1e50; - for (i = 0; i < nr_vertices; i++) { + for (i = min_index; i < nr_vertices; i++) { const float *v = (float *) ((ubyte *) vertices + i * vertex_size); if (v[0] < xmin) xmin = v[0]; @@ -204,6 +211,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, render->prim_type = cvbr->prim; render->num_indexes = nr_indices; + render->min_index = min_index; /* append indices after render command */ memcpy(render + 1, indices, nr_indices * 2); @@ -214,6 +222,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, render->vertex_size = 4 * cell->vertex_info.size; render->num_verts = nr_vertices; if (ALLOW_INLINE_VERTS && + min_index == 0 && vertex_bytes <= cell_batch_free_space(cell)) { /* vertex data inlined, after indices */ void *dst = cell_batch_alloc(cell, vertex_bytes); diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 5a5b17dd896..3c9efb4741f 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -253,7 +253,7 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) /* we'll DMA into these buffers */ ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB; const uint vertex_size = render->vertex_size; /* in bytes */ - const uint total_vertex_bytes = render->num_verts * vertex_size; + /*const*/ uint total_vertex_bytes = render->num_verts * vertex_size; const ubyte *vertices; const ushort *indexes; uint i, j; @@ -289,9 +289,21 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) } else { /* Begin DMA fetch of vertex buffer */ - void *src = spu.init.buffers[render->vertex_buf]; - mfc_get(vertex_data, /* dest */ - (unsigned int) src, + ubyte *src = spu.init.buffers[render->vertex_buf]; + ubyte *dest = vertex_data; + + /* skip vertex data we won't use */ +#if 01 + src += render->min_index * vertex_size; + dest += render->min_index * vertex_size; + total_vertex_bytes -= render->min_index * vertex_size; +#endif + ASSERT(total_vertex_bytes % 16 == 0); + ASSERT_ALIGN16(dest); + ASSERT_ALIGN16(src); + + mfc_get(dest, /* in vertex_data[] array */ + (unsigned int) src, /* src in main memory */ total_vertex_bytes, /* size */ TAG_VERTEX_BUFFER, 0, /* tid */ -- cgit v1.2.3 From c474e0d6ed7b654ef750d088df2b26d8215f20ec Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 18:09:16 -0700 Subject: Cell: add a few null texture tests --- src/mesa/pipe/cell/ppu/cell_state_emit.c | 13 ++++++++++--- src/mesa/pipe/cell/ppu/cell_texture.c | 3 ++- 2 files changed, 12 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_state_emit.c b/src/mesa/pipe/cell/ppu/cell_state_emit.c index 391ff454ac2..702184416bf 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_emit.c +++ b/src/mesa/pipe/cell/ppu/cell_state_emit.c @@ -74,9 +74,16 @@ cell_emit_state(struct cell_context *cell) if (cell->dirty & CELL_NEW_TEXTURE) { struct cell_command_texture texture; - texture.start = cell->texture[0]->tiled_data; - texture.width = cell->texture[0]->base.width[0]; - texture.height = cell->texture[0]->base.height[0]; + if (cell->texture[0]) { + texture.start = cell->texture[0]->tiled_data; + texture.width = cell->texture[0]->base.width[0]; + texture.height = cell->texture[0]->base.height[0]; + } + else { + texture.start = NULL; + texture.width = 0; + texture.height = 0; + } emit_state_cmd(cell, CELL_CMD_STATE_TEXTURE, &texture, sizeof(struct cell_command_texture)); diff --git a/src/mesa/pipe/cell/ppu/cell_texture.c b/src/mesa/pipe/cell/ppu/cell_texture.c index acbe4c79f05..2cf60229394 100644 --- a/src/mesa/pipe/cell/ppu/cell_texture.c +++ b/src/mesa/pipe/cell/ppu/cell_texture.c @@ -234,7 +234,8 @@ cell_update_texture_mapping(struct cell_context *cell) { uint face = 0, level = 0, zslice = 0; - cell_tile_texture(cell, cell->texture[0]); + if (cell->texture[0]) + cell_tile_texture(cell, cell->texture[0]); #if 0 if (cell->tex_surf && cell->tex_map) { pipe_surface_unmap(cell->tex_surf); -- cgit v1.2.3 From e308dc4465fb3869a8423ed9608da35b426ef993 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 18:17:30 -0700 Subject: Cell: move cmd_render() into new spu_render.c file --- src/mesa/pipe/cell/spu/Makefile | 1 + src/mesa/pipe/cell/spu/spu_main.c | 206 +------------------------------ src/mesa/pipe/cell/spu/spu_main.h | 1 + src/mesa/pipe/cell/spu/spu_render.c | 240 ++++++++++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_render.h | 38 ++++++ 5 files changed, 283 insertions(+), 203 deletions(-) create mode 100644 src/mesa/pipe/cell/spu/spu_render.c create mode 100644 src/mesa/pipe/cell/spu/spu_render.h diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index 011fdcefe3e..d5b30e1f27d 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -17,6 +17,7 @@ PROG_SPU_EMBED_O = $(PROG)_spu-embed.o SOURCES = \ spu_main.c \ + spu_render.c \ spu_texture.c \ spu_tile.c \ spu_tri.c diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 3c9efb4741f..6e02f2c9642 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -34,8 +34,8 @@ #include #include "spu_main.h" +#include "spu_render.h" #include "spu_texture.h" -#include "spu_tri.h" #include "spu_tile.h" #include "pipe/cell/common.h" #include "pipe/p_defines.h" @@ -47,7 +47,7 @@ helpful headers: /opt/ibm/cell-sdk/prototype/sysroot/usr/include/libmisc.h */ -static boolean Debug = FALSE; +boolean Debug = FALSE; struct spu_global spu; @@ -61,7 +61,7 @@ wait_on_mask(unsigned tagMask) } -static void +static INLINE void wait_on_mask_all(unsigned tagMask) { mfc_write_tag_mask( tagMask ); @@ -192,206 +192,6 @@ cmd_clear_surface(const struct cell_command_clear_surface *clear) } -/** - * Given a rendering command's bounding box (in pixels) compute the - * location of the corresponding screen tile bounding box. - */ -static INLINE void -tile_bounding_box(const struct cell_command_render *render, - uint *txmin, uint *tymin, - uint *box_num_tiles, uint *box_width_tiles) -{ -#if 0 - /* Debug: full-window bounding box */ - uint txmax = spu.fb.width_tiles - 1; - uint tymax = spu.fb.height_tiles - 1; - *txmin = 0; - *tymin = 0; - *box_num_tiles = spu.fb.width_tiles * spu.fb.height_tiles; - *box_width_tiles = spu.fb.width_tiles; - (void) render; - (void) txmax; - (void) tymax; -#else - uint txmax, tymax, box_height_tiles; - - *txmin = (uint) render->xmin / TILE_SIZE; - *tymin = (uint) render->ymin / TILE_SIZE; - txmax = (uint) render->xmax / TILE_SIZE; - tymax = (uint) render->ymax / TILE_SIZE; - *box_width_tiles = txmax - *txmin + 1; - box_height_tiles = tymax - *tymin + 1; - *box_num_tiles = *box_width_tiles * box_height_tiles; -#endif -#if 0 - printf("SPU %u: bounds: %g, %g ... %g, %g\n", spu.init.id, - render->xmin, render->ymin, render->xmax, render->ymax); - printf("SPU %u: tiles: %u, %u .. %u, %u\n", - spu.init.id, *txmin, *tymin, txmax, tymax); - ASSERT(render->xmin <= render->xmax); - ASSERT(render->ymin <= render->ymax); -#endif -} - - -/** Check if the tile at (tx,ty) belongs to this SPU */ -static INLINE boolean -my_tile(uint tx, uint ty) -{ - return (spu.fb.width_tiles * ty + tx) % spu.init.num_spus == spu.init.id; -} - - -/** - * Render primitives - * \param pos_incr returns value indicating how may words to skip after - * this command in the batch buffer - */ -static void -cmd_render(const struct cell_command_render *render, uint *pos_incr) -{ - /* we'll DMA into these buffers */ - ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB; - const uint vertex_size = render->vertex_size; /* in bytes */ - /*const*/ uint total_vertex_bytes = render->num_verts * vertex_size; - const ubyte *vertices; - const ushort *indexes; - uint i, j; - - - if (Debug) { - printf("SPU %u: RENDER prim %u, num_vert=%u num_ind=%u " - "inline_vert=%u\n", - spu.init.id, - render->prim_type, - render->num_verts, - render->num_indexes, - render->inline_verts); - - /* - printf(" bound: %g, %g .. %g, %g\n", - render->xmin, render->ymin, render->xmax, render->ymax); - */ - } - - ASSERT(sizeof(*render) % 4 == 0); - ASSERT(total_vertex_bytes % 16 == 0); - - /* indexes are right after the render command in the batch buffer */ - indexes = (const ushort *) (render + 1); - *pos_incr = (render->num_indexes * 2 + 3) / 4; - - - if (render->inline_verts) { - /* Vertices are right after indexes in batch buffer */ - vertices = (const ubyte *) (render + 1) + *pos_incr * 4; - *pos_incr = *pos_incr + total_vertex_bytes / 4; - } - else { - /* Begin DMA fetch of vertex buffer */ - ubyte *src = spu.init.buffers[render->vertex_buf]; - ubyte *dest = vertex_data; - - /* skip vertex data we won't use */ -#if 01 - src += render->min_index * vertex_size; - dest += render->min_index * vertex_size; - total_vertex_bytes -= render->min_index * vertex_size; -#endif - ASSERT(total_vertex_bytes % 16 == 0); - ASSERT_ALIGN16(dest); - ASSERT_ALIGN16(src); - - mfc_get(dest, /* in vertex_data[] array */ - (unsigned int) src, /* src in main memory */ - total_vertex_bytes, /* size */ - TAG_VERTEX_BUFFER, - 0, /* tid */ - 0 /* rid */); - - vertices = vertex_data; - - wait_on_mask(1 << TAG_VERTEX_BUFFER); - } - - - /** - ** find tiles which intersect the prim bounding box - **/ - uint txmin, tymin, box_width_tiles, box_num_tiles; - tile_bounding_box(render, &txmin, &tymin, - &box_num_tiles, &box_width_tiles); - - - /* make sure any pending clears have completed */ - wait_on_mask(1 << TAG_SURFACE_CLEAR); /* XXX temporary */ - - - /** - ** loop over tiles, rendering tris - **/ - for (i = 0; i < box_num_tiles; i++) { - const uint tx = txmin + i % box_width_tiles; - const uint ty = tymin + i / box_width_tiles; - - ASSERT(tx < spu.fb.width_tiles); - ASSERT(ty < spu.fb.height_tiles); - - if (!my_tile(tx, ty)) - continue; - - /* Start fetching color/z tiles. We'll wait for completion when - * we need read/write to them later in triangle rasterization. - */ - if (spu.depth_stencil.depth.enabled) { - if (tile_status_z[ty][tx] != TILE_STATUS_CLEAR) { - get_tile(tx, ty, &ztile, TAG_READ_TILE_Z, 1); - } - } - - if (tile_status[ty][tx] != TILE_STATUS_CLEAR) { - get_tile(tx, ty, &ctile, TAG_READ_TILE_COLOR, 0); - } - - ASSERT(render->prim_type == PIPE_PRIM_TRIANGLES); - ASSERT(render->num_indexes % 3 == 0); - - /* loop over tris */ - for (j = 0; j < render->num_indexes; j += 3) { - const float *v0, *v1, *v2; - - v0 = (const float *) (vertices + indexes[j+0] * vertex_size); - v1 = (const float *) (vertices + indexes[j+1] * vertex_size); - v2 = (const float *) (vertices + indexes[j+2] * vertex_size); - - tri_draw(v0, v1, v2, tx, ty); - } - - /* write color/z tiles back to main framebuffer, if dirtied */ - if (tile_status[ty][tx] == TILE_STATUS_DIRTY) { - put_tile(tx, ty, &ctile, TAG_WRITE_TILE_COLOR, 0); - tile_status[ty][tx] = TILE_STATUS_DEFINED; - } - if (spu.depth_stencil.depth.enabled) { - if (tile_status_z[ty][tx] == TILE_STATUS_DIRTY) { - put_tile(tx, ty, &ztile, TAG_WRITE_TILE_Z, 1); - tile_status_z[ty][tx] = TILE_STATUS_DEFINED; - } - } - - /* XXX move these... */ - wait_on_mask(1 << TAG_WRITE_TILE_COLOR); - if (spu.depth_stencil.depth.enabled) { - wait_on_mask(1 << TAG_WRITE_TILE_Z); - } - } - - if (Debug) - printf("SPU %u: RENDER done\n", - spu.init.id); -} - - static void cmd_release_verts(const struct cell_command_release_verts *release) { diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 480c54ebd08..009e046ba5e 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -70,6 +70,7 @@ struct spu_global extern struct spu_global spu; +extern boolean Debug; diff --git a/src/mesa/pipe/cell/spu/spu_render.c b/src/mesa/pipe/cell/spu/spu_render.c new file mode 100644 index 00000000000..21a286a23d3 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_render.c @@ -0,0 +1,240 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include +#include +#include + +#include "spu_main.h" +#include "spu_render.h" +#include "spu_tri.h" +#include "spu_tile.h" +#include "pipe/cell/common.h" + + + +/** + * Given a rendering command's bounding box (in pixels) compute the + * location of the corresponding screen tile bounding box. + */ +static INLINE void +tile_bounding_box(const struct cell_command_render *render, + uint *txmin, uint *tymin, + uint *box_num_tiles, uint *box_width_tiles) +{ +#if 0 + /* Debug: full-window bounding box */ + uint txmax = spu.fb.width_tiles - 1; + uint tymax = spu.fb.height_tiles - 1; + *txmin = 0; + *tymin = 0; + *box_num_tiles = spu.fb.width_tiles * spu.fb.height_tiles; + *box_width_tiles = spu.fb.width_tiles; + (void) render; + (void) txmax; + (void) tymax; +#else + uint txmax, tymax, box_height_tiles; + + *txmin = (uint) render->xmin / TILE_SIZE; + *tymin = (uint) render->ymin / TILE_SIZE; + txmax = (uint) render->xmax / TILE_SIZE; + tymax = (uint) render->ymax / TILE_SIZE; + *box_width_tiles = txmax - *txmin + 1; + box_height_tiles = tymax - *tymin + 1; + *box_num_tiles = *box_width_tiles * box_height_tiles; +#endif +#if 0 + printf("SPU %u: bounds: %g, %g ... %g, %g\n", spu.init.id, + render->xmin, render->ymin, render->xmax, render->ymax); + printf("SPU %u: tiles: %u, %u .. %u, %u\n", + spu.init.id, *txmin, *tymin, txmax, tymax); + ASSERT(render->xmin <= render->xmax); + ASSERT(render->ymin <= render->ymax); +#endif +} + + +/** Check if the tile at (tx,ty) belongs to this SPU */ +static INLINE boolean +my_tile(uint tx, uint ty) +{ + return (spu.fb.width_tiles * ty + tx) % spu.init.num_spus == spu.init.id; +} + + +/** + * Render primitives + * \param pos_incr returns value indicating how may words to skip after + * this command in the batch buffer + */ +void +cmd_render(const struct cell_command_render *render, uint *pos_incr) +{ + /* we'll DMA into these buffers */ + ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB; + const uint vertex_size = render->vertex_size; /* in bytes */ + /*const*/ uint total_vertex_bytes = render->num_verts * vertex_size; + const ubyte *vertices; + const ushort *indexes; + uint i, j; + + + if (Debug) { + printf("SPU %u: RENDER prim %u, num_vert=%u num_ind=%u " + "inline_vert=%u\n", + spu.init.id, + render->prim_type, + render->num_verts, + render->num_indexes, + render->inline_verts); + + /* + printf(" bound: %g, %g .. %g, %g\n", + render->xmin, render->ymin, render->xmax, render->ymax); + */ + } + + ASSERT(sizeof(*render) % 4 == 0); + ASSERT(total_vertex_bytes % 16 == 0); + + /* indexes are right after the render command in the batch buffer */ + indexes = (const ushort *) (render + 1); + *pos_incr = (render->num_indexes * 2 + 3) / 4; + + + if (render->inline_verts) { + /* Vertices are right after indexes in batch buffer */ + vertices = (const ubyte *) (render + 1) + *pos_incr * 4; + *pos_incr = *pos_incr + total_vertex_bytes / 4; + } + else { + /* Begin DMA fetch of vertex buffer */ + ubyte *src = spu.init.buffers[render->vertex_buf]; + ubyte *dest = vertex_data; + + /* skip vertex data we won't use */ +#if 01 + src += render->min_index * vertex_size; + dest += render->min_index * vertex_size; + total_vertex_bytes -= render->min_index * vertex_size; +#endif + ASSERT(total_vertex_bytes % 16 == 0); + ASSERT_ALIGN16(dest); + ASSERT_ALIGN16(src); + + mfc_get(dest, /* in vertex_data[] array */ + (unsigned int) src, /* src in main memory */ + total_vertex_bytes, /* size */ + TAG_VERTEX_BUFFER, + 0, /* tid */ + 0 /* rid */); + + vertices = vertex_data; + + wait_on_mask(1 << TAG_VERTEX_BUFFER); + } + + + /** + ** find tiles which intersect the prim bounding box + **/ + uint txmin, tymin, box_width_tiles, box_num_tiles; + tile_bounding_box(render, &txmin, &tymin, + &box_num_tiles, &box_width_tiles); + + + /* make sure any pending clears have completed */ + wait_on_mask(1 << TAG_SURFACE_CLEAR); /* XXX temporary */ + + + /** + ** loop over tiles, rendering tris + **/ + for (i = 0; i < box_num_tiles; i++) { + const uint tx = txmin + i % box_width_tiles; + const uint ty = tymin + i / box_width_tiles; + + ASSERT(tx < spu.fb.width_tiles); + ASSERT(ty < spu.fb.height_tiles); + + if (!my_tile(tx, ty)) + continue; + + /* Start fetching color/z tiles. We'll wait for completion when + * we need read/write to them later in triangle rasterization. + */ + if (spu.depth_stencil.depth.enabled) { + if (tile_status_z[ty][tx] != TILE_STATUS_CLEAR) { + get_tile(tx, ty, &ztile, TAG_READ_TILE_Z, 1); + } + } + + if (tile_status[ty][tx] != TILE_STATUS_CLEAR) { + get_tile(tx, ty, &ctile, TAG_READ_TILE_COLOR, 0); + } + + ASSERT(render->prim_type == PIPE_PRIM_TRIANGLES); + ASSERT(render->num_indexes % 3 == 0); + + /* loop over tris */ + for (j = 0; j < render->num_indexes; j += 3) { + const float *v0, *v1, *v2; + + v0 = (const float *) (vertices + indexes[j+0] * vertex_size); + v1 = (const float *) (vertices + indexes[j+1] * vertex_size); + v2 = (const float *) (vertices + indexes[j+2] * vertex_size); + + tri_draw(v0, v1, v2, tx, ty); + } + + /* write color/z tiles back to main framebuffer, if dirtied */ + if (tile_status[ty][tx] == TILE_STATUS_DIRTY) { + put_tile(tx, ty, &ctile, TAG_WRITE_TILE_COLOR, 0); + tile_status[ty][tx] = TILE_STATUS_DEFINED; + } + if (spu.depth_stencil.depth.enabled) { + if (tile_status_z[ty][tx] == TILE_STATUS_DIRTY) { + put_tile(tx, ty, &ztile, TAG_WRITE_TILE_Z, 1); + tile_status_z[ty][tx] = TILE_STATUS_DEFINED; + } + } + + /* XXX move these... */ + wait_on_mask(1 << TAG_WRITE_TILE_COLOR); + if (spu.depth_stencil.depth.enabled) { + wait_on_mask(1 << TAG_WRITE_TILE_Z); + } + } + + if (Debug) + printf("SPU %u: RENDER done\n", + spu.init.id); +} + + diff --git a/src/mesa/pipe/cell/spu/spu_render.h b/src/mesa/pipe/cell/spu/spu_render.h new file mode 100644 index 00000000000..fbcdc5ec316 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_render.h @@ -0,0 +1,38 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef SPU_RENDER_H +#define SPU_RENDER_H + +#include "pipe/cell/common.h" + +extern void +cmd_render(const struct cell_command_render *render, uint *pos_incr); + +#endif /* SPU_RENDER_H */ + -- cgit v1.2.3 From 1c65928d8400a350993687d7039e5e47371ae8b8 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 28 Jan 2008 18:17:55 -0700 Subject: Cell: add OPT_FLAGS var --- configs/linux-cell | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/configs/linux-cell b/configs/linux-cell index 4f0086cc1ff..3d874491e41 100644 --- a/configs/linux-cell +++ b/configs/linux-cell @@ -10,11 +10,13 @@ CC = ppu32-gcc CXX = ppu32-g++ HOST_CC = gcc +OPT_FLAGS = -g + # Cell SDK location SDK = /opt/ibm/cell-sdk/prototype/sysroot/usr -CFLAGS = -g -Wall -Winline -fPIC -m32 -mabi=altivec -maltivec -I. -I$(SDK)/include -DGALLIUM_CELL +CFLAGS = $(OPT_FLAGS) -Wall -Winline -fPIC -m32 -mabi=altivec -maltivec -I. -I$(SDK)/include -DGALLIUM_CELL CXXFLAGS = $(CFLAGS) @@ -34,7 +36,7 @@ GL_LIB_DEPS = $(EXTRA_LIB_PATH) -lX11 -lXext -lm -lpthread \ SPU_CC = spu-gcc -SPU_CFLAGS = -g -W -Wall -Winline -Wmissing-prototypes -Wno-main -I. -I $(SDK)/spu/include -include spu_intrinsics.h -I $(TOP)/src/mesa/ +SPU_CFLAGS = $(OPT_FLAGS) -W -Wall -Winline -Wmissing-prototypes -Wno-main -I. -I $(SDK)/spu/include -include spu_intrinsics.h -I $(TOP)/src/mesa/ SPU_LFLAGS = -L$(SDK)/spu/lib -Wl,-N -lmisc -- cgit v1.2.3 From 7710b36d28859222f9b0bf03ab3d0cdf79d39c64 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 11:21:29 +0900 Subject: gallium: Add extern keyword to global. --- src/mesa/pipe/draw/draw_vf.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index 7d90f35b0f4..c0fa063c52f 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -195,7 +195,8 @@ struct draw_vf_format_info { const unsigned attrsize; }; -const struct draw_vf_format_info draw_vf_format_info[DRAW_EMIT_MAX]; +extern const struct draw_vf_format_info +draw_vf_format_info[DRAW_EMIT_MAX]; #endif -- cgit v1.2.3 From 8a88f5e40f75ac52d02c1afbcc7dd612904b4f78 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 18:01:17 +0900 Subject: gallium: Allow draw_vf usage to be controlled at runtime. --- src/mesa/pipe/draw/draw_vbuf.c | 149 ++++++++++++++++++++--------------------- 1 file changed, 71 insertions(+), 78 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vbuf.c b/src/mesa/pipe/draw/draw_vbuf.c index 8ca225c65aa..2309ed9f129 100644 --- a/src/mesa/pipe/draw/draw_vbuf.c +++ b/src/mesa/pipe/draw/draw_vbuf.c @@ -127,15 +127,9 @@ emit_vertex( struct vbuf_stage *vbuf, struct vertex_header *vertex ) { #if 0 - const struct vertex_info *vinfo = vbuf->vinfo; - - uint i; - uint count = 0; /* for debug/sanity */ - - assert(vinfo == vbuf->render->get_vertex_info(vbuf->render)); - -// fprintf(stderr, "emit vertex %d to %p\n", -// vbuf->nr_vertices, vbuf->vertex_ptr); + fprintf(stderr, "emit vertex %d to %p\n", + vbuf->nr_vertices, vbuf->vertex_ptr); +#endif if(vertex->vertex_id != UNDEFINED_VERTEX_ID) { if(vertex->vertex_id < vbuf->nr_vertices) @@ -148,75 +142,72 @@ emit_vertex( struct vbuf_stage *vbuf, vertex->vertex_id = vbuf->nr_vertices++; - for (i = 0; i < vinfo->num_attribs; i++) { - uint j = vinfo->src_index[i]; - switch (vinfo->emit[i]) { - case EMIT_OMIT: - /* no-op */ - break; - case EMIT_ALL: - /* just copy the whole vertex as-is to the vbuf */ - assert(i == 0); - assert(j == 0); - memcpy(vbuf->vertex_ptr, vertex, vinfo->size * 4); - vbuf->vertex_ptr += vinfo->size; - count += vinfo->size; - break; - case EMIT_1F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - count++; - break; - case EMIT_1F_PSIZE: - *vbuf->vertex_ptr++ = fui(vbuf->stage.draw->rasterizer->point_size); - count++; - break; - case EMIT_2F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); - count += 2; - break; - case EMIT_3F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][2]); - count += 3; - break; - case EMIT_4F: - *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][2]); - *vbuf->vertex_ptr++ = fui(vertex->data[j][3]); - count += 4; - break; - case EMIT_4UB: - *vbuf->vertex_ptr++ = pack_ub4(float_to_ubyte( vertex->data[j][2] ), - float_to_ubyte( vertex->data[j][1] ), - float_to_ubyte( vertex->data[j][0] ), - float_to_ubyte( vertex->data[j][3] )); - count += 1; - break; - default: - assert(0); + if(!vbuf->vf) { + const struct vertex_info *vinfo = vbuf->vinfo; + uint i; + uint count = 0; /* for debug/sanity */ + + assert(vinfo == vbuf->render->get_vertex_info(vbuf->render)); + + for (i = 0; i < vinfo->num_attribs; i++) { + uint j = vinfo->src_index[i]; + switch (vinfo->emit[i]) { + case EMIT_OMIT: + /* no-op */ + break; + case EMIT_ALL: + /* just copy the whole vertex as-is to the vbuf */ + assert(i == 0); + assert(j == 0); + memcpy(vbuf->vertex_ptr, vertex, vinfo->size * 4); + vbuf->vertex_ptr += vinfo->size; + count += vinfo->size; + break; + case EMIT_1F: + *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); + count++; + break; + case EMIT_1F_PSIZE: + *vbuf->vertex_ptr++ = fui(vbuf->stage.draw->rasterizer->point_size); + count++; + break; + case EMIT_2F: + *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); + *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); + count += 2; + break; + case EMIT_3F: + *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); + *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); + *vbuf->vertex_ptr++ = fui(vertex->data[j][2]); + count += 3; + break; + case EMIT_4F: + *vbuf->vertex_ptr++ = fui(vertex->data[j][0]); + *vbuf->vertex_ptr++ = fui(vertex->data[j][1]); + *vbuf->vertex_ptr++ = fui(vertex->data[j][2]); + *vbuf->vertex_ptr++ = fui(vertex->data[j][3]); + count += 4; + break; + case EMIT_4UB: + *vbuf->vertex_ptr++ = pack_ub4(float_to_ubyte( vertex->data[j][2] ), + float_to_ubyte( vertex->data[j][1] ), + float_to_ubyte( vertex->data[j][0] ), + float_to_ubyte( vertex->data[j][3] )); + count += 1; + break; + default: + assert(0); + } } + assert(count == vinfo->size); } - assert(count == vinfo->size); -#else - if(vertex->vertex_id != UNDEFINED_VERTEX_ID) { - if(vertex->vertex_id < vbuf->nr_vertices) - return; - else - fprintf(stderr, "Bad vertex id 0x%04x (>= 0x%04x)\n", - vertex->vertex_id, vbuf->nr_vertices); - return; + else { + draw_vf_set_data(vbuf->vf, vertex->data); + draw_vf_emit_vertices(vbuf->vf, 1, vbuf->vertex_ptr); + + vbuf->vertex_ptr += vbuf->vertex_size/4; } - - vertex->vertex_id = vbuf->nr_vertices++; - - draw_vf_set_data(vbuf->vf, vertex->data); - draw_vf_emit_vertices(vbuf->vf, 1, vbuf->vertex_ptr); - - vbuf->vertex_ptr += vbuf->vertex_size/4; -#endif } @@ -229,6 +220,9 @@ vbuf_set_vf_attributes(struct vbuf_stage *vbuf ) uint count = 0; /* for debug/sanity */ unsigned nr_attrs = 0; + if(!vbuf->vf) + return; + // fprintf(stderr, "emit vertex %d to %p\n", // vbuf->nr_vertices, vbuf->vertex_ptr); @@ -625,9 +619,8 @@ struct draw_stage *draw_vbuf_stage( struct draw_context *draw, vbuf->prim = ~0; - vbuf->vf = draw_vf_create(); - if(!vbuf->vf) - vbuf_destroy(&vbuf->stage); + if(!GETENV("GALLIUM_NOVF")) + vbuf->vf = draw_vf_create(); return &vbuf->stage; } -- cgit v1.2.3 From c74f4a10f91acc4eca109c1be39fd320639bfa59 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 20:46:48 +0900 Subject: gallium: Emit constants. --- src/mesa/pipe/draw/draw_vf.c | 10 +++++++-- src/mesa/pipe/draw/draw_vf.h | 22 ++++++++++++++++++- src/mesa/pipe/draw/draw_vf_generic.c | 42 +++++++++++++++++++++++++----------- src/mesa/pipe/draw/draw_vf_sse.c | 4 ++++ 4 files changed, 62 insertions(+), 16 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 4fc2312ad1e..958d31933ba 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -182,6 +182,9 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, vf->attr[j].insert = draw_vf_format_info[format].insert; vf->attr[j].vertattrsize = draw_vf_format_info[format].attrsize; vf->attr[j].vertoffset = offset; + vf->attr[j].isconst = draw_vf_format_info[format].isconst; + if(vf->attr[j].isconst) + memcpy(vf->attr[j].data, &map[i].data, vf->attr[j].vertattrsize); if (DBG) _mesa_printf("%d: %s, offset %d\n", i, @@ -240,8 +243,11 @@ void draw_vf_set_data( struct draw_vertex_fetch *vf, for (j = 0; j < vf->attr_count; j++) { a[j].inputstride = 0; /* XXX: one-vertex-max ATM */ a[j].inputsize = 4; - a[j].do_insert = a[j].insert[4 - 1]; - a[j].inputptr = (uint8_t *)&data[a[j].attrib][0]; + a[j].do_insert = a[j].insert[4 - 1]; + if(a[j].isconst) + a[j].inputptr = a[j].data; + else + a[j].inputptr = (uint8_t *)&data[a[j].attrib][0]; } } diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index c0fa063c52f..911ea07bdfe 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -48,14 +48,30 @@ enum draw_vf_attr_format { DRAW_EMIT_4UB_4F_BGRA, /**< for color */ DRAW_EMIT_4UB_4F_ARGB, /**< for color */ DRAW_EMIT_4UB_4F_ABGR, /**< for color */ + DRAW_EMIT_1F_CONST, + DRAW_EMIT_2F_CONST, + DRAW_EMIT_3F_CONST, + DRAW_EMIT_4F_CONST, DRAW_EMIT_PAD, /**< leave a hole of 'offset' bytes */ DRAW_EMIT_MAX }; -struct draw_vf_attr_map { +struct draw_vf_attr_map +{ + /** Input attribute number */ unsigned attrib; + enum draw_vf_attr_format format; + unsigned offset; + + /** + * Constant data for DRAW_EMIT_*_CONST + */ + union { + uint8_t ub[4]; + float f[4]; + } data; }; struct draw_vertex_fetch; @@ -124,6 +140,9 @@ struct draw_vf_attr unsigned inputsize; unsigned inputstride; unsigned vertoffset; /**< position of the attrib in the vertex struct */ + + boolean isconst; /**< read from const data below */ + uint8_t data[16]; unsigned attrib; /**< which vertex attrib (0=position, etc) */ unsigned vertattrsize; /**< size of the attribute in bytes */ @@ -193,6 +212,7 @@ struct draw_vf_format_info { const char *name; draw_vf_insert_func insert[4]; const unsigned attrsize; + const boolean isconst; }; extern const struct draw_vf_format_info diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c index a16eb456b74..0caa798396d 100644 --- a/src/mesa/pipe/draw/draw_vf_generic.c +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -387,62 +387,78 @@ const struct draw_vf_format_info draw_vf_format_info[DRAW_EMIT_MAX] = { { "1f", { insert_1f_1, insert_1f_1, insert_1f_1, insert_1f_1 }, - sizeof(float) }, + sizeof(float), FALSE }, { "2f", { insert_2f_1, insert_2f_2, insert_2f_2, insert_2f_2 }, - 2 * sizeof(float) }, + 2 * sizeof(float), FALSE }, { "3f", { insert_3f_1, insert_3f_2, insert_3f_3, insert_3f_3 }, - 3 * sizeof(float) }, + 3 * sizeof(float), FALSE }, { "4f", { insert_4f_1, insert_4f_2, insert_4f_3, insert_4f_4 }, - 4 * sizeof(float) }, + 4 * sizeof(float), FALSE }, { "3f_xyw", { insert_3f_xyw_err, insert_3f_xyw_err, insert_3f_xyw_err, insert_3f_xyw_4 }, - 3 * sizeof(float) }, + 3 * sizeof(float), FALSE }, { "1ub_1f", { insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1, insert_1ub_1f_1 }, - sizeof(uint8_t) }, + sizeof(uint8_t), FALSE }, { "3ub_3f_rgb", { insert_3ub_3f_rgb_1, insert_3ub_3f_rgb_2, insert_3ub_3f_rgb_3, insert_3ub_3f_rgb_3 }, - 3 * sizeof(uint8_t) }, + 3 * sizeof(uint8_t), FALSE }, { "3ub_3f_bgr", { insert_3ub_3f_bgr_1, insert_3ub_3f_bgr_2, insert_3ub_3f_bgr_3, insert_3ub_3f_bgr_3 }, - 3 * sizeof(uint8_t) }, + 3 * sizeof(uint8_t), FALSE }, { "4ub_4f_rgba", { insert_4ub_4f_rgba_1, insert_4ub_4f_rgba_2, insert_4ub_4f_rgba_3, insert_4ub_4f_rgba_4 }, - 4 * sizeof(uint8_t) }, + 4 * sizeof(uint8_t), FALSE }, { "4ub_4f_bgra", { insert_4ub_4f_bgra_1, insert_4ub_4f_bgra_2, insert_4ub_4f_bgra_3, insert_4ub_4f_bgra_4 }, - 4 * sizeof(uint8_t) }, + 4 * sizeof(uint8_t), FALSE }, { "4ub_4f_argb", { insert_4ub_4f_argb_1, insert_4ub_4f_argb_2, insert_4ub_4f_argb_3, insert_4ub_4f_argb_4 }, - 4 * sizeof(uint8_t) }, + 4 * sizeof(uint8_t), FALSE }, { "4ub_4f_abgr", { insert_4ub_4f_abgr_1, insert_4ub_4f_abgr_2, insert_4ub_4f_abgr_3, insert_4ub_4f_abgr_4 }, - 4 * sizeof(uint8_t) }, + 4 * sizeof(uint8_t), FALSE }, + + { "1f_const", + { insert_1f_1, insert_1f_1, insert_1f_1, insert_1f_1 }, + sizeof(float), TRUE }, + + { "2f_const", + { insert_2f_1, insert_2f_2, insert_2f_2, insert_2f_2 }, + 2 * sizeof(float), TRUE }, + + { "3f_const", + { insert_3f_1, insert_3f_2, insert_3f_3, insert_3f_3 }, + 3 * sizeof(float), TRUE }, + + { "4f_const", + { insert_4f_1, insert_4f_2, insert_4f_3, insert_4f_4 }, + 4 * sizeof(float), TRUE }, { "pad", { NULL, NULL, NULL, NULL }, - 0 } + 0, FALSE }, }; diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c index 4036ded1d84..1389e6cfb9c 100644 --- a/src/mesa/pipe/draw/draw_vf_sse.c +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -388,18 +388,21 @@ static boolean build_vertex_emit( struct x86_program *p ) */ switch (a->format) { case DRAW_EMIT_1F: + case DRAW_EMIT_1F_CONST: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 1, x86_deref(srcECX), a->inputsize); emit_store(p, dest, 1, temp); update_src_ptr(p, srcECX, vfESI, a); break; case DRAW_EMIT_2F: + case DRAW_EMIT_2F_CONST: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 2, x86_deref(srcECX), a->inputsize); emit_store(p, dest, 2, temp); update_src_ptr(p, srcECX, vfESI, a); break; case DRAW_EMIT_3F: + case DRAW_EMIT_3F_CONST: /* Potentially the worst case - hardcode 2+1 copying: */ if (0) { @@ -423,6 +426,7 @@ static boolean build_vertex_emit( struct x86_program *p ) } break; case DRAW_EMIT_4F: + case DRAW_EMIT_4F_CONST: get_src_ptr(p, srcECX, vfESI, a); emit_load(p, temp, 4, x86_deref(srcECX), a->inputsize); emit_store(p, dest, 4, temp); -- cgit v1.2.3 From 2da0724e99785c2bf854fc8a7ba40765b0563088 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Tue, 29 Jan 2008 20:47:30 +0900 Subject: gallium: Emit point size as a constant. --- src/mesa/pipe/draw/draw_vbuf.c | 85 ++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 81 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vbuf.c b/src/mesa/pipe/draw/draw_vbuf.c index 2309ed9f129..92a8b9fbcfa 100644 --- a/src/mesa/pipe/draw/draw_vbuf.c +++ b/src/mesa/pipe/draw/draw_vbuf.c @@ -115,6 +115,70 @@ check_space( struct vbuf_stage *vbuf, unsigned nr ) } +#if 0 +static INLINE void +dump_emitted_vertex(const struct vertex_info *vinfo, const uint8_t *data) +{ + assert(vinfo == vbuf->render->get_vertex_info(vbuf->render)); + unsigned i, j, k; + + for (i = 0; i < vinfo->num_attribs; i++) { + j = vinfo->src_index[i]; + switch (vinfo->emit[i]) { + case EMIT_OMIT: + fprintf(stderr, "EMIT_OMIT:"); + break; + case EMIT_ALL: + assert(i == 0); + assert(j == 0); + fprintf(stderr, "EMIT_ALL:\t"); + for(k = 0; k < vinfo->size*4; ++k) + fprintf(stderr, "%02x ", *data++); + break; + case EMIT_1F: + fprintf(stderr, "EMIT_1F:\t"); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_1F_PSIZE: + fprintf(stderr, "EMIT_1F_PSIZE:\t"); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_2F: + fprintf(stderr, "EMIT_2F:\t"); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_3F: + fprintf(stderr, "EMIT_3F:\t"); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + data += sizeof(float); + break; + case EMIT_4F: + fprintf(stderr, "EMIT_4F:\t"); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + break; + case EMIT_4UB: + fprintf(stderr, "EMIT_4UB:\t"); + fprintf(stderr, "%u ", *data++); + fprintf(stderr, "%u ", *data++); + fprintf(stderr, "%u ", *data++); + fprintf(stderr, "%u ", *data++); + break; + default: + assert(0); + } + fprintf(stderr, "\n"); + } + fprintf(stderr, "\n"); +} +#endif + + /** * Extract the needed fields from post-transformed vertex and emit * a hardware(driver) vertex. @@ -190,7 +254,7 @@ emit_vertex( struct vbuf_stage *vbuf, count += 4; break; case EMIT_4UB: - *vbuf->vertex_ptr++ = pack_ub4(float_to_ubyte( vertex->data[j][2] ), + *vbuf->vertex_ptr++ = pack_ub4(float_to_ubyte( vertex->data[j][2] ), float_to_ubyte( vertex->data[j][1] ), float_to_ubyte( vertex->data[j][0] ), float_to_ubyte( vertex->data[j][3] )); @@ -201,6 +265,20 @@ emit_vertex( struct vbuf_stage *vbuf, } } assert(count == vinfo->size); +#if 0 + { + static float data[256]; + draw_vf_set_data(vbuf->vf, vertex->data); + draw_vf_emit_vertices(vbuf->vf, 1, data); + if(memcmp((uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size, data, vbuf->vertex_size)) { + fprintf(stderr, "With VF:\n"); + dump_emitted_vertex(vbuf->vinfo, (uint8_t *)data); + fprintf(stderr, "Without VF:\n"); + dump_emitted_vertex(vbuf->vinfo, (uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size); + assert(0); + } + } +#endif } else { draw_vf_set_data(vbuf->vf, vertex->data); @@ -297,11 +375,10 @@ vbuf_set_vf_attributes(struct vbuf_stage *vbuf ) count++; break; case EMIT_1F_PSIZE: - /* FIXME */ - assert(0); attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_PAD; + attrs[nr_attrs].format = DRAW_EMIT_1F_CONST; attrs[nr_attrs].offset = 0; + attrs[nr_attrs].data.f[0] = vbuf->stage.draw->rasterizer->point_size; nr_attrs++; count++; break; -- cgit v1.2.3 From 88469bf544b62fb0786d88383901914140afa56c Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Tue, 29 Jan 2008 12:37:07 +0000 Subject: gallium: don't rely on assert(0) for error handling - may be disabled --- src/mesa/state_tracker/st_draw.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 8ef50ee7684..c9b8e784855 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -298,6 +298,7 @@ st_draw_vbo(GLcontext *ctx, break; default: assert(0); + return; } /* get/create the index buffer object */ @@ -570,6 +571,7 @@ st_feedback_draw_vbo(GLcontext *ctx, break; default: assert(0); + return; } map = pipe->winsys->buffer_map(pipe->winsys, -- cgit v1.2.3 From d7d3c752368a236dd4755b00175d0e13212fac47 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Tue, 29 Jan 2008 12:37:47 +0000 Subject: gallium: streamline various unfilled & stippled paths --- src/mesa/pipe/draw/draw_prim.c | 158 +++++++++++++++++++++++++++++------------ 1 file changed, 113 insertions(+), 45 deletions(-) diff --git a/src/mesa/pipe/draw/draw_prim.c b/src/mesa/pipe/draw/draw_prim.c index 2a612a16733..41b3fddcc10 100644 --- a/src/mesa/pipe/draw/draw_prim.c +++ b/src/mesa/pipe/draw/draw_prim.c @@ -69,28 +69,46 @@ static void draw_prim_queue_flush( struct draw_context *draw ) * draw->pipeline->first is often changed by the first call to tri(), * line(), etc. */ - switch (draw->reduced_prim) { - case RP_TRI: - for (i = 0; i < draw->pq.queue_nr; i++) { - if (draw->pq.queue[i].reset_line_stipple) - draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); - - draw->pipeline.first->tri( draw->pipeline.first, &draw->pq.queue[i] ); + if (draw->rasterizer->line_stipple_enable) { + switch (draw->reduced_prim) { + case RP_TRI: + for (i = 0; i < draw->pq.queue_nr; i++) { + if (draw->pq.queue[i].reset_line_stipple) + draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); + + draw->pipeline.first->tri( draw->pipeline.first, &draw->pq.queue[i] ); + } + break; + case RP_LINE: + for (i = 0; i < draw->pq.queue_nr; i++) { + if (draw->pq.queue[i].reset_line_stipple) + draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); + + draw->pipeline.first->line( draw->pipeline.first, &draw->pq.queue[i] ); + } + break; + case RP_POINT: + draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); + for (i = 0; i < draw->pq.queue_nr; i++) + draw->pipeline.first->point( draw->pipeline.first, &draw->pq.queue[i] ); + break; } - break; - case RP_LINE: - for (i = 0; i < draw->pq.queue_nr; i++) { - if (draw->pq.queue[i].reset_line_stipple) - draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); - - draw->pipeline.first->line( draw->pipeline.first, &draw->pq.queue[i] ); + } + else { + switch (draw->reduced_prim) { + case RP_TRI: + for (i = 0; i < draw->pq.queue_nr; i++) + draw->pipeline.first->tri( draw->pipeline.first, &draw->pq.queue[i] ); + break; + case RP_LINE: + for (i = 0; i < draw->pq.queue_nr; i++) + draw->pipeline.first->line( draw->pipeline.first, &draw->pq.queue[i] ); + break; + case RP_POINT: + for (i = 0; i < draw->pq.queue_nr; i++) + draw->pipeline.first->point( draw->pipeline.first, &draw->pq.queue[i] ); + break; } - break; - case RP_POINT: - draw->pipeline.first->reset_stipple_counter( draw->pipeline.first ); - for (i = 0; i < draw->pq.queue_nr; i++) - draw->pipeline.first->point( draw->pipeline.first, &draw->pq.queue[i] ); - break; } draw->pq.queue_nr = 0; @@ -231,7 +249,7 @@ static void do_ef_triangle( struct draw_context *draw, } -static void do_quad( struct draw_context *draw, +static void do_ef_quad( struct draw_context *draw, unsigned v0, unsigned v1, unsigned v2, @@ -243,6 +261,16 @@ static void do_quad( struct draw_context *draw, do_ef_triangle( draw, 0, omitEdge3, v1, v2, v3 ); } +static void do_quad( struct draw_context *draw, + unsigned v0, + unsigned v1, + unsigned v2, + unsigned v3 ) +{ + do_triangle( draw, v0, v1, v3 ); + do_triangle( draw, v1, v2, v3 ); +} + /** * Main entrypoint to draw some number of points/lines/triangles @@ -252,6 +280,8 @@ draw_prim( struct draw_context *draw, unsigned prim, unsigned start, unsigned count ) { unsigned i; + boolean unfilled = (draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || + draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL); // _mesa_printf("%s (%d) %d/%d\n", __FUNCTION__, draw->prim, start, count ); @@ -289,24 +319,32 @@ draw_prim( struct draw_context *draw, break; case PIPE_PRIM_LINE_STRIP: - if (count >= 2) { - for (i = 1; i < count; i++) { - do_line( draw, - i == 1, - start + i - 1, - start + i ); - } + for (i = 1; i < count; i++) { + do_line( draw, + i == 1, + start + i - 1, + start + i ); } break; case PIPE_PRIM_TRIANGLES: - for (i = 0; i+2 < count; i += 3) { - do_ef_triangle( draw, - 1, - ~0, + if (unfilled) { + for (i = 0; i+2 < count; i += 3) { + do_ef_triangle( draw, + 1, + ~0, + start + i + 0, + start + i + 1, + start + i + 2 ); + } + } + else { + for (i = 0; i+2 < count; i += 3) { + do_triangle( draw, start + i + 0, start + i + 1, start + i + 2 ); + } } break; @@ -340,27 +378,49 @@ draw_prim( struct draw_context *draw, case PIPE_PRIM_QUADS: - for (i = 0; i+3 < count; i += 4) { - do_quad( draw, - start + i + 0, - start + i + 1, - start + i + 2, - start + i + 3); + if (unfilled) { + for (i = 0; i+3 < count; i += 4) { + do_ef_quad( draw, + start + i + 0, + start + i + 1, + start + i + 2, + start + i + 3); + } + } + else { + for (i = 0; i+3 < count; i += 4) { + do_quad( draw, + start + i + 0, + start + i + 1, + start + i + 2, + start + i + 3); + } } break; case PIPE_PRIM_QUAD_STRIP: - for (i = 0; i+3 < count; i += 2) { - do_quad( draw, - start + i + 2, - start + i + 0, - start + i + 1, - start + i + 3); + if (unfilled) { + for (i = 0; i+3 < count; i += 2) { + do_ef_quad( draw, + start + i + 2, + start + i + 0, + start + i + 1, + start + i + 3); + } + } + else { + for (i = 0; i+3 < count; i += 2) { + do_quad( draw, + start + i + 2, + start + i + 0, + start + i + 1, + start + i + 3); + } } break; case PIPE_PRIM_POLYGON: - if (count >= 3) { + if (unfilled) { unsigned ef_mask = (1<<2) | (1<<0); for (i = 0; i+2 < count; i++) { @@ -378,6 +438,14 @@ draw_prim( struct draw_context *draw, ef_mask &= ~(1<<2); } } + else { + for (i = 0; i+2 < count; i++) { + do_triangle( draw, + start + i + 1, + start + i + 2, + start + 0); + } + } break; default: -- cgit v1.2.3 From b63f994ec7742da53b4c32ff7ee8219bbd72c2ef Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Tue, 29 Jan 2008 15:17:56 +0000 Subject: gallium: weaken assert slightly --- src/mesa/pipe/draw/draw_vf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 958d31933ba..06b84b93cc8 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -174,7 +174,7 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, } else { - assert(vf->lookup[map[i].attrib] == 0); + assert(vf->lookup[map[i].attrib] == 0 || format == DRAW_EMIT_1F_CONST); vf->lookup[map[i].attrib] = &vf->attr[j]; vf->attr[j].attrib = map[i].attrib; -- cgit v1.2.3 From 2b47b5b413a6511ed45ce1e44a88822c35b084ee Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 29 Jan 2008 11:22:57 -0700 Subject: Cell: use _pack_rgba8() from pack_rgba8.h to do float[4]->uint color conversion texcyl.c is twice as fast now in non-texture mode --- src/mesa/pipe/cell/spu/spu_tri.c | 15 ++++----------- 1 file changed, 4 insertions(+), 11 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 19a231d9c4d..7c6a54134f8 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -29,6 +29,8 @@ * Triangle rendering within a tile. */ +#include + #include "pipe/p_compiler.h" #include "pipe/p_format.h" #include "pipe/p_util.h" @@ -38,7 +40,6 @@ #include "spu_tri.h" - /** * Simplified types taken from other parts of Gallium */ @@ -252,19 +253,11 @@ eval_z( struct setup_stage *setup, static INLINE uint pack_color(const float color[4]) { - uint r = (uint) (color[0] * 255.0); - uint g = (uint) (color[1] * 255.0); - uint b = (uint) (color[2] * 255.0); - uint a = (uint) (color[3] * 255.0); - r = MIN2(r, 255); - g = MIN2(g, 255); - b = MIN2(b, 255); - a = MIN2(a, 255); switch (spu.fb.color_format) { case PIPE_FORMAT_A8R8G8B8_UNORM: - return (a << 24) | (r << 16) | (g << 8) | b; + return _pack_rgba8(color[3], color[0], color[1], color[2]); case PIPE_FORMAT_B8G8R8A8_UNORM: - return (b << 24) | (g << 16) | (r << 8) | a; + return _pack_rgba8(color[2], color[1], color[0], color[3]); default: ASSERT(0); return 0; -- cgit v1.2.3 From a5273f0fac01f5864a1cfcb82d9302dd755375e9 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 30 Jan 2008 15:34:02 +1100 Subject: nouveau: 0xdeadc0de --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 23 ---------------------- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 --- 2 files changed, 26 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 288674f231b..4c235845b7a 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -246,29 +246,6 @@ nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, return 0; } -int -nouveau_bo_resize(struct nouveau_bo *bo, int size) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - int ret; - - if (!nvbo || nvbo->user) - return -EINVAL; - - if (nvbo->sysmem) { - nvbo->sysmem = realloc(nvbo->sysmem, size); - if (!nvbo->sysmem) - return -ENOMEM; - } else { - ret = nouveau_bo_realloc_gpu(nvbo, 0, size); - if (ret) - return ret; - } - - nvbo->base.size = size; - return 0; -} - void nouveau_bo_del(struct nouveau_bo **bo) { diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 3e886869d85..7ea4c654654 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -274,9 +274,6 @@ nouveau_bo_user(struct nouveau_device *, void *ptr, int size, extern int nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); -extern int -nouveau_bo_resize(struct nouveau_bo *, int size); - extern void nouveau_bo_del(struct nouveau_bo **); -- cgit v1.2.3 From fc36399f232942b3ff3975aac9e685d5f1363816 Mon Sep 17 00:00:00 2001 From: Michal Krol Date: Tue, 29 Jan 2008 16:41:10 +0100 Subject: gallium: Fix build on WinXP. --- src/mesa/pipe/draw/draw_clip.c | 2 +- src/mesa/pipe/draw/draw_vertex_fetch.c | 16 ++++++++-------- src/mesa/pipe/draw/draw_vertex_shader.c | 4 ++-- src/mesa/pipe/draw/draw_vf.c | 10 ++++++---- src/mesa/pipe/draw/draw_vf_generic.c | 2 -- src/mesa/pipe/pipebuffer/pb_buffer_fenced.c | 4 ++-- 6 files changed, 19 insertions(+), 19 deletions(-) diff --git a/src/mesa/pipe/draw/draw_clip.c b/src/mesa/pipe/draw/draw_clip.c index da20028904a..61130c5600a 100644 --- a/src/mesa/pipe/draw/draw_clip.c +++ b/src/mesa/pipe/draw/draw_clip.c @@ -406,7 +406,7 @@ clip_init_state( struct draw_stage *stage ) { struct clipper *clipper = clipper_stage( stage ); - clipper->flat = stage->draw->rasterizer->flatshade; + clipper->flat = stage->draw->rasterizer->flatshade ? TRUE : FALSE; if (clipper->flat) { const struct pipe_shader_state *vs = stage->draw->vertex_shader->state; diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index 89e4c256a7d..b23f487e741 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -326,6 +326,10 @@ static void fetch_xyz_rgb( struct draw_context *draw, const unsigned *elts, unsigned count ) { + const unsigned *pitch = draw->vertex_fetch.pitch; + const ubyte **src = draw->vertex_fetch.src_ptr; + int i; + assert(count <= 4); // _mesa_printf("%s\n", __FUNCTION__); @@ -333,10 +337,6 @@ static void fetch_xyz_rgb( struct draw_context *draw, /* loop over vertex attributes (vertex shader inputs) */ - const unsigned *pitch = draw->vertex_fetch.pitch; - const ubyte **src = draw->vertex_fetch.src_ptr; - int i; - for (i = 0; i < 4; i++) { { const float *in = (const float *)(src[0] + elts[i] * pitch[0]); @@ -366,15 +366,15 @@ static void fetch_xyz_rgb_st( struct draw_context *draw, const unsigned *elts, unsigned count ) { + const unsigned *pitch = draw->vertex_fetch.pitch; + const ubyte **src = draw->vertex_fetch.src_ptr; + int i; + assert(count <= 4); /* loop over vertex attributes (vertex shader inputs) */ - const unsigned *pitch = draw->vertex_fetch.pitch; - const ubyte **src = draw->vertex_fetch.src_ptr; - int i; - for (i = 0; i < 4; i++) { { const float *in = (const float *)(src[0] + elts[i] * pitch[0]); diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index 0806e23d6c4..b851da845f0 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -199,7 +199,7 @@ run_vertex_program(struct draw_context *draw, void draw_vertex_shader_queue_flush(struct draw_context *draw) { - unsigned i, j; + unsigned i; assert(draw->vs.queue_nr != 0); @@ -219,7 +219,7 @@ draw_vertex_shader_queue_flush(struct draw_context *draw) for (i = 0; i < draw->vs.queue_nr; i += 4) { struct vertex_header *dests[4]; unsigned elts[4]; - int n = MIN2(4, draw->vs.queue_nr - i); + int j, n = MIN2(4, draw->vs.queue_nr - i); for (j = 0; j < n; j++) { elts[j] = draw->vs.queue[i + j].elt; diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 06b84b93cc8..0debea1f128 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -32,7 +32,7 @@ #include "draw_vf.h" -#define DBG 0 +#define DRAW_VF_DBG 0 /* TODO: remove this */ @@ -166,9 +166,10 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, for (j = 0, i = 0; i < nr; i++) { const unsigned format = map[i].format; if (format == DRAW_EMIT_PAD) { - if (DBG) +#if (DRAW_VF_DBG) _mesa_printf("%d: pad %d, offset %d\n", i, map[i].offset, offset); +#endif offset += map[i].offset; @@ -186,10 +187,11 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, if(vf->attr[j].isconst) memcpy(vf->attr[j].data, &map[i].data, vf->attr[j].vertattrsize); - if (DBG) +#if (DRAW_VF_DBG) _mesa_printf("%d: %s, offset %d\n", i, draw_vf_format_info[format].name, vf->attr[j].vertoffset); +#endif offset += draw_vf_format_info[format].attrsize; j++; @@ -303,7 +305,7 @@ void draw_vf_destroy( struct draw_vertex_fetch *vf ) * to unify them, but this probably won't change until this * module gets another overhaul. */ - _mesa_exec_free((void *) fp->func); + //_mesa_exec_free((void *) fp->func); FREE(fp); } diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c index 0caa798396d..7f5f56ef9cc 100644 --- a/src/mesa/pipe/draw/draw_vf_generic.c +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -29,8 +29,6 @@ #include -#include "simple_list.h" - #include "pipe/p_compiler.h" #include "pipe/p_util.h" diff --git a/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c b/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c index 349647fe6e4..4cf4222db9d 100644 --- a/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c +++ b/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c @@ -145,7 +145,7 @@ _fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list, /* Do the delayed destroy: */ pb_reference(&fenced_buf->buffer, NULL); - free(fenced_buf); + FREE(fenced_buf); } } @@ -162,7 +162,7 @@ fenced_buffer_destroy(struct pb_buffer *buf) } else { pb_reference(&fenced_buf->buffer, NULL); - free(fenced_buf); + FREE(fenced_buf); } if ((fenced_list->numDelayed % fenced_list->checkDelayed) == 0) -- cgit v1.2.3 From cdb48e20d64b8dedcda2ee7f0636db223efef0fa Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Wed, 30 Jan 2008 15:24:56 +0900 Subject: gallium: Remove draw_vertex_fetch::lookup. It is not being used, and would be dangerous to use given the possibility of constants. --- src/mesa/pipe/draw/draw_vf.c | 5 ----- src/mesa/pipe/draw/draw_vf.h | 2 -- 2 files changed, 7 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 0debea1f128..64d9ed02a9d 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -161,8 +161,6 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, assert(nr < PIPE_ATTRIB_MAX); - memset(vf->lookup, 0, sizeof(vf->lookup)); - for (j = 0, i = 0; i < nr; i++) { const unsigned format = map[i].format; if (format == DRAW_EMIT_PAD) { @@ -175,9 +173,6 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, } else { - assert(vf->lookup[map[i].attrib] == 0 || format == DRAW_EMIT_1F_CONST); - vf->lookup[map[i].attrib] = &vf->attr[j]; - vf->attr[j].attrib = map[i].attrib; vf->attr[j].format = format; vf->attr[j].insert = draw_vf_format_info[format].insert; diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index 911ea07bdfe..09cf4d3a6ad 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -159,8 +159,6 @@ struct draw_vertex_fetch unsigned attr_count; unsigned vertex_stride; - struct draw_vf_attr *lookup[PIPE_ATTRIB_MAX]; - draw_vf_emit_func emit; /* Parameters and constants for codegen: -- cgit v1.2.3 From ee41d7afc922083de46cbdc491ee6052f3c4d45b Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Wed, 30 Jan 2008 16:46:41 +0900 Subject: gallium: Teach draw_vf about draw vertices. This reduces the emit overhead, which is significant since we're emiting one vertex at a time. --- src/mesa/pipe/draw/draw_vbuf.c | 147 ++--------------------------------- src/mesa/pipe/draw/draw_vf.c | 171 ++++++++++++++++++++++++++++++++++------- src/mesa/pipe/draw/draw_vf.h | 17 ++-- 3 files changed, 161 insertions(+), 174 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vbuf.c b/src/mesa/pipe/draw/draw_vbuf.c index 92a8b9fbcfa..ac03001d8f3 100644 --- a/src/mesa/pipe/draw/draw_vbuf.c +++ b/src/mesa/pipe/draw/draw_vbuf.c @@ -35,7 +35,6 @@ #include -#include #include "pipe/p_util.h" @@ -268,8 +267,7 @@ emit_vertex( struct vbuf_stage *vbuf, #if 0 { static float data[256]; - draw_vf_set_data(vbuf->vf, vertex->data); - draw_vf_emit_vertices(vbuf->vf, 1, data); + draw_vf_emit_vertex(vbuf->vf, vertex, data); if(memcmp((uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size, data, vbuf->vertex_size)) { fprintf(stderr, "With VF:\n"); dump_emitted_vertex(vbuf->vinfo, (uint8_t *)data); @@ -281,149 +279,13 @@ emit_vertex( struct vbuf_stage *vbuf, #endif } else { - draw_vf_set_data(vbuf->vf, vertex->data); - draw_vf_emit_vertices(vbuf->vf, 1, vbuf->vertex_ptr); + draw_vf_emit_vertex(vbuf->vf, vertex, vbuf->vertex_ptr); vbuf->vertex_ptr += vbuf->vertex_size/4; } } -static void -vbuf_set_vf_attributes(struct vbuf_stage *vbuf ) -{ - const struct vertex_info *vinfo = vbuf->vinfo; - struct draw_vf_attr_map attrs[PIPE_MAX_SHADER_INPUTS]; - uint i; - uint count = 0; /* for debug/sanity */ - unsigned nr_attrs = 0; - - if(!vbuf->vf) - return; - -// fprintf(stderr, "emit vertex %d to %p\n", -// vbuf->nr_vertices, vbuf->vertex_ptr); - -#if 0 - if(vertex->vertex_id != UNDEFINED_VERTEX_ID) { - if(vertex->vertex_id < vbuf->nr_vertices) - return; - else - fprintf(stderr, "Bad vertex id 0x%04x (>= 0x%04x)\n", - vertex->vertex_id, vbuf->nr_vertices); - return; - } -#endif - - for (i = 0; i < vinfo->num_attribs; i++) { - uint j = vinfo->src_index[i]; - switch (vinfo->emit[i]) { - case EMIT_OMIT: - /* no-op */ - break; - case EMIT_ALL: { - /* just copy the whole vertex as-is to the vbuf */ - unsigned k, s = vinfo->size; - assert(i == 0); - assert(j == 0); - /* copy the vertex header */ - /* XXX: we actually don't copy the header, just pad it */ - attrs[nr_attrs].attrib = 0; - attrs[nr_attrs].format = DRAW_EMIT_PAD; - attrs[nr_attrs].offset = offsetof(struct vertex_header, data); - s -= offsetof(struct vertex_header, data)/4; - count += offsetof(struct vertex_header, data)/4; - nr_attrs++; - /* copy the vertex data */ - for(k = 0; k < (s & ~0x3); k += 4) { - attrs[nr_attrs].attrib = k/4; - attrs[nr_attrs].format = DRAW_EMIT_4F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 4; - } - /* tail */ - /* XXX: actually, this shouldn't be needed */ - attrs[nr_attrs].attrib = k/4; - attrs[nr_attrs].offset = 0; - switch(s & 0x3) { - case 0: - break; - case 1: - attrs[nr_attrs].format = DRAW_EMIT_1F; - nr_attrs++; - count += 1; - break; - case 2: - attrs[nr_attrs].format = DRAW_EMIT_2F; - nr_attrs++; - count += 2; - break; - case 3: - attrs[nr_attrs].format = DRAW_EMIT_3F; - nr_attrs++; - count += 3; - break; - } - break; - } - case EMIT_1F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_1F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count++; - break; - case EMIT_1F_PSIZE: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_1F_CONST; - attrs[nr_attrs].offset = 0; - attrs[nr_attrs].data.f[0] = vbuf->stage.draw->rasterizer->point_size; - nr_attrs++; - count++; - break; - case EMIT_2F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_2F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 2; - break; - case EMIT_3F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_3F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 3; - break; - case EMIT_4F: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_4F; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 4; - break; - case EMIT_4UB: - attrs[nr_attrs].attrib = j; - attrs[nr_attrs].format = DRAW_EMIT_4UB_4F_BGRA; - attrs[nr_attrs].offset = 0; - nr_attrs++; - count += 1; - break; - default: - assert(0); - } - } - - assert(count == vinfo->size); - - draw_vf_set_vertex_attributes(vbuf->vf, - attrs, - nr_attrs, - vbuf->vertex_size); -} - - static void vbuf_tri( struct draw_stage *stage, struct prim_header *prim ) @@ -498,7 +360,10 @@ vbuf_set_prim( struct vbuf_stage *vbuf, uint newprim ) vbuf->vinfo = vinfo; vbuf->vertex_size = vertex_size; - vbuf_set_vf_attributes(vbuf); + if(vbuf->vf) + draw_vf_set_vertex_info(vbuf->vf, + vbuf->vinfo, + vbuf->stage.draw->rasterizer->point_size); if (!vbuf->vertices) vbuf_alloc_vertices(vbuf); diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 64d9ed02a9d..0da8e59ad68 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -26,6 +26,8 @@ */ +#include + #include "pipe/p_compiler.h" #include "pipe/p_util.h" @@ -151,10 +153,11 @@ static void choose_emit_func( struct draw_vertex_fetch *vf, -unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, - const struct draw_vf_attr_map *map, - unsigned nr, - unsigned vertex_stride ) +static unsigned +draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, + const struct draw_vf_attr_map *map, + unsigned nr, + unsigned vertex_stride ) { unsigned offset = 0; unsigned i, j; @@ -202,6 +205,133 @@ unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, } +void draw_vf_set_vertex_info( struct draw_vertex_fetch *vf, + const struct vertex_info *vinfo, + float point_size ) +{ + unsigned i, j, k; + struct draw_vf_attr *a = vf->attr; + struct draw_vf_attr_map attrs[PIPE_MAX_SHADER_INPUTS]; + unsigned count = 0; /* for debug/sanity */ + unsigned nr_attrs = 0; + + for (i = 0; i < vinfo->num_attribs; i++) { + j = vinfo->src_index[i]; + switch (vinfo->emit[i]) { + case EMIT_OMIT: + /* no-op */ + break; + case EMIT_ALL: { + /* just copy the whole vertex as-is to the vbuf */ + unsigned s = vinfo->size; + assert(i == 0); + assert(j == 0); + /* copy the vertex header */ + /* XXX: we actually don't copy the header, just pad it */ + attrs[nr_attrs].attrib = 0; + attrs[nr_attrs].format = DRAW_EMIT_PAD; + attrs[nr_attrs].offset = offsetof(struct vertex_header, data); + s -= offsetof(struct vertex_header, data)/4; + count += offsetof(struct vertex_header, data)/4; + nr_attrs++; + /* copy the vertex data */ + for(k = 0; k < (s & ~0x3); k += 4) { + attrs[nr_attrs].attrib = k/4; + attrs[nr_attrs].format = DRAW_EMIT_4F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 4; + } + /* tail */ + /* XXX: actually, this shouldn't be needed */ + attrs[nr_attrs].attrib = k/4; + attrs[nr_attrs].offset = 0; + switch(s & 0x3) { + case 0: + break; + case 1: + attrs[nr_attrs].format = DRAW_EMIT_1F; + nr_attrs++; + count += 1; + break; + case 2: + attrs[nr_attrs].format = DRAW_EMIT_2F; + nr_attrs++; + count += 2; + break; + case 3: + attrs[nr_attrs].format = DRAW_EMIT_3F; + nr_attrs++; + count += 3; + break; + } + break; + } + case EMIT_1F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_1F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count++; + break; + case EMIT_1F_PSIZE: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_1F_CONST; + attrs[nr_attrs].offset = 0; + attrs[nr_attrs].data.f[0] = point_size; + nr_attrs++; + count++; + break; + case EMIT_2F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_2F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 2; + break; + case EMIT_3F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_3F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 3; + break; + case EMIT_4F: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_4F; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 4; + break; + case EMIT_4UB: + attrs[nr_attrs].attrib = j; + attrs[nr_attrs].format = DRAW_EMIT_4UB_4F_BGRA; + attrs[nr_attrs].offset = 0; + nr_attrs++; + count += 1; + break; + default: + assert(0); + } + } + + assert(count == vinfo->size); + + draw_vf_set_vertex_attributes(vf, + attrs, + nr_attrs, + vinfo->size * sizeof(float) ); + + for (j = 0; j < vf->attr_count; j++) { + a[j].inputsize = 4; + a[j].do_insert = a[j].insert[4 - 1]; + if(a[j].isconst) { + a[j].inputptr = a[j].data; + a[j].inputstride = 0; + } + } +} + #if 0 /* Set attribute pointers, adjusted for start position: @@ -229,38 +359,27 @@ void draw_vf_set_sources( struct draw_vertex_fetch *vf, #endif -/* Set attribute pointers, adjusted for start position: +/** + * Emit a vertex to dest. */ -void draw_vf_set_data( struct draw_vertex_fetch *vf, - float data[][4]) +void draw_vf_emit_vertex( struct draw_vertex_fetch *vf, + struct vertex_header *vertex, + void *dest ) { struct draw_vf_attr *a = vf->attr; unsigned j; for (j = 0; j < vf->attr_count; j++) { - a[j].inputstride = 0; /* XXX: one-vertex-max ATM */ - a[j].inputsize = 4; - a[j].do_insert = a[j].insert[4 - 1]; - if(a[j].isconst) - a[j].inputptr = a[j].data; - else - a[j].inputptr = (uint8_t *)&data[a[j].attrib][0]; + if(!a[j].isconst) { + a[j].inputptr = (uint8_t *)&vertex->data[a[j].attrib][0]; + a[j].inputstride = 0; /* XXX: one-vertex-max ATM */ + } } + + vf->emit( vf, 1, (uint8_t*) dest ); } -/* Emit count VB vertices to dest. - */ -void draw_vf_emit_vertices( struct draw_vertex_fetch *vf, - unsigned count, - void *dest ) -{ - vf->emit( vf, count, (uint8_t*) dest ); -} - - - - struct draw_vertex_fetch *draw_vf_create( void ) { diff --git a/src/mesa/pipe/draw/draw_vf.h b/src/mesa/pipe/draw/draw_vf.h index 09cf4d3a6ad..e694b986754 100644 --- a/src/mesa/pipe/draw/draw_vf.h +++ b/src/mesa/pipe/draw/draw_vf.h @@ -33,6 +33,7 @@ #include "pipe/p_state.h" #include "draw_vertex.h" +#include "draw_private.h" // for vertex_header enum draw_vf_attr_format { @@ -78,11 +79,17 @@ struct draw_vertex_fetch; +#if 0 unsigned draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, const struct draw_vf_attr_map *map, unsigned nr, unsigned vertex_stride ); +#endif + +void draw_vf_set_vertex_info( struct draw_vertex_fetch *vf, + const struct vertex_info *vinfo, + float point_size ); #if 0 void @@ -92,13 +99,9 @@ draw_vf_set_sources( struct draw_vertex_fetch *vf, #endif void -draw_vf_set_data( struct draw_vertex_fetch *vf, - float data[][4]); - -void -draw_vf_emit_vertices( struct draw_vertex_fetch *vf, - unsigned count, - void *dest ); +draw_vf_emit_vertex( struct draw_vertex_fetch *vf, + struct vertex_header *vertex, + void *dest ); struct draw_vertex_fetch * draw_vf_create( void ); -- cgit v1.2.3 From 92d71f073006e05ef68e02dff92ae7ff40cfc470 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 11:49:26 -0700 Subject: Cell: move CELL_MAX_SPUS --- src/mesa/pipe/cell/common.h | 2 ++ src/mesa/pipe/cell/ppu/cell_context.h | 3 --- 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 90aa46a5343..d5e86863d48 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -61,6 +61,8 @@ #define ROUNDUP16(k) (((k) + 0xf) & ~0xf) +#define CELL_MAX_SPUS 6 + #define TILE_SIZE 32 diff --git a/src/mesa/pipe/cell/ppu/cell_context.h b/src/mesa/pipe/cell/ppu/cell_context.h index 7d234f3e452..65b89518ada 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.h +++ b/src/mesa/pipe/cell/ppu/cell_context.h @@ -38,9 +38,6 @@ #include "pipe/cell/common.h" -#define CELL_MAX_SPUS 6 - - struct cell_vbuf_render; struct cell_vertex_shader_state -- cgit v1.2.3 From ae6949659693385be2ccd4290338b58038ed8125 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 11:49:51 -0700 Subject: Cell: make wait_on_mask() static/inlined --- src/mesa/pipe/cell/spu/spu_main.c | 19 ------------------- src/mesa/pipe/cell/spu/spu_main.h | 23 +++++++++++++++++++++-- 2 files changed, 21 insertions(+), 21 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 6e02f2c9642..6886f283bec 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -31,7 +31,6 @@ #include #include -#include #include "spu_main.h" #include "spu_render.h" @@ -52,24 +51,6 @@ boolean Debug = FALSE; struct spu_global spu; -void -wait_on_mask(unsigned tagMask) -{ - mfc_write_tag_mask( tagMask ); - /* wait for completion of _any_ DMAs specified by tagMask */ - mfc_read_tag_status_any(); -} - - -static INLINE void -wait_on_mask_all(unsigned tagMask) -{ - mfc_write_tag_mask( tagMask ); - /* wait for completion of _any_ DMAs specified by tagMask */ - mfc_read_tag_status_all(); -} - - /** * Tell the PPU that this SPU has finished copying a buffer to * local store and that it may be reused by the PPU. diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 009e046ba5e..8908bf8bc07 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -29,6 +29,8 @@ #define SPU_MAIN_H +#include + #include "pipe/cell/common.h" #include "pipe/draw/draw_vertex.h" #include "pipe/p_state.h" @@ -90,8 +92,25 @@ extern boolean Debug; -extern void -wait_on_mask(unsigned tag); +static INLINE void +wait_on_mask(unsigned tagMask) +{ + mfc_write_tag_mask( tagMask ); + /* wait for completion of _any_ DMAs specified by tagMask */ + mfc_read_tag_status_any(); +} + + +static INLINE void +wait_on_mask_all(unsigned tagMask) +{ + mfc_write_tag_mask( tagMask ); + /* wait for completion of _any_ DMAs specified by tagMask */ + mfc_read_tag_status_all(); +} + + + static INLINE void -- cgit v1.2.3 From 69099004e62b8710bc0b360fd2938439b34c0079 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 11:56:14 -0700 Subject: Cell: check tile status before wait_on_mask() --- src/mesa/pipe/cell/spu/spu_tri.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 7c6a54134f8..01a47a48519 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -278,7 +278,7 @@ do_depth_test(struct setup_stage *setup, int x, int y, unsigned mask) /* now, _really_ clear the tile */ clear_z_tile(&ztile); } - else { + else if (tile_status_z[setup->ty][setup->tx] != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_Z); } @@ -403,7 +403,7 @@ emit_quad( struct setup_stage *setup, int x, int y, unsigned mask ) /* now, _really_ clear the tile */ clear_c_tile(&ctile); } - else { + else if (tile_status[setup->ty][setup->tx] != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_COLOR); } -- cgit v1.2.3 From ab5e8b33cb615b9267ec4d08173c3c83cfd6df3f Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 11:56:41 -0700 Subject: Cell: minor code refactoring, movement --- src/mesa/pipe/cell/spu/spu_render.c | 85 ++++++++++++++++++++++++------------- 1 file changed, 55 insertions(+), 30 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_render.c b/src/mesa/pipe/cell/spu/spu_render.c index 21a286a23d3..f506095116b 100644 --- a/src/mesa/pipe/cell/spu/spu_render.c +++ b/src/mesa/pipe/cell/spu/spu_render.c @@ -88,6 +88,55 @@ my_tile(uint tx, uint ty) } +/** + * Start fetching non-clear color/Z tiles from main memory + */ +static INLINE void +get_cz_tiles(uint tx, uint ty) +{ + if (spu.depth_stencil.depth.enabled) { + if (tile_status_z[ty][tx] != TILE_STATUS_CLEAR) { + get_tile(tx, ty, &ztile, TAG_READ_TILE_Z, 1); + } + } + + if (tile_status[ty][tx] != TILE_STATUS_CLEAR) { + get_tile(tx, ty, &ctile, TAG_READ_TILE_COLOR, 0); + } +} + + +/** + * Start putting dirty color/Z tiles back to main memory + */ +static INLINE void +put_cz_tiles(uint tx, uint ty) +{ + if (tile_status_z[ty][tx] == TILE_STATUS_DIRTY) { + put_tile(tx, ty, &ztile, TAG_WRITE_TILE_Z, 1); + tile_status_z[ty][tx] = TILE_STATUS_DEFINED; + } + + if (tile_status[ty][tx] == TILE_STATUS_DIRTY) { + put_tile(tx, ty, &ctile, TAG_WRITE_TILE_COLOR, 0); + tile_status[ty][tx] = TILE_STATUS_DEFINED; + } +} + + +/** + * Wait for 'put' of color/z tiles to complete. + */ +static INLINE void +wait_put_cz_tiles(void) +{ + wait_on_mask(1 << TAG_WRITE_TILE_COLOR); + if (spu.depth_stencil.depth.enabled) { + wait_on_mask(1 << TAG_WRITE_TILE_Z); + } +} + + /** * Render primitives * \param pos_incr returns value indicating how may words to skip after @@ -122,6 +171,9 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) ASSERT(sizeof(*render) % 4 == 0); ASSERT(total_vertex_bytes % 16 == 0); + ASSERT(render->prim_type == PIPE_PRIM_TRIANGLES); + ASSERT(render->num_indexes % 3 == 0); + /* indexes are right after the render command in the batch buffer */ indexes = (const ushort *) (render + 1); @@ -186,21 +238,7 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) if (!my_tile(tx, ty)) continue; - /* Start fetching color/z tiles. We'll wait for completion when - * we need read/write to them later in triangle rasterization. - */ - if (spu.depth_stencil.depth.enabled) { - if (tile_status_z[ty][tx] != TILE_STATUS_CLEAR) { - get_tile(tx, ty, &ztile, TAG_READ_TILE_Z, 1); - } - } - - if (tile_status[ty][tx] != TILE_STATUS_CLEAR) { - get_tile(tx, ty, &ctile, TAG_READ_TILE_COLOR, 0); - } - - ASSERT(render->prim_type == PIPE_PRIM_TRIANGLES); - ASSERT(render->num_indexes % 3 == 0); + get_cz_tiles(tx, ty); /* loop over tris */ for (j = 0; j < render->num_indexes; j += 3) { @@ -214,22 +252,9 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) } /* write color/z tiles back to main framebuffer, if dirtied */ - if (tile_status[ty][tx] == TILE_STATUS_DIRTY) { - put_tile(tx, ty, &ctile, TAG_WRITE_TILE_COLOR, 0); - tile_status[ty][tx] = TILE_STATUS_DEFINED; - } - if (spu.depth_stencil.depth.enabled) { - if (tile_status_z[ty][tx] == TILE_STATUS_DIRTY) { - put_tile(tx, ty, &ztile, TAG_WRITE_TILE_Z, 1); - tile_status_z[ty][tx] = TILE_STATUS_DEFINED; - } - } + put_cz_tiles(tx, ty); - /* XXX move these... */ - wait_on_mask(1 << TAG_WRITE_TILE_COLOR); - if (spu.depth_stencil.depth.enabled) { - wait_on_mask(1 << TAG_WRITE_TILE_Z); - } + wait_put_cz_tiles(); /* XXX seems unnecessary... */ } if (Debug) -- cgit v1.2.3 From ecb0013e2f4157caeb1e60c01ba06d6c8957e609 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 12:08:23 -0700 Subject: Cell: make 'setup' a regular var instead of passing around a pointer everywhere We'll never have more than one of these objects. Avoiding pointer deref improves performance a bit. --- src/mesa/pipe/cell/spu/spu_tri.c | 419 +++++++++++++++++++-------------------- 1 file changed, 209 insertions(+), 210 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 01a47a48519..5bb2cb12e37 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -135,6 +135,12 @@ struct setup_stage { }; + +static struct setup_stage setup; + + + + #if 0 /** * Basically a cast wrapper. @@ -147,33 +153,33 @@ static INLINE struct setup_stage *setup_stage( struct draw_stage *stage ) #if 0 /** - * Clip setup->quad against the scissor/surface bounds. + * Clip setup.quad against the scissor/surface bounds. */ static INLINE void quad_clip(struct setup_stage *setup) { - const struct pipe_scissor_state *cliprect = &setup->softpipe->cliprect; + const struct pipe_scissor_state *cliprect = &setup.softpipe->cliprect; const int minx = (int) cliprect->minx; const int maxx = (int) cliprect->maxx; const int miny = (int) cliprect->miny; const int maxy = (int) cliprect->maxy; - if (setup->quad.x0 >= maxx || - setup->quad.y0 >= maxy || - setup->quad.x0 + 1 < minx || - setup->quad.y0 + 1 < miny) { + if (setup.quad.x0 >= maxx || + setup.quad.y0 >= maxy || + setup.quad.x0 + 1 < minx || + setup.quad.y0 + 1 < miny) { /* totally clipped */ - setup->quad.mask = 0x0; + setup.quad.mask = 0x0; return; } - if (setup->quad.x0 < minx) - setup->quad.mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT); - if (setup->quad.y0 < miny) - setup->quad.mask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT); - if (setup->quad.x0 == maxx - 1) - setup->quad.mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); - if (setup->quad.y0 == maxy - 1) - setup->quad.mask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT); + if (setup.quad.x0 < minx) + setup.quad.mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT); + if (setup.quad.y0 < miny) + setup.quad.mask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT); + if (setup.quad.x0 == maxx - 1) + setup.quad.mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); + if (setup.quad.y0 == maxy - 1) + setup.quad.mask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT); } #endif @@ -185,9 +191,9 @@ static INLINE void clip_emit_quad(struct setup_stage *setup) { quad_clip(setup); - if (setup->quad.mask) { - struct softpipe_context *sp = setup->softpipe; - sp->quad.first->run(sp->quad.first, &setup->quad); + if (setup.quad.mask) { + struct softpipe_context *sp = setup.softpipe; + sp->quad.first->run(sp->quad.first, &setup.quad); } } #endif @@ -198,8 +204,7 @@ clip_emit_quad(struct setup_stage *setup) * Eg: four colors will be compute. */ static INLINE void -eval_coeff( struct setup_stage *setup, uint slot, - float x, float y, float result[4][4]) +eval_coeff(uint slot, float x, float y, float result[4][4]) { switch (spu.vertex_info.interp_mode[slot]) { case INTERP_CONSTANT: @@ -209,7 +214,7 @@ eval_coeff( struct setup_stage *setup, uint slot, result[QUAD_TOP_LEFT][i] = result[QUAD_TOP_RIGHT][i] = result[QUAD_BOTTOM_LEFT][i] = - result[QUAD_BOTTOM_RIGHT][i] = setup->coef[slot].a0[i]; + result[QUAD_BOTTOM_RIGHT][i] = setup.coef[slot].a0[i]; } } break; @@ -219,12 +224,12 @@ eval_coeff( struct setup_stage *setup, uint slot, default: { uint i; - const float *dadx = setup->coef[slot].dadx; - const float *dady = setup->coef[slot].dady; + const float *dadx = setup.coef[slot].dadx; + const float *dady = setup.coef[slot].dady; /* loop over XYZW comps */ for (i = 0; i < 4; i++) { - result[QUAD_TOP_LEFT][i] = setup->coef[slot].a0[i] + x * dadx[i] + y * dady[i]; + result[QUAD_TOP_LEFT][i] = setup.coef[slot].a0[i] + x * dadx[i] + y * dady[i]; result[QUAD_TOP_RIGHT][i] = result[0][i] + dadx[i]; result[QUAD_BOTTOM_LEFT][i] = result[0][i] + dady[i]; result[QUAD_BOTTOM_RIGHT][i] = result[0][i] + dadx[i] + dady[i]; @@ -235,15 +240,14 @@ eval_coeff( struct setup_stage *setup, uint slot, static INLINE void -eval_z( struct setup_stage *setup, - float x, float y, float result[4]) +eval_z(float x, float y, float result[4]) { const uint slot = 0; const uint i = 2; - const float *dadx = setup->coef[slot].dadx; - const float *dady = setup->coef[slot].dady; + const float *dadx = setup.coef[slot].dadx; + const float *dady = setup.coef[slot].dady; - result[QUAD_TOP_LEFT] = setup->coef[slot].a0[i] + x * dadx[i] + y * dady[i]; + result[QUAD_TOP_LEFT] = setup.coef[slot].a0[i] + x * dadx[i] + y * dady[i]; result[QUAD_TOP_RIGHT] = result[0] + dadx[i]; result[QUAD_BOTTOM_LEFT] = result[0] + dady[i]; result[QUAD_BOTTOM_RIGHT] = result[0] + dadx[i] + dady[i]; @@ -266,23 +270,23 @@ pack_color(const float color[4]) static uint -do_depth_test(struct setup_stage *setup, int x, int y, unsigned mask) +do_depth_test(int x, int y, unsigned mask) { - int ix = x - setup->cliprect_minx; - int iy = y - setup->cliprect_miny; + int ix = x - setup.cliprect_minx; + int iy = y - setup.cliprect_miny; float zvals[4]; - eval_z(setup, (float) x, (float) y, zvals); + eval_z((float) x, (float) y, zvals); - if (tile_status_z[setup->ty][setup->tx] == TILE_STATUS_CLEAR) { + if (tile_status_z[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_z_tile(&ztile); } - else if (tile_status_z[setup->ty][setup->tx] != TILE_STATUS_DIRTY) { + else if (tile_status_z[setup.ty][setup.tx] != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_Z); } - tile_status_z[setup->ty][setup->tx] = TILE_STATUS_DIRTY; + tile_status_z[setup.ty][setup.tx] = TILE_STATUS_DIRTY; if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { @@ -363,31 +367,31 @@ do_depth_test(struct setup_stage *setup, int x, int y, unsigned mask) * Emit a quad (pass to next stage). No clipping is done. */ static INLINE void -emit_quad( struct setup_stage *setup, int x, int y, unsigned mask ) +emit_quad( int x, int y, unsigned mask ) { #if 0 - struct softpipe_context *sp = setup->softpipe; - setup->quad.x0 = x; - setup->quad.y0 = y; - setup->quad.mask = mask; - sp->quad.first->run(sp->quad.first, &setup->quad); + struct softpipe_context *sp = setup.softpipe; + setup.quad.x0 = x; + setup.quad.y0 = y; + setup.quad.mask = mask; + sp->quad.first->run(sp->quad.first, &setup.quad); #else /* Cell: "write" quad fragments to the tile by setting prim color */ - const int ix = x - setup->cliprect_minx; - const int iy = y - setup->cliprect_miny; + const int ix = x - setup.cliprect_minx; + const int iy = y - setup.cliprect_miny; uint colors[4]; /* indexed by QUAD_x */ if (spu.texture.start) { float texcoords[4][4]; uint i; - eval_coeff(setup, 2, (float) x, (float) y, texcoords); + eval_coeff(2, (float) x, (float) y, texcoords); for (i = 0; i < 4; i++) { colors[i] = sample_texture(texcoords[i]); } } else { float fcolors[4][4]; - eval_coeff(setup, 1, (float) x, (float) y, fcolors); + eval_coeff(1, (float) x, (float) y, fcolors); colors[QUAD_TOP_LEFT] = pack_color(fcolors[QUAD_TOP_LEFT]); colors[QUAD_TOP_RIGHT] = pack_color(fcolors[QUAD_TOP_RIGHT]); colors[QUAD_BOTTOM_LEFT] = pack_color(fcolors[QUAD_BOTTOM_LEFT]); @@ -395,19 +399,19 @@ emit_quad( struct setup_stage *setup, int x, int y, unsigned mask ) } if (spu.depth_stencil.depth.enabled) { - mask &= do_depth_test(setup, x, y, mask); + mask &= do_depth_test(x, y, mask); } if (mask) { - if (tile_status[setup->ty][setup->tx] == TILE_STATUS_CLEAR) { + if (tile_status[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_c_tile(&ctile); } - else if (tile_status[setup->ty][setup->tx] != TILE_STATUS_DIRTY) { + else if (tile_status[setup.ty][setup.tx] != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_COLOR); } - tile_status[setup->ty][setup->tx] = TILE_STATUS_DIRTY; + tile_status[setup.ty][setup.tx] = TILE_STATUS_DIRTY; if (mask & MASK_TOP_LEFT) ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; @@ -439,20 +443,20 @@ static INLINE int block( int x ) * this is pretty nasty... may need to rework flush_spans again to * fix it, if possible. */ -static unsigned calculate_mask( struct setup_stage *setup, int x ) +static unsigned calculate_mask( int x ) { unsigned mask = 0x0; - if (x >= setup->span.left[0] && x < setup->span.right[0]) + if (x >= setup.span.left[0] && x < setup.span.right[0]) mask |= MASK_TOP_LEFT; - if (x >= setup->span.left[1] && x < setup->span.right[1]) + if (x >= setup.span.left[1] && x < setup.span.right[1]) mask |= MASK_BOTTOM_LEFT; - if (x+1 >= setup->span.left[0] && x+1 < setup->span.right[0]) + if (x+1 >= setup.span.left[0] && x+1 < setup.span.right[0]) mask |= MASK_TOP_RIGHT; - if (x+1 >= setup->span.left[1] && x+1 < setup->span.right[1]) + if (x+1 >= setup.span.left[1] && x+1 < setup.span.right[1]) mask |= MASK_BOTTOM_RIGHT; return mask; @@ -462,28 +466,28 @@ static unsigned calculate_mask( struct setup_stage *setup, int x ) /** * Render a horizontal span of quads */ -static void flush_spans( struct setup_stage *setup ) +static void flush_spans( void ) { int minleft, maxright; int x; - switch (setup->span.y_flags) { + switch (setup.span.y_flags) { case 0x3: /* both odd and even lines written (both quad rows) */ - minleft = MIN2(setup->span.left[0], setup->span.left[1]); - maxright = MAX2(setup->span.right[0], setup->span.right[1]); + minleft = MIN2(setup.span.left[0], setup.span.left[1]); + maxright = MAX2(setup.span.right[0], setup.span.right[1]); break; case 0x1: /* only even line written (quad top row) */ - minleft = setup->span.left[0]; - maxright = setup->span.right[0]; + minleft = setup.span.left[0]; + maxright = setup.span.right[0]; break; case 0x2: /* only odd line written (quad bottom row) */ - minleft = setup->span.left[1]; - maxright = setup->span.right[1]; + minleft = setup.span.left[1]; + maxright = setup.span.right[1]; break; default: @@ -494,31 +498,29 @@ static void flush_spans( struct setup_stage *setup ) * calculate_mask() could be simplified a bit... */ for (x = block(minleft); x <= block(maxright); x += 2) { - emit_quad( setup, x, setup->span.y, - calculate_mask( setup, x ) ); + emit_quad( x, setup.span.y, + calculate_mask( x ) ); } - setup->span.y = 0; - setup->span.y_flags = 0; - setup->span.right[0] = 0; - setup->span.right[1] = 0; + setup.span.y = 0; + setup.span.y_flags = 0; + setup.span.right[0] = 0; + setup.span.right[1] = 0; } #if DEBUG_VERTS -static void print_vertex(const struct setup_stage *setup, - const struct vertex_header *v) +static void print_vertex(const struct vertex_header *v) { int i; fprintf(stderr, "Vertex: (%p)\n", v); - for (i = 0; i < setup->quad.nr_attrs; i++) { + for (i = 0; i < setup.quad.nr_attrs; i++) { fprintf(stderr, " %d: %f %f %f %f\n", i, v->data[i][0], v->data[i][1], v->data[i][2], v->data[i][3]); } } #endif -static boolean setup_sort_vertices( struct setup_stage *setup, - const struct prim_header *prim ) +static boolean setup_sort_vertices(const struct prim_header *prim ) { const struct vertex_header *v0 = prim->v[0]; const struct vertex_header *v1 = prim->v[1]; @@ -526,12 +528,12 @@ static boolean setup_sort_vertices( struct setup_stage *setup, #if DEBUG_VERTS fprintf(stderr, "Triangle:\n"); - print_vertex(setup, v0); - print_vertex(setup, v1); - print_vertex(setup, v2); + print_vertex(v0); + print_vertex(v1); + print_vertex(v2); #endif - setup->vprovoke = v2; + setup.vprovoke = v2; /* determine bottom to top order of vertices */ { @@ -541,65 +543,65 @@ static boolean setup_sort_vertices( struct setup_stage *setup, if (y0 <= y1) { if (y1 <= y2) { /* y0<=y1<=y2 */ - setup->vmin = v0; - setup->vmid = v1; - setup->vmax = v2; + setup.vmin = v0; + setup.vmid = v1; + setup.vmax = v2; } else if (y2 <= y0) { /* y2<=y0<=y1 */ - setup->vmin = v2; - setup->vmid = v0; - setup->vmax = v1; + setup.vmin = v2; + setup.vmid = v0; + setup.vmax = v1; } else { /* y0<=y2<=y1 */ - setup->vmin = v0; - setup->vmid = v2; - setup->vmax = v1; + setup.vmin = v0; + setup.vmid = v2; + setup.vmax = v1; } } else { if (y0 <= y2) { /* y1<=y0<=y2 */ - setup->vmin = v1; - setup->vmid = v0; - setup->vmax = v2; + setup.vmin = v1; + setup.vmid = v0; + setup.vmax = v2; } else if (y2 <= y1) { /* y2<=y1<=y0 */ - setup->vmin = v2; - setup->vmid = v1; - setup->vmax = v0; + setup.vmin = v2; + setup.vmid = v1; + setup.vmax = v0; } else { /* y1<=y2<=y0 */ - setup->vmin = v1; - setup->vmid = v2; - setup->vmax = v0; + setup.vmin = v1; + setup.vmid = v2; + setup.vmax = v0; } } } /* Check if triangle is completely outside the tile bounds */ - if (setup->vmin->data[0][1] > setup->cliprect_maxy) + if (setup.vmin->data[0][1] > setup.cliprect_maxy) return FALSE; - if (setup->vmax->data[0][1] < setup->cliprect_miny) + if (setup.vmax->data[0][1] < setup.cliprect_miny) return FALSE; - if (setup->vmin->data[0][0] < setup->cliprect_minx && - setup->vmid->data[0][0] < setup->cliprect_minx && - setup->vmax->data[0][0] < setup->cliprect_minx) + if (setup.vmin->data[0][0] < setup.cliprect_minx && + setup.vmid->data[0][0] < setup.cliprect_minx && + setup.vmax->data[0][0] < setup.cliprect_minx) return FALSE; - if (setup->vmin->data[0][0] > setup->cliprect_maxx && - setup->vmid->data[0][0] > setup->cliprect_maxx && - setup->vmax->data[0][0] > setup->cliprect_maxx) + if (setup.vmin->data[0][0] > setup.cliprect_maxx && + setup.vmid->data[0][0] > setup.cliprect_maxx && + setup.vmax->data[0][0] > setup.cliprect_maxx) return FALSE; - setup->ebot.dx = setup->vmid->data[0][0] - setup->vmin->data[0][0]; - setup->ebot.dy = setup->vmid->data[0][1] - setup->vmin->data[0][1]; - setup->emaj.dx = setup->vmax->data[0][0] - setup->vmin->data[0][0]; - setup->emaj.dy = setup->vmax->data[0][1] - setup->vmin->data[0][1]; - setup->etop.dx = setup->vmax->data[0][0] - setup->vmid->data[0][0]; - setup->etop.dy = setup->vmax->data[0][1] - setup->vmid->data[0][1]; + setup.ebot.dx = setup.vmid->data[0][0] - setup.vmin->data[0][0]; + setup.ebot.dy = setup.vmid->data[0][1] - setup.vmin->data[0][1]; + setup.emaj.dx = setup.vmax->data[0][0] - setup.vmin->data[0][0]; + setup.emaj.dy = setup.vmax->data[0][1] - setup.vmin->data[0][1]; + setup.etop.dx = setup.vmax->data[0][0] - setup.vmid->data[0][0]; + setup.etop.dy = setup.vmax->data[0][1] - setup.vmid->data[0][1]; /* * Compute triangle's area. Use 1/area to compute partial @@ -612,13 +614,13 @@ static boolean setup_sort_vertices( struct setup_stage *setup, * use the prim->det value because its sign is correct. */ { - const float area = (setup->emaj.dx * setup->ebot.dy - - setup->ebot.dx * setup->emaj.dy); + const float area = (setup.emaj.dx * setup.ebot.dy - + setup.ebot.dx * setup.emaj.dy); - setup->oneoverarea = 1.0f / area; + setup.oneoverarea = 1.0f / area; /* _mesa_printf("%s one-over-area %f area %f det %f\n", - __FUNCTION__, setup->oneoverarea, area, prim->det ); + __FUNCTION__, setup.oneoverarea, area, prim->det ); */ } @@ -627,7 +629,7 @@ static boolean setup_sort_vertices( struct setup_stage *setup, * - the GLSL gl_FrontFacing fragment attribute (bool) * - two-sided stencil test */ - setup->quad.facing = (prim->det > 0.0) ^ (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW); + setup.quad.facing = (prim->det > 0.0) ^ (setup.softpipe->rasterizer->front_winding == PIPE_WINDING_CW); #endif return TRUE; @@ -637,22 +639,22 @@ static boolean setup_sort_vertices( struct setup_stage *setup, /** * Compute a0 for a constant-valued coefficient (GL_FLAT shading). * The value value comes from vertex->data[slot][i]. - * The result will be put into setup->coef[slot].a0[i]. + * The result will be put into setup.coef[slot].a0[i]. * \param slot which attribute slot * \param i which component of the slot (0..3) */ -static void const_coeff(struct setup_stage *setup, uint slot) +static void const_coeff(uint slot) { uint i; ASSERT(slot < PIPE_MAX_SHADER_INPUTS); for (i = 0; i < 4; i++) { - setup->coef[slot].dadx[i] = 0; - setup->coef[slot].dady[i] = 0; + setup.coef[slot].dadx[i] = 0; + setup.coef[slot].dady[i] = 0; /* need provoking vertex info! */ - setup->coef[slot].a0[i] = setup->vprovoke->data[slot][i]; + setup.coef[slot].a0[i] = setup.vprovoke->data[slot][i]; } } @@ -661,20 +663,19 @@ static void const_coeff(struct setup_stage *setup, uint slot) * Compute a0, dadx and dady for a linearly interpolated coefficient, * for a triangle. */ -static void tri_linear_coeff( struct setup_stage *setup, - uint slot, uint firstComp, uint lastComp ) +static void tri_linear_coeff( uint slot, uint firstComp, uint lastComp ) { uint i; for (i = firstComp; i < lastComp; i++) { - float botda = setup->vmid->data[slot][i] - setup->vmin->data[slot][i]; - float majda = setup->vmax->data[slot][i] - setup->vmin->data[slot][i]; - float a = setup->ebot.dy * majda - botda * setup->emaj.dy; - float b = setup->emaj.dx * botda - majda * setup->ebot.dx; + float botda = setup.vmid->data[slot][i] - setup.vmin->data[slot][i]; + float majda = setup.vmax->data[slot][i] - setup.vmin->data[slot][i]; + float a = setup.ebot.dy * majda - botda * setup.emaj.dy; + float b = setup.emaj.dx * botda - majda * setup.ebot.dx; ASSERT(slot < PIPE_MAX_SHADER_INPUTS); - setup->coef[slot].dadx[i] = a * setup->oneoverarea; - setup->coef[slot].dady[i] = b * setup->oneoverarea; + setup.coef[slot].dadx[i] = a * setup.oneoverarea; + setup.coef[slot].dady[i] = b * setup.oneoverarea; /* calculate a0 as the value which would be sampled for the * fragment at (0,0), taking into account that we want to sample at @@ -688,17 +689,17 @@ static void tri_linear_coeff( struct setup_stage *setup, * to define a0 as the sample at a pixel center somewhere near vmin * instead - i'll switch to this later. */ - setup->coef[slot].a0[i] = (setup->vmin->data[slot][i] - - (setup->coef[slot].dadx[i] * (setup->vmin->data[0][0] - 0.5f) + - setup->coef[slot].dady[i] * (setup->vmin->data[0][1] - 0.5f))); + setup.coef[slot].a0[i] = (setup.vmin->data[slot][i] - + (setup.coef[slot].dadx[i] * (setup.vmin->data[0][0] - 0.5f) + + setup.coef[slot].dady[i] * (setup.vmin->data[0][1] - 0.5f))); } /* _mesa_printf("attr[%d].%c: %f dx:%f dy:%f\n", slot, "xyzw"[i], - setup->coef[slot].a0[i], - setup->coef[slot].dadx[i], - setup->coef[slot].dady[i]); + setup.coef[slot].a0[i], + setup.coef[slot].dadx[i], + setup.coef[slot].dady[i]); */ } @@ -712,46 +713,45 @@ static void tri_linear_coeff( struct setup_stage *setup, * Later, when we compute the value at a particular fragment position we'll * divide the interpolated value by the interpolated W at that fragment. */ -static void tri_persp_coeff( struct setup_stage *setup, - unsigned slot, +static void tri_persp_coeff( unsigned slot, unsigned i ) { /* premultiply by 1/w: */ - float mina = setup->vmin->data[slot][i] * setup->vmin->data[0][3]; - float mida = setup->vmid->data[slot][i] * setup->vmid->data[0][3]; - float maxa = setup->vmax->data[slot][i] * setup->vmax->data[0][3]; + float mina = setup.vmin->data[slot][i] * setup.vmin->data[0][3]; + float mida = setup.vmid->data[slot][i] * setup.vmid->data[0][3]; + float maxa = setup.vmax->data[slot][i] * setup.vmax->data[0][3]; float botda = mida - mina; float majda = maxa - mina; - float a = setup->ebot.dy * majda - botda * setup->emaj.dy; - float b = setup->emaj.dx * botda - majda * setup->ebot.dx; + float a = setup.ebot.dy * majda - botda * setup.emaj.dy; + float b = setup.emaj.dx * botda - majda * setup.ebot.dx; /* printf("tri persp %d,%d: %f %f %f\n", slot, i, - setup->vmin->data[slot][i], - setup->vmid->data[slot][i], - setup->vmax->data[slot][i] + setup.vmin->data[slot][i], + setup.vmid->data[slot][i], + setup.vmax->data[slot][i] ); */ assert(slot < PIPE_MAX_SHADER_INPUTS); assert(i <= 3); - setup->coef[slot].dadx[i] = a * setup->oneoverarea; - setup->coef[slot].dady[i] = b * setup->oneoverarea; - setup->coef[slot].a0[i] = (mina - - (setup->coef[slot].dadx[i] * (setup->vmin->data[0][0] - 0.5f) + - setup->coef[slot].dady[i] * (setup->vmin->data[0][1] - 0.5f))); + setup.coef[slot].dadx[i] = a * setup.oneoverarea; + setup.coef[slot].dady[i] = b * setup.oneoverarea; + setup.coef[slot].a0[i] = (mina - + (setup.coef[slot].dadx[i] * (setup.vmin->data[0][0] - 0.5f) + + setup.coef[slot].dady[i] * (setup.vmin->data[0][1] - 0.5f))); } #endif /** - * Compute the setup->coef[] array dadx, dady, a0 values. - * Must be called after setup->vmin,vmid,vmax,vprovoke are initialized. + * Compute the setup.coef[] array dadx, dady, a0 values. + * Must be called after setup.vmin,vmid,vmax,vprovoke are initialized. */ -static void setup_tri_coefficients( struct setup_stage *setup ) +static void setup_tri_coefficients(void) { #if 1 uint i; @@ -761,17 +761,17 @@ static void setup_tri_coefficients( struct setup_stage *setup ) case INTERP_NONE: break; case INTERP_POS: - tri_linear_coeff(setup, i, 2, 3); + tri_linear_coeff(i, 2, 3); /* XXX interp W if PERSPECTIVE... */ break; case INTERP_CONSTANT: - const_coeff(setup, i); + const_coeff(i); break; case INTERP_LINEAR: - tri_linear_coeff(setup, i, 0, 4); + tri_linear_coeff(i, 0, 4); break; case INTERP_PERSPECTIVE: - tri_linear_coeff(setup, i, 0, 4); /* XXX temporary */ + tri_linear_coeff(i, 0, 4); /* XXX temporary */ break; default: ASSERT(0); @@ -781,35 +781,35 @@ static void setup_tri_coefficients( struct setup_stage *setup ) ASSERT(spu.vertex_info.interp_mode[0] == INTERP_POS); ASSERT(spu.vertex_info.interp_mode[1] == INTERP_LINEAR || spu.vertex_info.interp_mode[1] == INTERP_CONSTANT); - tri_linear_coeff(setup, 0, 2, 3); /* slot 0, z */ - tri_linear_coeff(setup, 1, 0, 4); /* slot 1, color */ + tri_linear_coeff(0, 2, 3); /* slot 0, z */ + tri_linear_coeff(1, 0, 4); /* slot 1, color */ #endif } -static void setup_tri_edges( struct setup_stage *setup ) +static void setup_tri_edges(void) { - float vmin_x = setup->vmin->data[0][0] + 0.5f; - float vmid_x = setup->vmid->data[0][0] + 0.5f; - - float vmin_y = setup->vmin->data[0][1] - 0.5f; - float vmid_y = setup->vmid->data[0][1] - 0.5f; - float vmax_y = setup->vmax->data[0][1] - 0.5f; - - setup->emaj.sy = CEILF(vmin_y); - setup->emaj.lines = (int) CEILF(vmax_y - setup->emaj.sy); - setup->emaj.dxdy = setup->emaj.dx / setup->emaj.dy; - setup->emaj.sx = vmin_x + (setup->emaj.sy - vmin_y) * setup->emaj.dxdy; - - setup->etop.sy = CEILF(vmid_y); - setup->etop.lines = (int) CEILF(vmax_y - setup->etop.sy); - setup->etop.dxdy = setup->etop.dx / setup->etop.dy; - setup->etop.sx = vmid_x + (setup->etop.sy - vmid_y) * setup->etop.dxdy; - - setup->ebot.sy = CEILF(vmin_y); - setup->ebot.lines = (int) CEILF(vmid_y - setup->ebot.sy); - setup->ebot.dxdy = setup->ebot.dx / setup->ebot.dy; - setup->ebot.sx = vmin_x + (setup->ebot.sy - vmin_y) * setup->ebot.dxdy; + float vmin_x = setup.vmin->data[0][0] + 0.5f; + float vmid_x = setup.vmid->data[0][0] + 0.5f; + + float vmin_y = setup.vmin->data[0][1] - 0.5f; + float vmid_y = setup.vmid->data[0][1] - 0.5f; + float vmax_y = setup.vmax->data[0][1] - 0.5f; + + setup.emaj.sy = CEILF(vmin_y); + setup.emaj.lines = (int) CEILF(vmax_y - setup.emaj.sy); + setup.emaj.dxdy = setup.emaj.dx / setup.emaj.dy; + setup.emaj.sx = vmin_x + (setup.emaj.sy - vmin_y) * setup.emaj.dxdy; + + setup.etop.sy = CEILF(vmid_y); + setup.etop.lines = (int) CEILF(vmax_y - setup.etop.sy); + setup.etop.dxdy = setup.etop.dx / setup.etop.dy; + setup.etop.sx = vmid_x + (setup.etop.sy - vmid_y) * setup.etop.dxdy; + + setup.ebot.sy = CEILF(vmin_y); + setup.ebot.lines = (int) CEILF(vmid_y - setup.ebot.sy); + setup.ebot.dxdy = setup.ebot.dx / setup.ebot.dy; + setup.ebot.sx = vmin_x + (setup.ebot.sy - vmin_y) * setup.ebot.dxdy; } @@ -817,15 +817,14 @@ static void setup_tri_edges( struct setup_stage *setup ) * Render the upper or lower half of a triangle. * Scissoring/cliprect is applied here too. */ -static void subtriangle( struct setup_stage *setup, - struct edge *eleft, +static void subtriangle( struct edge *eleft, struct edge *eright, unsigned lines ) { - const int minx = setup->cliprect_minx; - const int maxx = setup->cliprect_maxx; - const int miny = setup->cliprect_miny; - const int maxy = setup->cliprect_maxy; + const int minx = setup.cliprect_minx; + const int maxx = setup.cliprect_maxx; + const int miny = setup.cliprect_miny; + const int maxy = setup.cliprect_maxy; int y, start_y, finish_y; int sy = (int)eleft->sy; @@ -867,14 +866,14 @@ static void subtriangle( struct setup_stage *setup, if (left < right) { int _y = sy + y; - if (block(_y) != setup->span.y) { - flush_spans(setup); - setup->span.y = block(_y); + if (block(_y) != setup.span.y) { + flush_spans(); + setup.span.y = block(_y); } - setup->span.left[_y&1] = left; - setup->span.right[_y&1] = right; - setup->span.y_flags |= 1<<(_y&1); + setup.span.left[_y&1] = left; + setup.span.right[_y&1] = right; + setup.span.y_flags |= 1<<(_y&1); } } @@ -892,41 +891,41 @@ static void subtriangle( struct setup_stage *setup, * Do setup for triangle rasterization, then render the triangle. */ static void -setup_tri(struct setup_stage *setup, struct prim_header *prim) +setup_tri(struct prim_header *prim) { - if (!setup_sort_vertices( setup, prim )) { + if (!setup_sort_vertices( prim )) { return; /* totally clipped */ } - setup_tri_coefficients( setup ); - setup_tri_edges( setup ); + setup_tri_coefficients(); + setup_tri_edges(); #if 0 - setup->quad.prim = PRIM_TRI; + setup.quad.prim = PRIM_TRI; #endif - setup->span.y = 0; - setup->span.y_flags = 0; - setup->span.right[0] = 0; - setup->span.right[1] = 0; - /* setup->span.z_mode = tri_z_mode( setup->ctx ); */ + setup.span.y = 0; + setup.span.y_flags = 0; + setup.span.right[0] = 0; + setup.span.right[1] = 0; + /* setup.span.z_mode = tri_z_mode( setup.ctx ); */ /* init_constant_attribs( setup ); */ - if (setup->oneoverarea < 0.0) { + if (setup.oneoverarea < 0.0) { /* emaj on left: */ - subtriangle( setup, &setup->emaj, &setup->ebot, setup->ebot.lines ); - subtriangle( setup, &setup->emaj, &setup->etop, setup->etop.lines ); + subtriangle( &setup.emaj, &setup.ebot, setup.ebot.lines ); + subtriangle( &setup.emaj, &setup.etop, setup.etop.lines ); } else { /* emaj on right: */ - subtriangle( setup, &setup->ebot, &setup->emaj, setup->ebot.lines ); - subtriangle( setup, &setup->etop, &setup->emaj, setup->etop.lines ); + subtriangle( &setup.ebot, &setup.emaj, setup.ebot.lines ); + subtriangle( &setup.etop, &setup.emaj, setup.etop.lines ); } - flush_spans( setup ); + flush_spans(); } @@ -939,7 +938,7 @@ void tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) { struct prim_header tri; - struct setup_stage setup; + /*struct setup_stage setup;*/ tri.v[0] = (struct vertex_header *) v0; tri.v[1] = (struct vertex_header *) v1; @@ -954,5 +953,5 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) setup.cliprect_maxx = (tx + 1) * TILE_SIZE; setup.cliprect_maxy = (ty + 1) * TILE_SIZE; - setup_tri(&setup, &tri); + setup_tri(&tri); } -- cgit v1.2.3 From 0b762d65433445282267c2e6d1dc9ba4eb64af7d Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 12:13:04 -0700 Subject: Cell: fold setup_tri() into tri_draw() --- src/mesa/pipe/cell/spu/spu_tri.c | 62 ++++++++++++---------------------------- 1 file changed, 19 insertions(+), 43 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 5bb2cb12e37..1c615a6e6ad 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -47,9 +47,6 @@ struct vertex_header { float data[0][4]; }; -struct prim_header { - struct vertex_header *v[3]; -}; /* XXX fix this */ @@ -520,11 +517,10 @@ static void print_vertex(const struct vertex_header *v) } #endif -static boolean setup_sort_vertices(const struct prim_header *prim ) +static boolean setup_sort_vertices(const struct vertex_header *v0, + const struct vertex_header *v1, + const struct vertex_header *v2) { - const struct vertex_header *v0 = prim->v[0]; - const struct vertex_header *v1 = prim->v[1]; - const struct vertex_header *v2 = prim->v[2]; #if DEBUG_VERTS fprintf(stderr, "Triangle:\n"); @@ -888,22 +884,30 @@ static void subtriangle( struct edge *eleft, /** - * Do setup for triangle rasterization, then render the triangle. + * Draw triangle into tile at (tx, ty) (tile coords) + * The tile data should have already been fetched. */ -static void -setup_tri(struct prim_header *prim) +void +tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) { - if (!setup_sort_vertices( prim )) { + setup.tx = tx; + setup.ty = ty; + + /* set clipping bounds to tile bounds */ + setup.cliprect_minx = tx * TILE_SIZE; + setup.cliprect_miny = ty * TILE_SIZE; + setup.cliprect_maxx = (tx + 1) * TILE_SIZE; + setup.cliprect_maxy = (ty + 1) * TILE_SIZE; + + if (!setup_sort_vertices((struct vertex_header *) v0, + (struct vertex_header *) v1, + (struct vertex_header *) v2)) { return; /* totally clipped */ } setup_tri_coefficients(); setup_tri_edges(); -#if 0 - setup.quad.prim = PRIM_TRI; -#endif - setup.span.y = 0; setup.span.y_flags = 0; setup.span.right[0] = 0; @@ -927,31 +931,3 @@ setup_tri(struct prim_header *prim) flush_spans(); } - - - -/** - * Draw triangle into tile at (tx, ty) (tile coords) - * The tile data should have already been fetched. - */ -void -tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) -{ - struct prim_header tri; - /*struct setup_stage setup;*/ - - tri.v[0] = (struct vertex_header *) v0; - tri.v[1] = (struct vertex_header *) v1; - tri.v[2] = (struct vertex_header *) v2; - - setup.tx = tx; - setup.ty = ty; - - /* set clipping bounds to tile bounds */ - setup.cliprect_minx = tx * TILE_SIZE; - setup.cliprect_miny = ty * TILE_SIZE; - setup.cliprect_maxx = (tx + 1) * TILE_SIZE; - setup.cliprect_maxy = (ty + 1) * TILE_SIZE; - - setup_tri(&tri); -} -- cgit v1.2.3 From 1af8e381af83ba1344f063a52e9fbf932a77e5f5 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 15:26:51 -0700 Subject: Cell: start to SIMD-ize triangle attribute interpolation Using the spu_add(), etc intrinsics. About a 15% speed-up with some tests. --- src/mesa/pipe/cell/spu/spu_main.h | 7 ++ src/mesa/pipe/cell/spu/spu_texture.c | 6 +- src/mesa/pipe/cell/spu/spu_texture.h | 2 +- src/mesa/pipe/cell/spu/spu_tri.c | 126 +++++++++++++++++++---------------- 4 files changed, 79 insertions(+), 62 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 8908bf8bc07..73f9ed29d66 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -36,6 +36,13 @@ #include "pipe/p_state.h" +typedef union +{ + vector float v; + float f[4]; +} float4; + + struct spu_framebuffer { void *color_start; /**< addr of color surface in main memory */ void *depth_start; /**< addr of depth surface in main memory */ diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 6d566a5006d..7a1ca097c0a 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -128,11 +128,11 @@ get_tex_tile(uint i, uint j) * XXX this is extremely primitive for now. */ uint -sample_texture(const float *texcoord) +sample_texture(float4 texcoord) { /* wrap/repeat */ - uint i = (uint) (texcoord[0] * spu.texture.width) % spu.texture.width; - uint j = (uint) (texcoord[1] * spu.texture.height) % spu.texture.height; + uint i = (uint) (texcoord.f[0] * spu.texture.width) % spu.texture.width; + uint j = (uint) (texcoord.f[1] * spu.texture.height) % spu.texture.height; uint pos = get_tex_tile(i, j); uint texel = tex_tiles[pos].t32[j % TILE_SIZE][i % TILE_SIZE]; return texel; diff --git a/src/mesa/pipe/cell/spu/spu_texture.h b/src/mesa/pipe/cell/spu/spu_texture.h index b75b7ac44fd..938a42b5495 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.h +++ b/src/mesa/pipe/cell/spu/spu_texture.h @@ -37,7 +37,7 @@ invalidate_tex_cache(void); extern uint -sample_texture(const float *texcoord); +sample_texture(float4 texcoord); #endif /* SPU_TEXTURE_H */ diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 1c615a6e6ad..4fc6d908954 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -81,9 +81,9 @@ struct edge { struct interp_coef { - float a0[4]; - float dadx[4]; - float dady[4]; + float4 a0; + float4 dadx; + float4 dady; }; @@ -201,36 +201,31 @@ clip_emit_quad(struct setup_stage *setup) * Eg: four colors will be compute. */ static INLINE void -eval_coeff(uint slot, float x, float y, float result[4][4]) +eval_coeff(uint slot, float x, float y, float4 result[4]) { switch (spu.vertex_info.interp_mode[slot]) { case INTERP_CONSTANT: - { - uint i; - for (i = 0; i < 4; i++) { - result[QUAD_TOP_LEFT][i] = - result[QUAD_TOP_RIGHT][i] = - result[QUAD_BOTTOM_LEFT][i] = - result[QUAD_BOTTOM_RIGHT][i] = setup.coef[slot].a0[i]; - } - } + result[QUAD_TOP_LEFT] = + result[QUAD_TOP_RIGHT] = + result[QUAD_BOTTOM_LEFT] = + result[QUAD_BOTTOM_RIGHT] = setup.coef[slot].a0; break; case INTERP_LINEAR: /* fall-through, for now */ default: { - uint i; - const float *dadx = setup.coef[slot].dadx; - const float *dady = setup.coef[slot].dady; - - /* loop over XYZW comps */ - for (i = 0; i < 4; i++) { - result[QUAD_TOP_LEFT][i] = setup.coef[slot].a0[i] + x * dadx[i] + y * dady[i]; - result[QUAD_TOP_RIGHT][i] = result[0][i] + dadx[i]; - result[QUAD_BOTTOM_LEFT][i] = result[0][i] + dady[i]; - result[QUAD_BOTTOM_RIGHT][i] = result[0][i] + dadx[i] + dady[i]; - } + register vector float dadx = setup.coef[slot].dadx.v; + register vector float dady = setup.coef[slot].dady.v; + register vector float topLeft + = spu_add(setup.coef[slot].a0.v, + spu_add(spu_mul(spu_splats(x), dadx), + spu_mul(spu_splats(y), dady))); + + result[QUAD_TOP_LEFT].v = topLeft; + result[QUAD_TOP_RIGHT].v = spu_add(topLeft, dadx); + result[QUAD_BOTTOM_LEFT].v = spu_add(topLeft, dady); + result[QUAD_BOTTOM_RIGHT].v = spu_add(spu_add(topLeft, dadx), dady); } } } @@ -240,28 +235,46 @@ static INLINE void eval_z(float x, float y, float result[4]) { const uint slot = 0; - const uint i = 2; - const float *dadx = setup.coef[slot].dadx; - const float *dady = setup.coef[slot].dady; - - result[QUAD_TOP_LEFT] = setup.coef[slot].a0[i] + x * dadx[i] + y * dady[i]; - result[QUAD_TOP_RIGHT] = result[0] + dadx[i]; - result[QUAD_BOTTOM_LEFT] = result[0] + dady[i]; - result[QUAD_BOTTOM_RIGHT] = result[0] + dadx[i] + dady[i]; + const float dzdx = setup.coef[slot].dadx.f[2]; + const float dzdy = setup.coef[slot].dady.f[2]; + const float topLeft = setup.coef[slot].a0.f[2] + x * dzdx + y * dzdy; +#if 1 + result[QUAD_TOP_LEFT] = topLeft; + result[QUAD_TOP_RIGHT] = topLeft + dzdx; + result[QUAD_BOTTOM_LEFT] = topLeft + dzdy; + result[QUAD_BOTTOM_RIGHT] = topLeft + dzdx + dzdy; +#else + /* XXX vectorize */ + const vector float topLeftv = spu_splats(topLeft); + const vector float derivs + = (vector float) { 0.0, dzdx, dzdy, dzdx + dzdy }; + vector float *res = (vector float *) result; + *res = spu_add(topLeftv, derivs); +#endif } -static INLINE uint -pack_color(const float color[4]) +static INLINE void +pack_colors(uint uicolors[4], const float4 fcolors[4]) { + /* XXX grab the code for _pack_rgba8() and use the shuffle + * command to do the swizzling seen here. + */ switch (spu.fb.color_format) { case PIPE_FORMAT_A8R8G8B8_UNORM: - return _pack_rgba8(color[3], color[0], color[1], color[2]); + uicolors[0] = _pack_rgba8(fcolors[0].f[3], fcolors[0].f[0], fcolors[0].f[1], fcolors[0].f[2]); + uicolors[1] = _pack_rgba8(fcolors[1].f[3], fcolors[1].f[0], fcolors[1].f[1], fcolors[1].f[2]); + uicolors[2] = _pack_rgba8(fcolors[2].f[3], fcolors[2].f[0], fcolors[2].f[1], fcolors[2].f[2]); + uicolors[3] = _pack_rgba8(fcolors[3].f[3], fcolors[0].f[0], fcolors[3].f[1], fcolors[3].f[2]); + break; case PIPE_FORMAT_B8G8R8A8_UNORM: - return _pack_rgba8(color[2], color[1], color[0], color[3]); + uicolors[0] = _pack_rgba8(fcolors[0].f[2], fcolors[0].f[1], fcolors[0].f[0], fcolors[0].f[3]); + uicolors[1] = _pack_rgba8(fcolors[1].f[2], fcolors[1].f[1], fcolors[1].f[0], fcolors[1].f[3]); + uicolors[2] = _pack_rgba8(fcolors[2].f[2], fcolors[2].f[1], fcolors[2].f[0], fcolors[2].f[3]); + uicolors[3] = _pack_rgba8(fcolors[3].f[2], fcolors[3].f[1], fcolors[3].f[0], fcolors[3].f[3]); + break; default: ASSERT(0); - return 0; } } @@ -379,7 +392,7 @@ emit_quad( int x, int y, unsigned mask ) uint colors[4]; /* indexed by QUAD_x */ if (spu.texture.start) { - float texcoords[4][4]; + float4 texcoords[4]; uint i; eval_coeff(2, (float) x, (float) y, texcoords); for (i = 0; i < 4; i++) { @@ -387,12 +400,9 @@ emit_quad( int x, int y, unsigned mask ) } } else { - float fcolors[4][4]; + float4 fcolors[4]; eval_coeff(1, (float) x, (float) y, fcolors); - colors[QUAD_TOP_LEFT] = pack_color(fcolors[QUAD_TOP_LEFT]); - colors[QUAD_TOP_RIGHT] = pack_color(fcolors[QUAD_TOP_RIGHT]); - colors[QUAD_BOTTOM_LEFT] = pack_color(fcolors[QUAD_BOTTOM_LEFT]); - colors[QUAD_BOTTOM_RIGHT] = pack_color(fcolors[QUAD_BOTTOM_RIGHT]); + pack_colors(colors, fcolors); } if (spu.depth_stencil.depth.enabled) { @@ -645,12 +655,12 @@ static void const_coeff(uint slot) ASSERT(slot < PIPE_MAX_SHADER_INPUTS); for (i = 0; i < 4; i++) { - setup.coef[slot].dadx[i] = 0; - setup.coef[slot].dady[i] = 0; + setup.coef[slot].dadx.f[i] = 0; + setup.coef[slot].dady.f[i] = 0; /* need provoking vertex info! */ - setup.coef[slot].a0[i] = setup.vprovoke->data[slot][i]; + setup.coef[slot].a0.f[i] = setup.vprovoke->data[slot][i]; } } @@ -670,8 +680,8 @@ static void tri_linear_coeff( uint slot, uint firstComp, uint lastComp ) ASSERT(slot < PIPE_MAX_SHADER_INPUTS); - setup.coef[slot].dadx[i] = a * setup.oneoverarea; - setup.coef[slot].dady[i] = b * setup.oneoverarea; + setup.coef[slot].dadx.f[i] = a * setup.oneoverarea; + setup.coef[slot].dady.f[i] = b * setup.oneoverarea; /* calculate a0 as the value which would be sampled for the * fragment at (0,0), taking into account that we want to sample at @@ -685,17 +695,17 @@ static void tri_linear_coeff( uint slot, uint firstComp, uint lastComp ) * to define a0 as the sample at a pixel center somewhere near vmin * instead - i'll switch to this later. */ - setup.coef[slot].a0[i] = (setup.vmin->data[slot][i] - - (setup.coef[slot].dadx[i] * (setup.vmin->data[0][0] - 0.5f) + - setup.coef[slot].dady[i] * (setup.vmin->data[0][1] - 0.5f))); + setup.coef[slot].a0.f[i] = (setup.vmin->data[slot][i] - + (setup.coef[slot].dadx.f[i] * (setup.vmin->data[0][0] - 0.5f) + + setup.coef[slot].dady.f[i] * (setup.vmin->data[0][1] - 0.5f))); } /* _mesa_printf("attr[%d].%c: %f dx:%f dy:%f\n", slot, "xyzw"[i], setup.coef[slot].a0[i], - setup.coef[slot].dadx[i], - setup.coef[slot].dady[i]); + setup.coef[slot].dadx.f[i], + setup.coef[slot].dady.f[i]); */ } @@ -734,11 +744,11 @@ static void tri_persp_coeff( unsigned slot, assert(slot < PIPE_MAX_SHADER_INPUTS); assert(i <= 3); - setup.coef[slot].dadx[i] = a * setup.oneoverarea; - setup.coef[slot].dady[i] = b * setup.oneoverarea; - setup.coef[slot].a0[i] = (mina - - (setup.coef[slot].dadx[i] * (setup.vmin->data[0][0] - 0.5f) + - setup.coef[slot].dady[i] * (setup.vmin->data[0][1] - 0.5f))); + setup.coef[slot].dadx.f[i] = a * setup.oneoverarea; + setup.coef[slot].dady.f[i] = b * setup.oneoverarea; + setup.coef[slot].a0.f[i] = (mina - + (setup.coef[slot].dadx.f[i] * (setup.vmin->data[0][0] - 0.5f) + + setup.coef[slot].dady.f[i] * (setup.vmin->data[0][1] - 0.5f))); } #endif -- cgit v1.2.3 From 3e4306c594e0aa42b2dbf31d7437564466fadfcc Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 30 Jan 2008 20:40:26 -0700 Subject: Cell: prototype SIMD code for z testing --- src/mesa/pipe/cell/spu/spu_tile.h | 10 +++ src/mesa/pipe/cell/spu/spu_tri.c | 147 +++++++++++++++++++++++++++++--------- 2 files changed, 123 insertions(+), 34 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tile.h b/src/mesa/pipe/cell/spu/spu_tile.h index f83dc009c2c..18d1b3c1173 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.h +++ b/src/mesa/pipe/cell/spu/spu_tile.h @@ -42,6 +42,7 @@ typedef union { ushort t16[TILE_SIZE][TILE_SIZE]; uint t32[TILE_SIZE][TILE_SIZE]; + float4 f4[TILE_SIZE/2][TILE_SIZE/2]; } tile_t; @@ -83,9 +84,18 @@ clear_z_tile(tile_t *ztile) TILE_SIZE * TILE_SIZE); } else { + ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z32_UNORM); +#if SIMD_Z + union fi z; + z.f = 1.0; + memset32((uint*) ztile->t32, + z.i,/*spu.fb.depth_clear_value,*/ + TILE_SIZE * TILE_SIZE); +#else memset32((uint*) ztile->t32, spu.fb.depth_clear_value, TILE_SIZE * TILE_SIZE); +#endif } } diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 4fc6d908954..e436e153ece 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -40,6 +40,19 @@ #include "spu_tri.h" +/* + * If SIMD_Z=1 the Z buffer is floating point and we use vector instructions + * to do Z testing/updating. + */ +#define SIMD_Z 0 + +#if SIMD_Z +typedef vector unsigned int mask_t; +#else +typedef uint mask_t; +#endif + + /** * Simplified types taken from other parts of Gallium */ @@ -231,26 +244,16 @@ eval_coeff(uint slot, float x, float y, float4 result[4]) } -static INLINE void -eval_z(float x, float y, float result[4]) +static INLINE vector float +eval_z(float x, float y) { const uint slot = 0; const float dzdx = setup.coef[slot].dadx.f[2]; const float dzdy = setup.coef[slot].dady.f[2]; const float topLeft = setup.coef[slot].a0.f[2] + x * dzdx + y * dzdy; -#if 1 - result[QUAD_TOP_LEFT] = topLeft; - result[QUAD_TOP_RIGHT] = topLeft + dzdx; - result[QUAD_BOTTOM_LEFT] = topLeft + dzdy; - result[QUAD_BOTTOM_RIGHT] = topLeft + dzdx + dzdy; -#else - /* XXX vectorize */ const vector float topLeftv = spu_splats(topLeft); - const vector float derivs - = (vector float) { 0.0, dzdx, dzdy, dzdx + dzdy }; - vector float *res = (vector float *) result; - *res = spu_add(topLeftv, derivs); -#endif + const vector float derivs = (vector float) { 0.0, dzdx, dzdy, dzdx + dzdy }; + return spu_add(topLeftv, derivs); } @@ -279,14 +282,22 @@ pack_colors(uint uicolors[4], const float4 fcolors[4]) } -static uint -do_depth_test(int x, int y, unsigned mask) + +static unsigned int +do_depth_test(int x, int y, unsigned int mask) { + static const float4 zscale16 + = {.f={65535.0, 65535.0, 65535.0, 65535.0}}; + static const float4 zscale32 + = {.f={(float)0xffffffff, + (float)0xffffffff, + (float)0xffffffff, + (float)0xffffffff}}; int ix = x - setup.cliprect_minx; int iy = y - setup.cliprect_miny; - float zvals[4]; + float4 zvals; - eval_z((float) x, (float) y, zvals); + zvals.v = eval_z((float) x, (float) y); if (tile_status_z[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ @@ -300,9 +311,9 @@ do_depth_test(int x, int y, unsigned mask) if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { - const float zscale = 65535.0; + zvals.v = spu_mul(zvals.v, zscale16.v); if (mask & MASK_TOP_LEFT) { - uint z = (uint) (zvals[0] * zscale); + uint z = (uint) zvals.f[0]; if (z < ztile.t16[iy][ix]) ztile.t16[iy][ix] = z; else @@ -310,7 +321,7 @@ do_depth_test(int x, int y, unsigned mask) } if (mask & MASK_TOP_RIGHT) { - uint z = (uint) (zvals[1] * zscale); + uint z = (uint) zvals.f[1]; if (z < ztile.t16[iy][ix+1]) ztile.t16[iy][ix+1] = z; else @@ -318,7 +329,7 @@ do_depth_test(int x, int y, unsigned mask) } if (mask & MASK_BOTTOM_LEFT) { - uint z = (uint) (zvals[2] * zscale); + uint z = (uint) zvals.f[2]; if (z < ztile.t16[iy+1][ix]) ztile.t16[iy+1][ix] = z; else @@ -326,7 +337,7 @@ do_depth_test(int x, int y, unsigned mask) } if (mask & MASK_BOTTOM_RIGHT) { - uint z = (uint) (zvals[3] * zscale); + uint z = (uint) zvals.f[3]; if (z < ztile.t16[iy+1][ix+1]) ztile.t16[iy+1][ix+1] = z; else @@ -334,10 +345,10 @@ do_depth_test(int x, int y, unsigned mask) } } else { - const float zscale = (float) 0xffffffff; + zvals.v = spu_mul(zvals.v, zscale32.v); ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z32_UNORM); if (mask & MASK_TOP_LEFT) { - uint z = (uint) (zvals[0] * zscale); + uint z = (uint) zvals.f[0]; if (z < ztile.t32[iy][ix]) ztile.t32[iy][ix] = z; else @@ -345,7 +356,7 @@ do_depth_test(int x, int y, unsigned mask) } if (mask & MASK_TOP_RIGHT) { - uint z = (uint) (zvals[1] * zscale); + uint z = (uint) zvals.f[1]; if (z < ztile.t32[iy][ix+1]) ztile.t32[iy][ix+1] = z; else @@ -353,7 +364,7 @@ do_depth_test(int x, int y, unsigned mask) } if (mask & MASK_BOTTOM_LEFT) { - uint z = (uint) (zvals[2] * zscale); + uint z = (uint) zvals.f[2]; if (z < ztile.t32[iy+1][ix]) ztile.t32[iy+1][ix] = z; else @@ -361,7 +372,7 @@ do_depth_test(int x, int y, unsigned mask) } if (mask & MASK_BOTTOM_RIGHT) { - uint z = (uint) (zvals[3] * zscale); + uint z = (uint) zvals.f[3]; if (z < ztile.t32[iy+1][ix+1]) ztile.t32[iy+1][ix+1] = z; else @@ -373,11 +384,45 @@ do_depth_test(int x, int y, unsigned mask) } + + +static vector unsigned int +do_depth_test_simd(int x, int y, vector unsigned int quadmask) +{ + int ix = (x - setup.cliprect_minx) / 2; + int iy = (y - setup.cliprect_miny) / 2; + float4 zvals; + + vector unsigned int zmask; + + zvals.v = eval_z((float) x, (float) y); + + if (tile_status_z[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { + /* now, _really_ clear the tile */ + clear_z_tile(&ztile); + } + else if (tile_status_z[setup.ty][setup.tx] != TILE_STATUS_DIRTY) { + /* make sure we've got the tile from main mem */ + wait_on_mask(1 << TAG_READ_TILE_Z); + } + tile_status_z[setup.ty][setup.tx] = TILE_STATUS_DIRTY; + + /* XXX fetch Z value sooner to hide latency here */ + zmask = spu_cmpgt(ztile.f4[ix][iy].v, zvals.v); + zmask = spu_and(zmask, quadmask); + + ztile.f4[ix][iy].v = spu_sel(ztile.f4[ix][iy].v, zvals.v, zmask); + //ztile.f4[ix][iy].v = spu_sel(zvals.v, ztile.f4[ix][iy].v, mask4); + + return zmask; +} + + /** * Emit a quad (pass to next stage). No clipping is done. */ static INLINE void -emit_quad( int x, int y, unsigned mask ) +emit_quad( int x, int y, mask_t mask ) { #if 0 struct softpipe_context *sp = setup.softpipe; @@ -406,10 +451,17 @@ emit_quad( int x, int y, unsigned mask ) } if (spu.depth_stencil.depth.enabled) { - mask &= do_depth_test(x, y, mask); +#if SIMD_Z + mask = do_depth_test_simd(x, y, mask); +#else + mask = do_depth_test(x, y, mask); +#endif } - if (mask) { +#if !SIMD_Z + if (mask) +#endif + { if (tile_status[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_c_tile(&ctile); @@ -420,6 +472,21 @@ emit_quad( int x, int y, unsigned mask ) } tile_status[setup.ty][setup.tx] = TILE_STATUS_DIRTY; +#if SIMD_Z + if (spu_extract(mask, 0)) + ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; + if (spu_extract(mask, 1)) + ctile.t32[iy][ix+1] = colors[QUAD_TOP_RIGHT]; + if (spu_extract(mask, 2)) + ctile.t32[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; + if (spu_extract(mask, 3)) + ctile.t32[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; +#elif 0 + /* SIMD_Z with swizzled color buffer (someday) */ + vector float icolors = *((vector float *) &colors); + ctile.f4[iy/2][ix/2].v = spu_sel(ctile.f4[iy/2][ix/2].v, icolors, mask); + +#else if (mask & MASK_TOP_LEFT) ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; if (mask & MASK_TOP_RIGHT) @@ -428,7 +495,9 @@ emit_quad( int x, int y, unsigned mask ) ctile.t32[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; if (mask & MASK_BOTTOM_RIGHT) ctile.t32[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; +#endif } + #endif } @@ -450,8 +519,18 @@ static INLINE int block( int x ) * this is pretty nasty... may need to rework flush_spans again to * fix it, if possible. */ -static unsigned calculate_mask( int x ) +static mask_t calculate_mask( int x ) { +#if SIMD_Z + uint m0, m1, m2, m3; + + m0 = (x >= setup.span.left[0] && x < setup.span.right[0]) * ~0; + m1 = (x+1 >= setup.span.left[0] && x+1 < setup.span.right[0]) * ~0; + m2 = (x >= setup.span.left[1] && x < setup.span.right[1]) * ~0; + m3 = (x+1 >= setup.span.left[1] && x+1 < setup.span.right[1]) * ~0; + + return (vector unsigned int) {m0, m1, m2, m3}; +#else unsigned mask = 0x0; if (x >= setup.span.left[0] && x < setup.span.right[0]) @@ -467,6 +546,7 @@ static unsigned calculate_mask( int x ) mask |= MASK_BOTTOM_RIGHT; return mask; +#endif } @@ -505,8 +585,7 @@ static void flush_spans( void ) * calculate_mask() could be simplified a bit... */ for (x = block(minleft); x <= block(maxright); x += 2) { - emit_quad( x, setup.span.y, - calculate_mask( x ) ); + emit_quad( x, setup.span.y, calculate_mask( x ) ); } setup.span.y = 0; -- cgit v1.2.3 From 98eecdb4868c181476cbe2423adaa327eee4a02e Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Tue, 29 Jan 2008 10:37:18 -0800 Subject: Initial pass at vertex shader on SPU using TGSI VM All of the code is wired in on the SPU side, but it is not called from the PPU yet. Instruction / declaration fetch still needs to be implemented in spu_exec.c. --- src/mesa/pipe/cell/common.h | 38 + src/mesa/pipe/cell/spu/Makefile | 6 +- src/mesa/pipe/cell/spu/spu_exec.c | 2355 ++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_exec.h | 171 ++ src/mesa/pipe/cell/spu/spu_main.c | 28 + src/mesa/pipe/cell/spu/spu_util.c | 165 ++ src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 493 ++++++ src/mesa/pipe/cell/spu/spu_vertex_shader.c | 224 +++ src/mesa/pipe/cell/spu/spu_vertex_shader.h | 61 + 9 files changed, 3540 insertions(+), 1 deletion(-) create mode 100644 src/mesa/pipe/cell/spu/spu_exec.c create mode 100644 src/mesa/pipe/cell/spu/spu_exec.h create mode 100644 src/mesa/pipe/cell/spu/spu_util.c create mode 100644 src/mesa/pipe/cell/spu/spu_vertex_fetch.c create mode 100644 src/mesa/pipe/cell/spu/spu_vertex_shader.c create mode 100644 src/mesa/pipe/cell/spu/spu_vertex_shader.h diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index d5e86863d48..80a1425ec7e 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -83,6 +83,9 @@ #define CELL_CMD_STATE_SAMPLER 12 #define CELL_CMD_STATE_TEXTURE 13 #define CELL_CMD_STATE_VERTEX_INFO 14 +#define CELL_CMD_STATE_VIEWPORT 15 +#define CELL_CMD_STATE_VS_ARRAY_INFO 16 +#define CELL_CMD_VS_EXECUTE 17 #define CELL_NUM_BUFFERS 4 @@ -116,6 +119,41 @@ struct cell_command_clear_surface } ALIGN16_ATTRIB; +/** + * Array info used by the vertex shader's vertex puller. + */ +struct cell_array_info +{ + void *base; /**< Base address of the 0th element. */ + uint attr; /**< Attribute that this state if for. */ + uint pitch; /**< Byte pitch from one entry to the next. */ + enum pipe_format format; /**< Pipe format of each entry. */ +} ALIGN16_ATTRIB; + + +struct cell_shader_info +{ + unsigned processor; + unsigned num_outputs; + + void *declarations; + unsigned num_declarations; + void *instructions; + unsigned num_instructions; + void *uniforms; +} ALIGN16_ATTRIB; + + +struct cell_command_vs +{ + struct cell_shader_info shader; + void *elts; + unsigned num_elts; + unsigned bytes_per_elt; + void *vOut; +} ALIGN16_ATTRIB; + + struct cell_command_render { uint opcode; /**< CELL_CMD_RENDER */ diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index d5b30e1f27d..2d031bfbc61 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -20,7 +20,11 @@ SOURCES = \ spu_render.c \ spu_texture.c \ spu_tile.c \ - spu_tri.c + spu_tri.c \ + spu_exec.c \ + spu_util.c \ + spu_vertex_fetch.c \ + spu_vertex_shader.c SPU_OBJECTS = $(SOURCES:.c=.o) \ diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c new file mode 100644 index 00000000000..6888e97caf6 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -0,0 +1,2355 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * TGSI interpretor/executor. + * + * Flow control information: + * + * Since we operate on 'quads' (4 pixels or 4 vertices in parallel) + * flow control statements (IF/ELSE/ENDIF, LOOP/ENDLOOP) require special + * care since a condition may be true for some quad components but false + * for other components. + * + * We basically execute all statements (even if they're in the part of + * an IF/ELSE clause that's "not taken") and use a special mask to + * control writing to destination registers. This is the ExecMask. + * See store_dest(). + * + * The ExecMask is computed from three other masks (CondMask, LoopMask and + * ContMask) which are controlled by the flow control instructions (namely: + * (IF/ELSE/ENDIF, LOOP/ENDLOOP and CONT). + * + * + * Authors: + * Michal Krol + * Brian Paul + */ + +#include "pipe/p_compiler.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" +#include "pipe/tgsi/util/tgsi_parse.h" +#include "pipe/tgsi/util/tgsi_util.h" +#include "spu_exec.h" + +#define TILE_TOP_LEFT 0 +#define TILE_TOP_RIGHT 1 +#define TILE_BOTTOM_LEFT 2 +#define TILE_BOTTOM_RIGHT 3 + +/* + * Shorthand locations of various utility registers (_I = Index, _C = Channel) + */ +#define TEMP_0_I TGSI_EXEC_TEMP_00000000_I +#define TEMP_0_C TGSI_EXEC_TEMP_00000000_C +#define TEMP_7F_I TGSI_EXEC_TEMP_7FFFFFFF_I +#define TEMP_7F_C TGSI_EXEC_TEMP_7FFFFFFF_C +#define TEMP_80_I TGSI_EXEC_TEMP_80000000_I +#define TEMP_80_C TGSI_EXEC_TEMP_80000000_C +#define TEMP_FF_I TGSI_EXEC_TEMP_FFFFFFFF_I +#define TEMP_FF_C TGSI_EXEC_TEMP_FFFFFFFF_C +#define TEMP_1_I TGSI_EXEC_TEMP_ONE_I +#define TEMP_1_C TGSI_EXEC_TEMP_ONE_C +#define TEMP_2_I TGSI_EXEC_TEMP_TWO_I +#define TEMP_2_C TGSI_EXEC_TEMP_TWO_C +#define TEMP_128_I TGSI_EXEC_TEMP_128_I +#define TEMP_128_C TGSI_EXEC_TEMP_128_C +#define TEMP_M128_I TGSI_EXEC_TEMP_MINUS_128_I +#define TEMP_M128_C TGSI_EXEC_TEMP_MINUS_128_C +#define TEMP_KILMASK_I TGSI_EXEC_TEMP_KILMASK_I +#define TEMP_KILMASK_C TGSI_EXEC_TEMP_KILMASK_C +#define TEMP_OUTPUT_I TGSI_EXEC_TEMP_OUTPUT_I +#define TEMP_OUTPUT_C TGSI_EXEC_TEMP_OUTPUT_C +#define TEMP_PRIMITIVE_I TGSI_EXEC_TEMP_PRIMITIVE_I +#define TEMP_PRIMITIVE_C TGSI_EXEC_TEMP_PRIMITIVE_C +#define TEMP_R0 TGSI_EXEC_TEMP_R0 + +#define FOR_EACH_CHANNEL(CHAN)\ + for (CHAN = 0; CHAN < 4; CHAN++) + +#define IS_CHANNEL_ENABLED(INST, CHAN)\ + ((INST).FullDstRegisters[0].DstRegister.WriteMask & (1 << (CHAN))) + +#define IS_CHANNEL_ENABLED2(INST, CHAN)\ + ((INST).FullDstRegisters[1].DstRegister.WriteMask & (1 << (CHAN))) + +#define FOR_EACH_ENABLED_CHANNEL(INST, CHAN)\ + FOR_EACH_CHANNEL( CHAN )\ + if (IS_CHANNEL_ENABLED( INST, CHAN )) + +#define FOR_EACH_ENABLED_CHANNEL2(INST, CHAN)\ + FOR_EACH_CHANNEL( CHAN )\ + if (IS_CHANNEL_ENABLED2( INST, CHAN )) + + +/** The execution mask depends on the conditional mask and the loop mask */ +#define UPDATE_EXEC_MASK(MACH) \ + MACH->ExecMask = MACH->CondMask & MACH->LoopMask & MACH->ContMask & MACH->FuncMask + + +#define CHAN_X 0 +#define CHAN_Y 1 +#define CHAN_Z 2 +#define CHAN_W 3 + + + +/** + * Initialize machine state by expanding tokens to full instructions, + * allocating temporary storage, setting up constants, etc. + * After this, we can call spu_exec_machine_run() many times. + */ +void +spu_exec_machine_init(struct spu_exec_machine *mach, + uint numSamplers, + struct spu_sampler *samplers, + unsigned processor) +{ + uint i; + + mach->Samplers = samplers; + mach->Processor = processor; + mach->Addrs = &mach->Temps[TGSI_EXEC_NUM_TEMPS]; + + /* Setup constants. */ + for( i = 0; i < 4; i++ ) { + mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].u[i] = 0x00000000; + mach->Temps[TEMP_7F_I].xyzw[TEMP_7F_C].u[i] = 0x7FFFFFFF; + mach->Temps[TEMP_80_I].xyzw[TEMP_80_C].u[i] = 0x80000000; + mach->Temps[TEMP_FF_I].xyzw[TEMP_FF_C].u[i] = 0xFFFFFFFF; + mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].f[i] = 1.0f; + mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].f[i] = 2.0f; + mach->Temps[TEMP_128_I].xyzw[TEMP_128_C].f[i] = 128.0f; + mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C].f[i] = -128.0f; + } +} + + +static void +micro_abs( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->f[0] = (float) fabs( (double) src->f[0] ); + dst->f[1] = (float) fabs( (double) src->f[1] ); + dst->f[2] = (float) fabs( (double) src->f[2] ); + dst->f[3] = (float) fabs( (double) src->f[3] ); +} + +static void +micro_add( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->f[0] = src0->f[0] + src1->f[0]; + dst->f[1] = src0->f[1] + src1->f[1]; + dst->f[2] = src0->f[2] + src1->f[2]; + dst->f[3] = src0->f[3] + src1->f[3]; +} + +static void +micro_iadd( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->i[0] = src0->i[0] + src1->i[0]; + dst->i[1] = src0->i[1] + src1->i[1]; + dst->i[2] = src0->i[2] + src1->i[2]; + dst->i[3] = src0->i[3] + src1->i[3]; +} + +static void +micro_and( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] & src1->u[0]; + dst->u[1] = src0->u[1] & src1->u[1]; + dst->u[2] = src0->u[2] & src1->u[2]; + dst->u[3] = src0->u[3] & src1->u[3]; +} + +static void +micro_ceil( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) ceil( (double) src->f[0] ); + dst->f[1] = (float) ceil( (double) src->f[1] ); + dst->f[2] = (float) ceil( (double) src->f[2] ); + dst->f[3] = (float) ceil( (double) src->f[3] ); +#endif +} + +static void +micro_cos( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) cos( (double) src->f[0] ); + dst->f[1] = (float) cos( (double) src->f[1] ); + dst->f[2] = (float) cos( (double) src->f[2] ); + dst->f[3] = (float) cos( (double) src->f[3] ); +#endif +} + +static void +micro_ddx( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->f[0] = + dst->f[1] = + dst->f[2] = + dst->f[3] = src->f[TILE_BOTTOM_RIGHT] - src->f[TILE_BOTTOM_LEFT]; +} + +static void +micro_ddy( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->f[0] = + dst->f[1] = + dst->f[2] = + dst->f[3] = src->f[TILE_TOP_LEFT] - src->f[TILE_BOTTOM_LEFT]; +} + +static void +micro_div( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->f[0] = src0->f[0] / src1->f[0]; + dst->f[1] = src0->f[1] / src1->f[1]; + dst->f[2] = src0->f[2] / src1->f[2]; + dst->f[3] = src0->f[3] / src1->f[3]; +} + +static void +micro_udiv( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] / src1->u[0]; + dst->u[1] = src0->u[1] / src1->u[1]; + dst->u[2] = src0->u[2] / src1->u[2]; + dst->u[3] = src0->u[3] / src1->u[3]; +} + +static void +micro_eq( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2, + const union spu_exec_channel *src3 ) +{ + dst->f[0] = src0->f[0] == src1->f[0] ? src2->f[0] : src3->f[0]; + dst->f[1] = src0->f[1] == src1->f[1] ? src2->f[1] : src3->f[1]; + dst->f[2] = src0->f[2] == src1->f[2] ? src2->f[2] : src3->f[2]; + dst->f[3] = src0->f[3] == src1->f[3] ? src2->f[3] : src3->f[3]; +} + +static void +micro_ieq( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2, + const union spu_exec_channel *src3 ) +{ + dst->i[0] = src0->i[0] == src1->i[0] ? src2->i[0] : src3->i[0]; + dst->i[1] = src0->i[1] == src1->i[1] ? src2->i[1] : src3->i[1]; + dst->i[2] = src0->i[2] == src1->i[2] ? src2->i[2] : src3->i[2]; + dst->i[3] = src0->i[3] == src1->i[3] ? src2->i[3] : src3->i[3]; +} + +static void +micro_exp2( + union spu_exec_channel *dst, + const union spu_exec_channel *src) +{ +#if 0 + dst->f[0] = (float) pow( 2.0, (double) src->f[0] ); + dst->f[1] = (float) pow( 2.0, (double) src->f[1] ); + dst->f[2] = (float) pow( 2.0, (double) src->f[2] ); + dst->f[3] = (float) pow( 2.0, (double) src->f[3] ); +#endif +} + +static void +micro_f2it( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->i[0] = (int) src->f[0]; + dst->i[1] = (int) src->f[1]; + dst->i[2] = (int) src->f[2]; + dst->i[3] = (int) src->f[3]; +} + +static void +micro_f2ut( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->u[0] = (uint) src->f[0]; + dst->u[1] = (uint) src->f[1]; + dst->u[2] = (uint) src->f[2]; + dst->u[3] = (uint) src->f[3]; +} + +static void +micro_flr( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) floor( (double) src->f[0] ); + dst->f[1] = (float) floor( (double) src->f[1] ); + dst->f[2] = (float) floor( (double) src->f[2] ); + dst->f[3] = (float) floor( (double) src->f[3] ); +#endif +} + +static void +micro_frc( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = src->f[0] - (float) floor( (double) src->f[0] ); + dst->f[1] = src->f[1] - (float) floor( (double) src->f[1] ); + dst->f[2] = src->f[2] - (float) floor( (double) src->f[2] ); + dst->f[3] = src->f[3] - (float) floor( (double) src->f[3] ); +#endif +} + +static void +micro_ge( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2, + const union spu_exec_channel *src3 ) +{ + dst->f[0] = src0->f[0] >= src1->f[0] ? src2->f[0] : src3->f[0]; + dst->f[1] = src0->f[1] >= src1->f[1] ? src2->f[1] : src3->f[1]; + dst->f[2] = src0->f[2] >= src1->f[2] ? src2->f[2] : src3->f[2]; + dst->f[3] = src0->f[3] >= src1->f[3] ? src2->f[3] : src3->f[3]; +} + +static void +micro_i2f( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->f[0] = (float) src->i[0]; + dst->f[1] = (float) src->i[1]; + dst->f[2] = (float) src->i[2]; + dst->f[3] = (float) src->i[3]; +} + +static void +micro_lg2( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) log( (double) src->f[0] ) * 1.442695f; + dst->f[1] = (float) log( (double) src->f[1] ) * 1.442695f; + dst->f[2] = (float) log( (double) src->f[2] ) * 1.442695f; + dst->f[3] = (float) log( (double) src->f[3] ) * 1.442695f; +#endif +} + +static void +micro_lt( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2, + const union spu_exec_channel *src3 ) +{ + dst->f[0] = src0->f[0] < src1->f[0] ? src2->f[0] : src3->f[0]; + dst->f[1] = src0->f[1] < src1->f[1] ? src2->f[1] : src3->f[1]; + dst->f[2] = src0->f[2] < src1->f[2] ? src2->f[2] : src3->f[2]; + dst->f[3] = src0->f[3] < src1->f[3] ? src2->f[3] : src3->f[3]; +} + +static void +micro_ilt( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2, + const union spu_exec_channel *src3 ) +{ + dst->i[0] = src0->i[0] < src1->i[0] ? src2->i[0] : src3->i[0]; + dst->i[1] = src0->i[1] < src1->i[1] ? src2->i[1] : src3->i[1]; + dst->i[2] = src0->i[2] < src1->i[2] ? src2->i[2] : src3->i[2]; + dst->i[3] = src0->i[3] < src1->i[3] ? src2->i[3] : src3->i[3]; +} + +static void +micro_ult( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2, + const union spu_exec_channel *src3 ) +{ + dst->u[0] = src0->u[0] < src1->u[0] ? src2->u[0] : src3->u[0]; + dst->u[1] = src0->u[1] < src1->u[1] ? src2->u[1] : src3->u[1]; + dst->u[2] = src0->u[2] < src1->u[2] ? src2->u[2] : src3->u[2]; + dst->u[3] = src0->u[3] < src1->u[3] ? src2->u[3] : src3->u[3]; +} + +static void +micro_max( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->f[0] = src0->f[0] > src1->f[0] ? src0->f[0] : src1->f[0]; + dst->f[1] = src0->f[1] > src1->f[1] ? src0->f[1] : src1->f[1]; + dst->f[2] = src0->f[2] > src1->f[2] ? src0->f[2] : src1->f[2]; + dst->f[3] = src0->f[3] > src1->f[3] ? src0->f[3] : src1->f[3]; +} + +static void +micro_imax( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->i[0] = src0->i[0] > src1->i[0] ? src0->i[0] : src1->i[0]; + dst->i[1] = src0->i[1] > src1->i[1] ? src0->i[1] : src1->i[1]; + dst->i[2] = src0->i[2] > src1->i[2] ? src0->i[2] : src1->i[2]; + dst->i[3] = src0->i[3] > src1->i[3] ? src0->i[3] : src1->i[3]; +} + +static void +micro_umax( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] > src1->u[0] ? src0->u[0] : src1->u[0]; + dst->u[1] = src0->u[1] > src1->u[1] ? src0->u[1] : src1->u[1]; + dst->u[2] = src0->u[2] > src1->u[2] ? src0->u[2] : src1->u[2]; + dst->u[3] = src0->u[3] > src1->u[3] ? src0->u[3] : src1->u[3]; +} + +static void +micro_min( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->f[0] = src0->f[0] < src1->f[0] ? src0->f[0] : src1->f[0]; + dst->f[1] = src0->f[1] < src1->f[1] ? src0->f[1] : src1->f[1]; + dst->f[2] = src0->f[2] < src1->f[2] ? src0->f[2] : src1->f[2]; + dst->f[3] = src0->f[3] < src1->f[3] ? src0->f[3] : src1->f[3]; +} + +static void +micro_imin( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->i[0] = src0->i[0] < src1->i[0] ? src0->i[0] : src1->i[0]; + dst->i[1] = src0->i[1] < src1->i[1] ? src0->i[1] : src1->i[1]; + dst->i[2] = src0->i[2] < src1->i[2] ? src0->i[2] : src1->i[2]; + dst->i[3] = src0->i[3] < src1->i[3] ? src0->i[3] : src1->i[3]; +} + +static void +micro_umin( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] < src1->u[0] ? src0->u[0] : src1->u[0]; + dst->u[1] = src0->u[1] < src1->u[1] ? src0->u[1] : src1->u[1]; + dst->u[2] = src0->u[2] < src1->u[2] ? src0->u[2] : src1->u[2]; + dst->u[3] = src0->u[3] < src1->u[3] ? src0->u[3] : src1->u[3]; +} + +static void +micro_umod( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] % src1->u[0]; + dst->u[1] = src0->u[1] % src1->u[1]; + dst->u[2] = src0->u[2] % src1->u[2]; + dst->u[3] = src0->u[3] % src1->u[3]; +} + +static void +micro_mul( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->f[0] = src0->f[0] * src1->f[0]; + dst->f[1] = src0->f[1] * src1->f[1]; + dst->f[2] = src0->f[2] * src1->f[2]; + dst->f[3] = src0->f[3] * src1->f[3]; +} + +static void +micro_imul( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->i[0] = src0->i[0] * src1->i[0]; + dst->i[1] = src0->i[1] * src1->i[1]; + dst->i[2] = src0->i[2] * src1->i[2]; + dst->i[3] = src0->i[3] * src1->i[3]; +} + +static void +micro_imul64( + union spu_exec_channel *dst0, + union spu_exec_channel *dst1, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst1->i[0] = src0->i[0] * src1->i[0]; + dst1->i[1] = src0->i[1] * src1->i[1]; + dst1->i[2] = src0->i[2] * src1->i[2]; + dst1->i[3] = src0->i[3] * src1->i[3]; + dst0->i[0] = 0; + dst0->i[1] = 0; + dst0->i[2] = 0; + dst0->i[3] = 0; +} + +static void +micro_umul64( + union spu_exec_channel *dst0, + union spu_exec_channel *dst1, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst1->u[0] = src0->u[0] * src1->u[0]; + dst1->u[1] = src0->u[1] * src1->u[1]; + dst1->u[2] = src0->u[2] * src1->u[2]; + dst1->u[3] = src0->u[3] * src1->u[3]; + dst0->u[0] = 0; + dst0->u[1] = 0; + dst0->u[2] = 0; + dst0->u[3] = 0; +} + +static void +micro_movc( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1, + const union spu_exec_channel *src2 ) +{ + dst->u[0] = src0->u[0] ? src1->u[0] : src2->u[0]; + dst->u[1] = src0->u[1] ? src1->u[1] : src2->u[1]; + dst->u[2] = src0->u[2] ? src1->u[2] : src2->u[2]; + dst->u[3] = src0->u[3] ? src1->u[3] : src2->u[3]; +} + +static void +micro_neg( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->f[0] = -src->f[0]; + dst->f[1] = -src->f[1]; + dst->f[2] = -src->f[2]; + dst->f[3] = -src->f[3]; +} + +static void +micro_ineg( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->i[0] = -src->i[0]; + dst->i[1] = -src->i[1]; + dst->i[2] = -src->i[2]; + dst->i[3] = -src->i[3]; +} + +static void +micro_not( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->u[0] = ~src->u[0]; + dst->u[1] = ~src->u[1]; + dst->u[2] = ~src->u[2]; + dst->u[3] = ~src->u[3]; +} + +static void +micro_or( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] | src1->u[0]; + dst->u[1] = src0->u[1] | src1->u[1]; + dst->u[2] = src0->u[2] | src1->u[2]; + dst->u[3] = src0->u[3] | src1->u[3]; +} + +static void +micro_pow( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ +#if 0 + dst->f[0] = (float) pow( (double) src0->f[0], (double) src1->f[0] ); + dst->f[1] = (float) pow( (double) src0->f[1], (double) src1->f[1] ); + dst->f[2] = (float) pow( (double) src0->f[2], (double) src1->f[2] ); + dst->f[3] = (float) pow( (double) src0->f[3], (double) src1->f[3] ); +#endif +} + +static void +micro_rnd( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) floor( (double) (src->f[0] + 0.5f) ); + dst->f[1] = (float) floor( (double) (src->f[1] + 0.5f) ); + dst->f[2] = (float) floor( (double) (src->f[2] + 0.5f) ); + dst->f[3] = (float) floor( (double) (src->f[3] + 0.5f) ); +#endif +} + +static void +micro_shl( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->i[0] = src0->i[0] << src1->i[0]; + dst->i[1] = src0->i[1] << src1->i[1]; + dst->i[2] = src0->i[2] << src1->i[2]; + dst->i[3] = src0->i[3] << src1->i[3]; +} + +static void +micro_ishr( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->i[0] = src0->i[0] >> src1->i[0]; + dst->i[1] = src0->i[1] >> src1->i[1]; + dst->i[2] = src0->i[2] >> src1->i[2]; + dst->i[3] = src0->i[3] >> src1->i[3]; +} + +static void +micro_trunc( + union spu_exec_channel *dst, + const union spu_exec_channel *src0 ) +{ + dst->f[0] = (float) (int) src0->f[0]; + dst->f[1] = (float) (int) src0->f[1]; + dst->f[2] = (float) (int) src0->f[2]; + dst->f[3] = (float) (int) src0->f[3]; +} + +static void +micro_ushr( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] >> src1->u[0]; + dst->u[1] = src0->u[1] >> src1->u[1]; + dst->u[2] = src0->u[2] >> src1->u[2]; + dst->u[3] = src0->u[3] >> src1->u[3]; +} + +static void +micro_sin( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) sin( (double) src->f[0] ); + dst->f[1] = (float) sin( (double) src->f[1] ); + dst->f[2] = (float) sin( (double) src->f[2] ); + dst->f[3] = (float) sin( (double) src->f[3] ); +#endif +} + +static void +micro_sqrt( union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ +#if 0 + dst->f[0] = (float) sqrt( (double) src->f[0] ); + dst->f[1] = (float) sqrt( (double) src->f[1] ); + dst->f[2] = (float) sqrt( (double) src->f[2] ); + dst->f[3] = (float) sqrt( (double) src->f[3] ); +#endif +} + +static void +micro_sub( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->f[0] = src0->f[0] - src1->f[0]; + dst->f[1] = src0->f[1] - src1->f[1]; + dst->f[2] = src0->f[2] - src1->f[2]; + dst->f[3] = src0->f[3] - src1->f[3]; +} + +static void +micro_u2f( + union spu_exec_channel *dst, + const union spu_exec_channel *src ) +{ + dst->f[0] = (float) src->u[0]; + dst->f[1] = (float) src->u[1]; + dst->f[2] = (float) src->u[2]; + dst->f[3] = (float) src->u[3]; +} + +static void +micro_xor( + union spu_exec_channel *dst, + const union spu_exec_channel *src0, + const union spu_exec_channel *src1 ) +{ + dst->u[0] = src0->u[0] ^ src1->u[0]; + dst->u[1] = src0->u[1] ^ src1->u[1]; + dst->u[2] = src0->u[2] ^ src1->u[2]; + dst->u[3] = src0->u[3] ^ src1->u[3]; +} + +static void +fetch_src_file_channel( + const struct spu_exec_machine *mach, + const uint file, + const uint swizzle, + const union spu_exec_channel *index, + union spu_exec_channel *chan ) +{ + switch( swizzle ) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + switch( file ) { + case TGSI_FILE_CONSTANT: + chan->f[0] = mach->Consts[index->i[0]][swizzle]; + chan->f[1] = mach->Consts[index->i[1]][swizzle]; + chan->f[2] = mach->Consts[index->i[2]][swizzle]; + chan->f[3] = mach->Consts[index->i[3]][swizzle]; + break; + + case TGSI_FILE_INPUT: + chan->u[0] = mach->Inputs[index->i[0]].xyzw[swizzle].u[0]; + chan->u[1] = mach->Inputs[index->i[1]].xyzw[swizzle].u[1]; + chan->u[2] = mach->Inputs[index->i[2]].xyzw[swizzle].u[2]; + chan->u[3] = mach->Inputs[index->i[3]].xyzw[swizzle].u[3]; + break; + + case TGSI_FILE_TEMPORARY: + chan->u[0] = mach->Temps[index->i[0]].xyzw[swizzle].u[0]; + chan->u[1] = mach->Temps[index->i[1]].xyzw[swizzle].u[1]; + chan->u[2] = mach->Temps[index->i[2]].xyzw[swizzle].u[2]; + chan->u[3] = mach->Temps[index->i[3]].xyzw[swizzle].u[3]; + break; + + case TGSI_FILE_IMMEDIATE: + assert( index->i[0] < (int) mach->ImmLimit ); + assert( index->i[1] < (int) mach->ImmLimit ); + assert( index->i[2] < (int) mach->ImmLimit ); + assert( index->i[3] < (int) mach->ImmLimit ); + + chan->f[0] = mach->Imms[index->i[0]][swizzle]; + chan->f[1] = mach->Imms[index->i[1]][swizzle]; + chan->f[2] = mach->Imms[index->i[2]][swizzle]; + chan->f[3] = mach->Imms[index->i[3]][swizzle]; + break; + + case TGSI_FILE_ADDRESS: + chan->u[0] = mach->Addrs[index->i[0]].xyzw[swizzle].u[0]; + chan->u[1] = mach->Addrs[index->i[1]].xyzw[swizzle].u[1]; + chan->u[2] = mach->Addrs[index->i[2]].xyzw[swizzle].u[2]; + chan->u[3] = mach->Addrs[index->i[3]].xyzw[swizzle].u[3]; + break; + + case TGSI_FILE_OUTPUT: + /* vertex/fragment output vars can be read too */ + chan->u[0] = mach->Outputs[index->i[0]].xyzw[swizzle].u[0]; + chan->u[1] = mach->Outputs[index->i[1]].xyzw[swizzle].u[1]; + chan->u[2] = mach->Outputs[index->i[2]].xyzw[swizzle].u[2]; + chan->u[3] = mach->Outputs[index->i[3]].xyzw[swizzle].u[3]; + break; + + default: + assert( 0 ); + } + break; + + case TGSI_EXTSWIZZLE_ZERO: + *chan = mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]; + break; + + case TGSI_EXTSWIZZLE_ONE: + *chan = mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]; + break; + + default: + assert( 0 ); + } +} + +static void +fetch_source( + const struct spu_exec_machine *mach, + union spu_exec_channel *chan, + const struct tgsi_full_src_register *reg, + const uint chan_index ) +{ + union spu_exec_channel index; + uint swizzle; + + index.i[0] = + index.i[1] = + index.i[2] = + index.i[3] = reg->SrcRegister.Index; + + if (reg->SrcRegister.Indirect) { + union spu_exec_channel index2; + union spu_exec_channel indir_index; + + index2.i[0] = + index2.i[1] = + index2.i[2] = + index2.i[3] = reg->SrcRegisterInd.Index; + + swizzle = tgsi_util_get_src_register_swizzle(®->SrcRegisterInd, + CHAN_X); + fetch_src_file_channel( + mach, + reg->SrcRegisterInd.File, + swizzle, + &index2, + &indir_index ); + + index.i[0] += indir_index.i[0]; + index.i[1] += indir_index.i[1]; + index.i[2] += indir_index.i[2]; + index.i[3] += indir_index.i[3]; + } + + if( reg->SrcRegister.Dimension ) { + switch( reg->SrcRegister.File ) { + case TGSI_FILE_INPUT: + index.i[0] *= 17; + index.i[1] *= 17; + index.i[2] *= 17; + index.i[3] *= 17; + break; + case TGSI_FILE_CONSTANT: + index.i[0] *= 4096; + index.i[1] *= 4096; + index.i[2] *= 4096; + index.i[3] *= 4096; + break; + default: + assert( 0 ); + } + + index.i[0] += reg->SrcRegisterDim.Index; + index.i[1] += reg->SrcRegisterDim.Index; + index.i[2] += reg->SrcRegisterDim.Index; + index.i[3] += reg->SrcRegisterDim.Index; + + if (reg->SrcRegisterDim.Indirect) { + union spu_exec_channel index2; + union spu_exec_channel indir_index; + + index2.i[0] = + index2.i[1] = + index2.i[2] = + index2.i[3] = reg->SrcRegisterDimInd.Index; + + swizzle = tgsi_util_get_src_register_swizzle( ®->SrcRegisterDimInd, CHAN_X ); + fetch_src_file_channel( + mach, + reg->SrcRegisterDimInd.File, + swizzle, + &index2, + &indir_index ); + + index.i[0] += indir_index.i[0]; + index.i[1] += indir_index.i[1]; + index.i[2] += indir_index.i[2]; + index.i[3] += indir_index.i[3]; + } + } + + swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index ); + fetch_src_file_channel( + mach, + reg->SrcRegister.File, + swizzle, + &index, + chan ); + + switch (tgsi_util_get_full_src_register_sign_mode( reg, chan_index )) { + case TGSI_UTIL_SIGN_CLEAR: + micro_abs( chan, chan ); + break; + + case TGSI_UTIL_SIGN_SET: + micro_abs( chan, chan ); + micro_neg( chan, chan ); + break; + + case TGSI_UTIL_SIGN_TOGGLE: + micro_neg( chan, chan ); + break; + + case TGSI_UTIL_SIGN_KEEP: + break; + } + + if (reg->SrcRegisterExtMod.Complement) { + micro_sub( chan, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], chan ); + } +} + +static void +store_dest( + struct spu_exec_machine *mach, + const union spu_exec_channel *chan, + const struct tgsi_full_dst_register *reg, + const struct tgsi_full_instruction *inst, + uint chan_index ) +{ + union spu_exec_channel *dst; + + switch( reg->DstRegister.File ) { + case TGSI_FILE_NULL: + return; + + case TGSI_FILE_OUTPUT: + dst = &mach->Outputs[mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] + + reg->DstRegister.Index].xyzw[chan_index]; + break; + + case TGSI_FILE_TEMPORARY: + dst = &mach->Temps[reg->DstRegister.Index].xyzw[chan_index]; + break; + + case TGSI_FILE_ADDRESS: + dst = &mach->Addrs[reg->DstRegister.Index].xyzw[chan_index]; + break; + + default: + assert( 0 ); + return; + } + + switch (inst->Instruction.Saturate) + { + case TGSI_SAT_NONE: + if (mach->ExecMask & 0x1) + dst->i[0] = chan->i[0]; + if (mach->ExecMask & 0x2) + dst->i[1] = chan->i[1]; + if (mach->ExecMask & 0x4) + dst->i[2] = chan->i[2]; + if (mach->ExecMask & 0x8) + dst->i[3] = chan->i[3]; + break; + + case TGSI_SAT_ZERO_ONE: + /* XXX need to obey ExecMask here */ + micro_max(dst, chan, &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]); + micro_min(dst, dst, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]); + break; + + case TGSI_SAT_MINUS_PLUS_ONE: + assert( 0 ); + break; + + default: + assert( 0 ); + } +} + +#define FETCH(VAL,INDEX,CHAN)\ + fetch_source (mach, VAL, &inst->FullSrcRegisters[INDEX], CHAN) + +#define STORE(VAL,INDEX,CHAN)\ + store_dest (mach, VAL, &inst->FullDstRegisters[INDEX], inst, CHAN ) + + +/** + * Execute ARB-style KIL which is predicated by a src register. + * Kill fragment if any of the four values is less than zero. + */ +static void +exec_kilp(struct spu_exec_machine *mach, + const struct tgsi_full_instruction *inst) +{ + uint uniquemask; + uint chan_index; + uint kilmask = 0; /* bit 0 = pixel 0, bit 1 = pixel 1, etc */ + union spu_exec_channel r[1]; + + /* This mask stores component bits that were already tested. Note that + * we test if the value is less than zero, so 1.0 and 0.0 need not to be + * tested. */ + uniquemask = (1 << TGSI_EXTSWIZZLE_ZERO) | (1 << TGSI_EXTSWIZZLE_ONE); + + for (chan_index = 0; chan_index < 4; chan_index++) + { + uint swizzle; + uint i; + + /* unswizzle channel */ + swizzle = tgsi_util_get_full_src_register_extswizzle ( + &inst->FullSrcRegisters[0], + chan_index); + + /* check if the component has not been already tested */ + if (uniquemask & (1 << swizzle)) + continue; + uniquemask |= 1 << swizzle; + + FETCH(&r[0], 0, chan_index); + for (i = 0; i < 4; i++) + if (r[0].f[i] < 0.0f) + kilmask |= 1 << i; + } + + mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] |= kilmask; +} + + +/* + * Fetch a texel using STR texture coordinates. + */ +static void +fetch_texel( struct spu_sampler *sampler, + const union spu_exec_channel *s, + const union spu_exec_channel *t, + const union spu_exec_channel *p, + float lodbias, /* XXX should be float[4] */ + union spu_exec_channel *r, + union spu_exec_channel *g, + union spu_exec_channel *b, + union spu_exec_channel *a ) +{ + uint j; + float rgba[NUM_CHANNELS][QUAD_SIZE]; + + sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, rgba); + + for (j = 0; j < 4; j++) { + r->f[j] = rgba[0][j]; + g->f[j] = rgba[1][j]; + b->f[j] = rgba[2][j]; + a->f[j] = rgba[3][j]; + } +} + + +static void +exec_tex(struct spu_exec_machine *mach, + const struct tgsi_full_instruction *inst, + boolean biasLod) +{ + const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index; + union spu_exec_channel r[8]; + uint chan_index; + float lodBias; + + /* printf("Sampler %u unit %u\n", sampler, unit); */ + + switch (inst->InstructionExtTexture.Texture) { + case TGSI_TEXTURE_1D: + + FETCH(&r[0], 0, CHAN_X); + + switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) { + case TGSI_EXTSWIZZLE_W: + FETCH(&r[1], 0, CHAN_W); + micro_div( &r[0], &r[0], &r[1] ); + break; + + case TGSI_EXTSWIZZLE_ONE: + break; + + default: + assert (0); + } + + if (biasLod) { + FETCH(&r[1], 0, CHAN_W); + lodBias = r[2].f[0]; + } + else + lodBias = 0.0; + + fetch_texel(&mach->Samplers[unit], + &r[0], NULL, NULL, lodBias, /* S, T, P, BIAS */ + &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */ + break; + + case TGSI_TEXTURE_2D: + case TGSI_TEXTURE_RECT: + + FETCH(&r[0], 0, CHAN_X); + FETCH(&r[1], 0, CHAN_Y); + FETCH(&r[2], 0, CHAN_Z); + + switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) { + case TGSI_EXTSWIZZLE_W: + FETCH(&r[3], 0, CHAN_W); + micro_div( &r[0], &r[0], &r[3] ); + micro_div( &r[1], &r[1], &r[3] ); + micro_div( &r[2], &r[2], &r[3] ); + break; + + case TGSI_EXTSWIZZLE_ONE: + break; + + default: + assert (0); + } + + if (biasLod) { + FETCH(&r[3], 0, CHAN_W); + lodBias = r[3].f[0]; + } + else + lodBias = 0.0; + + fetch_texel(&mach->Samplers[unit], + &r[0], &r[1], &r[2], lodBias, /* inputs */ + &r[0], &r[1], &r[2], &r[3]); /* outputs */ + break; + + case TGSI_TEXTURE_3D: + case TGSI_TEXTURE_CUBE: + + FETCH(&r[0], 0, CHAN_X); + FETCH(&r[1], 0, CHAN_Y); + FETCH(&r[2], 0, CHAN_Z); + + switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) { + case TGSI_EXTSWIZZLE_W: + FETCH(&r[3], 0, CHAN_W); + micro_div( &r[0], &r[0], &r[3] ); + micro_div( &r[1], &r[1], &r[3] ); + micro_div( &r[2], &r[2], &r[3] ); + break; + + case TGSI_EXTSWIZZLE_ONE: + break; + + default: + assert (0); + } + + if (biasLod) { + FETCH(&r[3], 0, CHAN_W); + lodBias = r[3].f[0]; + } + else + lodBias = 0.0; + + fetch_texel(&mach->Samplers[unit], + &r[0], &r[1], &r[2], lodBias, + &r[0], &r[1], &r[2], &r[3]); + break; + + default: + assert (0); + } + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[chan_index], 0, chan_index ); + } +} + + + +static void +constant_interpolation( + struct spu_exec_machine *mach, + unsigned attrib, + unsigned chan ) +{ + unsigned i; + + for( i = 0; i < QUAD_SIZE; i++ ) { + mach->Inputs[attrib].xyzw[chan].f[i] = mach->InterpCoefs[attrib].a0[chan]; + } +} + +static void +linear_interpolation( + struct spu_exec_machine *mach, + unsigned attrib, + unsigned chan ) +{ + const float x = mach->QuadPos.xyzw[0].f[0]; + const float y = mach->QuadPos.xyzw[1].f[0]; + const float dadx = mach->InterpCoefs[attrib].dadx[chan]; + const float dady = mach->InterpCoefs[attrib].dady[chan]; + const float a0 = mach->InterpCoefs[attrib].a0[chan] + dadx * x + dady * y; + mach->Inputs[attrib].xyzw[chan].f[0] = a0; + mach->Inputs[attrib].xyzw[chan].f[1] = a0 + dadx; + mach->Inputs[attrib].xyzw[chan].f[2] = a0 + dady; + mach->Inputs[attrib].xyzw[chan].f[3] = a0 + dadx + dady; +} + +static void +perspective_interpolation( + struct spu_exec_machine *mach, + unsigned attrib, + unsigned chan ) +{ + const float x = mach->QuadPos.xyzw[0].f[0]; + const float y = mach->QuadPos.xyzw[1].f[0]; + const float dadx = mach->InterpCoefs[attrib].dadx[chan]; + const float dady = mach->InterpCoefs[attrib].dady[chan]; + const float a0 = mach->InterpCoefs[attrib].a0[chan] + dadx * x + dady * y; + const float *w = mach->QuadPos.xyzw[3].f; + /* divide by W here */ + mach->Inputs[attrib].xyzw[chan].f[0] = a0 / w[0]; + mach->Inputs[attrib].xyzw[chan].f[1] = (a0 + dadx) / w[1]; + mach->Inputs[attrib].xyzw[chan].f[2] = (a0 + dady) / w[2]; + mach->Inputs[attrib].xyzw[chan].f[3] = (a0 + dadx + dady) / w[3]; +} + + +typedef void (* interpolation_func)( + struct spu_exec_machine *mach, + unsigned attrib, + unsigned chan ); + +static void +exec_declaration(struct spu_exec_machine *mach, + const struct tgsi_full_declaration *decl) +{ + if( mach->Processor == TGSI_PROCESSOR_FRAGMENT ) { + if( decl->Declaration.File == TGSI_FILE_INPUT ) { + unsigned first, last, mask; + interpolation_func interp; + + assert( decl->Declaration.Declare == TGSI_DECLARE_RANGE ); + + first = decl->u.DeclarationRange.First; + last = decl->u.DeclarationRange.Last; + mask = decl->Declaration.UsageMask; + + switch( decl->Interpolation.Interpolate ) { + case TGSI_INTERPOLATE_CONSTANT: + interp = constant_interpolation; + break; + + case TGSI_INTERPOLATE_LINEAR: + interp = linear_interpolation; + break; + + case TGSI_INTERPOLATE_PERSPECTIVE: + interp = perspective_interpolation; + break; + + default: + assert( 0 ); + } + + if( mask == TGSI_WRITEMASK_XYZW ) { + unsigned i, j; + + for( i = first; i <= last; i++ ) { + for( j = 0; j < NUM_CHANNELS; j++ ) { + interp( mach, i, j ); + } + } + } + else { + unsigned i, j; + + for( j = 0; j < NUM_CHANNELS; j++ ) { + if( mask & (1 << j) ) { + for( i = first; i <= last; i++ ) { + interp( mach, i, j ); + } + } + } + } + } + } +} + +static void +exec_instruction( + struct spu_exec_machine *mach, + const struct tgsi_full_instruction *inst, + int *pc ) +{ + uint chan_index; + union spu_exec_channel r[8]; + + (*pc)++; + + switch (inst->Instruction.Opcode) { + case TGSI_OPCODE_ARL: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_f2it( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_MOV: + /* TGSI_OPCODE_SWZ */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_LIT: + if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X ); + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y ) || IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + FETCH( &r[0], 0, CHAN_X ); + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { + micro_max( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + STORE( &r[0], 0, CHAN_Y ); + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + FETCH( &r[1], 0, CHAN_Y ); + micro_max( &r[1], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + + FETCH( &r[2], 0, CHAN_W ); + micro_min( &r[2], &r[2], &mach->Temps[TEMP_128_I].xyzw[TEMP_128_C] ); + micro_max( &r[2], &r[2], &mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C] ); + micro_pow( &r[1], &r[1], &r[2] ); + micro_lt( &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + STORE( &r[0], 0, CHAN_Z ); + } + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W ); + } + break; + + case TGSI_OPCODE_RCP: + /* TGSI_OPCODE_RECIP */ + FETCH( &r[0], 0, CHAN_X ); + micro_div( &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &r[0] ); + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_RSQ: + /* TGSI_OPCODE_RECIPSQRT */ + FETCH( &r[0], 0, CHAN_X ); + micro_sqrt( &r[0], &r[0] ); + micro_div( &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &r[0] ); + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_EXP: + assert (0); + break; + + case TGSI_OPCODE_LOG: + assert (0); + break; + + case TGSI_OPCODE_MUL: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) + { + FETCH(&r[0], 0, chan_index); + FETCH(&r[1], 1, chan_index); + + micro_mul( &r[0], &r[0], &r[1] ); + + STORE(&r[0], 0, chan_index); + } + break; + + case TGSI_OPCODE_ADD: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_add( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_DP3: + /* TGSI_OPCODE_DOT3 */ + FETCH( &r[0], 0, CHAN_X ); + FETCH( &r[1], 1, CHAN_X ); + micro_mul( &r[0], &r[0], &r[1] ); + + FETCH( &r[1], 0, CHAN_Y ); + FETCH( &r[2], 1, CHAN_Y ); + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FETCH( &r[1], 0, CHAN_Z ); + FETCH( &r[2], 1, CHAN_Z ); + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_DP4: + /* TGSI_OPCODE_DOT4 */ + FETCH(&r[0], 0, CHAN_X); + FETCH(&r[1], 1, CHAN_X); + + micro_mul( &r[0], &r[0], &r[1] ); + + FETCH(&r[1], 0, CHAN_Y); + FETCH(&r[2], 1, CHAN_Y); + + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FETCH(&r[1], 0, CHAN_Z); + FETCH(&r[2], 1, CHAN_Z); + + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FETCH(&r[1], 0, CHAN_W); + FETCH(&r[2], 1, CHAN_W); + + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_DST: + if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X ); + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { + FETCH( &r[0], 0, CHAN_Y ); + FETCH( &r[1], 1, CHAN_Y); + micro_mul( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, CHAN_Y ); + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + FETCH( &r[0], 0, CHAN_Z ); + STORE( &r[0], 0, CHAN_Z ); + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) { + FETCH( &r[0], 1, CHAN_W ); + STORE( &r[0], 0, CHAN_W ); + } + break; + + case TGSI_OPCODE_MIN: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH(&r[0], 0, chan_index); + FETCH(&r[1], 1, chan_index); + + /* XXX use micro_min()?? */ + micro_lt( &r[0], &r[0], &r[1], &r[0], &r[1] ); + + STORE(&r[0], 0, chan_index); + } + break; + + case TGSI_OPCODE_MAX: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH(&r[0], 0, chan_index); + FETCH(&r[1], 1, chan_index); + + /* XXX use micro_max()?? */ + micro_lt( &r[0], &r[0], &r[1], &r[1], &r[0] ); + + STORE(&r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SLT: + /* TGSI_OPCODE_SETLT */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SGE: + /* TGSI_OPCODE_SETGE */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_ge( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_MAD: + /* TGSI_OPCODE_MADD */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_mul( &r[0], &r[0], &r[1] ); + FETCH( &r[1], 2, chan_index ); + micro_add( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SUB: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH(&r[0], 0, chan_index); + FETCH(&r[1], 1, chan_index); + + micro_sub( &r[0], &r[0], &r[1] ); + + STORE(&r[0], 0, chan_index); + } + break; + + case TGSI_OPCODE_LERP: + /* TGSI_OPCODE_LRP */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH(&r[0], 0, chan_index); + FETCH(&r[1], 1, chan_index); + FETCH(&r[2], 2, chan_index); + + micro_sub( &r[1], &r[1], &r[2] ); + micro_mul( &r[0], &r[0], &r[1] ); + micro_add( &r[0], &r[0], &r[2] ); + + STORE(&r[0], 0, chan_index); + } + break; + + case TGSI_OPCODE_CND: + assert (0); + break; + + case TGSI_OPCODE_CND0: + assert (0); + break; + + case TGSI_OPCODE_DOT2ADD: + /* TGSI_OPCODE_DP2A */ + assert (0); + break; + + case TGSI_OPCODE_INDEX: + assert (0); + break; + + case TGSI_OPCODE_NEGATE: + assert (0); + break; + + case TGSI_OPCODE_FRAC: + /* TGSI_OPCODE_FRC */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_frc( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_CLAMP: + assert (0); + break; + + case TGSI_OPCODE_FLOOR: + /* TGSI_OPCODE_FLR */ + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_flr( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_ROUND: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_rnd( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_EXPBASE2: + /* TGSI_OPCODE_EX2 */ + FETCH(&r[0], 0, CHAN_X); + + micro_pow( &r[0], &mach->Temps[TEMP_2_I].xyzw[TEMP_2_C], &r[0] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_LOGBASE2: + /* TGSI_OPCODE_LG2 */ + FETCH( &r[0], 0, CHAN_X ); + micro_lg2( &r[0], &r[0] ); + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_POWER: + /* TGSI_OPCODE_POW */ + FETCH(&r[0], 0, CHAN_X); + FETCH(&r[1], 1, CHAN_X); + + micro_pow( &r[0], &r[0], &r[1] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_CROSSPRODUCT: + /* TGSI_OPCODE_XPD */ + FETCH(&r[0], 0, CHAN_Y); + FETCH(&r[1], 1, CHAN_Z); + + micro_mul( &r[2], &r[0], &r[1] ); + + FETCH(&r[3], 0, CHAN_Z); + FETCH(&r[4], 1, CHAN_Y); + + micro_mul( &r[5], &r[3], &r[4] ); + micro_sub( &r[2], &r[2], &r[5] ); + + if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) { + STORE( &r[2], 0, CHAN_X ); + } + + FETCH(&r[2], 1, CHAN_X); + + micro_mul( &r[3], &r[3], &r[2] ); + + FETCH(&r[5], 0, CHAN_X); + + micro_mul( &r[1], &r[1], &r[5] ); + micro_sub( &r[3], &r[3], &r[1] ); + + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { + STORE( &r[3], 0, CHAN_Y ); + } + + micro_mul( &r[5], &r[5], &r[4] ); + micro_mul( &r[0], &r[0], &r[2] ); + micro_sub( &r[5], &r[5], &r[0] ); + + if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + STORE( &r[5], 0, CHAN_Z ); + } + + if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W ); + } + break; + + case TGSI_OPCODE_MULTIPLYMATRIX: + assert (0); + break; + + case TGSI_OPCODE_ABS: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH(&r[0], 0, chan_index); + + micro_abs( &r[0], &r[0] ); + + STORE(&r[0], 0, chan_index); + } + break; + + case TGSI_OPCODE_RCC: + assert (0); + break; + + case TGSI_OPCODE_DPH: + FETCH(&r[0], 0, CHAN_X); + FETCH(&r[1], 1, CHAN_X); + + micro_mul( &r[0], &r[0], &r[1] ); + + FETCH(&r[1], 0, CHAN_Y); + FETCH(&r[2], 1, CHAN_Y); + + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FETCH(&r[1], 0, CHAN_Z); + FETCH(&r[2], 1, CHAN_Z); + + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FETCH(&r[1], 1, CHAN_W); + + micro_add( &r[0], &r[0], &r[1] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_COS: + FETCH(&r[0], 0, CHAN_X); + + micro_cos( &r[0], &r[0] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_DDX: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_ddx( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_DDY: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_ddy( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_KILP: + exec_kilp (mach, inst); + break; + + case TGSI_OPCODE_KIL: + /* for enabled ExecMask bits, set the killed bit */ + mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] |= mach->ExecMask; + break; + + case TGSI_OPCODE_PK2H: + assert (0); + break; + + case TGSI_OPCODE_PK2US: + assert (0); + break; + + case TGSI_OPCODE_PK4B: + assert (0); + break; + + case TGSI_OPCODE_PK4UB: + assert (0); + break; + + case TGSI_OPCODE_RFL: + assert (0); + break; + + case TGSI_OPCODE_SEQ: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_eq( &r[0], &r[0], &r[1], + &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], + &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SFL: + assert (0); + break; + + case TGSI_OPCODE_SGT: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SIN: + FETCH( &r[0], 0, CHAN_X ); + micro_sin( &r[0], &r[0] ); + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SLE: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_ge( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SNE: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_eq( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_STR: + assert (0); + break; + + case TGSI_OPCODE_TEX: + /* simple texture lookup */ + /* src[0] = texcoord */ + /* src[1] = sampler unit */ + exec_tex(mach, inst, FALSE); + break; + + case TGSI_OPCODE_TXB: + /* Texture lookup with lod bias */ + /* src[0] = texcoord (src[0].w = load bias) */ + /* src[1] = sampler unit */ + exec_tex(mach, inst, TRUE); + break; + + case TGSI_OPCODE_TXD: + /* Texture lookup with explict partial derivatives */ + /* src[0] = texcoord */ + /* src[1] = d[strq]/dx */ + /* src[2] = d[strq]/dy */ + /* src[3] = sampler unit */ + assert (0); + break; + + case TGSI_OPCODE_TXL: + /* Texture lookup with explit LOD */ + /* src[0] = texcoord (src[0].w = load bias) */ + /* src[1] = sampler unit */ + exec_tex(mach, inst, TRUE); + break; + + case TGSI_OPCODE_UP2H: + assert (0); + break; + + case TGSI_OPCODE_UP2US: + assert (0); + break; + + case TGSI_OPCODE_UP4B: + assert (0); + break; + + case TGSI_OPCODE_UP4UB: + assert (0); + break; + + case TGSI_OPCODE_X2D: + assert (0); + break; + + case TGSI_OPCODE_ARA: + assert (0); + break; + + case TGSI_OPCODE_ARR: + assert (0); + break; + + case TGSI_OPCODE_BRA: + assert (0); + break; + + case TGSI_OPCODE_CAL: + /* skip the call if no execution channels are enabled */ + if (mach->ExecMask) { + /* do the call */ + + /* push the Cond, Loop, Cont stacks */ + assert(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING); + mach->CondStack[mach->CondStackTop++] = mach->CondMask; + assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING); + mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask; + assert(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING); + mach->ContStack[mach->ContStackTop++] = mach->ContMask; + + assert(mach->FuncStackTop < TGSI_EXEC_MAX_CALL_NESTING); + mach->FuncStack[mach->FuncStackTop++] = mach->FuncMask; + + /* note that PC was already incremented above */ + mach->CallStack[mach->CallStackTop++] = *pc; + *pc = inst->InstructionExtLabel.Label; + } + break; + + case TGSI_OPCODE_RET: + mach->FuncMask &= ~mach->ExecMask; + UPDATE_EXEC_MASK(mach); + + if (mach->ExecMask == 0x0) { + /* really return now (otherwise, keep executing */ + + if (mach->CallStackTop == 0) { + /* returning from main() */ + *pc = -1; + return; + } + *pc = mach->CallStack[--mach->CallStackTop]; + + /* pop the Cond, Loop, Cont stacks */ + assert(mach->CondStackTop > 0); + mach->CondMask = mach->CondStack[--mach->CondStackTop]; + assert(mach->LoopStackTop > 0); + mach->LoopMask = mach->LoopStack[--mach->LoopStackTop]; + assert(mach->ContStackTop > 0); + mach->ContMask = mach->ContStack[--mach->ContStackTop]; + assert(mach->FuncStackTop > 0); + mach->FuncMask = mach->FuncStack[--mach->FuncStackTop]; + + UPDATE_EXEC_MASK(mach); + } + break; + + case TGSI_OPCODE_SSG: + assert (0); + break; + + case TGSI_OPCODE_CMP: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH(&r[0], 0, chan_index); + FETCH(&r[1], 1, chan_index); + FETCH(&r[2], 2, chan_index); + + micro_lt( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[1], &r[2] ); + + STORE(&r[0], 0, chan_index); + } + break; + + case TGSI_OPCODE_SCS: + if( IS_CHANNEL_ENABLED( *inst, CHAN_X ) || IS_CHANNEL_ENABLED( *inst, CHAN_Y ) ) { + FETCH( &r[0], 0, CHAN_X ); + } + if( IS_CHANNEL_ENABLED( *inst, CHAN_X ) ) { + micro_cos( &r[1], &r[0] ); + STORE( &r[1], 0, CHAN_X ); + } + if( IS_CHANNEL_ENABLED( *inst, CHAN_Y ) ) { + micro_sin( &r[1], &r[0] ); + STORE( &r[1], 0, CHAN_Y ); + } + if( IS_CHANNEL_ENABLED( *inst, CHAN_Z ) ) { + STORE( &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], 0, CHAN_Z ); + } + if( IS_CHANNEL_ENABLED( *inst, CHAN_W ) ) { + STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W ); + } + break; + + case TGSI_OPCODE_NRM: + assert (0); + break; + + case TGSI_OPCODE_DIV: + assert( 0 ); + break; + + case TGSI_OPCODE_DP2: + FETCH( &r[0], 0, CHAN_X ); + FETCH( &r[1], 1, CHAN_X ); + micro_mul( &r[0], &r[0], &r[1] ); + + FETCH( &r[1], 0, CHAN_Y ); + FETCH( &r[2], 1, CHAN_Y ); + micro_mul( &r[1], &r[1], &r[2] ); + micro_add( &r[0], &r[0], &r[1] ); + + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_IF: + /* push CondMask */ + assert(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING); + mach->CondStack[mach->CondStackTop++] = mach->CondMask; + FETCH( &r[0], 0, CHAN_X ); + /* update CondMask */ + if( ! r[0].u[0] ) { + mach->CondMask &= ~0x1; + } + if( ! r[0].u[1] ) { + mach->CondMask &= ~0x2; + } + if( ! r[0].u[2] ) { + mach->CondMask &= ~0x4; + } + if( ! r[0].u[3] ) { + mach->CondMask &= ~0x8; + } + UPDATE_EXEC_MASK(mach); + /* Todo: If CondMask==0, jump to ELSE */ + break; + + case TGSI_OPCODE_ELSE: + /* invert CondMask wrt previous mask */ + { + uint prevMask; + assert(mach->CondStackTop > 0); + prevMask = mach->CondStack[mach->CondStackTop - 1]; + mach->CondMask = ~mach->CondMask & prevMask; + UPDATE_EXEC_MASK(mach); + /* Todo: If CondMask==0, jump to ENDIF */ + } + break; + + case TGSI_OPCODE_ENDIF: + /* pop CondMask */ + assert(mach->CondStackTop > 0); + mach->CondMask = mach->CondStack[--mach->CondStackTop]; + UPDATE_EXEC_MASK(mach); + break; + + case TGSI_OPCODE_END: + /* halt execution */ + *pc = -1; + break; + + case TGSI_OPCODE_REP: + assert (0); + break; + + case TGSI_OPCODE_ENDREP: + assert (0); + break; + + case TGSI_OPCODE_PUSHA: + assert (0); + break; + + case TGSI_OPCODE_POPA: + assert (0); + break; + + case TGSI_OPCODE_CEIL: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_ceil( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_I2F: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_i2f( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_NOT: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_not( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_TRUNC: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + micro_trunc( &r[0], &r[0] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SHL: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_shl( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SHR: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_ishr( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_AND: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_and( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_OR: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_or( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_MOD: + assert (0); + break; + + case TGSI_OPCODE_XOR: + FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { + FETCH( &r[0], 0, chan_index ); + FETCH( &r[1], 1, chan_index ); + micro_xor( &r[0], &r[0], &r[1] ); + STORE( &r[0], 0, chan_index ); + } + break; + + case TGSI_OPCODE_SAD: + assert (0); + break; + + case TGSI_OPCODE_TXF: + assert (0); + break; + + case TGSI_OPCODE_TXQ: + assert (0); + break; + + case TGSI_OPCODE_EMIT: + mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] += 16; + mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]]++; + break; + + case TGSI_OPCODE_ENDPRIM: + mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]++; + mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0; + break; + + case TGSI_OPCODE_LOOP: + /* fall-through (for now) */ + case TGSI_OPCODE_BGNLOOP2: + /* push LoopMask and ContMasks */ + assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING); + mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask; + assert(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING); + mach->ContStack[mach->ContStackTop++] = mach->ContMask; + break; + + case TGSI_OPCODE_ENDLOOP: + /* fall-through (for now at least) */ + case TGSI_OPCODE_ENDLOOP2: + /* Restore ContMask, but don't pop */ + assert(mach->ContStackTop > 0); + mach->ContMask = mach->ContStack[mach->ContStackTop - 1]; + if (mach->LoopMask) { + /* repeat loop: jump to instruction just past BGNLOOP */ + *pc = inst->InstructionExtLabel.Label + 1; + } + else { + /* exit loop: pop LoopMask */ + assert(mach->LoopStackTop > 0); + mach->LoopMask = mach->LoopStack[--mach->LoopStackTop]; + /* pop ContMask */ + assert(mach->ContStackTop > 0); + mach->ContMask = mach->ContStack[--mach->ContStackTop]; + } + UPDATE_EXEC_MASK(mach); + break; + + case TGSI_OPCODE_BRK: + /* turn off loop channels for each enabled exec channel */ + mach->LoopMask &= ~mach->ExecMask; + /* Todo: if mach->LoopMask == 0, jump to end of loop */ + UPDATE_EXEC_MASK(mach); + break; + + case TGSI_OPCODE_CONT: + /* turn off cont channels for each enabled exec channel */ + mach->ContMask &= ~mach->ExecMask; + /* Todo: if mach->LoopMask == 0, jump to end of loop */ + UPDATE_EXEC_MASK(mach); + break; + + case TGSI_OPCODE_BGNSUB: + /* no-op */ + break; + + case TGSI_OPCODE_ENDSUB: + /* no-op */ + break; + + case TGSI_OPCODE_NOISE1: + assert( 0 ); + break; + + case TGSI_OPCODE_NOISE2: + assert( 0 ); + break; + + case TGSI_OPCODE_NOISE3: + assert( 0 ); + break; + + case TGSI_OPCODE_NOISE4: + assert( 0 ); + break; + + case TGSI_OPCODE_NOP: + break; + + default: + assert( 0 ); + } +} + + +/** + * Run TGSI interpreter. + * \return bitmask of "alive" quad components + */ +uint +spu_exec_machine_run( struct spu_exec_machine *mach ) +{ + uint i; + int pc = 0; + + mach->CondMask = 0xf; + mach->LoopMask = 0xf; + mach->ContMask = 0xf; + mach->FuncMask = 0xf; + mach->ExecMask = 0xf; + + mach->CondStackTop = 0; /* temporarily subvert this assertion */ + assert(mach->CondStackTop == 0); + assert(mach->LoopStackTop == 0); + assert(mach->ContStackTop == 0); + assert(mach->CallStackTop == 0); + + mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] = 0; + mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] = 0; + + if( mach->Processor == TGSI_PROCESSOR_GEOMETRY ) { + mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0] = 0; + mach->Primitives[0] = 0; + } + + + /* execute declarations (interpolants) */ + for (i = 0; i < mach->NumDeclarations; i++) { + exec_declaration( mach, mach->Declarations+i ); + } + + /* execute instructions, until pc is set to -1 */ + while (pc != -1) { + assert(pc < mach->NumInstructions); + exec_instruction( mach, mach->Instructions + pc, &pc ); + } + +#if 0 + /* we scale from floats in [0,1] to Zbuffer ints in sp_quad_depth_test.c */ + if (mach->Processor == TGSI_PROCESSOR_FRAGMENT) { + /* + * Scale back depth component. + */ + for (i = 0; i < 4; i++) + mach->Outputs[0].xyzw[2].f[i] *= ctx->DrawBuffer->_DepthMaxF; + } +#endif + + return ~mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0]; +} + + diff --git a/src/mesa/pipe/cell/spu/spu_exec.h b/src/mesa/pipe/cell/spu/spu_exec.h new file mode 100644 index 00000000000..89e422ba48d --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_exec.h @@ -0,0 +1,171 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#if !defined SPU_EXEC_H +#define SPU_EXEC_H + +#include "pipe/p_compiler.h" +#include "pipe/tgsi/exec/tgsi_exec.h" + +#if defined __cplusplus +extern "C" { +#endif + +/** + * Registers may be treated as float, signed int or unsigned int. + */ +union spu_exec_channel +{ + float f[QUAD_SIZE]; + int i[QUAD_SIZE]; + unsigned u[QUAD_SIZE]; +}; + +/** + * A vector[RGBA] of channels[4 pixels] + */ +struct spu_exec_vector +{ + union spu_exec_channel xyzw[NUM_CHANNELS]; +}; + +/** + * For fragment programs, information for computing fragment input + * values from plane equation of the triangle/line. + */ +struct spu_interp_coef +{ + float a0[NUM_CHANNELS]; /* in an xyzw layout */ + float dadx[NUM_CHANNELS]; + float dady[NUM_CHANNELS]; +}; + + +struct softpipe_tile_cache; /**< Opaque to TGSI */ + +/** + * Information for sampling textures, which must be implemented + * by code outside the TGSI executor. + */ +struct spu_sampler +{ + const struct pipe_sampler_state *state; + struct pipe_texture *texture; + /** Get samples for four fragments in a quad */ + void (*get_samples)(struct spu_sampler *sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]); + void *pipe; /*XXX temporary*/ + struct softpipe_tile_cache *cache; +}; + + +/** + * Run-time virtual machine state for executing TGSI shader. + */ +struct spu_exec_machine +{ + /* + * 32 program temporaries + * 4 internal temporaries + * 1 address + */ + struct spu_exec_vector Temps[TGSI_EXEC_NUM_TEMPS + + TGSI_EXEC_NUM_ADDRS + 1] + ALIGN16_ATTRIB; + + struct spu_exec_vector *Addrs; + + struct spu_sampler *Samplers; + + float Imms[TGSI_EXEC_NUM_IMMEDIATES][4]; + unsigned ImmLimit; + float (*Consts)[4]; + struct spu_exec_vector *Inputs; + struct spu_exec_vector *Outputs; + unsigned Processor; + + /* GEOMETRY processor only. */ + unsigned *Primitives; + + /* FRAGMENT processor only. */ + const struct spu_interp_coef *InterpCoefs; + struct spu_exec_vector QuadPos; + + /* Conditional execution masks */ + uint CondMask; /**< For IF/ELSE/ENDIF */ + uint LoopMask; /**< For BGNLOOP/ENDLOOP */ + uint ContMask; /**< For loop CONT statements */ + uint FuncMask; /**< For function calls */ + uint ExecMask; /**< = CondMask & LoopMask */ + + /** Condition mask stack (for nested conditionals) */ + uint CondStack[TGSI_EXEC_MAX_COND_NESTING]; + int CondStackTop; + + /** Loop mask stack (for nested loops) */ + uint LoopStack[TGSI_EXEC_MAX_LOOP_NESTING]; + int LoopStackTop; + + /** Loop continue mask stack (see comments in tgsi_exec.c) */ + uint ContStack[TGSI_EXEC_MAX_LOOP_NESTING]; + int ContStackTop; + + /** Function execution mask stack (for executing subroutine code) */ + uint FuncStack[TGSI_EXEC_MAX_CALL_NESTING]; + int FuncStackTop; + + /** Function call stack for saving/restoring the program counter */ + uint CallStack[TGSI_EXEC_MAX_CALL_NESTING]; + int CallStackTop; + + struct tgsi_full_instruction *Instructions; + uint NumInstructions; + + struct tgsi_full_declaration *Declarations; + uint NumDeclarations; +}; + + +extern void +spu_exec_machine_init(struct spu_exec_machine *mach, + uint numSamplers, + struct spu_sampler *samplers, + unsigned processor); + +extern uint +spu_exec_machine_run( struct spu_exec_machine *mach ); + + +#if defined __cplusplus +} /* extern "C" */ +#endif + +#endif /* SPU_EXEC_H */ diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 6886f283bec..9daa3ec735a 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -36,6 +36,7 @@ #include "spu_render.h" #include "spu_texture.h" #include "spu_tile.h" +#include "spu_vertex_shader.h" #include "pipe/cell/common.h" #include "pipe/p_defines.h" @@ -50,6 +51,7 @@ boolean Debug = FALSE; struct spu_global spu; +struct spu_vs_context draw; /** * Tell the PPU that this SPU has finished copying a buffer to @@ -264,6 +266,18 @@ cmd_state_vertex_info(const struct vertex_info *vinfo) } +static void +cmd_state_vs_array_info(const struct cell_array_info *vs_info) +{ + const unsigned attr = vs_info->attr; + + ASSERT(attr < PIPE_ATTRIB_MAX); + draw.vertex_fetch.src_ptr[attr] = vs_info->base; + draw.vertex_fetch.pitch[attr] = vs_info->pitch; + draw.vertex_fetch.format[attr] = vs_info->format; + draw.vertex_fetch.dirty = 1; +} + static void cmd_finish(void) @@ -374,6 +388,20 @@ cmd_batch(uint opcode) cmd_state_vertex_info((struct vertex_info *) &buffer[pos+1]); pos += (1 + sizeof(struct vertex_info) / 4); break; + case CELL_CMD_STATE_VIEWPORT: + (void) memcpy(& draw.viewport, &buffer[pos+1], + sizeof(struct pipe_viewport_state)); + pos += (1 + sizeof(struct pipe_viewport_state) / 4); + break; + case CELL_CMD_STATE_VS_ARRAY_INFO: + cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]); + pos += (1 + sizeof(struct cell_array_info) / 4); + break; + case CELL_CMD_VS_EXECUTE: + spu_execute_vertex_shader(&draw, + (struct cell_command_vs *) &buffer[pos+1]); + pos += (1 + sizeof(struct cell_command_vs) / 4); + break; default: printf("SPU %u: bad opcode: 0x%x\n", spu.init.id, buffer[pos]); ASSERT(0); diff --git a/src/mesa/pipe/cell/spu/spu_util.c b/src/mesa/pipe/cell/spu/spu_util.c new file mode 100644 index 00000000000..ac373240c1f --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_util.c @@ -0,0 +1,165 @@ +#include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" +#include "pipe/tgsi/util/tgsi_parse.h" +//#include "tgsi_build.h" +#include "pipe/tgsi/util/tgsi_util.h" + +unsigned +tgsi_util_get_src_register_swizzle( + const struct tgsi_src_register *reg, + unsigned component ) +{ + switch( component ) { + case 0: + return reg->SwizzleX; + case 1: + return reg->SwizzleY; + case 2: + return reg->SwizzleZ; + case 3: + return reg->SwizzleW; + default: + assert( 0 ); + } + return 0; +} + +unsigned +tgsi_util_get_src_register_extswizzle( + const struct tgsi_src_register_ext_swz *reg, + unsigned component ) +{ + switch( component ) { + case 0: + return reg->ExtSwizzleX; + case 1: + return reg->ExtSwizzleY; + case 2: + return reg->ExtSwizzleZ; + case 3: + return reg->ExtSwizzleW; + default: + assert( 0 ); + } + return 0; +} + +unsigned +tgsi_util_get_full_src_register_extswizzle( + const struct tgsi_full_src_register *reg, + unsigned component ) +{ + unsigned swizzle; + + /* + * First, calculate the extended swizzle for a given channel. This will give + * us either a channel index into the simple swizzle or a constant 1 or 0. + */ + swizzle = tgsi_util_get_src_register_extswizzle( + ®->SrcRegisterExtSwz, + component ); + + assert (TGSI_SWIZZLE_X == TGSI_EXTSWIZZLE_X); + assert (TGSI_SWIZZLE_Y == TGSI_EXTSWIZZLE_Y); + assert (TGSI_SWIZZLE_Z == TGSI_EXTSWIZZLE_Z); + assert (TGSI_SWIZZLE_W == TGSI_EXTSWIZZLE_W); + assert (TGSI_EXTSWIZZLE_ZERO > TGSI_SWIZZLE_W); + assert (TGSI_EXTSWIZZLE_ONE > TGSI_SWIZZLE_W); + + /* + * Second, calculate the simple swizzle for the unswizzled channel index. + * Leave the constants intact, they are not affected by the simple swizzle. + */ + if( swizzle <= TGSI_SWIZZLE_W ) { + swizzle = tgsi_util_get_src_register_swizzle( + ®->SrcRegister, + component ); + } + + return swizzle; +} + +unsigned +tgsi_util_get_src_register_extnegate( + const struct tgsi_src_register_ext_swz *reg, + unsigned component ) +{ + switch( component ) { + case 0: + return reg->NegateX; + case 1: + return reg->NegateY; + case 2: + return reg->NegateZ; + case 3: + return reg->NegateW; + default: + assert( 0 ); + } + return 0; +} + +void +tgsi_util_set_src_register_extnegate( + struct tgsi_src_register_ext_swz *reg, + unsigned negate, + unsigned component ) +{ + switch( component ) { + case 0: + reg->NegateX = negate; + break; + case 1: + reg->NegateY = negate; + break; + case 2: + reg->NegateZ = negate; + break; + case 3: + reg->NegateW = negate; + break; + default: + assert( 0 ); + } +} + +unsigned +tgsi_util_get_full_src_register_sign_mode( + const struct tgsi_full_src_register *reg, + unsigned component ) +{ + unsigned sign_mode; + + if( reg->SrcRegisterExtMod.Absolute ) { + /* Consider only the post-abs negation. */ + + if( reg->SrcRegisterExtMod.Negate ) { + sign_mode = TGSI_UTIL_SIGN_SET; + } + else { + sign_mode = TGSI_UTIL_SIGN_CLEAR; + } + } + else { + /* Accumulate the three negations. */ + + unsigned negate; + + negate = reg->SrcRegister.Negate; + if( tgsi_util_get_src_register_extnegate( ®->SrcRegisterExtSwz, component ) ) { + negate = !negate; + } + if( reg->SrcRegisterExtMod.Negate ) { + negate = !negate; + } + + if( negate ) { + sign_mode = TGSI_UTIL_SIGN_TOGGLE; + } + else { + sign_mode = TGSI_UTIL_SIGN_KEEP; + } + } + + return sign_mode; +} diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c new file mode 100644 index 00000000000..b8f8c52eed3 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -0,0 +1,493 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell + */ + +#include "pipe/p_util.h" +#include "pipe/p_state.h" +#include "pipe/p_shader_tokens.h" +#include "spu_exec.h" +#include "spu_vertex_shader.h" + + +#define DRAW_DBG 0 + + +/** + * Fetch a float[4] vertex attribute from memory, doing format/type + * conversion as needed. + * + * This is probably needed/dupliocated elsewhere, eg format + * conversion, texture sampling etc. + */ +#define FETCH_ATTRIB( NAME, SZ, CVT ) \ +static void \ +fetch_##NAME(const void *ptr, float *attrib) \ +{ \ + static const float defaults[4] = { 0,0,0,1 }; \ + int i; \ + \ + for (i = 0; i < SZ; i++) { \ + attrib[i] = CVT; \ + } \ + \ + for (; i < 4; i++) { \ + attrib[i] = defaults[i]; \ + } \ +} + +#define CVT_64_FLOAT (float) ((double *) ptr)[i] +#define CVT_32_FLOAT ((float *) ptr)[i] + +#define CVT_8_USCALED (float) ((unsigned char *) ptr)[i] +#define CVT_16_USCALED (float) ((unsigned short *) ptr)[i] +#define CVT_32_USCALED (float) ((unsigned int *) ptr)[i] + +#define CVT_8_SSCALED (float) ((char *) ptr)[i] +#define CVT_16_SSCALED (float) ((short *) ptr)[i] +#define CVT_32_SSCALED (float) ((int *) ptr)[i] + +#define CVT_8_UNORM (float) ((unsigned char *) ptr)[i] / 255.0f +#define CVT_16_UNORM (float) ((unsigned short *) ptr)[i] / 65535.0f +#define CVT_32_UNORM (float) ((unsigned int *) ptr)[i] / 4294967295.0f + +#define CVT_8_SNORM (float) ((char *) ptr)[i] / 127.0f +#define CVT_16_SNORM (float) ((short *) ptr)[i] / 32767.0f +#define CVT_32_SNORM (float) ((int *) ptr)[i] / 2147483647.0f + +FETCH_ATTRIB( R64G64B64A64_FLOAT, 4, CVT_64_FLOAT ) +FETCH_ATTRIB( R64G64B64_FLOAT, 3, CVT_64_FLOAT ) +FETCH_ATTRIB( R64G64_FLOAT, 2, CVT_64_FLOAT ) +FETCH_ATTRIB( R64_FLOAT, 1, CVT_64_FLOAT ) + +FETCH_ATTRIB( R32G32B32A32_FLOAT, 4, CVT_32_FLOAT ) +FETCH_ATTRIB( R32G32B32_FLOAT, 3, CVT_32_FLOAT ) +FETCH_ATTRIB( R32G32_FLOAT, 2, CVT_32_FLOAT ) +FETCH_ATTRIB( R32_FLOAT, 1, CVT_32_FLOAT ) + +FETCH_ATTRIB( R32G32B32A32_USCALED, 4, CVT_32_USCALED ) +FETCH_ATTRIB( R32G32B32_USCALED, 3, CVT_32_USCALED ) +FETCH_ATTRIB( R32G32_USCALED, 2, CVT_32_USCALED ) +FETCH_ATTRIB( R32_USCALED, 1, CVT_32_USCALED ) + +FETCH_ATTRIB( R32G32B32A32_SSCALED, 4, CVT_32_SSCALED ) +FETCH_ATTRIB( R32G32B32_SSCALED, 3, CVT_32_SSCALED ) +FETCH_ATTRIB( R32G32_SSCALED, 2, CVT_32_SSCALED ) +FETCH_ATTRIB( R32_SSCALED, 1, CVT_32_SSCALED ) + +FETCH_ATTRIB( R32G32B32A32_UNORM, 4, CVT_32_UNORM ) +FETCH_ATTRIB( R32G32B32_UNORM, 3, CVT_32_UNORM ) +FETCH_ATTRIB( R32G32_UNORM, 2, CVT_32_UNORM ) +FETCH_ATTRIB( R32_UNORM, 1, CVT_32_UNORM ) + +FETCH_ATTRIB( R32G32B32A32_SNORM, 4, CVT_32_SNORM ) +FETCH_ATTRIB( R32G32B32_SNORM, 3, CVT_32_SNORM ) +FETCH_ATTRIB( R32G32_SNORM, 2, CVT_32_SNORM ) +FETCH_ATTRIB( R32_SNORM, 1, CVT_32_SNORM ) + +FETCH_ATTRIB( R16G16B16A16_USCALED, 4, CVT_16_USCALED ) +FETCH_ATTRIB( R16G16B16_USCALED, 3, CVT_16_USCALED ) +FETCH_ATTRIB( R16G16_USCALED, 2, CVT_16_USCALED ) +FETCH_ATTRIB( R16_USCALED, 1, CVT_16_USCALED ) + +FETCH_ATTRIB( R16G16B16A16_SSCALED, 4, CVT_16_SSCALED ) +FETCH_ATTRIB( R16G16B16_SSCALED, 3, CVT_16_SSCALED ) +FETCH_ATTRIB( R16G16_SSCALED, 2, CVT_16_SSCALED ) +FETCH_ATTRIB( R16_SSCALED, 1, CVT_16_SSCALED ) + +FETCH_ATTRIB( R16G16B16A16_UNORM, 4, CVT_16_UNORM ) +FETCH_ATTRIB( R16G16B16_UNORM, 3, CVT_16_UNORM ) +FETCH_ATTRIB( R16G16_UNORM, 2, CVT_16_UNORM ) +FETCH_ATTRIB( R16_UNORM, 1, CVT_16_UNORM ) + +FETCH_ATTRIB( R16G16B16A16_SNORM, 4, CVT_16_SNORM ) +FETCH_ATTRIB( R16G16B16_SNORM, 3, CVT_16_SNORM ) +FETCH_ATTRIB( R16G16_SNORM, 2, CVT_16_SNORM ) +FETCH_ATTRIB( R16_SNORM, 1, CVT_16_SNORM ) + +FETCH_ATTRIB( R8G8B8A8_USCALED, 4, CVT_8_USCALED ) +FETCH_ATTRIB( R8G8B8_USCALED, 3, CVT_8_USCALED ) +FETCH_ATTRIB( R8G8_USCALED, 2, CVT_8_USCALED ) +FETCH_ATTRIB( R8_USCALED, 1, CVT_8_USCALED ) + +FETCH_ATTRIB( R8G8B8A8_SSCALED, 4, CVT_8_SSCALED ) +FETCH_ATTRIB( R8G8B8_SSCALED, 3, CVT_8_SSCALED ) +FETCH_ATTRIB( R8G8_SSCALED, 2, CVT_8_SSCALED ) +FETCH_ATTRIB( R8_SSCALED, 1, CVT_8_SSCALED ) + +FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) +FETCH_ATTRIB( R8G8B8_UNORM, 3, CVT_8_UNORM ) +FETCH_ATTRIB( R8G8_UNORM, 2, CVT_8_UNORM ) +FETCH_ATTRIB( R8_UNORM, 1, CVT_8_UNORM ) + +FETCH_ATTRIB( R8G8B8A8_SNORM, 4, CVT_8_SNORM ) +FETCH_ATTRIB( R8G8B8_SNORM, 3, CVT_8_SNORM ) +FETCH_ATTRIB( R8G8_SNORM, 2, CVT_8_SNORM ) +FETCH_ATTRIB( R8_SNORM, 1, CVT_8_SNORM ) + +FETCH_ATTRIB( A8R8G8B8_UNORM, 4, CVT_8_UNORM ) +//FETCH_ATTRIB( R8G8B8A8_UNORM, 4, CVT_8_UNORM ) + + + +static spu_fetch_func get_fetch_func( enum pipe_format format ) +{ +#if 0 + { + char tmp[80]; + pf_sprint_name(tmp, format); + _mesa_printf("%s: %s\n", __FUNCTION__, tmp); + } +#endif + + switch (format) { + case PIPE_FORMAT_R64_FLOAT: + return fetch_R64_FLOAT; + case PIPE_FORMAT_R64G64_FLOAT: + return fetch_R64G64_FLOAT; + case PIPE_FORMAT_R64G64B64_FLOAT: + return fetch_R64G64B64_FLOAT; + case PIPE_FORMAT_R64G64B64A64_FLOAT: + return fetch_R64G64B64A64_FLOAT; + + case PIPE_FORMAT_R32_FLOAT: + return fetch_R32_FLOAT; + case PIPE_FORMAT_R32G32_FLOAT: + return fetch_R32G32_FLOAT; + case PIPE_FORMAT_R32G32B32_FLOAT: + return fetch_R32G32B32_FLOAT; + case PIPE_FORMAT_R32G32B32A32_FLOAT: + return fetch_R32G32B32A32_FLOAT; + + case PIPE_FORMAT_R32_UNORM: + return fetch_R32_UNORM; + case PIPE_FORMAT_R32G32_UNORM: + return fetch_R32G32_UNORM; + case PIPE_FORMAT_R32G32B32_UNORM: + return fetch_R32G32B32_UNORM; + case PIPE_FORMAT_R32G32B32A32_UNORM: + return fetch_R32G32B32A32_UNORM; + + case PIPE_FORMAT_R32_USCALED: + return fetch_R32_USCALED; + case PIPE_FORMAT_R32G32_USCALED: + return fetch_R32G32_USCALED; + case PIPE_FORMAT_R32G32B32_USCALED: + return fetch_R32G32B32_USCALED; + case PIPE_FORMAT_R32G32B32A32_USCALED: + return fetch_R32G32B32A32_USCALED; + + case PIPE_FORMAT_R32_SNORM: + return fetch_R32_SNORM; + case PIPE_FORMAT_R32G32_SNORM: + return fetch_R32G32_SNORM; + case PIPE_FORMAT_R32G32B32_SNORM: + return fetch_R32G32B32_SNORM; + case PIPE_FORMAT_R32G32B32A32_SNORM: + return fetch_R32G32B32A32_SNORM; + + case PIPE_FORMAT_R32_SSCALED: + return fetch_R32_SSCALED; + case PIPE_FORMAT_R32G32_SSCALED: + return fetch_R32G32_SSCALED; + case PIPE_FORMAT_R32G32B32_SSCALED: + return fetch_R32G32B32_SSCALED; + case PIPE_FORMAT_R32G32B32A32_SSCALED: + return fetch_R32G32B32A32_SSCALED; + + case PIPE_FORMAT_R16_UNORM: + return fetch_R16_UNORM; + case PIPE_FORMAT_R16G16_UNORM: + return fetch_R16G16_UNORM; + case PIPE_FORMAT_R16G16B16_UNORM: + return fetch_R16G16B16_UNORM; + case PIPE_FORMAT_R16G16B16A16_UNORM: + return fetch_R16G16B16A16_UNORM; + + case PIPE_FORMAT_R16_USCALED: + return fetch_R16_USCALED; + case PIPE_FORMAT_R16G16_USCALED: + return fetch_R16G16_USCALED; + case PIPE_FORMAT_R16G16B16_USCALED: + return fetch_R16G16B16_USCALED; + case PIPE_FORMAT_R16G16B16A16_USCALED: + return fetch_R16G16B16A16_USCALED; + + case PIPE_FORMAT_R16_SNORM: + return fetch_R16_SNORM; + case PIPE_FORMAT_R16G16_SNORM: + return fetch_R16G16_SNORM; + case PIPE_FORMAT_R16G16B16_SNORM: + return fetch_R16G16B16_SNORM; + case PIPE_FORMAT_R16G16B16A16_SNORM: + return fetch_R16G16B16A16_SNORM; + + case PIPE_FORMAT_R16_SSCALED: + return fetch_R16_SSCALED; + case PIPE_FORMAT_R16G16_SSCALED: + return fetch_R16G16_SSCALED; + case PIPE_FORMAT_R16G16B16_SSCALED: + return fetch_R16G16B16_SSCALED; + case PIPE_FORMAT_R16G16B16A16_SSCALED: + return fetch_R16G16B16A16_SSCALED; + + case PIPE_FORMAT_R8_UNORM: + return fetch_R8_UNORM; + case PIPE_FORMAT_R8G8_UNORM: + return fetch_R8G8_UNORM; + case PIPE_FORMAT_R8G8B8_UNORM: + return fetch_R8G8B8_UNORM; + case PIPE_FORMAT_R8G8B8A8_UNORM: + return fetch_R8G8B8A8_UNORM; + + case PIPE_FORMAT_R8_USCALED: + return fetch_R8_USCALED; + case PIPE_FORMAT_R8G8_USCALED: + return fetch_R8G8_USCALED; + case PIPE_FORMAT_R8G8B8_USCALED: + return fetch_R8G8B8_USCALED; + case PIPE_FORMAT_R8G8B8A8_USCALED: + return fetch_R8G8B8A8_USCALED; + + case PIPE_FORMAT_R8_SNORM: + return fetch_R8_SNORM; + case PIPE_FORMAT_R8G8_SNORM: + return fetch_R8G8_SNORM; + case PIPE_FORMAT_R8G8B8_SNORM: + return fetch_R8G8B8_SNORM; + case PIPE_FORMAT_R8G8B8A8_SNORM: + return fetch_R8G8B8A8_SNORM; + + case PIPE_FORMAT_R8_SSCALED: + return fetch_R8_SSCALED; + case PIPE_FORMAT_R8G8_SSCALED: + return fetch_R8G8_SSCALED; + case PIPE_FORMAT_R8G8B8_SSCALED: + return fetch_R8G8B8_SSCALED; + case PIPE_FORMAT_R8G8B8A8_SSCALED: + return fetch_R8G8B8A8_SSCALED; + + case PIPE_FORMAT_A8R8G8B8_UNORM: + return fetch_A8R8G8B8_UNORM; + + case 0: + return NULL; /* not sure why this is needed */ + + default: + assert(0); + return NULL; + } +} + + +static void +transpose_4x4( float *out, const float *in ) +{ + /* This can be achieved in 12 sse instructions, plus the final + * stores I guess. This is probably a bit more than that - maybe + * 32 or so? + */ + out[0] = in[0]; out[1] = in[4]; out[2] = in[8]; out[3] = in[12]; + out[4] = in[1]; out[5] = in[5]; out[6] = in[9]; out[7] = in[13]; + out[8] = in[2]; out[9] = in[6]; out[10] = in[10]; out[11] = in[14]; + out[12] = in[3]; out[13] = in[7]; out[14] = in[11]; out[15] = in[15]; +} + + + +static void fetch_xyz_rgb( struct spu_vs_context *draw, + struct spu_exec_machine *machine, + const unsigned *elts, + unsigned count ) +{ + assert(count <= 4); + +// _mesa_printf("%s\n", __FUNCTION__); + + /* loop over vertex attributes (vertex shader inputs) + */ + + const unsigned *pitch = draw->vertex_fetch.pitch; + const ubyte **src = draw->vertex_fetch.src_ptr; + int i; + + for (i = 0; i < 4; i++) { + { + const float *in = (const float *)(src[0] + elts[i] * pitch[0]); + float *out = &machine->Inputs[0].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + + { + const float *in = (const float *)(src[1] + elts[i] * pitch[1]); + float *out = &machine->Inputs[1].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + } +} + + + + +static void fetch_xyz_rgb_st( struct spu_vs_context *draw, + struct spu_exec_machine *machine, + const unsigned *elts, + unsigned count ) +{ + assert(count <= 4); + + /* loop over vertex attributes (vertex shader inputs) + */ + + const unsigned *pitch = draw->vertex_fetch.pitch; + const ubyte **src = draw->vertex_fetch.src_ptr; + int i; + + for (i = 0; i < 4; i++) { + { + const float *in = (const float *)(src[0] + elts[i] * pitch[0]); + float *out = &machine->Inputs[0].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + + { + const float *in = (const float *)(src[1] + elts[i] * pitch[1]); + float *out = &machine->Inputs[1].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = in[2]; + out[12] = 1.0f; + } + + { + const float *in = (const float *)(src[2] + elts[i] * pitch[2]); + float *out = &machine->Inputs[1].xyzw[0].f[i]; + out[0] = in[0]; + out[4] = in[1]; + out[8] = 0.0f; + out[12] = 1.0f; + } + } +} + + + + +/** + * Fetch vertex attributes for 'count' vertices. + */ +static void generic_vertex_fetch( struct spu_vs_context *draw, + struct spu_exec_machine *machine, + const unsigned *elts, + unsigned count ) +{ + unsigned nr_attrs = draw->vertex_fetch.nr_attrs; + unsigned attr; + + assert(count <= 4); + +// _mesa_printf("%s %d\n", __FUNCTION__, count); + + /* loop over vertex attributes (vertex shader inputs) + */ + for (attr = 0; attr < nr_attrs; attr++) { + + const unsigned pitch = draw->vertex_fetch.pitch[attr]; + const ubyte *src = draw->vertex_fetch.src_ptr[attr]; + const spu_fetch_func fetch = draw->vertex_fetch.fetch[attr]; + unsigned i; + float p[4][4]; + + + /* Fetch four attributes for four vertices. + * + * Could fetch directly into AOS format, but this is meant to be + * a prototype for an sse implementation, which would have + * difficulties doing that. + */ + for (i = 0; i < count; i++) + fetch( src + elts[i] * pitch, p[i] ); + + /* Be nice and zero out any missing vertices: + */ + for (/* empty */; i < 4; i++) + p[i][0] = p[i][1] = p[i][2] = p[i][3] = 0; + + /* Transpose/swizzle into sse-friendly format. Currently + * assuming that all vertex shader inputs are float[4], but this + * isn't true -- if the vertex shader only wants tex0.xy, we + * could optimize for that. + * + * To do so fully without codegen would probably require an + * excessive number of fetch functions, but we could at least + * minimize the transpose step: + */ + transpose_4x4( (float *)&machine->Inputs[attr].xyzw[0].f[0], (float *)p ); + } +} + + +void spu_update_vertex_fetch( struct spu_vs_context *draw ) +{ + unsigned i; + + + for (i = 0; i < draw->vertex_fetch.nr_attrs; i++) { + draw->vertex_fetch.fetch[i] = + get_fetch_func(draw->vertex_fetch.format[i]); + } + + draw->vertex_fetch.fetch_func = generic_vertex_fetch; + + switch (draw->vertex_fetch.nr_attrs) { + case 2: + if (draw->vertex_fetch.format[0] == PIPE_FORMAT_R32G32B32_FLOAT && + draw->vertex_fetch.format[1] == PIPE_FORMAT_R32G32B32_FLOAT) + draw->vertex_fetch.fetch_func = fetch_xyz_rgb; + break; + case 3: + if (draw->vertex_fetch.format[0] == PIPE_FORMAT_R32G32B32_FLOAT && + draw->vertex_fetch.format[1] == PIPE_FORMAT_R32G32B32_FLOAT && + draw->vertex_fetch.format[2] == PIPE_FORMAT_R32G32_FLOAT) + draw->vertex_fetch.fetch_func = fetch_xyz_rgb_st; + break; + default: + break; + } +} diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.c b/src/mesa/pipe/cell/spu/spu_vertex_shader.c new file mode 100644 index 00000000000..e694ff729f1 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.c @@ -0,0 +1,224 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell + * Brian Paul + * Ian Romanick + */ + +#include "pipe/p_util.h" +#include "pipe/p_state.h" +#include "pipe/p_shader_tokens.h" +#include "spu_vertex_shader.h" +#include "spu_exec.h" +#include "pipe/draw/draw_private.h" +#include "pipe/draw/draw_context.h" +#include "pipe/cell/common.h" + +#define DBG_VS 0 + + +static INLINE unsigned +compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr) +{ + unsigned mask = 0; + unsigned i; + + /* Do the hardwired planes first: + */ + if (-clip[0] + clip[3] < 0) mask |= CLIP_RIGHT_BIT; + if ( clip[0] + clip[3] < 0) mask |= CLIP_LEFT_BIT; + if (-clip[1] + clip[3] < 0) mask |= CLIP_TOP_BIT; + if ( clip[1] + clip[3] < 0) mask |= CLIP_BOTTOM_BIT; + if (-clip[2] + clip[3] < 0) mask |= CLIP_FAR_BIT; + if ( clip[2] + clip[3] < 0) mask |= CLIP_NEAR_BIT; + + /* Followed by any remaining ones: + */ + for (i = 6; i < nr; i++) { + if (dot4(clip, plane[i]) < 0) + mask |= (1<machine; + unsigned int j; + + ALIGN16_DECL(struct spu_exec_vector, inputs, PIPE_ATTRIB_MAX); + ALIGN16_DECL(struct spu_exec_vector, outputs, PIPE_ATTRIB_MAX); + const float *scale = draw->viewport.scale; + const float *trans = draw->viewport.translate; + + assert(count <= 4); + + /* Consts does not require 16 byte alignment. */ + ASSERT_ALIGN16(draw->constants); + machine->Consts = (float (*)[4]) draw->constants; + + machine->Inputs = ALIGN16_ASSIGN(inputs); + machine->Outputs = ALIGN16_ASSIGN(outputs); + + spu_vertex_fetch( draw, machine, elts, count ); + + /* run shader */ + spu_exec_machine_run( machine ); + + + /* store machine results */ + for (j = 0; j < count; j++) { + unsigned slot; + float x, y, z, w; + + /* Handle attr[0] (position) specially: + * + * XXX: Computing the clipmask should be done in the vertex + * program as a set of DP4 instructions appended to the + * user-provided code. + */ + x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j]; + y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j]; + z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j]; + w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j]; + + vOut[j]->clipmask = compute_clipmask(vOut[j]->clip, draw->plane, + draw->nr_planes); + vOut[j]->edgeflag = 1; + + /* divide by w */ + w = 1.0f / w; + x *= w; + y *= w; + z *= w; + + /* Viewport mapping */ + vOut[j]->data[0][0] = x * scale[0] + trans[0]; + vOut[j]->data[0][1] = y * scale[1] + trans[1]; + vOut[j]->data[0][2] = z * scale[2] + trans[2]; + vOut[j]->data[0][3] = w; + +#if DBG_VS + printf("output[%d]win: %f %f %f %f\n", j, + vOut[j]->data[0][0], + vOut[j]->data[0][1], + vOut[j]->data[0][2], + vOut[j]->data[0][3]); +#endif + /* Remaining attributes are packed into sequential post-transform + * vertex attrib slots. + */ + for (slot = 1; slot < draw->num_vs_outputs; slot++) { + vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; + vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; + vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; + vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; +#if DBG_VS + printf("output[%d][%d]: %f %f %f %f\n", j, slot, + vOut[j]->data[slot][0], + vOut[j]->data[slot][1], + vOut[j]->data[slot][2], + vOut[j]->data[slot][3]); +#endif + } + } /* loop over vertices */ +} + + +static void +spu_bind_vertex_shader(struct spu_vs_context *draw, + void *uniforms, + void *planes, + unsigned nr_planes, + unsigned num_outputs + ) +{ + draw->constants = (float (*)[4]) uniforms; + + (void) memcpy(draw->plane, planes, sizeof(float) * 4 * nr_planes); + draw->nr_planes = nr_planes; + draw->num_vs_outputs = num_outputs; + + /* specify the shader to interpret/execute */ + spu_exec_machine_init(&draw->machine, + PIPE_MAX_SAMPLERS, + NULL /*samplers*/, + PIPE_SHADER_VERTEX); +} + + +void +spu_execute_vertex_shader(struct spu_vs_context *draw, + const struct cell_command_vs *vs) +{ + unsigned i; + unsigned j; + + draw->machine.Instructions = (struct tgsi_full_instruction *) + vs->shader.instructions; + draw->machine.NumInstructions = vs->shader.num_instructions; + + draw->machine.Declarations = (struct tgsi_full_declaration *) + vs->shader.declarations; + draw->machine.NumDeclarations = vs->shader.num_declarations; + + spu_bind_vertex_shader(draw, vs->shader.uniforms, + NULL, -1, + vs->shader.num_outputs); + + for (i = 0; i < vs->num_elts; i += 4) { + const unsigned batch_size = MIN2(vs->num_elts - i, 4); + unsigned elts[4]; + + for (j = 0; j < batch_size; j++) { + switch (vs->bytes_per_elt) { + case 1: elts[j] = ((unsigned char *) vs->elts)[i + j]; break; + case 2: elts[j] = ((unsigned short *)vs->elts)[i + j]; break; + case 4: elts[j] = ((unsigned int *) vs->elts)[i + j]; break; + } + } + + run_vertex_program(draw, elts, batch_size, + (struct vertex_header (*)[]) vs->vOut); + } +} diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.h b/src/mesa/pipe/cell/spu/spu_vertex_shader.h new file mode 100644 index 00000000000..c52f38fd026 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.h @@ -0,0 +1,61 @@ +#ifndef SPU_VERTEX_SHADER_H +#define SPU_VERTEX_SHADER_H + +#include "pipe/p_format.h" +#include "spu_exec.h" + +struct spu_vs_context; + +typedef void (*spu_fetch_func)(const void *ptr, float *attrib); +typedef void (*spu_full_fetch_func)( struct spu_vs_context *draw, + struct spu_exec_machine *machine, + const unsigned *elts, + unsigned count ); + +struct spu_vs_context { + struct pipe_viewport_state viewport; + + struct { + const ubyte *src_ptr[PIPE_ATTRIB_MAX]; + unsigned pitch[PIPE_ATTRIB_MAX]; + enum pipe_format format[PIPE_ATTRIB_MAX]; + unsigned nr_attrs; + boolean dirty; + + spu_fetch_func fetch[PIPE_ATTRIB_MAX]; + spu_full_fetch_func fetch_func; + } vertex_fetch; + + /* Clip derived state: + */ + float plane[12][4]; + unsigned nr_planes; + + struct spu_exec_machine machine; + const float (*constants)[4]; + + unsigned num_vs_outputs; +}; + +extern void spu_update_vertex_fetch(struct spu_vs_context *draw); + +static INLINE void spu_vertex_fetch(struct spu_vs_context *draw, + struct spu_exec_machine *machine, + const unsigned *elts, + unsigned count) +{ + if (draw->vertex_fetch.dirty) { + spu_update_vertex_fetch(draw); + draw->vertex_fetch.dirty = 0; + } + + (*draw->vertex_fetch.fetch_func)(draw, machine, elts, count); +} + +struct cell_command_vs; + +extern void +spu_execute_vertex_shader(struct spu_vs_context *draw, + const struct cell_command_vs *vs); + +#endif /* SPU_VERTEX_SHADER_H */ -- cgit v1.2.3 From 5028f0fcaca0d3a521f7ec130f4bbea2600bce16 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Tue, 29 Jan 2008 10:56:53 -0800 Subject: Initial pass at instruction / declaration fetch --- src/mesa/pipe/cell/spu/spu_exec.c | 22 ++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_main.h | 1 + 2 files changed, 23 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index 6888e97caf6..f43278198e1 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -50,6 +50,9 @@ * Brian Paul */ +#include +#include + #include "pipe/p_compiler.h" #include "pipe/p_state.h" #include "pipe/p_util.h" @@ -57,6 +60,7 @@ #include "pipe/tgsi/util/tgsi_parse.h" #include "pipe/tgsi/util/tgsi_util.h" #include "spu_exec.h" +#include "spu_main.h" #define TILE_TOP_LEFT 0 #define TILE_TOP_RIGHT 1 @@ -2329,12 +2333,30 @@ spu_exec_machine_run( struct spu_exec_machine *mach ) /* execute declarations (interpolants) */ for (i = 0; i < mach->NumDeclarations; i++) { + uint8_t buffer[sizeof(struct tgsi_full_declaration) + 32] ALIGN16_ATTRIB; + struct tgsi_full_declaration decl; + unsigned long decl_addr = (unsigned long) (mach->Declarations+i); + unsigned size = ((sizeof(decl) + (decl_addr & 0x0f) + 0x0f) & ~0x0f); + + mfc_get(buffer, decl_addr & ~0x0f, size, TAG_INSTRUCTION_FETCH, 0, 0); + wait_on_mask(1 << TAG_INSTRUCTION_FETCH); + + memcpy(& decl, buffer + (decl_addr & 0x0f), sizeof(decl)); exec_declaration( mach, mach->Declarations+i ); } /* execute instructions, until pc is set to -1 */ while (pc != -1) { + uint8_t buffer[sizeof(struct tgsi_full_instruction) + 32] ALIGN16_ATTRIB; + struct tgsi_full_instruction inst; + unsigned long inst_addr = (unsigned long) (mach->Instructions + pc); + unsigned size = ((sizeof(inst) + (inst_addr & 0x0f) + 0x0f) & ~0x0f); + assert(pc < mach->NumInstructions); + mfc_get(buffer, inst_addr & ~0x0f, size, TAG_INSTRUCTION_FETCH, 0, 0); + wait_on_mask(1 << TAG_INSTRUCTION_FETCH); + + memcpy(& inst, buffer + (inst_addr & 0x0f), sizeof(inst)); exec_instruction( mach, mach->Instructions + pc, &pc ); } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 73f9ed29d66..8be5268f52d 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -96,6 +96,7 @@ extern boolean Debug; #define TAG_BATCH_BUFFER 17 #define TAG_MISC 18 #define TAG_TEXTURE_TILE 19 +#define TAG_INSTRUCTION_FETCH 20 -- cgit v1.2.3 From fc4620554a3eed2a4032d9f6bd349acfd152682c Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Tue, 29 Jan 2008 11:28:06 -0800 Subject: Implement vertex fetch / vertex shader output write-back --- src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 32 +++++++++++---- src/mesa/pipe/cell/spu/spu_vertex_shader.c | 62 +++++++++++++++--------------- src/mesa/pipe/draw/draw_context.c | 5 ++- 3 files changed, 58 insertions(+), 41 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c index b8f8c52eed3..0192227d57f 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -30,11 +30,13 @@ * Keith Whitwell */ +#include #include "pipe/p_util.h" #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" #include "spu_exec.h" #include "spu_vertex_shader.h" +#include "spu_main.h" #define DRAW_DBG 0 @@ -412,16 +414,18 @@ static void fetch_xyz_rgb_st( struct spu_vs_context *draw, /** * Fetch vertex attributes for 'count' vertices. */ -static void generic_vertex_fetch( struct spu_vs_context *draw, - struct spu_exec_machine *machine, - const unsigned *elts, - unsigned count ) +static void generic_vertex_fetch(struct spu_vs_context *draw, + struct spu_exec_machine *machine, + const unsigned *elts, + unsigned count) { unsigned nr_attrs = draw->vertex_fetch.nr_attrs; unsigned attr; assert(count <= 4); + wait_on_mask(1 << TAG_VERTEX_BUFFER); + // _mesa_printf("%s %d\n", __FUNCTION__, count); /* loop over vertex attributes (vertex shader inputs) @@ -441,13 +445,23 @@ static void generic_vertex_fetch( struct spu_vs_context *draw, * a prototype for an sse implementation, which would have * difficulties doing that. */ - for (i = 0; i < count; i++) - fetch( src + elts[i] * pitch, p[i] ); + for (i = 0; i < count; i++) { + uint8_t buffer[32 + (sizeof(float) * 4)] ALIGN16_ATTRIB; + const unsigned long addr = src + elts[i] * pitch; + const unsigned size = (sizeof(float) * 4) + (addr & 0x0f); + + mfc_get(buffer, addr & ~0x0f, size, TAG_VERTEX_BUFFER, 0, 0); + wait_on_mask(1 << TAG_VERTEX_BUFFER); + + memcpy(& buffer, buffer + (addr & 0x0f), sizeof(float) * 4); + + fetch(buffer, p[i]); + } /* Be nice and zero out any missing vertices: */ for (/* empty */; i < 4; i++) - p[i][0] = p[i][1] = p[i][2] = p[i][3] = 0; + p[i][0] = p[i][1] = p[i][2] = p[i][3] = 0; /* Transpose/swizzle into sse-friendly format. Currently * assuming that all vertex shader inputs are float[4], but this @@ -475,6 +489,9 @@ void spu_update_vertex_fetch( struct spu_vs_context *draw ) draw->vertex_fetch.fetch_func = generic_vertex_fetch; + /* Disable the fast path because they don't use mfc_get yet. + */ +#if 0 switch (draw->vertex_fetch.nr_attrs) { case 2: if (draw->vertex_fetch.format[0] == PIPE_FORMAT_R32G32B32_FLOAT && @@ -490,4 +507,5 @@ void spu_update_vertex_fetch( struct spu_vs_context *draw ) default: break; } +#endif } diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.c b/src/mesa/pipe/cell/spu/spu_vertex_shader.c index e694ff729f1..595f54b0ebd 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.c @@ -32,6 +32,8 @@ * Ian Romanick */ +#include + #include "pipe/p_util.h" #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" @@ -40,9 +42,7 @@ #include "pipe/draw/draw_private.h" #include "pipe/draw/draw_context.h" #include "pipe/cell/common.h" - -#define DBG_VS 0 - +#include "spu_main.h" static INLINE unsigned compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr) @@ -110,6 +110,12 @@ run_vertex_program(struct spu_vs_context *draw, for (j = 0; j < count; j++) { unsigned slot; float x, y, z, w; + unsigned char buffer[sizeof(struct vertex_header) + + MAX_VERTEX_SIZE] ALIGN16_ATTRIB; + struct vertex_header *const tmpOut = + (struct vertex_header *) buffer; + const unsigned vert_size = sizeof(struct vertex_header) + + (sizeof(float) * 4 * draw->num_vs_outputs); /* Handle attr[0] (position) specially: * @@ -117,14 +123,14 @@ run_vertex_program(struct spu_vs_context *draw, * program as a set of DP4 instructions appended to the * user-provided code. */ - x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j]; - y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j]; - z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j]; - w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j]; + x = tmpOut->clip[0] = machine->Outputs[0].xyzw[0].f[j]; + y = tmpOut->clip[1] = machine->Outputs[0].xyzw[1].f[j]; + z = tmpOut->clip[2] = machine->Outputs[0].xyzw[2].f[j]; + w = tmpOut->clip[3] = machine->Outputs[0].xyzw[3].f[j]; - vOut[j]->clipmask = compute_clipmask(vOut[j]->clip, draw->plane, + tmpOut->clipmask = compute_clipmask(tmpOut->clip, draw->plane, draw->nr_planes); - vOut[j]->edgeflag = 1; + tmpOut->edgeflag = 1; /* divide by w */ w = 1.0f / w; @@ -133,35 +139,27 @@ run_vertex_program(struct spu_vs_context *draw, z *= w; /* Viewport mapping */ - vOut[j]->data[0][0] = x * scale[0] + trans[0]; - vOut[j]->data[0][1] = y * scale[1] + trans[1]; - vOut[j]->data[0][2] = z * scale[2] + trans[2]; - vOut[j]->data[0][3] = w; - -#if DBG_VS - printf("output[%d]win: %f %f %f %f\n", j, - vOut[j]->data[0][0], - vOut[j]->data[0][1], - vOut[j]->data[0][2], - vOut[j]->data[0][3]); -#endif + tmpOut->data[0][0] = x * scale[0] + trans[0]; + tmpOut->data[0][1] = y * scale[1] + trans[1]; + tmpOut->data[0][2] = z * scale[2] + trans[2]; + tmpOut->data[0][3] = w; + /* Remaining attributes are packed into sequential post-transform * vertex attrib slots. */ for (slot = 1; slot < draw->num_vs_outputs; slot++) { - vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; - vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; - vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; - vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; -#if DBG_VS - printf("output[%d][%d]: %f %f %f %f\n", j, slot, - vOut[j]->data[slot][0], - vOut[j]->data[slot][1], - vOut[j]->data[slot][2], - vOut[j]->data[slot][3]); -#endif + tmpOut->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; + tmpOut->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; + tmpOut->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; + tmpOut->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; } + + wait_on_mask(1 << TAG_VERTEX_BUFFER); + mfc_put(tmpOut, vOut[j], vert_size, TAG_VERTEX_BUFFER, 0, 0); + } /* loop over vertices */ + + wait_on_mask(1 << TAG_VERTEX_BUFFER); } diff --git a/src/mesa/pipe/draw/draw_context.c b/src/mesa/pipe/draw/draw_context.c index e8ca1f035bb..711bcd02f6f 100644 --- a/src/mesa/pipe/draw/draw_context.c +++ b/src/mesa/pipe/draw/draw_context.c @@ -71,10 +71,11 @@ struct draw_context *draw_create( void ) */ { uint i; - char *tmp = (char*) MALLOC( Elements(draw->vcache.vertex) * MAX_VERTEX_SIZE ); + const unsigned size = (MAX_VERTEX_SIZE + 0x0f) & ~0x0f; + char *tmp = align_malloc(Elements(draw->vcache.vertex) * size, 16); for (i = 0; i < Elements(draw->vcache.vertex); i++) - draw->vcache.vertex[i] = (struct vertex_header *)(tmp + i * MAX_VERTEX_SIZE); + draw->vcache.vertex[i] = (struct vertex_header *)(tmp + i * size); } draw->convert_wide_points = TRUE; -- cgit v1.2.3 From d798e7e2689338918218bbde5b8a5d68e3ca8c22 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Tue, 29 Jan 2008 11:43:04 -0800 Subject: Provide mechanism to hook in custom vertex shader cache flush function --- src/mesa/pipe/draw/draw_context.c | 2 ++ src/mesa/pipe/draw/draw_prim.c | 2 +- src/mesa/pipe/draw/draw_private.h | 5 +++++ 3 files changed, 8 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_context.c b/src/mesa/pipe/draw/draw_context.c index 711bcd02f6f..87f4969983a 100644 --- a/src/mesa/pipe/draw/draw_context.c +++ b/src/mesa/pipe/draw/draw_context.c @@ -78,6 +78,8 @@ struct draw_context *draw_create( void ) draw->vcache.vertex[i] = (struct vertex_header *)(tmp + i * size); } + draw->shader_queue_flush = draw_vertex_shader_queue_flush; + draw->convert_wide_points = TRUE; draw->convert_wide_lines = TRUE; diff --git a/src/mesa/pipe/draw/draw_prim.c b/src/mesa/pipe/draw/draw_prim.c index 41b3fddcc10..58400213d76 100644 --- a/src/mesa/pipe/draw/draw_prim.c +++ b/src/mesa/pipe/draw/draw_prim.c @@ -127,7 +127,7 @@ void draw_do_flush( struct draw_context *draw, unsigned flags ) if (flags >= DRAW_FLUSH_SHADER_QUEUE) { if (draw->vs.queue_nr) - draw_vertex_shader_queue_flush(draw); + (*draw->shader_queue_flush)(draw); if (flags >= DRAW_FLUSH_PRIM_QUEUE) { if (draw->pq.queue_nr) diff --git a/src/mesa/pipe/draw/draw_private.h b/src/mesa/pipe/draw/draw_private.h index 21de4006769..fea6d94ed86 100644 --- a/src/mesa/pipe/draw/draw_private.h +++ b/src/mesa/pipe/draw/draw_private.h @@ -240,6 +240,11 @@ struct draw_context unsigned queue_nr; } vs; + /** + * Run the vertex shader on all vertices in the vertex queue. + */ + void (*shader_queue_flush)(struct draw_context *draw); + /* Prim pipeline queue: */ struct { -- cgit v1.2.3 From 7abddcf123f5726ea8b20ffb53100524a5da55d8 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 10:43:23 -0800 Subject: Pass ptr to local memory copy instead of main memory to exec_instruction This was essentially a cut-and-paste bug when the instruction fetcher was added. Also, the test for TGSI_PROCESSOR_FRAGMENT was moved outside the loop for exec_declaration. --- src/mesa/pipe/cell/spu/spu_exec.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index f43278198e1..b3db6716d5a 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -2332,17 +2332,19 @@ spu_exec_machine_run( struct spu_exec_machine *mach ) /* execute declarations (interpolants) */ - for (i = 0; i < mach->NumDeclarations; i++) { - uint8_t buffer[sizeof(struct tgsi_full_declaration) + 32] ALIGN16_ATTRIB; - struct tgsi_full_declaration decl; - unsigned long decl_addr = (unsigned long) (mach->Declarations+i); - unsigned size = ((sizeof(decl) + (decl_addr & 0x0f) + 0x0f) & ~0x0f); + if( mach->Processor == TGSI_PROCESSOR_FRAGMENT ) { + for (i = 0; i < mach->NumDeclarations; i++) { + uint8_t buffer[sizeof(struct tgsi_full_declaration) + 32] ALIGN16_ATTRIB; + struct tgsi_full_declaration decl; + unsigned long decl_addr = (unsigned long) (mach->Declarations+i); + unsigned size = ((sizeof(decl) + (decl_addr & 0x0f) + 0x0f) & ~0x0f); - mfc_get(buffer, decl_addr & ~0x0f, size, TAG_INSTRUCTION_FETCH, 0, 0); - wait_on_mask(1 << TAG_INSTRUCTION_FETCH); + mfc_get(buffer, decl_addr & ~0x0f, size, TAG_INSTRUCTION_FETCH, 0, 0); + wait_on_mask(1 << TAG_INSTRUCTION_FETCH); - memcpy(& decl, buffer + (decl_addr & 0x0f), sizeof(decl)); - exec_declaration( mach, mach->Declarations+i ); + memcpy(& decl, buffer + (decl_addr & 0x0f), sizeof(decl)); + exec_declaration( mach, decl ); + } } /* execute instructions, until pc is set to -1 */ @@ -2357,7 +2359,7 @@ spu_exec_machine_run( struct spu_exec_machine *mach ) wait_on_mask(1 << TAG_INSTRUCTION_FETCH); memcpy(& inst, buffer + (inst_addr & 0x0f), sizeof(inst)); - exec_instruction( mach, mach->Instructions + pc, &pc ); + exec_instruction( mach, & inst, &pc ); } #if 0 -- cgit v1.2.3 From c1ffb57cdf817934470a2115f6bdca148bdae269 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 10:46:55 -0800 Subject: Missing amperstand in previous commit. Oops. --- src/mesa/pipe/cell/spu/spu_exec.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index b3db6716d5a..85b5815cad1 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -2343,7 +2343,7 @@ spu_exec_machine_run( struct spu_exec_machine *mach ) wait_on_mask(1 << TAG_INSTRUCTION_FETCH); memcpy(& decl, buffer + (decl_addr & 0x0f), sizeof(decl)); - exec_declaration( mach, decl ); + exec_declaration( mach, &decl ); } } -- cgit v1.2.3 From ea1d5c43b28f16d5ff3bcc750d46143a35597a9e Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 12:59:09 -0800 Subject: Fetch uniforms from main memory. --- src/mesa/pipe/cell/spu/spu_exec.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index 85b5815cad1..78f7d0962f8 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -791,12 +791,23 @@ fetch_src_file_channel( case TGSI_EXTSWIZZLE_Z: case TGSI_EXTSWIZZLE_W: switch( file ) { - case TGSI_FILE_CONSTANT: - chan->f[0] = mach->Consts[index->i[0]][swizzle]; - chan->f[1] = mach->Consts[index->i[1]][swizzle]; - chan->f[2] = mach->Consts[index->i[2]][swizzle]; - chan->f[3] = mach->Consts[index->i[3]][swizzle]; + case TGSI_FILE_CONSTANT: { + unsigned char buffer[32] ALIGN16_ATTRIB; + unsigned i; + + for (i = 0; i < 4; i++) { + const float *ptr = mach->Consts[index->i[i]]; + const uint64_t addr = (uint64_t)(uintptr_t) ptr; + const unsigned size = ((addr & 0x0f) == 0) ? 16 : 32; + + mfc_get(buffer, addr & ~0x0f, size, TAG_VERTEX_BUFFER, 0, 0); + wait_on_mask(1 << TAG_VERTEX_BUFFER); + + (void) memcpy(& chan->f[i], &buffer[(addr & 0x0f) + + (sizeof(float) * swizzle)], sizeof(float)); + } break; + } case TGSI_FILE_INPUT: chan->u[0] = mach->Inputs[index->i[0]].xyzw[swizzle].u[0]; -- cgit v1.2.3 From 805aacfe604eaf9b414336318b8e170dd898123c Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 17:26:22 -0800 Subject: Fix size calculation in attribute fetch. --- src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c index 0192227d57f..1e846868e38 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -446,14 +446,14 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, * difficulties doing that. */ for (i = 0; i < count; i++) { - uint8_t buffer[32 + (sizeof(float) * 4)] ALIGN16_ATTRIB; - const unsigned long addr = src + elts[i] * pitch; - const unsigned size = (sizeof(float) * 4) + (addr & 0x0f); + uint8_t buffer[32] ALIGN16_ATTRIB; + const unsigned long addr = src + (elts[i] * pitch); + const unsigned size = ((addr & 0x0f) == 0) ? 16 : 32; mfc_get(buffer, addr & ~0x0f, size, TAG_VERTEX_BUFFER, 0, 0); wait_on_mask(1 << TAG_VERTEX_BUFFER); - memcpy(& buffer, buffer + (addr & 0x0f), sizeof(float) * 4); + memmove(& buffer, buffer + (addr & 0x0f), 16); fetch(buffer, p[i]); } -- cgit v1.2.3 From bbbd5c166a780d70110d236d40d3babd9d0b6346 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 17:28:48 -0800 Subject: Implement micro_pow and micro_sqrt Unimplemented micro ops get assertions for now. --- src/mesa/pipe/cell/spu/spu_exec.c | 43 ++++++++++++++++++++++++++++----------- 1 file changed, 31 insertions(+), 12 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index 78f7d0962f8..168bada3bb0 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -52,6 +52,8 @@ #include #include +#include +#include #include "pipe/p_compiler.h" #include "pipe/p_state.h" @@ -207,6 +209,7 @@ micro_ceil( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = (float) ceil( (double) src->f[0] ); dst->f[1] = (float) ceil( (double) src->f[1] ); @@ -220,6 +223,7 @@ micro_cos( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = (float) cos( (double) src->f[0] ); dst->f[1] = (float) cos( (double) src->f[1] ); @@ -307,6 +311,7 @@ micro_exp2( union spu_exec_channel *dst, const union spu_exec_channel *src) { + ASSERT(0); #if 0 dst->f[0] = (float) pow( 2.0, (double) src->f[0] ); dst->f[1] = (float) pow( 2.0, (double) src->f[1] ); @@ -342,6 +347,7 @@ micro_flr( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = (float) floor( (double) src->f[0] ); dst->f[1] = (float) floor( (double) src->f[1] ); @@ -355,6 +361,7 @@ micro_frc( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = src->f[0] - (float) floor( (double) src->f[0] ); dst->f[1] = src->f[1] - (float) floor( (double) src->f[1] ); @@ -393,6 +400,7 @@ micro_lg2( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = (float) log( (double) src->f[0] ) * 1.442695f; dst->f[1] = (float) log( (double) src->f[1] ) * 1.442695f; @@ -649,12 +657,18 @@ micro_pow( const union spu_exec_channel *src0, const union spu_exec_channel *src1 ) { -#if 0 - dst->f[0] = (float) pow( (double) src0->f[0], (double) src1->f[0] ); - dst->f[1] = (float) pow( (double) src0->f[1], (double) src1->f[1] ); - dst->f[2] = (float) pow( (double) src0->f[2], (double) src1->f[2] ); - dst->f[3] = (float) pow( (double) src0->f[3], (double) src1->f[3] ); -#endif + vec_float4 s0 = (vec_float4) { + src0->f[0], src0->f[1], src0->f[2], src0->f[3] + }; + vec_float4 s1 = (vec_float4) { + src1->f[0], src1->f[1], src1->f[2], src1->f[3] + }; + vec_float4 d = _powf4(s0, s1); + + dst->f[0] = spu_extract(d, 0); + dst->f[1] = spu_extract(d, 1); + dst->f[2] = spu_extract(d, 2); + dst->f[3] = spu_extract(d, 3); } static void @@ -662,6 +676,7 @@ micro_rnd( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = (float) floor( (double) (src->f[0] + 0.5f) ); dst->f[1] = (float) floor( (double) (src->f[1] + 0.5f) ); @@ -722,6 +737,7 @@ micro_sin( union spu_exec_channel *dst, const union spu_exec_channel *src ) { + ASSERT(0); #if 0 dst->f[0] = (float) sin( (double) src->f[0] ); dst->f[1] = (float) sin( (double) src->f[1] ); @@ -734,12 +750,15 @@ static void micro_sqrt( union spu_exec_channel *dst, const union spu_exec_channel *src ) { -#if 0 - dst->f[0] = (float) sqrt( (double) src->f[0] ); - dst->f[1] = (float) sqrt( (double) src->f[1] ); - dst->f[2] = (float) sqrt( (double) src->f[2] ); - dst->f[3] = (float) sqrt( (double) src->f[3] ); -#endif + vec_float4 s = (vec_float4) { + src->f[0], src->f[1], src->f[2], src->f[3] + }; + vec_float4 d = _sqrtf4(s); + + dst->f[0] = spu_extract(d, 0); + dst->f[1] = spu_extract(d, 1); + dst->f[2] = spu_extract(d, 2); + dst->f[3] = spu_extract(d, 3); } static void -- cgit v1.2.3 From dcfe7e1dca656dd897b7b0bdebbed3cee6f9cfd9 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 19:24:40 -0800 Subject: Elts are always ints, pass vOut pointers in-line in command --- src/mesa/pipe/cell/common.h | 6 +++--- src/mesa/pipe/cell/spu/spu_vertex_shader.c | 14 ++------------ 2 files changed, 5 insertions(+), 15 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 80a1425ec7e..fbbdf728a12 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -144,13 +144,13 @@ struct cell_shader_info } ALIGN16_ATTRIB; +#define SPU_VERTS_PER_BATCH 64 struct cell_command_vs { struct cell_shader_info shader; - void *elts; unsigned num_elts; - unsigned bytes_per_elt; - void *vOut; + unsigned elts[SPU_VERTS_PER_BATCH]; + uint64_t vOut[SPU_VERTS_PER_BATCH]; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.c b/src/mesa/pipe/cell/spu/spu_vertex_shader.c index 595f54b0ebd..82165501c5c 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.c @@ -81,7 +81,7 @@ compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr) static void run_vertex_program(struct spu_vs_context *draw, unsigned elts[4], unsigned count, - struct vertex_header *vOut[]) + const uint64_t *vOut) { struct spu_exec_machine *machine = &draw->machine; unsigned int j; @@ -206,17 +206,7 @@ spu_execute_vertex_shader(struct spu_vs_context *draw, for (i = 0; i < vs->num_elts; i += 4) { const unsigned batch_size = MIN2(vs->num_elts - i, 4); - unsigned elts[4]; - - for (j = 0; j < batch_size; j++) { - switch (vs->bytes_per_elt) { - case 1: elts[j] = ((unsigned char *) vs->elts)[i + j]; break; - case 2: elts[j] = ((unsigned short *)vs->elts)[i + j]; break; - case 4: elts[j] = ((unsigned int *) vs->elts)[i + j]; break; - } - } - run_vertex_program(draw, elts, batch_size, - (struct vertex_header (*)[]) vs->vOut); + run_vertex_program(draw, & vs->elts[i], batch_size, &vs->vOut[i]); } } -- cgit v1.2.3 From fea350d91133174254f544de30032049199991bf Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 19:25:47 -0800 Subject: Set machine->Processor The default value is 0, which is TGSI_PROCESSOR_FRAGMENT...not correct for a vertex shader! --- src/mesa/pipe/cell/spu/spu_vertex_shader.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.c b/src/mesa/pipe/cell/spu/spu_vertex_shader.c index 82165501c5c..125b2c3a436 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.c @@ -93,7 +93,8 @@ run_vertex_program(struct spu_vs_context *draw, assert(count <= 4); - /* Consts does not require 16 byte alignment. */ + machine->Processor = TGSI_PROCESSOR_VERTEX; + ASSERT_ALIGN16(draw->constants); machine->Consts = (float (*)[4]) draw->constants; -- cgit v1.2.3 From 6996b6c0559b32926188efe7b23574a1076b75ec Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 19:30:15 -0800 Subject: Handle CELL_CMD_VS_EXECUTE *only* outside batch commands. --- src/mesa/pipe/cell/common.h | 3 ++- src/mesa/pipe/cell/spu/spu_main.c | 8 +++----- 2 files changed, 5 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index fbbdf728a12..a40cfb8210a 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -133,7 +133,6 @@ struct cell_array_info struct cell_shader_info { - unsigned processor; unsigned num_outputs; void *declarations; @@ -147,6 +146,7 @@ struct cell_shader_info #define SPU_VERTS_PER_BATCH 64 struct cell_command_vs { + uint opcode; /**< CELL_CMD_VS_EXECUTE */ struct cell_shader_info shader; unsigned num_elts; unsigned elts[SPU_VERTS_PER_BATCH]; @@ -190,6 +190,7 @@ struct cell_command struct cell_command_framebuffer fb; struct cell_command_clear_surface clear; struct cell_command_render render; + struct cell_command_vs vs; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 9daa3ec735a..7105c0f8977 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -397,11 +397,6 @@ cmd_batch(uint opcode) cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]); pos += (1 + sizeof(struct cell_array_info) / 4); break; - case CELL_CMD_VS_EXECUTE: - spu_execute_vertex_shader(&draw, - (struct cell_command_vs *) &buffer[pos+1]); - pos += (1 + sizeof(struct cell_command_vs) / 4); - break; default: printf("SPU %u: bad opcode: 0x%x\n", spu.init.id, buffer[pos]); ASSERT(0); @@ -470,6 +465,9 @@ main_loop(void) assert(pos_incr == 0); } break; + case CELL_CMD_VS_EXECUTE: + spu_execute_vertex_shader(&draw, &cmd.vs); + break; case CELL_CMD_BATCH: cmd_batch(opcode); break; -- cgit v1.2.3 From 738a4292b18e5513935af3902b4ed9d1997f90d1 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 19:33:30 -0800 Subject: Correctly read / write vertex header from / to main memory --- src/mesa/pipe/cell/spu/spu_vertex_shader.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.c b/src/mesa/pipe/cell/spu/spu_vertex_shader.c index 125b2c3a436..ea5ffae6bc6 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.c @@ -112,11 +112,16 @@ run_vertex_program(struct spu_vs_context *draw, unsigned slot; float x, y, z, w; unsigned char buffer[sizeof(struct vertex_header) - + MAX_VERTEX_SIZE] ALIGN16_ATTRIB; + + MAX_VERTEX_SIZE] ALIGN16_ATTRIB; struct vertex_header *const tmpOut = - (struct vertex_header *) buffer; - const unsigned vert_size = sizeof(struct vertex_header) - + (sizeof(float) * 4 * draw->num_vs_outputs); + (struct vertex_header *) buffer; + const unsigned vert_size = ROUNDUP16(sizeof(struct vertex_header) + + (sizeof(float) * 4 + * draw->num_vs_outputs)); + + mfc_get(tmpOut, vOut[j], vert_size, TAG_VERTEX_BUFFER, 0, 0); + wait_on_mask(1 << TAG_VERTEX_BUFFER); + /* Handle attr[0] (position) specially: * @@ -155,12 +160,8 @@ run_vertex_program(struct spu_vs_context *draw, tmpOut->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; } - wait_on_mask(1 << TAG_VERTEX_BUFFER); mfc_put(tmpOut, vOut[j], vert_size, TAG_VERTEX_BUFFER, 0, 0); - } /* loop over vertices */ - - wait_on_mask(1 << TAG_VERTEX_BUFFER); } -- cgit v1.2.3 From 42db5715988a23743084742bc52baaa568be6091 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 19:34:22 -0800 Subject: cell_array_info should not be 16-byte aligned Forcing cell_array_info to be 16-byte aligned makes it more difficult to stuff that state in batch commands. --- src/mesa/pipe/cell/common.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index a40cfb8210a..533ad2cf6eb 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -128,7 +128,7 @@ struct cell_array_info uint attr; /**< Attribute that this state if for. */ uint pitch; /**< Byte pitch from one entry to the next. */ enum pipe_format format; /**< Pipe format of each entry. */ -} ALIGN16_ATTRIB; +}; struct cell_shader_info -- cgit v1.2.3 From 524f99a4ab1f4dde0cab07ffd9e72a4d49c1e79d Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 19:40:24 -0800 Subject: Numerous small fixed to PPU-SPU vertex shader protocol --- src/mesa/pipe/cell/common.h | 19 ++++++++++++------- src/mesa/pipe/cell/spu/spu_vertex_shader.c | 27 ++++++++++++++++++++++----- 2 files changed, 34 insertions(+), 12 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 533ad2cf6eb..28b0c59a0ad 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -124,10 +124,10 @@ struct cell_command_clear_surface */ struct cell_array_info { - void *base; /**< Base address of the 0th element. */ - uint attr; /**< Attribute that this state if for. */ - uint pitch; /**< Byte pitch from one entry to the next. */ - enum pipe_format format; /**< Pipe format of each entry. */ + uint64_t base; /**< Base address of the 0th element. */ + uint attr; /**< Attribute that this state if for. */ + uint pitch; /**< Byte pitch from one entry to the next. */ + uint format; /**< Pipe format of each entry. */ }; @@ -135,11 +135,13 @@ struct cell_shader_info { unsigned num_outputs; - void *declarations; + uint64_t declarations; unsigned num_declarations; - void *instructions; + uint64_t instructions; unsigned num_instructions; - void *uniforms; + uint64_t uniforms; + uint64_t immediates; + unsigned num_immediates; } ALIGN16_ATTRIB; @@ -151,6 +153,9 @@ struct cell_command_vs unsigned num_elts; unsigned elts[SPU_VERTS_PER_BATCH]; uint64_t vOut[SPU_VERTS_PER_BATCH]; + float plane[12][4]; + unsigned nr_planes; + unsigned nr_attrs; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.c b/src/mesa/pipe/cell/spu/spu_vertex_shader.c index ea5ffae6bc6..c1cbbb6d1e9 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.c @@ -187,12 +187,22 @@ spu_bind_vertex_shader(struct spu_vs_context *draw, } +unsigned char immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32] + ALIGN16_ATTRIB; + void spu_execute_vertex_shader(struct spu_vs_context *draw, - const struct cell_command_vs *vs) + const struct cell_command_vs *vs) { unsigned i; - unsigned j; + + const uint64_t immediate_addr = vs->shader.immediates; + const unsigned immediate_size = + ROUNDUP16((sizeof(float) * 4 * vs->shader.num_immediates) + + (immediate_addr & 0x0f)); + + mfc_get(immediates, immediate_addr & ~0x0f, immediate_size, + TAG_VERTEX_BUFFER, 0, 0); draw->machine.Instructions = (struct tgsi_full_instruction *) vs->shader.instructions; @@ -202,10 +212,17 @@ spu_execute_vertex_shader(struct spu_vs_context *draw, vs->shader.declarations; draw->machine.NumDeclarations = vs->shader.num_declarations; + draw->vertex_fetch.nr_attrs = vs->nr_attrs; + + wait_on_mask(1 << TAG_VERTEX_BUFFER); + + (void) memcpy(& draw->machine.Imms, &immediates[immediate_addr & 0x0f], + sizeof(float) * 4 * vs->shader.num_immediates); + spu_bind_vertex_shader(draw, vs->shader.uniforms, - NULL, -1, - vs->shader.num_outputs); - + vs->plane, vs->nr_planes, + vs->shader.num_outputs); + for (i = 0; i < vs->num_elts; i += 4) { const unsigned batch_size = MIN2(vs->num_elts - i, 4); -- cgit v1.2.3 From 87cc80297c62d7f8f2f22415e5099e3fbb7d229b Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 20:10:45 -0800 Subject: Add driver_private field for drivers that hook shader_queue_flush. --- src/mesa/pipe/draw/draw_private.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/pipe/draw/draw_private.h b/src/mesa/pipe/draw/draw_private.h index fea6d94ed86..7782db04770 100644 --- a/src/mesa/pipe/draw/draw_private.h +++ b/src/mesa/pipe/draw/draw_private.h @@ -259,6 +259,8 @@ struct draw_context #ifdef MESA_LLVM struct gallivm_cpu_engine *engine; #endif + + void *driver_private; }; -- cgit v1.2.3 From bcaf0dd8d23b3e8562078b3a3e07ef99ca940881 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 20:12:00 -0800 Subject: Use SPUs for vertex shader processing --- src/mesa/pipe/cell/ppu/Makefile | 1 + src/mesa/pipe/cell/ppu/cell_context.c | 12 ++- src/mesa/pipe/cell/ppu/cell_context.h | 2 + src/mesa/pipe/cell/ppu/cell_vertex_shader.c | 118 ++++++++++++++++++++++++++++ 4 files changed, 132 insertions(+), 1 deletion(-) create mode 100644 src/mesa/pipe/cell/ppu/cell_vertex_shader.c diff --git a/src/mesa/pipe/cell/ppu/Makefile b/src/mesa/pipe/cell/ppu/Makefile index e7f2562da75..50060f5cd30 100644 --- a/src/mesa/pipe/cell/ppu/Makefile +++ b/src/mesa/pipe/cell/ppu/Makefile @@ -34,6 +34,7 @@ SOURCES = \ cell_surface.c \ cell_texture.c \ cell_vbuf.c \ + cell_vertex_shader.c \ cell_winsys.c diff --git a/src/mesa/pipe/cell/ppu/cell_context.c b/src/mesa/pipe/cell/ppu/cell_context.c index e8020a49bc3..4885cd0d2c7 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.c +++ b/src/mesa/pipe/cell/ppu/cell_context.c @@ -39,6 +39,7 @@ #include "pipe/p_winsys.h" #include "pipe/cell/common.h" #include "pipe/draw/draw_context.h" +#include "pipe/draw/draw_private.h" #include "cell_clear.h" #include "cell_context.h" #include "cell_draw_arrays.h" @@ -156,6 +157,15 @@ cell_destroy_context( struct pipe_context *pipe ) } +static struct draw_context * +cell_draw_create(struct cell_context *cell) +{ + struct draw_context *draw = draw_create(); + + draw->shader_queue_flush = cell_vertex_shader_queue_flush; + draw->driver_private = cell; + return draw; +} struct pipe_context * @@ -242,7 +252,7 @@ cell_create_context(struct pipe_winsys *winsys, struct cell_winsys *cws) cell_init_surface_functions(cell); - cell->draw = draw_create(); + cell->draw = cell_draw_create(cell); cell_init_vbuf(cell); draw_set_rasterize_stage(cell->draw, cell->vbuf); diff --git a/src/mesa/pipe/cell/ppu/cell_context.h b/src/mesa/pipe/cell/ppu/cell_context.h index 65b89518ada..3b63419b5eb 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.h +++ b/src/mesa/pipe/cell/ppu/cell_context.h @@ -126,6 +126,8 @@ cell_context(struct pipe_context *pipe) extern struct pipe_context * cell_create_context(struct pipe_winsys *ws, struct cell_winsys *cws); +extern void +cell_vertex_shader_queue_flush(struct draw_context *draw); diff --git a/src/mesa/pipe/cell/ppu/cell_vertex_shader.c b/src/mesa/pipe/cell/ppu/cell_vertex_shader.c new file mode 100644 index 00000000000..aef329a9024 --- /dev/null +++ b/src/mesa/pipe/cell/ppu/cell_vertex_shader.c @@ -0,0 +1,118 @@ +/* + * (C) Copyright IBM Corporation 2008 + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * AUTHORS, COPYRIGHT HOLDERS, AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/** + * \file cell_vertex_shader.c + * Vertex shader interface routines for Cell. + * + * \author Ian Romanick + */ + +#include "pipe/p_defines.h" +#include "pipe/p_context.h" +#include "pipe/p_winsys.h" + +#include "cell_context.h" +#include "cell_draw_arrays.h" +#include "cell_spu.h" +#include "cell_batch.h" + +#include "pipe/cell/common.h" +#include "pipe/draw/draw_context.h" +#include "pipe/draw/draw_private.h" + +/** + * Run the vertex shader on all vertices in the vertex queue. + * Called by the draw module when the vertx cache needs to be flushed. + */ +void +cell_vertex_shader_queue_flush(struct draw_context *draw) +{ + struct cell_context *const cell = + (struct cell_context *) draw->driver_private; + struct cell_command_vs *const vs = &cell_global.command[0].vs; + unsigned *batch; + struct cell_array_info array_info; + unsigned i, j; + + assert(draw->vs.queue_nr != 0); + + /* XXX: do this on statechange: + */ + draw_update_vertex_fetch(draw); + + for (i = 0; i < draw->vertex_fetch.nr_attrs; i++) { + array_info.opcode = CELL_CMD_STATE_VS_ARRAY_INFO; + assert(draw->vertex_fetch.src_ptr[i] != NULL); + array_info.base = (uintptr_t) draw->vertex_fetch.src_ptr[i]; + array_info.attr = i; + array_info.pitch = draw->vertex_fetch.pitch[i]; + array_info.format = draw->vertex_element[i].src_format; + + cell_batch_append(cell, & array_info, sizeof(array_info)); + } + + batch = cell_batch_alloc(cell, sizeof(unsigned) + + sizeof(struct pipe_viewport_state)); + batch[0] = CELL_CMD_STATE_VIEWPORT; + (void) memcpy(&batch[1], &draw->viewport, + sizeof(struct pipe_viewport_state)); + + cell_batch_flush(cell); + + vs->opcode = CELL_CMD_VS_EXECUTE; + vs->shader.num_outputs = draw->num_vs_outputs; + vs->shader.declarations = (uintptr_t) draw->machine.Declarations; + vs->shader.num_declarations = draw->machine.NumDeclarations; + vs->shader.instructions = (uintptr_t) draw->machine.Instructions; + vs->shader.num_instructions = draw->machine.NumInstructions; + vs->shader.uniforms = (uintptr_t) draw->user.constants; + vs->shader.immediates = (uintptr_t) draw->machine.Imms; + vs->shader.num_immediates = draw->machine.ImmLimit / 4; + vs->nr_attrs = draw->vertex_fetch.nr_attrs; + + (void) memcpy(vs->plane, draw->plane, sizeof(draw->plane)); + vs->nr_planes = draw->nr_planes; + + for (i = 0; i < draw->vs.queue_nr; i += SPU_VERTS_PER_BATCH) { + const unsigned n = MIN2(SPU_VERTS_PER_BATCH, draw->vs.queue_nr - i); + + for (j = 0; j < n; j++) { + vs->elts[j] = draw->vs.queue[i + j].elt; + vs->vOut[j] = (uintptr_t) draw->vs.queue[i + j].dest; + } + + for (/* empty */; j < SPU_VERTS_PER_BATCH; j++) { + vs->elts[j] = vs->elts[0]; + vs->vOut[j] = vs->vOut[0]; + } + + vs->num_elts = n; + send_mbox_message(cell_global.spe_contexts[0], CELL_CMD_VS_EXECUTE); + + cell_flush_int(& cell->pipe, PIPE_FLUSH_WAIT); + } + + draw->vs.queue_nr = 0; +} -- cgit v1.2.3 From 2029ee48b1b5856ef3c9b4307f018bc6bd61ea6e Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Wed, 30 Jan 2008 20:14:14 -0800 Subject: I don't know why using uint64_t for "base" doesn't work. Ugh. --- src/mesa/pipe/cell/common.h | 5 +++-- src/mesa/pipe/cell/spu/spu_main.c | 4 ++-- 2 files changed, 5 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 28b0c59a0ad..05aeed83ab5 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -124,11 +124,12 @@ struct cell_command_clear_surface */ struct cell_array_info { - uint64_t base; /**< Base address of the 0th element. */ + uint opcode; + uint base; /**< Base address of the 0th element. */ uint attr; /**< Attribute that this state if for. */ uint pitch; /**< Byte pitch from one entry to the next. */ uint format; /**< Pipe format of each entry. */ -}; +} ALIGN16_ATTRIB; struct cell_shader_info diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 7105c0f8977..d6393048f56 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -394,8 +394,8 @@ cmd_batch(uint opcode) pos += (1 + sizeof(struct pipe_viewport_state) / 4); break; case CELL_CMD_STATE_VS_ARRAY_INFO: - cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]); - pos += (1 + sizeof(struct cell_array_info) / 4); + cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos]); + pos += (sizeof(struct cell_array_info) / 4); break; default: printf("SPU %u: bad opcode: 0x%x\n", spu.init.id, buffer[pos]); -- cgit v1.2.3 From c285e06e93d01b9253dfc2fb1ab42480216b72d4 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Thu, 31 Jan 2008 13:14:35 +0900 Subject: gallium: Add SCons as alternative build system for Gallium. --- .gitignore | 2 + SConstruct | 214 +++++++++++++ src/mesa/SConscript | 435 +++++++++++++++++++++++++++ src/mesa/drivers/dri/SConscript | 48 +++ src/mesa/drivers/dri/intel_winsys/SConscript | 41 +++ src/mesa/pipe/SConscript | 9 + src/mesa/pipe/i915simple/SConscript | 29 ++ src/mesa/pipe/i965simple/SConscript | 55 ++++ src/mesa/pipe/softpipe/SConscript | 42 +++ 9 files changed, 875 insertions(+) create mode 100644 SConstruct create mode 100644 src/mesa/SConscript create mode 100644 src/mesa/drivers/dri/SConscript create mode 100644 src/mesa/drivers/dri/intel_winsys/SConscript create mode 100644 src/mesa/pipe/SConscript create mode 100644 src/mesa/pipe/i915simple/SConscript create mode 100644 src/mesa/pipe/i965simple/SConscript create mode 100644 src/mesa/pipe/softpipe/SConscript diff --git a/.gitignore b/.gitignore index 033e6e10bde..bf50291fc16 100644 --- a/.gitignore +++ b/.gitignore @@ -7,3 +7,5 @@ depend depend.bak lib lib64 +.sconsign* +config.py diff --git a/SConstruct b/SConstruct new file mode 100644 index 00000000000..db6161ed519 --- /dev/null +++ b/SConstruct @@ -0,0 +1,214 @@ +####################################################################### +# Top-level SConstruct + +import os +import sys + + +####################################################################### +# Configuration options +# +# For example, invoke scons as +# +# scons debug=1 dri=0 x86=1 +# +# to set configuration variables. Or you can write those options to a file +# named config.py: +# +# # config.py +# debug=1 +# dri=0 +# x86=1 +# +# Invoke +# +# scons -h +# +# to get the full list of options. See scons manpage for more info. +# + +# TODO: auto-detect defaults +opts = Options('config.py') +opts.Add(BoolOption('debug', 'build debug version', False)) +opts.Add(BoolOption('dri', 'build dri drivers', False)) +opts.Add(EnumOption('machine', 'use machine-specific assembly code', 'x86', + allowed_values=('generic', 'x86', 'x86-64'))) + +env = Environment(options = opts) +Help(opts.GenerateHelpText(env)) + +# for debugging +#print env.Dump() + +if 1: + # platform will be typically 'posix' or 'win32' + platform = env['PLATFORM'] +else: + # platform will be one of 'linux', 'freebsd', 'win32', 'darwin', etc. + platform = sys.platform + if platform == 'linux2': + platform = 'linux' + +# replicate options values in local variables +debug = env['debug'] +dri = env['dri'] +machine = env['machine'] + +# derived options +x86 = machine == 'x86' +gcc = platform == 'posix' +msvc = platform == 'win32' + +Export([ + 'debug', + 'x86', + 'dri', + 'platform', + 'gcc', + 'msvc', +]) + + +####################################################################### +# Environment setup +# +# TODO: put the compiler specific settings in seperate files +# TODO: auto-detect as much as possible + + +# Optimization flags +if gcc: + if debug: + env.Append(CFLAGS = '-O0 -g3') + env.Append(CXXFLAGS = '-O0 -g3') + else: + env.Append(CFLAGS = '-O3 -g3') + env.Append(CXXFLAGS = '-O3 -g3') + + env.Append(CFLAGS = '-Wall -Wmissing-prototypes -std=c99 -ffast-math -pedantic') + env.Append(CXXFLAGS = '-Wall -pedantic') + + # Be nice to Eclipse + env.Append(CFLAGS = '-fmessage-length=0') + env.Append(CXXFLAGS = '-fmessage-length=0') + +# Defines +env.Append(CPPDEFINES = [ + '_POSIX_SOURCE', + ('_POSIX_C_SOURCE', '199309L'), + '_SVID_SOURCE', + '_BSD_SOURCE', + '_GNU_SOURCE', + + 'PTHREADS', + 'HAVE_ALIAS', + 'HAVE_POSIX_MEMALIGN', +]) + +if debug: + env.Append(CPPDEFINES = ['DEBUG']) +else: + env.Append(CPPDEFINES = ['NDEBUG']) + + +# Includes +env.Append(CPPPATH = [ + '#/include', + '#/src/mesa', + '#/src/mesa/main', + '#/src/mesa/pipe', + + '/usr/X11R6/include', +]) + + +# x86 assembly +if x86: + env.Append(CPPDEFINES = [ + 'USE_X86_ASM', + 'USE_MMX_ASM', + 'USE_3DNOW_ASM', + 'USE_SSE_ASM', + ]) + if gcc: + env.Append(CFLAGS = '-m32') + env.Append(CXXFLAGS = '-m32') + +env.Append(LIBPATH = ['/usr/X11R6/lib']) + +env.Append(LIBS = [ + 'm', + 'pthread', + 'expat', + 'dl', +]) + +# DRI +if dri: + env.ParseConfig('pkg-config --cflags --libs libdrm') + env.Append(CPPDEFINES = [ + ('USE_EXTERNAL_DXTN_LIB', '1'), + 'IN_DRI_DRIVER', + 'GLX_DIRECT_RENDERING', + 'GLX_INDIRECT_RENDERING', + ]) + +# libGL +if 1: + env.Append(LIBS = [ + 'X11', + 'Xext', + 'Xxf86vm', + 'Xdamage', + 'Xfixes', + ]) + +Export('env') + + +####################################################################### +# Convenience Library Builder +# based on the stock StaticLibrary and SharedLibrary builders + +def createConvenienceLibBuilder(env): + """This is a utility function that creates the ConvenienceLibrary + Builder in an Environment if it is not there already. + + If it is already there, we return the existing one. + """ + + try: + convenience_lib = env['BUILDERS']['ConvenienceLibrary'] + except KeyError: + action_list = [ Action("$ARCOM", "$ARCOMSTR") ] + if env.Detect('ranlib'): + ranlib_action = Action("$RANLIBCOM", "$RANLIBCOMSTR") + action_list.append(ranlib_action) + + convenience_lib = Builder(action = action_list, + emitter = '$LIBEMITTER', + prefix = '$LIBPREFIX', + suffix = '$LIBSUFFIX', + src_suffix = '$SHOBJSUFFIX', + src_builder = 'SharedObject') + env['BUILDERS']['ConvenienceLibrary'] = convenience_lib + env['BUILDERS']['Library'] = convenience_lib + + return convenience_lib + +createConvenienceLibBuilder(env) + + +####################################################################### +# Invoke SConscripts + +# Put build output in a separate dir +# TODO: make build_dir depend on platform and build type (check +# http://www.scons.org/wiki/AdvancedBuildExample for an example) +build_dir = 'build' + +SConscript( + 'src/mesa/SConscript', + build_dir = build_dir, + duplicate = 0 # http://www.scons.org/doc/0.97/HTML/scons-user/x2261.html +) diff --git a/src/mesa/SConscript b/src/mesa/SConscript new file mode 100644 index 00000000000..70a98f31294 --- /dev/null +++ b/src/mesa/SConscript @@ -0,0 +1,435 @@ +####################################################################### +# SConscript for mesa +# +# TODO: Split this into per-module SConscripts + + +Import('*') + + +####################################################################### +# Core sources + +MAIN_SOURCES = [ + 'main/api_arrayelt.c', + 'main/api_loopback.c', + 'main/api_noop.c', + 'main/api_validate.c', + 'main/accum.c', + 'main/attrib.c', + 'main/arrayobj.c', + 'main/blend.c', + 'main/bufferobj.c', + 'main/buffers.c', + 'main/clip.c', + 'main/colortab.c', + 'main/context.c', + 'main/convolve.c', + 'main/debug.c', + 'main/depth.c', + 'main/depthstencil.c', + 'main/dlist.c', + 'main/drawpix.c', + 'main/enable.c', + 'main/enums.c', + 'main/eval.c', + 'main/execmem.c', + 'main/extensions.c', + 'main/fbobject.c', + 'main/feedback.c', + 'main/ffvertex_prog.c', + 'main/fog.c', + 'main/framebuffer.c', + 'main/get.c', + 'main/getstring.c', + 'main/hash.c', + 'main/hint.c', + 'main/histogram.c', + 'main/image.c', + 'main/imports.c', + 'main/light.c', + 'main/lines.c', + 'main/matrix.c', + 'main/mipmap.c', + 'main/mm.c', + 'main/pixel.c', + 'main/points.c', + 'main/polygon.c', + 'main/queryobj.c', + 'main/rastpos.c', + 'main/rbadaptors.c', + 'main/renderbuffer.c', + 'main/shaders.c', + 'main/state.c', + 'main/stencil.c', + 'main/texcompress.c', + 'main/texcompress_s3tc.c', + 'main/texcompress_fxt1.c', + 'main/texenvprogram.c', + 'main/texformat.c', + 'main/teximage.c', + 'main/texobj.c', + 'main/texrender.c', + 'main/texstate.c', + 'main/texstore.c', + 'main/varray.c', + 'main/vtxfmt.c', +] + +GLAPI_SOURCES = [ + 'main/dispatch.c', + 'glapi/glapi.c', + 'glapi/glthread.c', +] + +MATH_SOURCES = [ + 'math/m_debug_clip.c', + 'math/m_debug_norm.c', + 'math/m_debug_xform.c', + 'math/m_eval.c', + 'math/m_matrix.c', + 'math/m_translate.c', + 'math/m_vector.c', + 'math/m_xform.c', +] + +VBO_SOURCES = [ + 'vbo/vbo_context.c', + 'vbo/vbo_exec.c', + 'vbo/vbo_exec_api.c', + 'vbo/vbo_exec_array.c', + 'vbo/vbo_exec_draw.c', + 'vbo/vbo_exec_eval.c', + 'vbo/vbo_rebase.c', + 'vbo/vbo_split.c', + 'vbo/vbo_split_copy.c', + 'vbo/vbo_split_inplace.c', + 'vbo/vbo_save.c', + 'vbo/vbo_save_api.c', + 'vbo/vbo_save_draw.c', + 'vbo/vbo_save_loopback.c', +] + +VF_SOURCES = [ + 'vf/vf.c', + 'vf/vf_generic.c', + 'vf/vf_sse.c', +] + +DRAW_SOURCES = [ + 'pipe/draw/draw_clip.c', + 'pipe/draw/draw_context.c', + 'pipe/draw/draw_cull.c', + 'pipe/draw/draw_debug.c', + 'pipe/draw/draw_flatshade.c', + 'pipe/draw/draw_offset.c', + 'pipe/draw/draw_prim.c', + 'pipe/draw/draw_stipple.c', + 'pipe/draw/draw_twoside.c', + 'pipe/draw/draw_unfilled.c', + 'pipe/draw/draw_validate.c', + 'pipe/draw/draw_vbuf.c', + 'pipe/draw/draw_vertex.c', + 'pipe/draw/draw_vertex_cache.c', + 'pipe/draw/draw_vertex_fetch.c', + 'pipe/draw/draw_vertex_shader.c', + 'pipe/draw/draw_vertex_shader_llvm.c', + 'pipe/draw/draw_vf.c', + 'pipe/draw/draw_vf_generic.c', + 'pipe/draw/draw_vf_sse.c', + 'pipe/draw/draw_wide_prims.c', +] + +TGSIEXEC_SOURCES = [ + 'pipe/tgsi/exec/tgsi_exec.c', + 'pipe/tgsi/exec/tgsi_sse2.c', +] + +TGSIUTIL_SOURCES = [ + 'pipe/tgsi/util/tgsi_build.c', + 'pipe/tgsi/util/tgsi_dump.c', + 'pipe/tgsi/util/tgsi_parse.c', + 'pipe/tgsi/util/tgsi_util.c', +] + +STATECACHE_SOURCES = [ + 'pipe/cso_cache/cso_hash.c', + 'pipe/cso_cache/cso_cache.c', +] + +PIPEUTIL_SOURCES = [ + 'pipe/util/p_tile.c', + 'pipe/util/p_util.c', +] + +STATETRACKER_SOURCES = [ + 'state_tracker/st_atom.c', + 'state_tracker/st_atom_blend.c', + 'state_tracker/st_atom_clip.c', + 'state_tracker/st_atom_constbuf.c', + 'state_tracker/st_atom_depth.c', + 'state_tracker/st_atom_framebuffer.c', + 'state_tracker/st_atom_pixeltransfer.c', + 'state_tracker/st_atom_sampler.c', + 'state_tracker/st_atom_scissor.c', + 'state_tracker/st_atom_shader.c', + 'state_tracker/st_atom_rasterizer.c', + 'state_tracker/st_atom_stipple.c', + 'state_tracker/st_atom_texture.c', + 'state_tracker/st_atom_viewport.c', + 'state_tracker/st_cb_accum.c', + 'state_tracker/st_cb_bufferobjects.c', + 'state_tracker/st_cb_clear.c', + 'state_tracker/st_cb_flush.c', + 'state_tracker/st_cb_drawpixels.c', + 'state_tracker/st_cb_fbo.c', + 'state_tracker/st_cb_feedback.c', + 'state_tracker/st_cb_program.c', + 'state_tracker/st_cb_queryobj.c', + 'state_tracker/st_cb_rasterpos.c', + 'state_tracker/st_cb_readpixels.c', + 'state_tracker/st_cb_strings.c', + 'state_tracker/st_cb_texture.c', + 'state_tracker/st_cache.c', + 'state_tracker/st_context.c', + 'state_tracker/st_debug.c', + 'state_tracker/st_draw.c', + 'state_tracker/st_extensions.c', + 'state_tracker/st_format.c', + 'state_tracker/st_framebuffer.c', + 'state_tracker/st_mesa_to_tgsi.c', + 'state_tracker/st_program.c', + 'state_tracker/st_texture.c', +] + +SHADER_SOURCES = [ + 'shader/arbprogparse.c', + 'shader/arbprogram.c', + 'shader/atifragshader.c', + 'shader/grammar/grammar_mesa.c', + 'shader/nvfragparse.c', + 'shader/nvprogram.c', + 'shader/nvvertparse.c', + 'shader/program.c', + 'shader/prog_cache.c', + 'shader/prog_debug.c', + 'shader/prog_execute.c', + 'shader/prog_instruction.c', + 'shader/prog_parameter.c', + 'shader/prog_print.c', + 'shader/prog_statevars.c', + 'shader/programopt.c', + 'shader/shader_api.c', +] + +SLANG_SOURCES = [ + 'shader/slang/slang_builtin.c', + 'shader/slang/slang_codegen.c', + 'shader/slang/slang_compile.c', + 'shader/slang/slang_compile_function.c', + 'shader/slang/slang_compile_operation.c', + 'shader/slang/slang_compile_struct.c', + 'shader/slang/slang_compile_variable.c', + 'shader/slang/slang_emit.c', + 'shader/slang/slang_ir.c', + 'shader/slang/slang_label.c', + 'shader/slang/slang_library_noise.c', + 'shader/slang/slang_link.c', + 'shader/slang/slang_log.c', + 'shader/slang/slang_mem.c', + 'shader/slang/slang_preprocess.c', + 'shader/slang/slang_print.c', + 'shader/slang/slang_simplify.c', + 'shader/slang/slang_storage.c', + 'shader/slang/slang_typeinfo.c', + 'shader/slang/slang_vartable.c', + 'shader/slang/slang_utility.c', +] + + +####################################################################### +# Assembly sources + +ASM_C_SOURCES = [ + 'x86/common_x86.c', + 'x86/x86.c', + 'x86/3dnow.c', + 'x86/sse.c', + 'x86/rtasm/x86sse.c', + 'sparc/sparc.c', + 'ppc/common_ppc.c', + 'x86-64/x86-64.c', +] + +X86_SOURCES = [ + 'x86/common_x86_asm.S', + 'x86/x86_xform2.S', + 'x86/x86_xform3.S', + 'x86/x86_xform4.S', + 'x86/x86_cliptest.S', + 'x86/mmx_blend.S', + 'x86/3dnow_xform1.S', + 'x86/3dnow_xform2.S', + 'x86/3dnow_xform3.S', + 'x86/3dnow_xform4.S', + 'x86/3dnow_normal.S', + 'x86/sse_xform1.S', + 'x86/sse_xform2.S', + 'x86/sse_xform3.S', + 'x86/sse_xform4.S', + 'x86/sse_normal.S', + 'x86/read_rgba_span_x86.S', +] + +X86_API = [ + 'x86/glapi_x86.S', +] + +X86_64_SOURCES = [ + 'x86-64/xform4.S', +] + +X86_64_API = [ + 'x86-64/glapi_x86-64.S', +] + +SPARC_SOURCES = [ + 'sparc/clip.S', + 'sparc/norm.S', + 'sparc/xform.S', +] + +SPARC_API = [ + 'sparc/glapi_sparc.S', +] + +if x86: + ASM_SOURCES = ASM_C_SOURCES + X86_SOURCES + API_SOURCES = X86_API +else: + ASM_SOURCES = [] + API_SOURCES = [] + + +####################################################################### +# Driver sources + + +X11_DRIVER_SOURCES = [ + 'pipe/xlib/glxapi.c', + 'pipe/xlib/fakeglx.c', + 'pipe/xlib/xfonts.c', + 'pipe/xlib/xm_api.c', + 'pipe/xlib/xm_winsys.c', + 'pipe/xlib/xm_winsys_aub.c', + 'pipe/xlib/brw_aub.c', +] + +OSMESA_DRIVER_SOURCES = [ + 'drivers/osmesa/osmesa.c', +] + +GLIDE_DRIVER_SOURCES = [ + 'drivers/glide/fxapi.c', + 'drivers/glide/fxdd.c', + 'drivers/glide/fxddspan.c', + 'drivers/glide/fxddtex.c', + 'drivers/glide/fxsetup.c', + 'drivers/glide/fxtexman.c', + 'drivers/glide/fxtris.c', + 'drivers/glide/fxvb.c', + 'drivers/glide/fxglidew.c', + 'drivers/glide/fxg.c', +] + +SVGA_DRIVER_SOURCES = [ + 'drivers/svga/svgamesa.c', + 'drivers/svga/svgamesa8.c', + 'drivers/svga/svgamesa15.c', + 'drivers/svga/svgamesa16.c', + 'drivers/svga/svgamesa24.c', + 'drivers/svga/svgamesa32.c', +] + +FBDEV_DRIVER_SOURCES = [ + 'drivers/fbdev/glfbdev.c', +] + + +### All the core C sources + +SOLO_SOURCES = \ + MAIN_SOURCES + \ + MATH_SOURCES + \ + VBO_SOURCES + \ + VF_SOURCES + \ + DRAW_SOURCES + \ + TGSIEXEC_SOURCES + \ + TGSIUTIL_SOURCES + \ + PIPEUTIL_SOURCES + \ + STATECACHE_SOURCES + \ + STATETRACKER_SOURCES + \ + SHADER_SOURCES + \ + ASM_SOURCES + \ + SLANG_SOURCES + +CORE_SOURCES = \ + GLAPI_SOURCES + API_SOURCES + \ + SOLO_SOURCES + +ALL_SOURCES = \ + GLAPI_SOURCES + API_SOURCES + \ + SOLO_SOURCES + \ + ASM_SOURCES + \ + X11_DRIVER_SOURCES + \ + FBDEV_DRIVER_SOURCES + \ + OSMESA_DRIVER_SOURCES + + +###################################################################### +# Gallium sources + +SConscript([ + 'pipe/SConscript', +]) + + +###################################################################### +# libGL + +if not dri: + STAND_ALONE_DRIVER_SOURCES = \ + CORE_SOURCES + \ + X11_DRIVER_SOURCES + + Import( + 'softpipe', + 'i915simple', + 'i965simple' + ) + + pipe_drivers = [ + softpipe, + i965simple + ] + + env.SharedLibrary( + target ='GL', + source = STAND_ALONE_DRIVER_SOURCES, + LIBS = [softpipe, i965simple] + env['LIBS'], + ) + + +###################################################################### +# Driver sources + +if dri: + mesa = env.ConvenienceLibrary( + target = 'mesa', + source = SOLO_SOURCES, + ) + env.Prepend(LIBS = [mesa]) + + SConscript([ + 'drivers/dri/SConscript', + ]) diff --git a/src/mesa/drivers/dri/SConscript b/src/mesa/drivers/dri/SConscript new file mode 100644 index 00000000000..d32bd086696 --- /dev/null +++ b/src/mesa/drivers/dri/SConscript @@ -0,0 +1,48 @@ +Import('*') + +drienv = env.Clone() + +drienv.Replace(CPPPATH = [ + '#src/mesa/drivers/dri/common', + '#include', + '#include/GL/internal', + '#src/mesa', + '#src/mesa/main', + '#src/mesa/glapi', + '#src/mesa/math', + '#src/mesa/transform', + '#src/mesa/shader', + '#src/mesa/swrast', + '#src/mesa/swrast_setup', + '#src/egl/main', + '#src/egl/drivers/dri', +]) + +drienv.ParseConfig('pkg-config --cflags --libs libdrm') + +COMMON_GALLIUM_SOURCES = [ + '../common/utils.c', + '../common/vblank.c', + '../common/dri_util.c', + '../common/xmlconfig.c', +] + +COMMON_BM_SOURCES = [ + '../common/dri_bufmgr.c', + '../common/dri_drmpool.c', +] + +Export([ + 'drienv', + 'COMMON_GALLIUM_SOURCES', + 'COMMON_BM_SOURCES', +]) + +# TODO: Installation +#install: $(LIBNAME) +# $(INSTALL) -d $(DRI_DRIVER_INSTALL_DIR) +# $(INSTALL) -m 755 $(LIBNAME) $(DRI_DRIVER_INSTALL_DIR) + +SConscript([ + 'intel_winsys/SConscript', +]) diff --git a/src/mesa/drivers/dri/intel_winsys/SConscript b/src/mesa/drivers/dri/intel_winsys/SConscript new file mode 100644 index 00000000000..a7cc10450e3 --- /dev/null +++ b/src/mesa/drivers/dri/intel_winsys/SConscript @@ -0,0 +1,41 @@ +Import('*') + +env = drienv.Clone() + +env.Append(CPPPATH = [ + '../intel', + 'server' +]) + +#MINIGLX_SOURCES = server/intel_dri.c + +pipe_drivers = [ + softpipe, + i915simple +] + +DRIVER_SOURCES = [ + 'intel_winsys_pipe.c', + 'intel_winsys_softpipe.c', + 'intel_winsys_i915.c', + 'intel_batchbuffer.c', + 'intel_swapbuffers.c', + 'intel_context.c', + 'intel_lock.c', + 'intel_screen.c', + 'intel_batchpool.c', +] + +sources = \ + COMMON_GALLIUM_SOURCES + \ + COMMON_BM_SOURCES + \ + DRIVER_SOURCES + +# DRIVER_DEFINES = -I../intel $(shell pkg-config libdrm --atleast-version=2.3.1 \ +# && echo "-DDRM_VBLANK_FLIP=DRM_VBLANK_FLIP") + +env.SharedLibrary( + target ='i915tex_dri.so', + source = sources, + LIBS = pipe_drivers + env['LIBS'], +) \ No newline at end of file diff --git a/src/mesa/pipe/SConscript b/src/mesa/pipe/SConscript new file mode 100644 index 00000000000..d9c20e01007 --- /dev/null +++ b/src/mesa/pipe/SConscript @@ -0,0 +1,9 @@ +Import('*') + +#env = env.Clone() + +SConscript([ + 'softpipe/SConscript', + 'i915simple/SConscript', + 'i965simple/SConscript', +]) diff --git a/src/mesa/pipe/i915simple/SConscript b/src/mesa/pipe/i915simple/SConscript new file mode 100644 index 00000000000..f5fb96b995c --- /dev/null +++ b/src/mesa/pipe/i915simple/SConscript @@ -0,0 +1,29 @@ +Import('*') + +env = env.Clone() + +i915simple = env.ConvenienceLibrary( + target = 'i915simple', + source = [ + 'i915_blit.c', + 'i915_clear.c', + 'i915_context.c', + 'i915_debug.c', + 'i915_debug_fp.c', + 'i915_flush.c', + 'i915_fpc_emit.c', + 'i915_fpc_translate.c', + 'i915_prim_emit.c', + 'i915_prim_vbuf.c', + 'i915_state.c', + 'i915_state_derived.c', + 'i915_state_dynamic.c', + 'i915_state_emit.c', + 'i915_state_immediate.c', + 'i915_state_sampler.c', + 'i915_strings.c', + 'i915_surface.c', + 'i915_texture.c', + ]) + +Export('i915simple') diff --git a/src/mesa/pipe/i965simple/SConscript b/src/mesa/pipe/i965simple/SConscript new file mode 100644 index 00000000000..74621de84c9 --- /dev/null +++ b/src/mesa/pipe/i965simple/SConscript @@ -0,0 +1,55 @@ +Import('*') + +env = env.Clone() + +i965simple = env.ConvenienceLibrary( + target = 'i965simple', + source = [ + 'brw_blit.c', + 'brw_cc.c', + 'brw_clip.c', + 'brw_clip_line.c', + 'brw_clip_point.c', + 'brw_clip_state.c', + 'brw_clip_tri.c', + 'brw_clip_util.c', + 'brw_context.c', + 'brw_curbe.c', + 'brw_draw.c', + 'brw_draw_upload.c', + 'brw_eu.c', + 'brw_eu_debug.c', + 'brw_eu_emit.c', + 'brw_eu_util.c', + 'brw_flush.c', + 'brw_gs.c', + 'brw_gs_emit.c', + 'brw_gs_state.c', + 'brw_misc_state.c', + 'brw_sf.c', + 'brw_sf_emit.c', + 'brw_sf_state.c', + 'brw_shader_info.c', + 'brw_state.c', + 'brw_state_batch.c', + 'brw_state_cache.c', + 'brw_state_pool.c', + 'brw_state_upload.c', + 'brw_strings.c', + 'brw_surface.c', + 'brw_tex_layout.c', + 'brw_urb.c', + 'brw_util.c', + 'brw_vs.c', + 'brw_vs_emit.c', + 'brw_vs_state.c', + 'brw_wm.c', + 'brw_wm_decl.c', + 'brw_wm_glsl.c', + 'brw_wm_iz.c', + 'brw_wm_sampler_state.c', + 'brw_wm_state.c', + 'brw_wm_surface_state.c', + ]) + +Export('i965simple') diff --git a/src/mesa/pipe/softpipe/SConscript b/src/mesa/pipe/softpipe/SConscript new file mode 100644 index 00000000000..d581ee8d3ca --- /dev/null +++ b/src/mesa/pipe/softpipe/SConscript @@ -0,0 +1,42 @@ +Import('*') + +env = env.Clone() + +softpipe = env.ConvenienceLibrary( + target = 'softpipe', + source = [ + 'sp_clear.c', + 'sp_context.c', + 'sp_draw_arrays.c', + 'sp_flush.c', + 'sp_prim_setup.c', + 'sp_prim_vbuf.c', + 'sp_quad_alpha_test.c', + 'sp_quad_blend.c', + 'sp_quad_bufloop.c', + 'sp_quad.c', + 'sp_quad_colormask.c', + 'sp_quad_coverage.c', + 'sp_quad_depth_test.c', + 'sp_quad_earlyz.c', + 'sp_quad_fs.c', + 'sp_quad_occlusion.c', + 'sp_quad_output.c', + 'sp_quad_stencil.c', + 'sp_quad_stipple.c', + 'sp_query.c', + 'sp_state_blend.c', + 'sp_state_clip.c', + 'sp_state_derived.c', + 'sp_state_fs.c', + 'sp_state_rasterizer.c', + 'sp_state_sampler.c', + 'sp_state_surface.c', + 'sp_state_vertex.c', + 'sp_surface.c', + 'sp_tex_sample.c', + 'sp_texture.c', + 'sp_tile_cache.c', + ]) + +Export('softpipe') \ No newline at end of file -- cgit v1.2.3 From e11bfc990708326faf3db3db12f7941a1a3c6e0f Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Thu, 31 Jan 2008 14:21:49 +0900 Subject: gallium: Make the build output dir depend on the configuration. The build output dirs mimics the old config names: build/linux build/linux-dri build/linux-dri-x86 build/linux-dri-x86-debug ... --- SConstruct | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-) diff --git a/SConstruct b/SConstruct index db6161ed519..01732b0c528 100644 --- a/SConstruct +++ b/SConstruct @@ -2,6 +2,7 @@ # Top-level SConstruct import os +import os.path import sys @@ -40,7 +41,7 @@ Help(opts.GenerateHelpText(env)) # for debugging #print env.Dump() -if 1: +if 0: # platform will be typically 'posix' or 'win32' platform = env['PLATFORM'] else: @@ -56,7 +57,7 @@ machine = env['machine'] # derived options x86 = machine == 'x86' -gcc = platform == 'posix' +gcc = platform in ('posix', 'linux', 'freebsd', 'darwin') msvc = platform == 'win32' Export([ @@ -202,10 +203,20 @@ createConvenienceLibBuilder(env) ####################################################################### # Invoke SConscripts -# Put build output in a separate dir -# TODO: make build_dir depend on platform and build type (check -# http://www.scons.org/wiki/AdvancedBuildExample for an example) -build_dir = 'build' +# Put build output in a separate dir, which depends on the current configuration +# See also http://www.scons.org/wiki/AdvancedBuildExample +build_topdir = 'build' +build_subdir = platform +if dri: + build_subdir += "-dri" +if x86: + build_subdir += "-x86" +if debug: + build_subdir += "-debug" +build_dir = os.path.join(build_topdir, build_subdir) + +# TODO: Build several variants at the same time? +# http://www.scons.org/wiki/SimultaneousVariantBuilds SConscript( 'src/mesa/SConscript', -- cgit v1.2.3 From 526bed2b956d88e1b371b2e43402f8a160113497 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Thu, 31 Jan 2008 14:26:39 +0900 Subject: gallium: Portability fixes. --- src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c index a2657dac592..b6af7cdedc2 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c @@ -367,7 +367,7 @@ struct mm_pb_manager }; -static inline struct mm_pb_manager * +static INLINE struct mm_pb_manager * mm_pb_manager(struct pb_manager *mgr) { assert(mgr); @@ -385,7 +385,7 @@ struct mm_buffer }; -static inline struct mm_buffer * +static INLINE struct mm_buffer * mm_buffer(struct pb_buffer *buf) { assert(buf); -- cgit v1.2.3 From c7403b18476c9a8f767fa7d33cb1efd643b30dd6 Mon Sep 17 00:00:00 2001 From: Keith Whitwell Date: Thu, 31 Jan 2008 11:57:15 +0000 Subject: tgsi: Use ESI instead of EBX as temp reg on non-win32 --- src/mesa/pipe/tgsi/exec/tgsi_sse2.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/mesa/pipe/tgsi/exec/tgsi_sse2.c b/src/mesa/pipe/tgsi/exec/tgsi_sse2.c index f8660e7ad14..df0c6983013 100755 --- a/src/mesa/pipe/tgsi/exec/tgsi_sse2.c +++ b/src/mesa/pipe/tgsi/exec/tgsi_sse2.c @@ -198,9 +198,15 @@ get_output_base( void ) static struct x86_reg get_temp_base( void ) { +#ifdef WIN32 return x86_make_reg( file_REG32, reg_BX ); +#else + return x86_make_reg( + file_REG32, + reg_SI ); +#endif } static struct x86_reg -- cgit v1.2.3 From 4d3f3f749f0675c07de2809a9c0dc843f29e9873 Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 08:12:47 -0700 Subject: Cell: set GALLIUM_CELL_VS env var to enable SPU-based vertex transformation --- src/mesa/pipe/cell/ppu/cell_context.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_context.c b/src/mesa/pipe/cell/ppu/cell_context.c index 4885cd0d2c7..bbe1fd7a111 100644 --- a/src/mesa/pipe/cell/ppu/cell_context.c +++ b/src/mesa/pipe/cell/ppu/cell_context.c @@ -162,8 +162,12 @@ cell_draw_create(struct cell_context *cell) { struct draw_context *draw = draw_create(); - draw->shader_queue_flush = cell_vertex_shader_queue_flush; - draw->driver_private = cell; + if (getenv("GALLIUM_CELL_VS")) { + /* plug in SPU-based vertex transformation code */ + draw->shader_queue_flush = cell_vertex_shader_queue_flush; + draw->driver_private = cell; + } + return draw; } -- cgit v1.2.3 From 7978c749fb2a267e9575c1280557da4cd33e2380 Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 08:21:38 -0700 Subject: Cell: SIMD-ize const_coeff() --- src/mesa/pipe/cell/spu/spu_tri.c | 24 +++++++++--------------- 1 file changed, 9 insertions(+), 15 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index e436e153ece..08b8bf0c9c7 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -723,24 +723,18 @@ static boolean setup_sort_vertices(const struct vertex_header *v0, /** * Compute a0 for a constant-valued coefficient (GL_FLAT shading). - * The value value comes from vertex->data[slot][i]. - * The result will be put into setup.coef[slot].a0[i]. + * The value value comes from vertex->data[slot]. + * The result will be put into setup.coef[slot].a0. * \param slot which attribute slot - * \param i which component of the slot (0..3) */ -static void const_coeff(uint slot) +static INLINE void const_coeff(uint slot) { - uint i; - ASSERT(slot < PIPE_MAX_SHADER_INPUTS); - - for (i = 0; i < 4; i++) { - setup.coef[slot].dadx.f[i] = 0; - setup.coef[slot].dady.f[i] = 0; - - /* need provoking vertex info! - */ - setup.coef[slot].a0.f[i] = setup.vprovoke->data[slot][i]; - } + setup.coef[slot].dadx.v = (vector float) {0.0, 0.0, 0.0, 0.0}; + setup.coef[slot].dady.v = (vector float) {0.0, 0.0, 0.0, 0.0}; + setup.coef[slot].a0.f[0] = setup.vprovoke->data[slot][0]; + setup.coef[slot].a0.f[1] = setup.vprovoke->data[slot][1]; + setup.coef[slot].a0.f[2] = setup.vprovoke->data[slot][2]; + setup.coef[slot].a0.f[3] = setup.vprovoke->data[slot][3]; } -- cgit v1.2.3 From 6c59de9a7bcc025ba070b854d79bf4fb8dcccabf Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 13:32:29 -0700 Subject: gallium: fix get/put typo regression This came from commit f3aa4de034b0d791ce2e38e8aeb3b3abdb4e3b50 on 1/22/08. Fixes strange Z buffer glitches seen in progs/glsl/texdemo1.c --- src/mesa/pipe/softpipe/sp_tile_cache.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index 451e157abf9..ccf367a5e4f 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -415,8 +415,8 @@ sp_get_cached_tile(struct softpipe_context *softpipe, /* put dirty tile back in framebuffer */ if (tc->depth_stencil) { pipe_put_tile_raw(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); } else { pipe_put_tile_rgba(pipe, ps, @@ -441,9 +441,9 @@ sp_get_cached_tile(struct softpipe_context *softpipe, else { /* get new tile data from surface */ if (tc->depth_stencil) { - pipe_put_tile_raw(pipe, ps, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, - tile->data.depth32, 0/*STRIDE*/); + pipe_get_tile_raw(pipe, ps, + tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->data.depth32, 0/*STRIDE*/); } else { pipe_get_tile_rgba(pipe, ps, -- cgit v1.2.3 From 5ee218a02186188f5819e5a4e1e15296bbd57322 Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 13:36:00 -0700 Subject: gallium: Fix z clear bug when TILE_CLEAR_OPTIMIZATION==0 --- src/mesa/pipe/softpipe/sp_clear.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/pipe/softpipe/sp_clear.c b/src/mesa/pipe/softpipe/sp_clear.c index 571f64b38d7..8d295a30ca6 100644 --- a/src/mesa/pipe/softpipe/sp_clear.c +++ b/src/mesa/pipe/softpipe/sp_clear.c @@ -55,7 +55,9 @@ softpipe_clear(struct pipe_context *pipe, struct pipe_surface *ps, if (ps == sp_tile_cache_get_surface(softpipe->zsbuf_cache)) { sp_tile_cache_clear(softpipe->zsbuf_cache, clearValue); +#if TILE_CLEAR_OPTIMIZATION return; +#endif } for (i = 0; i < softpipe->framebuffer.num_cbufs; i++) { -- cgit v1.2.3 From 26fff001e786d88041d9db4c35949b50849f6a59 Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 13:37:01 -0700 Subject: gallium: comments about fragment Z computation --- src/mesa/pipe/softpipe/sp_quad_fs.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/src/mesa/pipe/softpipe/sp_quad_fs.c b/src/mesa/pipe/softpipe/sp_quad_fs.c index c9cc8afa0ce..90691c6065d 100644 --- a/src/mesa/pipe/softpipe/sp_quad_fs.c +++ b/src/mesa/pipe/softpipe/sp_quad_fs.c @@ -168,6 +168,11 @@ shade_quad( sizeof( quad->outputs.color ) ); } + /* + * XXX the following code for updating quad->outputs.depth + * isn't really needed if we did early z testing. + */ + /* store result Z */ if (qss->depthOutSlot >= 0) { /* output[slot] is new Z */ @@ -181,6 +186,10 @@ shade_quad( uint i; for (i = 0; i < 4; i++) { quad->outputs.depth[i] = machine->Inputs[0].xyzw[2].f[i]; + /* XXX not sure the above line is always correct. The following + * might be better: + quad->outputs.depth[i] = machine->QuadPos.xyzw[2].f[i]; + */ } } -- cgit v1.2.3 From a4c7c8a6ee15d793d08e448f9ca8e2100bbe748c Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 14:05:04 -0700 Subject: Fix problem in mapping vertex program outputs (found with "spring" game engine) If the vertex program writes to an output that's not consumed by the fragment program, map the vp output to an unused slot. --- src/mesa/state_tracker/st_atom_shader.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c index 1ed9333556e..91969185093 100644 --- a/src/mesa/state_tracker/st_atom_shader.c +++ b/src/mesa/state_tracker/st_atom_shader.c @@ -226,9 +226,11 @@ find_translated_vp(struct st_context *st, GLint fpInAttrib = vp_out_to_fp_in(outAttr); if (fpInAttrib >= 0) { GLuint fpInSlot = stfp->input_to_slot[fpInAttrib]; - GLuint vpOutSlot = stfp->fs->state.input_map[fpInSlot]; - xvp->output_to_slot[outAttr] = vpOutSlot; - numVpOuts++; + if (fpInSlot != ~0) { + GLuint vpOutSlot = stfp->fs->state.input_map[fpInSlot]; + xvp->output_to_slot[outAttr] = vpOutSlot; + numVpOuts++; + } } else if (outAttr == VERT_RESULT_PSIZ || outAttr == VERT_RESULT_BFC0 || @@ -247,7 +249,7 @@ find_translated_vp(struct st_context *st, * We could use this info to do dead code elimination in the * vertex program. */ - dummySlot = stfp->num_input_slots; + dummySlot = numVpOuts; /* Map vert program outputs that aren't used to the dummy slot */ for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) { -- cgit v1.2.3 From 9aa37ad5401959fb43d39724d6efb5088087e56e Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 14:35:25 -0700 Subject: gallium: fix problem in which texcoords and varying vars got mapped to the same slot This fixes the glsl/bump.c and glsl/texdemo1.c programs --- src/mesa/state_tracker/st_program.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index 1f1e6500e09..84a90940018 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -33,6 +33,7 @@ #include "main/imports.h" #include "main/mtypes.h" +#include "shader/prog_print.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" @@ -68,6 +69,7 @@ st_translate_vertex_program(struct st_context *st, struct pipe_shader_state vs; const struct cso_vertex_shader *cso; GLuint attr, i; + GLuint num_generic = 0; memset(&vs, 0, sizeof(vs)); @@ -117,7 +119,7 @@ st_translate_vertex_program(struct st_context *st, case VERT_ATTRIB_TEX6: case VERT_ATTRIB_TEX7: vs.input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - vs.input_semantic_index[slot] = attr - VERT_ATTRIB_TEX0; + vs.input_semantic_index[slot] = num_generic++; break; case VERT_ATTRIB_GENERIC0: case VERT_ATTRIB_GENERIC1: @@ -129,7 +131,7 @@ st_translate_vertex_program(struct st_context *st, case VERT_ATTRIB_GENERIC7: assert(attr < VERT_ATTRIB_MAX); vs.input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - vs.input_semantic_index[slot] = attr - VERT_ATTRIB_GENERIC0; + vs.input_semantic_index[slot] = num_generic++; break; default: assert(0); @@ -143,6 +145,7 @@ st_translate_vertex_program(struct st_context *st, vs.output_semantic_index[i] = 0; } + num_generic = 0; /* * Determine number of outputs, the (default) output register * mapping and the semantic information for each output. @@ -207,14 +210,14 @@ st_translate_vertex_program(struct st_context *st, case VERT_RESULT_TEX6: case VERT_RESULT_TEX7: vs.output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - vs.output_semantic_index[slot] = attr - VERT_RESULT_TEX0; + vs.output_semantic_index[slot] = num_generic++; break; case VERT_RESULT_VAR0: /* fall-through */ default: assert(attr - VERT_RESULT_VAR0 < MAX_VARYING); vs.output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - vs.output_semantic_index[slot] = attr - VERT_RESULT_VAR0; + vs.output_semantic_index[slot] = num_generic++; } } } @@ -258,6 +261,9 @@ st_translate_vertex_program(struct st_context *st, cso = st_cached_vs_state(st, &vs); stvp->cso = cso; + if (0) + _mesa_print_program(&stvp->Base.Base); + if (TGSI_DEBUG) tgsi_dump( tokensOut, 0 ); } @@ -286,6 +292,7 @@ st_translate_fragment_program(struct st_context *st, GLuint attr; const GLbitfield inputsRead = stfp->Base.Base.InputsRead; GLuint vslot = 0; + GLuint num_generic = 0; memset(&fs, 0, sizeof(fs)); @@ -338,14 +345,14 @@ st_translate_fragment_program(struct st_context *st, case FRAG_ATTRIB_TEX6: case FRAG_ATTRIB_TEX7: fs.input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - fs.input_semantic_index[slot] = attr - FRAG_ATTRIB_TEX0; + fs.input_semantic_index[slot] = num_generic++; interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE; break; case FRAG_ATTRIB_VAR0: /* fall-through */ default: fs.input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - fs.input_semantic_index[slot] = attr - FRAG_ATTRIB_VAR0; + fs.input_semantic_index[slot] = num_generic++; interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE; } } @@ -415,6 +422,9 @@ st_translate_fragment_program(struct st_context *st, cso = st_cached_fs_state(st, &fs); stfp->fs = cso; + if (0) + _mesa_print_program(&stfp->Base.Base); + if (TGSI_DEBUG) tgsi_dump( tokensOut, 0/*TGSI_DUMP_VERBOSE*/ ); -- cgit v1.2.3 From 41d1179fa68fd8987cd09f26f32416963d235744 Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 31 Jan 2008 17:05:43 -0700 Subject: fix typo --- src/mesa/pipe/p_shader_tokens.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/p_shader_tokens.h b/src/mesa/pipe/p_shader_tokens.h index e9d1d66bdab..3ce35310f6d 100644 --- a/src/mesa/pipe/p_shader_tokens.h +++ b/src/mesa/pipe/p_shader_tokens.h @@ -626,7 +626,7 @@ struct tgsi_src_register_ext /* * If tgsi_src_register_ext::Type is TGSI_SRC_REGISTER_EXT_TYPE_SWZ, - * it should be cast to tgsi_src_register_ext_extswz. + * it should be cast to tgsi_src_register_ext_swz. * * If tgsi_src_register_ext::Type is TGSI_SRC_REGISTER_EXT_TYPE_MOD, * it should be cast to tgsi_src_register_ext_mod. -- cgit v1.2.3 From 0c0c62dd2a7d8d5f8bbd91bc7443f27abb059ad4 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Thu, 31 Jan 2008 17:22:07 -0800 Subject: Fix using "ccache ppu-gcc" for CC and fix parallel builds CC wasn't quoted in a couple places in src/mesa/Makefile. Also, the OSMesa link was missing a dependency. --- src/mesa/Makefile | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/src/mesa/Makefile b/src/mesa/Makefile index b16d74bf493..720f1b2e026 100644 --- a/src/mesa/Makefile +++ b/src/mesa/Makefile @@ -125,24 +125,25 @@ osmesa-only: depend subdirs $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME) # Make the GL library $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(STAND_ALONE_OBJECTS) $(PIPE_LIB) $(CELL_LIB) $(CELL_LIB_SPU) $(LLVM_LIB) @ $(TOP)/bin/mklib -o $(GL_LIB) \ - -linker $(CC) \ + -linker "$(CC)" \ -major $(GL_MAJOR) -minor $(GL_MINOR) -patch $(GL_TINY) \ -install $(TOP)/$(LIB_DIR) \ $(MKLIB_OPTIONS) $(STAND_ALONE_OBJECTS) \ $(PIPE_LIB) $(CELL_LIB) $(CELL_LIB_SPU) $(LLVM_LIB) $(GL_LIB_DEPS) # Make the OSMesa library -$(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME): $(OSMESA_DRIVER_OBJECTS) $(OSMESA16_OBJECTS) +$(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME): $(OSMESA_DRIVER_OBJECTS) \ + $(OSMESA16_OBJECTS) $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) @ if [ "${DRIVER_DIRS}" = "osmesa" ] ; then \ $(TOP)/bin/mklib -o $(OSMESA_LIB) \ - -linker $(CC) \ + -linker "$(CC)" \ -major $(MESA_MAJOR) \ -minor $(MESA_MINOR) -patch $(MESA_TINY) \ -install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \ $(OSMESA_LIB_DEPS) $(OSMESA16_OBJECTS) ; \ else \ $(TOP)/bin/mklib -o $(OSMESA_LIB) \ - -linker $(CC) \ + -linker "$(CC)" \ -major $(MESA_MAJOR) \ -minor $(MESA_MINOR) -patch $(GL_TINY) \ -install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \ -- cgit v1.2.3 From 4f7dcb0e04cbdb95684bf415133c0e7861839f96 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 09:27:57 -0700 Subject: Cell: store current tile status in cur_tile_status_c/z, add TILE_STATUS_GETTING --- src/mesa/pipe/cell/spu/spu_render.c | 36 ++++++++++++++++----- src/mesa/pipe/cell/spu/spu_tile.c | 1 + src/mesa/pipe/cell/spu/spu_tile.h | 8 +++-- src/mesa/pipe/cell/spu/spu_tri.c | 62 ++++++++++++++++++++++++++++++------- src/mesa/pipe/cell/spu/spu_tri.h | 2 +- 5 files changed, 87 insertions(+), 22 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_render.c b/src/mesa/pipe/cell/spu/spu_render.c index f506095116b..ca54a103bdc 100644 --- a/src/mesa/pipe/cell/spu/spu_render.c +++ b/src/mesa/pipe/cell/spu/spu_render.c @@ -95,13 +95,15 @@ static INLINE void get_cz_tiles(uint tx, uint ty) { if (spu.depth_stencil.depth.enabled) { - if (tile_status_z[ty][tx] != TILE_STATUS_CLEAR) { + if (cur_tile_status_z != TILE_STATUS_CLEAR) { get_tile(tx, ty, &ztile, TAG_READ_TILE_Z, 1); + cur_tile_status_z = TILE_STATUS_GETTING; } } - if (tile_status[ty][tx] != TILE_STATUS_CLEAR) { + if (cur_tile_status_c != TILE_STATUS_CLEAR) { get_tile(tx, ty, &ctile, TAG_READ_TILE_COLOR, 0); + cur_tile_status_c = TILE_STATUS_GETTING; } } @@ -112,14 +114,24 @@ get_cz_tiles(uint tx, uint ty) static INLINE void put_cz_tiles(uint tx, uint ty) { - if (tile_status_z[ty][tx] == TILE_STATUS_DIRTY) { + if (cur_tile_status_z == TILE_STATUS_DIRTY) { + /* tile was modified and needs to be written back */ put_tile(tx, ty, &ztile, TAG_WRITE_TILE_Z, 1); - tile_status_z[ty][tx] = TILE_STATUS_DEFINED; + cur_tile_status_z = TILE_STATUS_DEFINED; + } + else if (cur_tile_status_z == TILE_STATUS_GETTING) { + /* tile was never used */ + cur_tile_status_z = TILE_STATUS_DEFINED; } - if (tile_status[ty][tx] == TILE_STATUS_DIRTY) { + if (cur_tile_status_c == TILE_STATUS_DIRTY) { + /* tile was modified and needs to be written back */ put_tile(tx, ty, &ctile, TAG_WRITE_TILE_COLOR, 0); - tile_status[ty][tx] = TILE_STATUS_DEFINED; + cur_tile_status_c = TILE_STATUS_DEFINED; + } + else if (cur_tile_status_c == TILE_STATUS_GETTING) { + /* tile was never used */ + cur_tile_status_c = TILE_STATUS_DEFINED; } } @@ -238,8 +250,13 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) if (!my_tile(tx, ty)) continue; + cur_tile_status_c = tile_status[ty][tx]; + cur_tile_status_z = tile_status_z[ty][tx]; + get_cz_tiles(tx, ty); + uint drawn = 0; + /* loop over tris */ for (j = 0; j < render->num_indexes; j += 3) { const float *v0, *v1, *v2; @@ -248,13 +265,18 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) v1 = (const float *) (vertices + indexes[j+1] * vertex_size); v2 = (const float *) (vertices + indexes[j+2] * vertex_size); - tri_draw(v0, v1, v2, tx, ty); + drawn += tri_draw(v0, v1, v2, tx, ty); } + //printf("SPU %u: drew %u of %u\n", spu.init.id, drawn, render->num_indexes/3); + /* write color/z tiles back to main framebuffer, if dirtied */ put_cz_tiles(tx, ty); wait_put_cz_tiles(); /* XXX seems unnecessary... */ + + tile_status[ty][tx] = cur_tile_status_c; + tile_status_z[ty][tx] = cur_tile_status_z; } if (Debug) diff --git a/src/mesa/pipe/cell/spu/spu_tile.c b/src/mesa/pipe/cell/spu/spu_tile.c index ca1352f9f8d..aea4785bc2f 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.c +++ b/src/mesa/pipe/cell/spu/spu_tile.c @@ -37,6 +37,7 @@ tile_t ztile ALIGN16_ATTRIB; ubyte tile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; ubyte tile_status_z[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; +ubyte cur_tile_status_c, cur_tile_status_z; void diff --git a/src/mesa/pipe/cell/spu/spu_tile.h b/src/mesa/pipe/cell/spu/spu_tile.h index 18d1b3c1173..1f123a2b7b4 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.h +++ b/src/mesa/pipe/cell/spu/spu_tile.h @@ -51,12 +51,16 @@ extern tile_t ztile ALIGN16_ATTRIB; #define TILE_STATUS_CLEAR 1 -#define TILE_STATUS_DEFINED 2 /**< defined pixel data */ -#define TILE_STATUS_DIRTY 3 /**< modified, but not put back yet */ +#define TILE_STATUS_DEFINED 2 /**< defined in FB, but not in local store */ +#define TILE_STATUS_CLEAN 3 /**< in local store, but not changed */ +#define TILE_STATUS_DIRTY 4 /**< modified locally, but not put back yet */ +#define TILE_STATUS_GETTING 5 /**< mfc_get() called but not yet arrived */ extern ubyte tile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; extern ubyte tile_status_z[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; +extern ubyte cur_tile_status_c, cur_tile_status_z; + void get_tile(uint tx, uint ty, tile_t *tile, int tag, int zBuf); diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 08b8bf0c9c7..a32878d9178 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -299,16 +299,23 @@ do_depth_test(int x, int y, unsigned int mask) zvals.v = eval_z((float) x, (float) y); - if (tile_status_z[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { + if (cur_tile_status_c == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_z_tile(&ztile); + cur_tile_status_z = TILE_STATUS_DIRTY; } - else if (tile_status_z[setup.ty][setup.tx] != TILE_STATUS_DIRTY) { + +#if 0 + if (cur_tile_status_z == TILE_STATUS_CLEAR) { + /* now, _really_ clear the tile */ + clear_z_tile(&ztile); + } + else if (cur_tile_status_z != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_Z); } - tile_status_z[setup.ty][setup.tx] = TILE_STATUS_DIRTY; - + cur_tile_status_z = TILE_STATUS_DIRTY; +#endif if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { zvals.v = spu_mul(zvals.v, zscale16.v); @@ -380,6 +387,9 @@ do_depth_test(int x, int y, unsigned int mask) } } + if (mask) + cur_tile_status_z = TILE_STATUS_DIRTY; + return mask; } @@ -397,15 +407,15 @@ do_depth_test_simd(int x, int y, vector unsigned int quadmask) zvals.v = eval_z((float) x, (float) y); - if (tile_status_z[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { + if (cur_tile_status_z == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_z_tile(&ztile); } - else if (tile_status_z[setup.ty][setup.tx] != TILE_STATUS_DIRTY) { + else if (cur_tile_status_z != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_Z); } - tile_status_z[setup.ty][setup.tx] = TILE_STATUS_DIRTY; + cur_tile_status_z = TILE_STATUS_DIRTY; /* XXX fetch Z value sooner to hide latency here */ zmask = spu_cmpgt(ztile.f4[ix][iy].v, zvals.v); @@ -462,15 +472,23 @@ emit_quad( int x, int y, mask_t mask ) if (mask) #endif { - if (tile_status[setup.ty][setup.tx] == TILE_STATUS_CLEAR) { + if (cur_tile_status_c == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_c_tile(&ctile); } - else if (tile_status[setup.ty][setup.tx] != TILE_STATUS_DIRTY) { + +#if 0 + if (cur_tile_status_c == TILE_STATUS_CLEAR) { + /* now, _really_ clear the tile */ + clear_c_tile(&ctile); + cur_tile_status_c = TILE_STATUS_DIRTY; + } + else if (cur_tile_status_c != TILE_STATUS_DIRTY) { /* make sure we've got the tile from main mem */ wait_on_mask(1 << TAG_READ_TILE_COLOR); } - tile_status[setup.ty][setup.tx] = TILE_STATUS_DIRTY; +#endif + cur_tile_status_c = TILE_STATUS_DIRTY; #if SIMD_Z if (spu_extract(mask, 0)) @@ -970,7 +988,7 @@ static void subtriangle( struct edge *eleft, * Draw triangle into tile at (tx, ty) (tile coords) * The tile data should have already been fetched. */ -void +boolean tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) { setup.tx = tx; @@ -985,7 +1003,7 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) if (!setup_sort_vertices((struct vertex_header *) v0, (struct vertex_header *) v1, (struct vertex_header *) v2)) { - return; /* totally clipped */ + return FALSE; /* totally clipped */ } setup_tri_coefficients(); @@ -999,6 +1017,24 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) /* init_constant_attribs( setup ); */ + if (cur_tile_status_c == TILE_STATUS_GETTING) { + /* wait for mfc_get() to complete */ + wait_on_mask(1 << TAG_READ_TILE_COLOR); + cur_tile_status_c = TILE_STATUS_CLEAN; + } + + ASSERT(cur_tile_status_c != TILE_STATUS_DEFINED); + + if (spu.depth_stencil.depth.enabled) { + if (cur_tile_status_z == TILE_STATUS_GETTING) { + /* wait for mfc_get() to complete */ + wait_on_mask(1 << TAG_READ_TILE_Z); + cur_tile_status_z = TILE_STATUS_CLEAN; + } + ASSERT(cur_tile_status_z != TILE_STATUS_DEFINED); + } + + if (setup.oneoverarea < 0.0) { /* emaj on left: */ @@ -1013,4 +1049,6 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) } flush_spans(); + + return TRUE; } diff --git a/src/mesa/pipe/cell/spu/spu_tri.h b/src/mesa/pipe/cell/spu/spu_tri.h index 86c42b63394..aa694dd7c93 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.h +++ b/src/mesa/pipe/cell/spu/spu_tri.h @@ -30,7 +30,7 @@ #define SPU_TRI_H -extern void +extern boolean tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty); -- cgit v1.2.3 From 7a0099b9f3f4cbdb0893a3f11da84326dcf86179 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 13:45:58 -0700 Subject: Cell: implement Z16 and Z32 testing with SIMD instructions. --- src/mesa/pipe/cell/spu/spu_tile.h | 3 +- src/mesa/pipe/cell/spu/spu_tri.c | 222 +++++-------------------------------- src/mesa/pipe/cell/spu/spu_ztest.h | 135 ++++++++++++++++++++++ 3 files changed, 163 insertions(+), 197 deletions(-) create mode 100644 src/mesa/pipe/cell/spu/spu_ztest.h diff --git a/src/mesa/pipe/cell/spu/spu_tile.h b/src/mesa/pipe/cell/spu/spu_tile.h index 1f123a2b7b4..4b1ef2a4c8d 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.h +++ b/src/mesa/pipe/cell/spu/spu_tile.h @@ -42,7 +42,8 @@ typedef union { ushort t16[TILE_SIZE][TILE_SIZE]; uint t32[TILE_SIZE][TILE_SIZE]; - float4 f4[TILE_SIZE/2][TILE_SIZE/2]; + vector unsigned short us8[TILE_SIZE/2][TILE_SIZE/4]; + vector unsigned int ui4[TILE_SIZE/2][TILE_SIZE/2]; } tile_t; diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index a32878d9178..a26a4f098da 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -39,18 +39,11 @@ #include "spu_tile.h" #include "spu_tri.h" +#include "spu_ztest.h" -/* - * If SIMD_Z=1 the Z buffer is floating point and we use vector instructions - * to do Z testing/updating. - */ -#define SIMD_Z 0 -#if SIMD_Z +/** Masks are uint[4] vectors with each element being 0 or 0xffffffff */ typedef vector unsigned int mask_t; -#else -typedef uint mask_t; -#endif /** @@ -282,20 +275,11 @@ pack_colors(uint uicolors[4], const float4 fcolors[4]) } - -static unsigned int -do_depth_test(int x, int y, unsigned int mask) +static INLINE mask_t +do_depth_test(int x, int y, mask_t quadmask) { - static const float4 zscale16 - = {.f={65535.0, 65535.0, 65535.0, 65535.0}}; - static const float4 zscale32 - = {.f={(float)0xffffffff, - (float)0xffffffff, - (float)0xffffffff, - (float)0xffffffff}}; - int ix = x - setup.cliprect_minx; - int iy = y - setup.cliprect_miny; float4 zvals; + mask_t mask; zvals.v = eval_z((float) x, (float) y); @@ -305,129 +289,20 @@ do_depth_test(int x, int y, unsigned int mask) cur_tile_status_z = TILE_STATUS_DIRTY; } -#if 0 - if (cur_tile_status_z == TILE_STATUS_CLEAR) { - /* now, _really_ clear the tile */ - clear_z_tile(&ztile); - } - else if (cur_tile_status_z != TILE_STATUS_DIRTY) { - /* make sure we've got the tile from main mem */ - wait_on_mask(1 << TAG_READ_TILE_Z); - } - cur_tile_status_z = TILE_STATUS_DIRTY; -#endif - if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { - zvals.v = spu_mul(zvals.v, zscale16.v); - if (mask & MASK_TOP_LEFT) { - uint z = (uint) zvals.f[0]; - if (z < ztile.t16[iy][ix]) - ztile.t16[iy][ix] = z; - else - mask &= ~MASK_TOP_LEFT; - } - - if (mask & MASK_TOP_RIGHT) { - uint z = (uint) zvals.f[1]; - if (z < ztile.t16[iy][ix+1]) - ztile.t16[iy][ix+1] = z; - else - mask &= ~MASK_TOP_RIGHT; - } - - if (mask & MASK_BOTTOM_LEFT) { - uint z = (uint) zvals.f[2]; - if (z < ztile.t16[iy+1][ix]) - ztile.t16[iy+1][ix] = z; - else - mask &= ~MASK_BOTTOM_LEFT; - } - - if (mask & MASK_BOTTOM_RIGHT) { - uint z = (uint) zvals.f[3]; - if (z < ztile.t16[iy+1][ix+1]) - ztile.t16[iy+1][ix+1] = z; - else - mask &= ~MASK_BOTTOM_RIGHT; - } + int ix = (x - setup.cliprect_minx) / 4; + int iy = (y - setup.cliprect_miny) / 2; + mask = spu_z16_test_less(zvals.v, &ztile.us8[iy][ix], x>>1, quadmask); } else { - zvals.v = spu_mul(zvals.v, zscale32.v); - ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z32_UNORM); - if (mask & MASK_TOP_LEFT) { - uint z = (uint) zvals.f[0]; - if (z < ztile.t32[iy][ix]) - ztile.t32[iy][ix] = z; - else - mask &= ~MASK_TOP_LEFT; - } - - if (mask & MASK_TOP_RIGHT) { - uint z = (uint) zvals.f[1]; - if (z < ztile.t32[iy][ix+1]) - ztile.t32[iy][ix+1] = z; - else - mask &= ~MASK_TOP_RIGHT; - } - - if (mask & MASK_BOTTOM_LEFT) { - uint z = (uint) zvals.f[2]; - if (z < ztile.t32[iy+1][ix]) - ztile.t32[iy+1][ix] = z; - else - mask &= ~MASK_BOTTOM_LEFT; - } - - if (mask & MASK_BOTTOM_RIGHT) { - uint z = (uint) zvals.f[3]; - if (z < ztile.t32[iy+1][ix+1]) - ztile.t32[iy+1][ix+1] = z; - else - mask &= ~MASK_BOTTOM_RIGHT; - } + int ix = (x - setup.cliprect_minx) / 2; + int iy = (y - setup.cliprect_miny) / 2; + mask = spu_z32_test_less(zvals.v, &ztile.ui4[iy][ix], quadmask); } - - if (mask) - cur_tile_status_z = TILE_STATUS_DIRTY; - return mask; } - - -static vector unsigned int -do_depth_test_simd(int x, int y, vector unsigned int quadmask) -{ - int ix = (x - setup.cliprect_minx) / 2; - int iy = (y - setup.cliprect_miny) / 2; - float4 zvals; - - vector unsigned int zmask; - - zvals.v = eval_z((float) x, (float) y); - - if (cur_tile_status_z == TILE_STATUS_CLEAR) { - /* now, _really_ clear the tile */ - clear_z_tile(&ztile); - } - else if (cur_tile_status_z != TILE_STATUS_DIRTY) { - /* make sure we've got the tile from main mem */ - wait_on_mask(1 << TAG_READ_TILE_Z); - } - cur_tile_status_z = TILE_STATUS_DIRTY; - - /* XXX fetch Z value sooner to hide latency here */ - zmask = spu_cmpgt(ztile.f4[ix][iy].v, zvals.v); - zmask = spu_and(zmask, quadmask); - - ztile.f4[ix][iy].v = spu_sel(ztile.f4[ix][iy].v, zvals.v, zmask); - //ztile.f4[ix][iy].v = spu_sel(zvals.v, ztile.f4[ix][iy].v, mask4); - - return zmask; -} - - /** * Emit a quad (pass to next stage). No clipping is done. */ @@ -461,36 +336,18 @@ emit_quad( int x, int y, mask_t mask ) } if (spu.depth_stencil.depth.enabled) { -#if SIMD_Z - mask = do_depth_test_simd(x, y, mask); -#else mask = do_depth_test(x, y, mask); -#endif } -#if !SIMD_Z - if (mask) -#endif - { - if (cur_tile_status_c == TILE_STATUS_CLEAR) { - /* now, _really_ clear the tile */ - clear_c_tile(&ctile); - } + /* If any bits in mask are set... */ + if (spu_extract(spu_orx(mask), 0)) { -#if 0 if (cur_tile_status_c == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ clear_c_tile(&ctile); - cur_tile_status_c = TILE_STATUS_DIRTY; } - else if (cur_tile_status_c != TILE_STATUS_DIRTY) { - /* make sure we've got the tile from main mem */ - wait_on_mask(1 << TAG_READ_TILE_COLOR); - } -#endif cur_tile_status_c = TILE_STATUS_DIRTY; -#if SIMD_Z if (spu_extract(mask, 0)) ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; if (spu_extract(mask, 1)) @@ -499,20 +356,11 @@ emit_quad( int x, int y, mask_t mask ) ctile.t32[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; if (spu_extract(mask, 3)) ctile.t32[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; -#elif 0 + +#if 0 /* SIMD_Z with swizzled color buffer (someday) */ vector float icolors = *((vector float *) &colors); ctile.f4[iy/2][ix/2].v = spu_sel(ctile.f4[iy/2][ix/2].v, icolors, mask); - -#else - if (mask & MASK_TOP_LEFT) - ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; - if (mask & MASK_TOP_RIGHT) - ctile.t32[iy][ix+1] = colors[QUAD_TOP_RIGHT]; - if (mask & MASK_BOTTOM_LEFT) - ctile.t32[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; - if (mask & MASK_BOTTOM_RIGHT) - ctile.t32[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; #endif } @@ -533,38 +381,20 @@ static INLINE int block( int x ) /** * Compute mask which indicates which pixels in the 2x2 quad are actually inside * the triangle's bounds. - * - * this is pretty nasty... may need to rework flush_spans again to - * fix it, if possible. + * The mask is a uint4 vector and each element will be 0 or 0xffffffff. */ -static mask_t calculate_mask( int x ) +static INLINE mask_t calculate_mask( int x ) { -#if SIMD_Z - uint m0, m1, m2, m3; - - m0 = (x >= setup.span.left[0] && x < setup.span.right[0]) * ~0; - m1 = (x+1 >= setup.span.left[0] && x+1 < setup.span.right[0]) * ~0; - m2 = (x >= setup.span.left[1] && x < setup.span.right[1]) * ~0; - m3 = (x+1 >= setup.span.left[1] && x+1 < setup.span.right[1]) * ~0; - - return (vector unsigned int) {m0, m1, m2, m3}; -#else - unsigned mask = 0x0; - - if (x >= setup.span.left[0] && x < setup.span.right[0]) - mask |= MASK_TOP_LEFT; - - if (x >= setup.span.left[1] && x < setup.span.right[1]) - mask |= MASK_BOTTOM_LEFT; - - if (x+1 >= setup.span.left[0] && x+1 < setup.span.right[0]) - mask |= MASK_TOP_RIGHT; - - if (x+1 >= setup.span.left[1] && x+1 < setup.span.right[1]) - mask |= MASK_BOTTOM_RIGHT; - + /* This is a little tricky. + * Use & instead of && to avoid branches. + * Use negation to convert true/false to ~0/0 values. + */ + mask_t mask; + mask = spu_insert(-((x >= setup.span.left[0]) & (x < setup.span.right[0])), mask, 0); + mask = spu_insert(-((x+1 >= setup.span.left[0]) & (x+1 < setup.span.right[0])), mask, 1); + mask = spu_insert(-((x >= setup.span.left[1]) & (x < setup.span.right[1])), mask, 2); + mask = spu_insert(-((x+1 >= setup.span.left[1]) & (x+1 < setup.span.right[1])), mask, 3); return mask; -#endif } diff --git a/src/mesa/pipe/cell/spu/spu_ztest.h b/src/mesa/pipe/cell/spu/spu_ztest.h new file mode 100644 index 00000000000..5fefb151765 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_ztest.h @@ -0,0 +1,135 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +/** + * Zbuffer/depth test code. + */ + + +#ifndef SPU_ZTEST_H +#define SPU_ZTEST_H + + +#ifdef __SPU__ +#include +#endif + + + +/** + * Perform Z testing for a 16-bit/value Z buffer. + * + * \param zvals vector of four fragment zvalues as floats + * \param zbuf ptr to vector of ushort[8] zbuffer values. Note that this + * contains the Z values for 2 quads, 8 pixels. + * \param x x coordinate of quad (only lsbit is significant) + * \param inMask indicates which fragments in the quad are alive + * \return new mask indicating which fragments are alive after ztest + */ +static INLINE vector unsigned int +spu_z16_test_less(vector float zvals, vector unsigned short *zbuf, + uint x, vector unsigned int inMask) +{ +#define ZERO 0x80 + vector unsigned int zvals_ui4, zbuf_ui4, mask; + + /* convert floats to uints in [0, 65535] */ + zvals_ui4 = spu_convtu(zvals, 32); /* convert to [0, 2^32] */ + zvals_ui4 = spu_rlmask(zvals_ui4, -16); /* right shift 16 */ + + /* XXX this conditional could be removed with a bit of work */ + if (x & 1) { + /* convert zbuffer values from ushorts to uints */ + /* gather lower four ushorts */ + zbuf_ui4 = spu_shuffle((vector unsigned int) *zbuf, + (vector unsigned int) *zbuf, + VEC_LITERAL(vector unsigned char, + ZERO, ZERO, 8, 9, ZERO, ZERO, 10, 11, + ZERO, ZERO, 12, 13, ZERO, ZERO, 14, 15)); + /* mask = (zbuf_ui4 < zvals_ui4) ? ~0 : 0 */ + mask = spu_cmpgt(zbuf_ui4, zvals_ui4); + /* mask &= inMask */ + mask = spu_and(mask, inMask); + /* zbuf = mask ? zval : zbuf */ + zbuf_ui4 = spu_sel(zbuf_ui4, zvals_ui4, mask); + /* convert zbuffer values from uints back to ushorts, preserve lower 4 */ + *zbuf = (vector unsigned short) + spu_shuffle(zbuf_ui4, (vector unsigned int) *zbuf, + VEC_LITERAL(vector unsigned char, + 16, 17, 18, 19, 20, 21, 22, 23, + 2, 3, 6, 7, 10, 11, 14, 15)); + } + else { + /* convert zbuffer values from ushorts to uints */ + /* gather upper four ushorts */ + zbuf_ui4 = spu_shuffle((vector unsigned int) *zbuf, + (vector unsigned int) *zbuf, + VEC_LITERAL(vector unsigned char, + ZERO, ZERO, 0, 1, ZERO, ZERO, 2, 3, + ZERO, ZERO, 4, 5, ZERO, ZERO, 6, 7)); + /* mask = (zbuf_ui4 < zvals_ui4) ? ~0 : 0 */ + mask = spu_cmpgt(zbuf_ui4, zvals_ui4); + /* mask &= inMask */ + mask = spu_and(mask, inMask); + /* zbuf = mask ? zval : zbuf */ + zbuf_ui4 = spu_sel(zbuf_ui4, zvals_ui4, mask); + /* convert zbuffer values from uints back to ushorts, preserve upper 4 */ + *zbuf = (vector unsigned short) + spu_shuffle(zbuf_ui4, (vector unsigned int) *zbuf, + VEC_LITERAL(vector unsigned char, + 2, 3, 6, 7, 10, 11, 14, 15, + 24, 25, 26, 27, 28, 29, 30, 31)); + } + return mask; +#undef ZERO +} + + +/** + * As above, but Zbuffer values as 32-bit uints + */ +static INLINE vector unsigned int +spu_z32_test_less(vector float zvals, vector unsigned int *zbuf_ptr, + vector unsigned int inMask) +{ + vector unsigned int zvals_ui4, mask, zbuf = *zbuf_ptr; + + /* convert floats to uints in [0, 0xffffffff] */ + zvals_ui4 = spu_convtu(zvals, 32); + /* mask = (zbuf < zvals_ui4) ? ~0 : 0 */ + mask = spu_cmpgt(zbuf, zvals_ui4); + /* mask &= inMask */ + mask = spu_and(mask, inMask); + /* zbuf = mask ? zval : zbuf */ + *zbuf_ptr = spu_sel(zbuf, zvals_ui4, mask); + + return mask; +} + + +#endif /* SPU_ZTEST_H */ -- cgit v1.2.3 From a3d5d7067e9928d64fef08893c0a0368c8d83e45 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 13:49:51 -0700 Subject: Cell: rename fields of the tile_t union --- src/mesa/pipe/cell/spu/spu_main.c | 8 ++++++++ src/mesa/pipe/cell/spu/spu_texture.c | 6 +++--- src/mesa/pipe/cell/spu/spu_tile.c | 4 ++-- src/mesa/pipe/cell/spu/spu_tile.h | 18 +++++------------- src/mesa/pipe/cell/spu/spu_tri.c | 8 ++++---- 5 files changed, 22 insertions(+), 22 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index d6393048f56..7d6e910ad55 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -36,6 +36,7 @@ #include "spu_render.h" #include "spu_texture.h" #include "spu_tile.h" +//#include "spu_test.h" #include "spu_vertex_shader.h" #include "pipe/cell/common.h" #include "pipe/p_defines.h" @@ -495,6 +496,7 @@ one_time_init(void) } + /* In some versions of the SDK the SPE main takes 'unsigned long' as a * parameter. In others it takes 'unsigned long long'. Use a define to * select between the two. @@ -515,6 +517,8 @@ main(main_param_t speid, main_param_t argp) (void) speid; + ASSERT(sizeof(tile_t) == TILE_SIZE * TILE_SIZE * 4); + one_time_init(); if (Debug) @@ -528,6 +532,10 @@ main(main_param_t speid, main_param_t argp) 0 /* rid */); wait_on_mask( 1 << tag ); +#if 0 + if (spu.init.id==0) + spu_test_misc(); +#endif main_loop(); diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 7a1ca097c0a..c1dc6bfe902 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -97,10 +97,10 @@ get_tex_tile(uint i, uint j) spu.init.id, src, tex_tiles[pos].t32); #endif - ASSERT_ALIGN16(tex_tiles[pos].t32); + ASSERT_ALIGN16(tex_tiles[pos].ui); ASSERT_ALIGN16(src); - mfc_get(tex_tiles[pos].t32, /* dest */ + mfc_get(tex_tiles[pos].ui, /* dest */ (unsigned int) src, bytes_per_tile, /* size */ TAG_TEXTURE_TILE, @@ -134,6 +134,6 @@ sample_texture(float4 texcoord) uint i = (uint) (texcoord.f[0] * spu.texture.width) % spu.texture.width; uint j = (uint) (texcoord.f[1] * spu.texture.height) % spu.texture.height; uint pos = get_tex_tile(i, j); - uint texel = tex_tiles[pos].t32[j % TILE_SIZE][i % TILE_SIZE]; + uint texel = tex_tiles[pos].ui[j % TILE_SIZE][i % TILE_SIZE]; return texel; } diff --git a/src/mesa/pipe/cell/spu/spu_tile.c b/src/mesa/pipe/cell/spu/spu_tile.c index aea4785bc2f..fd65c2b49c4 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.c +++ b/src/mesa/pipe/cell/spu/spu_tile.c @@ -56,7 +56,7 @@ get_tile(uint tx, uint ty, tile_t *tile, int tag, int zBuf) printf("get_tile: dest: %p src: 0x%x size: %d\n", tile, (unsigned int) src, bytesPerTile); */ - mfc_get(tile->t32, /* dest in local memory */ + mfc_get(tile->ui, /* dest in local memory */ (unsigned int) src, /* src in main memory */ bytesPerTile, tag, @@ -82,7 +82,7 @@ put_tile(uint tx, uint ty, const tile_t *tile, int tag, int zBuf) spu.init.id, tile, (unsigned int) dst, bytesPerTile); */ - mfc_put((void *) tile->t32, /* src in local memory */ + mfc_put((void *) tile->ui, /* src in local memory */ (unsigned int) dst, /* dst in main memory */ bytesPerTile, tag, diff --git a/src/mesa/pipe/cell/spu/spu_tile.h b/src/mesa/pipe/cell/spu/spu_tile.h index 4b1ef2a4c8d..85a0d558079 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.h +++ b/src/mesa/pipe/cell/spu/spu_tile.h @@ -40,8 +40,8 @@ typedef union { - ushort t16[TILE_SIZE][TILE_SIZE]; - uint t32[TILE_SIZE][TILE_SIZE]; + ushort us[TILE_SIZE][TILE_SIZE]; + uint ui[TILE_SIZE][TILE_SIZE]; vector unsigned short us8[TILE_SIZE/2][TILE_SIZE/4]; vector unsigned int ui4[TILE_SIZE/2][TILE_SIZE/2]; } tile_t; @@ -74,7 +74,7 @@ put_tile(uint tx, uint ty, const tile_t *tile, int tag, int zBuf); static INLINE void clear_c_tile(tile_t *ctile) { - memset32((uint*) ctile->t32, + memset32((uint*) ctile->ui, spu.fb.color_clear_value, TILE_SIZE * TILE_SIZE); } @@ -84,23 +84,15 @@ static INLINE void clear_z_tile(tile_t *ztile) { if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { - memset16((ushort*) ztile->t16, + memset16((ushort*) ztile->us, spu.fb.depth_clear_value, TILE_SIZE * TILE_SIZE); } else { ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z32_UNORM); -#if SIMD_Z - union fi z; - z.f = 1.0; - memset32((uint*) ztile->t32, - z.i,/*spu.fb.depth_clear_value,*/ - TILE_SIZE * TILE_SIZE); -#else - memset32((uint*) ztile->t32, + memset32((uint*) ztile->ui, spu.fb.depth_clear_value, TILE_SIZE * TILE_SIZE); -#endif } } diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index a26a4f098da..b04b6841c04 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -349,13 +349,13 @@ emit_quad( int x, int y, mask_t mask ) cur_tile_status_c = TILE_STATUS_DIRTY; if (spu_extract(mask, 0)) - ctile.t32[iy][ix] = colors[QUAD_TOP_LEFT]; + ctile.ui[iy][ix] = colors[QUAD_TOP_LEFT]; if (spu_extract(mask, 1)) - ctile.t32[iy][ix+1] = colors[QUAD_TOP_RIGHT]; + ctile.ui[iy][ix+1] = colors[QUAD_TOP_RIGHT]; if (spu_extract(mask, 2)) - ctile.t32[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; + ctile.ui[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; if (spu_extract(mask, 3)) - ctile.t32[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; + ctile.ui[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; #if 0 /* SIMD_Z with swizzled color buffer (someday) */ -- cgit v1.2.3 From 1b6b5db4e2e891c62cfb868f2c6ae7f27b29709d Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 14:02:22 -0700 Subject: Cell: move ztest before color interp/packing --- src/mesa/pipe/cell/spu/spu_tri.c | 43 ++++++++++++++++++++-------------------- 1 file changed, 21 insertions(+), 22 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index b04b6841c04..ae8fd17cc61 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -316,24 +316,6 @@ emit_quad( int x, int y, mask_t mask ) setup.quad.mask = mask; sp->quad.first->run(sp->quad.first, &setup.quad); #else - /* Cell: "write" quad fragments to the tile by setting prim color */ - const int ix = x - setup.cliprect_minx; - const int iy = y - setup.cliprect_miny; - uint colors[4]; /* indexed by QUAD_x */ - - if (spu.texture.start) { - float4 texcoords[4]; - uint i; - eval_coeff(2, (float) x, (float) y, texcoords); - for (i = 0; i < 4; i++) { - colors[i] = sample_texture(texcoords[i]); - } - } - else { - float4 fcolors[4]; - eval_coeff(1, (float) x, (float) y, fcolors); - pack_colors(colors, fcolors); - } if (spu.depth_stencil.depth.enabled) { mask = do_depth_test(x, y, mask); @@ -341,6 +323,23 @@ emit_quad( int x, int y, mask_t mask ) /* If any bits in mask are set... */ if (spu_extract(spu_orx(mask), 0)) { + const int ix = x - setup.cliprect_minx; + const int iy = y - setup.cliprect_miny; + uint colors[4]; /* indexed by QUAD_x */ + + if (spu.texture.start) { + float4 texcoords[4]; + uint i; + eval_coeff(2, (float) x, (float) y, texcoords); + for (i = 0; i < 4; i++) { + colors[i] = sample_texture(texcoords[i]); + } + } + else { + float4 fcolors[4]; + eval_coeff(1, (float) x, (float) y, fcolors); + pack_colors(colors, fcolors); + } if (cur_tile_status_c == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ @@ -348,6 +347,7 @@ emit_quad( int x, int y, mask_t mask ) } cur_tile_status_c = TILE_STATUS_DIRTY; +#if 1 if (spu_extract(mask, 0)) ctile.ui[iy][ix] = colors[QUAD_TOP_LEFT]; if (spu_extract(mask, 1)) @@ -356,11 +356,10 @@ emit_quad( int x, int y, mask_t mask ) ctile.ui[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; if (spu_extract(mask, 3)) ctile.ui[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; - -#if 0 +#else /* SIMD_Z with swizzled color buffer (someday) */ - vector float icolors = *((vector float *) &colors); - ctile.f4[iy/2][ix/2].v = spu_sel(ctile.f4[iy/2][ix/2].v, icolors, mask); + vector unsigned int uicolors = *((vector unsigned int *) &colors); + ctile.ui4[iy/2][ix/2] = spu_sel(ctile.ui4[iy/2][ix/2], uicolors, mask); #endif } -- cgit v1.2.3 From b25f1244ab8b082c01e45b1068c233f4c2b21fb3 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 15:20:07 -0700 Subject: Cell: remove commands from top-level while loop which should only appear in batch buffers --- src/mesa/pipe/cell/spu/spu_main.c | 20 ++++---------------- 1 file changed, 4 insertions(+), 16 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 7d6e910ad55..1760de02b71 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -447,34 +447,22 @@ main_loop(void) 0 /* rid */); wait_on_mask( 1 << tag ); + /* + * NOTE: most commands should be contained in a batch buffer + */ + switch (opcode & CELL_CMD_OPCODE_MASK) { case CELL_CMD_EXIT: if (Debug) printf("SPU %u: EXIT\n", spu.init.id); exitFlag = 1; break; - case CELL_CMD_STATE_FRAMEBUFFER: - cmd_state_framebuffer(&cmd.fb); - break; - case CELL_CMD_CLEAR_SURFACE: - cmd_clear_surface(&cmd.clear); - break; - case CELL_CMD_RENDER: - { - uint pos_incr; - cmd_render(&cmd.render, &pos_incr); - assert(pos_incr == 0); - } - break; case CELL_CMD_VS_EXECUTE: spu_execute_vertex_shader(&draw, &cmd.vs); break; case CELL_CMD_BATCH: cmd_batch(opcode); break; - case CELL_CMD_FINISH: - cmd_finish(); - break; default: printf("Bad opcode!\n"); } -- cgit v1.2.3 From 4fa69471c40832967d106a39d6cca9b872609fbd Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 15:24:00 -0700 Subject: Cell: deprecate some use of struct cell_command - it should go away completely Also, remove ALIGN16_ATTRIB from structs that no longer need it. --- src/mesa/pipe/cell/common.h | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 05aeed83ab5..7e193f31be6 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -105,7 +105,7 @@ struct cell_command_framebuffer int width, height; void *color_start, *depth_start; enum pipe_format color_format, depth_format; -} ALIGN16_ATTRIB; +}; /** @@ -116,7 +116,7 @@ struct cell_command_clear_surface uint opcode; uint surface; /**< Temporary: 0=color, 1=Z */ uint value; -} ALIGN16_ATTRIB; +}; /** @@ -173,7 +173,7 @@ struct cell_command_render uint dummy3; uint min_index; boolean inline_verts; -} ALIGN16_ATTRIB; +}; struct cell_command_release_verts @@ -191,11 +191,14 @@ struct cell_command_texture /** XXX unions don't seem to work */ +/* XXX this should go away; all commands should be placed in batch buffers */ struct cell_command { +#if 0 struct cell_command_framebuffer fb; struct cell_command_clear_surface clear; struct cell_command_render render; +#endif struct cell_command_vs vs; } ALIGN16_ATTRIB; -- cgit v1.2.3 From 245c6a4cd5753ce4ef0e5bfa739e5e6ac4e9ae4e Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 15:33:53 -0700 Subject: Cell: rename/move global vars Put tile-related globals into spu_global struct. Rename c/ztile fields to be more consistant. --- src/mesa/pipe/cell/spu/spu_main.c | 28 +++++++++++++------------- src/mesa/pipe/cell/spu/spu_main.h | 32 +++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_render.c | 40 ++++++++++++++++++------------------- src/mesa/pipe/cell/spu/spu_tile.c | 11 +--------- src/mesa/pipe/cell/spu/spu_tile.h | 27 ------------------------- src/mesa/pipe/cell/spu/spu_tri.c | 38 +++++++++++++++++------------------ 6 files changed, 86 insertions(+), 90 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 1760de02b71..8e3987f6ef5 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -92,24 +92,24 @@ really_clear_tiles(uint surfaceIndex) uint i; if (surfaceIndex == 0) { - clear_c_tile(&ctile); + clear_c_tile(&spu.ctile); for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) { uint tx = i % spu.fb.width_tiles; uint ty = i / spu.fb.width_tiles; - if (tile_status[ty][tx] == TILE_STATUS_CLEAR) { - put_tile(tx, ty, &ctile, TAG_SURFACE_CLEAR, 0); + if (spu.ctile_status[ty][tx] == TILE_STATUS_CLEAR) { + put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 0); } } } else { - clear_z_tile(&ztile); + clear_z_tile(&spu.ztile); for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) { uint tx = i % spu.fb.width_tiles; uint ty = i / spu.fb.width_tiles; - if (tile_status_z[ty][tx] == TILE_STATUS_CLEAR) - put_tile(tx, ty, &ctile, TAG_SURFACE_CLEAR, 1); + if (spu.ztile_status[ty][tx] == TILE_STATUS_CLEAR) + put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 1); } } @@ -133,11 +133,11 @@ cmd_clear_surface(const struct cell_command_clear_surface *clear) #if CLEAR_OPT /* set all tile's status to CLEAR */ if (clear->surface == 0) { - memset(tile_status, TILE_STATUS_CLEAR, sizeof(tile_status)); + memset(spu.ctile_status, TILE_STATUS_CLEAR, sizeof(spu.ctile_status)); spu.fb.color_clear_value = clear->value; } else { - memset(tile_status_z, TILE_STATUS_CLEAR, sizeof(tile_status_z)); + memset(spu.ztile_status, TILE_STATUS_CLEAR, sizeof(spu.ztile_status)); spu.fb.depth_clear_value = clear->value; } return; @@ -145,11 +145,11 @@ cmd_clear_surface(const struct cell_command_clear_surface *clear) if (clear->surface == 0) { spu.fb.color_clear_value = clear->value; - clear_c_tile(&ctile); + clear_c_tile(&spu.ctile); } else { spu.fb.depth_clear_value = clear->value; - clear_z_tile(&ztile); + clear_z_tile(&spu.ztile); } /* @@ -161,9 +161,9 @@ cmd_clear_surface(const struct cell_command_clear_surface *clear) uint tx = i % spu.fb.width_tiles; uint ty = i / spu.fb.width_tiles; if (clear->surface == 0) - put_tile(tx, ty, &ctile, TAG_SURFACE_CLEAR, 0); + put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 0); else - put_tile(tx, ty, &ztile, TAG_SURFACE_CLEAR, 1); + put_tile(tx, ty, &spu.ztile, TAG_SURFACE_CLEAR, 1); /* XXX we don't want this here, but it fixes bad tile results */ } @@ -478,8 +478,8 @@ main_loop(void) static void one_time_init(void) { - memset(tile_status, TILE_STATUS_DEFINED, sizeof(tile_status)); - memset(tile_status_z, TILE_STATUS_DEFINED, sizeof(tile_status_z)); + memset(spu.ctile_status, TILE_STATUS_DEFINED, sizeof(spu.ctile_status)); + memset(spu.ztile_status, TILE_STATUS_DEFINED, sizeof(spu.ztile_status)); invalidate_tex_cache(); } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 8be5268f52d..cce5e708021 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -36,6 +36,11 @@ #include "pipe/p_state.h" + +#define MAX_WIDTH 1024 +#define MAX_HEIGHT 1024 + + typedef union { vector float v; @@ -43,6 +48,21 @@ typedef union } float4; +typedef union { + ushort us[TILE_SIZE][TILE_SIZE]; + uint ui[TILE_SIZE][TILE_SIZE]; + vector unsigned short us8[TILE_SIZE/2][TILE_SIZE/4]; + vector unsigned int ui4[TILE_SIZE/2][TILE_SIZE/2]; +} tile_t; + + +#define TILE_STATUS_CLEAR 1 +#define TILE_STATUS_DEFINED 2 /**< defined in FB, but not in local store */ +#define TILE_STATUS_CLEAN 3 /**< in local store, but not changed */ +#define TILE_STATUS_DIRTY 4 /**< modified locally, but not put back yet */ +#define TILE_STATUS_GETTING 5 /**< mfc_get() called but not yet arrived */ + + struct spu_framebuffer { void *color_start; /**< addr of color surface in main memory */ void *depth_start; /**< addr of depth surface in main memory */ @@ -75,6 +95,18 @@ struct spu_global /* XXX more state to come */ + + /** current color and Z tiles */ + tile_t ctile ALIGN16_ATTRIB; + tile_t ztile ALIGN16_ATTRIB; + + /** Current tiles' status */ + ubyte cur_ctile_status, cur_ztile_status; + + /** Status of all tiles in framebuffer */ + ubyte ctile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; + ubyte ztile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; + } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_render.c b/src/mesa/pipe/cell/spu/spu_render.c index ca54a103bdc..ab711d67feb 100644 --- a/src/mesa/pipe/cell/spu/spu_render.c +++ b/src/mesa/pipe/cell/spu/spu_render.c @@ -95,15 +95,15 @@ static INLINE void get_cz_tiles(uint tx, uint ty) { if (spu.depth_stencil.depth.enabled) { - if (cur_tile_status_z != TILE_STATUS_CLEAR) { - get_tile(tx, ty, &ztile, TAG_READ_TILE_Z, 1); - cur_tile_status_z = TILE_STATUS_GETTING; + if (spu.cur_ztile_status != TILE_STATUS_CLEAR) { + get_tile(tx, ty, &spu.ztile, TAG_READ_TILE_Z, 1); + spu.cur_ztile_status = TILE_STATUS_GETTING; } } - if (cur_tile_status_c != TILE_STATUS_CLEAR) { - get_tile(tx, ty, &ctile, TAG_READ_TILE_COLOR, 0); - cur_tile_status_c = TILE_STATUS_GETTING; + if (spu.cur_ctile_status != TILE_STATUS_CLEAR) { + get_tile(tx, ty, &spu.ctile, TAG_READ_TILE_COLOR, 0); + spu.cur_ctile_status = TILE_STATUS_GETTING; } } @@ -114,24 +114,24 @@ get_cz_tiles(uint tx, uint ty) static INLINE void put_cz_tiles(uint tx, uint ty) { - if (cur_tile_status_z == TILE_STATUS_DIRTY) { + if (spu.cur_ztile_status == TILE_STATUS_DIRTY) { /* tile was modified and needs to be written back */ - put_tile(tx, ty, &ztile, TAG_WRITE_TILE_Z, 1); - cur_tile_status_z = TILE_STATUS_DEFINED; + put_tile(tx, ty, &spu.ztile, TAG_WRITE_TILE_Z, 1); + spu.cur_ztile_status = TILE_STATUS_DEFINED; } - else if (cur_tile_status_z == TILE_STATUS_GETTING) { + else if (spu.cur_ztile_status == TILE_STATUS_GETTING) { /* tile was never used */ - cur_tile_status_z = TILE_STATUS_DEFINED; + spu.cur_ztile_status = TILE_STATUS_DEFINED; } - if (cur_tile_status_c == TILE_STATUS_DIRTY) { + if (spu.cur_ctile_status == TILE_STATUS_DIRTY) { /* tile was modified and needs to be written back */ - put_tile(tx, ty, &ctile, TAG_WRITE_TILE_COLOR, 0); - cur_tile_status_c = TILE_STATUS_DEFINED; + put_tile(tx, ty, &spu.ctile, TAG_WRITE_TILE_COLOR, 0); + spu.cur_ctile_status = TILE_STATUS_DEFINED; } - else if (cur_tile_status_c == TILE_STATUS_GETTING) { + else if (spu.cur_ctile_status == TILE_STATUS_GETTING) { /* tile was never used */ - cur_tile_status_c = TILE_STATUS_DEFINED; + spu.cur_ctile_status = TILE_STATUS_DEFINED; } } @@ -250,8 +250,8 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) if (!my_tile(tx, ty)) continue; - cur_tile_status_c = tile_status[ty][tx]; - cur_tile_status_z = tile_status_z[ty][tx]; + spu.cur_ctile_status = spu.ctile_status[ty][tx]; + spu.cur_ztile_status = spu.ztile_status[ty][tx]; get_cz_tiles(tx, ty); @@ -275,8 +275,8 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) wait_put_cz_tiles(); /* XXX seems unnecessary... */ - tile_status[ty][tx] = cur_tile_status_c; - tile_status_z[ty][tx] = cur_tile_status_z; + spu.ctile_status[ty][tx] = spu.cur_ctile_status; + spu.ztile_status[ty][tx] = spu.cur_ztile_status; } if (Debug) diff --git a/src/mesa/pipe/cell/spu/spu_tile.c b/src/mesa/pipe/cell/spu/spu_tile.c index fd65c2b49c4..12dc2463283 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.c +++ b/src/mesa/pipe/cell/spu/spu_tile.c @@ -28,16 +28,7 @@ #include "spu_tile.h" - - - -tile_t ctile ALIGN16_ATTRIB; -tile_t ztile ALIGN16_ATTRIB; - -ubyte tile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; -ubyte tile_status_z[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; - -ubyte cur_tile_status_c, cur_tile_status_z; +#include "spu_main.h" void diff --git a/src/mesa/pipe/cell/spu/spu_tile.h b/src/mesa/pipe/cell/spu/spu_tile.h index 85a0d558079..e53340a55a4 100644 --- a/src/mesa/pipe/cell/spu/spu_tile.h +++ b/src/mesa/pipe/cell/spu/spu_tile.h @@ -35,33 +35,6 @@ #include "pipe/cell/common.h" -#define MAX_WIDTH 1024 -#define MAX_HEIGHT 1024 - - -typedef union { - ushort us[TILE_SIZE][TILE_SIZE]; - uint ui[TILE_SIZE][TILE_SIZE]; - vector unsigned short us8[TILE_SIZE/2][TILE_SIZE/4]; - vector unsigned int ui4[TILE_SIZE/2][TILE_SIZE/2]; -} tile_t; - - -extern tile_t ctile ALIGN16_ATTRIB; -extern tile_t ztile ALIGN16_ATTRIB; - - -#define TILE_STATUS_CLEAR 1 -#define TILE_STATUS_DEFINED 2 /**< defined in FB, but not in local store */ -#define TILE_STATUS_CLEAN 3 /**< in local store, but not changed */ -#define TILE_STATUS_DIRTY 4 /**< modified locally, but not put back yet */ -#define TILE_STATUS_GETTING 5 /**< mfc_get() called but not yet arrived */ - -extern ubyte tile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; -extern ubyte tile_status_z[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; - -extern ubyte cur_tile_status_c, cur_tile_status_z; - void get_tile(uint tx, uint ty, tile_t *tile, int tag, int zBuf); diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index ae8fd17cc61..6f61a3d816d 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -283,21 +283,21 @@ do_depth_test(int x, int y, mask_t quadmask) zvals.v = eval_z((float) x, (float) y); - if (cur_tile_status_c == TILE_STATUS_CLEAR) { + if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ - clear_z_tile(&ztile); - cur_tile_status_z = TILE_STATUS_DIRTY; + clear_z_tile(&spu.ztile); + spu.cur_ztile_status = TILE_STATUS_DIRTY; } if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { int ix = (x - setup.cliprect_minx) / 4; int iy = (y - setup.cliprect_miny) / 2; - mask = spu_z16_test_less(zvals.v, &ztile.us8[iy][ix], x>>1, quadmask); + mask = spu_z16_test_less(zvals.v, &spu.ztile.us8[iy][ix], x>>1, quadmask); } else { int ix = (x - setup.cliprect_minx) / 2; int iy = (y - setup.cliprect_miny) / 2; - mask = spu_z32_test_less(zvals.v, &ztile.ui4[iy][ix], quadmask); + mask = spu_z32_test_less(zvals.v, &spu.ztile.ui4[iy][ix], quadmask); } return mask; } @@ -341,25 +341,25 @@ emit_quad( int x, int y, mask_t mask ) pack_colors(colors, fcolors); } - if (cur_tile_status_c == TILE_STATUS_CLEAR) { + if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { /* now, _really_ clear the tile */ - clear_c_tile(&ctile); + clear_c_tile(&spu.ctile); } - cur_tile_status_c = TILE_STATUS_DIRTY; + spu.cur_ctile_status = TILE_STATUS_DIRTY; #if 1 if (spu_extract(mask, 0)) - ctile.ui[iy][ix] = colors[QUAD_TOP_LEFT]; + spu.ctile.ui[iy][ix] = colors[QUAD_TOP_LEFT]; if (spu_extract(mask, 1)) - ctile.ui[iy][ix+1] = colors[QUAD_TOP_RIGHT]; + spu.ctile.ui[iy][ix+1] = colors[QUAD_TOP_RIGHT]; if (spu_extract(mask, 2)) - ctile.ui[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; + spu.ctile.ui[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; if (spu_extract(mask, 3)) - ctile.ui[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; + spu.ctile.ui[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; #else /* SIMD_Z with swizzled color buffer (someday) */ vector unsigned int uicolors = *((vector unsigned int *) &colors); - ctile.ui4[iy/2][ix/2] = spu_sel(ctile.ui4[iy/2][ix/2], uicolors, mask); + spu.ctile.ui4[iy/2][ix/2] = spu_sel(spu.ctile.ui4[iy/2][ix/2], uicolors, mask); #endif } @@ -846,21 +846,21 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) /* init_constant_attribs( setup ); */ - if (cur_tile_status_c == TILE_STATUS_GETTING) { + if (spu.cur_ctile_status == TILE_STATUS_GETTING) { /* wait for mfc_get() to complete */ wait_on_mask(1 << TAG_READ_TILE_COLOR); - cur_tile_status_c = TILE_STATUS_CLEAN; + spu.cur_ctile_status = TILE_STATUS_CLEAN; } - ASSERT(cur_tile_status_c != TILE_STATUS_DEFINED); + ASSERT(spu.cur_ctile_status != TILE_STATUS_DEFINED); if (spu.depth_stencil.depth.enabled) { - if (cur_tile_status_z == TILE_STATUS_GETTING) { + if (spu.cur_ztile_status == TILE_STATUS_GETTING) { /* wait for mfc_get() to complete */ wait_on_mask(1 << TAG_READ_TILE_Z); - cur_tile_status_z = TILE_STATUS_CLEAN; + spu.cur_ztile_status = TILE_STATUS_CLEAN; } - ASSERT(cur_tile_status_z != TILE_STATUS_DEFINED); + ASSERT(spu.cur_ztile_status != TILE_STATUS_DEFINED); } -- cgit v1.2.3 From de5d995201d617aca729efbc2821efde4b05685f Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 15:45:02 -0700 Subject: Cell: New color packing functions (A8R8G8B8 and B8G8R8A8) --- src/mesa/pipe/cell/spu/spu_colorpack.h | 60 ++++++++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_tri.c | 22 +++++-------- 2 files changed, 69 insertions(+), 13 deletions(-) create mode 100644 src/mesa/pipe/cell/spu/spu_colorpack.h diff --git a/src/mesa/pipe/cell/spu/spu_colorpack.h b/src/mesa/pipe/cell/spu/spu_colorpack.h new file mode 100644 index 00000000000..56709bd9f38 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_colorpack.h @@ -0,0 +1,60 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + + +#ifndef SPU_COLORPACK_H +#define SPU_COLORPACK_H + + +#include +#include + + +static INLINE unsigned int +spu_pack_A8R8G8B8(vector float rgba) +{ + vector unsigned int out = spu_convtu(rgba, 32); + out = spu_shuffle(out, out, VEC_LITERAL(vector unsigned char, + 12, 0, 4, 8, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0)); + return spu_extract(out, 0); +} + + +static INLINE unsigned int +spu_pack_B8G8R8A8(vector float rgba) +{ + vector unsigned int out = spu_convtu(rgba, 32); + out = spu_shuffle(out, out, VEC_LITERAL(vector unsigned char, + 8, 4, 0, 12, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0)); + return spu_extract(out, 0); +} + + +#endif /* SPU_COLORPACK_H */ diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 6f61a3d816d..c82ca51000e 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -29,11 +29,10 @@ * Triangle rendering within a tile. */ -#include - #include "pipe/p_compiler.h" #include "pipe/p_format.h" #include "pipe/p_util.h" +#include "spu_colorpack.h" #include "spu_main.h" #include "spu_texture.h" #include "spu_tile.h" @@ -253,21 +252,18 @@ eval_z(float x, float y) static INLINE void pack_colors(uint uicolors[4], const float4 fcolors[4]) { - /* XXX grab the code for _pack_rgba8() and use the shuffle - * command to do the swizzling seen here. - */ switch (spu.fb.color_format) { case PIPE_FORMAT_A8R8G8B8_UNORM: - uicolors[0] = _pack_rgba8(fcolors[0].f[3], fcolors[0].f[0], fcolors[0].f[1], fcolors[0].f[2]); - uicolors[1] = _pack_rgba8(fcolors[1].f[3], fcolors[1].f[0], fcolors[1].f[1], fcolors[1].f[2]); - uicolors[2] = _pack_rgba8(fcolors[2].f[3], fcolors[2].f[0], fcolors[2].f[1], fcolors[2].f[2]); - uicolors[3] = _pack_rgba8(fcolors[3].f[3], fcolors[0].f[0], fcolors[3].f[1], fcolors[3].f[2]); + uicolors[0] = spu_pack_A8R8G8B8(fcolors[0].v); + uicolors[1] = spu_pack_A8R8G8B8(fcolors[1].v); + uicolors[2] = spu_pack_A8R8G8B8(fcolors[2].v); + uicolors[3] = spu_pack_A8R8G8B8(fcolors[3].v); break; case PIPE_FORMAT_B8G8R8A8_UNORM: - uicolors[0] = _pack_rgba8(fcolors[0].f[2], fcolors[0].f[1], fcolors[0].f[0], fcolors[0].f[3]); - uicolors[1] = _pack_rgba8(fcolors[1].f[2], fcolors[1].f[1], fcolors[1].f[0], fcolors[1].f[3]); - uicolors[2] = _pack_rgba8(fcolors[2].f[2], fcolors[2].f[1], fcolors[2].f[0], fcolors[2].f[3]); - uicolors[3] = _pack_rgba8(fcolors[3].f[2], fcolors[3].f[1], fcolors[3].f[0], fcolors[3].f[3]); + uicolors[0] = spu_pack_B8G8R8A8(fcolors[0].v); + uicolors[1] = spu_pack_B8G8R8A8(fcolors[1].v); + uicolors[2] = spu_pack_B8G8R8A8(fcolors[2].v); + uicolors[3] = spu_pack_B8G8R8A8(fcolors[3].v); break; default: ASSERT(0); -- cgit v1.2.3 From ae620d5c36c2742cb4a7429a461e02bd6dbef201 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 16:25:42 -0700 Subject: Cell: use global color_shuffle to remove a switch stmnt --- src/mesa/pipe/cell/spu/Makefile | 2 + src/mesa/pipe/cell/spu/spu_colorpack.h | 9 ++++ src/mesa/pipe/cell/spu/spu_main.c | 12 ++++++ src/mesa/pipe/cell/spu/spu_main.h | 3 ++ src/mesa/pipe/cell/spu/spu_tri.c | 76 ++++++++++++++-------------------- 5 files changed, 56 insertions(+), 46 deletions(-) diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index 2d031bfbc61..91a631b6992 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -8,6 +8,8 @@ TOP = ../../../../.. include $(TOP)/configs/linux-cell +OPT_FLAGS=-g +OPT_FLAGS=-O3 PROG = g3d PROG_SPU = $(PROG)_spu diff --git a/src/mesa/pipe/cell/spu/spu_colorpack.h b/src/mesa/pipe/cell/spu/spu_colorpack.h index 56709bd9f38..9977a6ece0e 100644 --- a/src/mesa/pipe/cell/spu/spu_colorpack.h +++ b/src/mesa/pipe/cell/spu/spu_colorpack.h @@ -57,4 +57,13 @@ spu_pack_B8G8R8A8(vector float rgba) } +static INLINE unsigned int +spu_pack_color_shuffle(vector float rgba, vector unsigned char shuffle) +{ + vector unsigned int out = spu_convtu(rgba, 32); + out = spu_shuffle(out, out, shuffle); + return spu_extract(out, 0); +} + + #endif /* SPU_COLORPACK_H */ diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 8e3987f6ef5..ba4d180cc0c 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -31,6 +31,7 @@ #include #include +#include #include "spu_main.h" #include "spu_render.h" @@ -217,6 +218,17 @@ cmd_state_framebuffer(const struct cell_command_framebuffer *cmd) spu.fb.zsize = 2; else spu.fb.zsize = 0; + + if (spu.fb.color_format == PIPE_FORMAT_A8R8G8B8_UNORM) + spu.color_shuffle = VEC_LITERAL(vector unsigned char, + 12, 0, 4, 8, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0); + else if (spu.fb.color_format == PIPE_FORMAT_B8G8R8A8_UNORM) + spu.color_shuffle = VEC_LITERAL(vector unsigned char, + 8, 4, 0, 12, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0); + else + ASSERT(0); } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index cce5e708021..7a12715b0bc 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -107,6 +107,9 @@ struct spu_global ubyte ctile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; ubyte ztile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; + + /** for converting RGBA to PIPE_FORMAT_x colors */ + vector unsigned char color_shuffle; } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index c82ca51000e..165e41a7819 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -249,28 +249,6 @@ eval_z(float x, float y) } -static INLINE void -pack_colors(uint uicolors[4], const float4 fcolors[4]) -{ - switch (spu.fb.color_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - uicolors[0] = spu_pack_A8R8G8B8(fcolors[0].v); - uicolors[1] = spu_pack_A8R8G8B8(fcolors[1].v); - uicolors[2] = spu_pack_A8R8G8B8(fcolors[2].v); - uicolors[3] = spu_pack_A8R8G8B8(fcolors[3].v); - break; - case PIPE_FORMAT_B8G8R8A8_UNORM: - uicolors[0] = spu_pack_B8G8R8A8(fcolors[0].v); - uicolors[1] = spu_pack_B8G8R8A8(fcolors[1].v); - uicolors[2] = spu_pack_B8G8R8A8(fcolors[2].v); - uicolors[3] = spu_pack_B8G8R8A8(fcolors[3].v); - break; - default: - ASSERT(0); - } -} - - static INLINE mask_t do_depth_test(int x, int y, mask_t quadmask) { @@ -321,38 +299,44 @@ emit_quad( int x, int y, mask_t mask ) if (spu_extract(spu_orx(mask), 0)) { const int ix = x - setup.cliprect_minx; const int iy = y - setup.cliprect_miny; - uint colors[4]; /* indexed by QUAD_x */ + + if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { + /* now, _really_ clear the tile */ + clear_c_tile(&spu.ctile); + } + spu.cur_ctile_status = TILE_STATUS_DIRTY; if (spu.texture.start) { + /* texture mapping */ float4 texcoords[4]; - uint i; eval_coeff(2, (float) x, (float) y, texcoords); - for (i = 0; i < 4; i++) { - colors[i] = sample_texture(texcoords[i]); - } + + if (spu_extract(mask, 0)) + spu.ctile.ui[iy][ix] = sample_texture(texcoords[0]); + if (spu_extract(mask, 1)) + spu.ctile.ui[iy][ix+1] = sample_texture(texcoords[1]); + if (spu_extract(mask, 2)) + spu.ctile.ui[iy+1][ix] = sample_texture(texcoords[2]); + if (spu_extract(mask, 3)) + spu.ctile.ui[iy+1][ix+1] = sample_texture(texcoords[3]); } else { - float4 fcolors[4]; - eval_coeff(1, (float) x, (float) y, fcolors); - pack_colors(colors, fcolors); + /* simple shading */ + const vector unsigned char shuffle = spu.color_shuffle; + float4 colors[4]; + eval_coeff(1, (float) x, (float) y, colors); + + if (spu_extract(mask, 0)) + spu.ctile.ui[iy][ix] = spu_pack_color_shuffle(colors[0].v, shuffle); + if (spu_extract(mask, 1)) + spu.ctile.ui[iy][ix+1] = spu_pack_color_shuffle(colors[1].v, shuffle); + if (spu_extract(mask, 2)) + spu.ctile.ui[iy+1][ix] = spu_pack_color_shuffle(colors[2].v, shuffle); + if (spu_extract(mask, 3)) + spu.ctile.ui[iy+1][ix+1] = spu_pack_color_shuffle(colors[3].v, shuffle); } - if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { - /* now, _really_ clear the tile */ - clear_c_tile(&spu.ctile); - } - spu.cur_ctile_status = TILE_STATUS_DIRTY; - -#if 1 - if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = colors[QUAD_TOP_LEFT]; - if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = colors[QUAD_TOP_RIGHT]; - if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = colors[QUAD_BOTTOM_LEFT]; - if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = colors[QUAD_BOTTOM_RIGHT]; -#else +#if 0 /* SIMD_Z with swizzled color buffer (someday) */ vector unsigned int uicolors = *((vector unsigned int *) &colors); spu.ctile.ui4[iy/2][ix/2] = spu_sel(spu.ctile.ui4[iy/2][ix/2], uicolors, mask); -- cgit v1.2.3 From c90a2dcc294cba738fd65089e1b6ffe1c9b3853e Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 16:42:09 -0700 Subject: Cell: move some tile get/clear code Also, we weren't marking the ztile as dirty after ztesting, fixes gears glitches. --- src/mesa/pipe/cell/spu/spu_tri.c | 27 ++++++++++++++++----------- 1 file changed, 16 insertions(+), 11 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 165e41a7819..f0758c42e73 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -257,12 +257,6 @@ do_depth_test(int x, int y, mask_t quadmask) zvals.v = eval_z((float) x, (float) y); - if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { - /* now, _really_ clear the tile */ - clear_z_tile(&spu.ztile); - spu.cur_ztile_status = TILE_STATUS_DIRTY; - } - if (spu.fb.depth_format == PIPE_FORMAT_Z16_UNORM) { int ix = (x - setup.cliprect_minx) / 4; int iy = (y - setup.cliprect_miny) / 2; @@ -273,6 +267,10 @@ do_depth_test(int x, int y, mask_t quadmask) int iy = (y - setup.cliprect_miny) / 2; mask = spu_z32_test_less(zvals.v, &spu.ztile.ui4[iy][ix], quadmask); } + + if (spu_extract(spu_orx(mask), 0)) + spu.cur_ztile_status = TILE_STATUS_DIRTY; + return mask; } @@ -300,10 +298,6 @@ emit_quad( int x, int y, mask_t mask ) const int ix = x - setup.cliprect_minx; const int iy = y - setup.cliprect_miny; - if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { - /* now, _really_ clear the tile */ - clear_c_tile(&spu.ctile); - } spu.cur_ctile_status = TILE_STATUS_DIRTY; if (spu.texture.start) { @@ -408,6 +402,18 @@ static void flush_spans( void ) return; } + + /* _really_ clear tiles now if needed */ + if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { + clear_c_tile(&spu.ctile); + spu.cur_ctile_status = TILE_STATUS_DIRTY; + } + if (spu.depth_stencil.depth.enabled && + spu.cur_ztile_status == TILE_STATUS_CLEAR) { + clear_z_tile(&spu.ztile); + spu.cur_ztile_status = TILE_STATUS_DIRTY; + } + /* XXX this loop could be moved into the above switch cases and * calculate_mask() could be simplified a bit... */ @@ -831,7 +837,6 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) wait_on_mask(1 << TAG_READ_TILE_COLOR); spu.cur_ctile_status = TILE_STATUS_CLEAN; } - ASSERT(spu.cur_ctile_status != TILE_STATUS_DEFINED); if (spu.depth_stencil.depth.enabled) { -- cgit v1.2.3 From 6522a0531fbb6b6d607969ee6b2c2a85cce8ad2b Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 1 Feb 2008 16:54:46 -0700 Subject: Cell: comment about emit_quad() mask --- src/mesa/pipe/cell/spu/spu_tri.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index f0758c42e73..83bb247b22e 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -277,6 +277,9 @@ do_depth_test(int x, int y, mask_t quadmask) /** * Emit a quad (pass to next stage). No clipping is done. + * Note: about 1/5 to 1/7 of the time, mask is zero and this function + * should be skipped. But adding the test for that slows things down + * overall. */ static INLINE void emit_quad( int x, int y, mask_t mask ) -- cgit v1.2.3 From 8bbedc3f4b7b281a60286ba573077a6e3e659f63 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 2 Feb 2008 14:21:06 +1100 Subject: nouveau: implement a bo_set_status() --- src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 133 ++++++++++++--------- .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 3 + .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 4 +- 3 files changed, 80 insertions(+), 60 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c index 4c235845b7a..6887ffa6886 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c @@ -79,37 +79,6 @@ nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, return 0; } -static int -nouveau_bo_realloc_gpu(struct nouveau_bo_priv *nvbo, uint32_t flags, int size) -{ - int ret; - - if (nvbo->drm.size && nvbo->drm.size != size) { - nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); - } - - if (size && !nvbo->drm.size) { - if (flags) { - nvbo->drm.flags = 0; - if (flags & NOUVEAU_BO_VRAM) - nvbo->drm.flags |= NOUVEAU_MEM_FB; - if (flags & NOUVEAU_BO_GART) - nvbo->drm.flags |= (NOUVEAU_MEM_AGP | - NOUVEAU_MEM_PCI); - nvbo->drm.flags |= NOUVEAU_MEM_MAPPED; - } - - ret = nouveau_mem_alloc(nvbo->base.device, size, - nvbo->drm.alignment, nvbo->drm.flags, - &nvbo->drm, &nvbo->map); - if (ret) { - assert(0); - } - } - - return 0; -} - static void nouveau_bo_tmp_del(void *priv) { @@ -183,26 +152,17 @@ nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, if (!nvbo) return -ENOMEM; nvbo->base.device = dev; + nvbo->base.size = size; + nvbo->base.handle = bo_to_ptr(nvbo); nvbo->drm.alignment = align; + nvbo->refcount = 1; - if (flags & NOUVEAU_BO_PIN) { - ret = nouveau_bo_realloc_gpu(nvbo, flags, size); - if (ret) { - free(nvbo); - return ret; - } - } else { - nvbo->sysmem = malloc(size); - if (!nvbo->sysmem) { - free(nvbo); - return -ENOMEM; - } + ret = nouveau_bo_set_status(&nvbo->base, flags); + if (ret) { + free(nvbo); + return ret; } - nvbo->base.size = size; - nvbo->base.offset = nvbo->drm.offset; - nvbo->base.handle = bo_to_ptr(nvbo); - nvbo->refcount = 1; *bo = &nvbo->base; return 0; } @@ -261,8 +221,7 @@ nouveau_bo_del(struct nouveau_bo **bo) if (nvbo->fence) nouveau_fence_wait(&nvbo->fence); - - nouveau_bo_realloc_gpu(nvbo, 0, 0); + nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); if (nvbo->sysmem && !nvbo->user) free(nvbo->sysmem); free(nvbo); @@ -303,6 +262,66 @@ nouveau_bo_upload(struct nouveau_bo_priv *nvbo) return 0; } +int +nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct drm_nouveau_mem_alloc new; + void *new_map = NULL, *new_sysmem = NULL; + unsigned new_flags = 0, ret; + + assert(!bo->map); + + /* Check current memtype vs requested, if they match do nothing */ + if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) + return 0; + if ((nvbo->drm.flags & NOUVEAU_MEM_AGP) && (flags & NOUVEAU_BO_GART)) + return 0; + if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) + return 0; + + memset(&new, 0x00, sizeof(new)); + + /* Allocate new memory */ + if (flags & NOUVEAU_BO_VRAM) + new_flags |= NOUVEAU_MEM_FB; + else + if (flags & NOUVEAU_BO_GART) + new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); + + if (new_flags) { + ret = nouveau_mem_alloc(bo->device, bo->size, + nvbo->drm.alignment, new_flags, + &new, &new_map); + if (ret) + return ret; + } else { + new_sysmem = malloc(bo->size); + } + + /* Copy old -> new */ + /*XXX: use M2MF */ + if (nvbo->sysmem || nvbo->map) { + nouveau_bo_map(bo, NOUVEAU_BO_RD); + memcpy(new_map, bo->map, bo->size); + nouveau_bo_unmap(bo); + } + + /* Free old memory */ + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem) + free(nvbo->sysmem); + + nvbo->drm = new; + nvbo->map = new_map; + nvbo->sysmem = new_sysmem; + bo->flags = flags; + bo->offset = nvbo->drm.offset; + return 0; +} + static int nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, struct nouveau_fence *fence, uint32_t flags) @@ -335,18 +354,14 @@ nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, struct nouveau_fence *fence, uint32_t flags) { struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + int ret; - if (!nvbo->drm.size) { - nouveau_bo_realloc_gpu(nvbo, flags, nvbo->base.size); - nouveau_bo_upload(nvbo); - if (!nvbo->user) { - free(nvbo->sysmem); - nvbo->sysmem = NULL; - } - } else - if (nvbo->user) { + ret = nouveau_bo_set_status(bo, flags); + if (ret) + return ret; + + if (nvbo->user) nouveau_bo_upload(nvbo); - } nvbo->offset = nvbo->drm.offset; if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h index 7ea4c654654..67e19f1cfe8 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h @@ -274,6 +274,9 @@ nouveau_bo_user(struct nouveau_device *, void *ptr, int size, extern int nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); +extern int +nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags); + extern void nouveau_bo_del(struct nouveau_bo **); diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index f2087aaf9a2..7d7fefa8016 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -90,6 +90,7 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; struct nouveau_device *dev = nvpws->nv->nv_screen->device; struct nouveau_pipe_buffer *nvbuf; + uint32_t flags = 0; nvbuf = calloc(1, sizeof(*nvbuf)); if (!nvbuf) @@ -99,7 +100,8 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, nvbuf->base.usage = usage; nvbuf->base.size = size; - if (nouveau_bo_new(dev, NOUVEAU_BO_LOCAL, alignment, size, &nvbuf->bo)) { + flags = NOUVEAU_BO_LOCAL; + if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { free(nvbuf); return NULL; } -- cgit v1.2.3 From 705022f98c32c44b94411ea13dfe4cbc899f5a77 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 3 Feb 2008 12:08:31 +1100 Subject: nouveau: avoid relocations where possible. Potential relocations are emitted as NOPs where they're needed. In the event a buffer moves, the pushbuf code will emit the relevant state changes into the NOPs. Just a start, more work is needed to get this looking how I want it to. --- .../drivers/dri/nouveau_winsys/nouveau_local.h | 3 +- .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 61 +++++---- src/mesa/pipe/nouveau/nouveau_bo.h | 1 + src/mesa/pipe/nouveau/nouveau_push.h | 13 +- src/mesa/pipe/nv40/nv40_fragprog.c | 5 + src/mesa/pipe/nv40/nv40_fragtex.c | 8 +- src/mesa/pipe/nv40/nv40_state.c | 32 ++++- src/mesa/pipe/nv40/nv40_state_emit.c | 144 ++++++++++++--------- 8 files changed, 170 insertions(+), 97 deletions(-) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h index 7a539c81a94..59febca2929 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h @@ -61,9 +61,8 @@ } while(0) #define OUT_RELOC(buf,data,flags,vor,tor) do { \ - nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur, \ + nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur++, \ buf, (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ } while(0) /* Raw data + flags depending on FB/TT buffer */ diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c index a34a5c18662..7d5eddb92ff 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c @@ -96,6 +96,31 @@ nouveau_pushbuf_init(struct nouveau_channel *chan) return 0; } +static uint32_t +nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo, + struct nouveau_pushbuf_reloc *r) +{ + uint32_t push; + + if (r->flags & NOUVEAU_BO_LOW) { + push = bo->offset + r->data; + } else + if (r->flags & NOUVEAU_BO_HIGH) { + push = (bo->offset + r->data) >> 32; + } else { + push = r->data; + } + + if (r->flags & NOUVEAU_BO_OR) { + if (bo->flags & NOUVEAU_BO_VRAM) + push |= r->vor; + else + push |= r->tor; + } + + return push; +} + /* This would be our TTM "superioctl" */ int nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) @@ -133,34 +158,20 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) if (bo->offset == nouveau_bo(bo)->offset && bo->flags == nouveau_bo(bo)->flags) { - /*XXX: could avoid reloc in this case, except with the - * current design we'd confuse the GPU quite a bit - * if we did this. Will fix soon. - */ + while ((r = ptr_to_pbrel(pbbo->relocs))) { + pbbo->relocs = r->next; + free(r); + } + + nvpb->buffers = pbbo->next; + free(pbbo); + continue; } bo->offset = nouveau_bo(bo)->offset; bo->flags = nouveau_bo(bo)->flags; while ((r = ptr_to_pbrel(pbbo->relocs))) { - uint32_t push; - - if (r->flags & NOUVEAU_BO_LOW) { - push = bo->offset + r->data; - } else - if (r->flags & NOUVEAU_BO_HIGH) { - push = (bo->offset + r->data) >> 32; - } else { - push = r->data; - } - - if (r->flags & NOUVEAU_BO_OR) { - if (bo->flags & NOUVEAU_BO_VRAM) - push |= r->vor; - else - push |= r->tor; - } - - *r->ptr = push; + *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); pbbo->relocs = r->next; free(r); } @@ -241,6 +252,10 @@ nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, r->vor = vor; r->tor = tor; + if (flags & NOUVEAU_BO_DUMMY) + *(uint32_t *)ptr = 0; + else + *(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r); return 0; } diff --git a/src/mesa/pipe/nouveau/nouveau_bo.h b/src/mesa/pipe/nouveau/nouveau_bo.h index 2b57ee92633..18020e9c652 100644 --- a/src/mesa/pipe/nouveau/nouveau_bo.h +++ b/src/mesa/pipe/nouveau/nouveau_bo.h @@ -35,6 +35,7 @@ #define NOUVEAU_BO_HIGH (1 << 7) #define NOUVEAU_BO_OR (1 << 8) #define NOUVEAU_BO_LOCAL (1 << 9) +#define NOUVEAU_BO_DUMMY (1 << 31) struct nouveau_bo { struct nouveau_device *device; diff --git a/src/mesa/pipe/nouveau/nouveau_push.h b/src/mesa/pipe/nouveau/nouveau_push.h index 117e3535cf7..679472669b9 100644 --- a/src/mesa/pipe/nouveau/nouveau_push.h +++ b/src/mesa/pipe/nouveau/nouveau_push.h @@ -44,9 +44,8 @@ #define OUT_RELOC(bo,data,flags,vor,tor) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ pc->nvws->push_reloc(pc->nvws->channel, \ - pc->nvws->channel->pushbuf->cur, \ + pc->nvws->channel->pushbuf->cur++, \ (bo), (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ } while(0) /* Raw data + flags depending on FB/TT buffer */ @@ -71,4 +70,14 @@ OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ } while(0) +/* A reloc which'll recombine into a NV_DMA_METHOD packet header */ +#define OUT_RELOCm(bo, flags, obj, mthd, size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + OUT_RELOCd((bo), (pc->obj->subc << 13) | ((size) << 18) | (mthd), \ + (flags), 0, 0); \ + pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + #endif diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 14897f97982..667eb89cb24 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -815,6 +815,11 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->on_hw = TRUE; } + BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); + OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, + NV40TCL_FP_ADDRESS_DMA1); BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); OUT_RING (fp->fp_control); diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 48d6eb629f5..7c5ecd5c566 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -104,7 +104,13 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) nv40->tex[unit].buffer = nv40mt->buffer; nv40->tex[unit].format = txf; - BEGIN_RING(curie, NV40TCL_TEX_WRAP(unit), 6); + BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); + OUT_RELOCl(nv40->tex[unit].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv40->tex[unit].buffer, nv40->tex[unit].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, + NV40TCL_TEX_FORMAT_DMA1); OUT_RING (ps->wrap); OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | (0x00078000) /* mipmap related? */); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index c619948b55a..bb435b106b4 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -603,33 +603,51 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 1); - OUT_RING (rt[0]->pitch * rt[0]->cpp); nv40->rt[0] = rt[0]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); + OUT_RING (rt[0]->pitch * rt[0]->cpp); + OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(curie, NV40TCL_COLOR1_PITCH, 2); - OUT_RING (rt[1]->pitch * rt[1]->cpp); nv40->rt[1] = rt[1]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); + OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { + nv40->rt[2] = rt[2]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); + OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); + OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); OUT_RING (rt[2]->pitch * rt[2]->cpp); - nv40->rt[2] = rt[2]->buffer; } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { + nv40->rt[3] = rt[3]->buffer; + BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); + OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); + OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); OUT_RING (rt[3]->pitch * rt[3]->cpp); - nv40->rt[3] = rt[3]->buffer; } if (zeta_format) { + nv40->zeta = zeta->buffer; + BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); + OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); OUT_RING (zeta->pitch * zeta->cpp); - nv40->zeta = zeta->buffer; } nv40->rt_enable = rt_enable; diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index c9a7a2e364b..66b98d5fab3 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -1,94 +1,114 @@ #include "nv40_context.h" #include "nv40_state.h" -void -nv40_emit_hw_state(struct nv40_context *nv40) +/* Emit relocs for every referenced buffer. + * + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. These relocs appear in the push buffer as + * NOPs, and will only be turned into state changes if a buffer + * actually moves. + */ +static void +nv40_state_emit_dummy_relocs(struct nv40_context *nv40) { - int i; - - if (nv40->dirty & NV40_NEW_FRAGPROG) { - nv40_fragprog_bind(nv40, nv40->fragprog.current); - /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ - } - - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { - nv40_fragtex_bind(nv40); - - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (1); - nv40->dirty &= ~NV40_NEW_FRAGPROG; - } - - if (nv40->dirty & NV40_NEW_VERTPROG) { - nv40_vertprog_bind(nv40, nv40->vertprog.current); - nv40->dirty &= ~NV40_NEW_VERTPROG; - } - - nv40->dirty_samplers = 0; - - /* Emit relocs for every referenced buffer. - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. This isn't very efficient, but we don't - * seem to take a significant performance hit. Will be improved - * at some point. Vertex arrays are emitted by nv40_vbo.c - */ + unsigned rt_flags, tx_flags, fp_flags; + int i; + + rt_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR | NOUVEAU_BO_DUMMY; + tx_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_DUMMY; + fp_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_DUMMY; /* Render targets */ if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR0_OFFSET, 1); - OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[0], rt_flags, + curie, NV40TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv40->rt[0], rt_flags); + OUT_RELOCm(nv40->rt[0], rt_flags, + curie, NV40TCL_COLOR0_OFFSET, 1); + OUT_RELOCl(nv40->rt[0], 0, rt_flags); } if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 1); - OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[1], rt_flags, + curie, NV40TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv40->rt[1], rt_flags); + OUT_RELOCm(nv40->rt[1], rt_flags, + curie, NV40TCL_COLOR1_OFFSET, 1); + OUT_RELOCl(nv40->rt[1], 0, rt_flags); } if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[2], rt_flags, + curie, NV40TCL_DMA_COLOR2, 1); + OUT_RELOCo(nv40->rt[2], rt_flags); + OUT_RELOCm(nv40->rt[2], rt_flags, + curie, NV40TCL_COLOR2_OFFSET, 1); + OUT_RELOCl(nv40->rt[2], 0, rt_flags); } if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->rt[3], rt_flags, + curie, NV40TCL_DMA_COLOR3, 1); + OUT_RELOCo(nv40->rt[3], rt_flags); + OUT_RELOCm(nv40->rt[3], rt_flags, + curie, NV40TCL_COLOR3_OFFSET, 1); + OUT_RELOCl(nv40->rt[3], 0, rt_flags); } if (nv40->zeta) { - BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_DMA_ZETA, 1); + OUT_RELOCo(nv40->zeta, rt_flags); + OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv40->zeta, 0, rt_flags); } /* Texture images */ for (i = 0; i < 16; i++) { if (!(nv40->fp_samplers & (1 << i))) continue; - BEGIN_RING(curie, NV40TCL_TEX_OFFSET(i), 2); - OUT_RELOCl(nv40->tex[i].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCm(nv40->tex[i].buffer, tx_flags, + curie, NV40TCL_TEX_OFFSET(i), 2); + OUT_RELOCl(nv40->tex[i].buffer, 0, tx_flags); OUT_RELOCd(nv40->tex[i].buffer, nv40->tex[i].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, + tx_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); } /* Fragment program */ - BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (nv40->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, - NV40TCL_FP_ADDRESS_DMA1); + OUT_RELOCm(nv40->fragprog.active->buffer, fp_flags, + curie, NV40TCL_FP_ADDRESS, 1); + OUT_RELOC (nv40->fragprog.active->buffer, 0, + fp_flags | NOUVEAU_BO_OR | NOUVEAU_BO_LOW, + NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); +} + +void +nv40_emit_hw_state(struct nv40_context *nv40) +{ + if (nv40->dirty & NV40_NEW_FRAGPROG) { + nv40_fragprog_bind(nv40, nv40->fragprog.current); + /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ + } + + if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { + nv40_fragtex_bind(nv40); + + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (1); + nv40->dirty &= ~NV40_NEW_FRAGPROG; + } + + if (nv40->dirty & NV40_NEW_VERTPROG) { + nv40_vertprog_bind(nv40, nv40->vertprog.current); + nv40->dirty &= ~NV40_NEW_VERTPROG; + } + + nv40->dirty_samplers = 0; + + nv40_state_emit_dummy_relocs(nv40); } -- cgit v1.2.3 From b111d266a0edf334e8c315129482005a224cb899 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 4 Feb 2008 01:37:07 +1100 Subject: nv40: the 0x4497 version of curie doesn't support index buffers. --- src/mesa/pipe/nv40/nv40_vbo.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 3fa8ddcb81b..0bb54c7610e 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -391,12 +391,17 @@ nv40_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { - if (indexSize != 1) { - nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, - mode, start, count); - } else { + struct nv40_context *nv40 = nv40_context(pipe); + + /* 0x4497 doesn't support real index buffers, and there doesn't appear + * to be support on any chipset for 8-bit indices. + */ + if (nv40->curie->grclass == NV44TCL || indexSize == 1) { nv40_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); + } else { + nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, + mode, start, count); } pipe->flush(pipe, 0); -- cgit v1.2.3 From 4605b7df0a98025999169254f08e532027a8a46a Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 08:45:33 -0700 Subject: Cell: insert some draw_flush() calls --- src/mesa/pipe/cell/ppu/cell_state_blend.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/src/mesa/pipe/cell/ppu/cell_state_blend.c b/src/mesa/pipe/cell/ppu/cell_state_blend.c index 34ae0128ea7..2c19aa39715 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_blend.c +++ b/src/mesa/pipe/cell/ppu/cell_state_blend.c @@ -29,6 +29,7 @@ */ #include "pipe/p_util.h" +#include "pipe/draw/draw_context.h" #include "cell_context.h" #include "cell_state.h" @@ -49,6 +50,8 @@ cell_bind_blend_state(struct pipe_context *pipe, void *blend) { struct cell_context *cell = cell_context(pipe); + draw_flush(cell->draw); + cell->blend = (const struct pipe_blend_state *)blend; cell->dirty |= CELL_NEW_BLEND; @@ -68,6 +71,8 @@ cell_set_blend_color(struct pipe_context *pipe, { struct cell_context *cell = cell_context(pipe); + draw_flush(cell->draw); + cell->blend_color = *blend_color; cell->dirty |= CELL_NEW_BLEND; @@ -93,6 +98,8 @@ cell_bind_depth_stencil_alpha_state(struct pipe_context *pipe, { struct cell_context *cell = cell_context(pipe); + draw_flush(cell->draw); + cell->depth_stencil = (const struct pipe_depth_stencil_alpha_state *) depth_stencil; -- cgit v1.2.3 From a0c35df4a0d1619b2d8593d35456ed50be3b03c1 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 08:46:44 -0700 Subject: Cell: clamp txmax, tymax in tile_bounding_box() Also, added some debug printfs --- src/mesa/pipe/cell/spu/spu_render.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_render.c b/src/mesa/pipe/cell/spu/spu_render.c index ab711d67feb..e8705eeebad 100644 --- a/src/mesa/pipe/cell/spu/spu_render.c +++ b/src/mesa/pipe/cell/spu/spu_render.c @@ -65,6 +65,10 @@ tile_bounding_box(const struct cell_command_render *render, *tymin = (uint) render->ymin / TILE_SIZE; txmax = (uint) render->xmax / TILE_SIZE; tymax = (uint) render->ymax / TILE_SIZE; + if (txmax >= spu.fb.width_tiles) + txmax = spu.fb.width_tiles-1; + if (tymax >= spu.fb.height_tiles) + tymax = spu.fb.height_tiles-1; *box_width_tiles = txmax - *txmin + 1; box_height_tiles = tymax - *tymin + 1; *box_num_tiles = *box_width_tiles * box_height_tiles; @@ -96,12 +100,14 @@ get_cz_tiles(uint tx, uint ty) { if (spu.depth_stencil.depth.enabled) { if (spu.cur_ztile_status != TILE_STATUS_CLEAR) { + //printf("SPU %u: getting Z tile %u, %u\n", spu.init.id, tx, ty); get_tile(tx, ty, &spu.ztile, TAG_READ_TILE_Z, 1); spu.cur_ztile_status = TILE_STATUS_GETTING; } } if (spu.cur_ctile_status != TILE_STATUS_CLEAR) { + //printf("SPU %u: getting C tile %u, %u\n", spu.init.id, tx, ty); get_tile(tx, ty, &spu.ctile, TAG_READ_TILE_COLOR, 0); spu.cur_ctile_status = TILE_STATUS_GETTING; } @@ -116,22 +122,26 @@ put_cz_tiles(uint tx, uint ty) { if (spu.cur_ztile_status == TILE_STATUS_DIRTY) { /* tile was modified and needs to be written back */ + //printf("SPU %u: put dirty Z tile %u, %u\n", spu.init.id, tx, ty); put_tile(tx, ty, &spu.ztile, TAG_WRITE_TILE_Z, 1); spu.cur_ztile_status = TILE_STATUS_DEFINED; } else if (spu.cur_ztile_status == TILE_STATUS_GETTING) { /* tile was never used */ spu.cur_ztile_status = TILE_STATUS_DEFINED; + //printf("SPU %u: put getting Z tile %u, %u\n", spu.init.id, tx, ty); } if (spu.cur_ctile_status == TILE_STATUS_DIRTY) { /* tile was modified and needs to be written back */ + //printf("SPU %u: put dirty C tile %u, %u\n", spu.init.id, tx, ty); put_tile(tx, ty, &spu.ctile, TAG_WRITE_TILE_COLOR, 0); spu.cur_ctile_status = TILE_STATUS_DEFINED; } else if (spu.cur_ctile_status == TILE_STATUS_GETTING) { /* tile was never used */ spu.cur_ctile_status = TILE_STATUS_DEFINED; + //printf("SPU %u: put getting C tile %u, %u\n", spu.init.id, tx, ty); } } -- cgit v1.2.3 From d83dedc937641de247bebbefad649719f619cdeb Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 08:53:18 -0700 Subject: Cell: move tile clear code to flush_spans() --- src/mesa/pipe/cell/spu/spu_tri.c | 51 ++++++++++++++++++++++------------------ 1 file changed, 28 insertions(+), 23 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 83bb247b22e..3f46e75d7ce 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -406,22 +406,44 @@ static void flush_spans( void ) } - /* _really_ clear tiles now if needed */ - if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { + /* OK, we're very likely to need the tile data now. + * clear or finish waiting if needed. + */ + if (spu.cur_ctile_status == TILE_STATUS_GETTING) { + /* wait for mfc_get() to complete */ + //printf("SPU: %u: waiting for ctile\n", spu.init.id); + wait_on_mask(1 << TAG_READ_TILE_COLOR); + spu.cur_ctile_status = TILE_STATUS_CLEAN; + } + else if (spu.cur_ctile_status == TILE_STATUS_CLEAR) { + //printf("SPU %u: clearing C tile %u, %u\n", spu.init.id, setup.tx, setup.ty); clear_c_tile(&spu.ctile); spu.cur_ctile_status = TILE_STATUS_DIRTY; } - if (spu.depth_stencil.depth.enabled && - spu.cur_ztile_status == TILE_STATUS_CLEAR) { - clear_z_tile(&spu.ztile); - spu.cur_ztile_status = TILE_STATUS_DIRTY; + ASSERT(spu.cur_ctile_status != TILE_STATUS_DEFINED); + + if (spu.depth_stencil.depth.enabled) { + if (spu.cur_ztile_status == TILE_STATUS_GETTING) { + /* wait for mfc_get() to complete */ + //printf("SPU: %u: waiting for ztile\n", spu.init.id); + wait_on_mask(1 << TAG_READ_TILE_Z); + spu.cur_ztile_status = TILE_STATUS_CLEAN; + } + else if (spu.cur_ztile_status == TILE_STATUS_CLEAR) { + //printf("SPU %u: clearing Z tile %u, %u\n", spu.init.id, setup.tx, setup.ty); + clear_z_tile(&spu.ztile); + spu.cur_ztile_status = TILE_STATUS_DIRTY; + } + ASSERT(spu.cur_ztile_status != TILE_STATUS_DEFINED); } /* XXX this loop could be moved into the above switch cases and * calculate_mask() could be simplified a bit... */ for (x = block(minleft); x <= block(maxright); x += 2) { +#if 1 emit_quad( x, setup.span.y, calculate_mask( x ) ); +#endif } setup.span.y = 0; @@ -835,23 +857,6 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty) /* init_constant_attribs( setup ); */ - if (spu.cur_ctile_status == TILE_STATUS_GETTING) { - /* wait for mfc_get() to complete */ - wait_on_mask(1 << TAG_READ_TILE_COLOR); - spu.cur_ctile_status = TILE_STATUS_CLEAN; - } - ASSERT(spu.cur_ctile_status != TILE_STATUS_DEFINED); - - if (spu.depth_stencil.depth.enabled) { - if (spu.cur_ztile_status == TILE_STATUS_GETTING) { - /* wait for mfc_get() to complete */ - wait_on_mask(1 << TAG_READ_TILE_Z); - spu.cur_ztile_status = TILE_STATUS_CLEAN; - } - ASSERT(spu.cur_ztile_status != TILE_STATUS_DEFINED); - } - - if (setup.oneoverarea < 0.0) { /* emaj on left: */ -- cgit v1.2.3 From f94e0396ed023c21087d1ea1a849b1f8124f9ffb Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 09:54:21 -0700 Subject: Cell: checkpoint: start to SIMD-ize texture sampling --- src/mesa/pipe/cell/spu/spu_main.c | 10 ++++++++++ src/mesa/pipe/cell/spu/spu_main.h | 4 ++++ src/mesa/pipe/cell/spu/spu_texture.c | 17 ++++++++++++++--- src/mesa/pipe/cell/spu/spu_texture.h | 2 +- src/mesa/pipe/cell/spu/spu_tri.c | 8 ++++---- 5 files changed, 33 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index ba4d180cc0c..412661061ab 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -263,6 +263,16 @@ cmd_state_texture(const struct cell_command_texture *texture) spu.init.id, texture->start, texture->width, texture->height); memcpy(&spu.texture, texture, sizeof(*texture)); + spu.tex_size = VEC_LITERAL(vector float, + spu.texture.width, + spu.texture.height, + 0.0, + 0.0); + spu.tex_size_mask = VEC_LITERAL(vector unsigned int, + spu.texture.width - 1, + spu.texture.height - 1, + 0, + 0); } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 7a12715b0bc..02b62ee5cd5 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -110,6 +110,10 @@ struct spu_global /** for converting RGBA to PIPE_FORMAT_x colors */ vector unsigned char color_shuffle; + + vector float tex_size; + vector unsigned int tex_size_mask; /**< == int(size - 1) */ + } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index c1dc6bfe902..1cf958806fc 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -128,12 +128,23 @@ get_tex_tile(uint i, uint j) * XXX this is extremely primitive for now. */ uint -sample_texture(float4 texcoord) +sample_texture(vector float texcoord) { +#if 0 /* wrap/repeat */ - uint i = (uint) (texcoord.f[0] * spu.texture.width) % spu.texture.width; - uint j = (uint) (texcoord.f[1] * spu.texture.height) % spu.texture.height; + uint i = (uint) (spu_extract(texcoord, 0) * spu.texture.width) % spu.texture.width; + uint j = (uint) (spu_extract(texcoord, 1) * spu.texture.height) % spu.texture.height; uint pos = get_tex_tile(i, j); uint texel = tex_tiles[pos].ui[j % TILE_SIZE][i % TILE_SIZE]; return texel; +#else + vector float tc = spu_mul(texcoord, spu.tex_size); + vector unsigned int itc = spu_convtu(tc, 0); + itc = spu_and(itc, spu.tex_size_mask); + uint i = spu_extract(itc, 0); + uint j = spu_extract(itc, 1); + uint pos = get_tex_tile(i, j); + uint texel = tex_tiles[pos].ui[j % TILE_SIZE][i % TILE_SIZE]; + return texel; +#endif } diff --git a/src/mesa/pipe/cell/spu/spu_texture.h b/src/mesa/pipe/cell/spu/spu_texture.h index 938a42b5495..5bc8e71879b 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.h +++ b/src/mesa/pipe/cell/spu/spu_texture.h @@ -37,7 +37,7 @@ invalidate_tex_cache(void); extern uint -sample_texture(float4 texcoord); +sample_texture(vector float texcoord); #endif /* SPU_TEXTURE_H */ diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 3f46e75d7ce..c148c75dd61 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -309,13 +309,13 @@ emit_quad( int x, int y, mask_t mask ) eval_coeff(2, (float) x, (float) y, texcoords); if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = sample_texture(texcoords[0]); + spu.ctile.ui[iy][ix] = sample_texture(texcoords[0].v); if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = sample_texture(texcoords[1]); + spu.ctile.ui[iy][ix+1] = sample_texture(texcoords[1].v); if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = sample_texture(texcoords[2]); + spu.ctile.ui[iy+1][ix] = sample_texture(texcoords[2].v); if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = sample_texture(texcoords[3]); + spu.ctile.ui[iy+1][ix+1] = sample_texture(texcoords[3].v); } else { /* simple shading */ -- cgit v1.2.3 From 09edd2e29e023b326ba3f6fff671dd1db3ab1eea Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 11:02:47 -0700 Subject: Cell: SIMD-ize more of texture sampling --- src/mesa/pipe/cell/spu/spu_texture.c | 66 ++++++++++++++++-------------------- 1 file changed, 29 insertions(+), 37 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 1cf958806fc..b52df970d07 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -26,6 +26,8 @@ **************************************************************************/ +#include + #include "pipe/p_compiler.h" #include "spu_main.h" #include "spu_texture.h" @@ -41,7 +43,7 @@ static tile_t tex_tiles[CACHE_SIZE] ALIGN16_ATTRIB; -static int tex_tile_x[CACHE_SIZE], tex_tile_y[CACHE_SIZE]; +static vector unsigned int tex_tile_xy[CACHE_SIZE]; @@ -53,20 +55,19 @@ invalidate_tex_cache(void) { /* XXX memset? */ uint i; - for (i = 0; i < CACHE_SIZE; i++) - tex_tile_x[i] = tex_tile_y[i] = -1; + for (i = 0; i < CACHE_SIZE; i++) { + tex_tile_xy[i] = VEC_LITERAL(vector unsigned int, ~0U, ~0U, ~0U, ~0U); + } } /** - * Return the cache pos/index which corresponds to texel (i,j) + * Return the cache pos/index which corresponds to tile (tx,ty) */ static INLINE uint -cache_pos(uint i, uint j) +cache_pos(vector unsigned int txty) { - uint tx = i / TILE_SIZE; - uint ty = j / TILE_SIZE; - uint pos = (tx + ty * 4) % CACHE_SIZE; + uint pos = (spu_extract(txty,0) + spu_extract(txty,1) * 4) % CACHE_SIZE; return pos; } @@ -76,26 +77,28 @@ cache_pos(uint i, uint j) * in the cache. */ static uint -get_tex_tile(uint i, uint j) +get_tex_tile(vector unsigned int ij) { - const int tx = i / TILE_SIZE; - const int ty = j / TILE_SIZE; - const uint pos = cache_pos(i, j); + /* tile address: tx,ty */ + const vector unsigned int txty = spu_rlmask(ij, -5); /* divide by 32 */ + const uint pos = cache_pos(txty); + + if ((spu_extract(tex_tile_xy[pos], 0) != spu_extract(txty, 0)) || + (spu_extract(tex_tile_xy[pos], 1) != spu_extract(txty, 1))) { - if (tex_tile_x[pos] != tx || tex_tile_y[pos] != ty) { /* texture cache miss, fetch tile from main memory */ const uint tiles_per_row = spu.texture.width / TILE_SIZE; const uint bytes_per_tile = sizeof(tile_t); const void *src = (const ubyte *) spu.texture.start - + (ty * tiles_per_row + tx) * bytes_per_tile; + + (spu_extract(txty,1) * tiles_per_row + spu_extract(txty,0)) * bytes_per_tile; printf("SPU %u: tex cache miss at %d, %d pos=%u old=%d,%d\n", - spu.init.id, tx, ty, pos, - tex_tile_x[pos], tex_tile_y[pos]); -#if 0 - printf("SPU %u: get tex tile from %p to %p\n", - spu.init.id, src, tex_tiles[pos].t32); -#endif + spu.init.id, + spu_extract(txty,0), + spu_extract(txty,1), + pos, + spu_extract(tex_tile_xy[pos],0), + spu_extract(tex_tile_xy[pos],1)); ASSERT_ALIGN16(tex_tiles[pos].ui); ASSERT_ALIGN16(src); @@ -109,8 +112,7 @@ get_tex_tile(uint i, uint j) wait_on_mask(1 << TAG_TEXTURE_TILE); - tex_tile_x[pos] = tx; - tex_tile_y[pos] = ty; + tex_tile_xy[pos] = txty; } else { #if 0 @@ -130,21 +132,11 @@ get_tex_tile(uint i, uint j) uint sample_texture(vector float texcoord) { -#if 0 - /* wrap/repeat */ - uint i = (uint) (spu_extract(texcoord, 0) * spu.texture.width) % spu.texture.width; - uint j = (uint) (spu_extract(texcoord, 1) * spu.texture.height) % spu.texture.height; - uint pos = get_tex_tile(i, j); - uint texel = tex_tiles[pos].ui[j % TILE_SIZE][i % TILE_SIZE]; - return texel; -#else vector float tc = spu_mul(texcoord, spu.tex_size); - vector unsigned int itc = spu_convtu(tc, 0); - itc = spu_and(itc, spu.tex_size_mask); - uint i = spu_extract(itc, 0); - uint j = spu_extract(itc, 1); - uint pos = get_tex_tile(i, j); - uint texel = tex_tiles[pos].ui[j % TILE_SIZE][i % TILE_SIZE]; + vector unsigned int itc = spu_convtu(tc, 0); /* convert to int */ + itc = spu_and(itc, spu.tex_size_mask); /* mask (GL_REPEAT) */ + vector unsigned int ij = spu_and(itc, TILE_SIZE-1); /* intra tile addr */ + uint pos = get_tex_tile(itc); + uint texel = tex_tiles[pos].ui[spu_extract(ij, 1)][spu_extract(ij, 0)]; return texel; -#endif } -- cgit v1.2.3 From 382651a4fafbea0c24e993933cb08a7ba19abcb7 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 12:50:16 -0700 Subject: Cell: added spu_unpack_color(), spu_pack_R8G8B8A8() --- src/mesa/pipe/cell/spu/spu_colorpack.h | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_colorpack.h b/src/mesa/pipe/cell/spu/spu_colorpack.h index 9977a6ece0e..0c93c065627 100644 --- a/src/mesa/pipe/cell/spu/spu_colorpack.h +++ b/src/mesa/pipe/cell/spu/spu_colorpack.h @@ -35,6 +35,17 @@ #include +static INLINE unsigned int +spu_pack_R8G8B8A8(vector float rgba) +{ + vector unsigned int out = spu_convtu(rgba, 32); + out = spu_shuffle(out, out, VEC_LITERAL(vector unsigned char, + 0, 4, 8, 12, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0)); + return spu_extract(out, 0); +} + + static INLINE unsigned int spu_pack_A8R8G8B8(vector float rgba) { @@ -66,4 +77,18 @@ spu_pack_color_shuffle(vector float rgba, vector unsigned char shuffle) } +static INLINE vector float +spu_unpack_color(uint color) +{ + vector unsigned int color_u4 = spu_splats(color); + color_u4 = spu_shuffle(color_u4, color_u4, + VEC_LITERAL(vector unsigned char, + 0, 0, 0, 0, + 5, 5, 5, 5, + 10, 10, 10, 10, + 15, 15, 15, 15)); + return spu_convtf(color_u4, 32); +} + + #endif /* SPU_COLORPACK_H */ -- cgit v1.2.3 From e3ff185eae2f49f4dac92f7e89558ed175251c25 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 12:50:42 -0700 Subject: Cell: implement basic bilinear texture sampler --- src/mesa/pipe/cell/spu/spu_texture.c | 67 ++++++++++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_texture.h | 4 +++ 2 files changed, 71 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index b52df970d07..26a5eefc482 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -32,6 +32,7 @@ #include "spu_main.h" #include "spu_texture.h" #include "spu_tile.h" +#include "spu_colorpack.h" /** @@ -140,3 +141,69 @@ sample_texture(vector float texcoord) uint texel = tex_tiles[pos].ui[spu_extract(ij, 1)][spu_extract(ij, 0)]; return texel; } + + +uint +sample_texture_bilinear(vector float texcoord) +{ + static const vector unsigned int offset10 = {1, 0, 0, 0}; + static const vector unsigned int offset01 = {0, 1, 0, 0}; + + vector float tc = spu_mul(texcoord, spu.tex_size); + /* itcST */ + vector unsigned int itc00 = spu_convtu(tc, 0); /* convert to int */ + vector unsigned int itc01 = spu_add(itc00, offset01); + vector unsigned int itc10 = spu_add(itc00, offset10); + vector unsigned int itc11 = spu_add(itc10, offset01); + + itc00 = spu_and(itc00, spu.tex_size_mask); /* mask (GL_REPEAT) */ + itc01 = spu_and(itc01, spu.tex_size_mask); /* mask (GL_REPEAT) */ + itc10 = spu_and(itc10, spu.tex_size_mask); /* mask (GL_REPEAT) */ + itc11 = spu_and(itc11, spu.tex_size_mask); /* mask (GL_REPEAT) */ + + /* intra tile addr */ + vector unsigned int ij00 = spu_and(itc00, TILE_SIZE-1); + vector unsigned int ij01 = spu_and(itc01, TILE_SIZE-1); + vector unsigned int ij10 = spu_and(itc10, TILE_SIZE-1); + vector unsigned int ij11 = spu_and(itc11, TILE_SIZE-1); + + uint pos00 = get_tex_tile(itc00); + uint pos01 = get_tex_tile(itc01); + uint pos10 = get_tex_tile(itc10); + uint pos11 = get_tex_tile(itc11); + + vector float texel00 = spu_unpack_color(tex_tiles[pos00].ui[spu_extract(ij00, 1)][spu_extract(ij00, 0)]); + vector float texel01 = spu_unpack_color(tex_tiles[pos01].ui[spu_extract(ij01, 1)][spu_extract(ij01, 0)]); + vector float texel10 = spu_unpack_color(tex_tiles[pos10].ui[spu_extract(ij10, 1)][spu_extract(ij10, 0)]); + vector float texel11 = spu_unpack_color(tex_tiles[pos11].ui[spu_extract(ij11, 1)][spu_extract(ij11, 0)]); + + /* Compute weighting factors in [0,1] + * Multiply texcoord by 1024, AND with 1023, convert back to float. + */ + vector float tc1024 = spu_mul(tc, spu_splats(1024.0f)); + vector signed int itc1024 = spu_convts(tc1024, 0); + itc1024 = spu_and(itc1024, spu_splats((1 << 10) - 1)); + vector float weight = spu_convtf(itc1024, 10); + + /* smeared frac and 1-frac */ + vector float sfrac = spu_splats(spu_extract(weight, 0)); + vector float tfrac = spu_splats(spu_extract(weight, 1)); + vector float sfrac1 = spu_sub(spu_splats(1.0f), sfrac); + vector float tfrac1 = spu_sub(spu_splats(1.0f), tfrac); + + /* multiply the samples (colors) by the S/T weights */ + texel00 = spu_mul(spu_mul(texel00, sfrac1), tfrac1); + texel10 = spu_mul(spu_mul(texel10, sfrac ), tfrac1); + texel01 = spu_mul(spu_mul(texel01, sfrac1), tfrac ); + texel11 = spu_mul(spu_mul(texel11, sfrac ), tfrac ); + + /* compute sum of weighted samples */ + vector float texel_sum = spu_add(texel00, texel01); + texel_sum = spu_add(texel_sum, texel10); + texel_sum = spu_add(texel_sum, texel11); + + /* convert to uint color */ + uint texel = spu_pack_R8G8B8A8(texel_sum); + + return texel; +} diff --git a/src/mesa/pipe/cell/spu/spu_texture.h b/src/mesa/pipe/cell/spu/spu_texture.h index 5bc8e71879b..25cbe9b3c64 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.h +++ b/src/mesa/pipe/cell/spu/spu_texture.h @@ -40,4 +40,8 @@ extern uint sample_texture(vector float texcoord); +extern uint +sample_texture_bilinear(vector float texcoord); + + #endif /* SPU_TEXTURE_H */ -- cgit v1.2.3 From 1bd182889b3dbb2f4c75d18184e7c76a5bfca248 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 13:16:10 -0700 Subject: Cell: improved bilinear filtering avoid calling get_tex_tile() if all texels are in same tile --- src/mesa/pipe/cell/spu/spu_texture.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 26a5eefc482..6e243f7fa3a 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -150,16 +150,17 @@ sample_texture_bilinear(vector float texcoord) static const vector unsigned int offset01 = {0, 1, 0, 0}; vector float tc = spu_mul(texcoord, spu.tex_size); - /* itcST */ + /* integer texcoords S,T: */ vector unsigned int itc00 = spu_convtu(tc, 0); /* convert to int */ vector unsigned int itc01 = spu_add(itc00, offset01); vector unsigned int itc10 = spu_add(itc00, offset10); vector unsigned int itc11 = spu_add(itc10, offset01); - itc00 = spu_and(itc00, spu.tex_size_mask); /* mask (GL_REPEAT) */ - itc01 = spu_and(itc01, spu.tex_size_mask); /* mask (GL_REPEAT) */ - itc10 = spu_and(itc10, spu.tex_size_mask); /* mask (GL_REPEAT) */ - itc11 = spu_and(itc11, spu.tex_size_mask); /* mask (GL_REPEAT) */ + /* mask (GL_REPEAT) */ + itc00 = spu_and(itc00, spu.tex_size_mask); + itc01 = spu_and(itc01, spu.tex_size_mask); + itc10 = spu_and(itc10, spu.tex_size_mask); + itc11 = spu_and(itc11, spu.tex_size_mask); /* intra tile addr */ vector unsigned int ij00 = spu_and(itc00, TILE_SIZE-1); @@ -167,11 +168,21 @@ sample_texture_bilinear(vector float texcoord) vector unsigned int ij10 = spu_and(itc10, TILE_SIZE-1); vector unsigned int ij11 = spu_and(itc11, TILE_SIZE-1); + /* get tile cache positions */ uint pos00 = get_tex_tile(itc00); - uint pos01 = get_tex_tile(itc01); - uint pos10 = get_tex_tile(itc10); - uint pos11 = get_tex_tile(itc11); + uint pos01, pos10, pos11; + if ((spu_extract(ij00, 0) < TILE_SIZE-1) && + (spu_extract(ij00, 1) < TILE_SIZE-1)) { + /* all texels are in the same tile */ + pos01 = pos10 = pos11 = pos00; + } + else { + pos01 = get_tex_tile(itc01); + pos10 = get_tex_tile(itc10); + pos11 = get_tex_tile(itc11); + } + /* get texels from tiles and convert to float[4] */ vector float texel00 = spu_unpack_color(tex_tiles[pos00].ui[spu_extract(ij00, 1)][spu_extract(ij00, 0)]); vector float texel01 = spu_unpack_color(tex_tiles[pos01].ui[spu_extract(ij01, 1)][spu_extract(ij01, 0)]); vector float texel10 = spu_unpack_color(tex_tiles[pos10].ui[spu_extract(ij10, 1)][spu_extract(ij10, 0)]); -- cgit v1.2.3 From 4080fef4732078e8861eb0d26d1f6e43aa27dd9e Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 13:23:07 -0700 Subject: Cell: choose bilinear vs. nearest filtering according to sampler state --- src/mesa/pipe/cell/spu/spu_main.c | 4 ++++ src/mesa/pipe/cell/spu/spu_main.h | 2 ++ src/mesa/pipe/cell/spu/spu_texture.c | 2 +- src/mesa/pipe/cell/spu/spu_texture.h | 2 +- src/mesa/pipe/cell/spu/spu_tri.c | 8 ++++---- 5 files changed, 12 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 412661061ab..48e016fc8b4 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -252,6 +252,10 @@ cmd_state_sampler(const struct pipe_sampler_state *state) spu.init.id); memcpy(&spu.sampler[0], state, sizeof(*state)); + if (spu.sampler[0].min_img_filter == PIPE_TEX_FILTER_LINEAR) + spu.sample_texture = sample_texture_bilinear; + else + spu.sample_texture = sample_texture_nearest; } diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index 02b62ee5cd5..fb98b0d8897 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -114,6 +114,8 @@ struct spu_global vector float tex_size; vector unsigned int tex_size_mask; /**< == int(size - 1) */ + uint (*sample_texture)(vector float texcoord); + } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 6e243f7fa3a..ecacf2ec887 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -131,7 +131,7 @@ get_tex_tile(vector unsigned int ij) * XXX this is extremely primitive for now. */ uint -sample_texture(vector float texcoord) +sample_texture_nearest(vector float texcoord) { vector float tc = spu_mul(texcoord, spu.tex_size); vector unsigned int itc = spu_convtu(tc, 0); /* convert to int */ diff --git a/src/mesa/pipe/cell/spu/spu_texture.h b/src/mesa/pipe/cell/spu/spu_texture.h index 25cbe9b3c64..0e000bfebfd 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.h +++ b/src/mesa/pipe/cell/spu/spu_texture.h @@ -37,7 +37,7 @@ invalidate_tex_cache(void); extern uint -sample_texture(vector float texcoord); +sample_texture_nearest(vector float texcoord); extern uint diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index c148c75dd61..7b422f71a8d 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -309,13 +309,13 @@ emit_quad( int x, int y, mask_t mask ) eval_coeff(2, (float) x, (float) y, texcoords); if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = sample_texture(texcoords[0].v); + spu.ctile.ui[iy][ix] = spu.sample_texture(texcoords[0].v); if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = sample_texture(texcoords[1].v); + spu.ctile.ui[iy][ix+1] = spu.sample_texture(texcoords[1].v); if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = sample_texture(texcoords[2].v); + spu.ctile.ui[iy+1][ix] = spu.sample_texture(texcoords[2].v); if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = sample_texture(texcoords[3].v); + spu.ctile.ui[iy+1][ix+1] = spu.sample_texture(texcoords[3].v); } else { /* simple shading */ -- cgit v1.2.3 From ab9b705c67d0d8c40949ac7e697a8b4ede666c50 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 15:06:10 -0700 Subject: Cell: emit blend state to SPUs --- src/mesa/pipe/cell/common.h | 3 ++- src/mesa/pipe/cell/ppu/cell_state_emit.c | 6 ++++++ src/mesa/pipe/cell/spu/spu_main.c | 17 +++++++++++++++++ src/mesa/pipe/cell/spu/spu_main.h | 1 + 4 files changed, 26 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index 7e193f31be6..d861e82d332 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -85,7 +85,8 @@ #define CELL_CMD_STATE_VERTEX_INFO 14 #define CELL_CMD_STATE_VIEWPORT 15 #define CELL_CMD_STATE_VS_ARRAY_INFO 16 -#define CELL_CMD_VS_EXECUTE 17 +#define CELL_CMD_STATE_BLEND 17 +#define CELL_CMD_VS_EXECUTE 18 #define CELL_NUM_BUFFERS 4 diff --git a/src/mesa/pipe/cell/ppu/cell_state_emit.c b/src/mesa/pipe/cell/ppu/cell_state_emit.c index 702184416bf..3b2670f786b 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_emit.c +++ b/src/mesa/pipe/cell/ppu/cell_state_emit.c @@ -61,6 +61,12 @@ cell_emit_state(struct cell_context *cell) fb->height = cell->framebuffer.cbufs[0]->height; } + if (cell->dirty & CELL_NEW_BLEND) { + emit_state_cmd(cell, CELL_CMD_STATE_BLEND, + cell->blend, + sizeof(struct pipe_blend_state)); + } + if (cell->dirty & CELL_NEW_DEPTH_STENCIL) { emit_state_cmd(cell, CELL_CMD_STATE_DEPTH_STENCIL, cell->depth_stencil, diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 48e016fc8b4..9d8e6df0e30 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -232,6 +232,18 @@ cmd_state_framebuffer(const struct cell_command_framebuffer *cmd) } +static void +cmd_state_blend(const struct pipe_blend_state *state) +{ + if (Debug) + printf("SPU %u: BLEND: ztest %d\n", + spu.init.id, + state->blend_enable); + + memcpy(&spu.blend, state, sizeof(*state)); +} + + static void cmd_state_depth_stencil(const struct pipe_depth_stencil_alpha_state *state) { @@ -398,6 +410,11 @@ cmd_batch(uint opcode) cmd_finish(); pos += 1; break; + case CELL_CMD_STATE_BLEND: + cmd_state_blend((struct pipe_blend_state *) + &buffer[pos+1]); + pos += (1 + sizeof(struct pipe_blend_state) / 4); + break; case CELL_CMD_STATE_DEPTH_STENCIL: cmd_state_depth_stencil((struct pipe_depth_stencil_alpha_state *) &buffer[pos+1]); diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index fb98b0d8897..b22d5635518 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -86,6 +86,7 @@ struct spu_global struct cell_init_info init; struct spu_framebuffer fb; + struct pipe_blend_state blend_stencil; struct pipe_depth_stencil_alpha_state depth_stencil; struct pipe_blend_state blend; struct pipe_sampler_state sampler[PIPE_MAX_SAMPLERS]; -- cgit v1.2.3 From b4f5575add6ad4c8b0f960e10641f361dad74606 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 15:10:35 -0700 Subject: Cell: replace float 4 with vector float in eval_coeff() --- src/mesa/pipe/cell/spu/spu_tri.c | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 7b422f71a8d..199afa1aa69 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -32,6 +32,7 @@ #include "pipe/p_compiler.h" #include "pipe/p_format.h" #include "pipe/p_util.h" +#include "spu_blend.h" #include "spu_colorpack.h" #include "spu_main.h" #include "spu_texture.h" @@ -206,14 +207,14 @@ clip_emit_quad(struct setup_stage *setup) * Eg: four colors will be compute. */ static INLINE void -eval_coeff(uint slot, float x, float y, float4 result[4]) +eval_coeff(uint slot, float x, float y, vector float result[4]) { switch (spu.vertex_info.interp_mode[slot]) { case INTERP_CONSTANT: result[QUAD_TOP_LEFT] = result[QUAD_TOP_RIGHT] = result[QUAD_BOTTOM_LEFT] = - result[QUAD_BOTTOM_RIGHT] = setup.coef[slot].a0; + result[QUAD_BOTTOM_RIGHT] = setup.coef[slot].a0.v; break; case INTERP_LINEAR: @@ -227,10 +228,10 @@ eval_coeff(uint slot, float x, float y, float4 result[4]) spu_add(spu_mul(spu_splats(x), dadx), spu_mul(spu_splats(y), dady))); - result[QUAD_TOP_LEFT].v = topLeft; - result[QUAD_TOP_RIGHT].v = spu_add(topLeft, dadx); - result[QUAD_BOTTOM_LEFT].v = spu_add(topLeft, dady); - result[QUAD_BOTTOM_RIGHT].v = spu_add(spu_add(topLeft, dadx), dady); + result[QUAD_TOP_LEFT] = topLeft; + result[QUAD_TOP_RIGHT] = spu_add(topLeft, dadx); + result[QUAD_BOTTOM_LEFT] = spu_add(topLeft, dady); + result[QUAD_BOTTOM_RIGHT] = spu_add(spu_add(topLeft, dadx), dady); } } } @@ -305,32 +306,32 @@ emit_quad( int x, int y, mask_t mask ) if (spu.texture.start) { /* texture mapping */ - float4 texcoords[4]; + vector float texcoords[4]; eval_coeff(2, (float) x, (float) y, texcoords); if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = spu.sample_texture(texcoords[0].v); + spu.ctile.ui[iy][ix] = spu.sample_texture(texcoords[0]); if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = spu.sample_texture(texcoords[1].v); + spu.ctile.ui[iy][ix+1] = spu.sample_texture(texcoords[1]); if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = spu.sample_texture(texcoords[2].v); + spu.ctile.ui[iy+1][ix] = spu.sample_texture(texcoords[2]); if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = spu.sample_texture(texcoords[3].v); + spu.ctile.ui[iy+1][ix+1] = spu.sample_texture(texcoords[3]); } else { /* simple shading */ const vector unsigned char shuffle = spu.color_shuffle; - float4 colors[4]; + vector float colors[4]; eval_coeff(1, (float) x, (float) y, colors); if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = spu_pack_color_shuffle(colors[0].v, shuffle); + spu.ctile.ui[iy][ix] = spu_pack_color_shuffle(colors[0], shuffle); if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = spu_pack_color_shuffle(colors[1].v, shuffle); + spu.ctile.ui[iy][ix+1] = spu_pack_color_shuffle(colors[1], shuffle); if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = spu_pack_color_shuffle(colors[2].v, shuffle); + spu.ctile.ui[iy+1][ix] = spu_pack_color_shuffle(colors[2], shuffle); if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = spu_pack_color_shuffle(colors[3].v, shuffle); + spu.ctile.ui[iy+1][ix+1] = spu_pack_color_shuffle(colors[3], shuffle); } #if 0 -- cgit v1.2.3 From 21461014b2446208fefae0aabe8232c66d5b3057 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 15:17:50 -0700 Subject: Cell: some basic blending code --- src/mesa/pipe/cell/spu/Makefile | 1 + src/mesa/pipe/cell/spu/spu_blend.c | 62 ++++++++++++++++++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_blend.h | 37 +++++++++++++++++++++++ src/mesa/pipe/cell/spu/spu_tri.c | 5 +++ 4 files changed, 105 insertions(+) create mode 100644 src/mesa/pipe/cell/spu/spu_blend.c create mode 100644 src/mesa/pipe/cell/spu/spu_blend.h diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index 91a631b6992..66f16cde9b8 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -19,6 +19,7 @@ PROG_SPU_EMBED_O = $(PROG)_spu-embed.o SOURCES = \ spu_main.c \ + spu_blend.c \ spu_render.c \ spu_texture.c \ spu_tile.c \ diff --git a/src/mesa/pipe/cell/spu/spu_blend.c b/src/mesa/pipe/cell/spu/spu_blend.c new file mode 100644 index 00000000000..23ec0eeb451 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_blend.c @@ -0,0 +1,62 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include "spu_main.h" +#include "spu_blend.h" +#include "spu_colorpack.h" + + +void +blend_quad(uint itx, uint ity, vector float colors[4]) +{ + /* simple SRC_ALPHA, ONE_MINUS_SRC_ALPHA blending */ + vector float fbc00 = spu_unpack_color(spu.ctile.ui[ity][itx]); + vector float fbc01 = spu_unpack_color(spu.ctile.ui[ity][itx+1]); + vector float fbc10 = spu_unpack_color(spu.ctile.ui[ity+1][itx]); + vector float fbc11 = spu_unpack_color(spu.ctile.ui[ity+1][itx+1]); + + vector float alpha00 = spu_splats(spu_extract(colors[0], 3)); + vector float alpha01 = spu_splats(spu_extract(colors[1], 3)); + vector float alpha10 = spu_splats(spu_extract(colors[2], 3)); + vector float alpha11 = spu_splats(spu_extract(colors[3], 3)); + + vector float one_minus_alpha00 = spu_sub(spu_splats(1.0f), alpha00); + vector float one_minus_alpha01 = spu_sub(spu_splats(1.0f), alpha01); + vector float one_minus_alpha10 = spu_sub(spu_splats(1.0f), alpha10); + vector float one_minus_alpha11 = spu_sub(spu_splats(1.0f), alpha11); + + colors[0] = spu_add(spu_mul(colors[0], alpha00), + spu_mul(fbc00, one_minus_alpha00)); + colors[1] = spu_add(spu_mul(colors[1], alpha01), + spu_mul(fbc01, one_minus_alpha01)); + colors[2] = spu_add(spu_mul(colors[2], alpha10), + spu_mul(fbc10, one_minus_alpha10)); + colors[3] = spu_add(spu_mul(colors[3], alpha11), + spu_mul(fbc11, one_minus_alpha11)); +} + diff --git a/src/mesa/pipe/cell/spu/spu_blend.h b/src/mesa/pipe/cell/spu/spu_blend.h new file mode 100644 index 00000000000..2b594b578b4 --- /dev/null +++ b/src/mesa/pipe/cell/spu/spu_blend.h @@ -0,0 +1,37 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef SPU_BLEND_H +#define SPU_BLEND_H + + +extern void +blend_quad(uint itx, uint ity, vector float colors[4]); + + +#endif /* SPU_BLEND_H */ diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 199afa1aa69..89aaca9a72e 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -324,6 +324,11 @@ emit_quad( int x, int y, mask_t mask ) vector float colors[4]; eval_coeff(1, (float) x, (float) y, colors); +#if 0 + if (spu.blend.blend_enable) + blend_quad(ix % TILE_SIZE, iy % TILE_SIZE, colors); +#endif + if (spu_extract(mask, 0)) spu.ctile.ui[iy][ix] = spu_pack_color_shuffle(colors[0], shuffle); if (spu_extract(mask, 1)) -- cgit v1.2.3 From 2f8268aa02949828b8b14c252e2cc4e8f61c5f4e Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 16:04:50 -0700 Subject: Cell: fix typo --- src/mesa/pipe/cell/spu/spu_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 9d8e6df0e30..b0311db1aab 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -236,7 +236,7 @@ static void cmd_state_blend(const struct pipe_blend_state *state) { if (Debug) - printf("SPU %u: BLEND: ztest %d\n", + printf("SPU %u: BLEND: enabled %d\n", spu.init.id, state->blend_enable); -- cgit v1.2.3 From 71e6cd0b66be784aa3feb86101b7a62d17735f56 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 16:05:13 -0700 Subject: Cell: added spu_unpack_A8R8G8B8() --- src/mesa/pipe/cell/spu/spu_colorpack.h | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_colorpack.h b/src/mesa/pipe/cell/spu/spu_colorpack.h index 0c93c065627..57ea3525c27 100644 --- a/src/mesa/pipe/cell/spu/spu_colorpack.h +++ b/src/mesa/pipe/cell/spu/spu_colorpack.h @@ -91,4 +91,19 @@ spu_unpack_color(uint color) } +static INLINE vector float +spu_unpack_A8R8G8B8(uint color) +{ + vector unsigned int color_u4 = spu_splats(color); + color_u4 = spu_shuffle(color_u4, color_u4, + VEC_LITERAL(vector unsigned char, + 5, 5, 5, 5, + 10, 10, 10, 10, + 15, 15, 15, 15, + 0, 0, 0, 0)); + + return spu_convtf(color_u4, 32); +} + + #endif /* SPU_COLORPACK_H */ -- cgit v1.2.3 From 790eec9666ae6cc37ce4ba54cceff97e9eeb5ce3 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 16:06:51 -0700 Subject: Cell: texture sampler functions always return vector float now Texture colors look the same now, regardless of X display/pixel format --- src/mesa/pipe/cell/spu/spu_main.h | 2 +- src/mesa/pipe/cell/spu/spu_texture.c | 19 ++++++++----------- src/mesa/pipe/cell/spu/spu_texture.h | 4 ++-- src/mesa/pipe/cell/spu/spu_tri.c | 36 ++++++++++++++++++------------------ 4 files changed, 29 insertions(+), 32 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index b22d5635518..cfd4d727290 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -115,7 +115,7 @@ struct spu_global vector float tex_size; vector unsigned int tex_size_mask; /**< == int(size - 1) */ - uint (*sample_texture)(vector float texcoord); + vector float (*sample_texture)(vector float texcoord); } ALIGN16_ATTRIB; diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index ecacf2ec887..9ee2b45e24e 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -130,7 +130,7 @@ get_tex_tile(vector unsigned int ij) * Get texture sample at texcoord. * XXX this is extremely primitive for now. */ -uint +vector float sample_texture_nearest(vector float texcoord) { vector float tc = spu_mul(texcoord, spu.tex_size); @@ -139,11 +139,11 @@ sample_texture_nearest(vector float texcoord) vector unsigned int ij = spu_and(itc, TILE_SIZE-1); /* intra tile addr */ uint pos = get_tex_tile(itc); uint texel = tex_tiles[pos].ui[spu_extract(ij, 1)][spu_extract(ij, 0)]; - return texel; + return spu_unpack_A8R8G8B8(texel); } -uint +vector float sample_texture_bilinear(vector float texcoord) { static const vector unsigned int offset10 = {1, 0, 0, 0}; @@ -183,10 +183,10 @@ sample_texture_bilinear(vector float texcoord) } /* get texels from tiles and convert to float[4] */ - vector float texel00 = spu_unpack_color(tex_tiles[pos00].ui[spu_extract(ij00, 1)][spu_extract(ij00, 0)]); - vector float texel01 = spu_unpack_color(tex_tiles[pos01].ui[spu_extract(ij01, 1)][spu_extract(ij01, 0)]); - vector float texel10 = spu_unpack_color(tex_tiles[pos10].ui[spu_extract(ij10, 1)][spu_extract(ij10, 0)]); - vector float texel11 = spu_unpack_color(tex_tiles[pos11].ui[spu_extract(ij11, 1)][spu_extract(ij11, 0)]); + vector float texel00 = spu_unpack_A8R8G8B8(tex_tiles[pos00].ui[spu_extract(ij00, 1)][spu_extract(ij00, 0)]); + vector float texel01 = spu_unpack_A8R8G8B8(tex_tiles[pos01].ui[spu_extract(ij01, 1)][spu_extract(ij01, 0)]); + vector float texel10 = spu_unpack_A8R8G8B8(tex_tiles[pos10].ui[spu_extract(ij10, 1)][spu_extract(ij10, 0)]); + vector float texel11 = spu_unpack_A8R8G8B8(tex_tiles[pos11].ui[spu_extract(ij11, 1)][spu_extract(ij11, 0)]); /* Compute weighting factors in [0,1] * Multiply texcoord by 1024, AND with 1023, convert back to float. @@ -213,8 +213,5 @@ sample_texture_bilinear(vector float texcoord) texel_sum = spu_add(texel_sum, texel10); texel_sum = spu_add(texel_sum, texel11); - /* convert to uint color */ - uint texel = spu_pack_R8G8B8A8(texel_sum); - - return texel; + return texel_sum; } diff --git a/src/mesa/pipe/cell/spu/spu_texture.h b/src/mesa/pipe/cell/spu/spu_texture.h index 0e000bfebfd..95eb87080f1 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.h +++ b/src/mesa/pipe/cell/spu/spu_texture.h @@ -36,11 +36,11 @@ extern void invalidate_tex_cache(void); -extern uint +extern vector float sample_texture_nearest(vector float texcoord); -extern uint +extern vector float sample_texture_bilinear(vector float texcoord); diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 89aaca9a72e..4c6de56eda5 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -301,6 +301,8 @@ emit_quad( int x, int y, mask_t mask ) if (spu_extract(spu_orx(mask), 0)) { const int ix = x - setup.cliprect_minx; const int iy = y - setup.cliprect_miny; + const vector unsigned char shuffle = spu.color_shuffle; + vector float colors[4]; spu.cur_ctile_status = TILE_STATUS_DIRTY; @@ -310,34 +312,32 @@ emit_quad( int x, int y, mask_t mask ) eval_coeff(2, (float) x, (float) y, texcoords); if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = spu.sample_texture(texcoords[0]); + colors[0] = spu.sample_texture(texcoords[0]); if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = spu.sample_texture(texcoords[1]); + colors[1] = spu.sample_texture(texcoords[1]); if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = spu.sample_texture(texcoords[2]); + colors[2] = spu.sample_texture(texcoords[2]); if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = spu.sample_texture(texcoords[3]); + colors[3] = spu.sample_texture(texcoords[3]); } else { /* simple shading */ - const vector unsigned char shuffle = spu.color_shuffle; - vector float colors[4]; eval_coeff(1, (float) x, (float) y, colors); + } -#if 0 - if (spu.blend.blend_enable) - blend_quad(ix % TILE_SIZE, iy % TILE_SIZE, colors); +#if 1 + if (spu.blend.blend_enable) + blend_quad(ix % TILE_SIZE, iy % TILE_SIZE, colors); #endif - if (spu_extract(mask, 0)) - spu.ctile.ui[iy][ix] = spu_pack_color_shuffle(colors[0], shuffle); - if (spu_extract(mask, 1)) - spu.ctile.ui[iy][ix+1] = spu_pack_color_shuffle(colors[1], shuffle); - if (spu_extract(mask, 2)) - spu.ctile.ui[iy+1][ix] = spu_pack_color_shuffle(colors[2], shuffle); - if (spu_extract(mask, 3)) - spu.ctile.ui[iy+1][ix+1] = spu_pack_color_shuffle(colors[3], shuffle); - } + if (spu_extract(mask, 0)) + spu.ctile.ui[iy][ix] = spu_pack_color_shuffle(colors[0], shuffle); + if (spu_extract(mask, 1)) + spu.ctile.ui[iy][ix+1] = spu_pack_color_shuffle(colors[1], shuffle); + if (spu_extract(mask, 2)) + spu.ctile.ui[iy+1][ix] = spu_pack_color_shuffle(colors[2], shuffle); + if (spu_extract(mask, 3)) + spu.ctile.ui[iy+1][ix+1] = spu_pack_color_shuffle(colors[3], shuffle); #if 0 /* SIMD_Z with swizzled color buffer (someday) */ -- cgit v1.2.3 From d17e3362592c58f0d5d47745fd97b3b31d1a684a Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 16:48:00 -0700 Subject: Cell: fix small sampling error in sample_texture_bilinear() --- src/mesa/pipe/cell/spu/spu_texture.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 9ee2b45e24e..01ff33a8573 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -150,6 +150,8 @@ sample_texture_bilinear(vector float texcoord) static const vector unsigned int offset01 = {0, 1, 0, 0}; vector float tc = spu_mul(texcoord, spu.tex_size); + tc = spu_add(tc, spu_splats(-0.5f)); /* half texel bias */ + /* integer texcoords S,T: */ vector unsigned int itc00 = spu_convtu(tc, 0); /* convert to int */ vector unsigned int itc01 = spu_add(itc00, offset01); -- cgit v1.2.3 From 4540e01978280389ed219aa0a4b4f39db280961f Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 16:48:36 -0700 Subject: Cell: move float4 typedef (temporary datatype) --- src/mesa/pipe/cell/spu/spu_main.h | 7 ------- src/mesa/pipe/cell/spu/spu_tri.c | 6 ++++++ 2 files changed, 6 insertions(+), 7 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_main.h b/src/mesa/pipe/cell/spu/spu_main.h index cfd4d727290..1710a175123 100644 --- a/src/mesa/pipe/cell/spu/spu_main.h +++ b/src/mesa/pipe/cell/spu/spu_main.h @@ -41,13 +41,6 @@ #define MAX_HEIGHT 1024 -typedef union -{ - vector float v; - float f[4]; -} float4; - - typedef union { ushort us[TILE_SIZE][TILE_SIZE]; uint ui[TILE_SIZE][TILE_SIZE]; diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 4c6de56eda5..688c8646ab2 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -45,6 +45,12 @@ /** Masks are uint[4] vectors with each element being 0 or 0xffffffff */ typedef vector unsigned int mask_t; +typedef union +{ + vector float v; + float f[4]; +} float4; + /** * Simplified types taken from other parts of Gallium -- cgit v1.2.3 From f603652c30c40f7f7948fbdc79a3479016d8073f Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 18:02:21 -0700 Subject: Cell: don't use VEC_LITERAL macro, doesn't work w/ SDK 3.0 --- src/mesa/pipe/cell/spu/spu_colorpack.h | 41 +++++++++++++++++----------------- src/mesa/pipe/cell/spu/spu_ztest.h | 24 ++++++++++---------- 2 files changed, 33 insertions(+), 32 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_colorpack.h b/src/mesa/pipe/cell/spu/spu_colorpack.h index 57ea3525c27..e9fee8a3a61 100644 --- a/src/mesa/pipe/cell/spu/spu_colorpack.h +++ b/src/mesa/pipe/cell/spu/spu_colorpack.h @@ -31,7 +31,6 @@ #define SPU_COLORPACK_H -#include #include @@ -39,9 +38,11 @@ static INLINE unsigned int spu_pack_R8G8B8A8(vector float rgba) { vector unsigned int out = spu_convtu(rgba, 32); - out = spu_shuffle(out, out, VEC_LITERAL(vector unsigned char, - 0, 4, 8, 12, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0)); + + out = spu_shuffle(out, out, ((vector unsigned char) { + 0, 4, 8, 12, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0 }) ); + return spu_extract(out, 0); } @@ -50,9 +51,9 @@ static INLINE unsigned int spu_pack_A8R8G8B8(vector float rgba) { vector unsigned int out = spu_convtu(rgba, 32); - out = spu_shuffle(out, out, VEC_LITERAL(vector unsigned char, - 12, 0, 4, 8, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0)); + out = spu_shuffle(out, out, ((vector unsigned char) { + 12, 0, 4, 8, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0}) ); return spu_extract(out, 0); } @@ -61,9 +62,9 @@ static INLINE unsigned int spu_pack_B8G8R8A8(vector float rgba) { vector unsigned int out = spu_convtu(rgba, 32); - out = spu_shuffle(out, out, VEC_LITERAL(vector unsigned char, - 8, 4, 0, 12, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0)); + out = spu_shuffle(out, out, ((vector unsigned char) { + 8, 4, 0, 12, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0}) ); return spu_extract(out, 0); } @@ -82,11 +83,11 @@ spu_unpack_color(uint color) { vector unsigned int color_u4 = spu_splats(color); color_u4 = spu_shuffle(color_u4, color_u4, - VEC_LITERAL(vector unsigned char, - 0, 0, 0, 0, - 5, 5, 5, 5, - 10, 10, 10, 10, - 15, 15, 15, 15)); + ((vector unsigned char) { + 0, 0, 0, 0, + 5, 5, 5, 5, + 10, 10, 10, 10, + 15, 15, 15, 15}) ); return spu_convtf(color_u4, 32); } @@ -96,11 +97,11 @@ spu_unpack_A8R8G8B8(uint color) { vector unsigned int color_u4 = spu_splats(color); color_u4 = spu_shuffle(color_u4, color_u4, - VEC_LITERAL(vector unsigned char, - 5, 5, 5, 5, - 10, 10, 10, 10, - 15, 15, 15, 15, - 0, 0, 0, 0)); + ((vector unsigned char) { + 5, 5, 5, 5, + 10, 10, 10, 10, + 15, 15, 15, 15, + 0, 0, 0, 0}) ); return spu_convtf(color_u4, 32); } diff --git a/src/mesa/pipe/cell/spu/spu_ztest.h b/src/mesa/pipe/cell/spu/spu_ztest.h index 5fefb151765..ce8ad003393 100644 --- a/src/mesa/pipe/cell/spu/spu_ztest.h +++ b/src/mesa/pipe/cell/spu/spu_ztest.h @@ -68,9 +68,9 @@ spu_z16_test_less(vector float zvals, vector unsigned short *zbuf, /* gather lower four ushorts */ zbuf_ui4 = spu_shuffle((vector unsigned int) *zbuf, (vector unsigned int) *zbuf, - VEC_LITERAL(vector unsigned char, - ZERO, ZERO, 8, 9, ZERO, ZERO, 10, 11, - ZERO, ZERO, 12, 13, ZERO, ZERO, 14, 15)); + ((vector unsigned char) { + ZERO, ZERO, 8, 9, ZERO, ZERO, 10, 11, + ZERO, ZERO, 12, 13, ZERO, ZERO, 14, 15})); /* mask = (zbuf_ui4 < zvals_ui4) ? ~0 : 0 */ mask = spu_cmpgt(zbuf_ui4, zvals_ui4); /* mask &= inMask */ @@ -80,18 +80,18 @@ spu_z16_test_less(vector float zvals, vector unsigned short *zbuf, /* convert zbuffer values from uints back to ushorts, preserve lower 4 */ *zbuf = (vector unsigned short) spu_shuffle(zbuf_ui4, (vector unsigned int) *zbuf, - VEC_LITERAL(vector unsigned char, - 16, 17, 18, 19, 20, 21, 22, 23, - 2, 3, 6, 7, 10, 11, 14, 15)); + ((vector unsigned char) { + 16, 17, 18, 19, 20, 21, 22, 23, + 2, 3, 6, 7, 10, 11, 14, 15})); } else { /* convert zbuffer values from ushorts to uints */ /* gather upper four ushorts */ zbuf_ui4 = spu_shuffle((vector unsigned int) *zbuf, (vector unsigned int) *zbuf, - VEC_LITERAL(vector unsigned char, - ZERO, ZERO, 0, 1, ZERO, ZERO, 2, 3, - ZERO, ZERO, 4, 5, ZERO, ZERO, 6, 7)); + ((vector unsigned char) { + ZERO, ZERO, 0, 1, ZERO, ZERO, 2, 3, + ZERO, ZERO, 4, 5, ZERO, ZERO, 6, 7})); /* mask = (zbuf_ui4 < zvals_ui4) ? ~0 : 0 */ mask = spu_cmpgt(zbuf_ui4, zvals_ui4); /* mask &= inMask */ @@ -101,9 +101,9 @@ spu_z16_test_less(vector float zvals, vector unsigned short *zbuf, /* convert zbuffer values from uints back to ushorts, preserve upper 4 */ *zbuf = (vector unsigned short) spu_shuffle(zbuf_ui4, (vector unsigned int) *zbuf, - VEC_LITERAL(vector unsigned char, - 2, 3, 6, 7, 10, 11, 14, 15, - 24, 25, 26, 27, 28, 29, 30, 31)); + ((vector unsigned char) { + 2, 3, 6, 7, 10, 11, 14, 15, + 24, 25, 26, 27, 28, 29, 30, 31})); } return mask; #undef ZERO -- cgit v1.2.3 From 6a3f1ea91d3d8e4c47144cda422db9db761be94d Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 18:03:05 -0700 Subject: Cell: don't use VEC_LITERAL macro, doesn't work w/ SDK 3.0 --- src/mesa/pipe/cell/spu/spu_texture.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_texture.c b/src/mesa/pipe/cell/spu/spu_texture.c index 01ff33a8573..3962aaa4a9b 100644 --- a/src/mesa/pipe/cell/spu/spu_texture.c +++ b/src/mesa/pipe/cell/spu/spu_texture.c @@ -26,8 +26,6 @@ **************************************************************************/ -#include - #include "pipe/p_compiler.h" #include "spu_main.h" #include "spu_texture.h" @@ -57,7 +55,7 @@ invalidate_tex_cache(void) /* XXX memset? */ uint i; for (i = 0; i < CACHE_SIZE; i++) { - tex_tile_xy[i] = VEC_LITERAL(vector unsigned int, ~0U, ~0U, ~0U, ~0U); + tex_tile_xy[i] = ((vector unsigned int) { ~0U, ~0U, ~0U, ~0U }); } } -- cgit v1.2.3 From 535abe4037920960b37a23392142cc556d4cbcc4 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 4 Feb 2008 18:05:37 -0700 Subject: Cell: fix some alignment issues by aligning commands to 8-byte boundaries Contributed by Ian Romanick. Also, temporarily disable inlined vertex buffers. They need to be 16-byte aligned... --- src/mesa/pipe/cell/common.h | 16 ++++---- src/mesa/pipe/cell/ppu/cell_batch.c | 4 +- src/mesa/pipe/cell/ppu/cell_flush.c | 2 +- src/mesa/pipe/cell/ppu/cell_state_emit.c | 3 +- src/mesa/pipe/cell/ppu/cell_vbuf.c | 4 +- src/mesa/pipe/cell/ppu/cell_vertex_shader.c | 22 ++++++----- src/mesa/pipe/cell/spu/spu_main.c | 58 +++++++++++++---------------- src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 7 ++-- src/mesa/pipe/cell/spu/spu_vertex_shader.h | 2 +- 9 files changed, 57 insertions(+), 61 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index d861e82d332..cf8fc94ebf1 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -57,6 +57,9 @@ /** round up value to next multiple of 4 */ #define ROUNDUP4(k) (((k) + 0x3) & ~0x3) +/** round up value to next multiple of 8 */ +#define ROUNDUP8(k) (((k) + 0x7) & ~0x7) + /** round up value to next multiple of 16 */ #define ROUNDUP16(k) (((k) + 0xf) & ~0xf) @@ -102,7 +105,7 @@ */ struct cell_command_framebuffer { - uint opcode; + uint64_t opcode; int width, height; void *color_start, *depth_start; enum pipe_format color_format, depth_format; @@ -114,7 +117,7 @@ struct cell_command_framebuffer */ struct cell_command_clear_surface { - uint opcode; + uint64_t opcode; uint surface; /**< Temporary: 0=color, 1=Z */ uint value; }; @@ -125,8 +128,7 @@ struct cell_command_clear_surface */ struct cell_array_info { - uint opcode; - uint base; /**< Base address of the 0th element. */ + uint64_t base; /**< Base address of the 0th element. */ uint attr; /**< Attribute that this state if for. */ uint pitch; /**< Byte pitch from one entry to the next. */ uint format; /**< Pipe format of each entry. */ @@ -150,7 +152,7 @@ struct cell_shader_info #define SPU_VERTS_PER_BATCH 64 struct cell_command_vs { - uint opcode; /**< CELL_CMD_VS_EXECUTE */ + uint64_t opcode; /**< CELL_CMD_VS_EXECUTE */ struct cell_shader_info shader; unsigned num_elts; unsigned elts[SPU_VERTS_PER_BATCH]; @@ -163,7 +165,7 @@ struct cell_command_vs struct cell_command_render { - uint opcode; /**< CELL_CMD_RENDER */ + uint64_t opcode; /**< CELL_CMD_RENDER */ uint prim_type; /**< PIPE_PRIM_x */ uint num_verts; uint vertex_size; /**< bytes per vertex */ @@ -179,7 +181,7 @@ struct cell_command_render struct cell_command_release_verts { - int opcode; /**< CELL_CMD_RELEASE_VERTS */ + uint64_t opcode; /**< CELL_CMD_RELEASE_VERTS */ uint vertex_buf; /**< in [0, CELL_NUM_BUFFERS-1] */ }; diff --git a/src/mesa/pipe/cell/ppu/cell_batch.c b/src/mesa/pipe/cell/ppu/cell_batch.c index 2d032fc9026..2fb49711b2d 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.c +++ b/src/mesa/pipe/cell/ppu/cell_batch.c @@ -136,7 +136,7 @@ cell_batch_append(struct cell_context *cell, const void *data, uint bytes) { uint size; - ASSERT(bytes % 4 == 0); + ASSERT(bytes % 8 == 0); ASSERT(bytes <= CELL_BUFFER_SIZE); ASSERT(cell->cur_batch >= 0); @@ -171,7 +171,7 @@ cell_batch_alloc(struct cell_context *cell, uint bytes) void *pos; uint size; - ASSERT(bytes % 4 == 0); + ASSERT(bytes % 8 == 0); ASSERT(bytes <= CELL_BUFFER_SIZE); assert(cell->cur_batch >= 0); diff --git a/src/mesa/pipe/cell/ppu/cell_flush.c b/src/mesa/pipe/cell/ppu/cell_flush.c index cf4e676645b..f62bc4650ce 100644 --- a/src/mesa/pipe/cell/ppu/cell_flush.c +++ b/src/mesa/pipe/cell/ppu/cell_flush.c @@ -59,7 +59,7 @@ cell_flush_int(struct pipe_context *pipe, unsigned flags) flushing = TRUE; if (flags & PIPE_FLUSH_WAIT) { - uint *cmd = (uint *) cell_batch_alloc(cell, sizeof(uint)); + uint64_t *cmd = (uint64_t *) cell_batch_alloc(cell, sizeof(uint64_t)); *cmd = CELL_CMD_FINISH; } diff --git a/src/mesa/pipe/cell/ppu/cell_state_emit.c b/src/mesa/pipe/cell/ppu/cell_state_emit.c index 3b2670f786b..5d2a7864493 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_emit.c +++ b/src/mesa/pipe/cell/ppu/cell_state_emit.c @@ -37,7 +37,8 @@ static void emit_state_cmd(struct cell_context *cell, uint cmd, const void *state, uint state_size) { - uint *dst = (uint *) cell_batch_alloc(cell, sizeof(uint) + state_size); + uint64_t *dst = (uint64_t *) + cell_batch_alloc(cell, ROUNDUP8(sizeof(uint64_t) + state_size)); *dst = cmd; memcpy(dst + 1, state, state_size); } diff --git a/src/mesa/pipe/cell/ppu/cell_vbuf.c b/src/mesa/pipe/cell/ppu/cell_vbuf.c index e63b34cf525..0fee61821a8 100644 --- a/src/mesa/pipe/cell/ppu/cell_vbuf.c +++ b/src/mesa/pipe/cell/ppu/cell_vbuf.c @@ -40,7 +40,7 @@ /** Allow vertex data to be inlined after RENDER command */ -#define ALLOW_INLINE_VERTS 1 +#define ALLOW_INLINE_VERTS 0 /** @@ -197,7 +197,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, /* build/insert batch RENDER command */ { - const uint index_bytes = ROUNDUP4(nr_indices * 2); + const uint index_bytes = ROUNDUP8(nr_indices * 2); const uint vertex_bytes = nr_vertices * 4 * cell->vertex_info.size; const uint batch_size = sizeof(struct cell_command_render) diff --git a/src/mesa/pipe/cell/ppu/cell_vertex_shader.c b/src/mesa/pipe/cell/ppu/cell_vertex_shader.c index aef329a9024..80dd500b345 100644 --- a/src/mesa/pipe/cell/ppu/cell_vertex_shader.c +++ b/src/mesa/pipe/cell/ppu/cell_vertex_shader.c @@ -52,8 +52,8 @@ cell_vertex_shader_queue_flush(struct draw_context *draw) struct cell_context *const cell = (struct cell_context *) draw->driver_private; struct cell_command_vs *const vs = &cell_global.command[0].vs; - unsigned *batch; - struct cell_array_info array_info; + uint64_t *batch; + struct cell_array_info *array_info; unsigned i, j; assert(draw->vs.queue_nr != 0); @@ -63,17 +63,19 @@ cell_vertex_shader_queue_flush(struct draw_context *draw) draw_update_vertex_fetch(draw); for (i = 0; i < draw->vertex_fetch.nr_attrs; i++) { - array_info.opcode = CELL_CMD_STATE_VS_ARRAY_INFO; - assert(draw->vertex_fetch.src_ptr[i] != NULL); - array_info.base = (uintptr_t) draw->vertex_fetch.src_ptr[i]; - array_info.attr = i; - array_info.pitch = draw->vertex_fetch.pitch[i]; - array_info.format = draw->vertex_element[i].src_format; + batch = cell_batch_alloc(cell, sizeof(batch[0]) + sizeof(*array_info)); + + batch[0] = CELL_CMD_STATE_VS_ARRAY_INFO; - cell_batch_append(cell, & array_info, sizeof(array_info)); + array_info = (struct cell_array_info *) &batch[1]; + assert(draw->vertex_fetch.src_ptr[i] != NULL); + array_info->base = (uintptr_t) draw->vertex_fetch.src_ptr[i]; + array_info->attr = i; + array_info->pitch = draw->vertex_fetch.pitch[i]; + array_info->format = draw->vertex_element[i].src_format; } - batch = cell_batch_alloc(cell, sizeof(unsigned) + batch = cell_batch_alloc(cell, sizeof(batch[0]) + sizeof(struct pipe_viewport_state)); batch[0] = CELL_CMD_STATE_VIEWPORT; (void) memcpy(&batch[1], &draw->viewport, diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index b0311db1aab..4f126d5e5bb 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -31,7 +31,6 @@ #include #include -#include #include "spu_main.h" #include "spu_render.h" @@ -220,13 +219,13 @@ cmd_state_framebuffer(const struct cell_command_framebuffer *cmd) spu.fb.zsize = 0; if (spu.fb.color_format == PIPE_FORMAT_A8R8G8B8_UNORM) - spu.color_shuffle = VEC_LITERAL(vector unsigned char, - 12, 0, 4, 8, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0); + spu.color_shuffle = ((vector unsigned char) { + 12, 0, 4, 8, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0}); else if (spu.fb.color_format == PIPE_FORMAT_B8G8R8A8_UNORM) - spu.color_shuffle = VEC_LITERAL(vector unsigned char, - 8, 4, 0, 12, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0); + spu.color_shuffle = ((vector unsigned char) { + 8, 4, 0, 12, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0}); else ASSERT(0); } @@ -279,16 +278,10 @@ cmd_state_texture(const struct cell_command_texture *texture) spu.init.id, texture->start, texture->width, texture->height); memcpy(&spu.texture, texture, sizeof(*texture)); - spu.tex_size = VEC_LITERAL(vector float, - spu.texture.width, - spu.texture.height, - 0.0, - 0.0); - spu.tex_size_mask = VEC_LITERAL(vector unsigned int, - spu.texture.width - 1, - spu.texture.height - 1, - 0, - 0); + spu.tex_size = (vector float) + { spu.texture.width, spu.texture.height, 0.0, 0.0}; + spu.tex_size_mask = (vector unsigned int) + { spu.texture.width - 1, spu.texture.height - 1, 0, 0 }; } @@ -341,8 +334,8 @@ cmd_batch(uint opcode) { const uint buf = (opcode >> 8) & 0xff; uint size = (opcode >> 16); - uint buffer[CELL_BUFFER_SIZE / 4] ALIGN16_ATTRIB; - const uint usize = size / sizeof(uint); + uint64_t buffer[CELL_BUFFER_SIZE / 8] ALIGN16_ATTRIB; + const unsigned usize = size / sizeof(buffer[0]); uint pos; if (Debug) @@ -377,7 +370,7 @@ cmd_batch(uint opcode) struct cell_command_framebuffer *fb = (struct cell_command_framebuffer *) &buffer[pos]; cmd_state_framebuffer(fb); - pos += sizeof(*fb) / 4; + pos += sizeof(*fb) / 8; } break; case CELL_CMD_CLEAR_SURFACE: @@ -385,7 +378,7 @@ cmd_batch(uint opcode) struct cell_command_clear_surface *clr = (struct cell_command_clear_surface *) &buffer[pos]; cmd_clear_surface(clr); - pos += sizeof(*clr) / 4; + pos += sizeof(*clr) / 8; } break; case CELL_CMD_RENDER: @@ -394,7 +387,7 @@ cmd_batch(uint opcode) = (struct cell_command_render *) &buffer[pos]; uint pos_incr; cmd_render(render, &pos_incr); - pos += sizeof(*render) / 4 + pos_incr; + pos += sizeof(*render) / 8 + ((pos_incr + 1) / 2); } break; case CELL_CMD_RELEASE_VERTS: @@ -402,8 +395,7 @@ cmd_batch(uint opcode) struct cell_command_release_verts *release = (struct cell_command_release_verts *) &buffer[pos]; cmd_release_verts(release); - ASSERT(sizeof(*release) == 8); - pos += sizeof(*release) / 4; + pos += sizeof(*release) / 8; } break; case CELL_CMD_FINISH: @@ -413,36 +405,36 @@ cmd_batch(uint opcode) case CELL_CMD_STATE_BLEND: cmd_state_blend((struct pipe_blend_state *) &buffer[pos+1]); - pos += (1 + sizeof(struct pipe_blend_state) / 4); + pos += (1 + ROUNDUP8(sizeof(struct pipe_blend_state)) / 8); break; case CELL_CMD_STATE_DEPTH_STENCIL: cmd_state_depth_stencil((struct pipe_depth_stencil_alpha_state *) &buffer[pos+1]); - pos += (1 + sizeof(struct pipe_depth_stencil_alpha_state) / 4); + pos += (1 + ROUNDUP8(sizeof(struct pipe_depth_stencil_alpha_state)) / 8); break; case CELL_CMD_STATE_SAMPLER: cmd_state_sampler((struct pipe_sampler_state *) &buffer[pos+1]); - pos += (1 + sizeof(struct pipe_sampler_state) / 4); + pos += (1 + ROUNDUP8(sizeof(struct pipe_sampler_state)) / 8); break; case CELL_CMD_STATE_TEXTURE: cmd_state_texture((struct cell_command_texture *) &buffer[pos+1]); - pos += (1 + sizeof(struct cell_command_texture) / 4); + pos += (1 + ROUNDUP8(sizeof(struct cell_command_texture)) / 8); break; case CELL_CMD_STATE_VERTEX_INFO: cmd_state_vertex_info((struct vertex_info *) &buffer[pos+1]); - pos += (1 + sizeof(struct vertex_info) / 4); + pos += (1 + ROUNDUP8(sizeof(struct vertex_info)) / 8); break; case CELL_CMD_STATE_VIEWPORT: (void) memcpy(& draw.viewport, &buffer[pos+1], sizeof(struct pipe_viewport_state)); - pos += (1 + sizeof(struct pipe_viewport_state) / 4); + pos += (1 + ROUNDUP8(sizeof(struct pipe_viewport_state)) / 8); break; case CELL_CMD_STATE_VS_ARRAY_INFO: - cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos]); - pos += (sizeof(struct cell_array_info) / 4); + cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]); + pos += (1 + ROUNDUP8(sizeof(struct cell_array_info)) / 8); break; default: - printf("SPU %u: bad opcode: 0x%x\n", spu.init.id, buffer[pos]); + printf("SPU %u: bad opcode: 0x%llx\n", spu.init.id, buffer[pos]); ASSERT(0); break; } diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c index 1e846868e38..5b0f2a6470c 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -431,9 +431,8 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, /* loop over vertex attributes (vertex shader inputs) */ for (attr = 0; attr < nr_attrs; attr++) { - - const unsigned pitch = draw->vertex_fetch.pitch[attr]; - const ubyte *src = draw->vertex_fetch.src_ptr[attr]; + const unsigned pitch = draw->vertex_fetch.pitch[attr]; + const uint64_t src = draw->vertex_fetch.src_ptr[attr]; const spu_fetch_func fetch = draw->vertex_fetch.fetch[attr]; unsigned i; float p[4][4]; @@ -447,7 +446,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, */ for (i = 0; i < count; i++) { uint8_t buffer[32] ALIGN16_ATTRIB; - const unsigned long addr = src + (elts[i] * pitch); + const uint64_t addr = src + (elts[i] * pitch); const unsigned size = ((addr & 0x0f) == 0) ? 16 : 32; mfc_get(buffer, addr & ~0x0f, size, TAG_VERTEX_BUFFER, 0, 0); diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.h b/src/mesa/pipe/cell/spu/spu_vertex_shader.h index c52f38fd026..b261ab44a21 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.h +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.h @@ -16,7 +16,7 @@ struct spu_vs_context { struct pipe_viewport_state viewport; struct { - const ubyte *src_ptr[PIPE_ATTRIB_MAX]; + uint64_t src_ptr[PIPE_ATTRIB_MAX]; unsigned pitch[PIPE_ATTRIB_MAX]; enum pipe_format format[PIPE_ATTRIB_MAX]; unsigned nr_attrs; -- cgit v1.2.3 From 7a1b2f4078789aedf16158c41682c9d28a531d20 Mon Sep 17 00:00:00 2001 From: Michal Krol Date: Tue, 5 Feb 2008 07:50:56 -0700 Subject: gallium: Use align_free to free aligned memory. --- src/mesa/pipe/draw/draw_context.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_context.c b/src/mesa/pipe/draw/draw_context.c index 87f4969983a..b15f57c8248 100644 --- a/src/mesa/pipe/draw/draw_context.c +++ b/src/mesa/pipe/draw/draw_context.c @@ -106,7 +106,7 @@ void draw_destroy( struct draw_context *draw ) if (draw->pipeline.rasterize) draw->pipeline.rasterize->destroy( draw->pipeline.rasterize ); tgsi_exec_machine_free_data(&draw->machine); - FREE( draw->vcache.vertex[0] ); /* Frees all the vertices. */ + align_free( draw->vcache.vertex[0] ); /* Frees all the vertices. */ FREE( draw ); } -- cgit v1.2.3 From e9147bfab40d26fb8f8c0794e9a3fdcf14ca57dd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 6 Feb 2008 00:26:49 +1100 Subject: nv40: cleanup state handling a bit --- src/mesa/pipe/nouveau/nouveau_stateobj.h | 139 +++++++++ src/mesa/pipe/nv40/nv40_context.h | 18 +- src/mesa/pipe/nv40/nv40_fragprog.c | 18 +- src/mesa/pipe/nv40/nv40_fragtex.c | 38 +-- src/mesa/pipe/nv40/nv40_state.c | 464 ++++++++++++++++--------------- src/mesa/pipe/nv40/nv40_state.h | 68 +---- src/mesa/pipe/nv40/nv40_state_emit.c | 70 +---- src/mesa/pipe/nv40/nv40_vbo.c | 81 +++--- 8 files changed, 462 insertions(+), 434 deletions(-) create mode 100644 src/mesa/pipe/nouveau/nouveau_stateobj.h diff --git a/src/mesa/pipe/nouveau/nouveau_stateobj.h b/src/mesa/pipe/nouveau/nouveau_stateobj.h new file mode 100644 index 00000000000..8dfc0e9e9ac --- /dev/null +++ b/src/mesa/pipe/nouveau/nouveau_stateobj.h @@ -0,0 +1,139 @@ +#ifndef __NOUVEAU_STATEOBJ_H__ +#define __NOUVEAU_STATEOBJ_H__ + +struct nouveau_stateobj_reloc { + struct pipe_buffer *bo; + + unsigned offset; + unsigned packet; + + unsigned data; + unsigned flags; + unsigned vor; + unsigned tor; +}; + +struct nouveau_stateobj { + int refcount; + + unsigned *push; + struct nouveau_stateobj_reloc *reloc; + + unsigned *cur; + unsigned cur_packet; + unsigned cur_reloc; +}; + +static inline struct nouveau_stateobj * +so_new(unsigned push, unsigned reloc) +{ + struct nouveau_stateobj *so; + + so = malloc(sizeof(struct nouveau_stateobj)); + so->refcount = 0; + so->push = malloc(sizeof(unsigned) * push); + so->reloc = malloc(sizeof(struct nouveau_stateobj_reloc) * reloc); + + so->cur = so->push; + so->cur_reloc = so->cur_packet = 0; + + return so; +} + +static inline void +so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) +{ + struct nouveau_stateobj *so; + + so = *pso; + if (so) { + if (--so->refcount <= 0) { + free(so->push); + free(so->reloc); + free(so); + } + *pso = NULL; + } + + if (ref) { + ref->refcount++; + *pso = ref; + } +} + +static inline void +so_data(struct nouveau_stateobj *so, unsigned data) +{ + (*so->cur++) = (data); + so->cur_packet += 4; +} + +static inline void +so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, + unsigned mthd, unsigned size) +{ + so->cur_packet = (gr->subc << 13) | (1 << 18) | (mthd - 4); + so_data(so, (gr->subc << 13) | (size << 18) | mthd); +} + +static inline void +so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + struct nouveau_stateobj_reloc *r = &so->reloc[so->cur_reloc++]; + + r->bo = bo; + r->offset = so->cur - so->push; + r->packet = so->cur_packet; + r->data = data; + r->flags = flags; + r->vor = vor; + r->tor = tor; + so_data(so, data); +} + +static inline void +so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) +{ + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + unsigned nr, i; + + nr = so->cur - so->push; + if (pb->remaining < nr) + nvws->push_flush(nvws->channel, nr); + pb->remaining -= nr; + + memcpy(pb->cur, so->push, nr * 4); + for (i = 0; i < so->cur_reloc; i++) { + struct nouveau_stateobj_reloc *r = &so->reloc[i]; + + nvws->push_reloc(nvws->channel, pb->cur + r->offset, r->bo, + r->data, r->flags, r->vor, r->tor); + } + pb->cur += nr; +} + +static inline void +so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) +{ + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + unsigned i; + + i = so->cur_reloc << 1; + if (nvws->channel->pushbuf->remaining < i) + nvws->push_flush(nvws->channel, i); + nvws->channel->pushbuf->remaining -= i; + + for (i = 0; i < so->cur_reloc; i++) { + struct nouveau_stateobj_reloc *r = &so->reloc[i]; + + nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->packet, + (r->flags & + (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART)) | + NOUVEAU_BO_DUMMY, 0, 0); + nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->data, + r->flags | NOUVEAU_BO_DUMMY, r->vor, r->tor); + } +} + +#endif diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 2d069619d7c..1a31f00ad6a 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -13,6 +13,7 @@ #define NOUVEAU_PUSH_CONTEXT(ctx) \ struct nv40_context *ctx = nv40 #include "pipe/nouveau/nouveau_push.h" +#include "pipe/nouveau/nouveau_stateobj.h" #include "nv40_state.h" @@ -47,20 +48,9 @@ struct nv40_context { unsigned fp_samplers; unsigned vp_samplers; - uint32_t rt_enable; - struct pipe_buffer *rt[4]; - struct pipe_buffer *zeta; - - struct { - struct pipe_buffer *buffer; - uint32_t format; - } tex[16]; - - unsigned vb_enable; - struct { - struct pipe_buffer *buffer; - unsigned delta; - } vb[16]; + struct nouveau_stateobj *so_framebuffer; + struct nouveau_stateobj *so_fragtex[16]; + struct nouveau_stateobj *so_vtxbuf; struct { struct nouveau_resource *exec_heap; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 667eb89cb24..e650c975413 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -759,6 +759,7 @@ void nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) { struct pipe_winsys *ws = nv40->pipe.winsys; + struct nouveau_stateobj *so; int i; if (!fp->translated) { @@ -815,13 +816,16 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) fp->on_hw = TRUE; } - BEGIN_RING(curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (fp->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, - NV40TCL_FP_ADDRESS_DMA1); - BEGIN_RING(curie, NV40TCL_FP_CONTROL, 1); - OUT_RING (fp->fp_control); + so = so_new(4, 1); + so_method(so, nv40->curie, NV40TCL_FP_ADDRESS, 1); + so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); + so_method(so, nv40->curie, NV40TCL_FP_CONTROL, 1); + so_data (so, fp->fp_control); + + so_emit(nv40->nvws, so); + so_ref(so, &fp->so); nv40->fragprog.active = fp; } diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 7c5ecd5c566..283d49704af 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -59,8 +59,10 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; struct pipe_texture *pt = &nv40mt->base; struct nv40_texture_format *tf; + struct nouveau_stateobj *so; uint32_t txf, txs, txp; int swizzled = 0; /*XXX: implement in region code? */ + unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; tf = nv40_fragtex_format(pt->format); if (!tf) @@ -101,25 +103,24 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txs = tf->swizzle; - nv40->tex[unit].buffer = nv40mt->buffer; - nv40->tex[unit].format = txf; - - BEGIN_RING(curie, NV40TCL_TEX_OFFSET(unit), 8); - OUT_RELOCl(nv40->tex[unit].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv40->tex[unit].buffer, nv40->tex[unit].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, - NV40TCL_TEX_FORMAT_DMA1); - OUT_RING (ps->wrap); - OUT_RING (NV40TCL_TEX_ENABLE_ENABLE | ps->en | + so = so_new(16, 2); + so_method(so, nv40->curie, NV40TCL_TEX_OFFSET(unit), 8); + so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, + NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); + so_data (so, ps->wrap); + so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en | (0x00078000) /* mipmap related? */); - OUT_RING (txs); - OUT_RING (ps->filt | 0x3fd6 /*voodoo*/); - OUT_RING ((pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); - OUT_RING (ps->bcol); - BEGIN_RING(curie, NV40TCL_TEX_SIZE1(unit), 1); - OUT_RING ((pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); + so_data (so, txs); + so_data (so, ps->filt | 0x3fd6 /*voodoo*/); + so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | + pt->height[0]); + so_data (so, ps->bcol); + so_method(so, nv40->curie, NV40TCL_TEX_SIZE1(unit), 1); + so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); + + so_emit(nv40->nvws, so); + so_ref (so, &nv40->so_fragtex[unit]); } void @@ -133,6 +134,7 @@ nv40_fragtex_bind(struct nv40_context *nv40) unit = ffs(samplers) - 1; samplers &= ~(1 << unit); + so_ref(NULL, &nv40->so_fragtex[unit]); BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); OUT_RING (0); } diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index bb435b106b4..125134afdcd 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -9,59 +9,59 @@ static void * nv40_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { - struct nv40_blend_state *cb; - - cb = malloc(sizeof(struct nv40_blend_state)); - - cb->b_enable = cso->blend_enable ? 1 : 0; - cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | - (nvgl_blend_func(cso->rgb_src_factor))); - cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | - (nvgl_blend_func(cso->rgb_dst_factor))); - cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | - (nvgl_blend_eqn(cso->rgb_func))); + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(16, 0); + + if (cso->blend_enable) { + so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 3); + so_data (so, 1); + so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | + nvgl_blend_func(cso->rgb_src_factor)); + so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | + nvgl_blend_func(cso->rgb_dst_factor)); + so_method(so, nv40->curie, NV40TCL_BLEND_EQUATION, 1); + so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | + nvgl_blend_eqn(cso->rgb_func)); + } else { + so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 1); + so_data (so, 0); + } - cb->l_enable = cso->logicop_enable ? 1 : 0; - cb->l_op = nvgl_logicop_func(cso->logicop_func); + so_method(so, nv40->curie, NV40TCL_COLOR_MASK, 1); + so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); - cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | - ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | - ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | - ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + if (cso->logicop_enable) { + so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + so_data (so, 1); + so_data (so, nvgl_logicop_func(cso->logicop_func)); + } else { + so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); + so_data (so, 0); + } - cb->d_enable = cso->dither ? 1 : 0; + so_method(so, nv40->curie, NV40TCL_DITHER_ENABLE, 1); + so_data (so, cso->dither ? 1 : 0); - return (void *)cb; + return (void *)so; } static void nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_blend_state *cb = hwcso; - BEGIN_RING(curie, NV40TCL_DITHER_ENABLE, 1); - OUT_RING (cb->d_enable); - - BEGIN_RING(curie, NV40TCL_BLEND_ENABLE, 3); - OUT_RING (cb->b_enable); - OUT_RING (cb->b_srcfunc); - OUT_RING (cb->b_dstfunc); - BEGIN_RING(curie, NV40TCL_BLEND_EQUATION, 1); - OUT_RING (cb->b_eqn); - - BEGIN_RING(curie, NV40TCL_COLOR_MASK, 1); - OUT_RING (cb->c_mask); - - BEGIN_RING(curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (cb->l_enable); - OUT_RING (cb->l_op); + so_emit(nv40->nvws, hwcso); } static void nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); } @@ -261,8 +261,8 @@ static void * nv40_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { - struct nv40_rasterizer_state *rs; - int i; + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(32, 0); /*XXX: ignored: * light_twoside @@ -272,165 +272,163 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, * multisample * offset_units / offset_scale */ - rs = malloc(sizeof(struct nv40_rasterizer_state)); - - rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; - rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; - rs->line_smooth_en = cso->line_smooth ? 1 : 0; - rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; - rs->line_stipple = (cso->line_stipple_pattern << 16) | - cso->line_stipple_factor; + so_method(so, nv40->curie, NV40TCL_SHADE_MODEL, 1); + so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : + NV40TCL_SHADE_MODEL_SMOOTH); - rs->point_size = *(uint32_t*)&cso->point_size; + so_method(so, nv40->curie, NV40TCL_LINE_WIDTH, 2); + so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); + so_data (so, cso->line_smooth ? 1 : 0); + so_method(so, nv40->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + so_data (so, cso->line_stipple_enable ? 1 : 0); + so_data (so, (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor); - rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; - rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; + so_method(so, nv40->curie, NV40TCL_POINT_SIZE, 1); + so_data (so, fui(cso->point_size)); + so_method(so, nv40->curie, NV40TCL_POLYGON_MODE_FRONT, 6); if (cso->front_winding == PIPE_WINDING_CCW) { - rs->front_face = NV40TCL_FRONT_FACE_CCW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV40TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_CW: + so_data(so, NV40TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, 0); + break; + } + so_data(so, NV40TCL_FRONT_FACE_CCW); } else { - rs->front_face = NV40TCL_FRONT_FACE_CW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV40TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_CW: + so_data(so, NV40TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, 0); + break; + } + so_data(so, NV40TCL_FRONT_FACE_CW); } + so_data(so, cso->poly_smooth ? 1 : 0); + so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CCW) - rs->cull_face = NV40TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV40TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_CW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CW) - rs->cull_face = NV40TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV40TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_BOTH: - rs->cull_face_en = 1; - rs->cull_face = NV40TCL_CULL_FACE_FRONT_AND_BACK; - break; - case PIPE_WINDING_NONE: - default: - rs->cull_face_en = 0; - rs->cull_face = 0; - break; - } + so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, cso->poly_stipple_enable ? 1 : 0); + so_method(so, nv40->curie, NV40TCL_POINT_SPRITE, 1); if (cso->point_sprite) { - rs->point_sprite = (1 << 0); + unsigned psctl = (1 << 0), i; + for (i = 0; i < 8; i++) { if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) - rs->point_sprite |= (1 << (8 + i)); + psctl |= (1 << (8 + i)); } + + so_data(so, psctl); } else { - rs->point_sprite = 0; + so_data(so, 0); } - return (void *)rs; + return (void *)so; } static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_rasterizer_state *rs = hwcso; - - BEGIN_RING(curie, NV40TCL_SHADE_MODEL, 1); - OUT_RING (rs->shade_model); - - BEGIN_RING(curie, NV40TCL_LINE_WIDTH, 2); - OUT_RING (rs->line_width); - OUT_RING (rs->line_smooth_en); - BEGIN_RING(curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); - OUT_RING (rs->line_stipple_en); - OUT_RING (rs->line_stipple); - - BEGIN_RING(curie, NV40TCL_POINT_SIZE, 1); - OUT_RING (rs->point_size); - - BEGIN_RING(curie, NV40TCL_POLYGON_MODE_FRONT, 6); - OUT_RING (rs->poly_mode_front); - OUT_RING (rs->poly_mode_back); - OUT_RING (rs->cull_face); - OUT_RING (rs->front_face); - OUT_RING (rs->poly_smooth_en); - OUT_RING (rs->cull_face_en); - - BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); - OUT_RING (rs->poly_stipple_en); - - BEGIN_RING(curie, NV40TCL_POINT_SPRITE, 1); - OUT_RING (rs->point_sprite); + + so_emit(nv40->nvws, hwcso); } static void nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); -} + struct nouveau_stateobj *so = hwcso; -static void -nv40_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, - unsigned idx, struct nv40_stencil_push *hw) -{ - hw->enable = cso->stencil[idx].enabled ? 1 : 0; - hw->wmask = cso->stencil[idx].write_mask; - hw->func = nvgl_comparison_op(cso->stencil[idx].func); - hw->ref = cso->stencil[idx].ref_value; - hw->vmask = cso->stencil[idx].value_mask; - hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); - hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); - hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); + so_ref(NULL, &so); } static void * nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { - struct nv40_depth_stencil_alpha_state *hw; - - hw = malloc(sizeof(struct nv40_depth_stencil_alpha_state)); - - hw->depth.func = nvgl_comparison_op(cso->depth.func); - hw->depth.write_enable = cso->depth.writemask ? 1 : 0; - hw->depth.test_enable = cso->depth.enabled ? 1 : 0; - - nv40_translate_stencil(cso, 0, &hw->stencil.front); - nv40_translate_stencil(cso, 1, &hw->stencil.back); + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(32, 0); + + so_method(so, nv40->curie, NV40TCL_DEPTH_FUNC, 3); + so_data (so, nvgl_comparison_op(cso->depth.func)); + so_data (so, cso->depth.writemask ? 1 : 0); + so_data (so, cso->depth.enabled ? 1 : 0); + + so_method(so, nv40->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_data (so, cso->alpha.enabled ? 1 : 0); + so_data (so, nvgl_comparison_op(cso->alpha.func)); + so_data (so, float_to_ubyte(cso->alpha.ref)); + + if (cso->stencil[0].enabled) { + so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_data (so, cso->stencil[0].enabled ? 1 : 0); + so_data (so, cso->stencil[0].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[0].func)); + so_data (so, cso->stencil[0].ref_value); + so_data (so, cso->stencil[0].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); + } else { + so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_data (so, 0); + } - hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; - hw->alpha.func = nvgl_comparison_op(cso->alpha.func); - hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + if (cso->stencil[1].enabled) { + so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_data (so, cso->stencil[1].enabled ? 1 : 0); + so_data (so, cso->stencil[1].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[1].func)); + so_data (so, cso->stencil[1].ref_value); + so_data (so, cso->stencil[1].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); + } else { + so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_data (so, 0); + } - return (void *)hw; + return (void *)so; } static void nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_depth_stencil_alpha_state *hw = hwcso; - - BEGIN_RING(curie, NV40TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&hw->depth, 3); - BEGIN_RING(curie, NV40TCL_STENCIL_FRONT_ENABLE, 16); - OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); - OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); - BEGIN_RING(curie, NV40TCL_ALPHA_TEST_ENABLE, 3); - OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); + + so_emit(nv40->nvws, hwcso); } static void nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); } static void * @@ -502,12 +500,16 @@ nv40_set_blend_color(struct pipe_context *pipe, const struct pipe_blend_color *bcol) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(2, 0); + + so_method(so, nv40->curie, NV40TCL_BLEND_COLOR, 1); + so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0))); - BEGIN_RING(curie, NV40TCL_BLEND_COLOR, 1); - OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0)); + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void @@ -540,6 +542,8 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, struct pipe_surface *rt[4], *zeta; uint32_t rt_enable, rt_format, w, h; int i, colour_format = 0, zeta_format = 0; + struct nouveau_stateobj *so = so_new(64, 10); + unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; rt_enable = 0; for (i = 0; i < 4; i++) { @@ -603,66 +607,78 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - nv40->rt[0] = rt[0]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv40->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR0_PITCH, 2); - OUT_RING (rt[0]->pitch * rt[0]->cpp); - OUT_RELOCl(nv40->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR0, 1); + so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR0_PITCH, 2); + so_data (so, rt[0]->pitch * rt[0]->cpp); + so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - nv40->rt[1] = rt[1]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv40->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR1_OFFSET, 2); - OUT_RELOCl(nv40->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RING (rt[1]->pitch * rt[1]->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR1, 1); + so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR1_OFFSET, 2); + so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - nv40->rt[2] = rt[2]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(nv40->rt[2], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(nv40->rt[2], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR2_PITCH, 1); - OUT_RING (rt[2]->pitch * rt[2]->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR2, 1); + so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR2_OFFSET, 1); + so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->curie, NV40TCL_COLOR2_PITCH, 1); + so_data (so, rt[2]->pitch * rt[2]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - nv40->rt[3] = rt[3]->buffer; - BEGIN_RING(curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(nv40->rt[3], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(nv40->rt[3], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_COLOR3_PITCH, 1); - OUT_RING (rt[3]->pitch * rt[3]->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_COLOR3, 1); + so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_COLOR3_OFFSET, 1); + so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->curie, NV40TCL_COLOR3_PITCH, 1); + so_data (so, rt[3]->pitch * rt[3]->cpp); } if (zeta_format) { - nv40->zeta = zeta->buffer; - BEGIN_RING(curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(nv40->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv40->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(curie, NV40TCL_ZETA_PITCH, 1); - OUT_RING (zeta->pitch * zeta->cpp); + so_method(so, nv40->curie, NV40TCL_DMA_ZETA, 1); + so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->curie, NV40TCL_ZETA_OFFSET, 1); + so_reloc (so, zeta->buffer, zeta->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->curie, NV40TCL_ZETA_PITCH, 1); + so_data (so, zeta->pitch * zeta->cpp); } - nv40->rt_enable = rt_enable; - BEGIN_RING(curie, NV40TCL_RT_ENABLE, 1); - OUT_RING (rt_enable); - BEGIN_RING(curie, NV40TCL_RT_HORIZ, 3); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - OUT_RING (rt_format); - BEGIN_RING(curie, NV40TCL_VIEWPORT_HORIZ, 2); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - BEGIN_RING(curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0); - OUT_RING (((h - 1) << 16) | 0); + so_method(so, nv40->curie, NV40TCL_RT_ENABLE, 1); + so_data (so, rt_enable); + so_method(so, nv40->curie, NV40TCL_RT_HORIZ, 3); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_data (so, rt_format); + so_method(so, nv40->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_method(so, nv40->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_data (so, ((w - 1) << 16) | 0); + so_data (so, ((h - 1) << 16) | 0); + + so_emit(nv40->nvws, so); + so_ref (so, &nv40->so_framebuffer); } static void @@ -670,9 +686,15 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(33, 0); + unsigned i; - BEGIN_RING(curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); - OUT_RINGp ((uint32_t *)stipple->stipple, 32); + so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, stipple->stipple[i]); + + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void @@ -680,10 +702,14 @@ nv40_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(3, 0); + + so_method(so, nv40->curie, NV40TCL_SCISSOR_HORIZ, 2); + so_data (so, ((s->maxx - s->minx) << 16) | s->minx); + so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - BEGIN_RING(curie, NV40TCL_SCISSOR_HORIZ, 2); - OUT_RING (((s->maxx - s->minx) << 16) | s->minx); - OUT_RING (((s->maxy - s->miny) << 16) | s->miny); + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void @@ -691,16 +717,20 @@ nv40_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) { struct nv40_context *nv40 = nv40_context(pipe); - - BEGIN_RING(curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); - OUT_RINGf (vpt->translate[0]); - OUT_RINGf (vpt->translate[1]); - OUT_RINGf (vpt->translate[2]); - OUT_RINGf (vpt->translate[3]); - OUT_RINGf (vpt->scale[0]); - OUT_RINGf (vpt->scale[1]); - OUT_RINGf (vpt->scale[2]); - OUT_RINGf (vpt->scale[3]); + struct nouveau_stateobj *so = so_new(9, 0); + + so_method(so, nv40->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(vpt->translate[0])); + so_data (so, fui(vpt->translate[1])); + so_data (so, fui(vpt->translate[2])); + so_data (so, fui(vpt->translate[3])); + so_data (so, fui(vpt->scale[0])); + so_data (so, fui(vpt->scale[1])); + so_data (so, fui(vpt->scale[2])); + so_data (so, fui(vpt->scale[3])); + + so_emit(nv40->nvws, so); + so_ref(NULL, &so); } static void diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h index 24eea36db3a..e82ab9de98a 100644 --- a/src/mesa/pipe/nv40/nv40_state.h +++ b/src/mesa/pipe/nv40/nv40_state.h @@ -3,20 +3,6 @@ #include "pipe/p_state.h" -struct nv40_blend_state { - uint32_t b_enable; - uint32_t b_srcfunc; - uint32_t b_dstfunc; - uint32_t b_eqn; - - uint32_t l_enable; - uint32_t l_op; - - uint32_t c_mask; - - uint32_t d_enable; -}; - struct nv40_sampler_state { uint32_t fmt; uint32_t wrap; @@ -25,29 +11,6 @@ struct nv40_sampler_state { uint32_t bcol; }; -struct nv40_rasterizer_state { - uint32_t shade_model; - - uint32_t line_width; - uint32_t line_smooth_en; - uint32_t line_stipple_en; - uint32_t line_stipple; - - uint32_t point_size; - - uint32_t poly_smooth_en; - uint32_t poly_stipple_en; - - uint32_t poly_mode_front; - uint32_t poly_mode_back; - - uint32_t front_face; - uint32_t cull_face; - uint32_t cull_face_en; - - uint32_t point_sprite; -}; - struct nv40_vertex_program_exec { uint32_t data[4]; boolean has_branch_offset; @@ -99,36 +62,7 @@ struct nv40_fragment_program { struct pipe_buffer *buffer; uint32_t fp_control; -}; - -struct nv40_stencil_push { - uint32_t enable; - uint32_t wmask; - uint32_t func; - uint32_t ref; - uint32_t vmask; - uint32_t fail; - uint32_t zfail; - uint32_t zpass; -}; - -struct nv40_depth_stencil_alpha_state { - struct { - uint32_t func; - uint32_t write_enable; - uint32_t test_enable; - } depth; - - struct { - struct nv40_stencil_push back; - struct nv40_stencil_push front; - } stencil; - - struct { - uint32_t enabled; - uint32_t func; - uint32_t ref; - } alpha; + struct nouveau_stateobj *so; }; struct nv40_miptree { diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 66b98d5fab3..3a22cd4bd56 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -11,77 +11,15 @@ static void nv40_state_emit_dummy_relocs(struct nv40_context *nv40) { - unsigned rt_flags, tx_flags, fp_flags; - int i; + unsigned i; - rt_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR | NOUVEAU_BO_DUMMY; - tx_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_DUMMY; - fp_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_DUMMY; - - /* Render targets */ - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - OUT_RELOCm(nv40->rt[0], rt_flags, - curie, NV40TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv40->rt[0], rt_flags); - OUT_RELOCm(nv40->rt[0], rt_flags, - curie, NV40TCL_COLOR0_OFFSET, 1); - OUT_RELOCl(nv40->rt[0], 0, rt_flags); - } - - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - OUT_RELOCm(nv40->rt[1], rt_flags, - curie, NV40TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv40->rt[1], rt_flags); - OUT_RELOCm(nv40->rt[1], rt_flags, - curie, NV40TCL_COLOR1_OFFSET, 1); - OUT_RELOCl(nv40->rt[1], 0, rt_flags); - } - - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - OUT_RELOCm(nv40->rt[2], rt_flags, - curie, NV40TCL_DMA_COLOR2, 1); - OUT_RELOCo(nv40->rt[2], rt_flags); - OUT_RELOCm(nv40->rt[2], rt_flags, - curie, NV40TCL_COLOR2_OFFSET, 1); - OUT_RELOCl(nv40->rt[2], 0, rt_flags); - } - - if (nv40->rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - OUT_RELOCm(nv40->rt[3], rt_flags, - curie, NV40TCL_DMA_COLOR3, 1); - OUT_RELOCo(nv40->rt[3], rt_flags); - OUT_RELOCm(nv40->rt[3], rt_flags, - curie, NV40TCL_COLOR3_OFFSET, 1); - OUT_RELOCl(nv40->rt[3], 0, rt_flags); - } - - if (nv40->zeta) { - OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_DMA_ZETA, 1); - OUT_RELOCo(nv40->zeta, rt_flags); - OUT_RELOCm(nv40->zeta, rt_flags, curie, NV40TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv40->zeta, 0, rt_flags); - } - - /* Texture images */ + so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); for (i = 0; i < 16; i++) { if (!(nv40->fp_samplers & (1 << i))) continue; - OUT_RELOCm(nv40->tex[i].buffer, tx_flags, - curie, NV40TCL_TEX_OFFSET(i), 2); - OUT_RELOCl(nv40->tex[i].buffer, 0, tx_flags); - OUT_RELOCd(nv40->tex[i].buffer, nv40->tex[i].format, - tx_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, - NV40TCL_TEX_FORMAT_DMA1); + so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); } - - /* Fragment program */ - OUT_RELOCm(nv40->fragprog.active->buffer, fp_flags, - curie, NV40TCL_FP_ADDRESS, 1); - OUT_RELOC (nv40->fragprog.active->buffer, 0, - fp_flags | NOUVEAU_BO_OR | NOUVEAU_BO_LOW, - NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); + so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); } void diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 0bb54c7610e..e2cb3fda8f2 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -97,13 +97,13 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, } static void -nv40_vbo_arrays_update(struct nv40_context *nv40) +nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, + unsigned ib_format) { struct nv40_vertex_program *vp = nv40->vertprog.active; - uint32_t inputs, vtxfmt[16]; - int hw, num_hw; - - nv40->vb_enable = 0; + struct nouveau_stateobj *vtxbuf, *vtxfmt; + unsigned inputs, hw, num_hw; + unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; inputs = vp->ir; for (hw = 0; hw < 16 && inputs; hw++) { @@ -114,73 +114,64 @@ nv40_vbo_arrays_update(struct nv40_context *nv40) } num_hw++; + vtxbuf = so_new(20, 18); + so_method(vtxbuf, nv40->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + vtxfmt = so_new(17, 0); + so_method(vtxfmt, nv40->curie, NV40TCL_VTXFMT(0), num_hw); + inputs = vp->ir; for (hw = 0; hw < num_hw; hw++) { struct pipe_vertex_element *ve; struct pipe_vertex_buffer *vb; if (!(inputs & (1 << hw))) { - vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); continue; } ve = &nv40->vtxelt[hw]; vb = &nv40->vtxbuf[ve->vertex_buffer_index]; - if (vb->pitch == 0) { - vtxfmt[hw] = NV40TCL_VTXFMT_TYPE_FLOAT; - if (nv40_vbo_static_attrib(nv40, hw, ve, vb) == TRUE) - continue; + if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; } - nv40->vb_enable |= (1 << hw); - nv40->vb[hw].delta = vb->buffer_offset + ve->src_offset; - nv40->vb[hw].buffer = vb->buffer; + so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, + vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + 0, NV40TCL_VTXBUF_ADDRESS_DMA1); + so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | + (nv40_vbo_ncomp(ve->src_format) << + NV40TCL_VTXFMT_SIZE_SHIFT) | + nv40_vbo_type(ve->src_format))); + } - vtxfmt[hw] = ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_ncomp(ve->src_format) << - NV40TCL_VTXFMT_SIZE_SHIFT) | - nv40_vbo_type(ve->src_format)); + if (ib) { + so_method(vtxbuf, nv40->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); } - BEGIN_RING(curie, NV40TCL_VTXFMT(0), num_hw); - OUT_RINGp (vtxfmt, num_hw); + so_emit(nv40->nvws, vtxfmt); + so_emit(nv40->nvws, vtxbuf); + so_ref (vtxbuf, &nv40->so_vtxbuf); + so_ref (NULL, &vtxfmt); } static boolean nv40_vbo_validate_state(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned ib_format) { - unsigned inputs; - nv40_emit_hw_state(nv40); - - if (nv40->dirty & NV40_NEW_ARRAYS) { - nv40_vbo_arrays_update(nv40); + if (nv40->dirty & NV40_NEW_ARRAYS || ib) { + nv40_vbo_arrays_update(nv40, ib, ib_format); nv40->dirty &= ~NV40_NEW_ARRAYS; } - inputs = nv40->vb_enable; - while (inputs) { - unsigned a = ffs(inputs) - 1; - - inputs &= ~(1 << a); - - BEGIN_RING(curie, NV40TCL_VTXBUF_ADDRESS(a), 1); - OUT_RELOC (nv40->vb[a].buffer, nv40->vb[a].delta, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV40TCL_VTXBUF_ADDRESS_DMA1); - } - - if (ib) { - BEGIN_RING(curie, NV40TCL_IDXBUF_ADDRESS, 2); - OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD); - OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_OR, - 0, NV40TCL_IDXBUF_FORMAT_DMA1); - } + so_emit_reloc_markers(nv40->nvws, nv40->so_vtxbuf); BEGIN_RING(curie, 0x1710, 1); OUT_RING (0); /* vtx cache flush */ -- cgit v1.2.3 From d8642b830911825c30531a2e422fcd8d2d487f74 Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Fri, 1 Feb 2008 14:58:38 -0800 Subject: Vectorize all micro ops Fold single instruction micro ops inline. Remove unused micro ops. --- src/mesa/pipe/cell/spu/spu_exec.c | 912 ++++++++++---------------------------- src/mesa/pipe/cell/spu/spu_exec.h | 1 + 2 files changed, 230 insertions(+), 683 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index 168bada3bb0..1ac9c031e32 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -52,8 +52,15 @@ #include #include -#include +#include +#include +#include +#include +#include #include +#include +#include +#include #include "pipe/p_compiler.h" #include "pipe/p_state.h" @@ -157,643 +164,175 @@ spu_exec_machine_init(struct spu_exec_machine *mach, } -static void -micro_abs( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - dst->f[0] = (float) fabs( (double) src->f[0] ); - dst->f[1] = (float) fabs( (double) src->f[1] ); - dst->f[2] = (float) fabs( (double) src->f[2] ); - dst->f[3] = (float) fabs( (double) src->f[3] ); -} - -static void -micro_add( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->f[0] = src0->f[0] + src1->f[0]; - dst->f[1] = src0->f[1] + src1->f[1]; - dst->f[2] = src0->f[2] + src1->f[2]; - dst->f[3] = src0->f[3] + src1->f[3]; -} - -static void -micro_iadd( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] + src1->i[0]; - dst->i[1] = src0->i[1] + src1->i[1]; - dst->i[2] = src0->i[2] + src1->i[2]; - dst->i[3] = src0->i[3] + src1->i[3]; -} - -static void -micro_and( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] & src1->u[0]; - dst->u[1] = src0->u[1] & src1->u[1]; - dst->u[2] = src0->u[2] & src1->u[2]; - dst->u[3] = src0->u[3] & src1->u[3]; -} - -static void -micro_ceil( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - ASSERT(0); -#if 0 - dst->f[0] = (float) ceil( (double) src->f[0] ); - dst->f[1] = (float) ceil( (double) src->f[1] ); - dst->f[2] = (float) ceil( (double) src->f[2] ); - dst->f[3] = (float) ceil( (double) src->f[3] ); -#endif -} - -static void -micro_cos( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - ASSERT(0); -#if 0 - dst->f[0] = (float) cos( (double) src->f[0] ); - dst->f[1] = (float) cos( (double) src->f[1] ); - dst->f[2] = (float) cos( (double) src->f[2] ); - dst->f[3] = (float) cos( (double) src->f[3] ); -#endif -} - -static void -micro_ddx( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - dst->f[0] = - dst->f[1] = - dst->f[2] = - dst->f[3] = src->f[TILE_BOTTOM_RIGHT] - src->f[TILE_BOTTOM_LEFT]; -} - -static void -micro_ddy( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - dst->f[0] = - dst->f[1] = - dst->f[2] = - dst->f[3] = src->f[TILE_TOP_LEFT] - src->f[TILE_BOTTOM_LEFT]; -} - -static void -micro_div( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->f[0] = src0->f[0] / src1->f[0]; - dst->f[1] = src0->f[1] / src1->f[1]; - dst->f[2] = src0->f[2] / src1->f[2]; - dst->f[3] = src0->f[3] / src1->f[3]; -} - -static void -micro_udiv( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] / src1->u[0]; - dst->u[1] = src0->u[1] / src1->u[1]; - dst->u[2] = src0->u[2] / src1->u[2]; - dst->u[3] = src0->u[3] / src1->u[3]; -} - -static void -micro_eq( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2, - const union spu_exec_channel *src3 ) -{ - dst->f[0] = src0->f[0] == src1->f[0] ? src2->f[0] : src3->f[0]; - dst->f[1] = src0->f[1] == src1->f[1] ? src2->f[1] : src3->f[1]; - dst->f[2] = src0->f[2] == src1->f[2] ? src2->f[2] : src3->f[2]; - dst->f[3] = src0->f[3] == src1->f[3] ? src2->f[3] : src3->f[3]; -} - -static void -micro_ieq( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2, - const union spu_exec_channel *src3 ) -{ - dst->i[0] = src0->i[0] == src1->i[0] ? src2->i[0] : src3->i[0]; - dst->i[1] = src0->i[1] == src1->i[1] ? src2->i[1] : src3->i[1]; - dst->i[2] = src0->i[2] == src1->i[2] ? src2->i[2] : src3->i[2]; - dst->i[3] = src0->i[3] == src1->i[3] ? src2->i[3] : src3->i[3]; -} - -static void -micro_exp2( - union spu_exec_channel *dst, - const union spu_exec_channel *src) -{ - ASSERT(0); -#if 0 - dst->f[0] = (float) pow( 2.0, (double) src->f[0] ); - dst->f[1] = (float) pow( 2.0, (double) src->f[1] ); - dst->f[2] = (float) pow( 2.0, (double) src->f[2] ); - dst->f[3] = (float) pow( 2.0, (double) src->f[3] ); -#endif -} - -static void -micro_f2it( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - dst->i[0] = (int) src->f[0]; - dst->i[1] = (int) src->f[1]; - dst->i[2] = (int) src->f[2]; - dst->i[3] = (int) src->f[3]; -} - -static void -micro_f2ut( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - dst->u[0] = (uint) src->f[0]; - dst->u[1] = (uint) src->f[1]; - dst->u[2] = (uint) src->f[2]; - dst->u[3] = (uint) src->f[3]; -} - -static void -micro_flr( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - ASSERT(0); -#if 0 - dst->f[0] = (float) floor( (double) src->f[0] ); - dst->f[1] = (float) floor( (double) src->f[1] ); - dst->f[2] = (float) floor( (double) src->f[2] ); - dst->f[3] = (float) floor( (double) src->f[3] ); -#endif -} - -static void -micro_frc( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - ASSERT(0); -#if 0 - dst->f[0] = src->f[0] - (float) floor( (double) src->f[0] ); - dst->f[1] = src->f[1] - (float) floor( (double) src->f[1] ); - dst->f[2] = src->f[2] - (float) floor( (double) src->f[2] ); - dst->f[3] = src->f[3] - (float) floor( (double) src->f[3] ); -#endif -} - -static void -micro_ge( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2, - const union spu_exec_channel *src3 ) -{ - dst->f[0] = src0->f[0] >= src1->f[0] ? src2->f[0] : src3->f[0]; - dst->f[1] = src0->f[1] >= src1->f[1] ? src2->f[1] : src3->f[1]; - dst->f[2] = src0->f[2] >= src1->f[2] ? src2->f[2] : src3->f[2]; - dst->f[3] = src0->f[3] >= src1->f[3] ? src2->f[3] : src3->f[3]; -} - -static void -micro_i2f( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - dst->f[0] = (float) src->i[0]; - dst->f[1] = (float) src->i[1]; - dst->f[2] = (float) src->i[2]; - dst->f[3] = (float) src->i[3]; -} - -static void -micro_lg2( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - ASSERT(0); -#if 0 - dst->f[0] = (float) log( (double) src->f[0] ) * 1.442695f; - dst->f[1] = (float) log( (double) src->f[1] ) * 1.442695f; - dst->f[2] = (float) log( (double) src->f[2] ) * 1.442695f; - dst->f[3] = (float) log( (double) src->f[3] ) * 1.442695f; -#endif -} - -static void -micro_lt( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2, - const union spu_exec_channel *src3 ) +static INLINE qword +micro_abs(qword src) { - dst->f[0] = src0->f[0] < src1->f[0] ? src2->f[0] : src3->f[0]; - dst->f[1] = src0->f[1] < src1->f[1] ? src2->f[1] : src3->f[1]; - dst->f[2] = src0->f[2] < src1->f[2] ? src2->f[2] : src3->f[2]; - dst->f[3] = src0->f[3] < src1->f[3] ? src2->f[3] : src3->f[3]; + return si_rotmi(si_shli(src, 1), -1); } -static void -micro_ilt( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2, - const union spu_exec_channel *src3 ) +static INLINE qword +micro_ceil(qword src) { - dst->i[0] = src0->i[0] < src1->i[0] ? src2->i[0] : src3->i[0]; - dst->i[1] = src0->i[1] < src1->i[1] ? src2->i[1] : src3->i[1]; - dst->i[2] = src0->i[2] < src1->i[2] ? src2->i[2] : src3->i[2]; - dst->i[3] = src0->i[3] < src1->i[3] ? src2->i[3] : src3->i[3]; + return (qword) _ceilf4((vec_float4) src); } -static void -micro_ult( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2, - const union spu_exec_channel *src3 ) +static INLINE qword +micro_cos(qword src) { - dst->u[0] = src0->u[0] < src1->u[0] ? src2->u[0] : src3->u[0]; - dst->u[1] = src0->u[1] < src1->u[1] ? src2->u[1] : src3->u[1]; - dst->u[2] = src0->u[2] < src1->u[2] ? src2->u[2] : src3->u[2]; - dst->u[3] = src0->u[3] < src1->u[3] ? src2->u[3] : src3->u[3]; + return (qword) _cosf4((vec_float4) src); } -static void -micro_max( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static const qword br_shuf = { + TILE_BOTTOM_RIGHT + 0, TILE_BOTTOM_RIGHT + 1, + TILE_BOTTOM_RIGHT + 2, TILE_BOTTOM_RIGHT + 3, + TILE_BOTTOM_RIGHT + 0, TILE_BOTTOM_RIGHT + 1, + TILE_BOTTOM_RIGHT + 2, TILE_BOTTOM_RIGHT + 3, + TILE_BOTTOM_RIGHT + 0, TILE_BOTTOM_RIGHT + 1, + TILE_BOTTOM_RIGHT + 2, TILE_BOTTOM_RIGHT + 3, + TILE_BOTTOM_RIGHT + 0, TILE_BOTTOM_RIGHT + 1, + TILE_BOTTOM_RIGHT + 2, TILE_BOTTOM_RIGHT + 3, +}; + +static const qword bl_shuf = { + TILE_BOTTOM_LEFT + 0, TILE_BOTTOM_LEFT + 1, + TILE_BOTTOM_LEFT + 2, TILE_BOTTOM_LEFT + 3, + TILE_BOTTOM_LEFT + 0, TILE_BOTTOM_LEFT + 1, + TILE_BOTTOM_LEFT + 2, TILE_BOTTOM_LEFT + 3, + TILE_BOTTOM_LEFT + 0, TILE_BOTTOM_LEFT + 1, + TILE_BOTTOM_LEFT + 2, TILE_BOTTOM_LEFT + 3, + TILE_BOTTOM_LEFT + 0, TILE_BOTTOM_LEFT + 1, + TILE_BOTTOM_LEFT + 2, TILE_BOTTOM_LEFT + 3, +}; + +static const qword tl_shuf = { + TILE_TOP_LEFT + 0, TILE_TOP_LEFT + 1, + TILE_TOP_LEFT + 2, TILE_TOP_LEFT + 3, + TILE_TOP_LEFT + 0, TILE_TOP_LEFT + 1, + TILE_TOP_LEFT + 2, TILE_TOP_LEFT + 3, + TILE_TOP_LEFT + 0, TILE_TOP_LEFT + 1, + TILE_TOP_LEFT + 2, TILE_TOP_LEFT + 3, + TILE_TOP_LEFT + 0, TILE_TOP_LEFT + 1, + TILE_TOP_LEFT + 2, TILE_TOP_LEFT + 3, +}; + +static qword +micro_ddx(qword src) { - dst->f[0] = src0->f[0] > src1->f[0] ? src0->f[0] : src1->f[0]; - dst->f[1] = src0->f[1] > src1->f[1] ? src0->f[1] : src1->f[1]; - dst->f[2] = src0->f[2] > src1->f[2] ? src0->f[2] : src1->f[2]; - dst->f[3] = src0->f[3] > src1->f[3] ? src0->f[3] : src1->f[3]; -} + qword bottom_right = si_shufb(src, src, br_shuf); + qword bottom_left = si_shufb(src, src, bl_shuf); -static void -micro_imax( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] > src1->i[0] ? src0->i[0] : src1->i[0]; - dst->i[1] = src0->i[1] > src1->i[1] ? src0->i[1] : src1->i[1]; - dst->i[2] = src0->i[2] > src1->i[2] ? src0->i[2] : src1->i[2]; - dst->i[3] = src0->i[3] > src1->i[3] ? src0->i[3] : src1->i[3]; + return si_fs(bottom_right, bottom_left); } -static void -micro_umax( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_ddy(qword src) { - dst->u[0] = src0->u[0] > src1->u[0] ? src0->u[0] : src1->u[0]; - dst->u[1] = src0->u[1] > src1->u[1] ? src0->u[1] : src1->u[1]; - dst->u[2] = src0->u[2] > src1->u[2] ? src0->u[2] : src1->u[2]; - dst->u[3] = src0->u[3] > src1->u[3] ? src0->u[3] : src1->u[3]; -} + qword top_left = si_shufb(src, src, tl_shuf); + qword bottom_left = si_shufb(src, src, bl_shuf); -static void -micro_min( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->f[0] = src0->f[0] < src1->f[0] ? src0->f[0] : src1->f[0]; - dst->f[1] = src0->f[1] < src1->f[1] ? src0->f[1] : src1->f[1]; - dst->f[2] = src0->f[2] < src1->f[2] ? src0->f[2] : src1->f[2]; - dst->f[3] = src0->f[3] < src1->f[3] ? src0->f[3] : src1->f[3]; + return si_fs(top_left, bottom_left); } -static void -micro_imin( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static INLINE qword +micro_div(qword src0, qword src1) { - dst->i[0] = src0->i[0] < src1->i[0] ? src0->i[0] : src1->i[0]; - dst->i[1] = src0->i[1] < src1->i[1] ? src0->i[1] : src1->i[1]; - dst->i[2] = src0->i[2] < src1->i[2] ? src0->i[2] : src1->i[2]; - dst->i[3] = src0->i[3] < src1->i[3] ? src0->i[3] : src1->i[3]; + return (qword) _divf4((vec_float4) src0, (vec_float4) src1); } -static void -micro_umin( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_flr(qword src) { - dst->u[0] = src0->u[0] < src1->u[0] ? src0->u[0] : src1->u[0]; - dst->u[1] = src0->u[1] < src1->u[1] ? src0->u[1] : src1->u[1]; - dst->u[2] = src0->u[2] < src1->u[2] ? src0->u[2] : src1->u[2]; - dst->u[3] = src0->u[3] < src1->u[3] ? src0->u[3] : src1->u[3]; + return (qword) _floorf4((vec_float4) src); } -static void -micro_umod( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_frc(qword src) { - dst->u[0] = src0->u[0] % src1->u[0]; - dst->u[1] = src0->u[1] % src1->u[1]; - dst->u[2] = src0->u[2] % src1->u[2]; - dst->u[3] = src0->u[3] % src1->u[3]; + return si_fs(src, (qword) _floorf4((vec_float4) src)); } -static void -micro_mul( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->f[0] = src0->f[0] * src1->f[0]; - dst->f[1] = src0->f[1] * src1->f[1]; - dst->f[2] = src0->f[2] * src1->f[2]; - dst->f[3] = src0->f[3] * src1->f[3]; -} - -static void -micro_imul( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->i[0] = src0->i[0] * src1->i[0]; - dst->i[1] = src0->i[1] * src1->i[1]; - dst->i[2] = src0->i[2] * src1->i[2]; - dst->i[3] = src0->i[3] * src1->i[3]; -} - -static void -micro_imul64( - union spu_exec_channel *dst0, - union spu_exec_channel *dst1, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst1->i[0] = src0->i[0] * src1->i[0]; - dst1->i[1] = src0->i[1] * src1->i[1]; - dst1->i[2] = src0->i[2] * src1->i[2]; - dst1->i[3] = src0->i[3] * src1->i[3]; - dst0->i[0] = 0; - dst0->i[1] = 0; - dst0->i[2] = 0; - dst0->i[3] = 0; -} - -static void -micro_umul64( - union spu_exec_channel *dst0, - union spu_exec_channel *dst1, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst1->u[0] = src0->u[0] * src1->u[0]; - dst1->u[1] = src0->u[1] * src1->u[1]; - dst1->u[2] = src0->u[2] * src1->u[2]; - dst1->u[3] = src0->u[3] * src1->u[3]; - dst0->u[0] = 0; - dst0->u[1] = 0; - dst0->u[2] = 0; - dst0->u[3] = 0; -} - -static void -micro_movc( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1, - const union spu_exec_channel *src2 ) -{ - dst->u[0] = src0->u[0] ? src1->u[0] : src2->u[0]; - dst->u[1] = src0->u[1] ? src1->u[1] : src2->u[1]; - dst->u[2] = src0->u[2] ? src1->u[2] : src2->u[2]; - dst->u[3] = src0->u[3] ? src1->u[3] : src2->u[3]; -} - -static void -micro_neg( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) +static INLINE qword +micro_ge(qword src0, qword src1) { - dst->f[0] = -src->f[0]; - dst->f[1] = -src->f[1]; - dst->f[2] = -src->f[2]; - dst->f[3] = -src->f[3]; + return si_or(si_fceq(src0, src1), si_fcgt(src0, src1)); } -static void -micro_ineg( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) +static qword +micro_lg2(qword src) { - dst->i[0] = -src->i[0]; - dst->i[1] = -src->i[1]; - dst->i[2] = -src->i[2]; - dst->i[3] = -src->i[3]; + return (qword) _log2f4((vec_float4) src); } -static void -micro_not( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) +static INLINE qword +micro_lt(qword src0, qword src1) { - dst->u[0] = ~src->u[0]; - dst->u[1] = ~src->u[1]; - dst->u[2] = ~src->u[2]; - dst->u[3] = ~src->u[3]; -} + const qword tmp = si_or(si_fceq(src0, src1), si_fcgt(src0, src1)); -static void -micro_or( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) -{ - dst->u[0] = src0->u[0] | src1->u[0]; - dst->u[1] = src0->u[1] | src1->u[1]; - dst->u[2] = src0->u[2] | src1->u[2]; - dst->u[3] = src0->u[3] | src1->u[3]; + return si_xori(tmp, 0xff); } -static void -micro_pow( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static INLINE qword +micro_max(qword src0, qword src1) { - vec_float4 s0 = (vec_float4) { - src0->f[0], src0->f[1], src0->f[2], src0->f[3] - }; - vec_float4 s1 = (vec_float4) { - src1->f[0], src1->f[1], src1->f[2], src1->f[3] - }; - vec_float4 d = _powf4(s0, s1); - - dst->f[0] = spu_extract(d, 0); - dst->f[1] = spu_extract(d, 1); - dst->f[2] = spu_extract(d, 2); - dst->f[3] = spu_extract(d, 3); + return si_selb(src1, src0, si_fcgt(src0, src1)); } -static void -micro_rnd( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) +static INLINE qword +micro_min(qword src0, qword src1) { - ASSERT(0); -#if 0 - dst->f[0] = (float) floor( (double) (src->f[0] + 0.5f) ); - dst->f[1] = (float) floor( (double) (src->f[1] + 0.5f) ); - dst->f[2] = (float) floor( (double) (src->f[2] + 0.5f) ); - dst->f[3] = (float) floor( (double) (src->f[3] + 0.5f) ); -#endif + return si_selb(src0, src1, si_fcgt(src0, src1)); } -static void -micro_shl( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_neg(qword src) { - dst->i[0] = src0->i[0] << src1->i[0]; - dst->i[1] = src0->i[1] << src1->i[1]; - dst->i[2] = src0->i[2] << src1->i[2]; - dst->i[3] = src0->i[3] << src1->i[3]; + return si_xor(src, (qword) spu_splats(0x80000000)); } -static void -micro_ishr( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_set_sign(qword src) { - dst->i[0] = src0->i[0] >> src1->i[0]; - dst->i[1] = src0->i[1] >> src1->i[1]; - dst->i[2] = src0->i[2] >> src1->i[2]; - dst->i[3] = src0->i[3] >> src1->i[3]; + return si_or(src, (qword) spu_splats(0x80000000)); } -static void -micro_trunc( - union spu_exec_channel *dst, - const union spu_exec_channel *src0 ) +static qword +micro_pow(qword src0, qword src1) { - dst->f[0] = (float) (int) src0->f[0]; - dst->f[1] = (float) (int) src0->f[1]; - dst->f[2] = (float) (int) src0->f[2]; - dst->f[3] = (float) (int) src0->f[3]; + return (qword) _powf4((vec_float4) src0, (vec_float4) src1); } -static void -micro_ushr( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_rnd(qword src) { - dst->u[0] = src0->u[0] >> src1->u[0]; - dst->u[1] = src0->u[1] >> src1->u[1]; - dst->u[2] = src0->u[2] >> src1->u[2]; - dst->u[3] = src0->u[3] >> src1->u[3]; -} + const qword half = (qword) spu_splats(0.5f); -static void -micro_sin( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) -{ - ASSERT(0); -#if 0 - dst->f[0] = (float) sin( (double) src->f[0] ); - dst->f[1] = (float) sin( (double) src->f[1] ); - dst->f[2] = (float) sin( (double) src->f[2] ); - dst->f[3] = (float) sin( (double) src->f[3] ); -#endif + /* May be able to use _roundf4. There may be some difference, though. + */ + return (qword) _floorf4((vec_float4) si_fa(src, half)); } -static void -micro_sqrt( union spu_exec_channel *dst, - const union spu_exec_channel *src ) +static INLINE qword +micro_ishr(qword src0, qword src1) { - vec_float4 s = (vec_float4) { - src->f[0], src->f[1], src->f[2], src->f[3] - }; - vec_float4 d = _sqrtf4(s); - - dst->f[0] = spu_extract(d, 0); - dst->f[1] = spu_extract(d, 1); - dst->f[2] = spu_extract(d, 2); - dst->f[3] = spu_extract(d, 3); + return si_rotma(src0, si_sfi(src1, 0)); } -static void -micro_sub( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static qword +micro_trunc(qword src) { - dst->f[0] = src0->f[0] - src1->f[0]; - dst->f[1] = src0->f[1] - src1->f[1]; - dst->f[2] = src0->f[2] - src1->f[2]; - dst->f[3] = src0->f[3] - src1->f[3]; + return (qword) _truncf4((vec_float4) src); } -static void -micro_u2f( - union spu_exec_channel *dst, - const union spu_exec_channel *src ) +static qword +micro_sin(qword src) { - dst->f[0] = (float) src->u[0]; - dst->f[1] = (float) src->u[1]; - dst->f[2] = (float) src->u[2]; - dst->f[3] = (float) src->u[3]; + return (qword) _sinf4((vec_float4) src); } -static void -micro_xor( - union spu_exec_channel *dst, - const union spu_exec_channel *src0, - const union spu_exec_channel *src1 ) +static INLINE qword +micro_sqrt(qword src) { - dst->u[0] = src0->u[0] ^ src1->u[0]; - dst->u[1] = src0->u[1] ^ src1->u[1]; - dst->u[2] = src0->u[2] ^ src1->u[2]; - dst->u[3] = src0->u[3] ^ src1->u[3]; + return (qword) _sqrtf4((vec_float4) src); } static void @@ -983,16 +522,15 @@ fetch_source( switch (tgsi_util_get_full_src_register_sign_mode( reg, chan_index )) { case TGSI_UTIL_SIGN_CLEAR: - micro_abs( chan, chan ); + chan->q = micro_abs(chan->q); break; case TGSI_UTIL_SIGN_SET: - micro_abs( chan, chan ); - micro_neg( chan, chan ); + chan->q = micro_set_sign(chan->q); break; case TGSI_UTIL_SIGN_TOGGLE: - micro_neg( chan, chan ); + chan->q = micro_neg(chan->q); break; case TGSI_UTIL_SIGN_KEEP: @@ -1000,7 +538,7 @@ fetch_source( } if (reg->SrcRegisterExtMod.Complement) { - micro_sub( chan, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], chan ); + chan->q = si_fs(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, chan->q); } } @@ -1051,8 +589,8 @@ store_dest( case TGSI_SAT_ZERO_ONE: /* XXX need to obey ExecMask here */ - micro_max(dst, chan, &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]); - micro_min(dst, dst, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]); + dst->q = micro_max(chan->q, mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].q); + dst->q = micro_min(dst->q, mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q); break; case TGSI_SAT_MINUS_PLUS_ONE: @@ -1162,7 +700,7 @@ exec_tex(struct spu_exec_machine *mach, switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) { case TGSI_EXTSWIZZLE_W: FETCH(&r[1], 0, CHAN_W); - micro_div( &r[0], &r[0], &r[1] ); + r[0].q = micro_div(r[0].q, r[1].q); break; case TGSI_EXTSWIZZLE_ONE: @@ -1194,9 +732,9 @@ exec_tex(struct spu_exec_machine *mach, switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) { case TGSI_EXTSWIZZLE_W: FETCH(&r[3], 0, CHAN_W); - micro_div( &r[0], &r[0], &r[3] ); - micro_div( &r[1], &r[1], &r[3] ); - micro_div( &r[2], &r[2], &r[3] ); + r[0].q = micro_div(r[0].q, r[3].q); + r[1].q = micro_div(r[1].q, r[3].q); + r[2].q = micro_div(r[2].q, r[3].q); break; case TGSI_EXTSWIZZLE_ONE: @@ -1228,9 +766,9 @@ exec_tex(struct spu_exec_machine *mach, switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) { case TGSI_EXTSWIZZLE_W: FETCH(&r[3], 0, CHAN_W); - micro_div( &r[0], &r[0], &r[3] ); - micro_div( &r[1], &r[1], &r[3] ); - micro_div( &r[2], &r[2], &r[3] ); + r[0].q = micro_div(r[0].q, r[3].q); + r[1].q = micro_div(r[1].q, r[3].q); + r[2].q = micro_div(r[2].q, r[3].q); break; case TGSI_EXTSWIZZLE_ONE: @@ -1389,7 +927,7 @@ exec_instruction( case TGSI_OPCODE_ARL: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_f2it( &r[0], &r[0] ); + r[0].q = si_cflts(r[0].q, 0); STORE( &r[0], 0, chan_index ); } break; @@ -1409,22 +947,27 @@ exec_instruction( if (IS_CHANNEL_ENABLED( *inst, CHAN_Y ) || IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { FETCH( &r[0], 0, CHAN_X ); - if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { - micro_max( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { + r[0].q = micro_max(r[0].q, mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].q); STORE( &r[0], 0, CHAN_Y ); } - if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { - FETCH( &r[1], 0, CHAN_Y ); - micro_max( &r[1], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); - - FETCH( &r[2], 0, CHAN_W ); - micro_min( &r[2], &r[2], &mach->Temps[TEMP_128_I].xyzw[TEMP_128_C] ); - micro_max( &r[2], &r[2], &mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C] ); - micro_pow( &r[1], &r[1], &r[2] ); - micro_lt( &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); - STORE( &r[0], 0, CHAN_Z ); - } + if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { + FETCH( &r[1], 0, CHAN_Y ); + r[1].q = micro_max(r[1].q, mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].q); + + FETCH( &r[2], 0, CHAN_W ); + r[2].q = micro_min(r[2].q, mach->Temps[TEMP_128_I].xyzw[TEMP_128_C].q); + r[2].q = micro_max(r[2].q, mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C].q); + r[1].q = micro_pow(r[1].q, r[2].q); + + /* r0 = (r0 > 0.0) ? r1 : 0.0 + */ + r[0].q = si_fcgt(r[0].q, mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].q); + r[0].q = si_selb(mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].q, r[1].q, + r[0].q); + STORE( &r[0], 0, CHAN_Z ); + } } if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) { @@ -1435,7 +978,7 @@ exec_instruction( case TGSI_OPCODE_RCP: /* TGSI_OPCODE_RECIP */ FETCH( &r[0], 0, CHAN_X ); - micro_div( &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &r[0] ); + r[0].q = micro_div(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); } @@ -1444,8 +987,8 @@ exec_instruction( case TGSI_OPCODE_RSQ: /* TGSI_OPCODE_RECIPSQRT */ FETCH( &r[0], 0, CHAN_X ); - micro_sqrt( &r[0], &r[0] ); - micro_div( &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &r[0] ); + r[0].q = micro_sqrt(r[0].q); + r[0].q = micro_div(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); } @@ -1465,7 +1008,7 @@ exec_instruction( FETCH(&r[0], 0, chan_index); FETCH(&r[1], 1, chan_index); - micro_mul( &r[0], &r[0], &r[1] ); + r[0].q = si_fm(r[0].q, r[1].q); STORE(&r[0], 0, chan_index); } @@ -1475,7 +1018,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fa(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; @@ -1484,17 +1027,16 @@ exec_instruction( /* TGSI_OPCODE_DOT3 */ FETCH( &r[0], 0, CHAN_X ); FETCH( &r[1], 1, CHAN_X ); - micro_mul( &r[0], &r[0], &r[1] ); + r[0].q = si_fm(r[0].q, r[1].q); FETCH( &r[1], 0, CHAN_Y ); FETCH( &r[2], 1, CHAN_Y ); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); + FETCH( &r[1], 0, CHAN_Z ); FETCH( &r[2], 1, CHAN_Z ); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -1506,25 +1048,22 @@ exec_instruction( FETCH(&r[0], 0, CHAN_X); FETCH(&r[1], 1, CHAN_X); - micro_mul( &r[0], &r[0], &r[1] ); + r[0].q = si_fm(r[0].q, r[1].q); FETCH(&r[1], 0, CHAN_Y); FETCH(&r[2], 1, CHAN_Y); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FETCH(&r[1], 0, CHAN_Z); FETCH(&r[2], 1, CHAN_Z); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FETCH(&r[1], 0, CHAN_W); FETCH(&r[2], 1, CHAN_W); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -1539,7 +1078,7 @@ exec_instruction( if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { FETCH( &r[0], 0, CHAN_Y ); FETCH( &r[1], 1, CHAN_Y); - micro_mul( &r[0], &r[0], &r[1] ); + r[0].q = si_fm(r[0].q, r[1].q); STORE( &r[0], 0, CHAN_Y ); } @@ -1559,8 +1098,7 @@ exec_instruction( FETCH(&r[0], 0, chan_index); FETCH(&r[1], 1, chan_index); - /* XXX use micro_min()?? */ - micro_lt( &r[0], &r[0], &r[1], &r[0], &r[1] ); + r[0].q = micro_min(r[0].q, r[1].q); STORE(&r[0], 0, chan_index); } @@ -1571,8 +1109,7 @@ exec_instruction( FETCH(&r[0], 0, chan_index); FETCH(&r[1], 1, chan_index); - /* XXX use micro_max()?? */ - micro_lt( &r[0], &r[0], &r[1], &r[1], &r[0] ); + r[0].q = micro_max(r[0].q, r[1].q); STORE(&r[0], 0, chan_index ); } @@ -1583,7 +1120,10 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + + r[0].q = micro_ge(r[0].q, r[1].q); + r[0].q = si_xori(r[0].q, 0xff); + STORE( &r[0], 0, chan_index ); } break; @@ -1593,7 +1133,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_ge( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + r[0].q = micro_ge(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; @@ -1603,9 +1143,8 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_mul( &r[0], &r[0], &r[1] ); - FETCH( &r[1], 2, chan_index ); - micro_add( &r[0], &r[0], &r[1] ); + FETCH( &r[2], 2, chan_index ); + r[0].q = si_fma(r[0].q, r[1].q, r[2].q); STORE( &r[0], 0, chan_index ); } break; @@ -1615,7 +1154,7 @@ exec_instruction( FETCH(&r[0], 0, chan_index); FETCH(&r[1], 1, chan_index); - micro_sub( &r[0], &r[0], &r[1] ); + r[0].q = si_fs(r[0].q, r[1].q); STORE(&r[0], 0, chan_index); } @@ -1628,9 +1167,8 @@ exec_instruction( FETCH(&r[1], 1, chan_index); FETCH(&r[2], 2, chan_index); - micro_sub( &r[1], &r[1], &r[2] ); - micro_mul( &r[0], &r[0], &r[1] ); - micro_add( &r[0], &r[0], &r[2] ); + r[1].q = si_fs(r[1].q, r[2].q); + r[0].q = si_fma(r[0].q, r[1].q, r[2].q); STORE(&r[0], 0, chan_index); } @@ -1661,7 +1199,7 @@ exec_instruction( /* TGSI_OPCODE_FRC */ FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_frc( &r[0], &r[0] ); + r[0].q = micro_frc(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -1674,7 +1212,7 @@ exec_instruction( /* TGSI_OPCODE_FLR */ FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_flr( &r[0], &r[0] ); + r[0].q = micro_flr(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -1682,7 +1220,7 @@ exec_instruction( case TGSI_OPCODE_ROUND: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_rnd( &r[0], &r[0] ); + r[0].q = micro_rnd(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -1691,7 +1229,7 @@ exec_instruction( /* TGSI_OPCODE_EX2 */ FETCH(&r[0], 0, CHAN_X); - micro_pow( &r[0], &mach->Temps[TEMP_2_I].xyzw[TEMP_2_C], &r[0] ); + r[0].q = micro_pow(mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].q, r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -1701,7 +1239,7 @@ exec_instruction( case TGSI_OPCODE_LOGBASE2: /* TGSI_OPCODE_LG2 */ FETCH( &r[0], 0, CHAN_X ); - micro_lg2( &r[0], &r[0] ); + r[0].q = micro_lg2(r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); } @@ -1712,7 +1250,7 @@ exec_instruction( FETCH(&r[0], 0, CHAN_X); FETCH(&r[1], 1, CHAN_X); - micro_pow( &r[0], &r[0], &r[1] ); + r[0].q = micro_pow(r[0].q, r[1].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -1723,35 +1261,34 @@ exec_instruction( /* TGSI_OPCODE_XPD */ FETCH(&r[0], 0, CHAN_Y); FETCH(&r[1], 1, CHAN_Z); - - micro_mul( &r[2], &r[0], &r[1] ); - FETCH(&r[3], 0, CHAN_Z); FETCH(&r[4], 1, CHAN_Y); - micro_mul( &r[5], &r[3], &r[4] ); - micro_sub( &r[2], &r[2], &r[5] ); + /* r2 = (r0 * r1) - (r3 * r5) + */ + r[2].q = si_fm(r[3].q, r[5].q); + r[2].q = si_fms(r[0].q, r[1].q, r[2].q); if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) { STORE( &r[2], 0, CHAN_X ); } FETCH(&r[2], 1, CHAN_X); - - micro_mul( &r[3], &r[3], &r[2] ); - FETCH(&r[5], 0, CHAN_X); - micro_mul( &r[1], &r[1], &r[5] ); - micro_sub( &r[3], &r[3], &r[1] ); + /* r3 = (r3 * r2) - (r1 * r5) + */ + r[1].q = si_fm(r[1].q, r[5].q); + r[3].q = si_fms(r[3].q, r[2].q, r[1].q); if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) { STORE( &r[3], 0, CHAN_Y ); } - micro_mul( &r[5], &r[5], &r[4] ); - micro_mul( &r[0], &r[0], &r[2] ); - micro_sub( &r[5], &r[5], &r[0] ); + /* r5 = (r5 * r4) - (r0 * r2) + */ + r[0].q = si_fm(r[0].q, r[2].q); + r[5].q = si_fms(r[5].q, r[4].q, r[0].q); if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) { STORE( &r[5], 0, CHAN_Z ); @@ -1770,7 +1307,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH(&r[0], 0, chan_index); - micro_abs( &r[0], &r[0] ); + r[0].q = micro_abs(r[0].q); STORE(&r[0], 0, chan_index); } @@ -1784,23 +1321,21 @@ exec_instruction( FETCH(&r[0], 0, CHAN_X); FETCH(&r[1], 1, CHAN_X); - micro_mul( &r[0], &r[0], &r[1] ); + r[0].q = si_fm(r[0].q, r[1].q); FETCH(&r[1], 0, CHAN_Y); FETCH(&r[2], 1, CHAN_Y); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FETCH(&r[1], 0, CHAN_Z); FETCH(&r[2], 1, CHAN_Z); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FETCH(&r[1], 1, CHAN_W); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fa(r[0].q, r[1].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -1810,7 +1345,7 @@ exec_instruction( case TGSI_OPCODE_COS: FETCH(&r[0], 0, CHAN_X); - micro_cos( &r[0], &r[0] ); + r[0].q = micro_cos(r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -1820,7 +1355,7 @@ exec_instruction( case TGSI_OPCODE_DDX: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_ddx( &r[0], &r[0] ); + r[0].q = micro_ddx(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -1828,7 +1363,7 @@ exec_instruction( case TGSI_OPCODE_DDY: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_ddy( &r[0], &r[0] ); + r[0].q = micro_ddy(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -1866,9 +1401,9 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_eq( &r[0], &r[0], &r[1], - &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], - &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] ); + + r[0].q = si_fceq(r[0].q, r[1].q); + STORE( &r[0], 0, chan_index ); } break; @@ -1881,14 +1416,14 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + r[0].q = si_fcgt(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; case TGSI_OPCODE_SIN: FETCH( &r[0], 0, CHAN_X ); - micro_sin( &r[0], &r[0] ); + r[0].q = micro_sin(r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); } @@ -1898,7 +1433,10 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_ge( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + + r[0].q = si_fcgt(r[0].q, r[1].q); + r[0].q = si_xori(r[0].q, 0xff); + STORE( &r[0], 0, chan_index ); } break; @@ -1907,7 +1445,10 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_eq( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] ); + + r[0].q = si_fceq(r[0].q, r[1].q); + r[0].q = si_xori(r[0].q, 0xff); + STORE( &r[0], 0, chan_index ); } break; @@ -2038,7 +1579,11 @@ exec_instruction( FETCH(&r[1], 1, chan_index); FETCH(&r[2], 2, chan_index); - micro_lt( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[1], &r[2] ); + /* r0 = (r0 < 0.0) ? r1 : r2 + */ + r[3].q = si_xor(r[3].q, r[3].q); + r[0].q = micro_lt(r[0].q, r[3].q); + r[0].q = si_selb(r[1].q, r[2].q, r[0].q); STORE(&r[0], 0, chan_index); } @@ -2049,11 +1594,11 @@ exec_instruction( FETCH( &r[0], 0, CHAN_X ); } if( IS_CHANNEL_ENABLED( *inst, CHAN_X ) ) { - micro_cos( &r[1], &r[0] ); + r[1].q = micro_cos(r[0].q); STORE( &r[1], 0, CHAN_X ); } if( IS_CHANNEL_ENABLED( *inst, CHAN_Y ) ) { - micro_sin( &r[1], &r[0] ); + r[1].q = micro_sin(r[0].q); STORE( &r[1], 0, CHAN_Y ); } if( IS_CHANNEL_ENABLED( *inst, CHAN_Z ) ) { @@ -2075,12 +1620,11 @@ exec_instruction( case TGSI_OPCODE_DP2: FETCH( &r[0], 0, CHAN_X ); FETCH( &r[1], 1, CHAN_X ); - micro_mul( &r[0], &r[0], &r[1] ); + r[0].q = si_fm(r[0].q, r[1].q); FETCH( &r[1], 0, CHAN_Y ); FETCH( &r[2], 1, CHAN_Y ); - micro_mul( &r[1], &r[1], &r[2] ); - micro_add( &r[0], &r[0], &r[1] ); + r[0].q = si_fma(r[1].q, r[2].q, r[0].q); FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { STORE( &r[0], 0, chan_index ); @@ -2152,7 +1696,7 @@ exec_instruction( case TGSI_OPCODE_CEIL: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_ceil( &r[0], &r[0] ); + r[0].q = micro_ceil(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -2160,7 +1704,7 @@ exec_instruction( case TGSI_OPCODE_I2F: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_i2f( &r[0], &r[0] ); + r[0].q = si_csflt(r[0].q, 0); STORE( &r[0], 0, chan_index ); } break; @@ -2168,7 +1712,7 @@ exec_instruction( case TGSI_OPCODE_NOT: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_not( &r[0], &r[0] ); + r[0].q = si_xorbi(r[0].q, 0xff); STORE( &r[0], 0, chan_index ); } break; @@ -2176,7 +1720,7 @@ exec_instruction( case TGSI_OPCODE_TRUNC: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); - micro_trunc( &r[0], &r[0] ); + r[0].q = micro_trunc(r[0].q); STORE( &r[0], 0, chan_index ); } break; @@ -2185,7 +1729,9 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_shl( &r[0], &r[0], &r[1] ); + + r[0].q = si_shl(r[0].q, r[1].q); + STORE( &r[0], 0, chan_index ); } break; @@ -2194,7 +1740,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_ishr( &r[0], &r[0], &r[1] ); + r[0].q = micro_ishr(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; @@ -2203,7 +1749,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_and( &r[0], &r[0], &r[1] ); + r[0].q = si_and(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; @@ -2212,7 +1758,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_or( &r[0], &r[0], &r[1] ); + r[0].q = si_or(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; @@ -2225,7 +1771,7 @@ exec_instruction( FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); - micro_xor( &r[0], &r[0], &r[1] ); + r[0].q = si_xor(r[0].q, r[1].q); STORE( &r[0], 0, chan_index ); } break; diff --git a/src/mesa/pipe/cell/spu/spu_exec.h b/src/mesa/pipe/cell/spu/spu_exec.h index 89e422ba48d..b4c7661ef67 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.h +++ b/src/mesa/pipe/cell/spu/spu_exec.h @@ -43,6 +43,7 @@ union spu_exec_channel float f[QUAD_SIZE]; int i[QUAD_SIZE]; unsigned u[QUAD_SIZE]; + qword q; }; /** -- cgit v1.2.3 From 40147bd83556a41916892892193cc72d7977927a Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Fri, 1 Feb 2008 17:12:20 -0800 Subject: Vectorize vertex puller --- src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 186 +++++++++-------------------- src/mesa/pipe/cell/spu/spu_vertex_shader.h | 4 +- 2 files changed, 61 insertions(+), 129 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c index 5b0f2a6470c..4133fbba179 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -42,6 +42,8 @@ #define DRAW_DBG 0 +static const vec_float4 defaults = { 0.0, 0.0, 0.0, 1.0 }; + /** * Fetch a float[4] vertex attribute from memory, doing format/type * conversion as needed. @@ -50,19 +52,16 @@ * conversion, texture sampling etc. */ #define FETCH_ATTRIB( NAME, SZ, CVT ) \ -static void \ -fetch_##NAME(const void *ptr, float *attrib) \ +static qword \ +fetch_##NAME(const void *ptr) \ { \ - static const float defaults[4] = { 0,0,0,1 }; \ + vec_float4 attrib = defaults; \ int i; \ \ for (i = 0; i < SZ; i++) { \ - attrib[i] = CVT; \ - } \ - \ - for (; i < 4; i++) { \ - attrib[i] = defaults[i]; \ + attrib = spu_insert(CVT, attrib, i); \ } \ + return (qword) attrib; \ } #define CVT_64_FLOAT (float) ((double *) ptr)[i] @@ -309,106 +308,59 @@ static spu_fetch_func get_fetch_func( enum pipe_format format ) } -static void -transpose_4x4( float *out, const float *in ) -{ - /* This can be achieved in 12 sse instructions, plus the final - * stores I guess. This is probably a bit more than that - maybe - * 32 or so? - */ - out[0] = in[0]; out[1] = in[4]; out[2] = in[8]; out[3] = in[12]; - out[4] = in[1]; out[5] = in[5]; out[6] = in[9]; out[7] = in[13]; - out[8] = in[2]; out[9] = in[6]; out[10] = in[10]; out[11] = in[14]; - out[12] = in[3]; out[13] = in[7]; out[14] = in[11]; out[15] = in[15]; -} - - - -static void fetch_xyz_rgb( struct spu_vs_context *draw, - struct spu_exec_machine *machine, - const unsigned *elts, - unsigned count ) +void +spu_transpose_4x4(qword *out, const qword *in) { - assert(count <= 4); - -// _mesa_printf("%s\n", __FUNCTION__); - - /* loop over vertex attributes (vertex shader inputs) - */ - - const unsigned *pitch = draw->vertex_fetch.pitch; - const ubyte **src = draw->vertex_fetch.src_ptr; - int i; - - for (i = 0; i < 4; i++) { + static const qword masks[8] = { { - const float *in = (const float *)(src[0] + elts[i] * pitch[0]); - float *out = &machine->Inputs[0].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } - + 0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13, + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + }, { - const float *in = (const float *)(src[1] + elts[i] * pitch[1]); - float *out = &machine->Inputs[1].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } - } -} - - - - -static void fetch_xyz_rgb_st( struct spu_vs_context *draw, - struct spu_exec_machine *machine, - const unsigned *elts, - unsigned count ) -{ - assert(count <= 4); - - /* loop over vertex attributes (vertex shader inputs) - */ - - const unsigned *pitch = draw->vertex_fetch.pitch; - const ubyte **src = draw->vertex_fetch.src_ptr; - int i; - - for (i = 0; i < 4; i++) { + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + 0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13, + }, + + { + 0x04, 0x05, 0x06, 0x07, 0x14, 0x15, 0x16, 0x17, + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + }, + { + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + 0x04, 0x05, 0x06, 0x07, 0x14, 0x15, 0x16, 0x17, + }, + + { + 0x08, 0x09, 0x0a, 0x0b, 0x18, 0x19, 0x1a, 0x1b, + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + }, + { + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + 0x08, 0x09, 0x0a, 0x0b, 0x18, 0x19, 0x1a, 0x1b, + }, + + { + 0x0c, 0x0d, 0x0e, 0x0f, 0x1c, 0x1d, 0x1e, 0x1f, + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + }, { - const float *in = (const float *)(src[0] + elts[i] * pitch[0]); - float *out = &machine->Inputs[0].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } + 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, + 0x0c, 0x0d, 0x0e, 0x0f, 0x1c, 0x1d, 0x1e, 0x1f, + }, + }; - { - const float *in = (const float *)(src[1] + elts[i] * pitch[1]); - float *out = &machine->Inputs[1].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = in[2]; - out[12] = 1.0f; - } + out[0] = si_shufb(in[0], in[1], masks[0]); + out[0] = si_or(out[0], si_shufb(in[2], in[3], masks[1])); - { - const float *in = (const float *)(src[2] + elts[i] * pitch[2]); - float *out = &machine->Inputs[1].xyzw[0].f[i]; - out[0] = in[0]; - out[4] = in[1]; - out[8] = 0.0f; - out[12] = 1.0f; - } - } -} + out[1] = si_shufb(in[0], in[1], masks[2]); + out[1] = si_or(out[1], si_shufb(in[2], in[3], masks[3])); + out[2] = si_shufb(in[0], in[1], masks[4]); + out[2] = si_or(out[2], si_shufb(in[2], in[3], masks[5])); + out[3] = si_shufb(in[0], in[1], masks[6]); + out[3] = si_or(out[3], si_shufb(in[2], in[3], masks[7])); +} /** @@ -435,7 +387,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, const uint64_t src = draw->vertex_fetch.src_ptr[attr]; const spu_fetch_func fetch = draw->vertex_fetch.fetch[attr]; unsigned i; - float p[4][4]; + qword p[4]; /* Fetch four attributes for four vertices. @@ -452,17 +404,15 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, mfc_get(buffer, addr & ~0x0f, size, TAG_VERTEX_BUFFER, 0, 0); wait_on_mask(1 << TAG_VERTEX_BUFFER); - memmove(& buffer, buffer + (addr & 0x0f), 16); - - fetch(buffer, p[i]); + p[i] = (*fetch)(buffer + (addr & 0x0f)); } /* Be nice and zero out any missing vertices: */ for (/* empty */; i < 4; i++) - p[i][0] = p[i][1] = p[i][2] = p[i][3] = 0; - - /* Transpose/swizzle into sse-friendly format. Currently + p[i] = si_xor(p[i], p[i]); + + /* Transpose/swizzle into vector-friendly format. Currently * assuming that all vertex shader inputs are float[4], but this * isn't true -- if the vertex shader only wants tex0.xy, we * could optimize for that. @@ -471,7 +421,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, * excessive number of fetch functions, but we could at least * minimize the transpose step: */ - transpose_4x4( (float *)&machine->Inputs[attr].xyzw[0].f[0], (float *)p ); + spu_transpose_4x4(&machine->Inputs[attr].xyzw[0].q, p); } } @@ -487,24 +437,4 @@ void spu_update_vertex_fetch( struct spu_vs_context *draw ) } draw->vertex_fetch.fetch_func = generic_vertex_fetch; - - /* Disable the fast path because they don't use mfc_get yet. - */ -#if 0 - switch (draw->vertex_fetch.nr_attrs) { - case 2: - if (draw->vertex_fetch.format[0] == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_fetch.format[1] == PIPE_FORMAT_R32G32B32_FLOAT) - draw->vertex_fetch.fetch_func = fetch_xyz_rgb; - break; - case 3: - if (draw->vertex_fetch.format[0] == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_fetch.format[1] == PIPE_FORMAT_R32G32B32_FLOAT && - draw->vertex_fetch.format[2] == PIPE_FORMAT_R32G32_FLOAT) - draw->vertex_fetch.fetch_func = fetch_xyz_rgb_st; - break; - default: - break; - } -#endif } diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.h b/src/mesa/pipe/cell/spu/spu_vertex_shader.h index b261ab44a21..2435b7ddae0 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.h +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.h @@ -6,7 +6,7 @@ struct spu_vs_context; -typedef void (*spu_fetch_func)(const void *ptr, float *attrib); +typedef qword (*spu_fetch_func)(const void *ptr); typedef void (*spu_full_fetch_func)( struct spu_vs_context *draw, struct spu_exec_machine *machine, const unsigned *elts, @@ -39,6 +39,8 @@ struct spu_vs_context { extern void spu_update_vertex_fetch(struct spu_vs_context *draw); +extern void spu_transpose_4x4(qword *out, const qword *in); + static INLINE void spu_vertex_fetch(struct spu_vs_context *draw, struct spu_exec_machine *machine, const unsigned *elts, -- cgit v1.2.3 From f33c8119abbe5980b793961298978dddc5b0563f Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Fri, 1 Feb 2008 17:14:09 -0800 Subject: More semi-trivial vectorization in the shader VM --- src/mesa/pipe/cell/spu/spu_exec.c | 62 +++++++++++++++++---------------------- 1 file changed, 27 insertions(+), 35 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index 1ac9c031e32..1bd8687d419 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -70,6 +70,7 @@ #include "pipe/tgsi/util/tgsi_util.h" #include "spu_exec.h" #include "spu_main.h" +#include "spu_vertex_shader.h" #define TILE_TOP_LEFT 0 #define TILE_TOP_RIGHT 1 @@ -144,23 +145,27 @@ spu_exec_machine_init(struct spu_exec_machine *mach, struct spu_sampler *samplers, unsigned processor) { + qword zero; + qword not_zero; uint i; mach->Samplers = samplers; mach->Processor = processor; mach->Addrs = &mach->Temps[TGSI_EXEC_NUM_TEMPS]; + zero = si_xor(zero, zero); + not_zero = si_xori(zero, 0xff); + /* Setup constants. */ - for( i = 0; i < 4; i++ ) { - mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].u[i] = 0x00000000; - mach->Temps[TEMP_7F_I].xyzw[TEMP_7F_C].u[i] = 0x7FFFFFFF; - mach->Temps[TEMP_80_I].xyzw[TEMP_80_C].u[i] = 0x80000000; - mach->Temps[TEMP_FF_I].xyzw[TEMP_FF_C].u[i] = 0xFFFFFFFF; - mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].f[i] = 1.0f; - mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].f[i] = 2.0f; - mach->Temps[TEMP_128_I].xyzw[TEMP_128_C].f[i] = 128.0f; - mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C].f[i] = -128.0f; - } + mach->Temps[TEMP_0_I].xyzw[TEMP_0_C].q = zero; + mach->Temps[TEMP_FF_I].xyzw[TEMP_FF_C].q = not_zero; + mach->Temps[TEMP_7F_I].xyzw[TEMP_7F_C].q = si_shli(not_zero, -1); + mach->Temps[TEMP_80_I].xyzw[TEMP_80_C].q = si_shli(not_zero, 31); + + mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q = (qword) spu_splats(1.0f); + mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].q = (qword) spu_splats(2.0f); + mach->Temps[TEMP_128_I].xyzw[TEMP_128_C].q = (qword) spu_splats(128.0f); + mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C].q = (qword) spu_splats(-128.0f); } @@ -459,25 +464,16 @@ fetch_source( &index2, &indir_index ); - index.i[0] += indir_index.i[0]; - index.i[1] += indir_index.i[1]; - index.i[2] += indir_index.i[2]; - index.i[3] += indir_index.i[3]; + index.q = si_a(index.q, indir_index.q); } if( reg->SrcRegister.Dimension ) { switch( reg->SrcRegister.File ) { case TGSI_FILE_INPUT: - index.i[0] *= 17; - index.i[1] *= 17; - index.i[2] *= 17; - index.i[3] *= 17; + index.q = si_mpyi(index.q, 17); break; case TGSI_FILE_CONSTANT: - index.i[0] *= 4096; - index.i[1] *= 4096; - index.i[2] *= 4096; - index.i[3] *= 4096; + index.q = si_shli(index.q, 12); break; default: assert( 0 ); @@ -505,10 +501,7 @@ fetch_source( &index2, &indir_index ); - index.i[0] += indir_index.i[0]; - index.i[1] += indir_index.i[1]; - index.i[2] += indir_index.i[2]; - index.i[3] += indir_index.i[3]; + index.q = si_a(index.q, indir_index.q); } } @@ -666,17 +659,16 @@ fetch_texel( struct spu_sampler *sampler, union spu_exec_channel *b, union spu_exec_channel *a ) { - uint j; - float rgba[NUM_CHANNELS][QUAD_SIZE]; + qword rgba[4]; + qword out[4]; - sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, rgba); + sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, (float *) rgba); - for (j = 0; j < 4; j++) { - r->f[j] = rgba[0][j]; - g->f[j] = rgba[1][j]; - b->f[j] = rgba[2][j]; - a->f[j] = rgba[3][j]; - } + spu_transpose_4x4(out, rgba); + r->q = out[0]; + g->q = out[1]; + b->q = out[2]; + a->q = out[3]; } -- cgit v1.2.3 From 76702d5fcd1d8341a099adfabef94a0f847ca06f Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Mon, 4 Feb 2008 16:03:55 -0800 Subject: Add some debug messages --- src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c index 4133fbba179..cfa449e8130 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -378,7 +378,10 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, wait_on_mask(1 << TAG_VERTEX_BUFFER); -// _mesa_printf("%s %d\n", __FUNCTION__, count); +#if DRAW_DBG + printf("SPU: %s count = %u, nr_attrs = %u\n", + __FUNCTION__, count, nr_attrs); +#endif /* loop over vertex attributes (vertex shader inputs) */ @@ -401,6 +404,9 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, const uint64_t addr = src + (elts[i] * pitch); const unsigned size = ((addr & 0x0f) == 0) ? 16 : 32; +#if DRAW_DBG + printf("SPU: fetching = 0x%llx\n", addr); +#endif mfc_get(buffer, addr & ~0x0f, size, TAG_VERTEX_BUFFER, 0, 0); wait_on_mask(1 << TAG_VERTEX_BUFFER); -- cgit v1.2.3 From ac07631d85d0f1d30b1feba23f0f2f2c6549466d Mon Sep 17 00:00:00 2001 From: Ian Romanick Date: Tue, 5 Feb 2008 09:43:52 -0800 Subject: Use _transpose_matrix4x4 from Cell SDK instead of my own version --- src/mesa/pipe/cell/spu/spu_exec.c | 3 +- src/mesa/pipe/cell/spu/spu_vertex_fetch.c | 59 ++---------------------------- src/mesa/pipe/cell/spu/spu_vertex_shader.h | 2 - 3 files changed, 5 insertions(+), 59 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_exec.c b/src/mesa/pipe/cell/spu/spu_exec.c index 1bd8687d419..e51008b9b3c 100644 --- a/src/mesa/pipe/cell/spu/spu_exec.c +++ b/src/mesa/pipe/cell/spu/spu_exec.c @@ -52,6 +52,7 @@ #include #include +#include #include #include #include @@ -664,7 +665,7 @@ fetch_texel( struct spu_sampler *sampler, sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, (float *) rgba); - spu_transpose_4x4(out, rgba); + _transpose_matrix4x4(out, rgba); r->q = out[0]; g->q = out[1]; b->q = out[2]; diff --git a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c index cfa449e8130..6e86a919ce4 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_fetch.c +++ b/src/mesa/pipe/cell/spu/spu_vertex_fetch.c @@ -31,6 +31,8 @@ */ #include +#include + #include "pipe/p_util.h" #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" @@ -308,61 +310,6 @@ static spu_fetch_func get_fetch_func( enum pipe_format format ) } -void -spu_transpose_4x4(qword *out, const qword *in) -{ - static const qword masks[8] = { - { - 0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13, - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - }, - { - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - 0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13, - }, - - { - 0x04, 0x05, 0x06, 0x07, 0x14, 0x15, 0x16, 0x17, - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - }, - { - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - 0x04, 0x05, 0x06, 0x07, 0x14, 0x15, 0x16, 0x17, - }, - - { - 0x08, 0x09, 0x0a, 0x0b, 0x18, 0x19, 0x1a, 0x1b, - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - }, - { - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - 0x08, 0x09, 0x0a, 0x0b, 0x18, 0x19, 0x1a, 0x1b, - }, - - { - 0x0c, 0x0d, 0x0e, 0x0f, 0x1c, 0x1d, 0x1e, 0x1f, - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - }, - { - 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, - 0x0c, 0x0d, 0x0e, 0x0f, 0x1c, 0x1d, 0x1e, 0x1f, - }, - }; - - out[0] = si_shufb(in[0], in[1], masks[0]); - out[0] = si_or(out[0], si_shufb(in[2], in[3], masks[1])); - - out[1] = si_shufb(in[0], in[1], masks[2]); - out[1] = si_or(out[1], si_shufb(in[2], in[3], masks[3])); - - out[2] = si_shufb(in[0], in[1], masks[4]); - out[2] = si_or(out[2], si_shufb(in[2], in[3], masks[5])); - - out[3] = si_shufb(in[0], in[1], masks[6]); - out[3] = si_or(out[3], si_shufb(in[2], in[3], masks[7])); -} - - /** * Fetch vertex attributes for 'count' vertices. */ @@ -427,7 +374,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, * excessive number of fetch functions, but we could at least * minimize the transpose step: */ - spu_transpose_4x4(&machine->Inputs[attr].xyzw[0].q, p); + _transpose_matrix4x4(&machine->Inputs[attr].xyzw[0].q, p); } } diff --git a/src/mesa/pipe/cell/spu/spu_vertex_shader.h b/src/mesa/pipe/cell/spu/spu_vertex_shader.h index 2435b7ddae0..c96b93ff0ac 100644 --- a/src/mesa/pipe/cell/spu/spu_vertex_shader.h +++ b/src/mesa/pipe/cell/spu/spu_vertex_shader.h @@ -39,8 +39,6 @@ struct spu_vs_context { extern void spu_update_vertex_fetch(struct spu_vs_context *draw); -extern void spu_transpose_4x4(qword *out, const qword *in); - static INLINE void spu_vertex_fetch(struct spu_vs_context *draw, struct spu_exec_machine *machine, const unsigned *elts, -- cgit v1.2.3 From 48aad039398df0024126ff5892a62aca77b65547 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 14:21:01 -0700 Subject: Cell: added cell_batch_alloc_aligned() --- src/mesa/pipe/cell/ppu/cell_batch.c | 26 ++++++++++++++++++++------ src/mesa/pipe/cell/ppu/cell_batch.h | 4 ++++ 2 files changed, 24 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_batch.c b/src/mesa/pipe/cell/ppu/cell_batch.c index 2fb49711b2d..f45e5f25b64 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.c +++ b/src/mesa/pipe/cell/ppu/cell_batch.c @@ -157,7 +157,7 @@ cell_batch_append(struct cell_context *cell, const void *data, uint bytes) size = 0; } - assert(size + bytes <= CELL_BUFFER_SIZE); + ASSERT(size + bytes <= CELL_BUFFER_SIZE); memcpy(cell->buffer[cell->cur_batch] + size, data, bytes); @@ -167,14 +167,22 @@ cell_batch_append(struct cell_context *cell, const void *data, uint bytes) void * cell_batch_alloc(struct cell_context *cell, uint bytes) +{ + return cell_batch_alloc_aligned(cell, bytes, 1); +} + + +void * +cell_batch_alloc_aligned(struct cell_context *cell, uint bytes, + uint alignment) { void *pos; - uint size; + uint size, padbytes; ASSERT(bytes % 8 == 0); ASSERT(bytes <= CELL_BUFFER_SIZE); - - assert(cell->cur_batch >= 0); + ASSERT(alignment > 0); + ASSERT(cell->cur_batch >= 0); #ifdef ASSERT { @@ -188,12 +196,18 @@ cell_batch_alloc(struct cell_context *cell, uint bytes) size = cell->buffer_size[cell->cur_batch]; - if (size + bytes > CELL_BUFFER_SIZE) { + padbytes = (alignment - (size % alignment)) % alignment; + + if (padbytes + size + bytes > CELL_BUFFER_SIZE) { cell_batch_flush(cell); size = 0; } + else { + size += padbytes; + } - assert(size + bytes <= CELL_BUFFER_SIZE); + ASSERT(size % alignment == 0); + ASSERT(size + bytes <= CELL_BUFFER_SIZE); pos = (void *) (cell->buffer[cell->cur_batch] + size); diff --git a/src/mesa/pipe/cell/ppu/cell_batch.h b/src/mesa/pipe/cell/ppu/cell_batch.h index f4f37314a48..a6eee0a8b18 100644 --- a/src/mesa/pipe/cell/ppu/cell_batch.h +++ b/src/mesa/pipe/cell/ppu/cell_batch.h @@ -50,5 +50,9 @@ cell_batch_append(struct cell_context *cell, const void *data, uint bytes); extern void * cell_batch_alloc(struct cell_context *cell, uint bytes); +extern void * +cell_batch_alloc_aligned(struct cell_context *cell, uint bytes, + uint alignment); + #endif /* CELL_BATCH_H */ -- cgit v1.2.3 From 2ec5ae5e504d8bf82cdecae8569dc12b8d62c055 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 14:21:48 -0700 Subject: Cell: remove dummy fields, update/add some comments --- src/mesa/pipe/cell/common.h | 14 ++++++-------- 1 file changed, 6 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/cell/common.h b/src/mesa/pipe/cell/common.h index cf8fc94ebf1..4de514c3586 100644 --- a/src/mesa/pipe/cell/common.h +++ b/src/mesa/pipe/cell/common.h @@ -51,7 +51,7 @@ /** for sanity checking */ #define ASSERT_ALIGN16(ptr) \ - assert((((unsigned long) (ptr)) & 0xf) == 0); + ASSERT((((unsigned long) (ptr)) & 0xf) == 0); /** round up value to next multiple of 4 */ @@ -105,7 +105,7 @@ */ struct cell_command_framebuffer { - uint64_t opcode; + uint64_t opcode; /**< CELL_CMD_FRAMEBUFFER */ int width, height; void *color_start, *depth_start; enum pipe_format color_format, depth_format; @@ -117,7 +117,7 @@ struct cell_command_framebuffer */ struct cell_command_clear_surface { - uint64_t opcode; + uint64_t opcode; /**< CELL_CMD_CLEAR_SURFACE */ uint surface; /**< Temporary: 0=color, 1=Z */ uint value; }; @@ -128,8 +128,8 @@ struct cell_command_clear_surface */ struct cell_array_info { - uint64_t base; /**< Base address of the 0th element. */ - uint attr; /**< Attribute that this state if for. */ + uint64_t base; /**< Base address of the 0th element. */ + uint attr; /**< Attribute that this state is for. */ uint pitch; /**< Byte pitch from one entry to the next. */ uint format; /**< Pipe format of each entry. */ } ALIGN16_ATTRIB; @@ -169,11 +169,9 @@ struct cell_command_render uint prim_type; /**< PIPE_PRIM_x */ uint num_verts; uint vertex_size; /**< bytes per vertex */ - uint dummy; /* XXX this dummy field works around a compiler bug */ uint num_indexes; uint vertex_buf; /**< which cell->buffer[] contains the vertex data */ - float xmin, dummy2, ymin, xmax, ymax; /* XXX another dummy field */ - uint dummy3; + float xmin, ymin, xmax, ymax; /* XXX another dummy field */ uint min_index; boolean inline_verts; }; -- cgit v1.2.3 From df2ab198eb49333f01c8f10bea2033bea732d755 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 14:23:34 -0700 Subject: Cell: re-enable inlined vertex buffers Vertex data must be on a 16-byte address/offset so SIMD operations will work properly in the SPU code. --- src/mesa/pipe/cell/ppu/cell_vbuf.c | 12 +++++------- src/mesa/pipe/cell/spu/spu_main.c | 3 ++- src/mesa/pipe/cell/spu/spu_render.c | 12 ++++++++---- 3 files changed, 15 insertions(+), 12 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_vbuf.c b/src/mesa/pipe/cell/ppu/cell_vbuf.c index 0fee61821a8..e9fafe492ee 100644 --- a/src/mesa/pipe/cell/ppu/cell_vbuf.c +++ b/src/mesa/pipe/cell/ppu/cell_vbuf.c @@ -40,7 +40,7 @@ /** Allow vertex data to be inlined after RENDER command */ -#define ALLOW_INLINE_VERTS 0 +#define ALLOW_INLINE_VERTS 1 /** @@ -199,9 +199,7 @@ cell_vbuf_draw(struct vbuf_render *vbr, { const uint index_bytes = ROUNDUP8(nr_indices * 2); const uint vertex_bytes = nr_vertices * 4 * cell->vertex_info.size; - - const uint batch_size = sizeof(struct cell_command_render) - + index_bytes; + const uint batch_size = sizeof(struct cell_command_render) + index_bytes; struct cell_command_render *render = (struct cell_command_render *) @@ -223,9 +221,9 @@ cell_vbuf_draw(struct vbuf_render *vbr, render->num_verts = nr_vertices; if (ALLOW_INLINE_VERTS && min_index == 0 && - vertex_bytes <= cell_batch_free_space(cell)) { - /* vertex data inlined, after indices */ - void *dst = cell_batch_alloc(cell, vertex_bytes); + vertex_bytes + 16 <= cell_batch_free_space(cell)) { + /* vertex data inlined, after indices, at 16-byte boundary */ + void *dst = cell_batch_alloc_aligned(cell, vertex_bytes, 16); memcpy(dst, vertices, vertex_bytes); render->inline_verts = TRUE; render->vertex_buf = ~0; diff --git a/src/mesa/pipe/cell/spu/spu_main.c b/src/mesa/pipe/cell/spu/spu_main.c index 4f126d5e5bb..e375197fe60 100644 --- a/src/mesa/pipe/cell/spu/spu_main.c +++ b/src/mesa/pipe/cell/spu/spu_main.c @@ -387,7 +387,7 @@ cmd_batch(uint opcode) = (struct cell_command_render *) &buffer[pos]; uint pos_incr; cmd_render(render, &pos_incr); - pos += sizeof(*render) / 8 + ((pos_incr + 1) / 2); + pos += pos_incr; } break; case CELL_CMD_RELEASE_VERTS: @@ -541,6 +541,7 @@ main(main_param_t speid, main_param_t argp) (void) speid; ASSERT(sizeof(tile_t) == TILE_SIZE * TILE_SIZE * 4); + ASSERT(sizeof(struct cell_command_render) % 8 == 0); one_time_init(); diff --git a/src/mesa/pipe/cell/spu/spu_render.c b/src/mesa/pipe/cell/spu/spu_render.c index e8705eeebad..932fb500b3f 100644 --- a/src/mesa/pipe/cell/spu/spu_render.c +++ b/src/mesa/pipe/cell/spu/spu_render.c @@ -171,6 +171,7 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB; const uint vertex_size = render->vertex_size; /* in bytes */ /*const*/ uint total_vertex_bytes = render->num_verts * vertex_size; + uint index_bytes; const ubyte *vertices; const ushort *indexes; uint i, j; @@ -199,13 +200,16 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr) /* indexes are right after the render command in the batch buffer */ indexes = (const ushort *) (render + 1); - *pos_incr = (render->num_indexes * 2 + 3) / 4; + index_bytes = ROUNDUP8(render->num_indexes * 2); + *pos_incr = index_bytes / 8 + sizeof(*render) / 8; if (render->inline_verts) { - /* Vertices are right after indexes in batch buffer */ - vertices = (const ubyte *) (render + 1) + *pos_incr * 4; - *pos_incr = *pos_incr + total_vertex_bytes / 4; + /* Vertices are after indexes in batch buffer at next 16-byte addr */ + vertices = (const ubyte *) render + (*pos_incr * 8); + vertices = (const ubyte *) align_pointer((void *) vertices, 16); + ASSERT_ALIGN16(vertices); + *pos_incr = ((vertices + total_vertex_bytes) - (ubyte *) render) / 8; } else { /* Begin DMA fetch of vertex buffer */ -- cgit v1.2.3 From 14f1f2523b0186fe65f13b1981a782768d4f8376 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 15:07:36 -0700 Subject: Cell: SIMD-ize tri_linear_coeff(), use vector float for vertex attributes in struct vertex_header --- src/mesa/pipe/cell/spu/spu_tri.c | 112 ++++++++++++++++++++++++++------------- 1 file changed, 75 insertions(+), 37 deletions(-) diff --git a/src/mesa/pipe/cell/spu/spu_tri.c b/src/mesa/pipe/cell/spu/spu_tri.c index 688c8646ab2..be9624cf7d9 100644 --- a/src/mesa/pipe/cell/spu/spu_tri.c +++ b/src/mesa/pipe/cell/spu/spu_tri.c @@ -56,7 +56,7 @@ typedef union * Simplified types taken from other parts of Gallium */ struct vertex_header { - float data[0][4]; + vector float data[1]; }; @@ -476,6 +476,7 @@ static void print_vertex(const struct vertex_header *v) } #endif + static boolean setup_sort_vertices(const struct vertex_header *v0, const struct vertex_header *v1, const struct vertex_header *v2) @@ -492,9 +493,9 @@ static boolean setup_sort_vertices(const struct vertex_header *v0, /* determine bottom to top order of vertices */ { - float y0 = v0->data[0][1]; - float y1 = v1->data[0][1]; - float y2 = v2->data[0][1]; + float y0 = spu_extract(v0->data[0], 1); + float y1 = spu_extract(v1->data[0], 1); + float y2 = spu_extract(v2->data[0], 1); if (y0 <= y1) { if (y1 <= y2) { /* y0<=y1<=y2 */ @@ -538,25 +539,25 @@ static boolean setup_sort_vertices(const struct vertex_header *v0, } /* Check if triangle is completely outside the tile bounds */ - if (setup.vmin->data[0][1] > setup.cliprect_maxy) + if (spu_extract(setup.vmin->data[0], 1) > setup.cliprect_maxy) return FALSE; - if (setup.vmax->data[0][1] < setup.cliprect_miny) + if (spu_extract(setup.vmax->data[0], 1) < setup.cliprect_miny) return FALSE; - if (setup.vmin->data[0][0] < setup.cliprect_minx && - setup.vmid->data[0][0] < setup.cliprect_minx && - setup.vmax->data[0][0] < setup.cliprect_minx) + if (spu_extract(setup.vmin->data[0], 0) < setup.cliprect_minx && + spu_extract(setup.vmid->data[0], 0) < setup.cliprect_minx && + spu_extract(setup.vmax->data[0], 0) < setup.cliprect_minx) return FALSE; - if (setup.vmin->data[0][0] > setup.cliprect_maxx && - setup.vmid->data[0][0] > setup.cliprect_maxx && - setup.vmax->data[0][0] > setup.cliprect_maxx) + if (spu_extract(setup.vmin->data[0], 0) > setup.cliprect_maxx && + spu_extract(setup.vmid->data[0], 0) > setup.cliprect_maxx && + spu_extract(setup.vmax->data[0], 0) > setup.cliprect_maxx) return FALSE; - setup.ebot.dx = setup.vmid->data[0][0] - setup.vmin->data[0][0]; - setup.ebot.dy = setup.vmid->data[0][1] - setup.vmin->data[0][1]; - setup.emaj.dx = setup.vmax->data[0][0] - setup.vmin->data[0][0]; - setup.emaj.dy = setup.vmax->data[0][1] - setup.vmin->data[0][1]; - setup.etop.dx = setup.vmax->data[0][0] - setup.vmid->data[0][0]; - setup.etop.dy = setup.vmax->data[0][1] - setup.vmid->data[0][1]; + setup.ebot.dx = spu_extract(setup.vmid->data[0], 0) - spu_extract(setup.vmin->data[0], 0); + setup.ebot.dy = spu_extract(setup.vmid->data[0], 1) - spu_extract(setup.vmin->data[0], 1); + setup.emaj.dx = spu_extract(setup.vmax->data[0], 0) - spu_extract(setup.vmin->data[0], 0); + setup.emaj.dy = spu_extract(setup.vmax->data[0], 1) - spu_extract(setup.vmin->data[0], 1); + setup.etop.dx = spu_extract(setup.vmax->data[0], 0) - spu_extract(setup.vmid->data[0], 0); + setup.etop.dy = spu_extract(setup.vmax->data[0], 1) - spu_extract(setup.vmid->data[0], 1); /* * Compute triangle's area. Use 1/area to compute partial @@ -597,14 +598,12 @@ static boolean setup_sort_vertices(const struct vertex_header *v0, * The result will be put into setup.coef[slot].a0. * \param slot which attribute slot */ -static INLINE void const_coeff(uint slot) +static INLINE void +const_coeff(uint slot) { setup.coef[slot].dadx.v = (vector float) {0.0, 0.0, 0.0, 0.0}; setup.coef[slot].dady.v = (vector float) {0.0, 0.0, 0.0, 0.0}; - setup.coef[slot].a0.f[0] = setup.vprovoke->data[slot][0]; - setup.coef[slot].a0.f[1] = setup.vprovoke->data[slot][1]; - setup.coef[slot].a0.f[2] = setup.vprovoke->data[slot][2]; - setup.coef[slot].a0.f[3] = setup.vprovoke->data[slot][3]; + setup.coef[slot].a0.v = setup.vprovoke->data[slot]; } @@ -612,12 +611,19 @@ static INLINE void const_coeff(uint slot) * Compute a0, dadx and dady for a linearly interpolated coefficient, * for a triangle. */ -static void tri_linear_coeff( uint slot, uint firstComp, uint lastComp ) +static INLINE void +tri_linear_coeff(uint slot, uint firstComp, uint lastComp) { uint i; + const float *vmin_d = (float *) &setup.vmin->data[slot]; + const float *vmid_d = (float *) &setup.vmid->data[slot]; + const float *vmax_d = (float *) &setup.vmax->data[slot]; + const float x = spu_extract(setup.vmin->data[0], 0) - 0.5f; + const float y = spu_extract(setup.vmin->data[0], 1) - 0.5f; + for (i = firstComp; i < lastComp; i++) { - float botda = setup.vmid->data[slot][i] - setup.vmin->data[slot][i]; - float majda = setup.vmax->data[slot][i] - setup.vmin->data[slot][i]; + float botda = vmid_d[i] - vmin_d[i]; + float majda = vmax_d[i] - vmin_d[i]; float a = setup.ebot.dy * majda - botda * setup.emaj.dy; float b = setup.emaj.dx * botda - majda * setup.ebot.dx; @@ -638,9 +644,9 @@ static void tri_linear_coeff( uint slot, uint firstComp, uint lastComp ) * to define a0 as the sample at a pixel center somewhere near vmin * instead - i'll switch to this later. */ - setup.coef[slot].a0.f[i] = (setup.vmin->data[slot][i] - - (setup.coef[slot].dadx.f[i] * (setup.vmin->data[0][0] - 0.5f) + - setup.coef[slot].dady.f[i] * (setup.vmin->data[0][1] - 0.5f))); + setup.coef[slot].a0.f[i] = (vmin_d[i] - + (setup.coef[slot].dadx.f[i] * x + + setup.coef[slot].dady.f[i] * y)); } /* @@ -653,6 +659,37 @@ static void tri_linear_coeff( uint slot, uint firstComp, uint lastComp ) } +/** + * As above, but interp setup all four vector components. + */ +static INLINE void +tri_linear_coeff4(uint slot) +{ + const vector float vmin_d = setup.vmin->data[slot]; + const vector float vmid_d = setup.vmid->data[slot]; + const vector float vmax_d = setup.vmax->data[slot]; + const vector float xxxx = spu_splats(spu_extract(setup.vmin->data[0], 0) - 0.5f); + const vector float yyyy = spu_splats(spu_extract(setup.vmin->data[0], 1) - 0.5f); + + vector float botda = vmid_d - vmin_d; + vector float majda = vmax_d - vmin_d; + + vector float a = spu_sub(spu_mul(spu_splats(setup.ebot.dy), majda), + spu_mul(botda, spu_splats(setup.emaj.dy))); + vector float b = spu_sub(spu_mul(spu_splats(setup.emaj.dx), botda), + spu_mul(majda, spu_splats(setup.ebot.dx))); + + setup.coef[slot].dadx.v = spu_mul(a, spu_splats(setup.oneoverarea)); + setup.coef[slot].dady.v = spu_mul(b, spu_splats(setup.oneoverarea)); + + vector float tempx = spu_mul(setup.coef[slot].dadx.v, xxxx); + vector float tempy = spu_mul(setup.coef[slot].dady.v, yyyy); + + setup.coef[slot].a0.v = spu_sub(vmin_d, spu_add(tempx, tempy)); +} + + + #if 0 /** * Compute a0, dadx and dady for a perspective-corrected interpolant, @@ -710,17 +747,18 @@ static void setup_tri_coefficients(void) case INTERP_NONE: break; case INTERP_POS: - tri_linear_coeff(i, 2, 3); + /*tri_linear_coeff(i, 2, 3);*/ /* XXX interp W if PERSPECTIVE... */ + tri_linear_coeff4(i); break; case INTERP_CONSTANT: const_coeff(i); break; case INTERP_LINEAR: - tri_linear_coeff(i, 0, 4); + tri_linear_coeff4(i); break; case INTERP_PERSPECTIVE: - tri_linear_coeff(i, 0, 4); /* XXX temporary */ + tri_linear_coeff4(i); /* temporary */ break; default: ASSERT(0); @@ -738,12 +776,12 @@ static void setup_tri_coefficients(void) static void setup_tri_edges(void) { - float vmin_x = setup.vmin->data[0][0] + 0.5f; - float vmid_x = setup.vmid->data[0][0] + 0.5f; + float vmin_x = spu_extract(setup.vmin->data[0], 0) + 0.5f; + float vmid_x = spu_extract(setup.vmid->data[0], 0) + 0.5f; - float vmin_y = setup.vmin->data[0][1] - 0.5f; - float vmid_y = setup.vmid->data[0][1] - 0.5f; - float vmax_y = setup.vmax->data[0][1] - 0.5f; + float vmin_y = spu_extract(setup.vmin->data[0], 1) - 0.5f; + float vmid_y = spu_extract(setup.vmid->data[0], 1) - 0.5f; + float vmax_y = spu_extract(setup.vmax->data[0], 1) - 0.5f; setup.emaj.sy = CEILF(vmin_y); setup.emaj.lines = (int) CEILF(vmax_y - setup.emaj.sy); -- cgit v1.2.3 From d154f6a24b7c7265306d43fcb3b43dc759ad9bd2 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 15:12:18 -0700 Subject: Cell: remove accidentally added OPT_FLAGS lines --- src/mesa/pipe/cell/spu/Makefile | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/mesa/pipe/cell/spu/Makefile b/src/mesa/pipe/cell/spu/Makefile index 66f16cde9b8..f202971d738 100644 --- a/src/mesa/pipe/cell/spu/Makefile +++ b/src/mesa/pipe/cell/spu/Makefile @@ -8,8 +8,6 @@ TOP = ../../../../.. include $(TOP)/configs/linux-cell -OPT_FLAGS=-g -OPT_FLAGS=-O3 PROG = g3d PROG_SPU = $(PROG)_spu -- cgit v1.2.3 From 152ea0b42484c4173eb5eb4d8ecd6a79207b2310 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Mon, 4 Feb 2008 17:02:08 +0900 Subject: gallium: Portability guidelines. --- src/mesa/pipe/README.portability | 43 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) create mode 100644 src/mesa/pipe/README.portability diff --git a/src/mesa/pipe/README.portability b/src/mesa/pipe/README.portability new file mode 100644 index 00000000000..c70ca774da0 --- /dev/null +++ b/src/mesa/pipe/README.portability @@ -0,0 +1,43 @@ + CROSS-PLATFORM PORTABILITY GUIDELINES FOR GALLIUM3D + + += General Considerations = + +The state tracker and winsys driver support a rather limited number of +platforms. However, the pipe drivers are meant to run in a wide number of +platforms. Hence the pipe drivers, the auxiliary modules, and all public +headers in general, should stricly follow these guidelines to ensure + + += Compiler Support = + +* Include the p_compiler.h. + +* Don't use the 'inline' keyword, use the INLINE macro in p_compiler.h instead. + +* Cast explicitly when converting to integer types of smaller sizes. + +* Cast explicitly when converting between float, double and integral types. + +* Don't use named struct initializers. + +* Don't use variable number of macro arguments. Use static inline functions +instead. + + += Standard Library = + +* Avoid including standard library headers. Most standard library functions are +not available in Windows Kernel Mode. Use the appropriate p_*.h include. + +== Memory Allocation == + +* Use MALLOC, CALLOC, FREE instead of the malloc, calloc, free functions. + +* Use align_pointer() function defined in p_util.h for aligning pointers in a +portable way. + +== Debugging == + +TODO + -- cgit v1.2.3 From 25c29080f772ea175b0582031d483ca79f70f8ac Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Wed, 6 Feb 2008 13:27:49 +0900 Subject: gallium: Cross-platform debugging helpers. --- src/mesa/SConscript | 1 + src/mesa/pipe/p_debug.h | 79 ++++++++++++++++++++++++++++++++++++++++++++ src/mesa/pipe/util/p_debug.c | 70 +++++++++++++++++++++++++++++++++++++++ src/mesa/sources | 1 + 4 files changed, 151 insertions(+) create mode 100644 src/mesa/pipe/p_debug.h create mode 100644 src/mesa/pipe/util/p_debug.c diff --git a/src/mesa/SConscript b/src/mesa/SConscript index 70a98f31294..faf8c84872c 100644 --- a/src/mesa/SConscript +++ b/src/mesa/SConscript @@ -158,6 +158,7 @@ STATECACHE_SOURCES = [ ] PIPEUTIL_SOURCES = [ + 'pipe/util/p_debug.c', 'pipe/util/p_tile.c', 'pipe/util/p_util.c', ] diff --git a/src/mesa/pipe/p_debug.h b/src/mesa/pipe/p_debug.h new file mode 100644 index 00000000000..b037eba2a35 --- /dev/null +++ b/src/mesa/pipe/p_debug.h @@ -0,0 +1,79 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * @file + * Cross-platform debugging helpers. + * + * For now it just has assert and printf replacements, but it might be extended + * with stack trace reports and more advanced logging in the near future. + * + * @author Jose Fonseca + */ + +#ifndef P_DEBUG_H_ +#define P_DEBUG_H_ + +#ifdef __cplusplus +extern "C" { +#endif + + +#ifdef DBG +#ifndef DEBUG +#define DEBUG 1 +#endif +#else +#ifndef NDEBUG +#define NDEBUG 1 +#endif +#endif + + +void debug_printf(const char *format, ...); +void debug_assert_fail(const char *expr, const char *file, unsigned line); + +/** Assert macro */ +#ifdef DEBUG +#define debug_assert(expr) ((expr) ? (void)0 : debug_assert_fail(#expr, __FILE__, __LINE__)) +#else +#define debug_assert(expr) ((void)0) +#endif + + +#ifdef assert +#warning Standard C Library assert macro usage detected. +#undef assert +#endif +#define assert(expr) debug_assert(expr) + + +#ifdef __cplusplus +} +#endif + +#endif /* P_DEBUG_H_ */ diff --git a/src/mesa/pipe/util/p_debug.c b/src/mesa/pipe/util/p_debug.c new file mode 100644 index 00000000000..faa093e57cc --- /dev/null +++ b/src/mesa/pipe/util/p_debug.c @@ -0,0 +1,70 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include + +#ifdef WIN32 +#include +#include +#else +#include +#include +#endif + +#include "pipe/p_debug.h" +#include "pipe/p_compiler.h" + + +void debug_printf(const char *format, ...) +{ + va_list ap; + va_start( ap, format ); +#ifdef WIN32 + EngDebugPrint("Gallium3D: ", (PCHAR)format, ap); +#else + vfprintf(stderr, format, ap); +#endif + va_end( ap ); +} + + +static INLINE debug_abort(void) +{ +#ifdef WIN32 + EngDebugBreak(); +#else + abort(); +#endif +} + + +void debug_assert_fail(const char *expr, const char *file, unsigned line) +{ + debug_printf("%s:%i: Assertion `%s' failed."); + debug_abort(); +} diff --git a/src/mesa/sources b/src/mesa/sources index e31d8cc4667..c0087f76e68 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -195,6 +195,7 @@ STATECACHE_SOURCES = \ pipe/cso_cache/cso_cache.c PIPEUTIL_SOURCES = \ + pipe/util/p_debug.c \ pipe/util/p_tile.c \ pipe/util/p_util.c -- cgit v1.2.3 From d432583d69fbd68bef79f1bf2ab0976ea67ed0bc Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Wed, 6 Feb 2008 14:36:50 +0900 Subject: gallium: Update scons instructions. Propagate user environment. --- SConstruct | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/SConstruct b/SConstruct index 01732b0c528..47f9b5389bb 100644 --- a/SConstruct +++ b/SConstruct @@ -11,7 +11,7 @@ import sys # # For example, invoke scons as # -# scons debug=1 dri=0 x86=1 +# scons debug=1 dri=0 machine=x86 # # to set configuration variables. Or you can write those options to a file # named config.py: @@ -19,7 +19,7 @@ import sys # # config.py # debug=1 # dri=0 -# x86=1 +# machine='x86' # # Invoke # @@ -35,7 +35,9 @@ opts.Add(BoolOption('dri', 'build dri drivers', False)) opts.Add(EnumOption('machine', 'use machine-specific assembly code', 'x86', allowed_values=('generic', 'x86', 'x86-64'))) -env = Environment(options = opts) +env = Environment( + options = opts, + ENV = os.environ) Help(opts.GenerateHelpText(env)) # for debugging -- cgit v1.2.3 From 7ff0b6782a8fc24c4d8df2535fa070b10c416dfa Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Wed, 6 Feb 2008 14:37:24 +0900 Subject: gallium: Add forgotten return type. --- src/mesa/pipe/util/p_debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/util/p_debug.c b/src/mesa/pipe/util/p_debug.c index faa093e57cc..9303c970cca 100644 --- a/src/mesa/pipe/util/p_debug.c +++ b/src/mesa/pipe/util/p_debug.c @@ -53,7 +53,7 @@ void debug_printf(const char *format, ...) } -static INLINE debug_abort(void) +static INLINE void debug_abort(void) { #ifdef WIN32 EngDebugBreak(); -- cgit v1.2.3 From 560416b263d10dae5d235b4cdaf44699181da74a Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Wed, 6 Feb 2008 14:37:49 +0900 Subject: gallium: Use p_debug.h instead of non-portable stdio.h/assert.h functions. --- src/mesa/pipe/draw/draw_prim.c | 18 +++---- src/mesa/pipe/draw/draw_vbuf.c | 63 ++++++++++++------------- src/mesa/pipe/draw/draw_vertex_fetch.c | 8 ++-- src/mesa/pipe/draw/draw_vertex_shader.c | 6 +-- src/mesa/pipe/draw/draw_vertex_shader_llvm.c | 4 +- src/mesa/pipe/draw/draw_vf.c | 4 +- src/mesa/pipe/draw/draw_vf_generic.c | 3 +- src/mesa/pipe/draw/draw_vf_sse.c | 6 +-- src/mesa/pipe/i915simple/i915_fpc_translate.c | 14 +++--- src/mesa/pipe/i915simple/i915_prim_vbuf.c | 3 +- src/mesa/pipe/i915simple/i915_state_derived.c | 2 +- src/mesa/pipe/i915simple/i915_state_emit.c | 2 +- src/mesa/pipe/i915simple/i915_state_immediate.c | 2 +- src/mesa/pipe/i915simple/i915_state_sampler.c | 2 +- src/mesa/pipe/i965simple/brw_cc.c | 2 +- src/mesa/pipe/i965simple/brw_curbe.c | 6 +-- src/mesa/pipe/i965simple/brw_eu_debug.c | 10 ++-- src/mesa/pipe/i965simple/brw_eu_emit.c | 4 +- src/mesa/pipe/i965simple/brw_sf.c | 8 ++-- src/mesa/pipe/i965simple/brw_sf_emit.c | 14 +++--- src/mesa/pipe/i965simple/brw_state.c | 2 +- src/mesa/pipe/i965simple/brw_state_cache.c | 6 +-- src/mesa/pipe/i965simple/brw_state_pool.c | 4 +- src/mesa/pipe/i965simple/brw_urb.c | 6 +-- src/mesa/pipe/i965simple/brw_vs_emit.c | 2 +- src/mesa/pipe/i965simple/brw_wm.c | 2 +- src/mesa/pipe/i965simple/brw_wm_glsl.c | 2 +- src/mesa/pipe/i965simple/brw_wm_sampler_state.c | 2 +- src/mesa/pipe/p_compiler.h | 3 +- src/mesa/pipe/p_format.h | 3 ++ src/mesa/pipe/p_util.h | 4 -- src/mesa/pipe/pipebuffer/pb_buffer.h | 4 +- src/mesa/pipe/pipebuffer/pb_buffer_fenced.c | 4 +- src/mesa/pipe/pipebuffer/pb_buffer_fenced.h | 2 +- src/mesa/pipe/pipebuffer/pb_buffer_malloc.c | 4 +- src/mesa/pipe/pipebuffer/pb_bufmgr_fenced.c | 4 +- src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c | 21 ++++----- src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c | 8 ++-- src/mesa/pipe/softpipe/sp_prim_setup.c | 24 +++++----- src/mesa/pipe/softpipe/sp_quad_fs.c | 10 ++-- src/mesa/pipe/softpipe/sp_tile_cache.c | 4 +- src/mesa/pipe/tgsi/exec/tgsi_exec.c | 6 +-- src/mesa/pipe/tgsi/exec/tgsi_sse2.c | 44 ++++++++--------- src/mesa/pipe/tgsi/util/tgsi_build.c | 1 + src/mesa/pipe/tgsi/util/tgsi_dump.c | 3 ++ src/mesa/pipe/tgsi/util/tgsi_parse.c | 1 + src/mesa/pipe/tgsi/util/tgsi_util.c | 1 + 47 files changed, 176 insertions(+), 182 deletions(-) diff --git a/src/mesa/pipe/draw/draw_prim.c b/src/mesa/pipe/draw/draw_prim.c index 58400213d76..51e2242719a 100644 --- a/src/mesa/pipe/draw/draw_prim.c +++ b/src/mesa/pipe/draw/draw_prim.c @@ -30,6 +30,8 @@ * Keith Whitwell */ +#include "pipe/p_debug.h" + #include "draw_private.h" #include "draw_context.h" @@ -60,8 +62,8 @@ static void draw_prim_queue_flush( struct draw_context *draw ) unsigned i; if (0) - fprintf(stdout,"Flushing with %d prims, %d verts\n", - draw->pq.queue_nr, draw->vs.queue_nr); + debug_printf("Flushing with %d prims, %d verts\n", + draw->pq.queue_nr, draw->vs.queue_nr); assert (draw->pq.queue_nr != 0); @@ -120,9 +122,9 @@ static void draw_prim_queue_flush( struct draw_context *draw ) void draw_do_flush( struct draw_context *draw, unsigned flags ) { if (0) - fprintf(stdout,"Flushing with %d verts, %d prims\n", - draw->vs.queue_nr, - draw->pq.queue_nr ); + debug_printf("Flushing with %d verts, %d prims\n", + draw->vs.queue_nr, + draw->pq.queue_nr ); if (flags >= DRAW_FLUSH_SHADER_QUEUE) { @@ -157,11 +159,11 @@ static struct prim_header *get_queued_prim( struct draw_context *draw, unsigned nr_verts ) { if (!draw_vertex_cache_check_space( draw, nr_verts )) { -// fprintf(stderr, "v"); +// debug_printf("v"); draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE ); } else if (draw->pq.queue_nr == PRIM_QUEUE_LENGTH) { -// fprintf(stderr, "p"); +// debug_printf("p"); draw_do_flush( draw, DRAW_FLUSH_PRIM_QUEUE ); } @@ -283,7 +285,7 @@ draw_prim( struct draw_context *draw, boolean unfilled = (draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL || draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL); -// _mesa_printf("%s (%d) %d/%d\n", __FUNCTION__, draw->prim, start, count ); +// debug_printf("%s (%d) %d/%d\n", __FUNCTION__, draw->prim, start, count ); switch (prim) { case PIPE_PRIM_POINTS: diff --git a/src/mesa/pipe/draw/draw_vbuf.c b/src/mesa/pipe/draw/draw_vbuf.c index ac03001d8f3..be96c8fdeb1 100644 --- a/src/mesa/pipe/draw/draw_vbuf.c +++ b/src/mesa/pipe/draw/draw_vbuf.c @@ -34,8 +34,7 @@ */ -#include - +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "draw_vbuf.h" @@ -125,55 +124,55 @@ dump_emitted_vertex(const struct vertex_info *vinfo, const uint8_t *data) j = vinfo->src_index[i]; switch (vinfo->emit[i]) { case EMIT_OMIT: - fprintf(stderr, "EMIT_OMIT:"); + debug_printf("EMIT_OMIT:"); break; case EMIT_ALL: assert(i == 0); assert(j == 0); - fprintf(stderr, "EMIT_ALL:\t"); + debug_printf("EMIT_ALL:\t"); for(k = 0; k < vinfo->size*4; ++k) - fprintf(stderr, "%02x ", *data++); + debug_printf("%02x ", *data++); break; case EMIT_1F: - fprintf(stderr, "EMIT_1F:\t"); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + debug_printf("EMIT_1F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); break; case EMIT_1F_PSIZE: - fprintf(stderr, "EMIT_1F_PSIZE:\t"); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + debug_printf("EMIT_1F_PSIZE:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); break; case EMIT_2F: - fprintf(stderr, "EMIT_2F:\t"); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + debug_printf("EMIT_2F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); break; case EMIT_3F: - fprintf(stderr, "EMIT_3F:\t"); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + debug_printf("EMIT_3F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); data += sizeof(float); break; case EMIT_4F: - fprintf(stderr, "EMIT_4F:\t"); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); - fprintf(stderr, "%f ", *(float *)data); data += sizeof(float); + debug_printf("EMIT_4F:\t"); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); + debug_printf("%f ", *(float *)data); data += sizeof(float); break; case EMIT_4UB: - fprintf(stderr, "EMIT_4UB:\t"); - fprintf(stderr, "%u ", *data++); - fprintf(stderr, "%u ", *data++); - fprintf(stderr, "%u ", *data++); - fprintf(stderr, "%u ", *data++); + debug_printf("EMIT_4UB:\t"); + debug_printf("%u ", *data++); + debug_printf("%u ", *data++); + debug_printf("%u ", *data++); + debug_printf("%u ", *data++); break; default: assert(0); } - fprintf(stderr, "\n"); + debug_printf("\n"); } - fprintf(stderr, "\n"); + debug_printf("\n"); } #endif @@ -190,7 +189,7 @@ emit_vertex( struct vbuf_stage *vbuf, struct vertex_header *vertex ) { #if 0 - fprintf(stderr, "emit vertex %d to %p\n", + debug_printf("emit vertex %d to %p\n", vbuf->nr_vertices, vbuf->vertex_ptr); #endif @@ -198,7 +197,7 @@ emit_vertex( struct vbuf_stage *vbuf, if(vertex->vertex_id < vbuf->nr_vertices) return; else - fprintf(stderr, "Bad vertex id 0x%04x (>= 0x%04x)\n", + debug_printf("Bad vertex id 0x%04x (>= 0x%04x)\n", vertex->vertex_id, vbuf->nr_vertices); return; } @@ -269,9 +268,9 @@ emit_vertex( struct vbuf_stage *vbuf, static float data[256]; draw_vf_emit_vertex(vbuf->vf, vertex, data); if(memcmp((uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size, data, vbuf->vertex_size)) { - fprintf(stderr, "With VF:\n"); + debug_printf("With VF:\n"); dump_emitted_vertex(vbuf->vinfo, (uint8_t *)data); - fprintf(stderr, "Without VF:\n"); + debug_printf("Without VF:\n"); dump_emitted_vertex(vbuf->vinfo, (uint8_t *)vbuf->vertex_ptr - vbuf->vertex_size); assert(0); } diff --git a/src/mesa/pipe/draw/draw_vertex_fetch.c b/src/mesa/pipe/draw/draw_vertex_fetch.c index b23f487e741..e13df04605c 100644 --- a/src/mesa/pipe/draw/draw_vertex_fetch.c +++ b/src/mesa/pipe/draw/draw_vertex_fetch.c @@ -162,7 +162,7 @@ static fetch_func get_fetch_func( enum pipe_format format ) { char tmp[80]; pf_sprint_name(tmp, format); - _mesa_printf("%s: %s\n", __FUNCTION__, tmp); + debug_printf("%s: %s\n", __FUNCTION__, tmp); } #endif @@ -332,7 +332,7 @@ static void fetch_xyz_rgb( struct draw_context *draw, assert(count <= 4); -// _mesa_printf("%s\n", __FUNCTION__); +// debug_printf("%s\n", __FUNCTION__); /* loop over vertex attributes (vertex shader inputs) */ @@ -421,7 +421,7 @@ static void generic_vertex_fetch( struct draw_context *draw, assert(count <= 4); -// _mesa_printf("%s %d\n", __FUNCTION__, count); +// debug_printf("%s %d\n", __FUNCTION__, count); /* loop over vertex attributes (vertex shader inputs) */ @@ -467,7 +467,7 @@ void draw_update_vertex_fetch( struct draw_context *draw ) { unsigned nr_attrs, i; -// _mesa_printf("%s\n", __FUNCTION__); +// debug_printf("%s\n", __FUNCTION__); /* this may happend during context init */ if (!draw->vertex_shader) diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index b851da845f0..e6590eafcc8 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -166,7 +166,7 @@ run_vertex_program(struct draw_context *draw, vOut[j]->data[0][3] = w; #if DBG_VS - printf("output[%d]win: %f %f %f %f\n", j, + debug_printf("output[%d]win: %f %f %f %f\n", j, vOut[j]->data[0][0], vOut[j]->data[0][1], vOut[j]->data[0][2], @@ -181,7 +181,7 @@ run_vertex_program(struct draw_context *draw, vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; #if DBG_VS - printf("output[%d][%d]: %f %f %f %f\n", j, slot, + debug_printf("output[%d][%d]: %f %f %f %f\n", j, slot, vOut[j]->data[slot][0], vOut[j]->data[slot][1], vOut[j]->data[slot][2], @@ -207,7 +207,7 @@ draw_vertex_shader_queue_flush(struct draw_context *draw) */ draw_update_vertex_fetch( draw ); -// fprintf(stderr, " q(%d) ", draw->vs.queue_nr ); +// debug_printf( " q(%d) ", draw->vs.queue_nr ); #ifdef MESA_LLVM if (draw->vertex_shader->llvm_prog) { draw_vertex_shader_queue_flush_llvm(draw); diff --git a/src/mesa/pipe/draw/draw_vertex_shader_llvm.c b/src/mesa/pipe/draw/draw_vertex_shader_llvm.c index 4228c4f3889..63551c993e7 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader_llvm.c +++ b/src/mesa/pipe/draw/draw_vertex_shader_llvm.c @@ -152,7 +152,7 @@ void draw_vertex_shader_queue_flush_llvm(struct draw_context *draw) z = vOut->clip[2] = dests[0][2]; w = vOut->clip[3] = dests[0][3]; #if DBG - printf("output %d: %f %f %f %f\n", 0, x, y, z, w); + debug_printf("output %d: %f %f %f %f\n", 0, x, y, z, w); #endif vOut->clipmask = compute_clipmask(vOut->clip, draw->plane, draw->nr_planes); @@ -179,7 +179,7 @@ void draw_vertex_shader_queue_flush_llvm(struct draw_context *draw) vOut->data[slot][3] = dests[slot][3]; #if DBG - printf("output %d: %f %f %f %f\n", slot, + debug_printf("output %d: %f %f %f %f\n", slot, vOut->data[slot][0], vOut->data[slot][1], vOut->data[slot][2], diff --git a/src/mesa/pipe/draw/draw_vf.c b/src/mesa/pipe/draw/draw_vf.c index 0da8e59ad68..f23d7fcec5c 100644 --- a/src/mesa/pipe/draw/draw_vf.c +++ b/src/mesa/pipe/draw/draw_vf.c @@ -168,7 +168,7 @@ draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, const unsigned format = map[i].format; if (format == DRAW_EMIT_PAD) { #if (DRAW_VF_DBG) - _mesa_printf("%d: pad %d, offset %d\n", i, + debug_printf("%d: pad %d, offset %d\n", i, map[i].offset, offset); #endif @@ -186,7 +186,7 @@ draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf, memcpy(vf->attr[j].data, &map[i].data, vf->attr[j].vertattrsize); #if (DRAW_VF_DBG) - _mesa_printf("%d: %s, offset %d\n", i, + debug_printf("%d: %s, offset %d\n", i, draw_vf_format_info[format].name, vf->attr[j].vertoffset); #endif diff --git a/src/mesa/pipe/draw/draw_vf_generic.c b/src/mesa/pipe/draw/draw_vf_generic.c index 7f5f56ef9cc..7a60a9db9c3 100644 --- a/src/mesa/pipe/draw/draw_vf_generic.c +++ b/src/mesa/pipe/draw/draw_vf_generic.c @@ -27,9 +27,8 @@ */ -#include - #include "pipe/p_compiler.h" +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "draw_vf.h" diff --git a/src/mesa/pipe/draw/draw_vf_sse.c b/src/mesa/pipe/draw/draw_vf_sse.c index 1389e6cfb9c..1ad2ae756dd 100644 --- a/src/mesa/pipe/draw/draw_vf_sse.c +++ b/src/mesa/pipe/draw/draw_vf_sse.c @@ -453,7 +453,7 @@ static boolean build_vertex_emit( struct x86_program *p ) update_src_ptr(p, srcECX, vfESI, a); } else { - fprintf(stderr, "Can't emit 1ub %x %x %d\n", + debug_printf("Can't emit 1ub %x %x %d\n", a->vertoffset, a[-1].vertoffset, a[-1].vertattrsize ); return FALSE; } @@ -499,7 +499,7 @@ static boolean build_vertex_emit( struct x86_program *p ) j++; /* NOTE: two attrs consumed */ } else { - fprintf(stderr, "Can't emit 3ub\n"); + debug_printf("Can't emit 3ub\n"); } return FALSE; /* add this later */ break; @@ -532,7 +532,7 @@ static boolean build_vertex_emit( struct x86_program *p ) update_src_ptr(p, srcECX, vfESI, a); break; default: - fprintf(stderr, "unknown a[%d].format %d\n", j, a->format); + debug_printf("unknown a[%d].format %d\n", j, a->format); return FALSE; /* catch any new opcodes */ } diff --git a/src/mesa/pipe/i915simple/i915_fpc_translate.c b/src/mesa/pipe/i915simple/i915_fpc_translate.c index 0185512aeb8..868f0c7e046 100644 --- a/src/mesa/pipe/i915simple/i915_fpc_translate.c +++ b/src/mesa/pipe/i915simple/i915_fpc_translate.c @@ -100,7 +100,7 @@ negate(int reg, int x, int y, int z, int w) static void i915_use_passthrough_shader(struct i915_context *i915) { - fprintf(stderr, "**** Using i915 pass-through fragment shader\n"); + debug_printf("**** Using i915 pass-through fragment shader\n"); i915->current.program = (uint *) MALLOC(sizeof(passthrough)); if (i915->current.program) { @@ -119,12 +119,12 @@ i915_program_error(struct i915_fp_compile *p, const char *msg, ...) va_list args; char buffer[1024]; - fprintf(stderr, "i915_program_error: "); + debug_printf("i915_program_error: "); va_start( args, msg ); vsprintf( buffer, msg, args ); va_end( args ); - fprintf(stderr, buffer); - fprintf(stderr, "\n"); + debug_printf(buffer); + debug_printf("\n"); p->error = 1; } @@ -169,7 +169,7 @@ src_vector(struct i915_fp_compile *p, switch (sem_name) { case TGSI_SEMANTIC_POSITION: - fprintf(stderr, "SKIP SEM POS\n"); + debug_printf("SKIP SEM POS\n"); /* assert(p->wpos_tex != -1); src = i915_emit_decl(p, REG_TYPE_T, p->wpos_tex, D0_CHANNEL_ALL); @@ -913,7 +913,7 @@ i915_translate_instructions(struct i915_fp_compile *p, ind = parse.FullToken.FullDeclaration.u.DeclarationRange.First; sem = parse.FullToken.FullDeclaration.Semantic.SemanticName; semi = parse.FullToken.FullDeclaration.Semantic.SemanticIndex; - /*printf("FS Input DECL [%u] sem %u\n", ind, sem);*/ + /*debug_printf("FS Input DECL [%u] sem %u\n", ind, sem);*/ p->input_semantic_name[ind] = sem; p->input_semantic_index[ind] = semi; } @@ -924,7 +924,7 @@ i915_translate_instructions(struct i915_fp_compile *p, ind = parse.FullToken.FullDeclaration.u.DeclarationRange.First; sem = parse.FullToken.FullDeclaration.Semantic.SemanticName; semi = parse.FullToken.FullDeclaration.Semantic.SemanticIndex; - /*printf("FS Output DECL [%u] sem %u\n", ind, sem);*/ + /*debug_printf("FS Output DECL [%u] sem %u\n", ind, sem);*/ p->output_semantic_name[ind] = sem; p->output_semantic_index[ind] = semi; } diff --git a/src/mesa/pipe/i915simple/i915_prim_vbuf.c b/src/mesa/pipe/i915simple/i915_prim_vbuf.c index 39154b2488b..e069773fd4e 100644 --- a/src/mesa/pipe/i915simple/i915_prim_vbuf.c +++ b/src/mesa/pipe/i915simple/i915_prim_vbuf.c @@ -38,9 +38,8 @@ */ -#include - #include "pipe/draw/draw_vbuf.h" +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" diff --git a/src/mesa/pipe/i915simple/i915_state_derived.c b/src/mesa/pipe/i915simple/i915_state_derived.c index 62741e30f80..653983e4a99 100644 --- a/src/mesa/pipe/i915simple/i915_state_derived.c +++ b/src/mesa/pipe/i915simple/i915_state_derived.c @@ -87,7 +87,7 @@ static void calculate_vertex_layout( struct i915_context *i915 ) } break; case TGSI_SEMANTIC_FOG: - fprintf(stderr, "i915 fogcoord not implemented yet\n"); + debug_printf("i915 fogcoord not implemented yet\n"); draw_emit_vertex_attr(&vinfo, EMIT_1F, INTERP_PERSPECTIVE, src++); break; default: diff --git a/src/mesa/pipe/i915simple/i915_state_emit.c b/src/mesa/pipe/i915simple/i915_state_emit.c index 657f5238938..3339287f498 100644 --- a/src/mesa/pipe/i915simple/i915_state_emit.c +++ b/src/mesa/pipe/i915simple/i915_state_emit.c @@ -107,7 +107,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) ) * 3/2; /* plus 50% margin */ #if 0 - fprintf (stderr, "i915_emit_hardware_state: %d dwords, %d relocs\n", dwords, relocs); + debug_printf("i915_emit_hardware_state: %d dwords, %d relocs\n", dwords, relocs); #endif if(!BEGIN_BATCH(dwords, relocs)) { diff --git a/src/mesa/pipe/i915simple/i915_state_immediate.c b/src/mesa/pipe/i915simple/i915_state_immediate.c index 752d25f2331..07031fc6c5b 100644 --- a/src/mesa/pipe/i915simple/i915_state_immediate.c +++ b/src/mesa/pipe/i915simple/i915_state_immediate.c @@ -97,7 +97,7 @@ static void upload_S2S4(struct i915_context *i915) LIS2 = i915->current.vertex_info.hwfmt[1]; LIS4 = i915->current.vertex_info.hwfmt[0]; /* - printf("LIS2: 0x%x LIS4: 0x%x\n", LIS2, LIS4); + debug_printf("LIS2: 0x%x LIS4: 0x%x\n", LIS2, LIS4); */ assert(LIS4); /* should never be zero? */ } diff --git a/src/mesa/pipe/i915simple/i915_state_sampler.c b/src/mesa/pipe/i915simple/i915_state_sampler.c index 59408b6ba06..0dbbc5241d5 100644 --- a/src/mesa/pipe/i915simple/i915_state_sampler.c +++ b/src/mesa/pipe/i915simple/i915_state_sampler.c @@ -169,7 +169,7 @@ translate_texture_format(enum pipe_format pipeFormat) case PIPE_FORMAT_S8Z24_UNORM: return (MAPSURF_32BIT | MT_32BIT_xL824); default: - fprintf(stderr, "i915: translate_texture_format() bad image format %x\n", + debug_printf("i915: translate_texture_format() bad image format %x\n", pipeFormat); assert(0); return 0; diff --git a/src/mesa/pipe/i965simple/brw_cc.c b/src/mesa/pipe/i965simple/brw_cc.c index dcee7318958..337e4f95f69 100644 --- a/src/mesa/pipe/i965simple/brw_cc.c +++ b/src/mesa/pipe/i965simple/brw_cc.c @@ -58,7 +58,7 @@ static int brw_translate_compare_func(int func) return BRW_COMPAREFUNCTION_ALWAYS; } - fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func); + debug_printf("Unknown value in %s: %x\n", __FUNCTION__, func); return BRW_COMPAREFUNCTION_ALWAYS; } diff --git a/src/mesa/pipe/i965simple/brw_curbe.c b/src/mesa/pipe/i965simple/brw_curbe.c index 2733eb4e75c..52bbd525c16 100644 --- a/src/mesa/pipe/i965simple/brw_curbe.c +++ b/src/mesa/pipe/i965simple/brw_curbe.c @@ -273,10 +273,10 @@ static void upload_constant_buffer(struct brw_context *brw) if (1) { for (i = 0; i < sz; i+=4) - _mesa_printf("curbe %d.%d: %f %f %f %f\n", i/8, i&4, + debug_printf("curbe %d.%d: %f %f %f %f\n", i/8, i&4, buf[i+0], buf[i+1], buf[i+2], buf[i+3]); - _mesa_printf("last_buf %p buf %p sz %d/%d cmp %d\n", + debug_printf("last_buf %p buf %p sz %d/%d cmp %d\n", brw->curbe.last_buf, buf, bufsz, brw->curbe.last_bufsz, brw->curbe.last_buf ? memcmp(buf, brw->curbe.last_buf, bufsz) : -1); @@ -299,7 +299,7 @@ static void upload_constant_buffer(struct brw_context *brw) bufsz, 1 << 6, &brw->curbe.gs_offset)) { - _mesa_printf("out of GS memory for curbe\n"); + debug_printf("out of GS memory for curbe\n"); assert(0); return; } diff --git a/src/mesa/pipe/i965simple/brw_eu_debug.c b/src/mesa/pipe/i965simple/brw_eu_debug.c index be692f65022..4a94ddefa6a 100644 --- a/src/mesa/pipe/i965simple/brw_eu_debug.c +++ b/src/mesa/pipe/i965simple/brw_eu_debug.c @@ -30,6 +30,8 @@ */ +#include "pipe/p_debug.h" + #include "brw_eu.h" void brw_print_reg( struct brw_reg hwreg ) @@ -52,7 +54,7 @@ void brw_print_reg( struct brw_reg hwreg ) "f" }; - _mesa_printf("%s%s", + debug_printf("%s%s", hwreg.abs ? "abs/" : "", hwreg.negate ? "-" : ""); @@ -63,17 +65,17 @@ void brw_print_reg( struct brw_reg hwreg ) hwreg.width == BRW_WIDTH_8 && hwreg.hstride == BRW_HORIZONTAL_STRIDE_1 && hwreg.type == BRW_REGISTER_TYPE_F) { - _mesa_printf("vec%d", hwreg.nr); + debug_printf("vec%d", hwreg.nr); } else if (hwreg.file == BRW_GENERAL_REGISTER_FILE && hwreg.vstride == BRW_VERTICAL_STRIDE_0 && hwreg.width == BRW_WIDTH_1 && hwreg.hstride == BRW_HORIZONTAL_STRIDE_0 && hwreg.type == BRW_REGISTER_TYPE_F) { - _mesa_printf("scl%d.%d", hwreg.nr, hwreg.subnr / 4); + debug_printf("scl%d.%d", hwreg.nr, hwreg.subnr / 4); } else { - _mesa_printf("%s%d.%d<%d;%d,%d>:%s", + debug_printf("%s%d.%d<%d;%d,%d>:%s", file[hwreg.file], hwreg.nr, hwreg.subnr / type_sz(hwreg.type), diff --git a/src/mesa/pipe/i965simple/brw_eu_emit.c b/src/mesa/pipe/i965simple/brw_eu_emit.c index 2423536dd19..400a80b6fba 100644 --- a/src/mesa/pipe/i965simple/brw_eu_emit.c +++ b/src/mesa/pipe/i965simple/brw_eu_emit.c @@ -953,7 +953,7 @@ void brw_SAMPLE(struct brw_compile *p, boolean need_stall = 0; if(writemask == 0) { -/* _mesa_printf("%s: zero writemask??\n", __FUNCTION__); */ +/* debug_printf("%s: zero writemask??\n", __FUNCTION__); */ return; } @@ -985,7 +985,7 @@ void brw_SAMPLE(struct brw_compile *p, if (newmask != writemask) { need_stall = 1; -/* _mesa_printf("need stall %x %x\n", newmask , writemask); */ +/* debug_printf("need stall %x %x\n", newmask , writemask); */ } else { struct brw_reg m1 = brw_message_reg(msg_reg_nr); diff --git a/src/mesa/pipe/i965simple/brw_sf.c b/src/mesa/pipe/i965simple/brw_sf.c index b89b2e40875..7c83b81c858 100644 --- a/src/mesa/pipe/i965simple/brw_sf.c +++ b/src/mesa/pipe/i965simple/brw_sf.c @@ -175,7 +175,7 @@ static void upload_sf_prog( struct brw_context *brw ) //int semantic = parse.FullToken.FullDeclaration.Semantic.SemanticName; //int semantic_index = parse.FullToken.FullDeclaration.Semantic.SemanticIndex; - fprintf(stderr, "fs input %d..%d interp mode %d\n", first, last, interp_mode); + debug_printf("fs input %d..%d interp mode %d\n", first, last, interp_mode); switch (interp_mode) { case TGSI_INTERPOLATE_CONSTANT: @@ -213,9 +213,9 @@ static void upload_sf_prog( struct brw_context *brw ) key.linear_mask |= 1; key.const_mask <<= 1; - fprintf(stderr, "key.persp_mask: %x\n", key.persp_mask); - fprintf(stderr, "key.linear_mask: %x\n", key.linear_mask); - fprintf(stderr, "key.const_mask: %x\n", key.const_mask); + debug_printf("key.persp_mask: %x\n", key.persp_mask); + debug_printf("key.linear_mask: %x\n", key.linear_mask); + debug_printf("key.const_mask: %x\n", key.const_mask); // key.do_point_sprite = brw->attribs.Point->PointSprite; diff --git a/src/mesa/pipe/i965simple/brw_sf_emit.c b/src/mesa/pipe/i965simple/brw_sf_emit.c index 6ff5254ff73..78d6fa5e9e5 100644 --- a/src/mesa/pipe/i965simple/brw_sf_emit.c +++ b/src/mesa/pipe/i965simple/brw_sf_emit.c @@ -137,8 +137,8 @@ static boolean calculate_masks( struct brw_sf_compile *c, unsigned persp_mask = c->key.persp_mask; unsigned linear_mask = c->key.linear_mask; - fprintf(stderr, "persp_mask: %x\n", persp_mask); - fprintf(stderr, "linear_mask: %x\n", linear_mask); + debug_printf("persp_mask: %x\n", persp_mask); + debug_printf("linear_mask: %x\n", linear_mask); *pc_persp = 0; *pc_linear = 0; @@ -162,9 +162,9 @@ static boolean calculate_masks( struct brw_sf_compile *c, *pc_linear |= 0xf0; } - fprintf(stderr, "pc: %x\n", *pc); - fprintf(stderr, "pc_persp: %x\n", *pc_persp); - fprintf(stderr, "pc_linear: %x\n", *pc_linear); + debug_printf("pc: %x\n", *pc); + debug_printf("pc_persp: %x\n", *pc_persp); + debug_printf("pc_linear: %x\n", *pc_linear); return is_last_attr; @@ -177,7 +177,7 @@ void brw_emit_tri_setup( struct brw_sf_compile *c ) struct brw_compile *p = &c->func; unsigned i; - fprintf(stderr, "%s START ==============\n", __FUNCTION__); + debug_printf("%s START ==============\n", __FUNCTION__); c->nr_verts = 3; alloc_regs(c); @@ -250,7 +250,7 @@ void brw_emit_tri_setup( struct brw_sf_compile *c ) } } - fprintf(stderr, "%s DONE ==============\n", __FUNCTION__); + debug_printf("%s DONE ==============\n", __FUNCTION__); } diff --git a/src/mesa/pipe/i965simple/brw_state.c b/src/mesa/pipe/i965simple/brw_state.c index daf14ff4ff2..95dfce88e4a 100644 --- a/src/mesa/pipe/i965simple/brw_state.c +++ b/src/mesa/pipe/i965simple/brw_state.c @@ -225,7 +225,7 @@ static void brw_bind_vs_state(struct pipe_context *pipe, void *vs) brw->attribs.VertexProgram = (struct brw_vertex_program *)vs; brw->state.dirty.brw |= BRW_NEW_VS; - printf("YYYYYYYYYYYYY BINDING VERTEX SHADER\n"); + debug_printf("YYYYYYYYYYYYY BINDING VERTEX SHADER\n"); } static void brw_delete_vs_state(struct pipe_context *pipe, void *shader) diff --git a/src/mesa/pipe/i965simple/brw_state_cache.c b/src/mesa/pipe/i965simple/brw_state_cache.c index c5738733f4e..b3a5124461d 100644 --- a/src/mesa/pipe/i965simple/brw_state_cache.c +++ b/src/mesa/pipe/i965simple/brw_state_cache.c @@ -149,7 +149,7 @@ unsigned brw_upload_cache( struct brw_cache *cache, if (!brw_pool_alloc(cache->pool, data_size, 1 << 6, &offset)) { /* Should not be possible: */ - printf("brw_pool_alloc failed\n"); + debug_printf("brw_pool_alloc failed\n"); exit(1); } @@ -177,7 +177,7 @@ unsigned brw_upload_cache( struct brw_cache *cache, } if (BRW_DEBUG & DEBUG_STATE) - printf("upload %s: %d bytes to pool buffer %p offset %x\n", + debug_printf("upload %s: %d bytes to pool buffer %p offset %x\n", cache->name, data_size, (void*)cache->pool->buffer, @@ -416,7 +416,7 @@ void brw_clear_all_caches( struct brw_context *brw ) int i; if (BRW_DEBUG & DEBUG_STATE) - fprintf(stderr, "%s\n", __FUNCTION__); + debug_printf("%s\n", __FUNCTION__); for (i = 0; i < BRW_MAX_CACHE; i++) clear_cache(&brw->cache[i]); diff --git a/src/mesa/pipe/i965simple/brw_state_pool.c b/src/mesa/pipe/i965simple/brw_state_pool.c index 7c67f0ee25d..f3174bfe0ae 100644 --- a/src/mesa/pipe/i965simple/brw_state_pool.c +++ b/src/mesa/pipe/i965simple/brw_state_pool.c @@ -58,7 +58,7 @@ boolean brw_pool_alloc( struct brw_mem_pool *pool, size = align(size, 4); if (pool->offset + fixup + size >= pool->size) { - printf("%s failed\n", __FUNCTION__); + debug_printf("%s failed\n", __FUNCTION__); assert(0); exit(0); } @@ -74,7 +74,7 @@ static void brw_invalidate_pool( struct brw_mem_pool *pool ) { if (BRW_DEBUG & DEBUG_STATE) - printf("\n\n\n %s \n\n\n", __FUNCTION__); + debug_printf("\n\n\n %s \n\n\n", __FUNCTION__); pool->offset = 0; diff --git a/src/mesa/pipe/i965simple/brw_urb.c b/src/mesa/pipe/i965simple/brw_urb.c index b284526aa6e..101a4367b90 100644 --- a/src/mesa/pipe/i965simple/brw_urb.c +++ b/src/mesa/pipe/i965simple/brw_urb.c @@ -120,18 +120,18 @@ static void recalculate_urb_fence( struct brw_context *brw ) * entries and the values for minimum nr of entries * provided above. */ - fprintf(stderr, "couldn't calculate URB layout!\n"); + debug_printf("couldn't calculate URB layout!\n"); exit(1); } if (BRW_DEBUG & (DEBUG_URB|DEBUG_FALLBACKS)) - printf("URB CONSTRAINED\n"); + debug_printf("URB CONSTRAINED\n"); } else brw->urb.constrained = 0; if (BRW_DEBUG & DEBUG_URB) - printf("URB fence: %d ..VS.. %d ..GS.. %d ..CLP.. %d ..SF.. %d ..CS.. %d\n", + debug_printf("URB fence: %d ..VS.. %d ..GS.. %d ..CLP.. %d ..SF.. %d ..CS.. %d\n", brw->urb.vs_start, brw->urb.gs_start, brw->urb.clip_start, diff --git a/src/mesa/pipe/i965simple/brw_vs_emit.c b/src/mesa/pipe/i965simple/brw_vs_emit.c index b32c233dd22..98915ba1016 100644 --- a/src/mesa/pipe/i965simple/brw_vs_emit.c +++ b/src/mesa/pipe/i965simple/brw_vs_emit.c @@ -1228,7 +1228,7 @@ static void process_instruction(struct brw_vs_compile *c, case TGSI_OPCODE_ENDSUB: break; default: - printf("Unsupport opcode %d in vertex shader\n", inst->Instruction.Opcode); + debug_printf("Unsupport opcode %d in vertex shader\n", inst->Instruction.Opcode); break; } diff --git a/src/mesa/pipe/i965simple/brw_wm.c b/src/mesa/pipe/i965simple/brw_wm.c index 0ee0fbed511..539b1707444 100644 --- a/src/mesa/pipe/i965simple/brw_wm.c +++ b/src/mesa/pipe/i965simple/brw_wm.c @@ -57,7 +57,7 @@ static void do_wm_prog( struct brw_context *brw, c->pixel_w = brw_null_reg(); - fprintf(stderr, "XXXXXXXX FP\n"); + debug_printf("XXXXXXXX FP\n"); brw_wm_glsl_emit(c); diff --git a/src/mesa/pipe/i965simple/brw_wm_glsl.c b/src/mesa/pipe/i965simple/brw_wm_glsl.c index f4b5c13c069..d95645d1085 100644 --- a/src/mesa/pipe/i965simple/brw_wm_glsl.c +++ b/src/mesa/pipe/i965simple/brw_wm_glsl.c @@ -982,7 +982,7 @@ static void brw_wm_emit_instruction( struct brw_wm_compile *c, break; default: - _mesa_printf("unsupported IR in fragment shader %d\n", + debug_printf("unsupported IR in fragment shader %d\n", inst->Instruction.Opcode); } #if 0 diff --git a/src/mesa/pipe/i965simple/brw_wm_sampler_state.c b/src/mesa/pipe/i965simple/brw_wm_sampler_state.c index cfb430eb09c..de42ffc5b1e 100644 --- a/src/mesa/pipe/i965simple/brw_wm_sampler_state.c +++ b/src/mesa/pipe/i965simple/brw_wm_sampler_state.c @@ -71,7 +71,7 @@ static int intel_translate_shadow_compare_func(unsigned func) return COMPAREFUNC_NEVER; } - fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func); + debug_printf("Unknown value in %s: %x\n", __FUNCTION__, func); return COMPAREFUNC_NEVER; } diff --git a/src/mesa/pipe/p_compiler.h b/src/mesa/pipe/p_compiler.h index e939d9cd9b9..30cd729c564 100644 --- a/src/mesa/pipe/p_compiler.h +++ b/src/mesa/pipe/p_compiler.h @@ -28,10 +28,9 @@ #ifndef P_COMPILER_H #define P_COMPILER_H -#include + #include #include -#include #if defined(_WIN32) && !defined(__WIN32__) diff --git a/src/mesa/pipe/p_format.h b/src/mesa/pipe/p_format.h index 9f60cdbb04d..c9ad3243158 100644 --- a/src/mesa/pipe/p_format.h +++ b/src/mesa/pipe/p_format.h @@ -28,7 +28,10 @@ #ifndef PIPE_FORMAT_H #define PIPE_FORMAT_H +#include // for sprintf + #include "p_compiler.h" +#include "p_debug.h" /** * The PIPE_FORMAT is a 32-bit wide bitfield that encodes all the information diff --git a/src/mesa/pipe/p_util.h b/src/mesa/pipe/p_util.h index 059528787d6..4780ed78184 100644 --- a/src/mesa/pipe/p_util.h +++ b/src/mesa/pipe/p_util.h @@ -381,10 +381,6 @@ static INLINE int align(int value, int alignment) return (value + alignment - 1) & ~(alignment - 1); } -/* Convenient... - */ -extern void _mesa_printf(const char *str, ...); - /* util/p_util.c */ diff --git a/src/mesa/pipe/pipebuffer/pb_buffer.h b/src/mesa/pipe/pipebuffer/pb_buffer.h index 17551b3b50e..97beb5f72a9 100644 --- a/src/mesa/pipe/pipebuffer/pb_buffer.h +++ b/src/mesa/pipe/pipebuffer/pb_buffer.h @@ -44,10 +44,8 @@ #define PB_BUFFER_H_ -#include -#include - #include "pipe/p_compiler.h" +#include "pipe/p_debug.h" #include "pipe/p_state.h" #include "pipe/p_inlines.h" diff --git a/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c b/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c index 4cf4222db9d..f4fc3f6d714 100644 --- a/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c +++ b/src/mesa/pipe/pipebuffer/pb_buffer_fenced.c @@ -34,12 +34,10 @@ */ -#include -#include - #include "linked_list.h" #include "p_compiler.h" +#include "p_debug.h" #include "p_winsys.h" #include "p_thread.h" #include "p_util.h" diff --git a/src/mesa/pipe/pipebuffer/pb_buffer_fenced.h b/src/mesa/pipe/pipebuffer/pb_buffer_fenced.h index 09082a53909..c40b9c75e18 100644 --- a/src/mesa/pipe/pipebuffer/pb_buffer_fenced.h +++ b/src/mesa/pipe/pipebuffer/pb_buffer_fenced.h @@ -51,7 +51,7 @@ #define PB_BUFFER_FENCED_H_ -#include +#include "pipe/p_debug.h" struct pipe_winsys; diff --git a/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c b/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c index 2151f1d691e..c1b7759874c 100644 --- a/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c +++ b/src/mesa/pipe/pipebuffer/pb_buffer_malloc.c @@ -34,9 +34,7 @@ */ -#include -#include - +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pb_buffer.h" diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_fenced.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_fenced.c index 3b341c64c28..c535d3276c9 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_fenced.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_fenced.c @@ -34,9 +34,7 @@ */ -#include -#include - +#include "p_debug.h" #include "p_util.h" #include "pb_buffer.h" diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c index b6af7cdedc2..8b1b51c0e28 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_mm.c @@ -34,11 +34,10 @@ */ -#include - #include "linked_list.h" #include "p_defines.h" +#include "p_debug.h" #include "p_thread.h" #include "p_util.h" #include "pb_buffer.h" @@ -69,28 +68,28 @@ struct mem_block static void mmDumpMemInfo(const struct mem_block *heap) { - fprintf(stderr, "Memory heap %p:\n", (void *)heap); + debug_printf("Memory heap %p:\n", (void *)heap); if (heap == 0) { - fprintf(stderr, " heap == 0\n"); + debug_printf(" heap == 0\n"); } else { const struct mem_block *p; for(p = heap->next; p != heap; p = p->next) { - fprintf(stderr, " Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size, + debug_printf(" Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size, p->free ? 'F':'.', p->reserved ? 'R':'.'); } - fprintf(stderr, "\nFree list:\n"); + debug_printf("\nFree list:\n"); for(p = heap->next_free; p != heap; p = p->next_free) { - fprintf(stderr, " FREE Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size, + debug_printf(" FREE Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size, p->free ? 'F':'.', p->reserved ? 'R':'.'); } } - fprintf(stderr, "End of memory blocks\n"); + debug_printf("End of memory blocks\n"); } #endif @@ -308,11 +307,11 @@ mmFreeMem(struct mem_block *b) return 0; if (b->free) { - fprintf(stderr, "block already free\n"); + debug_printf("block already free\n"); return -1; } if (b->reserved) { - fprintf(stderr, "block is reserved\n"); + debug_printf("block is reserved\n"); return -1; } @@ -479,7 +478,7 @@ mm_bufmgr_create_buffer(struct pb_manager *mgr, mm_buf->block = mmAllocMem(mm->heap, size, mm->align2, 0); if(!mm_buf->block) { - fprintf(stderr, "warning: heap full\n"); + debug_printf("warning: heap full\n"); #if 0 mmDumpMemInfo(mm->heap); #endif diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c index f80c7e34c07..bcd4b3e2571 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c @@ -35,12 +35,10 @@ */ -#include -#include - #include "linked_list.h" #include "p_compiler.h" +#include "p_debug.h" #include "p_thread.h" #include "p_defines.h" #include "p_util.h" @@ -178,7 +176,7 @@ pool_bufmgr_create_buffer(struct pb_manager *mgr, if (pool->numFree == 0) { _glthread_UNLOCK_MUTEX(pool->mutex); - fprintf(stderr, "warning: out of fixed size buffer objects\n"); + debug_printf("warning: out of fixed size buffer objects\n"); return NULL; } @@ -186,7 +184,7 @@ pool_bufmgr_create_buffer(struct pb_manager *mgr, if (item == &pool->free) { _glthread_UNLOCK_MUTEX(pool->mutex); - fprintf(stderr, "error: fixed size buffer pool corruption\n"); + debug_printf("error: fixed size buffer pool corruption\n"); return NULL; } diff --git a/src/mesa/pipe/softpipe/sp_prim_setup.c b/src/mesa/pipe/softpipe/sp_prim_setup.c index b17801d13da..7478b2336b6 100644 --- a/src/mesa/pipe/softpipe/sp_prim_setup.c +++ b/src/mesa/pipe/softpipe/sp_prim_setup.c @@ -251,9 +251,9 @@ static void print_vertex(const struct setup_stage *setup, const struct vertex_header *v) { int i; - fprintf(stderr, "Vertex: (%p)\n", v); + debug_printf("Vertex: (%p)\n", v); for (i = 0; i < setup->quad.nr_attrs; i++) { - fprintf(stderr, " %d: %f %f %f %f\n", i, + debug_printf(" %d: %f %f %f %f\n", i, v->data[i][0], v->data[i][1], v->data[i][2], v->data[i][3]); } } @@ -267,7 +267,7 @@ static boolean setup_sort_vertices( struct setup_stage *setup, const struct vertex_header *v2 = prim->v[2]; #if DEBUG_VERTS - fprintf(stderr, "Triangle:\n"); + debug_printf("Triangle:\n"); print_vertex(setup, v0); print_vertex(setup, v1); print_vertex(setup, v2); @@ -345,7 +345,7 @@ static boolean setup_sort_vertices( struct setup_stage *setup, setup->oneoverarea = 1.0f / area; /* - _mesa_printf("%s one-over-area %f area %f det %f\n", + debug_printf("%s one-over-area %f area %f det %f\n", __FUNCTION__, setup->oneoverarea, area, prim->det ); */ } @@ -419,7 +419,7 @@ static void tri_linear_coeff( struct setup_stage *setup, dady * (setup->vmin->data[0][1] - 0.5f))); /* - _mesa_printf("attr[%d].%c: %f dx:%f dy:%f\n", + debug_printf("attr[%d].%c: %f dx:%f dy:%f\n", slot, "xyzw"[i], setup->coef[slot].a0[i], setup->coef[slot].dadx[i], @@ -453,10 +453,10 @@ static void tri_persp_coeff( struct setup_stage *setup, float dady = b * setup->oneoverarea; /* - printf("tri persp %d,%d: %f %f %f\n", vertSlot, i, - setup->vmin->data[vertSlot][i], - setup->vmid->data[vertSlot][i], - setup->vmax->data[vertSlot][i] + debug_printf("tri persp %d,%d: %f %f %f\n", vertSlot, i, + setup->vmin->data[vertSlot][i], + setup->vmid->data[vertSlot][i], + setup->vmax->data[vertSlot][i] ); */ assert(i <= 3); @@ -619,7 +619,7 @@ static void subtriangle( struct setup_stage *setup, finish_y -= sy; /* - _mesa_printf("%s %d %d\n", __FUNCTION__, start_y, finish_y); + debug_printf("%s %d %d\n", __FUNCTION__, start_y, finish_y); */ for (y = start_y; y < finish_y; y++) { @@ -671,7 +671,7 @@ static void setup_tri( struct draw_stage *stage, struct setup_stage *setup = setup_stage( stage ); /* - _mesa_printf("%s\n", __FUNCTION__ ); + debug_printf("%s\n", __FUNCTION__ ); */ setup_sort_vertices( setup, prim ); @@ -1124,7 +1124,7 @@ setup_point(struct draw_stage *stage, struct prim_header *prim) int ix, iy; /* - printf("(%f, %f) -> X:%d..%d Y:%d..%d\n", x, y, xmin, xmax,ymin,ymax); + debug_printf("(%f, %f) -> X:%d..%d Y:%d..%d\n", x, y, xmin, xmax,ymin,ymax); */ for (iy = iymin; iy <= iymax; iy += 2) { uint rowMask = 0xf; diff --git a/src/mesa/pipe/softpipe/sp_quad_fs.c b/src/mesa/pipe/softpipe/sp_quad_fs.c index 90691c6065d..b5d7dfca1ca 100644 --- a/src/mesa/pipe/softpipe/sp_quad_fs.c +++ b/src/mesa/pipe/softpipe/sp_quad_fs.c @@ -223,13 +223,13 @@ shade_quad_llvm(struct quad_stage *qs, inputs[2][0][1] = fy + 1.0f; inputs[3][0][1] = fy + 1.0f; #if DLLVM - printf("MASK = %d\n", quad->mask); + debug_printf("MASK = %d\n", quad->mask); #endif gallivm_prog_inputs_interpolate(llvm, inputs, quad->coef); #if DLLVM for (int i = 0; i < 4; ++i) { for (int j = 0; j < 2; ++j) { - printf("IN(%d,%d) [%f %f %f %f]\n", i, j, + debug_printf("IN(%d,%d) [%f %f %f %f]\n", i, j, inputs[i][j][0], inputs[i][j][1], inputs[i][j][2], inputs[i][j][3]); } } @@ -240,7 +240,7 @@ shade_quad_llvm(struct quad_stage *qs, softpipe->mapped_constants[PIPE_SHADER_FRAGMENT], qss->samplers); #if DLLVM - printf("OUT LLVM = 1[%f %f %f %f], 2[%f %f %f %f]\n", + debug_printf("OUT LLVM = 1[%f %f %f %f], 2[%f %f %f %f]\n", dests[0][0][0], dests[0][0][1], dests[0][0][2], dests[0][0][3], dests[0][1][0], dests[0][1][1], dests[0][1][2], dests[0][1][3]); #endif @@ -260,7 +260,7 @@ shade_quad_llvm(struct quad_stage *qs, } #if DLLVM for (int i = 0; i < QUAD_SIZE; ++i) { - printf("QLLVM%d(%d) [%f, %f, %f, %f]\n", i, qss->colorOutSlot, + debug_printf("QLLVM%d(%d) [%f, %f, %f, %f]\n", i, qss->colorOutSlot, quad->outputs.color[0][i], quad->outputs.color[1][i], quad->outputs.color[2][i], @@ -284,7 +284,7 @@ shade_quad_llvm(struct quad_stage *qs, } } #if DLLVM - printf("D [%f, %f, %f, %f] mask = %d\n", + debug_printf("D [%f, %f, %f, %f] mask = %d\n", quad->outputs.depth[0], quad->outputs.depth[1], quad->outputs.depth[2], diff --git a/src/mesa/pipe/softpipe/sp_tile_cache.c b/src/mesa/pipe/softpipe/sp_tile_cache.c index ccf367a5e4f..1597361b82f 100644 --- a/src/mesa/pipe/softpipe/sp_tile_cache.c +++ b/src/mesa/pipe/softpipe/sp_tile_cache.c @@ -341,7 +341,7 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, } } #if 0 - printf("num cleared: %u\n", numCleared); + debug_printf("num cleared: %u\n", numCleared); #endif } @@ -384,7 +384,7 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, #endif #if 0 - printf("flushed tiles in use: %d\n", inuse); + debug_printf("flushed tiles in use: %d\n", inuse); #endif } diff --git a/src/mesa/pipe/tgsi/exec/tgsi_exec.c b/src/mesa/pipe/tgsi/exec/tgsi_exec.c index dcc39362a9b..463ff0d9dab 100644 --- a/src/mesa/pipe/tgsi/exec/tgsi_exec.c +++ b/src/mesa/pipe/tgsi/exec/tgsi_exec.c @@ -143,7 +143,7 @@ tgsi_exec_prepare( struct tgsi_exec_machine *mach ) k = tgsi_parse_init( &parse, mach->Tokens ); if (k != TGSI_PARSE_OK) { - fprintf(stderr, "Problem parsing!\n"); + debug_printf("Problem parsing!\n"); return; } @@ -249,7 +249,7 @@ tgsi_exec_machine_init( k = tgsi_parse_init (&parse, mach->Tokens); if (k != TGSI_PARSE_OK) { - fprintf( stderr, "Problem parsing!\n" ); + debug_printf( "Problem parsing!\n" ); return; } @@ -1236,7 +1236,7 @@ exec_tex(struct tgsi_exec_machine *mach, uint chan_index; float lodBias; - /* printf("Sampler %u unit %u\n", sampler, unit); */ + /* debug_printf("Sampler %u unit %u\n", sampler, unit); */ switch (inst->InstructionExtTexture.Texture) { case TGSI_TEXTURE_1D: diff --git a/src/mesa/pipe/tgsi/exec/tgsi_sse2.c b/src/mesa/pipe/tgsi/exec/tgsi_sse2.c index df0c6983013..f2180082f15 100755 --- a/src/mesa/pipe/tgsi/exec/tgsi_sse2.c +++ b/src/mesa/pipe/tgsi/exec/tgsi_sse2.c @@ -48,28 +48,28 @@ _print_reg( case file_REG32: switch( reg.idx ) { case reg_AX: - printf( "EAX" ); + debug_printf( "EAX" ); break; case reg_CX: - printf( "ECX" ); + debug_printf( "ECX" ); break; case reg_DX: - printf( "EDX" ); + debug_printf( "EDX" ); break; case reg_BX: - printf( "EBX" ); + debug_printf( "EBX" ); break; case reg_SP: - printf( "ESP" ); + debug_printf( "ESP" ); break; case reg_BP: - printf( "EBP" ); + debug_printf( "EBP" ); break; case reg_SI: - printf( "ESI" ); + debug_printf( "ESI" ); break; case reg_DI: - printf( "EDI" ); + debug_printf( "EDI" ); break; } break; @@ -77,7 +77,7 @@ _print_reg( assert( 0 ); break; case file_XMM: - printf( "XMM%u", reg.idx ); + debug_printf( "XMM%u", reg.idx ); break; case file_x87: assert( 0 ); @@ -92,35 +92,35 @@ _fill( unsigned count = 10 - strlen( op ); while( count-- ) { - printf( " " ); + debug_printf( " " ); } } -#define DUMP_START() printf( "\nsse-dump start ----------------" ) -#define DUMP_END() printf( "\nsse-dump end ----------------\n" ) -#define DUMP( OP ) printf( "\n%s", OP ) +#define DUMP_START() debug_printf( "\nsse-dump start ----------------" ) +#define DUMP_END() debug_printf( "\nsse-dump end ----------------\n" ) +#define DUMP( OP ) debug_printf( "\n%s", OP ) #define DUMP_I( OP, I ) do {\ - printf( "\n%s", OP );\ + debug_printf( "\n%s", OP );\ _fill( OP );\ - printf( "%u", I ); } while( 0 ) + debug_printf( "%u", I ); } while( 0 ) #define DUMP_R( OP, R0 ) do {\ - printf( "\n%s", OP );\ + debug_printf( "\n%s", OP );\ _fill( OP );\ _print_reg( R0 ); } while( 0 ) #define DUMP_RR( OP, R0, R1 ) do {\ - printf( "\n%s", OP );\ + debug_printf( "\n%s", OP );\ _fill( OP );\ _print_reg( R0 );\ - printf( ", " );\ + debug_printf( ", " );\ _print_reg( R1 ); } while( 0 ) #define DUMP_RRI( OP, R0, R1, I ) do {\ - printf( "\n%s", OP );\ + debug_printf( "\n%s", OP );\ _fill( OP );\ _print_reg( R0 );\ - printf( ", " );\ + debug_printf( ", " );\ _print_reg( R1 );\ - printf( ", " );\ - printf( "%u", I ); } while( 0 ) + debug_printf( ", " );\ + debug_printf( "%u", I ); } while( 0 ) #else diff --git a/src/mesa/pipe/tgsi/util/tgsi_build.c b/src/mesa/pipe/tgsi/util/tgsi_build.c index 67f7d2c2c28..a00ff1c2a5f 100644 --- a/src/mesa/pipe/tgsi/util/tgsi_build.c +++ b/src/mesa/pipe/tgsi/util/tgsi_build.c @@ -1,3 +1,4 @@ +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi_build.h" diff --git a/src/mesa/pipe/tgsi/util/tgsi_dump.c b/src/mesa/pipe/tgsi/util/tgsi_dump.c index cdbc0dbc9ca..b5c54847e0b 100644 --- a/src/mesa/pipe/tgsi/util/tgsi_dump.c +++ b/src/mesa/pipe/tgsi/util/tgsi_dump.c @@ -25,6 +25,9 @@ * **************************************************************************/ +#include + +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi_dump.h" diff --git a/src/mesa/pipe/tgsi/util/tgsi_parse.c b/src/mesa/pipe/tgsi/util/tgsi_parse.c index f0f8d44ac20..bf6b89ce564 100644 --- a/src/mesa/pipe/tgsi/util/tgsi_parse.c +++ b/src/mesa/pipe/tgsi/util/tgsi_parse.c @@ -25,6 +25,7 @@ * **************************************************************************/ +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi_parse.h" diff --git a/src/mesa/pipe/tgsi/util/tgsi_util.c b/src/mesa/pipe/tgsi/util/tgsi_util.c index 1e76b0f133a..4cdd89182ad 100644 --- a/src/mesa/pipe/tgsi/util/tgsi_util.c +++ b/src/mesa/pipe/tgsi/util/tgsi_util.c @@ -1,3 +1,4 @@ +#include "pipe/p_debug.h" #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi_parse.h" -- cgit v1.2.3 From bf3101afdc5d315f1fb42eb74ec1b8b0d4101aae Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Thu, 7 Feb 2008 01:07:49 +0900 Subject: gallium: Bring latest fixes. --- src/mesa/pipe/p_debug.h | 9 ++++++++- src/mesa/pipe/util/p_debug.c | 16 +++++++++++----- 2 files changed, 19 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/p_debug.h b/src/mesa/pipe/p_debug.h index b037eba2a35..2a11627b365 100644 --- a/src/mesa/pipe/p_debug.h +++ b/src/mesa/pipe/p_debug.h @@ -38,6 +38,10 @@ #ifndef P_DEBUG_H_ #define P_DEBUG_H_ + +#include + + #ifdef __cplusplus extern "C" { #endif @@ -55,8 +59,12 @@ extern "C" { void debug_printf(const char *format, ...); + +void debug_vprintf(const char *format, va_list ap); + void debug_assert_fail(const char *expr, const char *file, unsigned line); + /** Assert macro */ #ifdef DEBUG #define debug_assert(expr) ((expr) ? (void)0 : debug_assert_fail(#expr, __FILE__, __LINE__)) @@ -66,7 +74,6 @@ void debug_assert_fail(const char *expr, const char *file, unsigned line); #ifdef assert -#warning Standard C Library assert macro usage detected. #undef assert #endif #define assert(expr) debug_assert(expr) diff --git a/src/mesa/pipe/util/p_debug.c b/src/mesa/pipe/util/p_debug.c index 9303c970cca..b9607a6ba7f 100644 --- a/src/mesa/pipe/util/p_debug.c +++ b/src/mesa/pipe/util/p_debug.c @@ -40,16 +40,22 @@ #include "pipe/p_compiler.h" -void debug_printf(const char *format, ...) +void debug_vprintf(const char *format, va_list ap) { - va_list ap; - va_start( ap, format ); #ifdef WIN32 EngDebugPrint("Gallium3D: ", (PCHAR)format, ap); #else vfprintf(stderr, format, ap); #endif - va_end( ap ); +} + + +void debug_printf(const char *format, ...) +{ + va_list ap; + va_start(ap, format); + debug_vprintf(format, ap); + va_end(ap); } @@ -65,6 +71,6 @@ static INLINE void debug_abort(void) void debug_assert_fail(const char *expr, const char *file, unsigned line) { - debug_printf("%s:%i: Assertion `%s' failed."); + debug_printf("%s:%i: Assertion `%s' failed.\n", file, line, expr); debug_abort(); } -- cgit v1.2.3 From 61e59234d072ce78770047f9f08e0bb92e2fb1c5 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 16:32:15 -0700 Subject: gallium: add bitmap/drawpixels texcoord bias support The state tracker will call pipe->get_paramf(PIPE_CAP_BITMAP_TEXCOORD_BIAS) to get a bias factor for adjusting the texcoords used in bitmap/drawpixels. This allows us to compensate for small differences in rasterization from one device to another. --- src/mesa/pipe/p_defines.h | 2 +- src/mesa/state_tracker/st_cb_drawpixels.c | 7 +++++-- src/mesa/state_tracker/st_context.h | 2 ++ src/mesa/state_tracker/st_extensions.c | 3 +++ 4 files changed, 11 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/p_defines.h b/src/mesa/pipe/p_defines.h index 85adf2d61df..0bf53ecb79d 100644 --- a/src/mesa/pipe/p_defines.h +++ b/src/mesa/pipe/p_defines.h @@ -265,6 +265,6 @@ enum pipe_texture_target { #define PIPE_CAP_MAX_POINT_WIDTH_AA 17 #define PIPE_CAP_MAX_TEXTURE_ANISOTROPY 18 #define PIPE_CAP_MAX_TEXTURE_LOD_BIAS 19 - +#define PIPE_CAP_BITMAP_TEXCOORD_BIAS 20 #endif diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 6b44cba2e40..34d420fcffa 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -581,10 +581,13 @@ draw_quad_colored(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, GLfloat x1, GLfloat y1, const GLfloat *color, GLboolean invertTex) { + GLfloat bias = ctx->st->bitmap_texcoord_bias; GLfloat verts[4][3][4]; /* four verts, three attribs, XYZW */ GLuint i; - GLfloat sLeft = 0.0, sRight = 1.0; - GLfloat tTop = invertTex, tBot = 1.0 - tTop; + GLfloat xBias = bias / (x1-x0); + GLfloat yBias = bias / (y1-y0); + GLfloat sLeft = 0.0 + xBias, sRight = 1.0 + xBias; + GLfloat tTop = invertTex - yBias, tBot = 1.0 - tTop - yBias; /* upper-left */ verts[0][0][0] = x0; /* attr[0].x */ diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index 2b6f8743f38..a756055898d 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -143,6 +143,8 @@ struct st_context GLfloat polygon_offset_scale; /* ?? */ + GLfloat bitmap_texcoord_bias; + /** Mapping from VERT_RESULT_x to post-transformed vertex slot */ const GLuint *vertex_result_to_slot; diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c index 0157bdd6b32..97d28d77c4b 100644 --- a/src/mesa/state_tracker/st_extensions.c +++ b/src/mesa/state_tracker/st_extensions.c @@ -106,6 +106,9 @@ void st_init_limits(struct st_context *st) c->MaxTextureLodBias = pipe->get_paramf(pipe, PIPE_CAP_MAX_TEXTURE_LOD_BIAS); + + st->bitmap_texcoord_bias + = pipe->get_paramf(pipe, PIPE_CAP_BITMAP_TEXCOORD_BIAS); } -- cgit v1.2.3 From 35caa43e68a20b39574a740e420bbca7be3b7649 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 17:46:31 -0700 Subject: clean-up --- src/mesa/state_tracker/st_atom_texture.c | 19 +++++++------------ 1 file changed, 7 insertions(+), 12 deletions(-) diff --git a/src/mesa/state_tracker/st_atom_texture.c b/src/mesa/state_tracker/st_atom_texture.c index fb21d29c409..b3859f18cb6 100644 --- a/src/mesa/state_tracker/st_atom_texture.c +++ b/src/mesa/state_tracker/st_atom_texture.c @@ -36,7 +36,6 @@ #include "st_atom.h" #include "st_cb_texture.h" #include "pipe/p_context.h" -#include "pipe/p_defines.h" /** @@ -46,18 +45,14 @@ static void update_textures(struct st_context *st) { - GLuint s; - /* ST_NEW_FRAGMENT_PROGRAM */ struct gl_fragment_program *fprog = st->ctx->FragmentProgram._Current; + GLuint unit; - for (s = 0; s < st->ctx->Const.MaxTextureCoordUnits; s++) { - GLuint su = fprog->Base.SamplerUnits[s]; - - struct gl_texture_object *texObj - = st->ctx->Texture.Unit[su]._Current; - + for (unit = 0; unit < st->ctx->Const.MaxTextureCoordUnits; unit++) { + const GLuint su = fprog->Base.SamplerUnits[unit]; + struct gl_texture_object *texObj = st->ctx->Texture.Unit[su]._Current; struct pipe_texture *pt; if (texObj) { @@ -75,9 +70,9 @@ update_textures(struct st_context *st) * this table before being deleted, otherwise the pointer * comparison below could fail. */ - if (st->state.sampler_texture[s] != pt) { - st->state.sampler_texture[s] = pt; - st->pipe->set_sampler_texture(st->pipe, s, pt); + if (st->state.sampler_texture[unit] != pt) { + st->state.sampler_texture[unit] = pt; + st->pipe->set_sampler_texture(st->pipe, unit, pt); } } } -- cgit v1.2.3 From e232b3ffc18d5b89f9472db6b96499d8a8907b63 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 17:55:16 -0700 Subject: comments, clean-ups, consts --- src/mesa/state_tracker/st_texture.c | 30 ++++++++++++++++++------------ src/mesa/state_tracker/st_texture.h | 8 ++++---- 2 files changed, 22 insertions(+), 16 deletions(-) diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 15cc458be8a..741f36c2a7d 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -59,6 +59,10 @@ target_to_target(GLenum target) } #endif + +/** + * Allocate a new pipe_texture object + */ struct pipe_texture * st_texture_create(struct st_context *st, enum pipe_texture_target target, @@ -100,17 +104,16 @@ st_texture_create(struct st_context *st, } - - -/* Can the image be pulled into a unified mipmap texture. This mirrors - * the completeness test in a lot of ways. +/** + * Check if a texture image be pulled into a unified mipmap texture. + * This mirrors the completeness test in a lot of ways. * * Not sure whether I want to pass gl_texture_image here. */ GLboolean -st_texture_match_image(struct pipe_texture *pt, - struct gl_texture_image *image, - GLuint face, GLuint level) +st_texture_match_image(const struct pipe_texture *pt, + const struct gl_texture_image *image, + GLuint face, GLuint level) { /* Images with borders are never pulled into mipmap textures. */ @@ -189,6 +192,7 @@ st_texture_image_map(struct st_context *st, struct st_texture_image *stImage, return pipe_surface_map(stImage->surface); } + void st_texture_image_unmap(struct st_texture_image *stImage) { @@ -201,7 +205,8 @@ st_texture_image_unmap(struct st_texture_image *stImage) -/* Upload data to a rectangular sub-region. Lots of choices how to do this: +/** + * Upload data to a rectangular sub-region. Lots of choices how to do this: * * - memcpy by span to current destination * - upload data as new buffer and blit @@ -261,13 +266,14 @@ st_texture_image_data(struct pipe_context *pipe, } } + /* Copy mipmap image between textures */ void st_texture_image_copy(struct pipe_context *pipe, - struct pipe_texture *dst, - GLuint face, GLuint level, - struct pipe_texture *src) + struct pipe_texture *dst, + GLuint face, GLuint level, + struct pipe_texture *src) { GLuint width = src->width[level]; GLuint height = src->height[level]; @@ -278,6 +284,7 @@ st_texture_image_copy(struct pipe_context *pipe, if (dst->compressed) height /= 4; + for (i = 0; i < depth; i++) { dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); src_surface = pipe->get_tex_surface(pipe, src, face, level, i); @@ -292,5 +299,4 @@ st_texture_image_copy(struct pipe_context *pipe, pipe_surface_reference(&dst_surface, NULL); pipe_surface_reference(&src_surface, NULL); } - } diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index d8b1bcad9d5..0b87a494c38 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -47,11 +47,11 @@ st_texture_create(struct st_context *st, GLuint compress_byte); -/* Check if an image fits an existing texture +/* Check if an image fits into an existing texture object. */ extern GLboolean -st_texture_match_image(struct pipe_texture *pt, - struct gl_texture_image *image, +st_texture_match_image(const struct pipe_texture *pt, + const struct gl_texture_image *image, GLuint face, GLuint level); /* Return a pointer to an image within a texture. Return image stride as @@ -73,7 +73,7 @@ extern const GLuint * st_texture_depth_offsets(struct pipe_texture *pt, GLuint level); -/* Return the linear offset of an image relative to the start of its region: +/* Return the linear offset of an image relative to the start of its region. */ extern GLuint st_texture_image_offset(const struct pipe_texture *pt, -- cgit v1.2.3 From 1ad145a950827e67eee4c5892cea8be9a9cf0dc7 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 18:01:18 -0700 Subject: added comment --- src/mesa/state_tracker/st_atom_texture.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/state_tracker/st_atom_texture.c b/src/mesa/state_tracker/st_atom_texture.c index b3859f18cb6..2a836d630b2 100644 --- a/src/mesa/state_tracker/st_atom_texture.c +++ b/src/mesa/state_tracker/st_atom_texture.c @@ -59,6 +59,7 @@ update_textures(struct st_context *st) GLboolean flush, retval; retval = st_finalize_texture(st->ctx, st->pipe, texObj, &flush); + /* XXX retval indicates whether there's a texture border */ pt = st_get_texobj_texture(texObj); } -- cgit v1.2.3 From 20aa31a2447a4bda378bf3d2d78c078d748b8271 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 18:13:28 -0700 Subject: clean-ups --- src/mesa/state_tracker/st_cb_texture.c | 67 +++++++++++++++------------------- 1 file changed, 29 insertions(+), 38 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index ba0950e2950..eee94baa20c 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -76,13 +76,13 @@ struct st_texture_object - static INLINE struct st_texture_object * st_texture_object(struct gl_texture_object *obj) { return (struct st_texture_object *) obj; } + static INLINE struct st_texture_image * st_texture_image(struct gl_texture_image *img) { @@ -122,32 +122,28 @@ gl_target_to_pipe(GLenum target) } +/** + * Return nominal bytes per texel for a compressed format, 0 for non-compressed + * format. + */ static int compressed_num_bytes(GLuint mesaFormat) { - int bytes = 0; switch(mesaFormat) { - case MESA_FORMAT_RGB_FXT1: case MESA_FORMAT_RGBA_FXT1: case MESA_FORMAT_RGB_DXT1: case MESA_FORMAT_RGBA_DXT1: - bytes = 2; - break; - + return 2; case MESA_FORMAT_RGBA_DXT3: case MESA_FORMAT_RGBA_DXT5: - bytes = 4; + return 4; default: - break; + return 0; } - - return bytes; } - - static GLboolean st_IsTextureResident(GLcontext * ctx, struct gl_texture_object *texObj) { @@ -164,7 +160,6 @@ st_IsTextureResident(GLcontext * ctx, struct gl_texture_object *texObj) } - static struct gl_texture_image * st_NewTextureImage(GLcontext * ctx) { @@ -216,8 +211,6 @@ st_FreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage) } - - /* ================================================================ * From linux kernel i386 header files, copes with odd sizes better * than COPY_DWORDS would: @@ -302,7 +295,7 @@ logbase2(int n) static void guess_and_alloc_texture(struct st_context *st, struct st_texture_object *stObj, - struct st_texture_image *stImage) + const struct st_texture_image *stImage) { GLuint firstLevel; GLuint lastLevel; @@ -487,21 +480,18 @@ try_pbo_upload(GLcontext *ctx, - - - - static void st_TexImage(GLcontext * ctx, - GLint dims, - GLenum target, GLint level, - GLint internalFormat, - GLint width, GLint height, GLint depth, - GLint border, - GLenum format, GLenum type, const void *pixels, - const struct gl_pixelstore_attrib *unpack, - struct gl_texture_object *texObj, - struct gl_texture_image *texImage, GLsizei imageSize, int compressed) + GLint dims, + GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint height, GLint depth, + GLint border, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *unpack, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage, + GLsizei imageSize, int compressed) { struct st_texture_object *stObj = st_texture_object(texObj); struct st_texture_image *stImage = st_texture_image(texImage); @@ -524,7 +514,7 @@ st_TexImage(GLcontext * ctx, /* choose the texture format */ texImage->TexFormat = st_ChooseTextureFormat(ctx, internalFormat, - format, type); + format, type); _mesa_set_fetch_functions(texImage, dims); @@ -536,7 +526,8 @@ st_TexImage(GLcontext * ctx, ctx->Driver.CompressedTextureSize(ctx, texImage->Width, texImage->Height, texImage->Depth, texImage->TexFormat->MesaFormat); - } else { + } + else { texelBytes = texImage->TexFormat->TexelBytes; /* Minimum pitch of 32 bytes */ @@ -669,7 +660,7 @@ st_TexImage(GLcontext * ctx, * conversion and copy: */ if (compressed) { - memcpy(texImage->Data, pixels, imageSize); + memcpy(texImage->Data, pixels, imageSize); } else { GLuint srcImageStride = _mesa_image_image_stride(unpack, width, height, @@ -1401,7 +1392,10 @@ copy_image_data_to_texture(struct st_context *st, } -/* +/** + * Called during state validation. When this function is finished, + * the texture object should be ready for rendering. + * \return GL_FALSE if a texture border is present, GL_TRUE otherwise */ GLboolean st_finalize_texture(GLcontext *ctx, @@ -1410,11 +1404,10 @@ st_finalize_texture(GLcontext *ctx, GLboolean *needFlush) { struct st_texture_object *stObj = st_texture_object(tObj); + const GLuint nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1; int comp_byte = 0; int cpp; - GLuint face, i; - GLuint nr_faces = 0; struct st_texture_image *firstImage; *needFlush = GL_FALSE; @@ -1426,8 +1419,7 @@ st_finalize_texture(GLcontext *ctx, /* What levels must the texture include at a minimum? */ calculate_first_last_level(stObj); - firstImage = - st_texture_image(stObj->base.Image[0][stObj->firstLevel]); + firstImage = st_texture_image(stObj->base.Image[0][stObj->firstLevel]); /* Fallback case: */ @@ -1503,7 +1495,6 @@ st_finalize_texture(GLcontext *ctx, /* Pull in any images not in the object's texture: */ - nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1; for (face = 0; face < nr_faces; face++) { for (i = stObj->firstLevel; i <= stObj->lastLevel; i++) { struct st_texture_image *stImage = -- cgit v1.2.3 From 08ffa00d15c4871a22f0670a8aacd7a3995a6769 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 18:15:03 -0700 Subject: Added ctx->Driver.GenerateMipmap() driver hook --- src/mesa/drivers/common/driverfuncs.c | 2 ++ src/mesa/main/dd.h | 7 +++++ src/mesa/main/fbobject.c | 2 +- src/mesa/main/texstore.c | 48 +++++++++++++++++------------------ 4 files changed, 34 insertions(+), 25 deletions(-) diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c index 33caf7dae1d..b5b383b4e42 100644 --- a/src/mesa/drivers/common/driverfuncs.c +++ b/src/mesa/drivers/common/driverfuncs.c @@ -28,6 +28,7 @@ #include "buffers.h" #include "context.h" #include "framebuffer.h" +#include "mipmap.h" #include "program.h" #include "prog_execute.h" #include "queryobj.h" @@ -99,6 +100,7 @@ _mesa_init_driver_functions(struct dd_function_table *driver) driver->CopyTexSubImage1D = _swrast_copy_texsubimage1d; driver->CopyTexSubImage2D = _swrast_copy_texsubimage2d; driver->CopyTexSubImage3D = _swrast_copy_texsubimage3d; + driver->GenerateMipmap = _mesa_generate_mipmap; driver->TestProxyTexImage = _mesa_test_proxy_teximage; driver->CompressedTexImage1D = _mesa_store_compressed_teximage1d; driver->CompressedTexImage2D = _mesa_store_compressed_teximage2d; diff --git a/src/mesa/main/dd.h b/src/mesa/main/dd.h index 3bec3bd4331..c2ef67ba6d9 100644 --- a/src/mesa/main/dd.h +++ b/src/mesa/main/dd.h @@ -332,6 +332,13 @@ struct dd_function_table { GLint x, GLint y, GLsizei width, GLsizei height ); + /** + * Called by glGenerateMipmap() or when GL_GENERATE_MIPMAP_SGIS is enabled. + */ + void (*GenerateMipmap)(GLcontext *ctx, GLenum target, + const struct gl_texture_unit *texUnit, + struct gl_texture_object *texObj); + /** * Called by glTexImage[123]D when user specifies a proxy texture * target. diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 963e35d6786..13cbd354243 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -1560,7 +1560,7 @@ _mesa_GenerateMipmapEXT(GLenum target) /* XXX this might not handle cube maps correctly */ _mesa_lock_texture(ctx, texObj); - _mesa_generate_mipmap(ctx, target, texUnit, texObj); + ctx->Driver.GenerateMipmap(ctx, target, texUnit, texObj); _mesa_unlock_texture(ctx, texObj); } diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c index 30be65525e0..26ca4f1bd5c 100644 --- a/src/mesa/main/texstore.c +++ b/src/mesa/main/texstore.c @@ -2917,9 +2917,9 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3003,9 +3003,9 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3079,9 +3079,9 @@ _mesa_store_teximage3d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3127,9 +3127,9 @@ _mesa_store_texsubimage1d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3182,9 +3182,9 @@ _mesa_store_texsubimage2d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3237,9 +3237,9 @@ _mesa_store_texsubimage3d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3313,9 +3313,9 @@ _mesa_store_compressed_teximage2d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack); @@ -3425,9 +3425,9 @@ _mesa_store_compressed_texsubimage2d(GLcontext *ctx, GLenum target, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, + &ctx->Texture.Unit[ctx->Texture.CurrentUnit], + texObj); } _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack); -- cgit v1.2.3 From 30c9e12f8d20a950b39577ea2b4a2e7a850dfe10 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 5 Feb 2008 18:15:39 -0700 Subject: plug in ctx->Driver.GenerateMipmap function --- src/mesa/state_tracker/st_cb_texture.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index eee94baa20c..15c5359360e 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -30,6 +30,7 @@ #include "main/enums.h" #include "main/image.h" #include "main/macros.h" +#include "main/mipmap.h" #include "main/texcompress.h" #include "main/texformat.h" #include "main/teximage.h" @@ -1531,6 +1532,7 @@ st_init_texture_functions(struct dd_function_table *functions) functions->CopyTexSubImage1D = st_CopyTexSubImage1D; functions->CopyTexSubImage2D = st_CopyTexSubImage2D; functions->CopyTexSubImage3D = st_CopyTexSubImage3D; + functions->GenerateMipmap = _mesa_generate_mipmap; functions->GetTexImage = st_GetTexImage; -- cgit v1.2.3 From 20f16a6ae44f55c2efa03a2c9deb07b66fe1b0e0 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 08:44:19 -0700 Subject: clean-ups in guess_and_alloc_texture() --- src/mesa/state_tracker/st_cb_texture.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 15c5359360e..8db4a804ab8 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -284,7 +284,12 @@ logbase2(int n) } -/* Otherwise, store it in memory if (Border != 0) or (any dimension == +/** + * Allocate a pipe_texture object for the given st_texture_object using + * the given st_texture_image to guess the mipmap size/levels. + * + * [comments...] + * Otherwise, store it in memory if (Border != 0) or (any dimension == * 1). * * Otherwise, if max_level >= level >= min_level, create texture with @@ -303,11 +308,12 @@ guess_and_alloc_texture(struct st_context *st, GLuint width = stImage->base.Width; GLuint height = stImage->base.Height; GLuint depth = stImage->base.Depth; - GLuint l2width, l2height, l2depth; GLuint i, comp_byte = 0; DBG("%s\n", __FUNCTION__); + assert(!stObj->pt); + if (stImage->base.Border) return; @@ -349,15 +355,15 @@ guess_and_alloc_texture(struct st_context *st, lastLevel = firstLevel; } else { - l2width = logbase2(width); - l2height = logbase2(height); - l2depth = logbase2(depth); + GLuint l2width = logbase2(width); + GLuint l2height = logbase2(height); + GLuint l2depth = logbase2(depth); lastLevel = firstLevel + MAX2(MAX2(l2width, l2height), l2depth); } - assert(!stObj->pt); if (stImage->base.IsCompressed) comp_byte = compressed_num_bytes(stImage->base.TexFormat->MesaFormat); + stObj->pt = st_texture_create(st, gl_target_to_pipe(stObj->base.Target), st_mesa_format_to_pipe_format(stImage->base.TexFormat->MesaFormat), -- cgit v1.2.3 From 64ca0678eeeb39831fcfb309ac48561b1981d360 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 09:24:30 -0700 Subject: gallium: change pipe->texture_create() to operate like the CSO functions Now, pass in a template object and return a new object. --- src/mesa/pipe/cell/ppu/cell_texture.c | 31 ++++++++++++++--------------- src/mesa/pipe/cell/ppu/cell_texture.h | 5 +++-- src/mesa/pipe/i915simple/i915_texture.c | 17 ++++++++-------- src/mesa/pipe/i915simple/i915_texture.h | 5 +++-- src/mesa/pipe/i965simple/brw_tex_layout.c | 15 +++++++------- src/mesa/pipe/i965simple/brw_tex_layout.h | 4 ++-- src/mesa/pipe/p_context.h | 4 ++-- src/mesa/pipe/softpipe/sp_texture.c | 33 +++++++++++++++---------------- src/mesa/pipe/softpipe/sp_texture.h | 5 +++-- src/mesa/state_tracker/st_texture.c | 31 ++++++++++++----------------- 10 files changed, 73 insertions(+), 77 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_texture.c b/src/mesa/pipe/cell/ppu/cell_texture.c index 2cf60229394..df178d9ca24 100644 --- a/src/mesa/pipe/cell/ppu/cell_texture.c +++ b/src/mesa/pipe/cell/ppu/cell_texture.c @@ -79,31 +79,30 @@ cell_texture_layout(struct cell_texture * spt) } -void -cell_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) +struct pipe_texture * +cell_texture_create(struct pipe_context *pipe, const struct pipe_texture *templat) { - struct cell_texture *spt = REALLOC(*pt, sizeof(struct pipe_texture), - sizeof(struct cell_texture)); + struct cell_texture *spt = CALLOC_STRUCT(cell_texture); + if (!spt) + return NULL; - if (spt) { - memset(&spt->base + 1, 0, - sizeof(struct cell_texture) - sizeof(struct pipe_texture)); + spt->base = *templat; - cell_texture_layout(spt); + cell_texture_layout(spt); - spt->buffer = pipe->winsys->buffer_create(pipe->winsys, 32, - PIPE_BUFFER_USAGE_PIXEL, - spt->buffer_size); + spt->buffer = pipe->winsys->buffer_create(pipe->winsys, 32, + PIPE_BUFFER_USAGE_PIXEL, + spt->buffer_size); - if (!spt->buffer) { - FREE(spt); - spt = NULL; - } + if (!spt->buffer) { + FREE(spt); + return NULL; } - *pt = &spt->base; + return &spt->base; } + void cell_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) { diff --git a/src/mesa/pipe/cell/ppu/cell_texture.h b/src/mesa/pipe/cell/ppu/cell_texture.h index bd434c87762..0264fed88e6 100644 --- a/src/mesa/pipe/cell/ppu/cell_texture.h +++ b/src/mesa/pipe/cell/ppu/cell_texture.h @@ -60,8 +60,9 @@ cell_texture(struct pipe_texture *pt) -extern void -cell_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); +extern struct pipe_texture * +cell_texture_create(struct pipe_context *pipe, + const struct pipe_texture *templat); extern void cell_texture_release(struct pipe_context *pipe, struct pipe_texture **pt); diff --git a/src/mesa/pipe/i915simple/i915_texture.c b/src/mesa/pipe/i915simple/i915_texture.c index 61944fe7d9a..6faeab134ab 100644 --- a/src/mesa/pipe/i915simple/i915_texture.c +++ b/src/mesa/pipe/i915simple/i915_texture.c @@ -477,17 +477,17 @@ i945_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) return TRUE; } -void -i915_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) + +struct pipe_texture * +i915_texture_create(struct pipe_context *pipe, + const struct pipe_texture *templat) { - struct i915_texture *tex = REALLOC(*pt, sizeof(struct pipe_texture), - sizeof(struct i915_texture)); + struct i915_texture *tex = CALLOC_STRUCT(i915_texture); if (tex) { struct i915_context *i915 = i915_context(pipe); - memset(&tex->base + 1, 0, - sizeof(struct i915_texture) - sizeof(struct pipe_texture)); + tex->base = *templat; if (i915->flags.is_i945 ? i945_miptree_layout(pipe, tex) : i915_miptree_layout(pipe, tex)) @@ -498,13 +498,14 @@ i915_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) if (!tex->buffer) { FREE(tex); - tex = NULL; + return NULL; } } - *pt = &tex->base; + return &tex->base; } + void i915_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) { diff --git a/src/mesa/pipe/i915simple/i915_texture.h b/src/mesa/pipe/i915simple/i915_texture.h index 84a0502e81a..330d111dc78 100644 --- a/src/mesa/pipe/i915simple/i915_texture.h +++ b/src/mesa/pipe/i915simple/i915_texture.h @@ -6,8 +6,9 @@ struct pipe_context; struct pipe_texture; -extern void -i915_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); +struct pipe_texture * +i915_texture_create(struct pipe_context *pipe, + const struct pipe_texture *templat); extern void i915_texture_release(struct pipe_context *pipe, struct pipe_texture **pt); diff --git a/src/mesa/pipe/i965simple/brw_tex_layout.c b/src/mesa/pipe/i965simple/brw_tex_layout.c index b8b6b579e22..405fd1f7949 100644 --- a/src/mesa/pipe/i965simple/brw_tex_layout.c +++ b/src/mesa/pipe/i965simple/brw_tex_layout.c @@ -299,15 +299,14 @@ static boolean brw_miptree_layout(struct pipe_context *pipe, struct brw_texture return TRUE; } -void -brw_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) + +struct pipe_texture * +brw_texture_create(struct pipe_context *pipe, const struct pipe_texture *templat) { - struct brw_texture *tex = REALLOC(*pt, sizeof(struct pipe_texture), - sizeof(struct brw_texture)); + struct brw_texture *tex = CALLOC_STRUCT(brw_texture); if (tex) { - memset(&tex->base + 1, 0, - sizeof(struct brw_texture) - sizeof(struct pipe_texture)); + tex->base = *templat; if (brw_miptree_layout(pipe, tex)) tex->buffer = pipe->winsys->buffer_create(pipe->winsys, 64, @@ -317,11 +316,11 @@ brw_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) if (!tex->buffer) { FREE(tex); - tex = NULL; + return NULL; } } - *pt = &tex->base; + return &tex->base; } void diff --git a/src/mesa/pipe/i965simple/brw_tex_layout.h b/src/mesa/pipe/i965simple/brw_tex_layout.h index 15e275058af..cfd6b1ef3ae 100644 --- a/src/mesa/pipe/i965simple/brw_tex_layout.h +++ b/src/mesa/pipe/i965simple/brw_tex_layout.h @@ -6,8 +6,8 @@ struct pipe_context; struct pipe_texture; -extern void -brw_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); +extern struct pipe_texture * +brw_texture_create(struct pipe_context *pipe, const struct pipe_texture *templat); extern void brw_texture_release(struct pipe_context *pipe, struct pipe_texture **pt); diff --git a/src/mesa/pipe/p_context.h b/src/mesa/pipe/p_context.h index 0dda06c53ba..92a1cd70c42 100644 --- a/src/mesa/pipe/p_context.h +++ b/src/mesa/pipe/p_context.h @@ -199,8 +199,8 @@ struct pipe_context { /* * Texture functions */ - void (*texture_create)(struct pipe_context *pipe, - struct pipe_texture **pt); + struct pipe_texture * (*texture_create)(struct pipe_context *pipe, + const struct pipe_texture *templat); void (*texture_release)(struct pipe_context *pipe, struct pipe_texture **pt); diff --git a/src/mesa/pipe/softpipe/sp_texture.c b/src/mesa/pipe/softpipe/sp_texture.c index 172234843d8..fd2cc3dbbb4 100644 --- a/src/mesa/pipe/softpipe/sp_texture.c +++ b/src/mesa/pipe/softpipe/sp_texture.c @@ -79,31 +79,30 @@ softpipe_texture_layout(struct softpipe_texture * spt) } -void -softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt) +struct pipe_texture * +softpipe_texture_create(struct pipe_context *pipe, + const struct pipe_texture *templat) { - struct softpipe_texture *spt = REALLOC(*pt, sizeof(struct pipe_texture), - sizeof(struct softpipe_texture)); - - if (spt) { - memset(&spt->base + 1, 0, - sizeof(struct softpipe_texture) - sizeof(struct pipe_texture)); + struct softpipe_texture *spt = CALLOC_STRUCT(softpipe_texture); + if (!spt) + return NULL; - softpipe_texture_layout(spt); + spt->base = *templat; - spt->buffer = pipe->winsys->buffer_create(pipe->winsys, 32, - PIPE_BUFFER_USAGE_PIXEL, - spt->buffer_size); + softpipe_texture_layout(spt); - if (!spt->buffer) { - FREE(spt); - spt = NULL; - } + spt->buffer = pipe->winsys->buffer_create(pipe->winsys, 32, + PIPE_BUFFER_USAGE_PIXEL, + spt->buffer_size); + if (!spt->buffer) { + FREE(spt); + return NULL; } - *pt = &spt->base; + return &spt->base; } + void softpipe_texture_release(struct pipe_context *pipe, struct pipe_texture **pt) { diff --git a/src/mesa/pipe/softpipe/sp_texture.h b/src/mesa/pipe/softpipe/sp_texture.h index c6cf370351d..fa646c0de9f 100644 --- a/src/mesa/pipe/softpipe/sp_texture.h +++ b/src/mesa/pipe/softpipe/sp_texture.h @@ -55,8 +55,9 @@ softpipe_texture(struct pipe_texture *pt) -extern void -softpipe_texture_create(struct pipe_context *pipe, struct pipe_texture **pt); +extern struct pipe_texture * +softpipe_texture_create(struct pipe_context *pipe, + const struct pipe_texture *templat); extern void softpipe_texture_release(struct pipe_context *pipe, struct pipe_texture **pt); diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 741f36c2a7d..844a9f80d85 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -74,7 +74,7 @@ st_texture_create(struct st_context *st, GLuint depth0, GLuint compress_byte) { - struct pipe_texture *pt = CALLOC_STRUCT(pipe_texture); + struct pipe_texture pt; assert(target <= PIPE_TEXTURE_CUBE); @@ -82,25 +82,20 @@ st_texture_create(struct st_context *st, _mesa_lookup_enum_by_nr(target), _mesa_lookup_enum_by_nr(format), first_level, last_level); - if (!pt) - return NULL; - assert(format); - pt->target = target; - pt->format = format; - pt->first_level = first_level; - pt->last_level = last_level; - pt->width[0] = width0; - pt->height[0] = height0; - pt->depth[0] = depth0; - pt->compressed = compress_byte ? 1 : 0; - pt->cpp = pt->compressed ? compress_byte : st_sizeof_format(format); - pt->refcount = 1; - - st->pipe->texture_create(st->pipe, &pt); - - return pt; + pt.target = target; + pt.format = format; + pt.first_level = first_level; + pt.last_level = last_level; + pt.width[0] = width0; + pt.height[0] = height0; + pt.depth[0] = depth0; + pt.compressed = compress_byte ? 1 : 0; + pt.cpp = pt.compressed ? compress_byte : st_sizeof_format(format); + pt.refcount = 1; + + return st->pipe->texture_create(st->pipe, &pt); } -- cgit v1.2.3 From 67155e8edff96f0a6e85580b0753041e67d3f99d Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 09:28:20 -0700 Subject: gallium: added mem_dup() --- src/mesa/pipe/p_util.h | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/src/mesa/pipe/p_util.h b/src/mesa/pipe/p_util.h index 4780ed78184..991ac447bac 100644 --- a/src/mesa/pipe/p_util.h +++ b/src/mesa/pipe/p_util.h @@ -183,6 +183,20 @@ align_free(void *ptr) +/** + * Duplicate of a block of memory + */ +static INLINE void * +mem_dup(const void *src, uint size) +{ + void *dup = malloc(size); + if (dup) + memcpy(dup, src, size); + return dup; +} + + + #define CLAMP( X, MIN, MAX ) ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) ) #define MIN2( A, B ) ( (A)<(B) ? (A) : (B) ) #define MAX2( A, B ) ( (A)>(B) ? (A) : (B) ) -- cgit v1.2.3 From 2a3b31c1c4e73529c93476161973dae13b31aee5 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 09:29:00 -0700 Subject: gallium: use mem_dup() --- src/mesa/pipe/softpipe/sp_state_blend.c | 10 +++------- src/mesa/pipe/softpipe/sp_state_rasterizer.c | 7 ++----- src/mesa/pipe/softpipe/sp_state_sampler.c | 4 +--- 3 files changed, 6 insertions(+), 15 deletions(-) diff --git a/src/mesa/pipe/softpipe/sp_state_blend.c b/src/mesa/pipe/softpipe/sp_state_blend.c index 160ca5cbc09..2d40d6bd8f5 100644 --- a/src/mesa/pipe/softpipe/sp_state_blend.c +++ b/src/mesa/pipe/softpipe/sp_state_blend.c @@ -32,13 +32,12 @@ #include "sp_context.h" #include "sp_state.h" + void * softpipe_create_blend_state(struct pipe_context *pipe, const struct pipe_blend_state *blend) { - struct pipe_blend_state *state = MALLOC( sizeof(struct pipe_blend_state) ); - memcpy(state, blend, sizeof(struct pipe_blend_state)); - return state; + return mem_dup(blend, sizeof(*blend)); } void softpipe_bind_blend_state( struct pipe_context *pipe, @@ -78,10 +77,7 @@ void * softpipe_create_depth_stencil_state(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *depth_stencil) { - struct pipe_depth_stencil_alpha_state *state = - MALLOC( sizeof(struct pipe_depth_stencil_alpha_state) ); - memcpy(state, depth_stencil, sizeof(struct pipe_depth_stencil_alpha_state)); - return state; + return mem_dup(depth_stencil, sizeof(*depth_stencil)); } void diff --git a/src/mesa/pipe/softpipe/sp_state_rasterizer.c b/src/mesa/pipe/softpipe/sp_state_rasterizer.c index ce8fa4f2b8b..53755099dde 100644 --- a/src/mesa/pipe/softpipe/sp_state_rasterizer.c +++ b/src/mesa/pipe/softpipe/sp_state_rasterizer.c @@ -35,12 +35,9 @@ void * softpipe_create_rasterizer_state(struct pipe_context *pipe, - const struct pipe_rasterizer_state *setup) + const struct pipe_rasterizer_state *rast) { - struct pipe_rasterizer_state *state = - MALLOC( sizeof(struct pipe_rasterizer_state) ); - memcpy(state, setup, sizeof(struct pipe_rasterizer_state)); - return state; + return mem_dup(rast, sizeof(*rast)); } void softpipe_bind_rasterizer_state(struct pipe_context *pipe, diff --git a/src/mesa/pipe/softpipe/sp_state_sampler.c b/src/mesa/pipe/softpipe/sp_state_sampler.c index 3842e71503b..51b4b782873 100644 --- a/src/mesa/pipe/softpipe/sp_state_sampler.c +++ b/src/mesa/pipe/softpipe/sp_state_sampler.c @@ -40,9 +40,7 @@ void * softpipe_create_sampler_state(struct pipe_context *pipe, const struct pipe_sampler_state *sampler) { - struct pipe_sampler_state *state = MALLOC( sizeof(struct pipe_sampler_state) ); - memcpy(state, sampler, sizeof(struct pipe_sampler_state)); - return state; + return mem_dup(sampler, sizeof(*sampler)); } void -- cgit v1.2.3 From 8ce9d29df9a8b0a43c60d946c85e2c871e12e911 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 09:35:33 -0700 Subject: Cell: use mem_dup() --- src/mesa/pipe/cell/ppu/cell_state_blend.c | 9 ++------- src/mesa/pipe/cell/ppu/cell_state_sampler.c | 4 +--- 2 files changed, 3 insertions(+), 10 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_state_blend.c b/src/mesa/pipe/cell/ppu/cell_state_blend.c index 2c19aa39715..4fc60548c85 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_blend.c +++ b/src/mesa/pipe/cell/ppu/cell_state_blend.c @@ -39,9 +39,7 @@ void * cell_create_blend_state(struct pipe_context *pipe, const struct pipe_blend_state *blend) { - struct pipe_blend_state *state = MALLOC(sizeof(struct pipe_blend_state)); - memcpy(state, blend, sizeof(struct pipe_blend_state)); - return state; + return mem_dup(blend, sizeof(*blend)); } @@ -85,10 +83,7 @@ void * cell_create_depth_stencil_alpha_state(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *depth_stencil) { - struct pipe_depth_stencil_alpha_state *state = - MALLOC(sizeof(struct pipe_depth_stencil_alpha_state)); - memcpy(state, depth_stencil, sizeof(struct pipe_depth_stencil_alpha_state)); - return state; + return mem_dup(depth_stencil, sizeof(*depth_stencil)); } diff --git a/src/mesa/pipe/cell/ppu/cell_state_sampler.c b/src/mesa/pipe/cell/ppu/cell_state_sampler.c index 317f7603bb7..ade6cc8338e 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_sampler.c +++ b/src/mesa/pipe/cell/ppu/cell_state_sampler.c @@ -40,9 +40,7 @@ void * cell_create_sampler_state(struct pipe_context *pipe, const struct pipe_sampler_state *sampler) { - struct pipe_sampler_state *state = MALLOC( sizeof(struct pipe_sampler_state) ); - memcpy(state, sampler, sizeof(struct pipe_sampler_state)); - return state; + return mem_dup(sampler, sizeof(*sampler)); } void -- cgit v1.2.3 From 0af89a60b436efe74d9ac381516776438861fe52 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 09:48:37 -0700 Subject: gallium: #include p_debug.h since we use assert --- src/mesa/pipe/p_util.h | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/pipe/p_util.h b/src/mesa/pipe/p_util.h index 991ac447bac..469920efee3 100644 --- a/src/mesa/pipe/p_util.h +++ b/src/mesa/pipe/p_util.h @@ -29,6 +29,7 @@ #define P_UTIL_H #include "p_compiler.h" +#include "p_debug.h" #include -- cgit v1.2.3 From 1d05c41c5b98df9f2f24645b141365f211dbddb6 Mon Sep 17 00:00:00 2001 From: Brian Date: Wed, 6 Feb 2008 09:48:56 -0700 Subject: Cell: silence unused var warnings --- src/mesa/pipe/cell/ppu/cell_state_fs.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/src/mesa/pipe/cell/ppu/cell_state_fs.c b/src/mesa/pipe/cell/ppu/cell_state_fs.c index 81c2ac14ddc..96a52273b01 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_fs.c +++ b/src/mesa/pipe/cell/ppu/cell_state_fs.c @@ -45,7 +45,7 @@ void * cell_create_fs_state(struct pipe_context *pipe, const struct pipe_shader_state *templ) { - struct cell_context *cell = cell_context(pipe); + /*struct cell_context *cell = cell_context(pipe);*/ struct cell_fragment_shader_state *state; state = CALLOC_STRUCT(cell_fragment_shader_state); @@ -94,8 +94,6 @@ cell_bind_fs_state(struct pipe_context *pipe, void *fs) void cell_delete_fs_state(struct pipe_context *pipe, void *fs) { - struct cell_context *cell = cell_context(pipe); - struct cell_fragment_shader_state *state = (struct cell_fragment_shader_state *) fs; -- cgit v1.2.3 From 87c8a1ba36b5deb9a0009a45a0b7a3a345c16126 Mon Sep 17 00:00:00 2001 From: José Fonseca Date: Thu, 7 Feb 2008 19:59:17 +0900 Subject: gallium: Conditionally use posix libraries/includes. --- .gitignore | 1 + SConstruct | 43 ++++++++++++++++++++++--------------------- 2 files changed, 23 insertions(+), 21 deletions(-) diff --git a/.gitignore b/.gitignore index bf50291fc16..b5e59dfc3ec 100644 --- a/.gitignore +++ b/.gitignore @@ -9,3 +9,4 @@ lib lib64 .sconsign* config.py +build diff --git a/SConstruct b/SConstruct index 47f9b5389bb..22a4072c933 100644 --- a/SConstruct +++ b/SConstruct @@ -95,19 +95,8 @@ if gcc: env.Append(CFLAGS = '-fmessage-length=0') env.Append(CXXFLAGS = '-fmessage-length=0') -# Defines -env.Append(CPPDEFINES = [ - '_POSIX_SOURCE', - ('_POSIX_C_SOURCE', '199309L'), - '_SVID_SOURCE', - '_BSD_SOURCE', - '_GNU_SOURCE', - - 'PTHREADS', - 'HAVE_ALIAS', - 'HAVE_POSIX_MEMALIGN', -]) +# Defines if debug: env.Append(CPPDEFINES = ['DEBUG']) else: @@ -120,8 +109,6 @@ env.Append(CPPPATH = [ '#/src/mesa', '#/src/mesa/main', '#/src/mesa/pipe', - - '/usr/X11R6/include', ]) @@ -137,14 +124,28 @@ if x86: env.Append(CFLAGS = '-m32') env.Append(CXXFLAGS = '-m32') -env.Append(LIBPATH = ['/usr/X11R6/lib']) -env.Append(LIBS = [ - 'm', - 'pthread', - 'expat', - 'dl', -]) +# Posix +if platform in ('posix', 'linux', 'freebsd', 'darwin'): + env.Append(CPPDEFINES = [ + '_POSIX_SOURCE', + ('_POSIX_C_SOURCE', '199309L'), + '_SVID_SOURCE', + '_BSD_SOURCE', + '_GNU_SOURCE', + + 'PTHREADS', + 'HAVE_POSIX_MEMALIGN', + ]) + env.Append(CPPPATH = ['/usr/X11R6/include']) + env.Append(LIBPATH = ['/usr/X11R6/lib']) + env.Append(LIBS = [ + 'm', + 'pthread', + 'expat', + 'dl', + ]) + # DRI if dri: -- cgit v1.2.3 From f41a4ee11ab3cbed8a4fd34866ed5771b8169ab7 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Thu, 7 Feb 2008 19:43:34 +0000 Subject: pipebuffer: Fix reversed assertion. --- src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c b/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c index bcd4b3e2571..04477a865a5 100644 --- a/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c +++ b/src/mesa/pipe/pipebuffer/pb_bufmgr_pool.c @@ -170,7 +170,7 @@ pool_bufmgr_create_buffer(struct pb_manager *mgr, struct list_head *item; assert(size == pool->bufSize); - assert(desc->alignment % pool->bufAlign == 0); + assert(pool->bufAlign % desc->alignment == 0); _glthread_LOCK_MUTEX(pool->mutex); -- cgit v1.2.3 From 99c3c2d038074209686559c42d7314fcaaf3953a Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Thu, 7 Feb 2008 19:44:42 +0000 Subject: tgsi: Fall back to interpreter instead of assert(0) on unimplemented SSE code. --- src/mesa/pipe/draw/draw_vertex_shader.c | 22 ++++++++++++++++------ src/mesa/pipe/tgsi/exec/tgsi_sse2.c | 3 +-- src/mesa/x86/rtasm/x86sse.c | 1 + 3 files changed, 18 insertions(+), 8 deletions(-) diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index e6590eafcc8..5ca93aa6152 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -121,11 +121,16 @@ run_vertex_program(struct draw_context *draw, = (struct draw_vertex_shader *)draw->vertex_shader; codegen_function func = (codegen_function) x86_get_func( &shader->sse2_program ); - func( - machine->Inputs, - machine->Outputs, - machine->Consts, - machine->Temps ); + + if (func) + func( + machine->Inputs, + machine->Outputs, + machine->Consts, + machine->Temps ); + else + /* interpreter */ + tgsi_exec_machine_run( machine ); } else #endif @@ -269,7 +274,12 @@ draw_create_vertex_shader(struct draw_context *draw, struct pipe_shader_state *sh = (struct pipe_shader_state *) shader; x86_init_func( &vs->sse2_program ); - tgsi_emit_sse2( (struct tgsi_token *) sh->tokens, &vs->sse2_program ); + if (!tgsi_emit_sse2( (struct tgsi_token *) sh->tokens, + &vs->sse2_program )) { + x86_release_func( (struct x86_function *) &vs->sse2_program ); + fprintf(stdout /*err*/, + "tgsi_emit_sse2() failed, falling back to interpreter\n"); + } } #endif diff --git a/src/mesa/pipe/tgsi/exec/tgsi_sse2.c b/src/mesa/pipe/tgsi/exec/tgsi_sse2.c index f2180082f15..40bacf85526 100755 --- a/src/mesa/pipe/tgsi/exec/tgsi_sse2.c +++ b/src/mesa/pipe/tgsi/exec/tgsi_sse2.c @@ -2254,8 +2254,7 @@ tgsi_emit_sse2( case TGSI_TOKEN_TYPE_IMMEDIATE: /* XXX implement this */ - assert(0); - break; + return 0; default: assert( 0 ); diff --git a/src/mesa/x86/rtasm/x86sse.c b/src/mesa/x86/rtasm/x86sse.c index 56c211eee06..f8da6e405f4 100644 --- a/src/mesa/x86/rtasm/x86sse.c +++ b/src/mesa/x86/rtasm/x86sse.c @@ -1137,6 +1137,7 @@ void x86_init_func_size( struct x86_function *p, unsigned code_size ) void x86_release_func( struct x86_function *p ) { _mesa_exec_free(p->store); + p->store = NULL; } -- cgit v1.2.3 From 1e2f5fb05a3f6720a1e7aa02e7ce12304991c6c7 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Fri, 8 Feb 2008 18:25:49 +0100 Subject: failover: several fixes to failover pipe module --- src/mesa/pipe/failover/fo_context.c | 2 + src/mesa/pipe/failover/fo_state.c | 79 ++++++++++++++++++++++++++----------- 2 files changed, 57 insertions(+), 24 deletions(-) diff --git a/src/mesa/pipe/failover/fo_context.c b/src/mesa/pipe/failover/fo_context.c index cf6c9fed502..7ce4a7df17e 100644 --- a/src/mesa/pipe/failover/fo_context.c +++ b/src/mesa/pipe/failover/fo_context.c @@ -114,6 +114,8 @@ struct pipe_context *failover_create( struct pipe_context *hw, if (failover == NULL) return NULL; + failover->hw = hw; + failover->sw = sw; failover->pipe.winsys = hw->winsys; failover->pipe.destroy = failover_destroy; failover->pipe.is_format_supported = hw->is_format_supported; diff --git a/src/mesa/pipe/failover/fo_state.c b/src/mesa/pipe/failover/fo_state.c index fa700b9674f..0fc5568da11 100644 --- a/src/mesa/pipe/failover/fo_state.c +++ b/src/mesa/pipe/failover/fo_state.c @@ -54,8 +54,8 @@ failover_create_blend_state( struct pipe_context *pipe, struct fo_state *state = malloc(sizeof(struct fo_state)); struct failover_context *failover = failover_context(pipe); - state->sw_state = failover->sw->create_blend_state(pipe, blend); - state->hw_state = failover->hw->create_blend_state(pipe, blend); + state->sw_state = failover->sw->create_blend_state(failover->sw, blend); + state->hw_state = failover->hw->create_blend_state(failover->hw, blend); return state; } @@ -68,6 +68,7 @@ failover_bind_blend_state( struct pipe_context *pipe, struct fo_state *state = (struct fo_state *)blend; failover->blend = state; failover->dirty |= FO_NEW_BLEND; + failover->sw->bind_blend_state( failover->sw, state->sw_state ); failover->hw->bind_blend_state( failover->hw, state->hw_state ); } @@ -78,8 +79,8 @@ failover_delete_blend_state( struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)blend; struct failover_context *failover = failover_context(pipe); - failover->sw->delete_blend_state(pipe, state->sw_state); - failover->hw->delete_blend_state(pipe, state->hw_state); + failover->sw->delete_blend_state(failover->sw, state->sw_state); + failover->hw->delete_blend_state(failover->hw, state->hw_state); state->sw_state = 0; state->hw_state = 0; free(state); @@ -93,6 +94,7 @@ failover_set_blend_color( struct pipe_context *pipe, failover->blend_color = *blend_color; failover->dirty |= FO_NEW_BLEND_COLOR; + failover->sw->set_blend_color( failover->sw, blend_color ); failover->hw->set_blend_color( failover->hw, blend_color ); } @@ -104,6 +106,7 @@ failover_set_clip_state( struct pipe_context *pipe, failover->clip = *clip; failover->dirty |= FO_NEW_CLIP; + failover->sw->set_clip_state( failover->sw, clip ); failover->hw->set_clip_state( failover->hw, clip ); } @@ -115,8 +118,8 @@ failover_create_depth_stencil_state(struct pipe_context *pipe, struct fo_state *state = malloc(sizeof(struct fo_state)); struct failover_context *failover = failover_context(pipe); - state->sw_state = failover->sw->create_depth_stencil_alpha_state(pipe, templ); - state->hw_state = failover->hw->create_depth_stencil_alpha_state(pipe, templ); + state->sw_state = failover->sw->create_depth_stencil_alpha_state(failover->sw, templ); + state->hw_state = failover->hw->create_depth_stencil_alpha_state(failover->hw, templ); return state; } @@ -129,6 +132,7 @@ failover_bind_depth_stencil_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state *)depth_stencil; failover->depth_stencil = state; failover->dirty |= FO_NEW_DEPTH_STENCIL; + failover->sw->bind_depth_stencil_alpha_state(failover->sw, state->sw_state); failover->hw->bind_depth_stencil_alpha_state(failover->hw, state->hw_state); } @@ -139,8 +143,8 @@ failover_delete_depth_stencil_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)ds; struct failover_context *failover = failover_context(pipe); - failover->sw->delete_depth_stencil_alpha_state(pipe, state->sw_state); - failover->hw->delete_depth_stencil_alpha_state(pipe, state->hw_state); + failover->sw->delete_depth_stencil_alpha_state(failover->sw, state->sw_state); + failover->hw->delete_depth_stencil_alpha_state(failover->hw, state->hw_state); state->sw_state = 0; state->hw_state = 0; free(state); @@ -154,6 +158,7 @@ failover_set_framebuffer_state(struct pipe_context *pipe, failover->framebuffer = *framebuffer; failover->dirty |= FO_NEW_FRAMEBUFFER; + failover->sw->set_framebuffer_state( failover->sw, framebuffer ); failover->hw->set_framebuffer_state( failover->hw, framebuffer ); } @@ -165,8 +170,8 @@ failover_create_fs_state(struct pipe_context *pipe, struct fo_state *state = malloc(sizeof(struct fo_state)); struct failover_context *failover = failover_context(pipe); - state->sw_state = failover->sw->create_fs_state(pipe, templ); - state->hw_state = failover->hw->create_fs_state(pipe, templ); + state->sw_state = failover->sw->create_fs_state(failover->sw, templ); + state->hw_state = failover->hw->create_fs_state(failover->hw, templ); return state; } @@ -178,6 +183,7 @@ failover_bind_fs_state(struct pipe_context *pipe, void *fs) struct fo_state *state = (struct fo_state*)fs; failover->fragment_shader = state; failover->dirty |= FO_NEW_FRAGMENT_SHADER; + failover->sw->bind_fs_state(failover->sw, state->sw_state); failover->hw->bind_fs_state(failover->hw, state->hw_state); } @@ -188,8 +194,8 @@ failover_delete_fs_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)fs; struct failover_context *failover = failover_context(pipe); - failover->sw->delete_fs_state(pipe, state->sw_state); - failover->hw->delete_fs_state(pipe, state->hw_state); + failover->sw->delete_fs_state(failover->sw, state->sw_state); + failover->hw->delete_fs_state(failover->hw, state->hw_state); state->sw_state = 0; state->hw_state = 0; free(state); @@ -202,8 +208,8 @@ failover_create_vs_state(struct pipe_context *pipe, struct fo_state *state = malloc(sizeof(struct fo_state)); struct failover_context *failover = failover_context(pipe); - state->sw_state = failover->sw->create_vs_state(pipe, templ); - state->hw_state = failover->hw->create_vs_state(pipe, templ); + state->sw_state = failover->sw->create_vs_state(failover->sw, templ); + state->hw_state = failover->hw->create_vs_state(failover->hw, templ); return state; } @@ -217,6 +223,7 @@ failover_bind_vs_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)vs; failover->vertex_shader = state; failover->dirty |= FO_NEW_VERTEX_SHADER; + failover->sw->bind_vs_state(failover->sw, state->sw_state); failover->hw->bind_vs_state(failover->hw, state->hw_state); } @@ -227,8 +234,8 @@ failover_delete_vs_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)vs; struct failover_context *failover = failover_context(pipe); - failover->sw->delete_vs_state(pipe, state->sw_state); - failover->hw->delete_vs_state(pipe, state->hw_state); + failover->sw->delete_vs_state(failover->sw, state->sw_state); + failover->hw->delete_vs_state(failover->hw, state->hw_state); state->sw_state = 0; state->hw_state = 0; free(state); @@ -242,6 +249,7 @@ failover_set_polygon_stipple( struct pipe_context *pipe, failover->poly_stipple = *stipple; failover->dirty |= FO_NEW_STIPPLE; + failover->sw->set_polygon_stipple( failover->sw, stipple ); failover->hw->set_polygon_stipple( failover->hw, stipple ); } @@ -253,8 +261,8 @@ failover_create_rasterizer_state(struct pipe_context *pipe, struct fo_state *state = malloc(sizeof(struct fo_state)); struct failover_context *failover = failover_context(pipe); - state->sw_state = failover->sw->create_rasterizer_state(pipe, templ); - state->hw_state = failover->hw->create_rasterizer_state(pipe, templ); + state->sw_state = failover->sw->create_rasterizer_state(failover->sw, templ); + state->hw_state = failover->hw->create_rasterizer_state(failover->hw, templ); return state; } @@ -268,6 +276,7 @@ failover_bind_rasterizer_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)raster; failover->rasterizer = state; failover->dirty |= FO_NEW_RASTERIZER; + failover->sw->bind_rasterizer_state(failover->sw, state->sw_state); failover->hw->bind_rasterizer_state(failover->hw, state->hw_state); } @@ -278,8 +287,8 @@ failover_delete_rasterizer_state(struct pipe_context *pipe, struct fo_state *state = (struct fo_state*)raster; struct failover_context *failover = failover_context(pipe); - failover->sw->delete_rasterizer_state(pipe, state->sw_state); - failover->hw->delete_rasterizer_state(pipe, state->hw_state); + failover->sw->delete_rasterizer_state(failover->sw, state->sw_state); + failover->hw->delete_rasterizer_state(failover->hw, state->hw_state); state->sw_state = 0; state->hw_state = 0; free(state); @@ -294,6 +303,7 @@ failover_set_scissor_state( struct pipe_context *pipe, failover->scissor = *scissor; failover->dirty |= FO_NEW_SCISSOR; + failover->sw->set_scissor_state( failover->sw, scissor ); failover->hw->set_scissor_state( failover->hw, scissor ); } @@ -305,8 +315,8 @@ failover_create_sampler_state(struct pipe_context *pipe, struct fo_state *state = malloc(sizeof(struct fo_state)); struct failover_context *failover = failover_context(pipe); - state->sw_state = failover->sw->create_sampler_state(pipe, templ); - state->hw_state = failover->hw->create_sampler_state(pipe, templ); + state->sw_state = failover->sw->create_sampler_state(failover->sw, templ); + state->hw_state = failover->hw->create_sampler_state(failover->hw, templ); return state; } @@ -320,6 +330,8 @@ failover_bind_sampler_state(struct pipe_context *pipe, failover->sampler[unit] = state; failover->dirty |= FO_NEW_SAMPLER; failover->dirty_sampler |= (1<sw->bind_sampler_state(failover->sw, unit, + state->sw_state); failover->hw->bind_sampler_state(failover->hw, unit, state->hw_state); } @@ -330,8 +342,8 @@ failover_delete_sampler_state(struct pipe_context *pipe, void *sampler) struct fo_state *state = (struct fo_state*)sampler; struct failover_context *failover = failover_context(pipe); - failover->sw->delete_sampler_state(pipe, state->sw_state); - failover->hw->delete_sampler_state(pipe, state->hw_state); + failover->sw->delete_sampler_state(failover->sw, state->sw_state); + failover->hw->delete_sampler_state(failover->hw, state->hw_state); state->sw_state = 0; state->hw_state = 0; free(state); @@ -348,6 +360,7 @@ failover_set_sampler_texture(struct pipe_context *pipe, failover->texture[unit] = texture; failover->dirty |= FO_NEW_TEXTURE; failover->dirty_texture |= (1<sw->set_sampler_texture( failover->sw, unit, texture ); failover->hw->set_sampler_texture( failover->hw, unit, texture ); } @@ -360,6 +373,7 @@ failover_set_viewport_state( struct pipe_context *pipe, failover->viewport = *viewport; failover->dirty |= FO_NEW_VIEWPORT; + failover->sw->set_viewport_state( failover->sw, viewport ); failover->hw->set_viewport_state( failover->hw, viewport ); } @@ -374,6 +388,7 @@ failover_set_vertex_buffer(struct pipe_context *pipe, failover->vertex_buffer[unit] = *vertex_buffer; failover->dirty |= FO_NEW_VERTEX_BUFFER; failover->dirty_vertex_buffer |= (1<sw->set_vertex_buffer( failover->sw, unit, vertex_buffer ); failover->hw->set_vertex_buffer( failover->hw, unit, vertex_buffer ); } @@ -388,9 +403,24 @@ failover_set_vertex_element(struct pipe_context *pipe, failover->vertex_element[unit] = *vertex_element; failover->dirty |= FO_NEW_VERTEX_ELEMENT; failover->dirty_vertex_element |= (1<sw->set_vertex_element( failover->sw, unit, vertex_element ); failover->hw->set_vertex_element( failover->hw, unit, vertex_element ); } +void +failover_set_constant_buffer(struct pipe_context *pipe, + uint shader, uint index, + const struct pipe_constant_buffer *buf) +{ + struct failover_context *failover = failover_context(pipe); + + assert(shader < PIPE_SHADER_TYPES); + assert(index == 0); + + failover->sw->set_constant_buffer(failover->sw, shader, index, buf); + failover->hw->set_constant_buffer(failover->hw, shader, index, buf); +} + void failover_init_state_functions( struct failover_context *failover ) @@ -423,4 +453,5 @@ failover_init_state_functions( struct failover_context *failover ) failover->pipe.set_viewport_state = failover_set_viewport_state; failover->pipe.set_vertex_buffer = failover_set_vertex_buffer; failover->pipe.set_vertex_element = failover_set_vertex_element; + failover->pipe.set_constant_buffer = failover_set_constant_buffer; } -- cgit v1.2.3 From e770d6adeb710fcd16ea6af9764121b8933315c0 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Fri, 8 Feb 2008 18:47:25 +0100 Subject: intel_winsys: remove leftover code --- src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c | 5 ----- 1 file changed, 5 deletions(-) diff --git a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c index 910c0d2cc5e..789a386500a 100644 --- a/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c +++ b/src/mesa/drivers/dri/intel_winsys/intel_winsys_pipe.c @@ -224,11 +224,6 @@ intel_i915_surface_alloc_storage(struct pipe_winsys *winsys, if(!surf->buffer) return -1; - if(ret) { - pipe_buffer_reference(winsys, &surf->buffer, NULL); - return ret; - } - return 0; } -- cgit v1.2.3 From 76dc41765f97b7550d691069fb53e699d5b07d95 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 14:45:58 -0700 Subject: Remove unused texunit parameter to ctx->Driver.GenerateMipmap() --- src/mesa/main/dd.h | 3 +-- src/mesa/main/fbobject.c | 2 +- src/mesa/main/mipmap.c | 1 - src/mesa/main/mipmap.h | 1 - src/mesa/main/texstore.c | 32 ++++++++------------------------ 5 files changed, 10 insertions(+), 29 deletions(-) diff --git a/src/mesa/main/dd.h b/src/mesa/main/dd.h index c2ef67ba6d9..37ef2a865b9 100644 --- a/src/mesa/main/dd.h +++ b/src/mesa/main/dd.h @@ -335,8 +335,7 @@ struct dd_function_table { /** * Called by glGenerateMipmap() or when GL_GENERATE_MIPMAP_SGIS is enabled. */ - void (*GenerateMipmap)(GLcontext *ctx, GLenum target, - const struct gl_texture_unit *texUnit, + void (*GenerateMipmap)(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj); /** diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 13cbd354243..6a8cba4d8a9 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -1560,7 +1560,7 @@ _mesa_GenerateMipmapEXT(GLenum target) /* XXX this might not handle cube maps correctly */ _mesa_lock_texture(ctx, texObj); - ctx->Driver.GenerateMipmap(ctx, target, texUnit, texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); _mesa_unlock_texture(ctx, texObj); } diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index 9f3db22b75c..1e61829e8f1 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -933,7 +933,6 @@ make_2d_stack_mipmap(const struct gl_texture_format *format, GLint border, */ void _mesa_generate_mipmap(GLcontext *ctx, GLenum target, - const struct gl_texture_unit *texUnit, struct gl_texture_object *texObj) { const struct gl_texture_image *srcImage; diff --git a/src/mesa/main/mipmap.h b/src/mesa/main/mipmap.h index df786032838..46e16902c86 100644 --- a/src/mesa/main/mipmap.h +++ b/src/mesa/main/mipmap.h @@ -30,7 +30,6 @@ extern void _mesa_generate_mipmap(GLcontext *ctx, GLenum target, - const struct gl_texture_unit *texUnit, struct gl_texture_object *texObj); diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c index 26ca4f1bd5c..a6a18910fc7 100644 --- a/src/mesa/main/texstore.c +++ b/src/mesa/main/texstore.c @@ -2917,9 +2917,7 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3003,9 +3001,7 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3079,9 +3075,7 @@ _mesa_store_teximage3d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3127,9 +3121,7 @@ _mesa_store_texsubimage1d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3182,9 +3174,7 @@ _mesa_store_texsubimage2d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3237,9 +3227,7 @@ _mesa_store_texsubimage3d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, packing); @@ -3313,9 +3301,7 @@ _mesa_store_compressed_teximage2d(GLcontext *ctx, GLenum target, GLint level, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack); @@ -3425,9 +3411,7 @@ _mesa_store_compressed_texsubimage2d(GLcontext *ctx, GLenum target, /* GL_SGIS_generate_mipmap */ if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - ctx->Driver.GenerateMipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } _mesa_unmap_teximage_pbo(ctx, &ctx->Unpack); -- cgit v1.2.3 From 184054fea12e4301c1ccc4cbe13594fe84f0ed78 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 14:46:47 -0700 Subject: gallium: added draw_flush() call in softpipe_bind_sampler_state() --- src/mesa/pipe/softpipe/sp_state_sampler.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/mesa/pipe/softpipe/sp_state_sampler.c b/src/mesa/pipe/softpipe/sp_state_sampler.c index 51b4b782873..291bbc40ad0 100644 --- a/src/mesa/pipe/softpipe/sp_state_sampler.c +++ b/src/mesa/pipe/softpipe/sp_state_sampler.c @@ -49,6 +49,8 @@ softpipe_bind_sampler_state(struct pipe_context *pipe, { struct softpipe_context *softpipe = softpipe_context(pipe); + draw_flush(softpipe->draw); + assert(unit < PIPE_MAX_SAMPLERS); softpipe->sampler[unit] = (struct pipe_sampler_state *)sampler; -- cgit v1.2.3 From d68ea8114abdf128907618e86c0077ad719a9920 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 14:51:32 -0700 Subject: gallium: added inClipCoords param to st_draw_vertices() to indicate coord system of vertices Also, export st_make_passthrough_vertex_shader() from st_cb_drawpixels.c --- src/mesa/state_tracker/st_cb_clear.c | 2 +- src/mesa/state_tracker/st_cb_drawpixels.c | 18 +++++++++--------- src/mesa/state_tracker/st_cb_drawpixels.h | 4 ++++ src/mesa/state_tracker/st_draw.c | 21 ++++++++++++--------- src/mesa/state_tracker/st_draw.h | 3 ++- 5 files changed, 28 insertions(+), 20 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 0cd469c156a..ab98b54bab4 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -251,7 +251,7 @@ draw_quad(GLcontext *ctx, verts[i][1][3] = color[3]; } - st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 2); + st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 2, GL_FALSE); } diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 34d420fcffa..07886e79829 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -355,8 +355,8 @@ make_fragment_shader_z(struct st_context *st) * Create a simple vertex shader that just passes through the * vertex position and texcoord (and optionally, color). */ -static struct st_vertex_program * -make_vertex_shader(struct st_context *st, GLboolean passColor) +struct st_vertex_program * +st_make_passthrough_vertex_shader(struct st_context *st, GLboolean passColor) { /* only make programs once and re-use */ static struct st_vertex_program *progs[2] = { NULL, NULL }; @@ -572,7 +572,7 @@ draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, verts[i][1][3] = 1.0; /*Q*/ } - st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 2); + st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 2, GL_FALSE); } @@ -625,7 +625,7 @@ draw_quad_colored(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, verts[i][2][3] = 1.0; /*Q*/ } - st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 3); + st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 3, GL_FALSE); } @@ -945,7 +945,7 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, if (format == GL_DEPTH_COMPONENT) { ps = st->state.framebuffer.zsbuf; stfp = make_fragment_shader_z(ctx->st); - stvp = make_vertex_shader(ctx->st, GL_TRUE); + stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE); color = ctx->Current.RasterColor; } else if (format == GL_STENCIL_INDEX) { @@ -956,7 +956,7 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, else { ps = st->state.framebuffer.cbufs[0]; stfp = combined_drawpix_fragment_program(ctx); - stvp = make_vertex_shader(ctx->st, GL_FALSE); + stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE); color = NULL; } @@ -1111,7 +1111,7 @@ st_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, struct st_context *st = ctx->st; struct pipe_texture *pt; - stvp = make_vertex_shader(ctx->st, GL_TRUE); + stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE); stfp = combined_bitmap_fragment_program(ctx); st_validate_state(st); @@ -1229,13 +1229,13 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, rbRead = st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer); color = NULL; stfp = combined_drawpix_fragment_program(ctx); - stvp = make_vertex_shader(ctx->st, GL_FALSE); + stvp = st_make_passthrough_vertex_shader(ctx->st, GL_FALSE); } else { rbRead = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer); color = ctx->Current.Attrib[VERT_ATTRIB_COLOR0]; stfp = make_fragment_shader_z(ctx->st); - stvp = make_vertex_shader(ctx->st, GL_TRUE); + stvp = st_make_passthrough_vertex_shader(ctx->st, GL_TRUE); } psRead = rbRead->surface; diff --git a/src/mesa/state_tracker/st_cb_drawpixels.h b/src/mesa/state_tracker/st_cb_drawpixels.h index 71ba4870200..b8b906f06bc 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.h +++ b/src/mesa/state_tracker/st_cb_drawpixels.h @@ -30,6 +30,10 @@ #define ST_CB_DRAWPIXELS_H +extern struct st_vertex_program * +st_make_passthrough_vertex_shader(struct st_context *st, GLboolean passColor); + + extern void st_init_drawpixels_functions(struct dd_function_table *functions); diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index c9b8e784855..ae9f5c8b117 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -354,7 +354,8 @@ st_draw_vbo(GLcontext *ctx, void st_draw_vertices(GLcontext *ctx, unsigned prim, unsigned numVertex, float *verts, - unsigned numAttribs) + unsigned numAttribs, + GLboolean inClipCoords) { const float width = ctx->DrawBuffer->Width; const float height = ctx->DrawBuffer->Height; @@ -367,14 +368,16 @@ st_draw_vertices(GLcontext *ctx, unsigned prim, assert(numAttribs > 0); - /* convert to clip coords */ - for (i = 0; i < numVertex; i++) { - float x = verts[i * numAttribs * 4 + 0]; - float y = verts[i * numAttribs * 4 + 1]; - x = x / width * 2.0 - 1.0; - y = y / height * 2.0 - 1.0; - verts[i * numAttribs * 4 + 0] = x; - verts[i * numAttribs * 4 + 1] = y; + if (!inClipCoords) { + /* convert to clip coords */ + for (i = 0; i < numVertex; i++) { + float x = verts[i * numAttribs * 4 + 0]; + float y = verts[i * numAttribs * 4 + 1]; + x = x / width * 2.0 - 1.0; + y = y / height * 2.0 - 1.0; + verts[i * numAttribs * 4 + 0] = x; + verts[i * numAttribs * 4 + 1] = y; + } } /* XXX create one-time */ diff --git a/src/mesa/state_tracker/st_draw.h b/src/mesa/state_tracker/st_draw.h index 89ee790c577..171bde57e55 100644 --- a/src/mesa/state_tracker/st_draw.h +++ b/src/mesa/state_tracker/st_draw.h @@ -62,7 +62,8 @@ st_feedback_draw_vbo(GLcontext *ctx, void st_draw_vertices(GLcontext *ctx, unsigned prim, unsigned numVertex, float *verts, - unsigned numAttribs); + unsigned numAttribs, + GLboolean inClipCoords); #endif -- cgit v1.2.3 From 44bb16c4d45584384f6fbbcc7207016421200891 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 14:54:18 -0700 Subject: gallium: initial implemenation of auto mipmap generation in state tracker Use hardware rendering to compute/render mipmap levels. The fallback path (which will be used for non-renderable texture formats) isn't working yet. --- src/mesa/sources | 1 + src/mesa/state_tracker/st_cb_texture.c | 11 +- src/mesa/state_tracker/st_context.c | 2 + src/mesa/state_tracker/st_gen_mipmap.c | 362 +++++++++++++++++++++++++++++++++ src/mesa/state_tracker/st_gen_mipmap.h | 46 +++++ 5 files changed, 415 insertions(+), 7 deletions(-) create mode 100644 src/mesa/state_tracker/st_gen_mipmap.c create mode 100644 src/mesa/state_tracker/st_gen_mipmap.h diff --git a/src/mesa/sources b/src/mesa/sources index c0087f76e68..84492c91ac8 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -234,6 +234,7 @@ STATETRACKER_SOURCES = \ state_tracker/st_extensions.c \ state_tracker/st_format.c \ state_tracker/st_framebuffer.c \ + state_tracker/st_gen_mipmap.c \ state_tracker/st_mesa_to_tgsi.c \ state_tracker/st_program.c \ state_tracker/st_texture.c diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 8db4a804ab8..3350254654b 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -42,6 +42,7 @@ #include "state_tracker/st_cb_texture.h" #include "state_tracker/st_format.h" #include "state_tracker/st_texture.h" +#include "state_tracker/st_gen_mipmap.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" @@ -703,13 +704,9 @@ st_TexImage(GLcontext * ctx, texImage->Data = NULL; } -#if 0 - /* GL_SGIS_generate_mipmap -- this can be accelerated now. - */ +#if 01 if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, - &ctx->Texture.Unit[ctx->Texture.CurrentUnit], - texObj); + ctx->Driver.GenerateMipmap(ctx, target, texObj); } #endif } @@ -1538,7 +1535,7 @@ st_init_texture_functions(struct dd_function_table *functions) functions->CopyTexSubImage1D = st_CopyTexSubImage1D; functions->CopyTexSubImage2D = st_CopyTexSubImage2D; functions->CopyTexSubImage3D = st_CopyTexSubImage3D; - functions->GenerateMipmap = _mesa_generate_mipmap; + functions->GenerateMipmap = st_generate_mipmap; functions->GetTexImage = st_GetTexImage; diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index 9c206c057a0..bf4618bed8e 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -49,6 +49,7 @@ #include "st_atom.h" #include "st_draw.h" #include "st_extensions.h" +#include "st_gen_mipmap.h" #include "st_program.h" #include "pipe/p_context.h" #include "pipe/p_winsys.h" @@ -96,6 +97,7 @@ st_create_context_priv( GLcontext *ctx, struct pipe_context *pipe ) st_init_atoms( st ); st_init_draw( st ); + st_init_generate_mipmap(st); /* we want all vertex data to be placed in buffer objects */ vbo_use_buffer_objects(ctx); diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c new file mode 100644 index 00000000000..16f9e4cd272 --- /dev/null +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -0,0 +1,362 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#include "main/imports.h" +#include "main/mipmap.h" +#include "main/teximage.h" + +#include "shader/prog_instruction.h" + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/cso_cache/cso_cache.h" + +#include "st_context.h" +#include "st_draw.h" +#include "st_gen_mipmap.h" +#include "st_program.h" +#include "st_cb_texture.h" + + + +static void *blend_cso = NULL; +static void *depthstencil_cso = NULL; +static void *rasterizer_cso = NULL; +static void *sampler_cso = NULL; + +static struct st_fragment_program *stfp = NULL; +static struct st_vertex_program *stvp = NULL; + + + +static struct st_fragment_program * +make_tex_fragment_program(GLcontext *ctx) +{ + struct st_fragment_program *stfp; + struct gl_program *p; + GLuint ic = 0; + + p = ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0); + if (!p) + return NULL; + + p->NumInstructions = 2; + + p->Instructions = _mesa_alloc_instructions(p->NumInstructions); + if (!p->Instructions) { + ctx->Driver.DeleteProgram(ctx, p); + return NULL; + } + _mesa_init_instructions(p->Instructions, p->NumInstructions); + + /* TEX result.color, fragment.texcoord[0], texture[0], 2D; */ + p->Instructions[ic].Opcode = OPCODE_TEX; + p->Instructions[ic].DstReg.File = PROGRAM_OUTPUT; + p->Instructions[ic].DstReg.Index = FRAG_RESULT_COLR; + p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT; + p->Instructions[ic].SrcReg[0].Index = FRAG_ATTRIB_TEX0; + p->Instructions[ic].TexSrcUnit = 0; + p->Instructions[ic].TexSrcTarget = TEXTURE_2D_INDEX; + ic++; + + /* END; */ + p->Instructions[ic++].Opcode = OPCODE_END; + + assert(ic == p->NumInstructions); + + p->InputsRead = FRAG_BIT_TEX0; + p->OutputsWritten = (1 << FRAG_RESULT_COLR); + + stfp = (struct st_fragment_program *) p; + + st_translate_fragment_program(ctx->st, stfp, NULL, + stfp->tokens, ST_MAX_SHADER_TOKENS); + + return stfp; +} + + + + +/** + * one-time init for generate mipmap + * XXX Note: there may be other times we need no-op/simple state like this. + * In that case, some code refactoring would be good. + */ +void +st_init_generate_mipmap(struct st_context *st) +{ + struct pipe_context *pipe = st->pipe; + struct pipe_blend_state blend; + struct pipe_rasterizer_state rasterizer; + struct pipe_sampler_state sampler; + struct pipe_depth_stencil_alpha_state depthstencil; + + assert(!blend_cso); + + memset(&blend, 0, sizeof(blend)); + blend.colormask = PIPE_MASK_RGBA; + blend_cso = pipe->create_blend_state(pipe, &blend); + + memset(&depthstencil, 0, sizeof(depthstencil)); + depthstencil_cso = pipe->create_depth_stencil_alpha_state(pipe, &depthstencil); + + memset(&rasterizer, 0, sizeof(rasterizer)); + rasterizer_cso = pipe->create_rasterizer_state(pipe, &rasterizer); + + memset(&sampler, 0, sizeof(sampler)); + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.normalized_coords = 1; + sampler_cso = pipe->create_sampler_state(pipe, &sampler); + + stfp = make_tex_fragment_program(st->ctx); + stvp = st_make_passthrough_vertex_shader(st, GL_FALSE); +} + + +void +st_destroy_generate_mipmpap(struct st_context *st) +{ + struct pipe_context *pipe = st->pipe; + + pipe->delete_blend_state(pipe, blend_cso); + pipe->delete_depth_stencil_alpha_state(pipe, depthstencil_cso); + pipe->delete_rasterizer_state(pipe, rasterizer_cso); + pipe->delete_sampler_state(pipe, sampler_cso); + + /* XXX free stfp, stvp */ + + blend_cso = NULL; + depthstencil_cso = NULL; + rasterizer_cso = NULL; + sampler_cso = NULL; +} + + +static void +simple_viewport(struct pipe_context *pipe, uint width, uint height) +{ + struct pipe_viewport_state vp; + + vp.scale[0] = 0.5 * width; + vp.scale[1] = -0.5 * height; + vp.scale[2] = 1.0; + vp.scale[3] = 1.0; + vp.translate[0] = 0.5 * width; + vp.translate[1] = 0.5 * height; + vp.translate[2] = 0.0; + vp.translate[3] = 0.0; + + pipe->set_viewport_state(pipe, &vp); +} + + + +/* + * Draw simple [-1,1]x[-1,1] quad + */ +static void +draw_quad(GLcontext *ctx) +{ + GLfloat verts[4][2][4]; /* four verts, two attribs, XYZW */ + GLuint i; + GLfloat sLeft = 0.0, sRight = 1.0; + GLfloat tTop = 1.0, tBot = 0.0; + GLfloat x0 = -1.0, x1 = 1.0; + GLfloat y0 = -1.0, y1 = 1.0; + + /* upper-left */ + verts[0][0][0] = x0; /* attr[0].x */ + verts[0][0][1] = y0; /* attr[0].y */ + verts[0][1][0] = sLeft; /* attr[1].s */ + verts[0][1][1] = tTop; /* attr[1].t */ + + /* upper-right */ + verts[1][0][0] = x1; + verts[1][0][1] = y0; + verts[1][1][0] = sRight; + verts[1][1][1] = tTop; + + /* lower-right */ + verts[2][0][0] = x1; + verts[2][0][1] = y1; + verts[2][1][0] = sRight; + verts[2][1][1] = tBot; + + /* lower-left */ + verts[3][0][0] = x0; + verts[3][0][1] = y1; + verts[3][1][0] = sLeft; + verts[3][1][1] = tBot; + + /* same for all verts: */ + for (i = 0; i < 4; i++) { + verts[i][0][2] = 0.0; /*Z*/ + verts[i][0][3] = 1.0; /*W*/ + verts[i][1][2] = 0.0; /*R*/ + verts[i][1][3] = 1.0; /*Q*/ + } + + st_draw_vertices(ctx, PIPE_PRIM_QUADS, 4, (float *) verts, 2, GL_TRUE); +} + + + +/** + * Generate mipmap levels using hardware rendering. + * \return TRUE if successful, FALSE if not possible + */ +static boolean +st_render_mipmap(struct st_context *st, + struct pipe_texture *pt, + uint baseLevel, uint lastLevel) +{ + struct pipe_context *pipe = st->pipe; + struct pipe_framebuffer_state fb; + const uint face = 0, zslice = 0; + const uint first_level_save = pt->first_level; + uint dstLevel; + + /* check if we can render in the texture's format */ + if (!pipe->is_format_supported(pipe, pt->format, PIPE_SURFACE)) { + return FALSE; + } + + /* init framebuffer state */ + memset(&fb, 0, sizeof(fb)); + fb.num_cbufs = 1; + + /* bind CSOs */ + pipe->bind_blend_state(pipe, blend_cso); + pipe->bind_depth_stencil_alpha_state(pipe, depthstencil_cso); + pipe->bind_rasterizer_state(pipe, rasterizer_cso); + pipe->bind_sampler_state(pipe, 0, sampler_cso); + + /* bind shaders */ + pipe->bind_fs_state(pipe, stfp->fs->data); + pipe->bind_vs_state(pipe, stvp->cso->data); + + for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) { + const uint srcLevel = dstLevel - 1; + + /* + * Setup framebuffer / dest surface + */ + fb.cbufs[0] = pipe->get_tex_surface(pipe, pt, face, dstLevel, zslice); + pipe->set_framebuffer_state(pipe, &fb); + + simple_viewport(pipe, pt->width[dstLevel], pt->height[dstLevel]); + + /* + * Setup src texture, override pt->first_level so we sample from + * the right mipmap level. + */ + pt->first_level = srcLevel; + pipe->set_sampler_texture(pipe, 0, pt); + + draw_quad(st->ctx); + } + + /* restore first_level */ + pt->first_level = first_level_save; + + /* restore pipe state */ + if (st->state.rasterizer) + pipe->bind_rasterizer_state(pipe, st->state.rasterizer->data); + if (st->state.fs) + pipe->bind_fs_state(pipe, st->state.fs->data); + if (st->state.vs) + pipe->bind_vs_state(pipe, st->state.vs->cso->data); + if (st->state.sampler[0]) + pipe->bind_sampler_state(pipe, 0, st->state.sampler[0]->data); + pipe->set_sampler_texture(pipe, 0, st->state.sampler_texture[0]); + pipe->set_viewport_state(pipe, &st->state.viewport); + + return TRUE; +} + + + +void +st_generate_mipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) +{ + struct st_context *st = ctx->st; + struct pipe_texture *pt = st_get_texobj_texture(texObj); + const uint baseLevel = texObj->BaseLevel; + const uint lastLevel = pt->last_level; + uint dstLevel; + + if (!st_render_mipmap(st, pt, baseLevel, lastLevel)) { + abort(); + /* XXX the following won't really work at this time */ + _mesa_generate_mipmap(ctx, target, texObj); + return; + } + + for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) { + const uint srcLevel = dstLevel - 1; + const struct gl_texture_image *srcImage + = _mesa_get_tex_image(ctx, texObj, target, srcLevel); + struct gl_texture_image *dstImage; + struct st_texture_image *stImage; + uint dstWidth = pt->width[dstLevel]; + uint dstHeight = pt->height[dstLevel]; + uint dstDepth = pt->depth[dstLevel]; + uint border = srcImage->Border; + + + dstImage = _mesa_get_tex_image(ctx, texObj, target, dstLevel); + if (!dstImage) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "generating mipmaps"); + return; + } + + if (dstImage->ImageOffsets) + _mesa_free(dstImage->ImageOffsets); + + /* Free old image data */ + if (dstImage->Data) + ctx->Driver.FreeTexImageData(ctx, dstImage); + + /* initialize new image */ + _mesa_init_teximage_fields(ctx, target, dstImage, dstWidth, dstHeight, + dstDepth, border, srcImage->InternalFormat); + + dstImage->TexFormat = srcImage->TexFormat; + + stImage = (struct st_texture_image *) dstImage; + stImage->pt = pt; + } + +} diff --git a/src/mesa/state_tracker/st_gen_mipmap.h b/src/mesa/state_tracker/st_gen_mipmap.h new file mode 100644 index 00000000000..7668c1e44e3 --- /dev/null +++ b/src/mesa/state_tracker/st_gen_mipmap.h @@ -0,0 +1,46 @@ +/************************************************************************** + * + * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef ST_GEN_MIPMAP_H +#define ST_GEN_MIPMAP_H + + +extern void +st_init_generate_mipmap(struct st_context *st); + + +extern void +st_destroy_generate_mipmpap(struct st_context *st); + + +extern void +st_generate_mipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj); + + +#endif /* ST_GEN_MIPMAP_H */ -- cgit v1.2.3 From fa0a651a3e849908a020b40f723ed347b2054dbf Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 14:55:33 -0700 Subject: fix comment typos --- src/mesa/pipe/tgsi/exec/tgsi_exec.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/tgsi/exec/tgsi_exec.c b/src/mesa/pipe/tgsi/exec/tgsi_exec.c index 463ff0d9dab..336ae1c8b6d 100644 --- a/src/mesa/pipe/tgsi/exec/tgsi_exec.c +++ b/src/mesa/pipe/tgsi/exec/tgsi_exec.c @@ -2010,7 +2010,7 @@ exec_instruction( case TGSI_OPCODE_TXB: /* Texture lookup with lod bias */ - /* src[0] = texcoord (src[0].w = load bias) */ + /* src[0] = texcoord (src[0].w = LOD bias) */ /* src[1] = sampler unit */ exec_tex(mach, inst, TRUE); break; @@ -2026,7 +2026,7 @@ exec_instruction( case TGSI_OPCODE_TXL: /* Texture lookup with explit LOD */ - /* src[0] = texcoord (src[0].w = load bias) */ + /* src[0] = texcoord (src[0].w = LOD) */ /* src[1] = sampler unit */ exec_tex(mach, inst, TRUE); break; -- cgit v1.2.3 From fc65fb54eec6562b158e38f9fc426b49174ba912 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 14:56:38 -0700 Subject: gallium: include st_cb_drawpixels.h --- src/mesa/state_tracker/st_gen_mipmap.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index 16f9e4cd272..a6ac9a55fb3 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -40,6 +40,7 @@ #include "st_draw.h" #include "st_gen_mipmap.h" #include "st_program.h" +#include "st_cb_drawpixels.h" #include "st_cb_texture.h" -- cgit v1.2.3 From 6aad1d9bbc2dd77b600c60e471da3f6e392c09ab Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 9 Feb 2008 14:08:54 +1100 Subject: nv40: delay all state emit until before draw --- src/mesa/pipe/nouveau/nouveau_stateobj.h | 2 +- src/mesa/pipe/nv40/nv40_context.h | 22 +++++++++++++++++++--- src/mesa/pipe/nv40/nv40_state.c | 26 +++++++++++++++++--------- src/mesa/pipe/nv40/nv40_state_emit.c | 25 +++++++++++++++++++++++++ src/mesa/pipe/nv40/nv40_vbo.c | 5 ++++- 5 files changed, 66 insertions(+), 14 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_stateobj.h b/src/mesa/pipe/nouveau/nouveau_stateobj.h index 8dfc0e9e9ac..58167a24de7 100644 --- a/src/mesa/pipe/nouveau/nouveau_stateobj.h +++ b/src/mesa/pipe/nouveau/nouveau_stateobj.h @@ -30,7 +30,7 @@ so_new(unsigned push, unsigned reloc) struct nouveau_stateobj *so; so = malloc(sizeof(struct nouveau_stateobj)); - so->refcount = 0; + so->refcount = 1; so->push = malloc(sizeof(unsigned) * push); so->reloc = malloc(sizeof(struct nouveau_stateobj_reloc) * reloc); diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 1a31f00ad6a..4aa34847e85 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -22,9 +22,18 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV40_NEW_VERTPROG (1 << 1) -#define NV40_NEW_FRAGPROG (1 << 2) -#define NV40_NEW_ARRAYS (1 << 3) +#define NV40_NEW_BLEND (1 << 0) +#define NV40_NEW_RAST (1 << 1) +#define NV40_NEW_ZSA (1 << 2) +#define NV40_NEW_SAMPLER (1 << 3) +#define NV40_NEW_FB (1 << 4) +#define NV40_NEW_STIPPLE (1 << 5) +#define NV40_NEW_SCISSOR (1 << 6) +#define NV40_NEW_VIEWPORT (1 << 7) +#define NV40_NEW_BCOL (1 << 8) +#define NV40_NEW_VERTPROG (1 << 9) +#define NV40_NEW_FRAGPROG (1 << 10) +#define NV40_NEW_ARRAYS (1 << 11) struct nv40_context { struct pipe_context pipe; @@ -51,6 +60,13 @@ struct nv40_context { struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; struct nouveau_stateobj *so_vtxbuf; + struct nouveau_stateobj *so_blend; + struct nouveau_stateobj *so_rast; + struct nouveau_stateobj *so_zsa; + struct nouveau_stateobj *so_bcol; + struct nouveau_stateobj *so_scissor; + struct nouveau_stateobj *so_viewport; + struct nouveau_stateobj *so_stipple; struct { struct nouveau_resource *exec_heap; diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 125134afdcd..ab53b03cb08 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -53,7 +53,8 @@ nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_emit(nv40->nvws, hwcso); + so_ref(hwcso, &nv40->so_blend); + nv40->dirty |= NV40_NEW_BLEND; } static void @@ -354,7 +355,8 @@ nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_emit(nv40->nvws, hwcso); + so_ref(hwcso, &nv40->so_rast); + nv40->dirty |= NV40_NEW_RAST; } static void @@ -420,7 +422,8 @@ nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_emit(nv40->nvws, hwcso); + so_ref(hwcso, &nv40->so_zsa); + nv40->dirty |= NV40_NEW_ZSA; } static void @@ -508,8 +511,9 @@ nv40_set_blend_color(struct pipe_context *pipe, (float_to_ubyte(bcol->color[1]) << 8) | (float_to_ubyte(bcol->color[2]) << 0))); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_bcol); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_BCOL; } static void @@ -677,8 +681,9 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, so_data (so, ((w - 1) << 16) | 0); so_data (so, ((h - 1) << 16) | 0); - so_emit(nv40->nvws, so); - so_ref (so, &nv40->so_framebuffer); + so_ref(so, &nv40->so_framebuffer); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_FB; } static void @@ -693,8 +698,9 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, for (i = 0; i < 32; i++) so_data(so, stipple->stipple[i]); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_stipple); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_STIPPLE; } static void @@ -708,8 +714,9 @@ nv40_set_scissor_state(struct pipe_context *pipe, so_data (so, ((s->maxx - s->minx) << 16) | s->minx); so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_scissor); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_SCISSOR; } static void @@ -729,8 +736,9 @@ nv40_set_viewport_state(struct pipe_context *pipe, so_data (so, fui(vpt->scale[2])); so_data (so, fui(vpt->scale[3])); - so_emit(nv40->nvws, so); + so_ref(so, &nv40->so_viewport); so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_VIEWPORT; } static void diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c index 3a22cd4bd56..a10c9955480 100644 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ b/src/mesa/pipe/nv40/nv40_state_emit.c @@ -25,6 +25,30 @@ nv40_state_emit_dummy_relocs(struct nv40_context *nv40) void nv40_emit_hw_state(struct nv40_context *nv40) { + if (nv40->dirty & NV40_NEW_FB) + so_emit(nv40->nvws, nv40->so_framebuffer); + + if (nv40->dirty & NV40_NEW_BLEND) + so_emit(nv40->nvws, nv40->so_blend); + + if (nv40->dirty & NV40_NEW_RAST) + so_emit(nv40->nvws, nv40->so_rast); + + if (nv40->dirty & NV40_NEW_ZSA) + so_emit(nv40->nvws, nv40->so_zsa); + + if (nv40->dirty & NV40_NEW_BCOL) + so_emit(nv40->nvws, nv40->so_bcol); + + if (nv40->dirty & NV40_NEW_SCISSOR) + so_emit(nv40->nvws, nv40->so_scissor); + + if (nv40->dirty & NV40_NEW_VIEWPORT) + so_emit(nv40->nvws, nv40->so_viewport); + + if (nv40->dirty & NV40_NEW_STIPPLE) + so_emit(nv40->nvws, nv40->so_stipple); + if (nv40->dirty & NV40_NEW_FRAGPROG) { nv40_fragprog_bind(nv40, nv40->fragprog.current); /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ @@ -46,6 +70,7 @@ nv40_emit_hw_state(struct nv40_context *nv40) } nv40->dirty_samplers = 0; + nv40->dirty = 0; nv40_state_emit_dummy_relocs(nv40); } diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index e2cb3fda8f2..fd1d884193b 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -158,6 +158,7 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, so_emit(nv40->nvws, vtxfmt); so_emit(nv40->nvws, vtxbuf); so_ref (vtxbuf, &nv40->so_vtxbuf); + so_ref (NULL, &vtxbuf); so_ref (NULL, &vtxfmt); } @@ -165,8 +166,10 @@ static boolean nv40_vbo_validate_state(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned ib_format) { + unsigned vdn = nv40->dirty & NV40_NEW_ARRAYS; + nv40_emit_hw_state(nv40); - if (nv40->dirty & NV40_NEW_ARRAYS || ib) { + if (vdn || ib) { nv40_vbo_arrays_update(nv40, ib, ib_format); nv40->dirty &= ~NV40_NEW_ARRAYS; } -- cgit v1.2.3 From fc38b21d2a27793f9473d0bb7f871d6a694e6923 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 9 Feb 2008 16:25:29 +1100 Subject: nouveau: interface updates --- src/mesa/pipe/nouveau/nouveau_gldefs.h | 2 +- src/mesa/pipe/nv40/nv40_context.c | 2 ++ src/mesa/pipe/nv40/nv40_fragprog.c | 1 + src/mesa/pipe/nv40/nv40_fragtex.c | 1 + src/mesa/pipe/nv40/nv40_miptree.c | 35 +++++++++++++++++----------------- src/mesa/pipe/nv50/nv50_miptree.c | 5 +++-- 6 files changed, 25 insertions(+), 21 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_gldefs.h b/src/mesa/pipe/nouveau/nouveau_gldefs.h index 8ba3bdef649..e1015c93a27 100644 --- a/src/mesa/pipe/nouveau/nouveau_gldefs.h +++ b/src/mesa/pipe/nouveau/nouveau_gldefs.h @@ -189,7 +189,7 @@ nvgl_primitive(unsigned prim) { case PIPE_PRIM_POLYGON: return 0x000a; default: - assert(0); + return 0x0001; } } diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 1351a79fe0a..49c67a8fd54 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -71,6 +71,8 @@ nv40_get_paramf(struct pipe_context *pipe, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index e650c975413..1165661010d 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -826,6 +826,7 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) so_emit(nv40->nvws, so); so_ref(so, &fp->so); + so_ref(NULL, &so); nv40->fragprog.active = fp; } diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 283d49704af..c87e3618317 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -121,6 +121,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_emit(nv40->nvws, so); so_ref (so, &nv40->so_fragtex[unit]); + so_ref (NULL, &so); } void diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 48062d1ae8a..df70feaa057 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -53,27 +53,26 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) nv40mt->total_size = offset; } -static void -nv40_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +static struct pipe_texture * +nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) { struct pipe_winsys *ws = pipe->winsys; - struct nv40_miptree *nv40mt; - - nv40mt = realloc(*pt, sizeof(struct nv40_miptree)); - if (!nv40mt) - return; - *pt = NULL; - - nv40_miptree_layout(nv40mt); - - nv40mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - nv40mt->total_size); - if (!nv40mt->buffer) { - free(nv40mt); - return; + struct nv40_miptree *mt; + + mt = malloc(sizeof(struct nv40_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + nv40_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + free(mt); + return NULL; } - - *pt = &nv40mt->base; + + return &mt->base; } static void diff --git a/src/mesa/pipe/nv50/nv50_miptree.c b/src/mesa/pipe/nv50/nv50_miptree.c index 51442d64f3e..0c034ed4387 100644 --- a/src/mesa/pipe/nv50/nv50_miptree.c +++ b/src/mesa/pipe/nv50/nv50_miptree.c @@ -4,10 +4,11 @@ #include "nv50_context.h" -static void -nv50_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +static struct pipe_texture * +nv50_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) { NOUVEAU_ERR("unimplemented\n"); + return NULL; } static void -- cgit v1.2.3 From ae78e6b549c8c67c0997f79bf1fdfac7929df92a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 10 Feb 2008 15:16:50 +1100 Subject: nv40: prep for multiple pipe contexts on a single hw channel --- src/mesa/pipe/nv40/nv40_context.c | 237 ++++++++++++++++++++----------------- src/mesa/pipe/nv40/nv40_context.h | 29 +++-- src/mesa/pipe/nv40/nv40_fragprog.c | 4 +- src/mesa/pipe/nv40/nv40_fragtex.c | 4 +- src/mesa/pipe/nv40/nv40_query.c | 11 +- src/mesa/pipe/nv40/nv40_state.c | 83 ++++++------- src/mesa/pipe/nv40/nv40_vbo.c | 8 +- src/mesa/pipe/nv40/nv40_vertprog.c | 4 +- 8 files changed, 206 insertions(+), 174 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 49c67a8fd54..302ad04c154 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -5,6 +5,10 @@ #include "nv40_context.h" +#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf +#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 +#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 + static const char * nv40_get_name(struct pipe_context *pipe) { @@ -93,7 +97,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) } if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv40->sync, 0); + nvws->notifier_reset(nv40->hw->sync, 0); BEGIN_RING(curie, 0x104, 1); OUT_RING (0); BEGIN_RING(curie, 0x100, 1); @@ -103,104 +107,32 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) FIRE_RING(); if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv40->sync, 0, 0, 2000); + nvws->notifier_wait(nv40->hw->sync, 0, 0, 2000); } static void -nv40_destroy(struct pipe_context *pipe) +nv40_channel_takedown(struct nv40_channel_context *cnv40) { - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; - - if (nv40->draw) - draw_destroy(nv40->draw); - - nvws->res_free(&nv40->vertprog.exec_heap); - nvws->res_free(&nv40->vertprog.data_heap); - - nvws->res_free(&nv40->query_heap); - nvws->notifier_free(&nv40->query); - - nvws->notifier_free(&nv40->sync); - - nvws->grobj_free(&nv40->curie); - - free(nv40); + struct nouveau_winsys *nvws = cnv40->nvws; + + nvws->res_free(&cnv40->vp_exec_heap); + nvws->res_free(&cnv40->vp_data_heap); + nvws->res_free(&cnv40->query_heap); + nvws->notifier_free(&cnv40->query); + nvws->notifier_free(&cnv40->sync); + nvws->grobj_free(&cnv40->curie); + free(cnv40); } -static boolean -nv40_init_hwctx(struct nv40_context *nv40, int curie_class) +static struct nv40_channel_context * +nv40_channel_init(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) { - struct nouveau_winsys *nvws = nv40->nvws; + struct nv40_channel_context *cnv40 = NULL; + struct nouveau_stateobj *so; + unsigned curie_class = 0; int ret; - ret = nvws->grobj_alloc(nvws, curie_class, &nv40->curie); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - BEGIN_RING(curie, NV40TCL_DMA_NOTIFY, 1); - OUT_RING (nv40->sync->handle); - BEGIN_RING(curie, NV40TCL_DMA_TEXTURE0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(curie, NV40TCL_DMA_COLOR1, 1); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(curie, NV40TCL_DMA_COLOR0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(curie, NV40TCL_DMA_VTXBUF0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(curie, NV40TCL_DMA_FENCE, 2); - OUT_RING (0); - OUT_RING (nv40->query->handle); - BEGIN_RING(curie, NV40TCL_DMA_UNK01AC, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(curie, NV40TCL_DMA_COLOR2, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - - BEGIN_RING(curie, 0x1ea4, 3); - OUT_RING (0x00000010); - OUT_RING (0x01000100); - OUT_RING (0xff800006); - - /* vtxprog output routing */ - BEGIN_RING(curie, 0x1fc4, 1); - OUT_RING (0x06144321); - BEGIN_RING(curie, 0x1fc8, 2); - OUT_RING (0xedcba987); - OUT_RING (0x00000021); - BEGIN_RING(curie, 0x1fd0, 1); - OUT_RING (0x00171615); - BEGIN_RING(curie, 0x1fd4, 1); - OUT_RING (0x001b1a19); - - BEGIN_RING(curie, 0x1ef8, 1); - OUT_RING (0x0020ffff); - BEGIN_RING(curie, 0x1d64, 1); - OUT_RING (0x00d30000); - BEGIN_RING(curie, 0x1e94, 1); - OUT_RING (0x00000001); - - FIRE_RING (); - return TRUE; -} - -#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf -#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 -#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 - -struct pipe_context * -nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv40_context *nv40; - int curie_class = 0, ret; - switch (chipset & 0xf0) { case 0x40: if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) @@ -218,65 +150,152 @@ nv40_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, } if (!curie_class) { - NOUVEAU_ERR("Unknown NV4x chipset: NV%02x\n", chipset); + NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset); return NULL; } - nv40 = CALLOC_STRUCT(nv40_context); - if (!nv40) + cnv40 = calloc(1, sizeof(struct nv40_channel_context)); + if (!cnv40) return NULL; - nv40->chipset = chipset; - nv40->nvws = nvws; + cnv40->chipset = chipset; + cnv40->nvws = nvws; /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &nv40->sync); + ret = nvws->notifier_alloc(nvws, 1, &cnv40->sync); if (ret) { NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv40_destroy(&nv40->pipe); + nv40_channel_takedown(cnv40); return NULL; } /* Query objects */ - ret = nvws->notifier_alloc(nvws, 32, &nv40->query); + ret = nvws->notifier_alloc(nvws, 32, &cnv40->query); if (ret) { NOUVEAU_ERR("Error initialising query objects: %d\n", ret); - nv40_destroy(&nv40->pipe); + nv40_channel_takedown(cnv40); return NULL; } - ret = nvws->res_init(&nv40->query_heap, 0, 32); + ret = nvws->res_init(&cnv40->query_heap, 0, 32); if (ret) { NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); - nv40_destroy(&nv40->pipe); + nv40_channel_takedown(cnv40); return NULL; } /* Vtxprog resources */ - if (nvws->res_init(&nv40->vertprog.exec_heap, 0, 512) || - nvws->res_init(&nv40->vertprog.data_heap, 0, 256)) { - nv40_destroy(&nv40->pipe); + if (nvws->res_init(&cnv40->vp_exec_heap, 0, 512) || + nvws->res_init(&cnv40->vp_data_heap, 0, 256)) { + nv40_channel_takedown(cnv40); return NULL; } + /* 3D object */ + ret = nvws->grobj_alloc(nvws, curie_class, &cnv40->curie); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + /* Static curie initialisation */ - if (!nv40_init_hwctx(nv40, curie_class)) { + so = so_new(128, 0); + so_method(so, cnv40->curie, NV40TCL_DMA_NOTIFY, 1); + so_data (so, cnv40->sync->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_TEXTURE0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR1, 1); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_VTXBUF0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_FENCE, 2); + so_data (so, 0); + so_data (so, cnv40->query->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_UNK01AC, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR2, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + + so_method(so, cnv40->curie, 0x1ea4, 3); + so_data (so, 0x00000010); + so_data (so, 0x01000100); + so_data (so, 0xff800006); + + /* vtxprog output routing */ + so_method(so, cnv40->curie, 0x1fc4, 1); + so_data (so, 0x06144321); + so_method(so, cnv40->curie, 0x1fc8, 2); + so_data (so, 0xedcba987); + so_data (so, 0x00000021); + so_method(so, cnv40->curie, 0x1fd0, 1); + so_data (so, 0x00171615); + so_method(so, cnv40->curie, 0x1fd4, 1); + so_data (so, 0x001b1a19); + + so_method(so, cnv40->curie, 0x1ef8, 1); + so_data (so, 0x0020ffff); + so_method(so, cnv40->curie, 0x1d64, 1); + so_data (so, 0x00d30000); + so_method(so, cnv40->curie, 0x1e94, 1); + so_data (so, 0x00000001); + + so_emit(nvws, so); + so_ref(NULL, &so); + nvws->push_flush(nvws->channel, 0); + + return cnv40; +} + +static void +nv40_destroy(struct pipe_context *pipe) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + if (nv40->draw) + draw_destroy(nv40->draw); + + if (nv40->hw) { + if (--nv40->hw->refcount == 0) + nv40_channel_takedown(nv40->hw); + } + + free(nv40); +} + +struct pipe_context * +nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv40_context *nv40; + + nv40 = calloc(1, sizeof(struct nv40_context)); + if (!nv40) + return NULL; + + nv40->hw = nv40_channel_init(ws, nvws, chipset); + if (!nv40->hw) { nv40_destroy(&nv40->pipe); return NULL; } - /* Pipe context setup */ - nv40->pipe.winsys = pipe_winsys; + nv40->chipset = chipset; + nv40->nvws = nvws; + nv40->pipe.winsys = ws; nv40->pipe.destroy = nv40_destroy; nv40->pipe.get_name = nv40_get_name; nv40->pipe.get_vendor = nv40_get_vendor; nv40->pipe.get_param = nv40_get_param; nv40->pipe.get_paramf = nv40_get_paramf; - nv40->pipe.draw_arrays = nv40_draw_arrays; nv40->pipe.draw_elements = nv40_draw_elements; nv40->pipe.clear = nv40_clear; - nv40->pipe.flush = nv40_flush; nv40_init_query_functions(nv40); diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index 4aa34847e85..d7c9ee7851e 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -11,7 +11,7 @@ #include "pipe/nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv40_context *ctx = nv40 + struct nv40_channel_context *ctx = nv40->hw #include "pipe/nouveau/nouveau_push.h" #include "pipe/nouveau/nouveau_stateobj.h" @@ -35,20 +35,34 @@ #define NV40_NEW_FRAGPROG (1 << 10) #define NV40_NEW_ARRAYS (1 << 11) -struct nv40_context { - struct pipe_context pipe; +struct nv40_channel_context { struct nouveau_winsys *nvws; + unsigned refcount; - struct draw_context *draw; + unsigned chipset; - int chipset; + /* HW graphics objects */ struct nouveau_grobj *curie; struct nouveau_notifier *sync; - /* query objects */ + /* Query object resources */ struct nouveau_notifier *query; struct nouveau_resource *query_heap; + /* Vtxprog resources */ + struct nouveau_resource *vp_exec_heap; + struct nouveau_resource *vp_data_heap; +}; + +struct nv40_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct nv40_channel_context *hw; + struct draw_context *draw; + + int chipset; + uint32_t dirty; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; @@ -69,9 +83,6 @@ struct nv40_context { struct nouveau_stateobj *so_stipple; struct { - struct nouveau_resource *exec_heap; - struct nouveau_resource *data_heap; - struct nv40_vertex_program *active; struct nv40_vertex_program *current; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index 1165661010d..cb7b4a5e70d 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -817,11 +817,11 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) } so = so_new(4, 1); - so_method(so, nv40->curie, NV40TCL_FP_ADDRESS, 1); + so_method(so, nv40->hw->curie, NV40TCL_FP_ADDRESS, 1); so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); - so_method(so, nv40->curie, NV40TCL_FP_CONTROL, 1); + so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); so_data (so, fp->fp_control); so_emit(nv40->nvws, so); diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index c87e3618317..d278ce18975 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -104,7 +104,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txs = tf->swizzle; so = so_new(16, 2); - so_method(so, nv40->curie, NV40TCL_TEX_OFFSET(unit), 8); + so_method(so, nv40->hw->curie, NV40TCL_TEX_OFFSET(unit), 8); so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); @@ -116,7 +116,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); so_data (so, ps->bcol); - so_method(so, nv40->curie, NV40TCL_TEX_SIZE1(unit), 1); + so_method(so, nv40->hw->curie, NV40TCL_TEX_SIZE1(unit), 1); so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); so_emit(nv40->nvws, so); diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index 06f41fe84f7..eb305e64446 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -45,9 +45,9 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - if (nv40->nvws->res_alloc(nv40->query_heap, 1, NULL, &q->object)) + if (nv40->nvws->res_alloc(nv40->hw->query_heap, 1, NULL, &q->object)) assert(0); - nv40->nvws->notifier_reset(nv40->query, q->object->start); + nv40->nvws->notifier_reset(nv40->hw->query, q->object->start); BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); OUT_RING (1); @@ -82,16 +82,17 @@ nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, if (!q->ready) { unsigned status; - status = nvws->notifier_status(nv40->query, q->object->start); + status = nvws->notifier_status(nv40->hw->query, + q->object->start); if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { if (wait == FALSE) return FALSE; - nvws->notifier_wait(nv40->query, q->object->start, + nvws->notifier_wait(nv40->hw->query, q->object->start, NV_NOTIFY_STATE_STATUS_COMPLETED, 0); } - q->result = nvws->notifier_retval(nv40->query, + q->result = nvws->notifier_retval(nv40->hw->query, q->object->start); q->ready = TRUE; nvws->res_free(&q->object); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index ab53b03cb08..80e94737efb 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -10,39 +10,40 @@ nv40_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_grobj *curie = nv40->hw->curie; struct nouveau_stateobj *so = so_new(16, 0); if (cso->blend_enable) { - so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 3); + so_method(so, curie, NV40TCL_BLEND_ENABLE, 3); so_data (so, 1); so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | nvgl_blend_func(cso->rgb_src_factor)); so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | nvgl_blend_func(cso->rgb_dst_factor)); - so_method(so, nv40->curie, NV40TCL_BLEND_EQUATION, 1); + so_method(so, curie, NV40TCL_BLEND_EQUATION, 1); so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | nvgl_blend_eqn(cso->rgb_func)); } else { - so_method(so, nv40->curie, NV40TCL_BLEND_ENABLE, 1); + so_method(so, curie, NV40TCL_BLEND_ENABLE, 1); so_data (so, 0); } - so_method(so, nv40->curie, NV40TCL_COLOR_MASK, 1); + so_method(so, curie, NV40TCL_COLOR_MASK, 1); so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); if (cso->logicop_enable) { - so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); so_data (so, 1); so_data (so, nvgl_logicop_func(cso->logicop_func)); } else { - so_method(so, nv40->curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); + so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); so_data (so, 0); } - so_method(so, nv40->curie, NV40TCL_DITHER_ENABLE, 1); + so_method(so, curie, NV40TCL_DITHER_ENABLE, 1); so_data (so, cso->dither ? 1 : 0); return (void *)so; @@ -274,22 +275,22 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, * offset_units / offset_scale */ - so_method(so, nv40->curie, NV40TCL_SHADE_MODEL, 1); + so_method(so, nv40->hw->curie, NV40TCL_SHADE_MODEL, 1); so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : NV40TCL_SHADE_MODEL_SMOOTH); - so_method(so, nv40->curie, NV40TCL_LINE_WIDTH, 2); + so_method(so, nv40->hw->curie, NV40TCL_LINE_WIDTH, 2); so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); so_data (so, cso->line_smooth ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + so_method(so, nv40->hw->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); so_data (so, cso->line_stipple_enable ? 1 : 0); so_data (so, (cso->line_stipple_pattern << 16) | cso->line_stipple_factor); - so_method(so, nv40->curie, NV40TCL_POINT_SIZE, 1); + so_method(so, nv40->hw->curie, NV40TCL_POINT_SIZE, 1); so_data (so, fui(cso->point_size)); - so_method(so, nv40->curie, NV40TCL_POLYGON_MODE_FRONT, 6); + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_MODE_FRONT, 6); if (cso->front_winding == PIPE_WINDING_CCW) { so_data(so, nvgl_polygon_mode(cso->fill_ccw)); so_data(so, nvgl_polygon_mode(cso->fill_cw)); @@ -330,10 +331,10 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, cso->poly_smooth ? 1 : 0); so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, cso->poly_stipple_enable ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_POINT_SPRITE, 1); + so_method(so, nv40->hw->curie, NV40TCL_POINT_SPRITE, 1); if (cso->point_sprite) { unsigned psctl = (1 << 0), i; @@ -374,18 +375,18 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(32, 0); - so_method(so, nv40->curie, NV40TCL_DEPTH_FUNC, 3); + so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); so_data (so, nvgl_comparison_op(cso->depth.func)); so_data (so, cso->depth.writemask ? 1 : 0); so_data (so, cso->depth.enabled ? 1 : 0); - so_method(so, nv40->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_method(so, nv40->hw->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); so_data (so, cso->alpha.enabled ? 1 : 0); so_data (so, nvgl_comparison_op(cso->alpha.func)); so_data (so, float_to_ubyte(cso->alpha.ref)); if (cso->stencil[0].enabled) { - so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); so_data (so, cso->stencil[0].enabled ? 1 : 0); so_data (so, cso->stencil[0].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[0].func)); @@ -395,12 +396,12 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); } else { - so_method(so, nv40->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); so_data (so, 0); } if (cso->stencil[1].enabled) { - so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); so_data (so, cso->stencil[1].enabled ? 1 : 0); so_data (so, cso->stencil[1].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[1].func)); @@ -410,7 +411,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); } else { - so_method(so, nv40->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); so_data (so, 0); } @@ -505,7 +506,7 @@ nv40_set_blend_color(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(2, 0); - so_method(so, nv40->curie, NV40TCL_BLEND_COLOR, 1); + so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | (float_to_ubyte(bcol->color[0]) << 16) | (float_to_ubyte(bcol->color[1]) << 8) | @@ -611,73 +612,73 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR0, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR0_PITCH, 2); + so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); so_data (so, rt[0]->pitch * rt[0]->cpp); so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR1, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR1_OFFSET, 2); + so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); so_data (so, rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR2, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR2_OFFSET, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->curie, NV40TCL_COLOR2_PITCH, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); so_data (so, rt[2]->pitch * rt[2]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - so_method(so, nv40->curie, NV40TCL_DMA_COLOR3, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_COLOR3_OFFSET, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->curie, NV40TCL_COLOR3_PITCH, 1); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); so_data (so, rt[3]->pitch * rt[3]->cpp); } if (zeta_format) { - so_method(so, nv40->curie, NV40TCL_DMA_ZETA, 1); + so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->curie, NV40TCL_ZETA_OFFSET, 1); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); so_reloc (so, zeta->buffer, zeta->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->curie, NV40TCL_ZETA_PITCH, 1); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); so_data (so, zeta->pitch * zeta->cpp); } - so_method(so, nv40->curie, NV40TCL_RT_ENABLE, 1); + so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); so_data (so, rt_enable); - so_method(so, nv40->curie, NV40TCL_RT_HORIZ, 3); + so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); so_data (so, (w << 16) | 0); so_data (so, (h << 16) | 0); so_data (so, rt_format); - so_method(so, nv40->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); so_data (so, (w << 16) | 0); so_data (so, (h << 16) | 0); - so_method(so, nv40->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); so_data (so, ((w - 1) << 16) | 0); so_data (so, ((h - 1) << 16) | 0); @@ -694,7 +695,7 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, struct nouveau_stateobj *so = so_new(33, 0); unsigned i; - so_method(so, nv40->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); for (i = 0; i < 32; i++) so_data(so, stipple->stipple[i]); @@ -710,7 +711,7 @@ nv40_set_scissor_state(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(3, 0); - so_method(so, nv40->curie, NV40TCL_SCISSOR_HORIZ, 2); + so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); so_data (so, ((s->maxx - s->minx) << 16) | s->minx); so_data (so, ((s->maxy - s->miny) << 16) | s->miny); @@ -726,7 +727,7 @@ nv40_set_viewport_state(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_stateobj *so = so_new(9, 0); - so_method(so, nv40->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); so_data (so, fui(vpt->translate[0])); so_data (so, fui(vpt->translate[1])); so_data (so, fui(vpt->translate[2])); diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index fd1d884193b..552058b3ae7 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -115,9 +115,9 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, num_hw++; vtxbuf = so_new(20, 18); - so_method(vtxbuf, nv40->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); vtxfmt = so_new(17, 0); - so_method(vtxfmt, nv40->curie, NV40TCL_VTXFMT(0), num_hw); + so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); inputs = vp->ir; for (hw = 0; hw < num_hw; hw++) { @@ -149,7 +149,7 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, } if (ib) { - so_method(vtxbuf, nv40->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, 0, NV40TCL_IDXBUF_FORMAT_DMA1); @@ -390,7 +390,7 @@ nv40_draw_elements(struct pipe_context *pipe, /* 0x4497 doesn't support real index buffers, and there doesn't appear * to be support on any chipset for 8-bit indices. */ - if (nv40->curie->grclass == NV44TCL || indexSize == 1) { + if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { nv40_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); } else { diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index e15ddbbcde5..415b3c70c79 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -648,7 +648,7 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) /* Allocate hw vtxprog exec slots */ if (!vp->exec) { - struct nouveau_resource *heap = nv40->vertprog.exec_heap; + struct nouveau_resource *heap = nv40->hw->vp_exec_heap; uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { @@ -668,7 +668,7 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) /* Allocate hw vtxprog const slots */ if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv40->vertprog.data_heap; + struct nouveau_resource *heap = nv40->hw->vp_data_heap; if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { while (heap->next && heap->size < vp->nr_consts) { -- cgit v1.2.3 From ce358b879c1823ab4fa04f56e208d679792667cc Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 10 Feb 2008 15:19:18 +1100 Subject: nouveau: memory leak --- src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c index 7d7fefa8016..e1a9271395b 100644 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c @@ -136,6 +136,7 @@ nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); nouveau_bo_del(&nvbuf->bo); + free(nvbuf); } static void * -- cgit v1.2.3 From 474f1a1d56fbb5472dd9bbf5828c413ae7e629dd Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Fri, 25 Jan 2008 06:36:35 -0500 Subject: add a stub of a lowering pass --- src/mesa/pipe/llvm/Makefile | 1 + src/mesa/pipe/llvm/gallivm.cpp | 2 ++ src/mesa/pipe/llvm/loweringpass.cpp | 17 +++++++++++++++++ src/mesa/pipe/llvm/loweringpass.h | 15 +++++++++++++++ 4 files changed, 35 insertions(+) create mode 100644 src/mesa/pipe/llvm/loweringpass.cpp create mode 100644 src/mesa/pipe/llvm/loweringpass.h diff --git a/src/mesa/pipe/llvm/Makefile b/src/mesa/pipe/llvm/Makefile index f655fb8340a..1e3ae988df3 100644 --- a/src/mesa/pipe/llvm/Makefile +++ b/src/mesa/pipe/llvm/Makefile @@ -8,6 +8,7 @@ LIBNAME = gallivm GALLIVM_SOURCES = \ gallivm.cpp \ instructions.cpp \ + loweringpass.cpp \ storage.cpp INC_SOURCES = gallivm_builtins.cpp llvm_base_shader.cpp diff --git a/src/mesa/pipe/llvm/gallivm.cpp b/src/mesa/pipe/llvm/gallivm.cpp index afa14468902..46e11c185a9 100644 --- a/src/mesa/pipe/llvm/gallivm.cpp +++ b/src/mesa/pipe/llvm/gallivm.cpp @@ -34,6 +34,7 @@ #include "gallivm.h" #include "instructions.h" +#include "loweringpass.h" #include "storage.h" #include "pipe/p_context.h" @@ -95,6 +96,7 @@ using namespace llvm; static int GLOBAL_ID = 0; static inline void AddStandardCompilePasses(PassManager &PM) { + PM.add(new LoweringPass()); PM.add(createVerifierPass()); // Verify that input is correct PM.add(createLowerSetJmpPass()); // Lower llvm.setjmp/.longjmp diff --git a/src/mesa/pipe/llvm/loweringpass.cpp b/src/mesa/pipe/llvm/loweringpass.cpp new file mode 100644 index 00000000000..556dbec3661 --- /dev/null +++ b/src/mesa/pipe/llvm/loweringpass.cpp @@ -0,0 +1,17 @@ +#include "loweringpass.h" + +using namespace llvm; + +char LoweringPass::ID = 0; +RegisterPass X("lowering", "Lowering Pass"); + +LoweringPass::LoweringPass() + : ModulePass((intptr_t)&ID) +{ +} + +bool LoweringPass::runOnModule(Module &m) +{ + llvm::cerr << "Hello: " << m.getModuleIdentifier() << "\n"; + return false; +} diff --git a/src/mesa/pipe/llvm/loweringpass.h b/src/mesa/pipe/llvm/loweringpass.h new file mode 100644 index 00000000000..f62dcf6ba73 --- /dev/null +++ b/src/mesa/pipe/llvm/loweringpass.h @@ -0,0 +1,15 @@ +#ifndef LOWERINGPASS_H +#define LOWERINGPASS_H + +#include "llvm/Pass.h" +#include "llvm/Module.h" + +struct LoweringPass : public llvm::ModulePass +{ + static char ID; + LoweringPass(); + + virtual bool runOnModule(llvm::Module &m); +}; + +#endif -- cgit v1.2.3 From 716206c190b1b1408c09807671d28dcc8906f855 Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Mon, 4 Feb 2008 10:07:02 -0500 Subject: rewrite the way we handle ir in llvm code introduce intermediate step gallivm_ir before compiling it down to the final llvm ir. --- src/mesa/pipe/cell/ppu/cell_state_fs.c | 2 +- src/mesa/pipe/draw/draw_vertex_shader.c | 27 +- src/mesa/pipe/llvm/Makefile | 1 + src/mesa/pipe/llvm/gallivm.cpp | 791 +++----------------------------- src/mesa/pipe/llvm/gallivm.h | 23 +- src/mesa/pipe/llvm/gallivm_p.h | 56 +++ src/mesa/pipe/llvm/llvm_base_shader.cpp | 435 ++++++++---------- src/mesa/pipe/llvm/llvm_entry.c | 14 +- src/mesa/pipe/llvm/tgsitollvm.cpp | 682 +++++++++++++++++++++++++++ src/mesa/pipe/llvm/tgsitollvm.h | 16 + src/mesa/pipe/softpipe/sp_state_fs.c | 2 +- 11 files changed, 1079 insertions(+), 970 deletions(-) create mode 100644 src/mesa/pipe/llvm/gallivm_p.h create mode 100644 src/mesa/pipe/llvm/tgsitollvm.cpp create mode 100644 src/mesa/pipe/llvm/tgsitollvm.h diff --git a/src/mesa/pipe/cell/ppu/cell_state_fs.c b/src/mesa/pipe/cell/ppu/cell_state_fs.c index 96a52273b01..3f46a87d189 100644 --- a/src/mesa/pipe/cell/ppu/cell_state_fs.c +++ b/src/mesa/pipe/cell/ppu/cell_state_fs.c @@ -67,7 +67,7 @@ cell_create_fs_state(struct pipe_context *pipe, #endif #ifdef MESA_LLVM - state->llvm_prog = gallivm_from_tgsi(state->shader.tokens, GALLIVM_FS); + state->llvm_prog = 0; if (!gallivm_global_cpu_engine()) { gallivm_cpu_engine_create(state->llvm_prog); } diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index 5ca93aa6152..9567283ff5c 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -113,7 +113,16 @@ run_vertex_program(struct draw_context *draw, draw->vertex_fetch.fetch_func( draw, machine, elts, count ); /* run shader */ -#if defined(__i386__) || defined(__386__) +#ifdef MESA_LLVM + if (1) { + struct gallivm_prog *prog = draw->vertex_shader->llvm_prog; + gallivm_prog_exec(prog, + machine->Inputs, + machine->Outputs, + machine->Consts, + 12, 12, 12); + } else +#elif defined(__i386__) || defined(__386__) if (draw->use_sse) { /* SSE */ /* cast away const */ @@ -212,13 +221,7 @@ draw_vertex_shader_queue_flush(struct draw_context *draw) */ draw_update_vertex_fetch( draw ); -// debug_printf( " q(%d) ", draw->vs.queue_nr ); -#ifdef MESA_LLVM - if (draw->vertex_shader->llvm_prog) { - draw_vertex_shader_queue_flush_llvm(draw); - return; - } -#endif +// fprintf(stderr, " q(%d) ", draw->vs.queue_nr ); /* run vertex shader on vertex cache entries, four per invokation */ for (i = 0; i < draw->vs.queue_nr; i += 4) { @@ -260,7 +263,13 @@ draw_create_vertex_shader(struct draw_context *draw, vs->state = shader; #ifdef MESA_LLVM - vs->llvm_prog = gallivm_from_tgsi(shader->tokens, GALLIVM_VS); + struct gallivm_ir *ir = gallivm_ir_new(GALLIVM_VS); + gallivm_ir_set_layout(ir, GALLIVM_SOA); + gallivm_ir_set_components(ir, 4); + gallivm_ir_fill_from_tgsi(ir, shader->tokens); + vs->llvm_prog = gallivm_ir_compile(ir); + gallivm_ir_delete(ir); + draw->engine = gallivm_global_cpu_engine(); if (!draw->engine) { draw->engine = gallivm_cpu_engine_create(vs->llvm_prog); diff --git a/src/mesa/pipe/llvm/Makefile b/src/mesa/pipe/llvm/Makefile index 1e3ae988df3..10ff7aacae8 100644 --- a/src/mesa/pipe/llvm/Makefile +++ b/src/mesa/pipe/llvm/Makefile @@ -9,6 +9,7 @@ GALLIVM_SOURCES = \ gallivm.cpp \ instructions.cpp \ loweringpass.cpp \ + tgsitollvm.cpp \ storage.cpp INC_SOURCES = gallivm_builtins.cpp llvm_base_shader.cpp diff --git a/src/mesa/pipe/llvm/gallivm.cpp b/src/mesa/pipe/llvm/gallivm.cpp index 46e11c185a9..cf9b0f64067 100644 --- a/src/mesa/pipe/llvm/gallivm.cpp +++ b/src/mesa/pipe/llvm/gallivm.cpp @@ -32,17 +32,17 @@ #ifdef MESA_LLVM #include "gallivm.h" +#include "gallivm_p.h" #include "instructions.h" #include "loweringpass.h" #include "storage.h" +#include "tgsitollvm.h" #include "pipe/p_context.h" #include "pipe/p_shader_tokens.h" -#include "pipe/tgsi/util/tgsi_parse.h" + #include "pipe/tgsi/exec/tgsi_exec.h" -#include "pipe/tgsi/util/tgsi_util.h" -#include "pipe/tgsi/util/tgsi_build.h" #include "pipe/tgsi/util/tgsi_dump.h" #include @@ -64,38 +64,23 @@ #include #include #include +#include #include #include #include -struct gallivm_interpolate { - int attrib; - int chan; - int type; -}; - -struct gallivm_prog { - llvm::Module *module; - void *function; - int num_consts; - int id; - enum gallivm_shader_type type; - - struct gallivm_interpolate interpolators[32*4]; //FIXME: this might not be enough for some shaders - int num_interp; -}; - struct gallivm_cpu_engine { llvm::ExecutionEngine *engine; }; -using namespace llvm; -#include "llvm_base_shader.cpp" - static int GLOBAL_ID = 0; -static inline void AddStandardCompilePasses(PassManager &PM) { +using namespace llvm; + +static inline +void AddStandardCompilePasses(PassManager &PM) +{ PM.add(new LoweringPass()); PM.add(createVerifierPass()); // Verify that input is correct @@ -152,691 +137,16 @@ static inline void AddStandardCompilePasses(PassManager &PM) { PM.add(createConstantMergePass()); // Merge dup global constants } -static inline void -add_interpolator(struct gallivm_prog *prog, - struct gallivm_interpolate *interp) -{ - prog->interpolators[prog->num_interp] = *interp; - ++prog->num_interp; -} - -static void -translate_declaration(struct gallivm_prog *prog, - llvm::Module *module, - Storage *storage, - struct tgsi_full_declaration *decl, - struct tgsi_full_declaration *fd) -{ - if (decl->Declaration.File == TGSI_FILE_INPUT) { - unsigned first, last, mask; - uint interp_method; - - assert(decl->Declaration.Declare == TGSI_DECLARE_RANGE); - - first = decl->u.DeclarationRange.First; - last = decl->u.DeclarationRange.Last; - mask = decl->Declaration.UsageMask; - - /* Do not touch WPOS.xy */ - if (first == 0) { - mask &= ~TGSI_WRITEMASK_XY; - if (mask == TGSI_WRITEMASK_NONE) { - first++; - if (first > last) { - return; - } - } - } - - interp_method = decl->Interpolation.Interpolate; - - if (mask == TGSI_WRITEMASK_XYZW) { - unsigned i, j; - - for (i = first; i <= last; i++) { - for (j = 0; j < NUM_CHANNELS; j++) { - //interp( mach, i, j ); - struct gallivm_interpolate interp; - interp.type = interp_method; - interp.attrib = i; - interp.chan = j; - add_interpolator(prog, &interp); - } - } - } else { - unsigned i, j; - for( j = 0; j < NUM_CHANNELS; j++ ) { - if( mask & (1 << j) ) { - for( i = first; i <= last; i++ ) { - struct gallivm_interpolate interp; - interp.type = interp_method; - interp.attrib = i; - interp.chan = j; - add_interpolator(prog, &interp); - } - } - } - } - } -} - - -static void -translate_immediate(Storage *storage, - struct tgsi_full_immediate *imm) -{ - float vec[4]; - int i; - for (i = 0; i < imm->Immediate.Size - 1; ++i) { - switch( imm->Immediate.DataType ) { - case TGSI_IMM_FLOAT32: - vec[i] = imm->u.ImmediateFloat32[i].Float; - break; - default: - assert( 0 ); - } - } - storage->addImmediate(vec); -} - -static inline llvm::Value * -swizzleVector(llvm::Value *val, struct tgsi_full_src_register *src, - Storage *storage) -{ - int swizzle = 0; - int start = 1000; - const int NO_SWIZZLE = TGSI_SWIZZLE_X * 1000 + TGSI_SWIZZLE_Y * 100 + - TGSI_SWIZZLE_Z * 10 + TGSI_SWIZZLE_W; - for (int k = 0; k < 4; ++k) { - swizzle += tgsi_util_get_full_src_register_extswizzle(src, k) * start; - start /= 10; - } - if (swizzle != NO_SWIZZLE) { - /*fprintf(stderr, "XXXXXXXX swizzle = %d\n", swizzle);*/ - val = storage->shuffleVector(val, swizzle); - } - return val; -} - -static void -translate_instruction(llvm::Module *module, - Storage *storage, - Instructions *instr, - struct tgsi_full_instruction *inst, - struct tgsi_full_instruction *fi, - unsigned instno) -{ - llvm::Value *inputs[4]; - inputs[0] = 0; - inputs[1] = 0; - inputs[2] = 0; - inputs[3] = 0; - - for (int i = 0; i < inst->Instruction.NumSrcRegs; ++i) { - struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i]; - llvm::Value *val = 0; - llvm::Value *indIdx = 0; - - if (src->SrcRegister.Indirect) { - indIdx = storage->addrElement(src->SrcRegisterInd.Index); - indIdx = storage->extractIndex(indIdx); - } - if (src->SrcRegister.File == TGSI_FILE_CONSTANT) { - val = storage->constElement(src->SrcRegister.Index, indIdx); - } else if (src->SrcRegister.File == TGSI_FILE_INPUT) { - val = storage->inputElement(src->SrcRegister.Index, indIdx); - } else if (src->SrcRegister.File == TGSI_FILE_TEMPORARY) { - val = storage->tempElement(src->SrcRegister.Index); - } else if (src->SrcRegister.File == TGSI_FILE_OUTPUT) { - val = storage->outputElement(src->SrcRegister.Index, indIdx); - } else if (src->SrcRegister.File == TGSI_FILE_IMMEDIATE) { - val = storage->immediateElement(src->SrcRegister.Index); - } else { - fprintf(stderr, "ERROR: not supported llvm source %d\n", src->SrcRegister.File); - return; - } - - inputs[i] = swizzleVector(val, src, storage); - } - - /*if (inputs[0]) - instr->printVector(inputs[0]); - if (inputs[1]) - instr->printVector(inputs[1]);*/ - llvm::Value *out = 0; - switch (inst->Instruction.Opcode) { - case TGSI_OPCODE_ARL: { - out = instr->arl(inputs[0]); - } - break; - case TGSI_OPCODE_MOV: { - out = inputs[0]; - } - break; - case TGSI_OPCODE_LIT: { - out = instr->lit(inputs[0]); - } - break; - case TGSI_OPCODE_RCP: { - out = instr->rcp(inputs[0]); - } - break; - case TGSI_OPCODE_RSQ: { - out = instr->rsq(inputs[0]); - } - break; - case TGSI_OPCODE_EXP: - break; - case TGSI_OPCODE_LOG: - break; - case TGSI_OPCODE_MUL: { - out = instr->mul(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_ADD: { - out = instr->add(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_DP3: { - out = instr->dp3(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_DP4: { - out = instr->dp4(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_DST: { - out = instr->dst(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_MIN: { - out = instr->min(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_MAX: { - out = instr->max(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_SLT: { - out = instr->slt(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_SGE: { - out = instr->sge(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_MAD: { - out = instr->madd(inputs[0], inputs[1], inputs[2]); - } - break; - case TGSI_OPCODE_SUB: { - out = instr->sub(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_LERP: { - out = instr->lerp(inputs[0], inputs[1], inputs[2]); - } - break; - case TGSI_OPCODE_CND: - break; - case TGSI_OPCODE_CND0: - break; - case TGSI_OPCODE_DOT2ADD: - break; - case TGSI_OPCODE_INDEX: - break; - case TGSI_OPCODE_NEGATE: - break; - case TGSI_OPCODE_FRAC: { - out = instr->frc(inputs[0]); - } - break; - case TGSI_OPCODE_CLAMP: - break; - case TGSI_OPCODE_FLOOR: { - out = instr->floor(inputs[0]); - } - break; - case TGSI_OPCODE_ROUND: - break; - case TGSI_OPCODE_EXPBASE2: { - out = instr->ex2(inputs[0]); - } - break; - case TGSI_OPCODE_LOGBASE2: { - out = instr->lg2(inputs[0]); - } - break; - case TGSI_OPCODE_POWER: { - out = instr->pow(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_CROSSPRODUCT: { - out = instr->cross(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_MULTIPLYMATRIX: - break; - case TGSI_OPCODE_ABS: { - out = instr->abs(inputs[0]); - } - break; - case TGSI_OPCODE_RCC: - break; - case TGSI_OPCODE_DPH: { - out = instr->dph(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_COS: { - out = instr->cos(inputs[0]); - } - break; - case TGSI_OPCODE_DDX: - break; - case TGSI_OPCODE_DDY: - break; - case TGSI_OPCODE_KILP: { - out = instr->kilp(inputs[0]); - storage->setKilElement(out); - return; - } - break; - case TGSI_OPCODE_PK2H: - break; - case TGSI_OPCODE_PK2US: - break; - case TGSI_OPCODE_PK4B: - break; - case TGSI_OPCODE_PK4UB: - break; - case TGSI_OPCODE_RFL: - break; - case TGSI_OPCODE_SEQ: - break; - case TGSI_OPCODE_SFL: - break; - case TGSI_OPCODE_SGT: { - out = instr->sgt(inputs[0], inputs[1]); - } - break; - case TGSI_OPCODE_SIN: { - out = instr->sin(inputs[0]); - } - break; - case TGSI_OPCODE_SLE: - break; - case TGSI_OPCODE_SNE: - break; - case TGSI_OPCODE_STR: - break; - case TGSI_OPCODE_TEX: - break; - case TGSI_OPCODE_TXD: - break; - case TGSI_OPCODE_UP2H: - break; - case TGSI_OPCODE_UP2US: - break; - case TGSI_OPCODE_UP4B: - break; - case TGSI_OPCODE_UP4UB: - break; - case TGSI_OPCODE_X2D: - break; - case TGSI_OPCODE_ARA: - break; - case TGSI_OPCODE_ARR: - break; - case TGSI_OPCODE_BRA: - break; - case TGSI_OPCODE_CAL: { - instr->cal(inst->InstructionExtLabel.Label, storage->inputPtr()); - return; - } - break; - case TGSI_OPCODE_RET: { - instr->end(); - return; - } - break; - case TGSI_OPCODE_SSG: - break; - case TGSI_OPCODE_CMP: { - out = instr->cmp(inputs[0], inputs[1], inputs[2]); - } - break; - case TGSI_OPCODE_SCS: { - out = instr->scs(inputs[0]); - } - break; - case TGSI_OPCODE_TXB: - break; - case TGSI_OPCODE_NRM: - break; - case TGSI_OPCODE_DIV: - break; - case TGSI_OPCODE_DP2: - break; - case TGSI_OPCODE_TXL: - break; - case TGSI_OPCODE_BRK: { - instr->brk(); - return; - } - break; - case TGSI_OPCODE_IF: { - instr->ifop(inputs[0]); - storage->setCurrentBlock(instr->currentBlock()); - return; //just update the state - } - break; - case TGSI_OPCODE_LOOP: - break; - case TGSI_OPCODE_REP: - break; - case TGSI_OPCODE_ELSE: { - instr->elseop(); - storage->setCurrentBlock(instr->currentBlock()); - return; //only state update - } - break; - case TGSI_OPCODE_ENDIF: { - instr->endif(); - storage->setCurrentBlock(instr->currentBlock()); - return; //just update the state - } - break; - case TGSI_OPCODE_ENDLOOP: - break; - case TGSI_OPCODE_ENDREP: - break; - case TGSI_OPCODE_PUSHA: - break; - case TGSI_OPCODE_POPA: - break; - case TGSI_OPCODE_CEIL: - break; - case TGSI_OPCODE_I2F: - break; - case TGSI_OPCODE_NOT: - break; - case TGSI_OPCODE_TRUNC: { - out = instr->trunc(inputs[0]); - } - break; - case TGSI_OPCODE_SHL: - break; - case TGSI_OPCODE_SHR: - break; - case TGSI_OPCODE_AND: - break; - case TGSI_OPCODE_OR: - break; - case TGSI_OPCODE_MOD: - break; - case TGSI_OPCODE_XOR: - break; - case TGSI_OPCODE_SAD: - break; - case TGSI_OPCODE_TXF: - break; - case TGSI_OPCODE_TXQ: - break; - case TGSI_OPCODE_CONT: - break; - case TGSI_OPCODE_EMIT: - break; - case TGSI_OPCODE_ENDPRIM: - break; - case TGSI_OPCODE_BGNLOOP2: { - instr->beginLoop(); - storage->setCurrentBlock(instr->currentBlock()); - return; - } - break; - case TGSI_OPCODE_BGNSUB: { - instr->bgnSub(instno); - storage->setCurrentBlock(instr->currentBlock()); - storage->pushTemps(); - return; - } - break; - case TGSI_OPCODE_ENDLOOP2: { - instr->endLoop(); - storage->setCurrentBlock(instr->currentBlock()); - return; - } - break; - case TGSI_OPCODE_ENDSUB: { - instr->endSub(); - storage->setCurrentBlock(instr->currentBlock()); - storage->popArguments(); - storage->popTemps(); - return; - } - break; - case TGSI_OPCODE_NOISE1: - break; - case TGSI_OPCODE_NOISE2: - break; - case TGSI_OPCODE_NOISE3: - break; - case TGSI_OPCODE_NOISE4: - break; - case TGSI_OPCODE_NOP: - break; - case TGSI_OPCODE_TEXBEM: - break; - case TGSI_OPCODE_TEXBEML: - break; - case TGSI_OPCODE_TEXREG2AR: - break; - case TGSI_OPCODE_TEXM3X2PAD: - break; - case TGSI_OPCODE_TEXM3X2TEX: - break; - case TGSI_OPCODE_TEXM3X3PAD: - break; - case TGSI_OPCODE_TEXM3X3TEX: - break; - case TGSI_OPCODE_TEXM3X3SPEC: - break; - case TGSI_OPCODE_TEXM3X3VSPEC: - break; - case TGSI_OPCODE_TEXREG2GB: - break; - case TGSI_OPCODE_TEXREG2RGB: - break; - case TGSI_OPCODE_TEXDP3TEX: - break; - case TGSI_OPCODE_TEXDP3: - break; - case TGSI_OPCODE_TEXM3X3: - break; - case TGSI_OPCODE_TEXM3X2DEPTH: - break; - case TGSI_OPCODE_TEXDEPTH: - break; - case TGSI_OPCODE_BEM: - break; - case TGSI_OPCODE_M4X3: - break; - case TGSI_OPCODE_M3X4: - break; - case TGSI_OPCODE_M3X3: - break; - case TGSI_OPCODE_M3X2: - break; - case TGSI_OPCODE_NRM4: - break; - case TGSI_OPCODE_CALLNZ: - break; - case TGSI_OPCODE_IFC: - break; - case TGSI_OPCODE_BREAKC: - break; - case TGSI_OPCODE_KIL: - break; - case TGSI_OPCODE_END: - instr->end(); - return; - break; - default: - fprintf(stderr, "ERROR: Unknown opcode %d\n", - inst->Instruction.Opcode); - assert(0); - break; - } - - if (!out) { - fprintf(stderr, "ERROR: unsupported opcode %d\n", - inst->Instruction.Opcode); - assert(!"Unsupported opcode"); - } - - /* # not sure if we need this */ - switch( inst->Instruction.Saturate ) { - case TGSI_SAT_NONE: - break; - case TGSI_SAT_ZERO_ONE: - /*TXT( "_SAT" );*/ - break; - case TGSI_SAT_MINUS_PLUS_ONE: - /*TXT( "_SAT[-1,1]" );*/ - break; - default: - assert( 0 ); - } - - /* store results */ - for (int i = 0; i < inst->Instruction.NumDstRegs; ++i) { - struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i]; - - if (dst->DstRegister.File == TGSI_FILE_OUTPUT) { - storage->setOutputElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); - } else if (dst->DstRegister.File == TGSI_FILE_TEMPORARY) { - storage->setTempElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); - } else if (dst->DstRegister.File == TGSI_FILE_ADDRESS) { - storage->setAddrElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); - } else { - fprintf(stderr, "ERROR: unsupported LLVM destination!"); - assert(!"wrong destination"); - } - } -} - -static llvm::Module * -tgsi_to_llvm(struct gallivm_prog *prog, const struct tgsi_token *tokens) -{ - llvm::Module *mod = createBaseShader(); - struct tgsi_parse_context parse; - struct tgsi_full_instruction fi; - struct tgsi_full_declaration fd; - unsigned instno = 0; - Function* shader = mod->getFunction("execute_shader"); - std::ostringstream stream; - if (prog->type == GALLIVM_VS) { - stream << "vs_shader"; - } else { - stream << "fs_shader"; - } - stream << prog->id; - std::string func_name = stream.str(); - shader->setName(func_name.c_str()); - - Function::arg_iterator args = shader->arg_begin(); - Value *ptr_INPUT = args++; - ptr_INPUT->setName("input"); - - BasicBlock *label_entry = new BasicBlock("entry", shader, 0); - - tgsi_parse_init(&parse, tokens); - - fi = tgsi_default_full_instruction(); - fd = tgsi_default_full_declaration(); - Storage storage(label_entry, ptr_INPUT); - Instructions instr(mod, shader, label_entry, &storage); - while(!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - translate_declaration(prog, mod, &storage, - &parse.FullToken.FullDeclaration, - &fd); - break; - - case TGSI_TOKEN_TYPE_IMMEDIATE: - translate_immediate(&storage, - &parse.FullToken.FullImmediate); - break; - - case TGSI_TOKEN_TYPE_INSTRUCTION: - translate_instruction(mod, &storage, &instr, - &parse.FullToken.FullInstruction, - &fi, instno); - ++instno; - break; - - default: - assert(0); - } - } - - tgsi_parse_free(&parse); - - prog->num_consts = storage.numConsts(); - return mod; -} - -/*! - Translates the TGSI tokens into LLVM format. Translated representation - is stored in the gallivm_prog and returned. - After calling this function the gallivm_prog can either be used with a custom - code generator to generate machine code for the GPU which the code generator - addresses or it can be jit compiled with gallivm_cpu_jit_compile and executed - with gallivm_prog_exec to run the module on the CPU. - */ -struct gallivm_prog * -gallivm_from_tgsi(const struct tgsi_token *tokens, enum gallivm_shader_type type) -{ - std::cout << "Creating llvm from: " <id = GLOBAL_ID; - gallivm->type = type; - tgsi_dump(tokens, 0); - - llvm::Module *mod = tgsi_to_llvm(gallivm, tokens); - gallivm->module = mod; - gallivm_prog_dump(gallivm, 0); - - /* Run optimization passes over it */ - PassManager passes; - passes.add(new TargetData(mod)); - AddStandardCompilePasses(passes); - passes.run(*mod); - - gallivm->module = mod; - - gallivm_prog_dump(gallivm, 0); - - return gallivm; -} - - void gallivm_prog_delete(struct gallivm_prog *prog) { - llvm::Module *mod = static_cast(prog->module); - delete mod; + delete prog->module; prog->module = 0; prog->function = 0; free(prog); } -typedef void (*vertex_shader_runner)(float (*ainputs)[PIPE_MAX_SHADER_INPUTS][4], - float (*dests)[PIPE_MAX_SHADER_INPUTS][4], +typedef void (*vertex_shader_runner)(void *ainputs, + void *dests, float (*aconsts)[4], int num_vertices, int num_inputs, @@ -850,8 +160,8 @@ typedef void (*vertex_shader_runner)(float (*ainputs)[PIPE_MAX_SHADER_INPUTS][4] function. */ int gallivm_prog_exec(struct gallivm_prog *prog, - float (*inputs)[PIPE_MAX_SHADER_INPUTS][4], - float (*dests)[PIPE_MAX_SHADER_INPUTS][4], + struct tgsi_exec_vector *inputs, + struct tgsi_exec_vector *dests, float (*consts)[4], int num_vertices, int num_inputs, @@ -943,18 +253,15 @@ int gallivm_fragment_shader_exec(struct gallivm_prog *prog, samplers); } -void gallivm_prog_dump(struct gallivm_prog *prog, const char *file_prefix) +void gallivm_ir_dump(struct gallivm_ir *ir, const char *file_prefix) { - llvm::Module *mod; - if (!prog || !prog->module) + if (!ir || !ir->module) return; - mod = static_cast(prog->module); - if (file_prefix) { std::ostringstream stream; stream << file_prefix; - stream << prog->id; + stream << ir->id; stream << ".ll"; std::string name = stream.str(); std::ofstream out(name.c_str()); @@ -962,12 +269,12 @@ void gallivm_prog_dump(struct gallivm_prog *prog, const char *file_prefix) std::cerr<<"Can't open file : "<module); out.close(); } else { - const llvm::Module::FunctionListType &funcs = mod->getFunctionList(); + const llvm::Module::FunctionListType &funcs = ir->module->getFunctionList(); llvm::Module::FunctionListType::const_iterator itr; - std::cout<<"; ---------- Start shader "<id<id<id<id<id = GLOBAL_ID; + ir->type = type; + return ir; +} +void gallivm_ir_set_layout(struct gallivm_ir *ir, + enum gallivm_vector_layout layout) +{ + ir->layout = layout; +} +void gallivm_ir_set_components(struct gallivm_ir *ir, int num) +{ + ir->num_components = num; +} +void gallivm_ir_fill_from_tgsi(struct gallivm_ir *ir, + const struct tgsi_token *tokens) +{ + std::cout << "Creating llvm from: " <module = mod; + gallivm_ir_dump(ir, 0); +} + +void gallivm_ir_delete(struct gallivm_ir *ir) +{ + delete ir->module; + free(ir); +} + +struct gallivm_prog * gallivm_ir_compile(struct gallivm_ir *ir) +{ + struct gallivm_prog *prog = + (struct gallivm_prog *)calloc(1, sizeof(struct gallivm_prog)); + llvm::Module *mod = llvm::CloneModule(ir->module); + prog->num_consts = ir->num_consts; + memcpy(prog->interpolators, ir->interpolators, sizeof(prog->interpolators)); + prog->num_interp = ir->num_interp; + + /* Run optimization passes over it */ + PassManager passes; + passes.add(new TargetData(mod)); + AddStandardCompilePasses(passes); + passes.run(*mod); + prog->module = mod; + + return prog; +} + +#endif /* MESA_LLVM */ diff --git a/src/mesa/pipe/llvm/gallivm.h b/src/mesa/pipe/llvm/gallivm.h index 4695de3127a..f9f5d5ee744 100644 --- a/src/mesa/pipe/llvm/gallivm.h +++ b/src/mesa/pipe/llvm/gallivm.h @@ -43,21 +43,36 @@ extern "C" { struct tgsi_token; +struct gallivm_ir; struct gallivm_prog; struct gallivm_cpu_engine; struct tgsi_interp_coef; struct tgsi_sampler; +struct tgsi_exec_vector; enum gallivm_shader_type { GALLIVM_VS, GALLIVM_FS }; -struct gallivm_prog *gallivm_from_tgsi(const struct tgsi_token *tokens, enum gallivm_shader_type type); -void gallivm_prog_delete(struct gallivm_prog *prog); +enum gallivm_vector_layout { + GALLIVM_AOS, + GALLIVM_SOA +}; + +struct gallivm_ir *gallivm_ir_new(enum gallivm_shader_type type); +void gallivm_ir_set_layout(struct gallivm_ir *prog, + enum gallivm_vector_layout layout); +void gallivm_ir_set_components(struct gallivm_ir *prog, int num); +void gallivm_ir_fill_from_tgsi(struct gallivm_ir *prog, + const struct tgsi_token *tokens); +void gallivm_ir_delete(struct gallivm_ir *prog); + +struct gallivm_prog *gallivm_ir_compile(struct gallivm_ir *ir); + int gallivm_prog_exec(struct gallivm_prog *prog, - float (*inputs)[PIPE_MAX_SHADER_INPUTS][4], - float (*dests)[PIPE_MAX_SHADER_INPUTS][4], + struct tgsi_exec_vector *inputs, + struct tgsi_exec_vector *dests, float (*consts)[4], int num_vertices, int num_inputs, diff --git a/src/mesa/pipe/llvm/gallivm_p.h b/src/mesa/pipe/llvm/gallivm_p.h new file mode 100644 index 00000000000..2c6e5e8f5f2 --- /dev/null +++ b/src/mesa/pipe/llvm/gallivm_p.h @@ -0,0 +1,56 @@ +#ifndef GALLIVM_P_H +#define GALLIVM_P_H + +#ifdef MESA_LLVM + +namespace llvm { + class Module; +} + +#if defined __cplusplus +extern "C" { +#endif + +enum gallivm_shader_type; +enum gallivm_vector_layout; + +struct gallivm_interpolate { + int attrib; + int chan; + int type; +}; + +struct gallivm_ir { + llvm::Module *module; + int id; + enum gallivm_shader_type type; + enum gallivm_vector_layout layout; + int num_components; + int num_consts; + + //FIXME: this might not be enough for some shaders + struct gallivm_interpolate interpolators[32*4]; + int num_interp; +}; + +struct gallivm_prog { + llvm::Module *module; + void *function; + + int id; + enum gallivm_shader_type type; + + int num_consts; + + //FIXME: this might not be enough for some shaders + struct gallivm_interpolate interpolators[32*4]; + int num_interp; +}; + +#endif /* MESA_LLVM */ + +#if defined __cplusplus +} // extern "C" +#endif + +#endif diff --git a/src/mesa/pipe/llvm/llvm_base_shader.cpp b/src/mesa/pipe/llvm/llvm_base_shader.cpp index b574b550aec..90a25a440ab 100644 --- a/src/mesa/pipe/llvm/llvm_base_shader.cpp +++ b/src/mesa/pipe/llvm/llvm_base_shader.cpp @@ -104,8 +104,8 @@ Module* createBaseShader() { /*isVarArg=*/false); std::vectorFuncTy_18_args; - FuncTy_18_args.push_back(PointerTy_12); - FuncTy_18_args.push_back(PointerTy_12); + FuncTy_18_args.push_back(PointerTy_9); + FuncTy_18_args.push_back(PointerTy_9); FuncTy_18_args.push_back(PointerTy_7); FuncTy_18_args.push_back(IntegerType::get(32)); FuncTy_18_args.push_back(IntegerType::get(32)); @@ -526,14 +526,23 @@ Module* createBaseShader() { BasicBlock* label_forbody_preheader_i = new BasicBlock("forbody.preheader.i",func_run_vertex_shader,0); BasicBlock* label_forbody_i = new BasicBlock("forbody.i",func_run_vertex_shader,0); BasicBlock* label_from_consts_exit = new BasicBlock("from_consts.exit",func_run_vertex_shader,0); - BasicBlock* label_forbody_preheader_91 = new BasicBlock("forbody.preheader",func_run_vertex_shader,0); - BasicBlock* label_forbody_92 = new BasicBlock("forbody",func_run_vertex_shader,0); - BasicBlock* label_afterfor_93 = new BasicBlock("afterfor",func_run_vertex_shader,0); // Block entry (label_entry_90) AllocaInst* ptr_consts = new AllocaInst(ArrayTy_20, "consts", label_entry_90); AllocaInst* ptr_temps = new AllocaInst(ArrayTy_22, "temps", label_entry_90); AllocaInst* ptr_args = new AllocaInst(StructTy_struct_ShaderInput, "args", label_entry_90); + std::vector ptr_tmp_indices; + ptr_tmp_indices.push_back(const_int32_29); + ptr_tmp_indices.push_back(const_int32_29); + Instruction* ptr_tmp = new GetElementPtrInst(ptr_args, ptr_tmp_indices.begin(), ptr_tmp_indices.end(), "tmp", label_entry_90); + CastInst* ptr_conv = new BitCastInst(ptr_results, PointerTy_0, "conv", label_entry_90); + StoreInst* void_91 = new StoreInst(ptr_conv, ptr_tmp, false, label_entry_90); + std::vector ptr_tmp2_indices; + ptr_tmp2_indices.push_back(const_int32_29); + ptr_tmp2_indices.push_back(const_int32_31); + Instruction* ptr_tmp2 = new GetElementPtrInst(ptr_args, ptr_tmp2_indices.begin(), ptr_tmp2_indices.end(), "tmp2", label_entry_90); + CastInst* ptr_conv4 = new BitCastInst(ptr_inputs, PointerTy_0, "conv4", label_entry_90); + StoreInst* void_92 = new StoreInst(ptr_conv4, ptr_tmp2, false, label_entry_90); ICmpInst* int1_cmp_i = new ICmpInst(ICmpInst::ICMP_SGT, int32_num_consts, const_int32_29, "cmp.i", label_entry_90); new BranchInst(label_forbody_preheader_i, label_from_consts_exit, int1_cmp_i, label_entry_90); @@ -544,17 +553,17 @@ Module* createBaseShader() { new BranchInst(label_forbody_i, label_forbody_preheader_i); // Block forbody.i (label_forbody_i) - Argument* fwdref_96 = new Argument(IntegerType::get(32)); + Argument* fwdref_95 = new Argument(IntegerType::get(32)); PHINode* int32_i_0_reg2mem_0_i = new PHINode(IntegerType::get(32), "i.0.reg2mem.0.i", label_forbody_i); int32_i_0_reg2mem_0_i->reserveOperandSpace(2); int32_i_0_reg2mem_0_i->addIncoming(const_int32_29, label_forbody_preheader_i); - int32_i_0_reg2mem_0_i->addIncoming(fwdref_96, label_forbody_i); + int32_i_0_reg2mem_0_i->addIncoming(fwdref_95, label_forbody_i); - Argument* fwdref_97 = new Argument(VectorTy_1); + Argument* fwdref_96 = new Argument(VectorTy_1); PHINode* packed_vec_0_reg2mem_0_i = new PHINode(VectorTy_1, "vec.0.reg2mem.0.i", label_forbody_i); packed_vec_0_reg2mem_0_i->reserveOperandSpace(2); packed_vec_0_reg2mem_0_i->addIncoming(const_packed_32, label_forbody_preheader_i); - packed_vec_0_reg2mem_0_i->addIncoming(fwdref_97, label_forbody_i); + packed_vec_0_reg2mem_0_i->addIncoming(fwdref_96, label_forbody_i); std::vector ptr_arraydecay_i_indices; ptr_arraydecay_i_indices.push_back(int32_i_0_reg2mem_0_i); @@ -584,80 +593,40 @@ Module* createBaseShader() { ptr_arrayidx34_i_indices.push_back(const_int32_29); ptr_arrayidx34_i_indices.push_back(int32_i_0_reg2mem_0_i); Instruction* ptr_arrayidx34_i = new GetElementPtrInst(ptr_consts, ptr_arrayidx34_i_indices.begin(), ptr_arrayidx34_i_indices.end(), "arrayidx34.i", label_forbody_i); - StoreInst* void_98 = new StoreInst(packed_tmp31_i, ptr_arrayidx34_i, false, label_forbody_i); - BinaryOperator* int32_indvar_next8 = BinaryOperator::create(Instruction::Add, int32_i_0_reg2mem_0_i, const_int32_31, "indvar.next8", label_forbody_i); - ICmpInst* int1_exitcond9 = new ICmpInst(ICmpInst::ICMP_EQ, int32_indvar_next8, int32_tmp10_i, "exitcond9", label_forbody_i); - new BranchInst(label_from_consts_exit, label_forbody_i, int1_exitcond9, label_forbody_i); + StoreInst* void_97 = new StoreInst(packed_tmp31_i, ptr_arrayidx34_i, false, label_forbody_i); + BinaryOperator* int32_indvar_next_98 = BinaryOperator::create(Instruction::Add, int32_i_0_reg2mem_0_i, const_int32_31, "indvar.next", label_forbody_i); + ICmpInst* int1_exitcond_99 = new ICmpInst(ICmpInst::ICMP_EQ, int32_indvar_next_98, int32_tmp10_i, "exitcond", label_forbody_i); + new BranchInst(label_from_consts_exit, label_forbody_i, int1_exitcond_99, label_forbody_i); // Block from_consts.exit (label_from_consts_exit) - std::vector ptr_tmp2_indices; - ptr_tmp2_indices.push_back(const_int32_29); - ptr_tmp2_indices.push_back(const_int32_34); - Instruction* ptr_tmp2 = new GetElementPtrInst(ptr_args, ptr_tmp2_indices.begin(), ptr_tmp2_indices.end(), "tmp2", label_from_consts_exit); - std::vector ptr_arraydecay3_indices; - ptr_arraydecay3_indices.push_back(const_int32_29); - ptr_arraydecay3_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay3 = new GetElementPtrInst(ptr_consts, ptr_arraydecay3_indices.begin(), ptr_arraydecay3_indices.end(), "arraydecay3", label_from_consts_exit); - StoreInst* void_100 = new StoreInst(ptr_arraydecay3, ptr_tmp2, false, label_from_consts_exit); - std::vector ptr_tmp4_indices; - ptr_tmp4_indices.push_back(const_int32_29); - ptr_tmp4_indices.push_back(const_int32_33); - Instruction* ptr_tmp4 = new GetElementPtrInst(ptr_args, ptr_tmp4_indices.begin(), ptr_tmp4_indices.end(), "tmp4", label_from_consts_exit); - std::vector ptr_arraydecay5_indices; - ptr_arraydecay5_indices.push_back(const_int32_29); - ptr_arraydecay5_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay5 = new GetElementPtrInst(ptr_temps, ptr_arraydecay5_indices.begin(), ptr_arraydecay5_indices.end(), "arraydecay5", label_from_consts_exit); - StoreInst* void_101 = new StoreInst(ptr_arraydecay5, ptr_tmp4, false, label_from_consts_exit); - ICmpInst* int1_cmp_102 = new ICmpInst(ICmpInst::ICMP_SGT, int32_num_vertices, const_int32_29, "cmp", label_from_consts_exit); - new BranchInst(label_forbody_preheader_91, label_afterfor_93, int1_cmp_102, label_from_consts_exit); - - // Block forbody.preheader (label_forbody_preheader_91) - std::vector ptr_tmp8_indices; - ptr_tmp8_indices.push_back(const_int32_29); - ptr_tmp8_indices.push_back(const_int32_29); - Instruction* ptr_tmp8 = new GetElementPtrInst(ptr_args, ptr_tmp8_indices.begin(), ptr_tmp8_indices.end(), "tmp8", label_forbody_preheader_91); - std::vector ptr_tmp12_indices; - ptr_tmp12_indices.push_back(const_int32_29); - ptr_tmp12_indices.push_back(const_int32_31); - Instruction* ptr_tmp12 = new GetElementPtrInst(ptr_args, ptr_tmp12_indices.begin(), ptr_tmp12_indices.end(), "tmp12", label_forbody_preheader_91); - BinaryOperator* int32_tmp_104 = BinaryOperator::create(Instruction::Add, int32_num_vertices, const_int32_30, "tmp", label_forbody_preheader_91); - ICmpInst* int1_tmp6 = new ICmpInst(ICmpInst::ICMP_SLT, int32_tmp_104, const_int32_29, "tmp6", label_forbody_preheader_91); - SelectInst* int32_tmp7 = new SelectInst(int1_tmp6, const_int32_31, int32_num_vertices, "tmp7", label_forbody_preheader_91); - new BranchInst(label_forbody_92, label_forbody_preheader_91); - - // Block forbody (label_forbody_92) - Argument* fwdref_107 = new Argument(IntegerType::get(32)); - PHINode* int32_i_0_reg2mem_0_106 = new PHINode(IntegerType::get(32), "i.0.reg2mem.0", label_forbody_92); - int32_i_0_reg2mem_0_106->reserveOperandSpace(2); - int32_i_0_reg2mem_0_106->addIncoming(const_int32_29, label_forbody_preheader_91); - int32_i_0_reg2mem_0_106->addIncoming(fwdref_107, label_forbody_92); - - std::vector ptr_arraydecay11_108_indices; - ptr_arraydecay11_108_indices.push_back(int32_i_0_reg2mem_0_106); - ptr_arraydecay11_108_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay11_108 = new GetElementPtrInst(ptr_results, ptr_arraydecay11_108_indices.begin(), ptr_arraydecay11_108_indices.end(), "arraydecay11", label_forbody_92); - StoreInst* void_109 = new StoreInst(ptr_arraydecay11_108, ptr_tmp8, false, label_forbody_92); - std::vector ptr_arraydecay16_indices; - ptr_arraydecay16_indices.push_back(int32_i_0_reg2mem_0_106); - ptr_arraydecay16_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay16 = new GetElementPtrInst(ptr_inputs, ptr_arraydecay16_indices.begin(), ptr_arraydecay16_indices.end(), "arraydecay16", label_forbody_92); - StoreInst* void_110 = new StoreInst(ptr_arraydecay16, ptr_tmp12, false, label_forbody_92); - CallInst* void_111 = new CallInst(func_execute_shader, ptr_args, "", label_forbody_92); - void_111->setCallingConv(CallingConv::C); - void_111->setTailCall(false);const ParamAttrsList *void_111_PAL = 0; - void_111->setParamAttrs(void_111_PAL); - - BinaryOperator* int32_indvar_next_112 = BinaryOperator::create(Instruction::Add, int32_i_0_reg2mem_0_106, const_int32_31, "indvar.next", label_forbody_92); - ICmpInst* int1_exitcond_113 = new ICmpInst(ICmpInst::ICMP_EQ, int32_indvar_next_112, int32_tmp7, "exitcond", label_forbody_92); - new BranchInst(label_afterfor_93, label_forbody_92, int1_exitcond_113, label_forbody_92); - - // Block afterfor (label_afterfor_93) - new ReturnInst(label_afterfor_93); + std::vector ptr_tmp7_indices; + ptr_tmp7_indices.push_back(const_int32_29); + ptr_tmp7_indices.push_back(const_int32_34); + Instruction* ptr_tmp7 = new GetElementPtrInst(ptr_args, ptr_tmp7_indices.begin(), ptr_tmp7_indices.end(), "tmp7", label_from_consts_exit); + std::vector ptr_arraydecay8_indices; + ptr_arraydecay8_indices.push_back(const_int32_29); + ptr_arraydecay8_indices.push_back(const_int32_29); + Instruction* ptr_arraydecay8 = new GetElementPtrInst(ptr_consts, ptr_arraydecay8_indices.begin(), ptr_arraydecay8_indices.end(), "arraydecay8", label_from_consts_exit); + StoreInst* void_101 = new StoreInst(ptr_arraydecay8, ptr_tmp7, false, label_from_consts_exit); + std::vector ptr_tmp9_indices; + ptr_tmp9_indices.push_back(const_int32_29); + ptr_tmp9_indices.push_back(const_int32_33); + Instruction* ptr_tmp9 = new GetElementPtrInst(ptr_args, ptr_tmp9_indices.begin(), ptr_tmp9_indices.end(), "tmp9", label_from_consts_exit); + std::vector ptr_arraydecay10_indices; + ptr_arraydecay10_indices.push_back(const_int32_29); + ptr_arraydecay10_indices.push_back(const_int32_29); + Instruction* ptr_arraydecay10 = new GetElementPtrInst(ptr_temps, ptr_arraydecay10_indices.begin(), ptr_arraydecay10_indices.end(), "arraydecay10", label_from_consts_exit); + StoreInst* void_102 = new StoreInst(ptr_arraydecay10, ptr_tmp9, false, label_from_consts_exit); + CallInst* void_103 = new CallInst(func_execute_shader, ptr_args, "", label_from_consts_exit); + void_103->setCallingConv(CallingConv::C); + void_103->setTailCall(false);const ParamAttrsList *void_103_PAL = 0; + void_103->setParamAttrs(void_103_PAL); + + new ReturnInst(label_from_consts_exit); // Resolve Forward References - fwdref_107->replaceAllUsesWith(int32_indvar_next_112); delete fwdref_107; - fwdref_97->replaceAllUsesWith(packed_tmp31_i); delete fwdref_97; - fwdref_96->replaceAllUsesWith(int32_indvar_next8); delete fwdref_96; + fwdref_96->replaceAllUsesWith(packed_tmp31_i); delete fwdref_96; + fwdref_95->replaceAllUsesWith(int32_indvar_next_98); delete fwdref_95; } @@ -668,195 +637,195 @@ Module* createBaseShader() { float_x->setName("x"); Value* float_y = args++; float_y->setName("y"); - Value* ptr_results_116 = args++; - ptr_results_116->setName("results"); - Value* ptr_inputs_117 = args++; - ptr_inputs_117->setName("inputs"); - Value* int32_num_inputs_118 = args++; - int32_num_inputs_118->setName("num_inputs"); - Value* ptr_aconsts_119 = args++; - ptr_aconsts_119->setName("aconsts"); - Value* int32_num_consts_120 = args++; - int32_num_consts_120->setName("num_consts"); + Value* ptr_results_105 = args++; + ptr_results_105->setName("results"); + Value* ptr_inputs_106 = args++; + ptr_inputs_106->setName("inputs"); + Value* int32_num_inputs_107 = args++; + int32_num_inputs_107->setName("num_inputs"); + Value* ptr_aconsts_108 = args++; + ptr_aconsts_108->setName("aconsts"); + Value* int32_num_consts_109 = args++; + int32_num_consts_109->setName("num_consts"); Value* ptr_samplers = args++; ptr_samplers->setName("samplers"); - BasicBlock* label_entry_121 = new BasicBlock("entry",func_run_fragment_shader,0); - BasicBlock* label_forbody_preheader_i_122 = new BasicBlock("forbody.preheader.i",func_run_fragment_shader,0); - BasicBlock* label_forbody_i_123 = new BasicBlock("forbody.i",func_run_fragment_shader,0); - BasicBlock* label_from_consts_exit_124 = new BasicBlock("from_consts.exit",func_run_fragment_shader,0); - - // Block entry (label_entry_121) - AllocaInst* ptr_consts_125 = new AllocaInst(ArrayTy_20, "consts", label_entry_121); - AllocaInst* ptr_temps_126 = new AllocaInst(ArrayTy_22, "temps", label_entry_121); - AllocaInst* ptr_args_127 = new AllocaInst(StructTy_struct_ShaderInput, "args", label_entry_121); - std::vector ptr_tmp_indices; - ptr_tmp_indices.push_back(const_int32_29); - ptr_tmp_indices.push_back(const_int32_35); - Instruction* ptr_tmp = new GetElementPtrInst(ptr_args_127, ptr_tmp_indices.begin(), ptr_tmp_indices.end(), "tmp", label_entry_121); - StoreInst* void_128 = new StoreInst(const_int32_29, ptr_tmp, false, label_entry_121); - ICmpInst* int1_cmp_i_129 = new ICmpInst(ICmpInst::ICMP_SGT, int32_num_consts_120, const_int32_29, "cmp.i", label_entry_121); - new BranchInst(label_forbody_preheader_i_122, label_from_consts_exit_124, int1_cmp_i_129, label_entry_121); - - // Block forbody.preheader.i (label_forbody_preheader_i_122) - BinaryOperator* int32_tmp_i_131 = BinaryOperator::create(Instruction::Add, int32_num_consts_120, const_int32_30, "tmp.i", label_forbody_preheader_i_122); - ICmpInst* int1_tmp9_i_132 = new ICmpInst(ICmpInst::ICMP_SLT, int32_tmp_i_131, const_int32_29, "tmp9.i", label_forbody_preheader_i_122); - SelectInst* int32_tmp10_i_133 = new SelectInst(int1_tmp9_i_132, const_int32_31, int32_num_consts_120, "tmp10.i", label_forbody_preheader_i_122); - new BranchInst(label_forbody_i_123, label_forbody_preheader_i_122); - - // Block forbody.i (label_forbody_i_123) - Argument* fwdref_136 = new Argument(IntegerType::get(32)); - PHINode* int32_i_0_reg2mem_0_i_135 = new PHINode(IntegerType::get(32), "i.0.reg2mem.0.i", label_forbody_i_123); - int32_i_0_reg2mem_0_i_135->reserveOperandSpace(2); - int32_i_0_reg2mem_0_i_135->addIncoming(const_int32_29, label_forbody_preheader_i_122); - int32_i_0_reg2mem_0_i_135->addIncoming(fwdref_136, label_forbody_i_123); - - Argument* fwdref_138 = new Argument(VectorTy_1); - PHINode* packed_vec_0_reg2mem_0_i_137 = new PHINode(VectorTy_1, "vec.0.reg2mem.0.i", label_forbody_i_123); - packed_vec_0_reg2mem_0_i_137->reserveOperandSpace(2); - packed_vec_0_reg2mem_0_i_137->addIncoming(const_packed_32, label_forbody_preheader_i_122); - packed_vec_0_reg2mem_0_i_137->addIncoming(fwdref_138, label_forbody_i_123); - - std::vector ptr_arraydecay_i_139_indices; - ptr_arraydecay_i_139_indices.push_back(int32_i_0_reg2mem_0_i_135); - ptr_arraydecay_i_139_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay_i_139 = new GetElementPtrInst(ptr_aconsts_119, ptr_arraydecay_i_139_indices.begin(), ptr_arraydecay_i_139_indices.end(), "arraydecay.i", label_forbody_i_123); - LoadInst* float_tmp5_i_140 = new LoadInst(ptr_arraydecay_i_139, "tmp5.i", false, label_forbody_i_123); - InsertElementInst* packed_tmp7_i_141 = new InsertElementInst(packed_vec_0_reg2mem_0_i_137, float_tmp5_i_140, const_int32_29, "tmp7.i", label_forbody_i_123); - std::vector ptr_arrayidx12_i_142_indices; - ptr_arrayidx12_i_142_indices.push_back(int32_i_0_reg2mem_0_i_135); - ptr_arrayidx12_i_142_indices.push_back(const_int32_31); - Instruction* ptr_arrayidx12_i_142 = new GetElementPtrInst(ptr_aconsts_119, ptr_arrayidx12_i_142_indices.begin(), ptr_arrayidx12_i_142_indices.end(), "arrayidx12.i", label_forbody_i_123); - LoadInst* float_tmp13_i_143 = new LoadInst(ptr_arrayidx12_i_142, "tmp13.i", false, label_forbody_i_123); - InsertElementInst* packed_tmp15_i_144 = new InsertElementInst(packed_tmp7_i_141, float_tmp13_i_143, const_int32_31, "tmp15.i", label_forbody_i_123); - std::vector ptr_arrayidx20_i_145_indices; - ptr_arrayidx20_i_145_indices.push_back(int32_i_0_reg2mem_0_i_135); - ptr_arrayidx20_i_145_indices.push_back(const_int32_33); - Instruction* ptr_arrayidx20_i_145 = new GetElementPtrInst(ptr_aconsts_119, ptr_arrayidx20_i_145_indices.begin(), ptr_arrayidx20_i_145_indices.end(), "arrayidx20.i", label_forbody_i_123); - LoadInst* float_tmp21_i_146 = new LoadInst(ptr_arrayidx20_i_145, "tmp21.i", false, label_forbody_i_123); - InsertElementInst* packed_tmp23_i_147 = new InsertElementInst(packed_tmp15_i_144, float_tmp21_i_146, const_int32_33, "tmp23.i", label_forbody_i_123); - std::vector ptr_arrayidx28_i_148_indices; - ptr_arrayidx28_i_148_indices.push_back(int32_i_0_reg2mem_0_i_135); - ptr_arrayidx28_i_148_indices.push_back(const_int32_34); - Instruction* ptr_arrayidx28_i_148 = new GetElementPtrInst(ptr_aconsts_119, ptr_arrayidx28_i_148_indices.begin(), ptr_arrayidx28_i_148_indices.end(), "arrayidx28.i", label_forbody_i_123); - LoadInst* float_tmp29_i_149 = new LoadInst(ptr_arrayidx28_i_148, "tmp29.i", false, label_forbody_i_123); - InsertElementInst* packed_tmp31_i_150 = new InsertElementInst(packed_tmp23_i_147, float_tmp29_i_149, const_int32_34, "tmp31.i", label_forbody_i_123); - std::vector ptr_arrayidx34_i_151_indices; - ptr_arrayidx34_i_151_indices.push_back(const_int32_29); - ptr_arrayidx34_i_151_indices.push_back(int32_i_0_reg2mem_0_i_135); - Instruction* ptr_arrayidx34_i_151 = new GetElementPtrInst(ptr_consts_125, ptr_arrayidx34_i_151_indices.begin(), ptr_arrayidx34_i_151_indices.end(), "arrayidx34.i", label_forbody_i_123); - StoreInst* void_152 = new StoreInst(packed_tmp31_i_150, ptr_arrayidx34_i_151, false, label_forbody_i_123); - BinaryOperator* int32_indvar_next7 = BinaryOperator::create(Instruction::Add, int32_i_0_reg2mem_0_i_135, const_int32_31, "indvar.next7", label_forbody_i_123); - ICmpInst* int1_exitcond8 = new ICmpInst(ICmpInst::ICMP_EQ, int32_indvar_next7, int32_tmp10_i_133, "exitcond8", label_forbody_i_123); - new BranchInst(label_from_consts_exit_124, label_forbody_i_123, int1_exitcond8, label_forbody_i_123); - - // Block from_consts.exit (label_from_consts_exit_124) + BasicBlock* label_entry_110 = new BasicBlock("entry",func_run_fragment_shader,0); + BasicBlock* label_forbody_preheader_i_111 = new BasicBlock("forbody.preheader.i",func_run_fragment_shader,0); + BasicBlock* label_forbody_i_112 = new BasicBlock("forbody.i",func_run_fragment_shader,0); + BasicBlock* label_from_consts_exit_113 = new BasicBlock("from_consts.exit",func_run_fragment_shader,0); + + // Block entry (label_entry_110) + AllocaInst* ptr_consts_114 = new AllocaInst(ArrayTy_20, "consts", label_entry_110); + AllocaInst* ptr_temps_115 = new AllocaInst(ArrayTy_22, "temps", label_entry_110); + AllocaInst* ptr_args_116 = new AllocaInst(StructTy_struct_ShaderInput, "args", label_entry_110); + std::vector ptr_tmp_117_indices; + ptr_tmp_117_indices.push_back(const_int32_29); + ptr_tmp_117_indices.push_back(const_int32_35); + Instruction* ptr_tmp_117 = new GetElementPtrInst(ptr_args_116, ptr_tmp_117_indices.begin(), ptr_tmp_117_indices.end(), "tmp", label_entry_110); + StoreInst* void_118 = new StoreInst(const_int32_29, ptr_tmp_117, false, label_entry_110); + ICmpInst* int1_cmp_i_119 = new ICmpInst(ICmpInst::ICMP_SGT, int32_num_consts_109, const_int32_29, "cmp.i", label_entry_110); + new BranchInst(label_forbody_preheader_i_111, label_from_consts_exit_113, int1_cmp_i_119, label_entry_110); + + // Block forbody.preheader.i (label_forbody_preheader_i_111) + BinaryOperator* int32_tmp_i_121 = BinaryOperator::create(Instruction::Add, int32_num_consts_109, const_int32_30, "tmp.i", label_forbody_preheader_i_111); + ICmpInst* int1_tmp9_i_122 = new ICmpInst(ICmpInst::ICMP_SLT, int32_tmp_i_121, const_int32_29, "tmp9.i", label_forbody_preheader_i_111); + SelectInst* int32_tmp10_i_123 = new SelectInst(int1_tmp9_i_122, const_int32_31, int32_num_consts_109, "tmp10.i", label_forbody_preheader_i_111); + new BranchInst(label_forbody_i_112, label_forbody_preheader_i_111); + + // Block forbody.i (label_forbody_i_112) + Argument* fwdref_126 = new Argument(IntegerType::get(32)); + PHINode* int32_i_0_reg2mem_0_i_125 = new PHINode(IntegerType::get(32), "i.0.reg2mem.0.i", label_forbody_i_112); + int32_i_0_reg2mem_0_i_125->reserveOperandSpace(2); + int32_i_0_reg2mem_0_i_125->addIncoming(const_int32_29, label_forbody_preheader_i_111); + int32_i_0_reg2mem_0_i_125->addIncoming(fwdref_126, label_forbody_i_112); + + Argument* fwdref_128 = new Argument(VectorTy_1); + PHINode* packed_vec_0_reg2mem_0_i_127 = new PHINode(VectorTy_1, "vec.0.reg2mem.0.i", label_forbody_i_112); + packed_vec_0_reg2mem_0_i_127->reserveOperandSpace(2); + packed_vec_0_reg2mem_0_i_127->addIncoming(const_packed_32, label_forbody_preheader_i_111); + packed_vec_0_reg2mem_0_i_127->addIncoming(fwdref_128, label_forbody_i_112); + + std::vector ptr_arraydecay_i_129_indices; + ptr_arraydecay_i_129_indices.push_back(int32_i_0_reg2mem_0_i_125); + ptr_arraydecay_i_129_indices.push_back(const_int32_29); + Instruction* ptr_arraydecay_i_129 = new GetElementPtrInst(ptr_aconsts_108, ptr_arraydecay_i_129_indices.begin(), ptr_arraydecay_i_129_indices.end(), "arraydecay.i", label_forbody_i_112); + LoadInst* float_tmp5_i_130 = new LoadInst(ptr_arraydecay_i_129, "tmp5.i", false, label_forbody_i_112); + InsertElementInst* packed_tmp7_i_131 = new InsertElementInst(packed_vec_0_reg2mem_0_i_127, float_tmp5_i_130, const_int32_29, "tmp7.i", label_forbody_i_112); + std::vector ptr_arrayidx12_i_132_indices; + ptr_arrayidx12_i_132_indices.push_back(int32_i_0_reg2mem_0_i_125); + ptr_arrayidx12_i_132_indices.push_back(const_int32_31); + Instruction* ptr_arrayidx12_i_132 = new GetElementPtrInst(ptr_aconsts_108, ptr_arrayidx12_i_132_indices.begin(), ptr_arrayidx12_i_132_indices.end(), "arrayidx12.i", label_forbody_i_112); + LoadInst* float_tmp13_i_133 = new LoadInst(ptr_arrayidx12_i_132, "tmp13.i", false, label_forbody_i_112); + InsertElementInst* packed_tmp15_i_134 = new InsertElementInst(packed_tmp7_i_131, float_tmp13_i_133, const_int32_31, "tmp15.i", label_forbody_i_112); + std::vector ptr_arrayidx20_i_135_indices; + ptr_arrayidx20_i_135_indices.push_back(int32_i_0_reg2mem_0_i_125); + ptr_arrayidx20_i_135_indices.push_back(const_int32_33); + Instruction* ptr_arrayidx20_i_135 = new GetElementPtrInst(ptr_aconsts_108, ptr_arrayidx20_i_135_indices.begin(), ptr_arrayidx20_i_135_indices.end(), "arrayidx20.i", label_forbody_i_112); + LoadInst* float_tmp21_i_136 = new LoadInst(ptr_arrayidx20_i_135, "tmp21.i", false, label_forbody_i_112); + InsertElementInst* packed_tmp23_i_137 = new InsertElementInst(packed_tmp15_i_134, float_tmp21_i_136, const_int32_33, "tmp23.i", label_forbody_i_112); + std::vector ptr_arrayidx28_i_138_indices; + ptr_arrayidx28_i_138_indices.push_back(int32_i_0_reg2mem_0_i_125); + ptr_arrayidx28_i_138_indices.push_back(const_int32_34); + Instruction* ptr_arrayidx28_i_138 = new GetElementPtrInst(ptr_aconsts_108, ptr_arrayidx28_i_138_indices.begin(), ptr_arrayidx28_i_138_indices.end(), "arrayidx28.i", label_forbody_i_112); + LoadInst* float_tmp29_i_139 = new LoadInst(ptr_arrayidx28_i_138, "tmp29.i", false, label_forbody_i_112); + InsertElementInst* packed_tmp31_i_140 = new InsertElementInst(packed_tmp23_i_137, float_tmp29_i_139, const_int32_34, "tmp31.i", label_forbody_i_112); + std::vector ptr_arrayidx34_i_141_indices; + ptr_arrayidx34_i_141_indices.push_back(const_int32_29); + ptr_arrayidx34_i_141_indices.push_back(int32_i_0_reg2mem_0_i_125); + Instruction* ptr_arrayidx34_i_141 = new GetElementPtrInst(ptr_consts_114, ptr_arrayidx34_i_141_indices.begin(), ptr_arrayidx34_i_141_indices.end(), "arrayidx34.i", label_forbody_i_112); + StoreInst* void_142 = new StoreInst(packed_tmp31_i_140, ptr_arrayidx34_i_141, false, label_forbody_i_112); + BinaryOperator* int32_indvar_next7 = BinaryOperator::create(Instruction::Add, int32_i_0_reg2mem_0_i_125, const_int32_31, "indvar.next7", label_forbody_i_112); + ICmpInst* int1_exitcond8 = new ICmpInst(ICmpInst::ICMP_EQ, int32_indvar_next7, int32_tmp10_i_123, "exitcond8", label_forbody_i_112); + new BranchInst(label_from_consts_exit_113, label_forbody_i_112, int1_exitcond8, label_forbody_i_112); + + // Block from_consts.exit (label_from_consts_exit_113) std::vector ptr_tmp3_indices; ptr_tmp3_indices.push_back(const_int32_29); ptr_tmp3_indices.push_back(const_int32_34); - Instruction* ptr_tmp3 = new GetElementPtrInst(ptr_args_127, ptr_tmp3_indices.begin(), ptr_tmp3_indices.end(), "tmp3", label_from_consts_exit_124); + Instruction* ptr_tmp3 = new GetElementPtrInst(ptr_args_116, ptr_tmp3_indices.begin(), ptr_tmp3_indices.end(), "tmp3", label_from_consts_exit_113); std::vector ptr_arraydecay4_indices; ptr_arraydecay4_indices.push_back(const_int32_29); ptr_arraydecay4_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay4 = new GetElementPtrInst(ptr_consts_125, ptr_arraydecay4_indices.begin(), ptr_arraydecay4_indices.end(), "arraydecay4", label_from_consts_exit_124); - StoreInst* void_154 = new StoreInst(ptr_arraydecay4, ptr_tmp3, false, label_from_consts_exit_124); + Instruction* ptr_arraydecay4 = new GetElementPtrInst(ptr_consts_114, ptr_arraydecay4_indices.begin(), ptr_arraydecay4_indices.end(), "arraydecay4", label_from_consts_exit_113); + StoreInst* void_144 = new StoreInst(ptr_arraydecay4, ptr_tmp3, false, label_from_consts_exit_113); std::vector ptr_tmp5_indices; ptr_tmp5_indices.push_back(const_int32_29); ptr_tmp5_indices.push_back(const_int32_33); - Instruction* ptr_tmp5 = new GetElementPtrInst(ptr_args_127, ptr_tmp5_indices.begin(), ptr_tmp5_indices.end(), "tmp5", label_from_consts_exit_124); + Instruction* ptr_tmp5 = new GetElementPtrInst(ptr_args_116, ptr_tmp5_indices.begin(), ptr_tmp5_indices.end(), "tmp5", label_from_consts_exit_113); std::vector ptr_arraydecay6_indices; ptr_arraydecay6_indices.push_back(const_int32_29); ptr_arraydecay6_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay6 = new GetElementPtrInst(ptr_temps_126, ptr_arraydecay6_indices.begin(), ptr_arraydecay6_indices.end(), "arraydecay6", label_from_consts_exit_124); - StoreInst* void_155 = new StoreInst(ptr_arraydecay6, ptr_tmp5, false, label_from_consts_exit_124); - std::vector ptr_tmp8_156_indices; - ptr_tmp8_156_indices.push_back(const_int32_29); - ptr_tmp8_156_indices.push_back(const_int32_31); - Instruction* ptr_tmp8_156 = new GetElementPtrInst(ptr_args_127, ptr_tmp8_156_indices.begin(), ptr_tmp8_156_indices.end(), "tmp8", label_from_consts_exit_124); - std::vector ptr_tmp12_157_indices; - ptr_tmp12_157_indices.push_back(const_int32_29); - ptr_tmp12_157_indices.push_back(const_int32_29); - Instruction* ptr_tmp12_157 = new GetElementPtrInst(ptr_args_127, ptr_tmp12_157_indices.begin(), ptr_tmp12_157_indices.end(), "tmp12", label_from_consts_exit_124); - std::vector ptr_arraydecay11_158_indices; - ptr_arraydecay11_158_indices.push_back(const_int32_29); - ptr_arraydecay11_158_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay11_158 = new GetElementPtrInst(ptr_inputs_117, ptr_arraydecay11_158_indices.begin(), ptr_arraydecay11_158_indices.end(), "arraydecay11", label_from_consts_exit_124); - StoreInst* void_159 = new StoreInst(ptr_arraydecay11_158, ptr_tmp8_156, false, label_from_consts_exit_124); - std::vector ptr_arraydecay16_160_indices; - ptr_arraydecay16_160_indices.push_back(const_int32_29); - ptr_arraydecay16_160_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay16_160 = new GetElementPtrInst(ptr_results_116, ptr_arraydecay16_160_indices.begin(), ptr_arraydecay16_160_indices.end(), "arraydecay16", label_from_consts_exit_124); - StoreInst* void_161 = new StoreInst(ptr_arraydecay16_160, ptr_tmp12_157, false, label_from_consts_exit_124); - StoreInst* void_162 = new StoreInst(const_int32_29, ptr_tmp, false, label_from_consts_exit_124); - CallInst* void_163 = new CallInst(func_execute_shader, ptr_args_127, "", label_from_consts_exit_124); - void_163->setCallingConv(CallingConv::C); - void_163->setTailCall(false);const ParamAttrsList *void_163_PAL = 0; - void_163->setParamAttrs(void_163_PAL); - - LoadInst* int32_tmp23 = new LoadInst(ptr_tmp, "tmp23", false, label_from_consts_exit_124); + Instruction* ptr_arraydecay6 = new GetElementPtrInst(ptr_temps_115, ptr_arraydecay6_indices.begin(), ptr_arraydecay6_indices.end(), "arraydecay6", label_from_consts_exit_113); + StoreInst* void_145 = new StoreInst(ptr_arraydecay6, ptr_tmp5, false, label_from_consts_exit_113); + std::vector ptr_tmp8_indices; + ptr_tmp8_indices.push_back(const_int32_29); + ptr_tmp8_indices.push_back(const_int32_31); + Instruction* ptr_tmp8 = new GetElementPtrInst(ptr_args_116, ptr_tmp8_indices.begin(), ptr_tmp8_indices.end(), "tmp8", label_from_consts_exit_113); + std::vector ptr_tmp12_indices; + ptr_tmp12_indices.push_back(const_int32_29); + ptr_tmp12_indices.push_back(const_int32_29); + Instruction* ptr_tmp12 = new GetElementPtrInst(ptr_args_116, ptr_tmp12_indices.begin(), ptr_tmp12_indices.end(), "tmp12", label_from_consts_exit_113); + std::vector ptr_arraydecay11_146_indices; + ptr_arraydecay11_146_indices.push_back(const_int32_29); + ptr_arraydecay11_146_indices.push_back(const_int32_29); + Instruction* ptr_arraydecay11_146 = new GetElementPtrInst(ptr_inputs_106, ptr_arraydecay11_146_indices.begin(), ptr_arraydecay11_146_indices.end(), "arraydecay11", label_from_consts_exit_113); + StoreInst* void_147 = new StoreInst(ptr_arraydecay11_146, ptr_tmp8, false, label_from_consts_exit_113); + std::vector ptr_arraydecay16_indices; + ptr_arraydecay16_indices.push_back(const_int32_29); + ptr_arraydecay16_indices.push_back(const_int32_29); + Instruction* ptr_arraydecay16 = new GetElementPtrInst(ptr_results_105, ptr_arraydecay16_indices.begin(), ptr_arraydecay16_indices.end(), "arraydecay16", label_from_consts_exit_113); + StoreInst* void_148 = new StoreInst(ptr_arraydecay16, ptr_tmp12, false, label_from_consts_exit_113); + StoreInst* void_149 = new StoreInst(const_int32_29, ptr_tmp_117, false, label_from_consts_exit_113); + CallInst* void_150 = new CallInst(func_execute_shader, ptr_args_116, "", label_from_consts_exit_113); + void_150->setCallingConv(CallingConv::C); + void_150->setTailCall(false);const ParamAttrsList *void_150_PAL = 0; + void_150->setParamAttrs(void_150_PAL); + + LoadInst* int32_tmp23 = new LoadInst(ptr_tmp_117, "tmp23", false, label_from_consts_exit_113); std::vector ptr_arraydecay11_1_indices; ptr_arraydecay11_1_indices.push_back(const_int32_31); ptr_arraydecay11_1_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay11_1 = new GetElementPtrInst(ptr_inputs_117, ptr_arraydecay11_1_indices.begin(), ptr_arraydecay11_1_indices.end(), "arraydecay11.1", label_from_consts_exit_124); - StoreInst* void_164 = new StoreInst(ptr_arraydecay11_1, ptr_tmp8_156, false, label_from_consts_exit_124); + Instruction* ptr_arraydecay11_1 = new GetElementPtrInst(ptr_inputs_106, ptr_arraydecay11_1_indices.begin(), ptr_arraydecay11_1_indices.end(), "arraydecay11.1", label_from_consts_exit_113); + StoreInst* void_151 = new StoreInst(ptr_arraydecay11_1, ptr_tmp8, false, label_from_consts_exit_113); std::vector ptr_arraydecay16_1_indices; ptr_arraydecay16_1_indices.push_back(const_int32_31); ptr_arraydecay16_1_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay16_1 = new GetElementPtrInst(ptr_results_116, ptr_arraydecay16_1_indices.begin(), ptr_arraydecay16_1_indices.end(), "arraydecay16.1", label_from_consts_exit_124); - StoreInst* void_165 = new StoreInst(ptr_arraydecay16_1, ptr_tmp12_157, false, label_from_consts_exit_124); - StoreInst* void_166 = new StoreInst(const_int32_29, ptr_tmp, false, label_from_consts_exit_124); - CallInst* void_167 = new CallInst(func_execute_shader, ptr_args_127, "", label_from_consts_exit_124); - void_167->setCallingConv(CallingConv::C); - void_167->setTailCall(false);const ParamAttrsList *void_167_PAL = 0; - void_167->setParamAttrs(void_167_PAL); - - LoadInst* int32_tmp23_1 = new LoadInst(ptr_tmp, "tmp23.1", false, label_from_consts_exit_124); - BinaryOperator* int32_shl_1 = BinaryOperator::create(Instruction::Shl, int32_tmp23_1, const_int32_31, "shl.1", label_from_consts_exit_124); - BinaryOperator* int32_or_1 = BinaryOperator::create(Instruction::Or, int32_shl_1, int32_tmp23, "or.1", label_from_consts_exit_124); + Instruction* ptr_arraydecay16_1 = new GetElementPtrInst(ptr_results_105, ptr_arraydecay16_1_indices.begin(), ptr_arraydecay16_1_indices.end(), "arraydecay16.1", label_from_consts_exit_113); + StoreInst* void_152 = new StoreInst(ptr_arraydecay16_1, ptr_tmp12, false, label_from_consts_exit_113); + StoreInst* void_153 = new StoreInst(const_int32_29, ptr_tmp_117, false, label_from_consts_exit_113); + CallInst* void_154 = new CallInst(func_execute_shader, ptr_args_116, "", label_from_consts_exit_113); + void_154->setCallingConv(CallingConv::C); + void_154->setTailCall(false);const ParamAttrsList *void_154_PAL = 0; + void_154->setParamAttrs(void_154_PAL); + + LoadInst* int32_tmp23_1 = new LoadInst(ptr_tmp_117, "tmp23.1", false, label_from_consts_exit_113); + BinaryOperator* int32_shl_1 = BinaryOperator::create(Instruction::Shl, int32_tmp23_1, const_int32_31, "shl.1", label_from_consts_exit_113); + BinaryOperator* int32_or_1 = BinaryOperator::create(Instruction::Or, int32_shl_1, int32_tmp23, "or.1", label_from_consts_exit_113); std::vector ptr_arraydecay11_2_indices; ptr_arraydecay11_2_indices.push_back(const_int32_33); ptr_arraydecay11_2_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay11_2 = new GetElementPtrInst(ptr_inputs_117, ptr_arraydecay11_2_indices.begin(), ptr_arraydecay11_2_indices.end(), "arraydecay11.2", label_from_consts_exit_124); - StoreInst* void_168 = new StoreInst(ptr_arraydecay11_2, ptr_tmp8_156, false, label_from_consts_exit_124); + Instruction* ptr_arraydecay11_2 = new GetElementPtrInst(ptr_inputs_106, ptr_arraydecay11_2_indices.begin(), ptr_arraydecay11_2_indices.end(), "arraydecay11.2", label_from_consts_exit_113); + StoreInst* void_155 = new StoreInst(ptr_arraydecay11_2, ptr_tmp8, false, label_from_consts_exit_113); std::vector ptr_arraydecay16_2_indices; ptr_arraydecay16_2_indices.push_back(const_int32_33); ptr_arraydecay16_2_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay16_2 = new GetElementPtrInst(ptr_results_116, ptr_arraydecay16_2_indices.begin(), ptr_arraydecay16_2_indices.end(), "arraydecay16.2", label_from_consts_exit_124); - StoreInst* void_169 = new StoreInst(ptr_arraydecay16_2, ptr_tmp12_157, false, label_from_consts_exit_124); - StoreInst* void_170 = new StoreInst(const_int32_29, ptr_tmp, false, label_from_consts_exit_124); - CallInst* void_171 = new CallInst(func_execute_shader, ptr_args_127, "", label_from_consts_exit_124); - void_171->setCallingConv(CallingConv::C); - void_171->setTailCall(false);const ParamAttrsList *void_171_PAL = 0; - void_171->setParamAttrs(void_171_PAL); - - LoadInst* int32_tmp23_2 = new LoadInst(ptr_tmp, "tmp23.2", false, label_from_consts_exit_124); - BinaryOperator* int32_shl_2 = BinaryOperator::create(Instruction::Shl, int32_tmp23_2, const_int32_33, "shl.2", label_from_consts_exit_124); - BinaryOperator* int32_or_2 = BinaryOperator::create(Instruction::Or, int32_shl_2, int32_or_1, "or.2", label_from_consts_exit_124); + Instruction* ptr_arraydecay16_2 = new GetElementPtrInst(ptr_results_105, ptr_arraydecay16_2_indices.begin(), ptr_arraydecay16_2_indices.end(), "arraydecay16.2", label_from_consts_exit_113); + StoreInst* void_156 = new StoreInst(ptr_arraydecay16_2, ptr_tmp12, false, label_from_consts_exit_113); + StoreInst* void_157 = new StoreInst(const_int32_29, ptr_tmp_117, false, label_from_consts_exit_113); + CallInst* void_158 = new CallInst(func_execute_shader, ptr_args_116, "", label_from_consts_exit_113); + void_158->setCallingConv(CallingConv::C); + void_158->setTailCall(false);const ParamAttrsList *void_158_PAL = 0; + void_158->setParamAttrs(void_158_PAL); + + LoadInst* int32_tmp23_2 = new LoadInst(ptr_tmp_117, "tmp23.2", false, label_from_consts_exit_113); + BinaryOperator* int32_shl_2 = BinaryOperator::create(Instruction::Shl, int32_tmp23_2, const_int32_33, "shl.2", label_from_consts_exit_113); + BinaryOperator* int32_or_2 = BinaryOperator::create(Instruction::Or, int32_shl_2, int32_or_1, "or.2", label_from_consts_exit_113); std::vector ptr_arraydecay11_3_indices; ptr_arraydecay11_3_indices.push_back(const_int32_34); ptr_arraydecay11_3_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay11_3 = new GetElementPtrInst(ptr_inputs_117, ptr_arraydecay11_3_indices.begin(), ptr_arraydecay11_3_indices.end(), "arraydecay11.3", label_from_consts_exit_124); - StoreInst* void_172 = new StoreInst(ptr_arraydecay11_3, ptr_tmp8_156, false, label_from_consts_exit_124); + Instruction* ptr_arraydecay11_3 = new GetElementPtrInst(ptr_inputs_106, ptr_arraydecay11_3_indices.begin(), ptr_arraydecay11_3_indices.end(), "arraydecay11.3", label_from_consts_exit_113); + StoreInst* void_159 = new StoreInst(ptr_arraydecay11_3, ptr_tmp8, false, label_from_consts_exit_113); std::vector ptr_arraydecay16_3_indices; ptr_arraydecay16_3_indices.push_back(const_int32_34); ptr_arraydecay16_3_indices.push_back(const_int32_29); - Instruction* ptr_arraydecay16_3 = new GetElementPtrInst(ptr_results_116, ptr_arraydecay16_3_indices.begin(), ptr_arraydecay16_3_indices.end(), "arraydecay16.3", label_from_consts_exit_124); - StoreInst* void_173 = new StoreInst(ptr_arraydecay16_3, ptr_tmp12_157, false, label_from_consts_exit_124); - StoreInst* void_174 = new StoreInst(const_int32_29, ptr_tmp, false, label_from_consts_exit_124); - CallInst* void_175 = new CallInst(func_execute_shader, ptr_args_127, "", label_from_consts_exit_124); - void_175->setCallingConv(CallingConv::C); - void_175->setTailCall(false);const ParamAttrsList *void_175_PAL = 0; - void_175->setParamAttrs(void_175_PAL); - - LoadInst* int32_tmp23_3 = new LoadInst(ptr_tmp, "tmp23.3", false, label_from_consts_exit_124); - BinaryOperator* int32_shl_3 = BinaryOperator::create(Instruction::Shl, int32_tmp23_3, const_int32_34, "shl.3", label_from_consts_exit_124); - BinaryOperator* int32_or_3 = BinaryOperator::create(Instruction::Or, int32_shl_3, int32_or_2, "or.3", label_from_consts_exit_124); - BinaryOperator* int32_neg = BinaryOperator::create(Instruction::Xor, int32_or_3, const_int32_30, "neg", label_from_consts_exit_124); - new ReturnInst(int32_neg, label_from_consts_exit_124); + Instruction* ptr_arraydecay16_3 = new GetElementPtrInst(ptr_results_105, ptr_arraydecay16_3_indices.begin(), ptr_arraydecay16_3_indices.end(), "arraydecay16.3", label_from_consts_exit_113); + StoreInst* void_160 = new StoreInst(ptr_arraydecay16_3, ptr_tmp12, false, label_from_consts_exit_113); + StoreInst* void_161 = new StoreInst(const_int32_29, ptr_tmp_117, false, label_from_consts_exit_113); + CallInst* void_162 = new CallInst(func_execute_shader, ptr_args_116, "", label_from_consts_exit_113); + void_162->setCallingConv(CallingConv::C); + void_162->setTailCall(false);const ParamAttrsList *void_162_PAL = 0; + void_162->setParamAttrs(void_162_PAL); + + LoadInst* int32_tmp23_3 = new LoadInst(ptr_tmp_117, "tmp23.3", false, label_from_consts_exit_113); + BinaryOperator* int32_shl_3 = BinaryOperator::create(Instruction::Shl, int32_tmp23_3, const_int32_34, "shl.3", label_from_consts_exit_113); + BinaryOperator* int32_or_3 = BinaryOperator::create(Instruction::Or, int32_shl_3, int32_or_2, "or.3", label_from_consts_exit_113); + BinaryOperator* int32_neg = BinaryOperator::create(Instruction::Xor, int32_or_3, const_int32_30, "neg", label_from_consts_exit_113); + new ReturnInst(int32_neg, label_from_consts_exit_113); // Resolve Forward References - fwdref_138->replaceAllUsesWith(packed_tmp31_i_150); delete fwdref_138; - fwdref_136->replaceAllUsesWith(int32_indvar_next7); delete fwdref_136; + fwdref_128->replaceAllUsesWith(packed_tmp31_i_140); delete fwdref_128; + fwdref_126->replaceAllUsesWith(int32_indvar_next7); delete fwdref_126; } diff --git a/src/mesa/pipe/llvm/llvm_entry.c b/src/mesa/pipe/llvm/llvm_entry.c index c3b34584e16..fa50b60e663 100644 --- a/src/mesa/pipe/llvm/llvm_entry.c +++ b/src/mesa/pipe/llvm/llvm_entry.c @@ -86,8 +86,8 @@ struct ShaderInput extern void execute_shader(struct ShaderInput *input); -void run_vertex_shader(float4 (*inputs)[16], - float4 (*results)[16], +void run_vertex_shader(void *inputs, + void *results, float (*aconsts)[4], int num_vertices, int num_inputs, @@ -98,16 +98,16 @@ void run_vertex_shader(float4 (*inputs)[16], float4 temps[128];//MAX_PROGRAM_TEMPS struct ShaderInput args; + args.dests = results; + args.inputs = inputs; + /*printf("XXX LLVM run_vertex_shader vertices = %d, inputs = %d, attribs = %d, consts = %d\n", num_vertices, num_inputs, num_attribs, num_consts);*/ from_consts(consts, aconsts, num_consts); args.consts = consts; args.temps = temps; - for (int i = 0; i < num_vertices; ++i) { - args.dests = results[i]; - args.inputs = inputs[i]; - execute_shader(&args); - } + + execute_shader(&args); } diff --git a/src/mesa/pipe/llvm/tgsitollvm.cpp b/src/mesa/pipe/llvm/tgsitollvm.cpp new file mode 100644 index 00000000000..eb9e1196f17 --- /dev/null +++ b/src/mesa/pipe/llvm/tgsitollvm.cpp @@ -0,0 +1,682 @@ +#include "tgsitollvm.h" + +#include "gallivm.h" +#include "gallivm_p.h" + +#include "storage.h" +#include "instructions.h" + +#include "pipe/p_shader_tokens.h" + +#include "pipe/tgsi/util/tgsi_parse.h" +#include "pipe/tgsi/exec/tgsi_exec.h" +#include "pipe/tgsi/util/tgsi_util.h" +#include "pipe/tgsi/util/tgsi_build.h" +#include "pipe/tgsi/util/tgsi_dump.h" + + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + + +#include +#include +#include + +using namespace llvm; +#include "llvm_base_shader.cpp" + +static inline void +add_interpolator(struct gallivm_ir *ir, + struct gallivm_interpolate *interp) +{ + ir->interpolators[ir->num_interp] = *interp; + ++ir->num_interp; +} + +static void +translate_declaration(struct gallivm_ir *prog, + llvm::Module *module, + Storage *storage, + struct tgsi_full_declaration *decl, + struct tgsi_full_declaration *fd) +{ + if (decl->Declaration.File == TGSI_FILE_INPUT) { + unsigned first, last, mask; + uint interp_method; + + assert(decl->Declaration.Declare == TGSI_DECLARE_RANGE); + + first = decl->u.DeclarationRange.First; + last = decl->u.DeclarationRange.Last; + mask = decl->Declaration.UsageMask; + + /* Do not touch WPOS.xy */ + if (first == 0) { + mask &= ~TGSI_WRITEMASK_XY; + if (mask == TGSI_WRITEMASK_NONE) { + first++; + if (first > last) { + return; + } + } + } + + interp_method = decl->Interpolation.Interpolate; + + if (mask == TGSI_WRITEMASK_XYZW) { + unsigned i, j; + + for (i = first; i <= last; i++) { + for (j = 0; j < NUM_CHANNELS; j++) { + //interp( mach, i, j ); + struct gallivm_interpolate interp; + interp.type = interp_method; + interp.attrib = i; + interp.chan = j; + add_interpolator(prog, &interp); + } + } + } else { + unsigned i, j; + for( j = 0; j < NUM_CHANNELS; j++ ) { + if( mask & (1 << j) ) { + for( i = first; i <= last; i++ ) { + struct gallivm_interpolate interp; + interp.type = interp_method; + interp.attrib = i; + interp.chan = j; + add_interpolator(prog, &interp); + } + } + } + } + } +} + + +static void +translate_immediate(Storage *storage, + struct tgsi_full_immediate *imm) +{ + float vec[4]; + int i; + for (i = 0; i < imm->Immediate.Size - 1; ++i) { + switch( imm->Immediate.DataType ) { + case TGSI_IMM_FLOAT32: + vec[i] = imm->u.ImmediateFloat32[i].Float; + break; + default: + assert( 0 ); + } + } + storage->addImmediate(vec); +} + +static inline llvm::Value * +swizzleVector(llvm::Value *val, struct tgsi_full_src_register *src, + Storage *storage) +{ + int swizzle = 0; + int start = 1000; + const int NO_SWIZZLE = TGSI_SWIZZLE_X * 1000 + TGSI_SWIZZLE_Y * 100 + + TGSI_SWIZZLE_Z * 10 + TGSI_SWIZZLE_W; + for (int k = 0; k < 4; ++k) { + swizzle += tgsi_util_get_full_src_register_extswizzle(src, k) * start; + start /= 10; + } + if (swizzle != NO_SWIZZLE) { + /*fprintf(stderr, "XXXXXXXX swizzle = %d\n", swizzle);*/ + val = storage->shuffleVector(val, swizzle); + } + return val; +} + +static void +translate_instruction(llvm::Module *module, + Storage *storage, + Instructions *instr, + struct tgsi_full_instruction *inst, + struct tgsi_full_instruction *fi, + unsigned instno) +{ + llvm::Value *inputs[4]; + inputs[0] = 0; + inputs[1] = 0; + inputs[2] = 0; + inputs[3] = 0; + + for (int i = 0; i < inst->Instruction.NumSrcRegs; ++i) { + struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i]; + llvm::Value *val = 0; + llvm::Value *indIdx = 0; + + if (src->SrcRegister.Indirect) { + indIdx = storage->addrElement(src->SrcRegisterInd.Index); + indIdx = storage->extractIndex(indIdx); + } + if (src->SrcRegister.File == TGSI_FILE_CONSTANT) { + val = storage->constElement(src->SrcRegister.Index, indIdx); + } else if (src->SrcRegister.File == TGSI_FILE_INPUT) { + val = storage->inputElement(src->SrcRegister.Index, indIdx); + } else if (src->SrcRegister.File == TGSI_FILE_TEMPORARY) { + val = storage->tempElement(src->SrcRegister.Index); + } else if (src->SrcRegister.File == TGSI_FILE_OUTPUT) { + val = storage->outputElement(src->SrcRegister.Index, indIdx); + } else if (src->SrcRegister.File == TGSI_FILE_IMMEDIATE) { + val = storage->immediateElement(src->SrcRegister.Index); + } else { + fprintf(stderr, "ERROR: not supported llvm source %d\n", src->SrcRegister.File); + return; + } + + inputs[i] = swizzleVector(val, src, storage); + } + + /*if (inputs[0]) + instr->printVector(inputs[0]); + if (inputs[1]) + instr->printVector(inputs[1]);*/ + llvm::Value *out = 0; + switch (inst->Instruction.Opcode) { + case TGSI_OPCODE_ARL: { + out = instr->arl(inputs[0]); + } + break; + case TGSI_OPCODE_MOV: { + out = inputs[0]; + } + break; + case TGSI_OPCODE_LIT: { + out = instr->lit(inputs[0]); + } + break; + case TGSI_OPCODE_RCP: { + out = instr->rcp(inputs[0]); + } + break; + case TGSI_OPCODE_RSQ: { + out = instr->rsq(inputs[0]); + } + break; + case TGSI_OPCODE_EXP: + break; + case TGSI_OPCODE_LOG: + break; + case TGSI_OPCODE_MUL: { + out = instr->mul(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_ADD: { + out = instr->add(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_DP3: { + out = instr->dp3(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_DP4: { + out = instr->dp4(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_DST: { + out = instr->dst(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_MIN: { + out = instr->min(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_MAX: { + out = instr->max(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_SLT: { + out = instr->slt(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_SGE: { + out = instr->sge(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_MAD: { + out = instr->madd(inputs[0], inputs[1], inputs[2]); + } + break; + case TGSI_OPCODE_SUB: { + out = instr->sub(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_LERP: { + out = instr->lerp(inputs[0], inputs[1], inputs[2]); + } + break; + case TGSI_OPCODE_CND: + break; + case TGSI_OPCODE_CND0: + break; + case TGSI_OPCODE_DOT2ADD: + break; + case TGSI_OPCODE_INDEX: + break; + case TGSI_OPCODE_NEGATE: + break; + case TGSI_OPCODE_FRAC: { + out = instr->frc(inputs[0]); + } + break; + case TGSI_OPCODE_CLAMP: + break; + case TGSI_OPCODE_FLOOR: { + out = instr->floor(inputs[0]); + } + break; + case TGSI_OPCODE_ROUND: + break; + case TGSI_OPCODE_EXPBASE2: { + out = instr->ex2(inputs[0]); + } + break; + case TGSI_OPCODE_LOGBASE2: { + out = instr->lg2(inputs[0]); + } + break; + case TGSI_OPCODE_POWER: { + out = instr->pow(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_CROSSPRODUCT: { + out = instr->cross(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_MULTIPLYMATRIX: + break; + case TGSI_OPCODE_ABS: { + out = instr->abs(inputs[0]); + } + break; + case TGSI_OPCODE_RCC: + break; + case TGSI_OPCODE_DPH: { + out = instr->dph(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_COS: { + out = instr->cos(inputs[0]); + } + break; + case TGSI_OPCODE_DDX: + break; + case TGSI_OPCODE_DDY: + break; + case TGSI_OPCODE_KILP: { + out = instr->kilp(inputs[0]); + storage->setKilElement(out); + return; + } + break; + case TGSI_OPCODE_PK2H: + break; + case TGSI_OPCODE_PK2US: + break; + case TGSI_OPCODE_PK4B: + break; + case TGSI_OPCODE_PK4UB: + break; + case TGSI_OPCODE_RFL: + break; + case TGSI_OPCODE_SEQ: + break; + case TGSI_OPCODE_SFL: + break; + case TGSI_OPCODE_SGT: { + out = instr->sgt(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_SIN: { + out = instr->sin(inputs[0]); + } + break; + case TGSI_OPCODE_SLE: + break; + case TGSI_OPCODE_SNE: + break; + case TGSI_OPCODE_STR: + break; + case TGSI_OPCODE_TEX: + break; + case TGSI_OPCODE_TXD: + break; + case TGSI_OPCODE_UP2H: + break; + case TGSI_OPCODE_UP2US: + break; + case TGSI_OPCODE_UP4B: + break; + case TGSI_OPCODE_UP4UB: + break; + case TGSI_OPCODE_X2D: + break; + case TGSI_OPCODE_ARA: + break; + case TGSI_OPCODE_ARR: + break; + case TGSI_OPCODE_BRA: + break; + case TGSI_OPCODE_CAL: { + instr->cal(inst->InstructionExtLabel.Label, storage->inputPtr()); + return; + } + break; + case TGSI_OPCODE_RET: { + instr->end(); + return; + } + break; + case TGSI_OPCODE_SSG: + break; + case TGSI_OPCODE_CMP: { + out = instr->cmp(inputs[0], inputs[1], inputs[2]); + } + break; + case TGSI_OPCODE_SCS: { + out = instr->scs(inputs[0]); + } + break; + case TGSI_OPCODE_TXB: + break; + case TGSI_OPCODE_NRM: + break; + case TGSI_OPCODE_DIV: + break; + case TGSI_OPCODE_DP2: + break; + case TGSI_OPCODE_TXL: + break; + case TGSI_OPCODE_BRK: { + instr->brk(); + return; + } + break; + case TGSI_OPCODE_IF: { + instr->ifop(inputs[0]); + storage->setCurrentBlock(instr->currentBlock()); + return; //just update the state + } + break; + case TGSI_OPCODE_LOOP: + break; + case TGSI_OPCODE_REP: + break; + case TGSI_OPCODE_ELSE: { + instr->elseop(); + storage->setCurrentBlock(instr->currentBlock()); + return; //only state update + } + break; + case TGSI_OPCODE_ENDIF: { + instr->endif(); + storage->setCurrentBlock(instr->currentBlock()); + return; //just update the state + } + break; + case TGSI_OPCODE_ENDLOOP: + break; + case TGSI_OPCODE_ENDREP: + break; + case TGSI_OPCODE_PUSHA: + break; + case TGSI_OPCODE_POPA: + break; + case TGSI_OPCODE_CEIL: + break; + case TGSI_OPCODE_I2F: + break; + case TGSI_OPCODE_NOT: + break; + case TGSI_OPCODE_TRUNC: { + out = instr->trunc(inputs[0]); + } + break; + case TGSI_OPCODE_SHL: + break; + case TGSI_OPCODE_SHR: + break; + case TGSI_OPCODE_AND: + break; + case TGSI_OPCODE_OR: + break; + case TGSI_OPCODE_MOD: + break; + case TGSI_OPCODE_XOR: + break; + case TGSI_OPCODE_SAD: + break; + case TGSI_OPCODE_TXF: + break; + case TGSI_OPCODE_TXQ: + break; + case TGSI_OPCODE_CONT: + break; + case TGSI_OPCODE_EMIT: + break; + case TGSI_OPCODE_ENDPRIM: + break; + case TGSI_OPCODE_BGNLOOP2: { + instr->beginLoop(); + storage->setCurrentBlock(instr->currentBlock()); + return; + } + break; + case TGSI_OPCODE_BGNSUB: { + instr->bgnSub(instno); + storage->setCurrentBlock(instr->currentBlock()); + storage->pushTemps(); + return; + } + break; + case TGSI_OPCODE_ENDLOOP2: { + instr->endLoop(); + storage->setCurrentBlock(instr->currentBlock()); + return; + } + break; + case TGSI_OPCODE_ENDSUB: { + instr->endSub(); + storage->setCurrentBlock(instr->currentBlock()); + storage->popArguments(); + storage->popTemps(); + return; + } + break; + case TGSI_OPCODE_NOISE1: + break; + case TGSI_OPCODE_NOISE2: + break; + case TGSI_OPCODE_NOISE3: + break; + case TGSI_OPCODE_NOISE4: + break; + case TGSI_OPCODE_NOP: + break; + case TGSI_OPCODE_TEXBEM: + break; + case TGSI_OPCODE_TEXBEML: + break; + case TGSI_OPCODE_TEXREG2AR: + break; + case TGSI_OPCODE_TEXM3X2PAD: + break; + case TGSI_OPCODE_TEXM3X2TEX: + break; + case TGSI_OPCODE_TEXM3X3PAD: + break; + case TGSI_OPCODE_TEXM3X3TEX: + break; + case TGSI_OPCODE_TEXM3X3SPEC: + break; + case TGSI_OPCODE_TEXM3X3VSPEC: + break; + case TGSI_OPCODE_TEXREG2GB: + break; + case TGSI_OPCODE_TEXREG2RGB: + break; + case TGSI_OPCODE_TEXDP3TEX: + break; + case TGSI_OPCODE_TEXDP3: + break; + case TGSI_OPCODE_TEXM3X3: + break; + case TGSI_OPCODE_TEXM3X2DEPTH: + break; + case TGSI_OPCODE_TEXDEPTH: + break; + case TGSI_OPCODE_BEM: + break; + case TGSI_OPCODE_M4X3: + break; + case TGSI_OPCODE_M3X4: + break; + case TGSI_OPCODE_M3X3: + break; + case TGSI_OPCODE_M3X2: + break; + case TGSI_OPCODE_NRM4: + break; + case TGSI_OPCODE_CALLNZ: + break; + case TGSI_OPCODE_IFC: + break; + case TGSI_OPCODE_BREAKC: + break; + case TGSI_OPCODE_KIL: + break; + case TGSI_OPCODE_END: + instr->end(); + return; + break; + default: + fprintf(stderr, "ERROR: Unknown opcode %d\n", + inst->Instruction.Opcode); + assert(0); + break; + } + + if (!out) { + fprintf(stderr, "ERROR: unsupported opcode %d\n", + inst->Instruction.Opcode); + assert(!"Unsupported opcode"); + } + + /* # not sure if we need this */ + switch( inst->Instruction.Saturate ) { + case TGSI_SAT_NONE: + break; + case TGSI_SAT_ZERO_ONE: + /*TXT( "_SAT" );*/ + break; + case TGSI_SAT_MINUS_PLUS_ONE: + /*TXT( "_SAT[-1,1]" );*/ + break; + default: + assert( 0 ); + } + + /* store results */ + for (int i = 0; i < inst->Instruction.NumDstRegs; ++i) { + struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i]; + + if (dst->DstRegister.File == TGSI_FILE_OUTPUT) { + storage->setOutputElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); + } else if (dst->DstRegister.File == TGSI_FILE_TEMPORARY) { + storage->setTempElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); + } else if (dst->DstRegister.File == TGSI_FILE_ADDRESS) { + storage->setAddrElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); + } else { + fprintf(stderr, "ERROR: unsupported LLVM destination!"); + assert(!"wrong destination"); + } + } +} + +llvm::Module * +tgsi_to_llvm(struct gallivm_ir *ir, const struct tgsi_token *tokens) +{ + llvm::Module *mod = createBaseShader(); + struct tgsi_parse_context parse; + struct tgsi_full_instruction fi; + struct tgsi_full_declaration fd; + unsigned instno = 0; + Function* shader = mod->getFunction("execute_shader"); + std::ostringstream stream; + if (ir->type == GALLIVM_VS) { + stream << "vs_shader"; + } else { + stream << "fs_shader"; + } + stream << ir->id; + std::string func_name = stream.str(); + shader->setName(func_name.c_str()); + + Function::arg_iterator args = shader->arg_begin(); + Value *ptr_INPUT = args++; + ptr_INPUT->setName("input"); + + BasicBlock *label_entry = new BasicBlock("entry", shader, 0); + + tgsi_parse_init(&parse, tokens); + + fi = tgsi_default_full_instruction(); + fd = tgsi_default_full_declaration(); + Storage storage(label_entry, ptr_INPUT); + Instructions instr(mod, shader, label_entry, &storage); + while(!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + translate_declaration(ir, mod, &storage, + &parse.FullToken.FullDeclaration, + &fd); + break; + + case TGSI_TOKEN_TYPE_IMMEDIATE: + translate_immediate(&storage, + &parse.FullToken.FullImmediate); + break; + + case TGSI_TOKEN_TYPE_INSTRUCTION: + translate_instruction(mod, &storage, &instr, + &parse.FullToken.FullInstruction, + &fi, instno); + ++instno; + break; + + default: + assert(0); + } + } + + tgsi_parse_free(&parse); + + ir->num_consts = storage.numConsts(); + return mod; +} diff --git a/src/mesa/pipe/llvm/tgsitollvm.h b/src/mesa/pipe/llvm/tgsitollvm.h new file mode 100644 index 00000000000..073ffb5749a --- /dev/null +++ b/src/mesa/pipe/llvm/tgsitollvm.h @@ -0,0 +1,16 @@ +#ifndef TGSITOLLVM_H +#define TGSITOLLVM_H + + +namespace llvm { + class Module; +} + +struct gallivm_ir; +struct tgsi_token; + + +llvm::Module * tgsi_to_llvm(struct gallivm_ir *ir, + const struct tgsi_token *tokens); + +#endif diff --git a/src/mesa/pipe/softpipe/sp_state_fs.c b/src/mesa/pipe/softpipe/sp_state_fs.c index 1430be7869a..598a70c8279 100644 --- a/src/mesa/pipe/softpipe/sp_state_fs.c +++ b/src/mesa/pipe/softpipe/sp_state_fs.c @@ -61,7 +61,7 @@ softpipe_create_fs_state(struct pipe_context *pipe, } #ifdef MESA_LLVM - state->llvm_prog = gallivm_from_tgsi(state->shader.tokens, GALLIVM_FS); + state->llvm_prog = 0; if (!gallivm_global_cpu_engine()) { gallivm_cpu_engine_create(state->llvm_prog); } -- cgit v1.2.3 From 8f15140943a65e3e1488c8b43b2dffb1cd0a299c Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Tue, 5 Feb 2008 03:09:24 -0500 Subject: rewrite the way cpu engine is handled --- src/mesa/pipe/draw/draw_vertex_shader.c | 9 +- src/mesa/pipe/llvm/Makefile | 1 + src/mesa/pipe/llvm/gallivm.cpp | 134 --------------------- src/mesa/pipe/llvm/gallivm.h | 37 +++--- src/mesa/pipe/llvm/gallivm_cpu.cpp | 204 ++++++++++++++++++++++++++++++++ 5 files changed, 227 insertions(+), 158 deletions(-) create mode 100644 src/mesa/pipe/llvm/gallivm_cpu.cpp diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index 9567283ff5c..574ac670572 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -116,11 +116,10 @@ run_vertex_program(struct draw_context *draw, #ifdef MESA_LLVM if (1) { struct gallivm_prog *prog = draw->vertex_shader->llvm_prog; - gallivm_prog_exec(prog, - machine->Inputs, - machine->Outputs, - machine->Consts, - 12, 12, 12); + gallivm_cpu_vs_exec(prog, + machine->Inputs, + machine->Outputs, + machine->Consts); } else #elif defined(__i386__) || defined(__386__) if (draw->use_sse) { diff --git a/src/mesa/pipe/llvm/Makefile b/src/mesa/pipe/llvm/Makefile index 10ff7aacae8..b333661c6f8 100644 --- a/src/mesa/pipe/llvm/Makefile +++ b/src/mesa/pipe/llvm/Makefile @@ -7,6 +7,7 @@ LIBNAME = gallivm GALLIVM_SOURCES = \ gallivm.cpp \ + gallivm_cpu.cpp \ instructions.cpp \ loweringpass.cpp \ tgsitollvm.cpp \ diff --git a/src/mesa/pipe/llvm/gallivm.cpp b/src/mesa/pipe/llvm/gallivm.cpp index cf9b0f64067..cb9a88f0b86 100644 --- a/src/mesa/pipe/llvm/gallivm.cpp +++ b/src/mesa/pipe/llvm/gallivm.cpp @@ -70,10 +70,6 @@ #include #include -struct gallivm_cpu_engine { - llvm::ExecutionEngine *engine; -}; - static int GLOBAL_ID = 0; using namespace llvm; @@ -145,38 +141,6 @@ void gallivm_prog_delete(struct gallivm_prog *prog) free(prog); } -typedef void (*vertex_shader_runner)(void *ainputs, - void *dests, - float (*aconsts)[4], - int num_vertices, - int num_inputs, - int num_attribs, - int num_consts); - - -/*! - This function is used to execute the gallivm_prog in software. Before calling - this function the gallivm_prog has to be JIT compiled with the gallivm_cpu_jit_compile - function. - */ -int gallivm_prog_exec(struct gallivm_prog *prog, - struct tgsi_exec_vector *inputs, - struct tgsi_exec_vector *dests, - float (*consts)[4], - int num_vertices, - int num_inputs, - int num_attribs) -{ - vertex_shader_runner runner = reinterpret_cast(prog->function); - assert(runner); - runner(inputs, dests, consts, num_vertices, num_inputs, - num_attribs, prog->num_consts); - - return 0; -} - - - static inline void constant_interpolation(float (*inputs)[16][4], const struct tgsi_interp_coef *coefs, @@ -231,28 +195,6 @@ perspective_interpolation(float (*inputs)[16][4], } } -typedef int (*fragment_shader_runner)(float x, float y, - float (*dests)[16][4], - float (*inputs)[16][4], - int num_attribs, - float (*consts)[4], int num_consts, - struct tgsi_sampler *samplers); - -int gallivm_fragment_shader_exec(struct gallivm_prog *prog, - float fx, float fy, - float (*dests)[16][4], - float (*inputs)[16][4], - float (*consts)[4], - struct tgsi_sampler *samplers) -{ - fragment_shader_runner runner = reinterpret_cast(prog->function); - assert(runner); - - return runner(fx, fy, dests, inputs, prog->num_interp, - consts, prog->num_consts, - samplers); -} - void gallivm_ir_dump(struct gallivm_ir *ir, const char *file_prefix) { if (!ir || !ir->module) @@ -292,82 +234,6 @@ void gallivm_ir_dump(struct gallivm_ir *ir, const char *file_prefix) } -static struct gallivm_cpu_engine *CPU = 0; - -static inline llvm::Function *func_for_shader(struct gallivm_prog *prog) -{ - llvm::Module *mod = prog->module; - llvm::Function *func = 0; - - switch (prog->type) { - case GALLIVM_VS: - func = mod->getFunction("run_vertex_shader"); - break; - case GALLIVM_FS: - func = mod->getFunction("run_fragment_shader"); - break; - default: - assert(!"Unknown shader type!"); - break; - } - return func; -} - -/*! - This function creates a CPU based execution engine for the given gallivm_prog. - gallivm_cpu_engine should be used as a singleton throughout the library. Before - executing gallivm_prog_exec one needs to call gallivm_cpu_jit_compile. - The gallivm_prog instance which is being passed to the constructor is being - automatically JIT compiled so one shouldn't call gallivm_cpu_jit_compile - with it again. - */ -struct gallivm_cpu_engine * gallivm_cpu_engine_create(struct gallivm_prog *prog) -{ - struct gallivm_cpu_engine *cpu = (struct gallivm_cpu_engine *) - calloc(1, sizeof(struct gallivm_cpu_engine)); - llvm::Module *mod = static_cast(prog->module); - llvm::ExistingModuleProvider *mp = new llvm::ExistingModuleProvider(mod); - llvm::ExecutionEngine *ee = llvm::ExecutionEngine::create(mp, false); - ee->DisableLazyCompilation(); - cpu->engine = ee; - - llvm::Function *func = func_for_shader(prog); - - prog->function = ee->getPointerToFunction(func); - CPU = cpu; - return cpu; -} - - -/*! - This function JIT compiles the given gallivm_prog with the given cpu based execution engine. - The reference to the generated machine code entry point will be stored - in the gallivm_prog program. After executing this function one can call gallivm_prog_exec - in order to execute the gallivm_prog on the CPU. - */ -void gallivm_cpu_jit_compile(struct gallivm_cpu_engine *cpu, struct gallivm_prog *prog) -{ - llvm::Module *mod = static_cast(prog->module); - llvm::ExistingModuleProvider *mp = new llvm::ExistingModuleProvider(mod); - llvm::ExecutionEngine *ee = cpu->engine; - assert(ee); - ee->DisableLazyCompilation(); - ee->addModuleProvider(mp); - - llvm::Function *func = func_for_shader(prog); - prog->function = ee->getPointerToFunction(func); -} - -void gallivm_cpu_engine_delete(struct gallivm_cpu_engine *cpu) -{ - free(cpu); -} - -struct gallivm_cpu_engine * gallivm_global_cpu_engine() -{ - return CPU; -} - void gallivm_prog_inputs_interpolate(struct gallivm_prog *prog, float (*inputs)[16][4], const struct tgsi_interp_coef *coef) diff --git a/src/mesa/pipe/llvm/gallivm.h b/src/mesa/pipe/llvm/gallivm.h index f9f5d5ee744..98eda56f81c 100644 --- a/src/mesa/pipe/llvm/gallivm.h +++ b/src/mesa/pipe/llvm/gallivm.h @@ -61,28 +61,16 @@ enum gallivm_vector_layout { }; struct gallivm_ir *gallivm_ir_new(enum gallivm_shader_type type); -void gallivm_ir_set_layout(struct gallivm_ir *prog, - enum gallivm_vector_layout layout); -void gallivm_ir_set_components(struct gallivm_ir *prog, int num); -void gallivm_ir_fill_from_tgsi(struct gallivm_ir *prog, - const struct tgsi_token *tokens); -void gallivm_ir_delete(struct gallivm_ir *prog); +void gallivm_ir_set_layout(struct gallivm_ir *prog, + enum gallivm_vector_layout layout); +void gallivm_ir_set_components(struct gallivm_ir *prog, int num); +void gallivm_ir_fill_from_tgsi(struct gallivm_ir *prog, + const struct tgsi_token *tokens); +void gallivm_ir_delete(struct gallivm_ir *prog); + struct gallivm_prog *gallivm_ir_compile(struct gallivm_ir *ir); -int gallivm_prog_exec(struct gallivm_prog *prog, - struct tgsi_exec_vector *inputs, - struct tgsi_exec_vector *dests, - float (*consts)[4], - int num_vertices, - int num_inputs, - int num_attribs); -int gallivm_fragment_shader_exec(struct gallivm_prog *prog, - float x, float y, - float (*dests)[PIPE_MAX_SHADER_INPUTS][4], - float (*inputs)[PIPE_MAX_SHADER_INPUTS][4], - float (*consts)[4], - struct tgsi_sampler *samplers); void gallivm_prog_inputs_interpolate(struct gallivm_prog *prog, float (*inputs)[PIPE_MAX_SHADER_INPUTS][4], const struct tgsi_interp_coef *coefs); @@ -91,9 +79,20 @@ void gallivm_prog_dump(struct gallivm_prog *prog, const char *file_prefix); struct gallivm_cpu_engine *gallivm_cpu_engine_create(struct gallivm_prog *prog); struct gallivm_cpu_engine *gallivm_global_cpu_engine(); +int gallivm_cpu_vs_exec(struct gallivm_prog *prog, + struct tgsi_exec_vector *inputs, + struct tgsi_exec_vector *dests, + float (*consts)[4]); +int gallivm_cpu_fs_exec(struct gallivm_prog *prog, + float x, float y, + float (*dests)[PIPE_MAX_SHADER_INPUTS][4], + float (*inputs)[PIPE_MAX_SHADER_INPUTS][4], + float (*consts)[4], + struct tgsi_sampler *samplers); void gallivm_cpu_jit_compile(struct gallivm_cpu_engine *ee, struct gallivm_prog *prog); void gallivm_cpu_engine_delete(struct gallivm_cpu_engine *ee); + #endif /* MESA_LLVM */ #if defined __cplusplus diff --git a/src/mesa/pipe/llvm/gallivm_cpu.cpp b/src/mesa/pipe/llvm/gallivm_cpu.cpp new file mode 100644 index 00000000000..5f1268bf4f8 --- /dev/null +++ b/src/mesa/pipe/llvm/gallivm_cpu.cpp @@ -0,0 +1,204 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Zack Rusin zack@tungstengraphics.com + */ +#ifdef MESA_LLVM + +#include "gallivm.h" +#include "gallivm_p.h" + +#include "instructions.h" +#include "loweringpass.h" +#include "storage.h" +#include "tgsitollvm.h" + +#include "pipe/p_context.h" +#include "pipe/p_shader_tokens.h" + +#include "pipe/tgsi/exec/tgsi_exec.h" +#include "pipe/tgsi/util/tgsi_dump.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +struct gallivm_cpu_engine { + llvm::ExecutionEngine *engine; +}; + +static struct gallivm_cpu_engine *CPU = 0; + +typedef int (*fragment_shader_runner)(float x, float y, + float (*dests)[16][4], + float (*inputs)[16][4], + int num_attribs, + float (*consts)[4], int num_consts, + struct tgsi_sampler *samplers); + +int gallivm_cpu_fs_exec(struct gallivm_prog *prog, + float fx, float fy, + float (*dests)[16][4], + float (*inputs)[16][4], + float (*consts)[4], + struct tgsi_sampler *samplers) +{ + fragment_shader_runner runner = reinterpret_cast(prog->function); + assert(runner); + + return runner(fx, fy, dests, inputs, prog->num_interp, + consts, prog->num_consts, + samplers); +} + +static inline llvm::Function *func_for_shader(struct gallivm_prog *prog) +{ + llvm::Module *mod = prog->module; + llvm::Function *func = 0; + + switch (prog->type) { + case GALLIVM_VS: + func = mod->getFunction("run_vertex_shader"); + break; + case GALLIVM_FS: + func = mod->getFunction("run_fragment_shader"); + break; + default: + assert(!"Unknown shader type!"); + break; + } + return func; +} + +/*! + This function creates a CPU based execution engine for the given gallivm_prog. + gallivm_cpu_engine should be used as a singleton throughout the library. Before + executing gallivm_prog_exec one needs to call gallivm_cpu_jit_compile. + The gallivm_prog instance which is being passed to the constructor is being + automatically JIT compiled so one shouldn't call gallivm_cpu_jit_compile + with it again. + */ +struct gallivm_cpu_engine * gallivm_cpu_engine_create(struct gallivm_prog *prog) +{ + struct gallivm_cpu_engine *cpu = (struct gallivm_cpu_engine *) + calloc(1, sizeof(struct gallivm_cpu_engine)); + llvm::Module *mod = static_cast(prog->module); + llvm::ExistingModuleProvider *mp = new llvm::ExistingModuleProvider(mod); + llvm::ExecutionEngine *ee = llvm::ExecutionEngine::create(mp, false); + ee->DisableLazyCompilation(); + cpu->engine = ee; + + llvm::Function *func = func_for_shader(prog); + + prog->function = ee->getPointerToFunction(func); + CPU = cpu; + return cpu; +} + + +/*! + This function JIT compiles the given gallivm_prog with the given cpu based execution engine. + The reference to the generated machine code entry point will be stored + in the gallivm_prog program. After executing this function one can call gallivm_prog_exec + in order to execute the gallivm_prog on the CPU. + */ +void gallivm_cpu_jit_compile(struct gallivm_cpu_engine *cpu, struct gallivm_prog *prog) +{ + llvm::Module *mod = static_cast(prog->module); + llvm::ExistingModuleProvider *mp = new llvm::ExistingModuleProvider(mod); + llvm::ExecutionEngine *ee = cpu->engine; + assert(ee); + /*FIXME : remove */ + ee->DisableLazyCompilation(); + ee->addModuleProvider(mp); + + llvm::Function *func = func_for_shader(prog); + prog->function = ee->getPointerToFunction(func); +} + +void gallivm_cpu_engine_delete(struct gallivm_cpu_engine *cpu) +{ + free(cpu); +} + +struct gallivm_cpu_engine * gallivm_global_cpu_engine() +{ + return CPU; +} + + +typedef void (*vertex_shader_runner)(void *ainputs, + void *dests, + float (*aconsts)[4], + int num_vertices, + int num_inputs, + int num_attribs, + int num_consts); + + +/*! + This function is used to execute the gallivm_prog in software. Before calling + this function the gallivm_prog has to be JIT compiled with the gallivm_cpu_jit_compile + function. + */ +int gallivm_cpu_vs_exec(struct gallivm_prog *prog, + struct tgsi_exec_vector *inputs, + struct tgsi_exec_vector *dests, + float (*consts)[4]) +{ + vertex_shader_runner runner = reinterpret_cast(prog->function); + assert(runner); + /*FIXME*/ + runner(inputs, dests, consts, 4, 4, 4, prog->num_consts); + + return 0; +} + +#endif -- cgit v1.2.3 From 6f04f0f8aa652595d7c53bda5eeb304145fd02f2 Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Wed, 6 Feb 2008 06:12:34 -0500 Subject: disable llvm for fragment shaders for now --- src/mesa/pipe/softpipe/sp_state_fs.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/mesa/pipe/softpipe/sp_state_fs.c b/src/mesa/pipe/softpipe/sp_state_fs.c index 598a70c8279..0b814fc2847 100644 --- a/src/mesa/pipe/softpipe/sp_state_fs.c +++ b/src/mesa/pipe/softpipe/sp_state_fs.c @@ -62,11 +62,15 @@ softpipe_create_fs_state(struct pipe_context *pipe, #ifdef MESA_LLVM state->llvm_prog = 0; + +#if 0 if (!gallivm_global_cpu_engine()) { gallivm_cpu_engine_create(state->llvm_prog); } else gallivm_cpu_jit_compile(gallivm_global_cpu_engine(), state->llvm_prog); +#endif + #elif defined(__i386__) || defined(__386__) if (softpipe->use_sse) { x86_init_func( &state->sse2_program ); -- cgit v1.2.3 From 74295558492060694910892d843e89708272a271 Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Mon, 11 Feb 2008 09:43:59 -0500 Subject: start genering soa type code in llvm paths --- src/mesa/pipe/llvm/Makefile | 4 +- src/mesa/pipe/llvm/gallivm.cpp | 3 + src/mesa/pipe/llvm/gallivm.h | 8 +- src/mesa/pipe/llvm/instructionssoa.cpp | 26 ++ src/mesa/pipe/llvm/instructionssoa.h | 55 ++++ src/mesa/pipe/llvm/storagesoa.cpp | 117 ++++++++ src/mesa/pipe/llvm/storagesoa.h | 71 +++++ src/mesa/pipe/llvm/tgsitollvm.cpp | 506 ++++++++++++++++++++++++++++++++- src/mesa/pipe/llvm/tgsitollvm.h | 4 + 9 files changed, 782 insertions(+), 12 deletions(-) create mode 100644 src/mesa/pipe/llvm/instructionssoa.cpp create mode 100644 src/mesa/pipe/llvm/instructionssoa.h create mode 100644 src/mesa/pipe/llvm/storagesoa.cpp create mode 100644 src/mesa/pipe/llvm/storagesoa.h diff --git a/src/mesa/pipe/llvm/Makefile b/src/mesa/pipe/llvm/Makefile index b333661c6f8..a0494ba966e 100644 --- a/src/mesa/pipe/llvm/Makefile +++ b/src/mesa/pipe/llvm/Makefile @@ -11,7 +11,9 @@ GALLIVM_SOURCES = \ instructions.cpp \ loweringpass.cpp \ tgsitollvm.cpp \ - storage.cpp + storage.cpp \ + storagesoa.cpp \ + instructionssoa.cpp INC_SOURCES = gallivm_builtins.cpp llvm_base_shader.cpp diff --git a/src/mesa/pipe/llvm/gallivm.cpp b/src/mesa/pipe/llvm/gallivm.cpp index cb9a88f0b86..b99dc6db5b8 100644 --- a/src/mesa/pipe/llvm/gallivm.cpp +++ b/src/mesa/pipe/llvm/gallivm.cpp @@ -288,6 +288,9 @@ void gallivm_ir_fill_from_tgsi(struct gallivm_ir *ir, std::cout << "Creating llvm from: " <module = mod; gallivm_ir_dump(ir, 0); diff --git a/src/mesa/pipe/llvm/gallivm.h b/src/mesa/pipe/llvm/gallivm.h index 98eda56f81c..b104520cb7f 100644 --- a/src/mesa/pipe/llvm/gallivm.h +++ b/src/mesa/pipe/llvm/gallivm.h @@ -61,12 +61,12 @@ enum gallivm_vector_layout { }; struct gallivm_ir *gallivm_ir_new(enum gallivm_shader_type type); -void gallivm_ir_set_layout(struct gallivm_ir *prog, +void gallivm_ir_set_layout(struct gallivm_ir *ir, enum gallivm_vector_layout layout); -void gallivm_ir_set_components(struct gallivm_ir *prog, int num); -void gallivm_ir_fill_from_tgsi(struct gallivm_ir *prog, +void gallivm_ir_set_components(struct gallivm_ir *ir, int num); +void gallivm_ir_fill_from_tgsi(struct gallivm_ir *ir, const struct tgsi_token *tokens); -void gallivm_ir_delete(struct gallivm_ir *prog); +void gallivm_ir_delete(struct gallivm_ir *ir); struct gallivm_prog *gallivm_ir_compile(struct gallivm_ir *ir); diff --git a/src/mesa/pipe/llvm/instructionssoa.cpp b/src/mesa/pipe/llvm/instructionssoa.cpp new file mode 100644 index 00000000000..9ac4d8fbc73 --- /dev/null +++ b/src/mesa/pipe/llvm/instructionssoa.cpp @@ -0,0 +1,26 @@ +#include "instructionssoa.h" + +InstructionsSoa::InstructionsSoa(llvm::Module *mod, llvm::Function *func, + llvm::BasicBlock *block, StorageSoa *storage) +{ +} + +std::vector InstructionsSoa::add(const std::vector in1, + const std::vector in2) +{ + std::vector res(4); + + return res; +} + +std::vector InstructionsSoa::mul(const std::vector in1, + const std::vector in2) +{ + std::vector res(4); + + return res; +} + +void InstructionsSoa::end() +{ +} diff --git a/src/mesa/pipe/llvm/instructionssoa.h b/src/mesa/pipe/llvm/instructionssoa.h new file mode 100644 index 00000000000..0b6b41cf056 --- /dev/null +++ b/src/mesa/pipe/llvm/instructionssoa.h @@ -0,0 +1,55 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef INSTRUCTIONSSOA_H +#define INSTRUCTIONSSOA_H + +#include + +namespace llvm { + class Module; + class Function; + class BasicBlock; + class Value; +} +class StorageSoa; + +class InstructionsSoa +{ +public: + InstructionsSoa(llvm::Module *mod, llvm::Function *func, + llvm::BasicBlock *block, StorageSoa *storage); + + std::vector add(const std::vector in1, + const std::vector in2); + std::vector mul(const std::vector in1, + const std::vector in2); + void end(); +}; + + +#endif diff --git a/src/mesa/pipe/llvm/storagesoa.cpp b/src/mesa/pipe/llvm/storagesoa.cpp new file mode 100644 index 00000000000..b2aca3557a9 --- /dev/null +++ b/src/mesa/pipe/llvm/storagesoa.cpp @@ -0,0 +1,117 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "storagesoa.h" + + +#include "pipe/p_shader_tokens.h" +#include +#include +#include + +#include +#include +#include +#include +#include + +using namespace llvm; + +StorageSoa::StorageSoa(llvm::BasicBlock *block, + llvm::Value *input, + llvm::Value *output, + llvm::Value *consts) +{ +} + +void StorageSoa::addImmediate(float *vec) +{ +} + +llvm::Value *StorageSoa::addrElement(int idx) const +{ + return 0; +} + +std::vector StorageSoa::inputElement(int idx, int swizzle, + llvm::Value *indIdx) +{ + std::vector res(4); + + return res; +} + +std::vector StorageSoa::constElement(int idx, int swizzle, + llvm::Value *indIdx) +{ + std::vector res(4); + + return res; +} + +std::vector StorageSoa::outputElement(int idx, int swizzle, + llvm::Value *indIdx) +{ + std::vector res(4); + + return res; +} + +std::vector StorageSoa::tempElement(int idx, int swizzle, + llvm::Value *indIdx) +{ + std::vector res(4); + + return res; +} + +std::vector StorageSoa::immediateElement(int idx, int swizzle) +{ + std::vector res(4); + + return res; +} + +llvm::Value * StorageSoa::extractIndex(llvm::Value *vec) +{ + return 0; +} + +void StorageSoa::storeOutput(int dstIdx, const std::vector &val, + int mask) +{ +} + +void StorageSoa::storeTemp(int idx, const std::vector &val, + int mask) +{ +} + +void StorageSoa::storeAddress(int idx, const std::vector &val, + int mask) +{ +} diff --git a/src/mesa/pipe/llvm/storagesoa.h b/src/mesa/pipe/llvm/storagesoa.h new file mode 100644 index 00000000000..551b0b97342 --- /dev/null +++ b/src/mesa/pipe/llvm/storagesoa.h @@ -0,0 +1,71 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef STORAGESOA_H +#define STORAGESOA_H + +#include + +namespace llvm { + class BasicBlock; + class Constant; + class ConstantInt; + class LoadInst; + class Value; + class VectorType; +} + +class StorageSoa +{ +public: + StorageSoa(llvm::BasicBlock *block, + llvm::Value *input, + llvm::Value *output, + llvm::Value *consts); + + void addImmediate(float *vec); + + llvm::Value * addrElement(int idx) const; + + std::vector inputElement(int idx, int swizzle, llvm::Value *indIdx =0); + std::vector constElement(int idx, int swizzle, llvm::Value *indIdx =0); + std::vector outputElement(int idx, int swizzle, llvm::Value *indIdx =0); + std::vector tempElement(int idx, int swizzle, llvm::Value *indIdx =0); + std::vector immediateElement(int idx, int swizzle); + + llvm::Value *extractIndex(llvm::Value *vec); + + void storeOutput(int dstIdx, const std::vector &val, + int mask); + void storeTemp(int idx, const std::vector &val, + int mask); + void storeAddress(int idx, const std::vector &val, + int mask); + +}; + +#endif diff --git a/src/mesa/pipe/llvm/tgsitollvm.cpp b/src/mesa/pipe/llvm/tgsitollvm.cpp index eb9e1196f17..bc4df610717 100644 --- a/src/mesa/pipe/llvm/tgsitollvm.cpp +++ b/src/mesa/pipe/llvm/tgsitollvm.cpp @@ -5,6 +5,8 @@ #include "storage.h" #include "instructions.h" +#include "storagesoa.h" +#include "instructionssoa.h" #include "pipe/p_shader_tokens.h" @@ -112,6 +114,14 @@ translate_declaration(struct gallivm_ir *prog, } } +static void +translate_declarationir(struct gallivm_ir *, + llvm::Module *, + StorageSoa *, + struct tgsi_full_declaration *, + struct tgsi_full_declaration *) +{ +} static void translate_immediate(Storage *storage, @@ -120,29 +130,56 @@ translate_immediate(Storage *storage, float vec[4]; int i; for (i = 0; i < imm->Immediate.Size - 1; ++i) { - switch( imm->Immediate.DataType ) { + switch (imm->Immediate.DataType) { case TGSI_IMM_FLOAT32: vec[i] = imm->u.ImmediateFloat32[i].Float; break; default: - assert( 0 ); + assert(0); } } storage->addImmediate(vec); } -static inline llvm::Value * -swizzleVector(llvm::Value *val, struct tgsi_full_src_register *src, - Storage *storage) + +static void +translate_immediateir(StorageSoa *storage, + struct tgsi_full_immediate *imm) +{ + float vec[4]; + int i; + for (i = 0; i < imm->Immediate.Size - 1; ++i) { + switch (imm->Immediate.DataType) { + case TGSI_IMM_FLOAT32: + vec[i] = imm->u.ImmediateFloat32[i].Float; + break; + default: + assert(0); + } + } + storage->addImmediate(vec); +} + +static inline int +swizzleInt(struct tgsi_full_src_register *src) { int swizzle = 0; int start = 1000; - const int NO_SWIZZLE = TGSI_SWIZZLE_X * 1000 + TGSI_SWIZZLE_Y * 100 + - TGSI_SWIZZLE_Z * 10 + TGSI_SWIZZLE_W; + for (int k = 0; k < 4; ++k) { swizzle += tgsi_util_get_full_src_register_extswizzle(src, k) * start; start /= 10; } + return swizzle; +} + +static inline llvm::Value * +swizzleVector(llvm::Value *val, struct tgsi_full_src_register *src, + Storage *storage) +{ + int swizzle = swizzleInt(src); + const int NO_SWIZZLE = TGSI_SWIZZLE_X * 1000 + TGSI_SWIZZLE_Y * 100 + + TGSI_SWIZZLE_Z * 10 + TGSI_SWIZZLE_W; if (swizzle != NO_SWIZZLE) { /*fprintf(stderr, "XXXXXXXX swizzle = %d\n", swizzle);*/ val = storage->shuffleVector(val, swizzle); @@ -617,6 +654,392 @@ translate_instruction(llvm::Module *module, } } + +static void +translate_instructionir(llvm::Module *module, + StorageSoa *storage, + InstructionsSoa *instr, + struct tgsi_full_instruction *inst, + struct tgsi_full_instruction *fi, + unsigned instno) +{ + std::vector< std::vector > inputs(inst->Instruction.NumSrcRegs); + + for (int i = 0; i < inst->Instruction.NumSrcRegs; ++i) { + struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i]; + std::vector val; + llvm::Value *indIdx = 0; + int swizzle = swizzleInt(src); + + if (src->SrcRegister.Indirect) { + indIdx = storage->addrElement(src->SrcRegisterInd.Index); + indIdx = storage->extractIndex(indIdx); + } + if (src->SrcRegister.File == TGSI_FILE_CONSTANT) { + val = storage->constElement(src->SrcRegister.Index, swizzle, indIdx); + } else if (src->SrcRegister.File == TGSI_FILE_INPUT) { + val = storage->inputElement(src->SrcRegister.Index, swizzle, indIdx); + } else if (src->SrcRegister.File == TGSI_FILE_TEMPORARY) { + val = storage->tempElement(src->SrcRegister.Index, swizzle); + } else if (src->SrcRegister.File == TGSI_FILE_OUTPUT) { + val = storage->outputElement(src->SrcRegister.Index, swizzle, indIdx); + } else if (src->SrcRegister.File == TGSI_FILE_IMMEDIATE) { + val = storage->immediateElement(src->SrcRegister.Index, swizzle); + } else { + fprintf(stderr, "ERROR: not supported llvm source %d\n", src->SrcRegister.File); + return; + } + + inputs[i] = val; + } + + std::vector out(4); + switch (inst->Instruction.Opcode) { + case TGSI_OPCODE_ARL: { + } + break; + case TGSI_OPCODE_MOV: { + out = inputs[0]; + } + break; + case TGSI_OPCODE_LIT: { + } + break; + case TGSI_OPCODE_RCP: { + } + break; + case TGSI_OPCODE_RSQ: { + } + break; + case TGSI_OPCODE_EXP: + break; + case TGSI_OPCODE_LOG: + break; + case TGSI_OPCODE_MUL: { + out = instr->mul(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_ADD: { + out = instr->add(inputs[0], inputs[1]); + } + break; + case TGSI_OPCODE_DP3: { + } + break; + case TGSI_OPCODE_DP4: { + } + break; + case TGSI_OPCODE_DST: { + } + break; + case TGSI_OPCODE_MIN: { + } + break; + case TGSI_OPCODE_MAX: { + } + break; + case TGSI_OPCODE_SLT: { + } + break; + case TGSI_OPCODE_SGE: { + } + break; + case TGSI_OPCODE_MAD: { + } + break; + case TGSI_OPCODE_SUB: { + } + break; + case TGSI_OPCODE_LERP: { + } + break; + case TGSI_OPCODE_CND: + break; + case TGSI_OPCODE_CND0: + break; + case TGSI_OPCODE_DOT2ADD: + break; + case TGSI_OPCODE_INDEX: + break; + case TGSI_OPCODE_NEGATE: + break; + case TGSI_OPCODE_FRAC: { + } + break; + case TGSI_OPCODE_CLAMP: + break; + case TGSI_OPCODE_FLOOR: { + } + break; + case TGSI_OPCODE_ROUND: + break; + case TGSI_OPCODE_EXPBASE2: { + } + break; + case TGSI_OPCODE_LOGBASE2: { + } + break; + case TGSI_OPCODE_POWER: { + } + break; + case TGSI_OPCODE_CROSSPRODUCT: { + } + break; + case TGSI_OPCODE_MULTIPLYMATRIX: + break; + case TGSI_OPCODE_ABS: { + } + break; + case TGSI_OPCODE_RCC: + break; + case TGSI_OPCODE_DPH: { + } + break; + case TGSI_OPCODE_COS: { + } + break; + case TGSI_OPCODE_DDX: + break; + case TGSI_OPCODE_DDY: + break; + case TGSI_OPCODE_KILP: { + } + break; + case TGSI_OPCODE_PK2H: + break; + case TGSI_OPCODE_PK2US: + break; + case TGSI_OPCODE_PK4B: + break; + case TGSI_OPCODE_PK4UB: + break; + case TGSI_OPCODE_RFL: + break; + case TGSI_OPCODE_SEQ: + break; + case TGSI_OPCODE_SFL: + break; + case TGSI_OPCODE_SGT: { + } + break; + case TGSI_OPCODE_SIN: { + } + break; + case TGSI_OPCODE_SLE: + break; + case TGSI_OPCODE_SNE: + break; + case TGSI_OPCODE_STR: + break; + case TGSI_OPCODE_TEX: + break; + case TGSI_OPCODE_TXD: + break; + case TGSI_OPCODE_UP2H: + break; + case TGSI_OPCODE_UP2US: + break; + case TGSI_OPCODE_UP4B: + break; + case TGSI_OPCODE_UP4UB: + break; + case TGSI_OPCODE_X2D: + break; + case TGSI_OPCODE_ARA: + break; + case TGSI_OPCODE_ARR: + break; + case TGSI_OPCODE_BRA: + break; + case TGSI_OPCODE_CAL: { + } + break; + case TGSI_OPCODE_RET: { + } + break; + case TGSI_OPCODE_SSG: + break; + case TGSI_OPCODE_CMP: { + } + break; + case TGSI_OPCODE_SCS: { + } + break; + case TGSI_OPCODE_TXB: + break; + case TGSI_OPCODE_NRM: + break; + case TGSI_OPCODE_DIV: + break; + case TGSI_OPCODE_DP2: + break; + case TGSI_OPCODE_TXL: + break; + case TGSI_OPCODE_BRK: { + } + break; + case TGSI_OPCODE_IF: { + } + break; + case TGSI_OPCODE_LOOP: + break; + case TGSI_OPCODE_REP: + break; + case TGSI_OPCODE_ELSE: { + } + break; + case TGSI_OPCODE_ENDIF: { + } + break; + case TGSI_OPCODE_ENDLOOP: + break; + case TGSI_OPCODE_ENDREP: + break; + case TGSI_OPCODE_PUSHA: + break; + case TGSI_OPCODE_POPA: + break; + case TGSI_OPCODE_CEIL: + break; + case TGSI_OPCODE_I2F: + break; + case TGSI_OPCODE_NOT: + break; + case TGSI_OPCODE_TRUNC: { + } + break; + case TGSI_OPCODE_SHL: + break; + case TGSI_OPCODE_SHR: + break; + case TGSI_OPCODE_AND: + break; + case TGSI_OPCODE_OR: + break; + case TGSI_OPCODE_MOD: + break; + case TGSI_OPCODE_XOR: + break; + case TGSI_OPCODE_SAD: + break; + case TGSI_OPCODE_TXF: + break; + case TGSI_OPCODE_TXQ: + break; + case TGSI_OPCODE_CONT: + break; + case TGSI_OPCODE_EMIT: + break; + case TGSI_OPCODE_ENDPRIM: + break; + case TGSI_OPCODE_BGNLOOP2: { + } + break; + case TGSI_OPCODE_BGNSUB: { + } + break; + case TGSI_OPCODE_ENDLOOP2: { + } + break; + case TGSI_OPCODE_ENDSUB: { + } + break; + case TGSI_OPCODE_NOISE1: + break; + case TGSI_OPCODE_NOISE2: + break; + case TGSI_OPCODE_NOISE3: + break; + case TGSI_OPCODE_NOISE4: + break; + case TGSI_OPCODE_NOP: + break; + case TGSI_OPCODE_TEXBEM: + break; + case TGSI_OPCODE_TEXBEML: + break; + case TGSI_OPCODE_TEXREG2AR: + break; + case TGSI_OPCODE_TEXM3X2PAD: + break; + case TGSI_OPCODE_TEXM3X2TEX: + break; + case TGSI_OPCODE_TEXM3X3PAD: + break; + case TGSI_OPCODE_TEXM3X3TEX: + break; + case TGSI_OPCODE_TEXM3X3SPEC: + break; + case TGSI_OPCODE_TEXM3X3VSPEC: + break; + case TGSI_OPCODE_TEXREG2GB: + break; + case TGSI_OPCODE_TEXREG2RGB: + break; + case TGSI_OPCODE_TEXDP3TEX: + break; + case TGSI_OPCODE_TEXDP3: + break; + case TGSI_OPCODE_TEXM3X3: + break; + case TGSI_OPCODE_TEXM3X2DEPTH: + break; + case TGSI_OPCODE_TEXDEPTH: + break; + case TGSI_OPCODE_BEM: + break; + case TGSI_OPCODE_M4X3: + break; + case TGSI_OPCODE_M3X4: + break; + case TGSI_OPCODE_M3X3: + break; + case TGSI_OPCODE_M3X2: + break; + case TGSI_OPCODE_NRM4: + break; + case TGSI_OPCODE_CALLNZ: + break; + case TGSI_OPCODE_IFC: + break; + case TGSI_OPCODE_BREAKC: + break; + case TGSI_OPCODE_KIL: + break; + case TGSI_OPCODE_END: + instr->end(); + return; + break; + default: + fprintf(stderr, "ERROR: Unknown opcode %d\n", + inst->Instruction.Opcode); + assert(0); + break; + } + + if (!out[0]) { + fprintf(stderr, "ERROR: unsupported opcode %d\n", + inst->Instruction.Opcode); + assert(!"Unsupported opcode"); + } + + /* store results */ + for (int i = 0; i < inst->Instruction.NumDstRegs; ++i) { + struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i]; + + if (dst->DstRegister.File == TGSI_FILE_OUTPUT) { + storage->storeOutput(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); + } else if (dst->DstRegister.File == TGSI_FILE_TEMPORARY) { + storage->storeTemp(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); + } else if (dst->DstRegister.File == TGSI_FILE_ADDRESS) { + storage->storeAddress(dst->DstRegister.Index, out, dst->DstRegister.WriteMask); + } else { + fprintf(stderr, "ERROR: unsupported LLVM destination!"); + assert(!"wrong destination"); + } + } +} + llvm::Module * tgsi_to_llvm(struct gallivm_ir *ir, const struct tgsi_token *tokens) { @@ -680,3 +1103,72 @@ tgsi_to_llvm(struct gallivm_ir *ir, const struct tgsi_token *tokens) ir->num_consts = storage.numConsts(); return mod; } + +llvm::Module * tgsi_to_llvmir(struct gallivm_ir *ir, + const struct tgsi_token *tokens) +{ + llvm::Module *mod = createBaseShader(); + struct tgsi_parse_context parse; + struct tgsi_full_instruction fi; + struct tgsi_full_declaration fd; + unsigned instno = 0; + Function* shader = mod->getFunction("execute_shader"); + std::ostringstream stream; + if (ir->type == GALLIVM_VS) { + stream << "vs_shader"; + } else { + stream << "fs_shader"; + } + stream << ir->id; + std::string func_name = stream.str(); + shader->setName(func_name.c_str()); + + Function::arg_iterator args = shader->arg_begin(); + Value *input = args++; + input->setName("input"); + Value *output = args++; + output->setName("output"); + Value *consts = args++; + consts->setName("consts"); + + BasicBlock *label_entry = new BasicBlock("entry", shader, 0); + + tgsi_parse_init(&parse, tokens); + + fi = tgsi_default_full_instruction(); + fd = tgsi_default_full_declaration(); + + StorageSoa storage(label_entry, input, output, consts); + InstructionsSoa instr(mod, shader, label_entry, &storage); + + while(!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + translate_declarationir(ir, mod, &storage, + &parse.FullToken.FullDeclaration, + &fd); + break; + + case TGSI_TOKEN_TYPE_IMMEDIATE: + translate_immediateir(&storage, + &parse.FullToken.FullImmediate); + break; + + case TGSI_TOKEN_TYPE_INSTRUCTION: + translate_instructionir(mod, &storage, &instr, + &parse.FullToken.FullInstruction, + &fi, instno); + ++instno; + break; + + default: + assert(0); + } + } + + tgsi_parse_free(&parse); + + return mod; +} diff --git a/src/mesa/pipe/llvm/tgsitollvm.h b/src/mesa/pipe/llvm/tgsitollvm.h index 073ffb5749a..7ada04d6299 100644 --- a/src/mesa/pipe/llvm/tgsitollvm.h +++ b/src/mesa/pipe/llvm/tgsitollvm.h @@ -13,4 +13,8 @@ struct tgsi_token; llvm::Module * tgsi_to_llvm(struct gallivm_ir *ir, const struct tgsi_token *tokens); + +llvm::Module * tgsi_to_llvmir(struct gallivm_ir *ir, + const struct tgsi_token *tokens); + #endif -- cgit v1.2.3 From a8637fb991e0b897a3574e0dc7d0ce15d7cbe8bc Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Mon, 11 Feb 2008 10:47:28 -0500 Subject: remove seperate llvm vs entry points they're not necessary anymore. we use the same paths as sse and tgsi code --- src/mesa/pipe/draw/draw_vertex_shader_llvm.c | 194 --------------------------- src/mesa/sources | 1 - 2 files changed, 195 deletions(-) delete mode 100644 src/mesa/pipe/draw/draw_vertex_shader_llvm.c diff --git a/src/mesa/pipe/draw/draw_vertex_shader_llvm.c b/src/mesa/pipe/draw/draw_vertex_shader_llvm.c deleted file mode 100644 index 63551c993e7..00000000000 --- a/src/mesa/pipe/draw/draw_vertex_shader_llvm.c +++ /dev/null @@ -1,194 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - /* - * Authors: - * Zack Rusin zack@tungstengraphics.com - */ - -#include "pipe/p_util.h" -#include "draw_private.h" -#include "draw_context.h" - -#ifdef MESA_LLVM - -#include "pipe/llvm/gallivm.h" -#include "pipe/p_shader_tokens.h" - -#define DBG 0 - -static INLINE void -fetch_attrib4(const void *ptr, enum pipe_format format, float attrib[4]) -{ - /* defaults */ - attrib[1] = 0.0; - attrib[2] = 0.0; - attrib[3] = 1.0; - switch (format) { - case PIPE_FORMAT_R32G32B32A32_FLOAT: - attrib[3] = ((float *) ptr)[3]; - /* fall-through */ - case PIPE_FORMAT_R32G32B32_FLOAT: - attrib[2] = ((float *) ptr)[2]; - /* fall-through */ - case PIPE_FORMAT_R32G32_FLOAT: - attrib[1] = ((float *) ptr)[1]; - /* fall-through */ - case PIPE_FORMAT_R32_FLOAT: - attrib[0] = ((float *) ptr)[0]; - break; - default: - assert(0); - } -} - - -/** - * Fetch vertex attributes for 'count' vertices. - */ -static INLINE -void vertex_fetch(struct draw_context *draw, - const unsigned elt, - float (*inputs)[4]) -{ - uint attr; - - /* loop over vertex attributes (vertex shader inputs) */ - for (attr = 0; attr < draw->vertex_shader->state->num_inputs; attr++) { - - unsigned buf = draw->vertex_element[attr].vertex_buffer_index; - const void *src - = (const void *) ((const ubyte *) draw->user.vbuffer[buf] - + draw->vertex_buffer[buf].buffer_offset - + draw->vertex_element[attr].src_offset - + elt * draw->vertex_buffer[buf].pitch); - fetch_attrib4(src, draw->vertex_element[attr].src_format, inputs[attr]); - } -} - -static INLINE unsigned -compute_clipmask(const float *clip, const float (*plane)[4], unsigned nr) -{ - unsigned mask = 0; - unsigned i; - - for (i = 0; i < nr; i++) { - if (dot4(clip, plane[i]) < 0) - mask |= (1<user.constants; - struct gallivm_prog *prog = draw->vertex_shader->llvm_prog; - const float *scale = draw->viewport.scale; - const float *trans = draw->viewport.translate; - /* fetch the inputs */ - for (i = 0; i < draw->vs.queue_nr; ++i) { - unsigned elt = draw->vs.queue[i].elt; - dests[i] = draw->vs.queue[i].dest; - vertex_fetch(draw, elt, inputs[i]); - } - - /* batch execute the shaders on all the vertices */ - gallivm_prog_exec(prog, inputs, outputs, consts, - draw->vs.queue_nr, - draw->vertex_shader->state->num_inputs, - draw->vertex_shader->state->num_outputs); - - - /* store machine results */ - for (int i = 0; i < draw->vs.queue_nr; ++i) { - unsigned slot; - float x, y, z, w; - struct vertex_header *vOut = draw->vs.queue[i].dest; - float (*dests)[4] = outputs[i]; - - /* Handle attr[0] (position) specially: - * - * XXX: Computing the clipmask should be done in the vertex - * program as a set of DP4 instructions appended to the - * user-provided code. - */ - x = vOut->clip[0] = dests[0][0]; - y = vOut->clip[1] = dests[0][1]; - z = vOut->clip[2] = dests[0][2]; - w = vOut->clip[3] = dests[0][3]; -#if DBG - debug_printf("output %d: %f %f %f %f\n", 0, x, y, z, w); -#endif - - vOut->clipmask = compute_clipmask(vOut->clip, draw->plane, draw->nr_planes); - vOut->edgeflag = 1; - /* divide by w */ - w = 1.0f / w; - x *= w; - y *= w; - z *= w; - - /* Viewport mapping */ - vOut->data[0][0] = x * scale[0] + trans[0]; - vOut->data[0][1] = y * scale[1] + trans[1]; - vOut->data[0][2] = z * scale[2] + trans[2]; - vOut->data[0][3] = w; - - /* Remaining attributes are packed into sequential post-transform - * vertex attrib slots. - */ - for (slot = 1; slot < draw->num_vs_outputs; slot++) { - vOut->data[slot][0] = dests[slot][0]; - vOut->data[slot][1] = dests[slot][1]; - vOut->data[slot][2] = dests[slot][2]; - vOut->data[slot][3] = dests[slot][3]; - -#if DBG - debug_printf("output %d: %f %f %f %f\n", slot, - vOut->data[slot][0], - vOut->data[slot][1], - vOut->data[slot][2], - vOut->data[slot][3]); -#endif - } - } /* loop over vertices */ - - draw->vs.queue_nr = 0; -} - -#endif /* MESA_LLVM */ diff --git a/src/mesa/sources b/src/mesa/sources index 84492c91ac8..96ae3dbca00 100644 --- a/src/mesa/sources +++ b/src/mesa/sources @@ -174,7 +174,6 @@ DRAW_SOURCES = \ pipe/draw/draw_vertex_cache.c \ pipe/draw/draw_vertex_fetch.c \ pipe/draw/draw_vertex_shader.c \ - pipe/draw/draw_vertex_shader_llvm.c \ pipe/draw/draw_vf.c \ pipe/draw/draw_vf_generic.c \ pipe/draw/draw_vf_sse.c \ -- cgit v1.2.3 From f554db1893749043fe5b2906f7075588be178884 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 16:35:44 -0700 Subject: checkpoint- remove dependencies on gl_texture_format to make code re-usable by state tracker --- src/mesa/main/mipmap.c | 424 +++++++++++++++++++++++++++++++------------------ 1 file changed, 266 insertions(+), 158 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index 1e61829e8f1..013dc3752ec 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -36,27 +36,205 @@ +static GLint +bytes_per_pixel(GLenum datatype, GLuint comps) +{ + GLint b = _mesa_sizeof_packed_type(datatype); + assert(b >= 0); + return b * comps; +} + + +static void +mesa_format_to_type_and_comps(const struct gl_texture_format *format, + GLenum *datatype, GLuint *comps) +{ + switch (format->MesaFormat) { + case MESA_FORMAT_RGBA8888: + case MESA_FORMAT_RGBA8888_REV: + case MESA_FORMAT_ARGB8888: + case MESA_FORMAT_ARGB8888_REV: + *datatype = CHAN_TYPE; + *comps = 4; + return; + case MESA_FORMAT_RGB888: + case MESA_FORMAT_BGR888: + *datatype = GL_UNSIGNED_BYTE; + *comps = 3; + return; + case MESA_FORMAT_RGB565: + case MESA_FORMAT_RGB565_REV: + *datatype = GL_UNSIGNED_SHORT_5_6_5; + *comps = 3; + return; + + case MESA_FORMAT_ARGB4444: + case MESA_FORMAT_ARGB4444_REV: + *datatype = GL_UNSIGNED_SHORT_4_4_4_4; + *comps = 4; + return; + + case MESA_FORMAT_ARGB1555: + case MESA_FORMAT_ARGB1555_REV: + *datatype = GL_UNSIGNED_SHORT_1_5_5_5_REV; + *comps = 3; + return; + + case MESA_FORMAT_AL88: + case MESA_FORMAT_AL88_REV: + *datatype = GL_UNSIGNED_BYTE; + *comps = 2; + return; + case MESA_FORMAT_RGB332: + *datatype = GL_UNSIGNED_BYTE_3_3_2; + *comps = 3; + return; + + case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: + case MESA_FORMAT_CI8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 1; + return; + + case MESA_FORMAT_YCBCR: + case MESA_FORMAT_YCBCR_REV: + *datatype = GL_UNSIGNED_SHORT; + *comps = 2; + return; + + case MESA_FORMAT_Z24_S8: + *datatype = GL_UNSIGNED_INT; + *comps = 1; /* XXX OK? */ + return; + + case MESA_FORMAT_Z16: + *datatype = GL_UNSIGNED_SHORT; + *comps = 1; + return; + + case MESA_FORMAT_Z32: + *datatype = GL_UNSIGNED_INT; + *comps = 1; + return; + + case MESA_FORMAT_SRGB8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 3; + return; + case MESA_FORMAT_SRGBA8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 4; + return; + case MESA_FORMAT_SL8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 1; + return; + case MESA_FORMAT_SLA8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 2; + return; + + case MESA_FORMAT_RGB_FXT1: + case MESA_FORMAT_RGBA_FXT1: + case MESA_FORMAT_RGB_DXT1: + case MESA_FORMAT_RGBA_DXT1: + case MESA_FORMAT_RGBA_DXT3: + case MESA_FORMAT_RGBA_DXT5: + /* XXX generate error instead? */ + *datatype = GL_UNSIGNED_BYTE; + *comps = 0; + return; + + case MESA_FORMAT_RGBA: + *datatype = CHAN_TYPE; + *comps = 4; + return; + case MESA_FORMAT_RGB: + *datatype = CHAN_TYPE; + *comps = 3; + return; + case MESA_FORMAT_LUMINANCE_ALPHA: + *datatype = CHAN_TYPE; + *comps = 2; + return; + case MESA_FORMAT_ALPHA: + case MESA_FORMAT_LUMINANCE: + case MESA_FORMAT_INTENSITY: + *datatype = CHAN_TYPE; + *comps = 1; + return; + + case MESA_FORMAT_RGBA_FLOAT32: + *datatype = GL_FLOAT; + *comps = 4; + return; + case MESA_FORMAT_RGBA_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 4; + return; + case MESA_FORMAT_RGB_FLOAT32: + *datatype = GL_FLOAT; + *comps = 3; + return; + case MESA_FORMAT_RGB_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 3; + return; + case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32: + *datatype = GL_FLOAT; + *comps = 2; + return; + case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 2; + return; + case MESA_FORMAT_ALPHA_FLOAT32: + case MESA_FORMAT_LUMINANCE_FLOAT32: + case MESA_FORMAT_INTENSITY_FLOAT32: + *datatype = GL_FLOAT; + *comps = 1; + return; + case MESA_FORMAT_ALPHA_FLOAT16: + case MESA_FORMAT_LUMINANCE_FLOAT16: + case MESA_FORMAT_INTENSITY_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 1; + return; + + default: + _mesa_problem(NULL, "bad texture format in mesa_format_to_type_and_comps"); + *datatype = 0; + *comps = 1; + } +} + + /** * Average together two rows of a source image to produce a single new * row in the dest image. It's legal for the two source rows to point * to the same data. The source width must be equal to either the * dest width or two times the dest width. + * \param datatype GL_UNSIGNED_BYTE, GL_UNSIGNED_SHORT, GL_FLOAT, etc. + * \param comps number of components per pixel (1..4) */ static void -do_row(const struct gl_texture_format *format, GLint srcWidth, +do_row(GLenum datatype, GLuint comps, GLint srcWidth, const GLvoid *srcRowA, const GLvoid *srcRowB, GLint dstWidth, GLvoid *dstRow) { const GLuint k0 = (srcWidth == dstWidth) ? 0 : 1; const GLuint colStride = (srcWidth == dstWidth) ? 1 : 2; + ASSERT(comps >= 1); + ASSERT(comps <= 4); + /* This assertion is no longer valid with non-power-of-2 textures assert(srcWidth == dstWidth || srcWidth == 2 * dstWidth); */ - switch (format->MesaFormat) { - case MESA_FORMAT_RGBA: - { + if (datatype == CHAN_TYPE && comps == 4) { GLuint i, j, k; const GLchan (*rowA)[4] = (const GLchan (*)[4]) srcRowA; const GLchan (*rowB)[4] = (const GLchan (*)[4]) srcRowB; @@ -72,10 +250,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) / 4; } - } - return; - case MESA_FORMAT_RGB: - { + } + else if (datatype == CHAN_TYPE && comps == 3) { GLuint i, j, k; const GLchan (*rowA)[3] = (const GLchan (*)[3]) srcRowA; const GLchan (*rowB)[3] = (const GLchan (*)[3]) srcRowB; @@ -89,12 +265,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; } - } - return; - case MESA_FORMAT_ALPHA: - case MESA_FORMAT_LUMINANCE: - case MESA_FORMAT_INTENSITY: - { + } + else if (datatype == CHAN_TYPE && comps == 1) { GLuint i, j, k; const GLchan *rowA = (const GLchan *) srcRowA; const GLchan *rowB = (const GLchan *) srcRowB; @@ -103,10 +275,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, i++, j += colStride, k += colStride) { dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; } - } - return; - case MESA_FORMAT_LUMINANCE_ALPHA: - { + } + else if (datatype == CHAN_TYPE && comps == 2) { GLuint i, j, k; const GLchan (*rowA)[2] = (const GLchan (*)[2]) srcRowA; const GLchan (*rowB)[2] = (const GLchan (*)[2]) srcRowB; @@ -118,10 +288,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; } - } - return; - case MESA_FORMAT_Z32: - { + } + else if (datatype == GL_UNSIGNED_INT && comps == 1) { GLuint i, j, k; const GLuint *rowA = (const GLuint *) srcRowA; const GLuint *rowB = (const GLuint *) srcRowB; @@ -130,10 +298,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, i++, j += colStride, k += colStride) { dst[i] = rowA[j] / 4 + rowA[k] / 4 + rowB[j] / 4 + rowB[k] / 4; } - } - return; - case MESA_FORMAT_Z16: - { + } + else if (datatype == GL_UNSIGNED_SHORT && comps == 1) { GLuint i, j, k; const GLushort *rowA = (const GLushort *) srcRowA; const GLushort *rowB = (const GLushort *) srcRowB; @@ -142,17 +308,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, i++, j += colStride, k += colStride) { dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; } - } - return; - /* Begin hardware formats */ - case MESA_FORMAT_RGBA8888: - case MESA_FORMAT_RGBA8888_REV: - case MESA_FORMAT_ARGB8888: - case MESA_FORMAT_ARGB8888_REV: -#if FEATURE_EXT_texture_sRGB - case MESA_FORMAT_SRGBA8: -#endif - { + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 4) { GLuint i, j, k; const GLubyte (*rowA)[4] = (const GLubyte (*)[4]) srcRowA; const GLubyte (*rowB)[4] = (const GLubyte (*)[4]) srcRowB; @@ -168,14 +325,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) / 4; } - } - return; - case MESA_FORMAT_RGB888: - case MESA_FORMAT_BGR888: -#if FEATURE_EXT_texture_sRGB - case MESA_FORMAT_SRGB8: -#endif - { + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 3) { GLuint i, j, k; const GLubyte (*rowA)[3] = (const GLubyte (*)[3]) srcRowA; const GLubyte (*rowB)[3] = (const GLubyte (*)[3]) srcRowB; @@ -189,11 +340,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; } - } - return; - case MESA_FORMAT_RGB565: - case MESA_FORMAT_RGB565_REV: - { + } + else if (datatype == GL_UNSIGNED_SHORT_5_6_5 && comps == 3) { GLuint i, j, k; const GLushort *rowA = (const GLushort *) srcRowA; const GLushort *rowB = (const GLushort *) srcRowB; @@ -217,11 +365,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; dst[i] = (blue << 11) | (green << 5) | red; } - } - return; - case MESA_FORMAT_ARGB4444: - case MESA_FORMAT_ARGB4444_REV: - { + } + else if (datatype == GL_UNSIGNED_SHORT_4_4_4_4 && comps == 4) { GLuint i, j, k; const GLushort *rowA = (const GLushort *) srcRowA; const GLushort *rowB = (const GLushort *) srcRowB; @@ -250,11 +395,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, const GLint alpha = (rowAa0 + rowAa1 + rowBa0 + rowBa1) >> 2; dst[i] = (alpha << 12) | (blue << 8) | (green << 4) | red; } - } - return; - case MESA_FORMAT_ARGB1555: - case MESA_FORMAT_ARGB1555_REV: /* XXX broken? */ - { + } + else if (datatype == GL_UNSIGNED_SHORT_1_5_5_5_REV && comps == 4) { GLuint i, j, k; const GLushort *rowA = (const GLushort *) srcRowA; const GLushort *rowB = (const GLushort *) srcRowB; @@ -283,14 +425,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, const GLint alpha = (rowAa0 + rowAa1 + rowBa0 + rowBa1) >> 2; dst[i] = (alpha << 15) | (blue << 10) | (green << 5) | red; } - } - return; - case MESA_FORMAT_AL88: - case MESA_FORMAT_AL88_REV: -#if FEATURE_EXT_texture_sRGB - case MESA_FORMAT_SLA8: -#endif - { + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 2) { GLuint i, j, k; const GLubyte (*rowA)[2] = (const GLubyte (*)[2]) srcRowA; const GLubyte (*rowB)[2] = (const GLubyte (*)[2]) srcRowB; @@ -302,10 +438,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) >> 2; } - } - return; - case MESA_FORMAT_RGB332: - { + } + else if (datatype == GL_UNSIGNED_BYTE_3_3_2 && comps == 3) { GLuint i, j, k; const GLubyte *rowA = (const GLubyte *) srcRowA; const GLubyte *rowB = (const GLubyte *) srcRowB; @@ -329,16 +463,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; dst[i] = (blue << 5) | (green << 2) | red; } - } - return; - case MESA_FORMAT_A8: - case MESA_FORMAT_L8: - case MESA_FORMAT_I8: - case MESA_FORMAT_CI8: -#if FEATURE_EXT_texture_sRGB - case MESA_FORMAT_SL8: -#endif - { + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 1) { GLuint i, j, k; const GLubyte *rowA = (const GLubyte *) srcRowA; const GLubyte *rowB = (const GLubyte *) srcRowB; @@ -347,10 +473,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, i++, j += colStride, k += colStride) { dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) >> 2; } - } - return; - case MESA_FORMAT_RGBA_FLOAT32: - { + } + else if (datatype == GL_FLOAT && comps == 4) { GLuint i, j, k; const GLfloat (*rowA)[4] = (const GLfloat (*)[4]) srcRowA; const GLfloat (*rowB)[4] = (const GLfloat (*)[4]) srcRowB; @@ -366,10 +490,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) * 0.25F; } - } - return; - case MESA_FORMAT_RGBA_FLOAT16: - { + } + else if (datatype == GL_HALF_FLOAT_ARB && comps == 4) { GLuint i, j, k, comp; const GLhalfARB (*rowA)[4] = (const GLhalfARB (*)[4]) srcRowA; const GLhalfARB (*rowB)[4] = (const GLhalfARB (*)[4]) srcRowB; @@ -385,10 +507,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } } - } - return; - case MESA_FORMAT_RGB_FLOAT32: - { + } + else if (datatype == GL_FLOAT && comps == 3) { GLuint i, j, k; const GLfloat (*rowA)[3] = (const GLfloat (*)[3]) srcRowA; const GLfloat (*rowB)[3] = (const GLfloat (*)[3]) srcRowB; @@ -402,10 +522,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) * 0.25F; } - } - return; - case MESA_FORMAT_RGB_FLOAT16: - { + } + else if (datatype == GL_HALF_FLOAT_ARB && comps == 3) { GLuint i, j, k, comp; const GLhalfARB (*rowA)[3] = (const GLhalfARB (*)[3]) srcRowA; const GLhalfARB (*rowB)[3] = (const GLhalfARB (*)[3]) srcRowB; @@ -421,10 +539,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } } - } - return; - case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32: - { + } + else if (datatype == GL_FLOAT && comps == 2) { GLuint i, j, k; const GLfloat (*rowA)[2] = (const GLfloat (*)[2]) srcRowA; const GLfloat (*rowB)[2] = (const GLfloat (*)[2]) srcRowB; @@ -436,10 +552,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) * 0.25F; } - } - return; - case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16: - { + } + else if (datatype == GL_HALF_FLOAT_ARB && comps == 2) { GLuint i, j, k, comp; const GLhalfARB (*rowA)[2] = (const GLhalfARB (*)[2]) srcRowA; const GLhalfARB (*rowB)[2] = (const GLhalfARB (*)[2]) srcRowB; @@ -455,12 +569,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } } - } - return; - case MESA_FORMAT_ALPHA_FLOAT32: - case MESA_FORMAT_LUMINANCE_FLOAT32: - case MESA_FORMAT_INTENSITY_FLOAT32: - { + } + else if (datatype == GL_FLOAT && comps == 1) { GLuint i, j, k; const GLfloat *rowA = (const GLfloat *) srcRowA; const GLfloat *rowB = (const GLfloat *) srcRowB; @@ -469,12 +579,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, i++, j += colStride, k += colStride) { dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) * 0.25F; } - } - return; - case MESA_FORMAT_ALPHA_FLOAT16: - case MESA_FORMAT_LUMINANCE_FLOAT16: - case MESA_FORMAT_INTENSITY_FLOAT16: - { + } + else if (datatype == GL_HALF_FLOAT_ARB && comps == 1) { GLuint i, j, k; const GLhalfARB *rowA = (const GLhalfARB *) srcRowA; const GLhalfARB *rowB = (const GLhalfARB *) srcRowB; @@ -488,10 +594,8 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, bk = _mesa_half_to_float(rowB[k]); dst[i] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } - } - return; - - default: + } + else { _mesa_problem(NULL, "bad format in do_row()"); } } @@ -504,11 +608,11 @@ do_row(const struct gl_texture_format *format, GLint srcWidth, */ static void -make_1d_mipmap(const struct gl_texture_format *format, GLint border, +make_1d_mipmap(GLenum datatype, GLuint comps, GLint border, GLint srcWidth, const GLubyte *srcPtr, GLint dstWidth, GLubyte *dstPtr) { - const GLint bpt = format->TexelBytes; + const GLint bpt = bytes_per_pixel(datatype, comps); const GLubyte *src; GLubyte *dst; @@ -517,7 +621,7 @@ make_1d_mipmap(const struct gl_texture_format *format, GLint border, dst = dstPtr + border * bpt; /* we just duplicate the input row, kind of hack, saves code */ - do_row(format, srcWidth - 2 * border, src, src, + do_row(datatype, comps, srcWidth - 2 * border, src, src, dstWidth - 2 * border, dst); if (border) { @@ -535,11 +639,11 @@ make_1d_mipmap(const struct gl_texture_format *format, GLint border, * XXX need to use the tex image's row stride! */ static void -make_2d_mipmap(const struct gl_texture_format *format, GLint border, +make_2d_mipmap(GLenum datatype, GLuint comps, GLint border, GLint srcWidth, GLint srcHeight, const GLubyte *srcPtr, GLint dstWidth, GLint dstHeight, GLubyte *dstPtr) { - const GLint bpt = format->TexelBytes; + const GLint bpt = bytes_per_pixel(datatype, comps); const GLint srcWidthNB = srcWidth - 2 * border; /* sizes w/out border */ const GLint dstWidthNB = dstWidth - 2 * border; const GLint dstHeightNB = dstHeight - 2 * border; @@ -558,7 +662,7 @@ make_2d_mipmap(const struct gl_texture_format *format, GLint border, dst = dstPtr + border * ((dstWidth + 1) * bpt); for (row = 0; row < dstHeightNB; row++) { - do_row(format, srcWidthNB, srcA, srcB, + do_row(datatype, comps, srcWidthNB, srcA, srcB, dstWidthNB, dst); srcA += 2 * srcRowStride; srcB += 2 * srcRowStride; @@ -580,12 +684,12 @@ make_2d_mipmap(const struct gl_texture_format *format, GLint border, MEMCPY(dstPtr + (dstWidth * dstHeight - 1) * bpt, srcPtr + (srcWidth * srcHeight - 1) * bpt, bpt); /* lower border */ - do_row(format, srcWidthNB, + do_row(datatype, comps, srcWidthNB, srcPtr + bpt, srcPtr + bpt, dstWidthNB, dstPtr + bpt); /* upper border */ - do_row(format, srcWidthNB, + do_row(datatype, comps, srcWidthNB, srcPtr + (srcWidth * (srcHeight - 1) + 1) * bpt, srcPtr + (srcWidth * (srcHeight - 1) + 1) * bpt, dstWidthNB, @@ -603,11 +707,11 @@ make_2d_mipmap(const struct gl_texture_format *format, GLint border, else { /* average two src pixels each dest pixel */ for (row = 0; row < dstHeightNB; row += 2) { - do_row(format, 1, + do_row(datatype, comps, 1, srcPtr + (srcWidth * (row * 2 + 1)) * bpt, srcPtr + (srcWidth * (row * 2 + 2)) * bpt, 1, dstPtr + (dstWidth * row + 1) * bpt); - do_row(format, 1, + do_row(datatype, comps, 1, srcPtr + (srcWidth * (row * 2 + 1) + srcWidth - 1) * bpt, srcPtr + (srcWidth * (row * 2 + 2) + srcWidth - 1) * bpt, 1, dstPtr + (dstWidth * row + 1 + dstWidth - 1) * bpt); @@ -618,13 +722,13 @@ make_2d_mipmap(const struct gl_texture_format *format, GLint border, static void -make_3d_mipmap(const struct gl_texture_format *format, GLint border, +make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, GLint srcWidth, GLint srcHeight, GLint srcDepth, const GLubyte *srcPtr, GLint dstWidth, GLint dstHeight, GLint dstDepth, GLubyte *dstPtr) { - const GLint bpt = format->TexelBytes; + const GLint bpt = bytes_per_pixel(datatype, comps); const GLint srcWidthNB = srcWidth - 2 * border; /* sizes w/out border */ const GLint srcDepthNB = srcDepth - 2 * border; const GLint dstWidthNB = dstWidth - 2 * border; @@ -694,13 +798,13 @@ make_3d_mipmap(const struct gl_texture_format *format, GLint border, for (row = 0; row < dstHeightNB; row++) { /* Average together two rows from first src image */ - do_row(format, srcWidthNB, srcImgARowA, srcImgARowB, + do_row(datatype, comps, srcWidthNB, srcImgARowA, srcImgARowB, srcWidthNB, tmpRowA); /* Average together two rows from second src image */ - do_row(format, srcWidthNB, srcImgBRowA, srcImgBRowB, + do_row(datatype, comps, srcWidthNB, srcImgBRowA, srcImgBRowB, srcWidthNB, tmpRowB); /* Average together the temp rows to make the final row */ - do_row(format, srcWidthNB, tmpRowA, tmpRowB, + do_row(datatype, comps, srcWidthNB, tmpRowA, tmpRowB, dstWidthNB, dstImgRow); /* advance to next rows */ srcImgARowA += bytesPerSrcRow + srcRowOffset; @@ -717,10 +821,10 @@ make_3d_mipmap(const struct gl_texture_format *format, GLint border, /* Luckily we can leverage the make_2d_mipmap() function here! */ if (border > 0) { /* do front border image */ - make_2d_mipmap(format, 1, srcWidth, srcHeight, srcPtr, + make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, srcPtr, dstWidth, dstHeight, dstPtr); /* do back border image */ - make_2d_mipmap(format, 1, srcWidth, srcHeight, + make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, srcPtr + bytesPerSrcImage * (srcDepth - 1), dstWidth, dstHeight, dstPtr + bytesPerDstImage * (dstDepth - 1)); @@ -768,28 +872,28 @@ make_3d_mipmap(const struct gl_texture_format *format, GLint border, /* do border along [img][row=0][col=0] */ src = srcPtr + (img * 2 + 1) * bytesPerSrcImage; dst = dstPtr + (img + 1) * bytesPerDstImage; - do_row(format, 1, src, src + srcImageOffset, 1, dst); + do_row(datatype, comps, 1, src, src + srcImageOffset, 1, dst); /* do border along [img][row=dstHeight-1][col=0] */ src = srcPtr + (img * 2 + 1) * bytesPerSrcImage + (srcHeight - 1) * bytesPerSrcRow; dst = dstPtr + (img + 1) * bytesPerDstImage + (dstHeight - 1) * bytesPerDstRow; - do_row(format, 1, src, src + srcImageOffset, 1, dst); + do_row(datatype, comps, 1, src, src + srcImageOffset, 1, dst); /* do border along [img][row=0][col=dstWidth-1] */ src = srcPtr + (img * 2 + 1) * bytesPerSrcImage + (srcWidth - 1) * bpt; dst = dstPtr + (img + 1) * bytesPerDstImage + (dstWidth - 1) * bpt; - do_row(format, 1, src, src + srcImageOffset, 1, dst); + do_row(datatype, comps, 1, src, src + srcImageOffset, 1, dst); /* do border along [img][row=dstHeight-1][col=dstWidth-1] */ src = srcPtr + (img * 2 + 1) * bytesPerSrcImage + (bytesPerSrcImage - bpt); dst = dstPtr + (img + 1) * bytesPerDstImage + (bytesPerDstImage - bpt); - do_row(format, 1, src, src + srcImageOffset, 1, dst); + do_row(datatype, comps, 1, src, src + srcImageOffset, 1, dst); } } } @@ -797,11 +901,11 @@ make_3d_mipmap(const struct gl_texture_format *format, GLint border, static void -make_1d_stack_mipmap(const struct gl_texture_format *format, GLint border, +make_1d_stack_mipmap(GLenum datatype, GLuint comps, GLint border, GLint srcWidth, const GLubyte *srcPtr, GLint dstWidth, GLint dstHeight, GLubyte *dstPtr) { - const GLint bpt = format->TexelBytes; + const GLint bpt = bytes_per_pixel(datatype, comps); const GLint srcWidthNB = srcWidth - 2 * border; /* sizes w/out border */ const GLint dstWidthNB = dstWidth - 2 * border; const GLint dstHeightNB = dstHeight - 2 * border; @@ -816,7 +920,7 @@ make_1d_stack_mipmap(const struct gl_texture_format *format, GLint border, dst = dstPtr + border * ((dstWidth + 1) * bpt); for (row = 0; row < dstHeightNB; row++) { - do_row(format, srcWidthNB, src, src, + do_row(datatype, comps, srcWidthNB, src, src, dstWidthNB, dst); src += srcRowStride; dst += dstRowStride; @@ -839,12 +943,12 @@ make_1d_stack_mipmap(const struct gl_texture_format *format, GLint border, * and \c make_2d_mipmap. */ static void -make_2d_stack_mipmap(const struct gl_texture_format *format, GLint border, +make_2d_stack_mipmap(GLenum datatype, GLuint comps, GLint border, GLint srcWidth, GLint srcHeight, const GLubyte *srcPtr, GLint dstWidth, GLint dstHeight, GLint dstDepth, GLubyte *dstPtr) { - const GLint bpt = format->TexelBytes; + const GLint bpt = bytes_per_pixel(datatype, comps); const GLint srcWidthNB = srcWidth - 2 * border; /* sizes w/out border */ const GLint dstWidthNB = dstWidth - 2 * border; const GLint dstHeightNB = dstHeight - 2 * border; @@ -866,7 +970,7 @@ make_2d_stack_mipmap(const struct gl_texture_format *format, GLint border, for (layer = 0; layer < dstDepthNB; layer++) { for (row = 0; row < dstHeightNB; row++) { - do_row(format, srcWidthNB, srcA, srcB, + do_row(datatype, comps, srcWidthNB, srcA, srcB, dstWidthNB, dst); srcA += 2 * srcRowStride; srcB += 2 * srcRowStride; @@ -888,12 +992,12 @@ make_2d_stack_mipmap(const struct gl_texture_format *format, GLint border, MEMCPY(dstPtr + (dstWidth * dstHeight - 1) * bpt, srcPtr + (srcWidth * srcHeight - 1) * bpt, bpt); /* lower border */ - do_row(format, srcWidthNB, + do_row(datatype, comps, srcWidthNB, srcPtr + bpt, srcPtr + bpt, dstWidthNB, dstPtr + bpt); /* upper border */ - do_row(format, srcWidthNB, + do_row(datatype, comps, srcWidthNB, srcPtr + (srcWidth * (srcHeight - 1) + 1) * bpt, srcPtr + (srcWidth * (srcHeight - 1) + 1) * bpt, dstWidthNB, @@ -911,11 +1015,11 @@ make_2d_stack_mipmap(const struct gl_texture_format *format, GLint border, else { /* average two src pixels each dest pixel */ for (row = 0; row < dstHeightNB; row += 2) { - do_row(format, 1, + do_row(datatype, comps, 1, srcPtr + (srcWidth * (row * 2 + 1)) * bpt, srcPtr + (srcWidth * (row * 2 + 2)) * bpt, 1, dstPtr + (dstWidth * row + 1) * bpt); - do_row(format, 1, + do_row(datatype, comps, 1, srcPtr + (srcWidth * (row * 2 + 1) + srcWidth - 1) * bpt, srcPtr + (srcWidth * (row * 2 + 2) + srcWidth - 1) * bpt, 1, dstPtr + (dstWidth * row + 1 + dstWidth - 1) * bpt); @@ -940,6 +1044,8 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, const GLubyte *srcData = NULL; GLubyte *dstData = NULL; GLint level, maxLevels; + GLenum datatype; + GLuint comps; ASSERT(texObj); /* XXX choose cube map face here??? */ @@ -1002,6 +1108,8 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, convertFormat = srcImage->TexFormat; } + mesa_format_to_type_and_comps(convertFormat, &datatype, &comps); + for (level = texObj->BaseLevel; level < texObj->MaxLevel && level < maxLevels - 1; level++) { /* generate image[level+1] from image[level] */ @@ -1118,7 +1226,7 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, */ switch (target) { case GL_TEXTURE_1D: - make_1d_mipmap(convertFormat, border, + make_1d_mipmap(datatype, comps, border, srcWidth, srcData, dstWidth, dstData); break; @@ -1129,22 +1237,22 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: - make_2d_mipmap(convertFormat, border, + make_2d_mipmap(datatype, comps, border, srcWidth, srcHeight, srcData, dstWidth, dstHeight, dstData); break; case GL_TEXTURE_3D: - make_3d_mipmap(convertFormat, border, + make_3d_mipmap(datatype, comps, border, srcWidth, srcHeight, srcDepth, srcData, dstWidth, dstHeight, dstDepth, dstData); break; case GL_TEXTURE_1D_ARRAY_EXT: - make_1d_stack_mipmap(convertFormat, border, + make_1d_stack_mipmap(datatype, comps, border, srcWidth, srcData, dstWidth, dstHeight, dstData); break; case GL_TEXTURE_2D_ARRAY_EXT: - make_2d_stack_mipmap(convertFormat, border, + make_2d_stack_mipmap(datatype, comps, border, srcWidth, srcHeight, srcData, dstWidth, dstHeight, dstDepth, dstData); break; -- cgit v1.2.3 From 9b3b230bf6b5833df65d706b68e887ae3bdcf950 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 16:38:28 -0700 Subject: checkpoint- consolidation in do_row() --- src/mesa/main/mipmap.c | 42 ++++++++++++++++-------------------------- 1 file changed, 16 insertions(+), 26 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index 013dc3752ec..c5f1c5bcbed 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -234,11 +234,11 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, assert(srcWidth == dstWidth || srcWidth == 2 * dstWidth); */ - if (datatype == CHAN_TYPE && comps == 4) { + if (datatype == GL_UNSIGNED_SHORT && comps == 4) { GLuint i, j, k; - const GLchan (*rowA)[4] = (const GLchan (*)[4]) srcRowA; - const GLchan (*rowB)[4] = (const GLchan (*)[4]) srcRowB; - GLchan (*dst)[4] = (GLchan (*)[4]) dstRow; + const GLushort (*rowA)[4] = (const GLushort (*)[4]) srcRowA; + const GLushort (*rowB)[4] = (const GLushort (*)[4]) srcRowB; + GLushort (*dst)[4] = (GLushort (*)[4]) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { dst[i][0] = (rowA[j][0] + rowA[k][0] + @@ -251,11 +251,11 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, rowB[j][3] + rowB[k][3]) / 4; } } - else if (datatype == CHAN_TYPE && comps == 3) { + else if (datatype == GL_UNSIGNED_SHORT && comps == 3) { GLuint i, j, k; - const GLchan (*rowA)[3] = (const GLchan (*)[3]) srcRowA; - const GLchan (*rowB)[3] = (const GLchan (*)[3]) srcRowB; - GLchan (*dst)[3] = (GLchan (*)[3]) dstRow; + const GLushort (*rowA)[3] = (const GLushort (*)[3]) srcRowA; + const GLushort (*rowB)[3] = (const GLushort (*)[3]) srcRowB; + GLushort (*dst)[3] = (GLushort (*)[3]) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { dst[i][0] = (rowA[j][0] + rowA[k][0] + @@ -266,21 +266,21 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, rowB[j][2] + rowB[k][2]) / 4; } } - else if (datatype == CHAN_TYPE && comps == 1) { + else if (datatype == GL_UNSIGNED_SHORT && comps == 1) { GLuint i, j, k; - const GLchan *rowA = (const GLchan *) srcRowA; - const GLchan *rowB = (const GLchan *) srcRowB; - GLchan *dst = (GLchan *) dstRow; + const GLushort *rowA = (const GLushort *) srcRowA; + const GLushort *rowB = (const GLushort *) srcRowB; + GLushort *dst = (GLushort *) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; } } - else if (datatype == CHAN_TYPE && comps == 2) { + else if (datatype == GL_UNSIGNED_SHORT && comps == 2) { GLuint i, j, k; - const GLchan (*rowA)[2] = (const GLchan (*)[2]) srcRowA; - const GLchan (*rowB)[2] = (const GLchan (*)[2]) srcRowB; - GLchan (*dst)[2] = (GLchan (*)[2]) dstRow; + const GLushort (*rowA)[2] = (const GLushort (*)[2]) srcRowA; + const GLushort (*rowB)[2] = (const GLushort (*)[2]) srcRowB; + GLushort (*dst)[2] = (GLushort (*)[2]) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { dst[i][0] = (rowA[j][0] + rowA[k][0] + @@ -299,16 +299,6 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, dst[i] = rowA[j] / 4 + rowA[k] / 4 + rowB[j] / 4 + rowB[k] / 4; } } - else if (datatype == GL_UNSIGNED_SHORT && comps == 1) { - GLuint i, j, k; - const GLushort *rowA = (const GLushort *) srcRowA; - const GLushort *rowB = (const GLushort *) srcRowB; - GLushort *dst = (GLushort *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; - } - } else if (datatype == GL_UNSIGNED_BYTE && comps == 4) { GLuint i, j, k; const GLubyte (*rowA)[4] = (const GLubyte (*)[4]) srcRowA; -- cgit v1.2.3 From 4bf4f6e029b85d9eb90f5649fc5635cf274e1fe1 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 16:40:39 -0700 Subject: re-indent do_row() --- src/mesa/main/mipmap.c | 594 ++++++++++++++++++++++++------------------------- 1 file changed, 288 insertions(+), 306 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index c5f1c5bcbed..22c7530e835 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -235,355 +235,337 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, */ if (datatype == GL_UNSIGNED_SHORT && comps == 4) { - GLuint i, j, k; - const GLushort (*rowA)[4] = (const GLushort (*)[4]) srcRowA; - const GLushort (*rowB)[4] = (const GLushort (*)[4]) srcRowB; - GLushort (*dst)[4] = (GLushort (*)[4]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) / 4; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) / 4; - dst[i][3] = (rowA[j][3] + rowA[k][3] + - rowB[j][3] + rowB[k][3]) / 4; - } + GLuint i, j, k; + const GLushort(*rowA)[4] = (const GLushort(*)[4]) srcRowA; + const GLushort(*rowB)[4] = (const GLushort(*)[4]) srcRowB; + GLushort(*dst)[4] = (GLushort(*)[4]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) / 4; + } } else if (datatype == GL_UNSIGNED_SHORT && comps == 3) { - GLuint i, j, k; - const GLushort (*rowA)[3] = (const GLushort (*)[3]) srcRowA; - const GLushort (*rowB)[3] = (const GLushort (*)[3]) srcRowB; - GLushort (*dst)[3] = (GLushort (*)[3]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) / 4; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) / 4; - } + GLuint i, j, k; + const GLushort(*rowA)[3] = (const GLushort(*)[3]) srcRowA; + const GLushort(*rowB)[3] = (const GLushort(*)[3]) srcRowB; + GLushort(*dst)[3] = (GLushort(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + } } else if (datatype == GL_UNSIGNED_SHORT && comps == 1) { - GLuint i, j, k; - const GLushort *rowA = (const GLushort *) srcRowA; - const GLushort *rowB = (const GLushort *) srcRowB; - GLushort *dst = (GLushort *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; - } + GLuint i, j, k; + const GLushort *rowA = (const GLushort *) srcRowA; + const GLushort *rowB = (const GLushort *) srcRowB; + GLushort *dst = (GLushort *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; + } } else if (datatype == GL_UNSIGNED_SHORT && comps == 2) { - GLuint i, j, k; - const GLushort (*rowA)[2] = (const GLushort (*)[2]) srcRowA; - const GLushort (*rowB)[2] = (const GLushort (*)[2]) srcRowB; - GLushort (*dst)[2] = (GLushort (*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) / 4; - } + GLuint i, j, k; + const GLushort(*rowA)[2] = (const GLushort(*)[2]) srcRowA; + const GLushort(*rowB)[2] = (const GLushort(*)[2]) srcRowB; + GLushort(*dst)[2] = (GLushort(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + } } else if (datatype == GL_UNSIGNED_INT && comps == 1) { - GLuint i, j, k; - const GLuint *rowA = (const GLuint *) srcRowA; - const GLuint *rowB = (const GLuint *) srcRowB; - GLfloat *dst = (GLfloat *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = rowA[j] / 4 + rowA[k] / 4 + rowB[j] / 4 + rowB[k] / 4; - } + GLuint i, j, k; + const GLuint *rowA = (const GLuint *) srcRowA; + const GLuint *rowB = (const GLuint *) srcRowB; + GLfloat *dst = (GLfloat *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i] = rowA[j] / 4 + rowA[k] / 4 + rowB[j] / 4 + rowB[k] / 4; + } } else if (datatype == GL_UNSIGNED_BYTE && comps == 4) { - GLuint i, j, k; - const GLubyte (*rowA)[4] = (const GLubyte (*)[4]) srcRowA; - const GLubyte (*rowB)[4] = (const GLubyte (*)[4]) srcRowB; - GLubyte (*dst)[4] = (GLubyte (*)[4]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) / 4; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) / 4; - dst[i][3] = (rowA[j][3] + rowA[k][3] + - rowB[j][3] + rowB[k][3]) / 4; - } + GLuint i, j, k; + const GLubyte(*rowA)[4] = (const GLubyte(*)[4]) srcRowA; + const GLubyte(*rowB)[4] = (const GLubyte(*)[4]) srcRowB; + GLubyte(*dst)[4] = (GLubyte(*)[4]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) / 4; + } } else if (datatype == GL_UNSIGNED_BYTE && comps == 3) { - GLuint i, j, k; - const GLubyte (*rowA)[3] = (const GLubyte (*)[3]) srcRowA; - const GLubyte (*rowB)[3] = (const GLubyte (*)[3]) srcRowB; - GLubyte (*dst)[3] = (GLubyte (*)[3]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) / 4; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) / 4; - } + GLuint i, j, k; + const GLubyte(*rowA)[3] = (const GLubyte(*)[3]) srcRowA; + const GLubyte(*rowB)[3] = (const GLubyte(*)[3]) srcRowB; + GLubyte(*dst)[3] = (GLubyte(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + } } else if (datatype == GL_UNSIGNED_SHORT_5_6_5 && comps == 3) { - GLuint i, j, k; - const GLushort *rowA = (const GLushort *) srcRowA; - const GLushort *rowB = (const GLushort *) srcRowB; - GLushort *dst = (GLushort *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - const GLint rowAr0 = rowA[j] & 0x1f; - const GLint rowAr1 = rowA[k] & 0x1f; - const GLint rowBr0 = rowB[j] & 0x1f; - const GLint rowBr1 = rowB[k] & 0x1f; - const GLint rowAg0 = (rowA[j] >> 5) & 0x3f; - const GLint rowAg1 = (rowA[k] >> 5) & 0x3f; - const GLint rowBg0 = (rowB[j] >> 5) & 0x3f; - const GLint rowBg1 = (rowB[k] >> 5) & 0x3f; - const GLint rowAb0 = (rowA[j] >> 11) & 0x1f; - const GLint rowAb1 = (rowA[k] >> 11) & 0x1f; - const GLint rowBb0 = (rowB[j] >> 11) & 0x1f; - const GLint rowBb1 = (rowB[k] >> 11) & 0x1f; - const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; - const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; - const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; - dst[i] = (blue << 11) | (green << 5) | red; - } + GLuint i, j, k; + const GLushort *rowA = (const GLushort *) srcRowA; + const GLushort *rowB = (const GLushort *) srcRowB; + GLushort *dst = (GLushort *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + const GLint rowAr0 = rowA[j] & 0x1f; + const GLint rowAr1 = rowA[k] & 0x1f; + const GLint rowBr0 = rowB[j] & 0x1f; + const GLint rowBr1 = rowB[k] & 0x1f; + const GLint rowAg0 = (rowA[j] >> 5) & 0x3f; + const GLint rowAg1 = (rowA[k] >> 5) & 0x3f; + const GLint rowBg0 = (rowB[j] >> 5) & 0x3f; + const GLint rowBg1 = (rowB[k] >> 5) & 0x3f; + const GLint rowAb0 = (rowA[j] >> 11) & 0x1f; + const GLint rowAb1 = (rowA[k] >> 11) & 0x1f; + const GLint rowBb0 = (rowB[j] >> 11) & 0x1f; + const GLint rowBb1 = (rowB[k] >> 11) & 0x1f; + const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; + const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; + const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; + dst[i] = (blue << 11) | (green << 5) | red; + } } else if (datatype == GL_UNSIGNED_SHORT_4_4_4_4 && comps == 4) { - GLuint i, j, k; - const GLushort *rowA = (const GLushort *) srcRowA; - const GLushort *rowB = (const GLushort *) srcRowB; - GLushort *dst = (GLushort *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - const GLint rowAr0 = rowA[j] & 0xf; - const GLint rowAr1 = rowA[k] & 0xf; - const GLint rowBr0 = rowB[j] & 0xf; - const GLint rowBr1 = rowB[k] & 0xf; - const GLint rowAg0 = (rowA[j] >> 4) & 0xf; - const GLint rowAg1 = (rowA[k] >> 4) & 0xf; - const GLint rowBg0 = (rowB[j] >> 4) & 0xf; - const GLint rowBg1 = (rowB[k] >> 4) & 0xf; - const GLint rowAb0 = (rowA[j] >> 8) & 0xf; - const GLint rowAb1 = (rowA[k] >> 8) & 0xf; - const GLint rowBb0 = (rowB[j] >> 8) & 0xf; - const GLint rowBb1 = (rowB[k] >> 8) & 0xf; - const GLint rowAa0 = (rowA[j] >> 12) & 0xf; - const GLint rowAa1 = (rowA[k] >> 12) & 0xf; - const GLint rowBa0 = (rowB[j] >> 12) & 0xf; - const GLint rowBa1 = (rowB[k] >> 12) & 0xf; - const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; - const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; - const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; - const GLint alpha = (rowAa0 + rowAa1 + rowBa0 + rowBa1) >> 2; - dst[i] = (alpha << 12) | (blue << 8) | (green << 4) | red; - } + GLuint i, j, k; + const GLushort *rowA = (const GLushort *) srcRowA; + const GLushort *rowB = (const GLushort *) srcRowB; + GLushort *dst = (GLushort *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + const GLint rowAr0 = rowA[j] & 0xf; + const GLint rowAr1 = rowA[k] & 0xf; + const GLint rowBr0 = rowB[j] & 0xf; + const GLint rowBr1 = rowB[k] & 0xf; + const GLint rowAg0 = (rowA[j] >> 4) & 0xf; + const GLint rowAg1 = (rowA[k] >> 4) & 0xf; + const GLint rowBg0 = (rowB[j] >> 4) & 0xf; + const GLint rowBg1 = (rowB[k] >> 4) & 0xf; + const GLint rowAb0 = (rowA[j] >> 8) & 0xf; + const GLint rowAb1 = (rowA[k] >> 8) & 0xf; + const GLint rowBb0 = (rowB[j] >> 8) & 0xf; + const GLint rowBb1 = (rowB[k] >> 8) & 0xf; + const GLint rowAa0 = (rowA[j] >> 12) & 0xf; + const GLint rowAa1 = (rowA[k] >> 12) & 0xf; + const GLint rowBa0 = (rowB[j] >> 12) & 0xf; + const GLint rowBa1 = (rowB[k] >> 12) & 0xf; + const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; + const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; + const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; + const GLint alpha = (rowAa0 + rowAa1 + rowBa0 + rowBa1) >> 2; + dst[i] = (alpha << 12) | (blue << 8) | (green << 4) | red; + } } else if (datatype == GL_UNSIGNED_SHORT_1_5_5_5_REV && comps == 4) { - GLuint i, j, k; - const GLushort *rowA = (const GLushort *) srcRowA; - const GLushort *rowB = (const GLushort *) srcRowB; - GLushort *dst = (GLushort *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - const GLint rowAr0 = rowA[j] & 0x1f; - const GLint rowAr1 = rowA[k] & 0x1f; - const GLint rowBr0 = rowB[j] & 0x1f; - const GLint rowBr1 = rowB[k] & 0xf; - const GLint rowAg0 = (rowA[j] >> 5) & 0x1f; - const GLint rowAg1 = (rowA[k] >> 5) & 0x1f; - const GLint rowBg0 = (rowB[j] >> 5) & 0x1f; - const GLint rowBg1 = (rowB[k] >> 5) & 0x1f; - const GLint rowAb0 = (rowA[j] >> 10) & 0x1f; - const GLint rowAb1 = (rowA[k] >> 10) & 0x1f; - const GLint rowBb0 = (rowB[j] >> 10) & 0x1f; - const GLint rowBb1 = (rowB[k] >> 10) & 0x1f; - const GLint rowAa0 = (rowA[j] >> 15) & 0x1; - const GLint rowAa1 = (rowA[k] >> 15) & 0x1; - const GLint rowBa0 = (rowB[j] >> 15) & 0x1; - const GLint rowBa1 = (rowB[k] >> 15) & 0x1; - const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; - const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; - const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; - const GLint alpha = (rowAa0 + rowAa1 + rowBa0 + rowBa1) >> 2; - dst[i] = (alpha << 15) | (blue << 10) | (green << 5) | red; - } + GLuint i, j, k; + const GLushort *rowA = (const GLushort *) srcRowA; + const GLushort *rowB = (const GLushort *) srcRowB; + GLushort *dst = (GLushort *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + const GLint rowAr0 = rowA[j] & 0x1f; + const GLint rowAr1 = rowA[k] & 0x1f; + const GLint rowBr0 = rowB[j] & 0x1f; + const GLint rowBr1 = rowB[k] & 0xf; + const GLint rowAg0 = (rowA[j] >> 5) & 0x1f; + const GLint rowAg1 = (rowA[k] >> 5) & 0x1f; + const GLint rowBg0 = (rowB[j] >> 5) & 0x1f; + const GLint rowBg1 = (rowB[k] >> 5) & 0x1f; + const GLint rowAb0 = (rowA[j] >> 10) & 0x1f; + const GLint rowAb1 = (rowA[k] >> 10) & 0x1f; + const GLint rowBb0 = (rowB[j] >> 10) & 0x1f; + const GLint rowBb1 = (rowB[k] >> 10) & 0x1f; + const GLint rowAa0 = (rowA[j] >> 15) & 0x1; + const GLint rowAa1 = (rowA[k] >> 15) & 0x1; + const GLint rowBa0 = (rowB[j] >> 15) & 0x1; + const GLint rowBa1 = (rowB[k] >> 15) & 0x1; + const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; + const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; + const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; + const GLint alpha = (rowAa0 + rowAa1 + rowBa0 + rowBa1) >> 2; + dst[i] = (alpha << 15) | (blue << 10) | (green << 5) | red; + } } else if (datatype == GL_UNSIGNED_BYTE && comps == 2) { - GLuint i, j, k; - const GLubyte (*rowA)[2] = (const GLubyte (*)[2]) srcRowA; - const GLubyte (*rowB)[2] = (const GLubyte (*)[2]) srcRowB; - GLubyte (*dst)[2] = (GLubyte (*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) >> 2; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) >> 2; - } + GLuint i, j, k; + const GLubyte(*rowA)[2] = (const GLubyte(*)[2]) srcRowA; + const GLubyte(*rowB)[2] = (const GLubyte(*)[2]) srcRowB; + GLubyte(*dst)[2] = (GLubyte(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) >> 2; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) >> 2; + } } else if (datatype == GL_UNSIGNED_BYTE_3_3_2 && comps == 3) { - GLuint i, j, k; - const GLubyte *rowA = (const GLubyte *) srcRowA; - const GLubyte *rowB = (const GLubyte *) srcRowB; - GLubyte *dst = (GLubyte *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - const GLint rowAr0 = rowA[j] & 0x3; - const GLint rowAr1 = rowA[k] & 0x3; - const GLint rowBr0 = rowB[j] & 0x3; - const GLint rowBr1 = rowB[k] & 0x3; - const GLint rowAg0 = (rowA[j] >> 2) & 0x7; - const GLint rowAg1 = (rowA[k] >> 2) & 0x7; - const GLint rowBg0 = (rowB[j] >> 2) & 0x7; - const GLint rowBg1 = (rowB[k] >> 2) & 0x7; - const GLint rowAb0 = (rowA[j] >> 5) & 0x7; - const GLint rowAb1 = (rowA[k] >> 5) & 0x7; - const GLint rowBb0 = (rowB[j] >> 5) & 0x7; - const GLint rowBb1 = (rowB[k] >> 5) & 0x7; - const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; - const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; - const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; - dst[i] = (blue << 5) | (green << 2) | red; - } + GLuint i, j, k; + const GLubyte *rowA = (const GLubyte *) srcRowA; + const GLubyte *rowB = (const GLubyte *) srcRowB; + GLubyte *dst = (GLubyte *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + const GLint rowAr0 = rowA[j] & 0x3; + const GLint rowAr1 = rowA[k] & 0x3; + const GLint rowBr0 = rowB[j] & 0x3; + const GLint rowBr1 = rowB[k] & 0x3; + const GLint rowAg0 = (rowA[j] >> 2) & 0x7; + const GLint rowAg1 = (rowA[k] >> 2) & 0x7; + const GLint rowBg0 = (rowB[j] >> 2) & 0x7; + const GLint rowBg1 = (rowB[k] >> 2) & 0x7; + const GLint rowAb0 = (rowA[j] >> 5) & 0x7; + const GLint rowAb1 = (rowA[k] >> 5) & 0x7; + const GLint rowBb0 = (rowB[j] >> 5) & 0x7; + const GLint rowBb1 = (rowB[k] >> 5) & 0x7; + const GLint red = (rowAr0 + rowAr1 + rowBr0 + rowBr1) >> 2; + const GLint green = (rowAg0 + rowAg1 + rowBg0 + rowBg1) >> 2; + const GLint blue = (rowAb0 + rowAb1 + rowBb0 + rowBb1) >> 2; + dst[i] = (blue << 5) | (green << 2) | red; + } } else if (datatype == GL_UNSIGNED_BYTE && comps == 1) { - GLuint i, j, k; - const GLubyte *rowA = (const GLubyte *) srcRowA; - const GLubyte *rowB = (const GLubyte *) srcRowB; - GLubyte *dst = (GLubyte *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) >> 2; - } + GLuint i, j, k; + const GLubyte *rowA = (const GLubyte *) srcRowA; + const GLubyte *rowB = (const GLubyte *) srcRowB; + GLubyte *dst = (GLubyte *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) >> 2; + } } else if (datatype == GL_FLOAT && comps == 4) { - GLuint i, j, k; - const GLfloat (*rowA)[4] = (const GLfloat (*)[4]) srcRowA; - const GLfloat (*rowB)[4] = (const GLfloat (*)[4]) srcRowB; - GLfloat (*dst)[4] = (GLfloat (*)[4]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) * 0.25F; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) * 0.25F; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) * 0.25F; - dst[i][3] = (rowA[j][3] + rowA[k][3] + - rowB[j][3] + rowB[k][3]) * 0.25F; - } + GLuint i, j, k; + const GLfloat(*rowA)[4] = (const GLfloat(*)[4]) srcRowA; + const GLfloat(*rowB)[4] = (const GLfloat(*)[4]) srcRowB; + GLfloat(*dst)[4] = (GLfloat(*)[4]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + + rowB[j][0] + rowB[k][0]) * 0.25F; + dst[i][1] = (rowA[j][1] + rowA[k][1] + + rowB[j][1] + rowB[k][1]) * 0.25F; + dst[i][2] = (rowA[j][2] + rowA[k][2] + + rowB[j][2] + rowB[k][2]) * 0.25F; + dst[i][3] = (rowA[j][3] + rowA[k][3] + + rowB[j][3] + rowB[k][3]) * 0.25F; + } } else if (datatype == GL_HALF_FLOAT_ARB && comps == 4) { - GLuint i, j, k, comp; - const GLhalfARB (*rowA)[4] = (const GLhalfARB (*)[4]) srcRowA; - const GLhalfARB (*rowB)[4] = (const GLhalfARB (*)[4]) srcRowB; - GLhalfARB (*dst)[4] = (GLhalfARB (*)[4]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - for (comp = 0; comp < 4; comp++) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j][comp]); - ak = _mesa_half_to_float(rowA[k][comp]); - bj = _mesa_half_to_float(rowB[j][comp]); - bk = _mesa_half_to_float(rowB[k][comp]); - dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } + GLuint i, j, k, comp; + const GLhalfARB(*rowA)[4] = (const GLhalfARB(*)[4]) srcRowA; + const GLhalfARB(*rowB)[4] = (const GLhalfARB(*)[4]) srcRowB; + GLhalfARB(*dst)[4] = (GLhalfARB(*)[4]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + for (comp = 0; comp < 4; comp++) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j][comp]); + ak = _mesa_half_to_float(rowA[k][comp]); + bj = _mesa_half_to_float(rowB[j][comp]); + bk = _mesa_half_to_float(rowB[k][comp]); + dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } + } } else if (datatype == GL_FLOAT && comps == 3) { - GLuint i, j, k; - const GLfloat (*rowA)[3] = (const GLfloat (*)[3]) srcRowA; - const GLfloat (*rowB)[3] = (const GLfloat (*)[3]) srcRowB; - GLfloat (*dst)[3] = (GLfloat (*)[3]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) * 0.25F; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) * 0.25F; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) * 0.25F; - } + GLuint i, j, k; + const GLfloat(*rowA)[3] = (const GLfloat(*)[3]) srcRowA; + const GLfloat(*rowB)[3] = (const GLfloat(*)[3]) srcRowB; + GLfloat(*dst)[3] = (GLfloat(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + + rowB[j][0] + rowB[k][0]) * 0.25F; + dst[i][1] = (rowA[j][1] + rowA[k][1] + + rowB[j][1] + rowB[k][1]) * 0.25F; + dst[i][2] = (rowA[j][2] + rowA[k][2] + + rowB[j][2] + rowB[k][2]) * 0.25F; + } } else if (datatype == GL_HALF_FLOAT_ARB && comps == 3) { - GLuint i, j, k, comp; - const GLhalfARB (*rowA)[3] = (const GLhalfARB (*)[3]) srcRowA; - const GLhalfARB (*rowB)[3] = (const GLhalfARB (*)[3]) srcRowB; - GLhalfARB (*dst)[3] = (GLhalfARB (*)[3]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - for (comp = 0; comp < 3; comp++) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j][comp]); - ak = _mesa_half_to_float(rowA[k][comp]); - bj = _mesa_half_to_float(rowB[j][comp]); - bk = _mesa_half_to_float(rowB[k][comp]); - dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } + GLuint i, j, k, comp; + const GLhalfARB(*rowA)[3] = (const GLhalfARB(*)[3]) srcRowA; + const GLhalfARB(*rowB)[3] = (const GLhalfARB(*)[3]) srcRowB; + GLhalfARB(*dst)[3] = (GLhalfARB(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + for (comp = 0; comp < 3; comp++) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j][comp]); + ak = _mesa_half_to_float(rowA[k][comp]); + bj = _mesa_half_to_float(rowB[j][comp]); + bk = _mesa_half_to_float(rowB[k][comp]); + dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } + } } else if (datatype == GL_FLOAT && comps == 2) { - GLuint i, j, k; - const GLfloat (*rowA)[2] = (const GLfloat (*)[2]) srcRowA; - const GLfloat (*rowB)[2] = (const GLfloat (*)[2]) srcRowB; - GLfloat (*dst)[2] = (GLfloat (*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) * 0.25F; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) * 0.25F; - } + GLuint i, j, k; + const GLfloat(*rowA)[2] = (const GLfloat(*)[2]) srcRowA; + const GLfloat(*rowB)[2] = (const GLfloat(*)[2]) srcRowB; + GLfloat(*dst)[2] = (GLfloat(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + + rowB[j][0] + rowB[k][0]) * 0.25F; + dst[i][1] = (rowA[j][1] + rowA[k][1] + + rowB[j][1] + rowB[k][1]) * 0.25F; + } } else if (datatype == GL_HALF_FLOAT_ARB && comps == 2) { - GLuint i, j, k, comp; - const GLhalfARB (*rowA)[2] = (const GLhalfARB (*)[2]) srcRowA; - const GLhalfARB (*rowB)[2] = (const GLhalfARB (*)[2]) srcRowB; - GLhalfARB (*dst)[2] = (GLhalfARB (*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - for (comp = 0; comp < 2; comp++) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j][comp]); - ak = _mesa_half_to_float(rowA[k][comp]); - bj = _mesa_half_to_float(rowB[j][comp]); - bk = _mesa_half_to_float(rowB[k][comp]); - dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } + GLuint i, j, k, comp; + const GLhalfARB(*rowA)[2] = (const GLhalfARB(*)[2]) srcRowA; + const GLhalfARB(*rowB)[2] = (const GLhalfARB(*)[2]) srcRowB; + GLhalfARB(*dst)[2] = (GLhalfARB(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + for (comp = 0; comp < 2; comp++) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j][comp]); + ak = _mesa_half_to_float(rowA[k][comp]); + bj = _mesa_half_to_float(rowB[j][comp]); + bk = _mesa_half_to_float(rowB[k][comp]); + dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } + } } else if (datatype == GL_FLOAT && comps == 1) { - GLuint i, j, k; - const GLfloat *rowA = (const GLfloat *) srcRowA; - const GLfloat *rowB = (const GLfloat *) srcRowB; - GLfloat *dst = (GLfloat *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) * 0.25F; - } + GLuint i, j, k; + const GLfloat *rowA = (const GLfloat *) srcRowA; + const GLfloat *rowB = (const GLfloat *) srcRowB; + GLfloat *dst = (GLfloat *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) * 0.25F; + } } else if (datatype == GL_HALF_FLOAT_ARB && comps == 1) { - GLuint i, j, k; - const GLhalfARB *rowA = (const GLhalfARB *) srcRowA; - const GLhalfARB *rowB = (const GLhalfARB *) srcRowB; - GLhalfARB *dst = (GLhalfARB *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j]); - ak = _mesa_half_to_float(rowA[k]); - bj = _mesa_half_to_float(rowB[j]); - bk = _mesa_half_to_float(rowB[k]); - dst[i] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } + GLuint i, j, k; + const GLhalfARB *rowA = (const GLhalfARB *) srcRowA; + const GLhalfARB *rowB = (const GLhalfARB *) srcRowB; + GLhalfARB *dst = (GLhalfARB *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j]); + ak = _mesa_half_to_float(rowA[k]); + bj = _mesa_half_to_float(rowB[j]); + bk = _mesa_half_to_float(rowB[k]); + dst[i] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); + } } else { _mesa_problem(NULL, "bad format in do_row()"); -- cgit v1.2.3 From 708a022959104303be554b7c0144dd75fe8d7b81 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 16:42:50 -0700 Subject: reorder cases in do_row() --- src/mesa/main/mipmap.c | 341 +++++++++++++++++++++++++------------------------ 1 file changed, 173 insertions(+), 168 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index 22c7530e835..db8ab65401f 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -234,7 +234,54 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, assert(srcWidth == dstWidth || srcWidth == 2 * dstWidth); */ - if (datatype == GL_UNSIGNED_SHORT && comps == 4) { + if (datatype == GL_UNSIGNED_BYTE && comps == 4) { + GLuint i, j, k; + const GLubyte(*rowA)[4] = (const GLubyte(*)[4]) srcRowA; + const GLubyte(*rowB)[4] = (const GLubyte(*)[4]) srcRowB; + GLubyte(*dst)[4] = (GLubyte(*)[4]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) / 4; + } + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 3) { + GLuint i, j, k; + const GLubyte(*rowA)[3] = (const GLubyte(*)[3]) srcRowA; + const GLubyte(*rowB)[3] = (const GLubyte(*)[3]) srcRowB; + GLubyte(*dst)[3] = (GLubyte(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + } + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 2) { + GLuint i, j, k; + const GLubyte(*rowA)[2] = (const GLubyte(*)[2]) srcRowA; + const GLubyte(*rowB)[2] = (const GLubyte(*)[2]) srcRowB; + GLubyte(*dst)[2] = (GLubyte(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) >> 2; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) >> 2; + } + } + else if (datatype == GL_UNSIGNED_BYTE && comps == 1) { + GLuint i, j, k; + const GLubyte *rowA = (const GLubyte *) srcRowA; + const GLubyte *rowB = (const GLubyte *) srcRowB; + GLubyte *dst = (GLubyte *) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) >> 2; + } + } + + else if (datatype == GL_UNSIGNED_SHORT && comps == 4) { GLuint i, j, k; const GLushort(*rowA)[4] = (const GLushort(*)[4]) srcRowA; const GLushort(*rowB)[4] = (const GLushort(*)[4]) srcRowB; @@ -259,6 +306,17 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; } } + else if (datatype == GL_UNSIGNED_SHORT && comps == 2) { + GLuint i, j, k; + const GLushort(*rowA)[2] = (const GLushort(*)[2]) srcRowA; + const GLushort(*rowB)[2] = (const GLushort(*)[2]) srcRowB; + GLushort(*dst)[2] = (GLushort(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; + dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + } + } else if (datatype == GL_UNSIGNED_SHORT && comps == 1) { GLuint i, j, k; const GLushort *rowA = (const GLushort *) srcRowA; @@ -269,52 +327,141 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) / 4; } } - else if (datatype == GL_UNSIGNED_SHORT && comps == 2) { + + else if (datatype == GL_FLOAT && comps == 4) { GLuint i, j, k; - const GLushort(*rowA)[2] = (const GLushort(*)[2]) srcRowA; - const GLushort(*rowB)[2] = (const GLushort(*)[2]) srcRowB; - GLushort(*dst)[2] = (GLushort(*)[2]) dstRow; + const GLfloat(*rowA)[4] = (const GLfloat(*)[4]) srcRowA; + const GLfloat(*rowB)[4] = (const GLfloat(*)[4]) srcRowB; + GLfloat(*dst)[4] = (GLfloat(*)[4]) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; + dst[i][0] = (rowA[j][0] + rowA[k][0] + + rowB[j][0] + rowB[k][0]) * 0.25F; + dst[i][1] = (rowA[j][1] + rowA[k][1] + + rowB[j][1] + rowB[k][1]) * 0.25F; + dst[i][2] = (rowA[j][2] + rowA[k][2] + + rowB[j][2] + rowB[k][2]) * 0.25F; + dst[i][3] = (rowA[j][3] + rowA[k][3] + + rowB[j][3] + rowB[k][3]) * 0.25F; } } - else if (datatype == GL_UNSIGNED_INT && comps == 1) { + else if (datatype == GL_FLOAT && comps == 3) { GLuint i, j, k; - const GLuint *rowA = (const GLuint *) srcRowA; - const GLuint *rowB = (const GLuint *) srcRowB; + const GLfloat(*rowA)[3] = (const GLfloat(*)[3]) srcRowA; + const GLfloat(*rowB)[3] = (const GLfloat(*)[3]) srcRowB; + GLfloat(*dst)[3] = (GLfloat(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + + rowB[j][0] + rowB[k][0]) * 0.25F; + dst[i][1] = (rowA[j][1] + rowA[k][1] + + rowB[j][1] + rowB[k][1]) * 0.25F; + dst[i][2] = (rowA[j][2] + rowA[k][2] + + rowB[j][2] + rowB[k][2]) * 0.25F; + } + } + else if (datatype == GL_FLOAT && comps == 2) { + GLuint i, j, k; + const GLfloat(*rowA)[2] = (const GLfloat(*)[2]) srcRowA; + const GLfloat(*rowB)[2] = (const GLfloat(*)[2]) srcRowB; + GLfloat(*dst)[2] = (GLfloat(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + dst[i][0] = (rowA[j][0] + rowA[k][0] + + rowB[j][0] + rowB[k][0]) * 0.25F; + dst[i][1] = (rowA[j][1] + rowA[k][1] + + rowB[j][1] + rowB[k][1]) * 0.25F; + } + } + else if (datatype == GL_FLOAT && comps == 1) { + GLuint i, j, k; + const GLfloat *rowA = (const GLfloat *) srcRowA; + const GLfloat *rowB = (const GLfloat *) srcRowB; GLfloat *dst = (GLfloat *) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { - dst[i] = rowA[j] / 4 + rowA[k] / 4 + rowB[j] / 4 + rowB[k] / 4; + dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) * 0.25F; + } + } + + else if (datatype == GL_HALF_FLOAT_ARB && comps == 4) { + GLuint i, j, k, comp; + const GLhalfARB(*rowA)[4] = (const GLhalfARB(*)[4]) srcRowA; + const GLhalfARB(*rowB)[4] = (const GLhalfARB(*)[4]) srcRowB; + GLhalfARB(*dst)[4] = (GLhalfARB(*)[4]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + for (comp = 0; comp < 4; comp++) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j][comp]); + ak = _mesa_half_to_float(rowA[k][comp]); + bj = _mesa_half_to_float(rowB[j][comp]); + bk = _mesa_half_to_float(rowB[k][comp]); + dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); + } + } + } + else if (datatype == GL_HALF_FLOAT_ARB && comps == 3) { + GLuint i, j, k, comp; + const GLhalfARB(*rowA)[3] = (const GLhalfARB(*)[3]) srcRowA; + const GLhalfARB(*rowB)[3] = (const GLhalfARB(*)[3]) srcRowB; + GLhalfARB(*dst)[3] = (GLhalfARB(*)[3]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + for (comp = 0; comp < 3; comp++) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j][comp]); + ak = _mesa_half_to_float(rowA[k][comp]); + bj = _mesa_half_to_float(rowB[j][comp]); + bk = _mesa_half_to_float(rowB[k][comp]); + dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); + } + } + } + else if (datatype == GL_HALF_FLOAT_ARB && comps == 2) { + GLuint i, j, k, comp; + const GLhalfARB(*rowA)[2] = (const GLhalfARB(*)[2]) srcRowA; + const GLhalfARB(*rowB)[2] = (const GLhalfARB(*)[2]) srcRowB; + GLhalfARB(*dst)[2] = (GLhalfARB(*)[2]) dstRow; + for (i = j = 0, k = k0; i < (GLuint) dstWidth; + i++, j += colStride, k += colStride) { + for (comp = 0; comp < 2; comp++) { + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j][comp]); + ak = _mesa_half_to_float(rowA[k][comp]); + bj = _mesa_half_to_float(rowB[j][comp]); + bk = _mesa_half_to_float(rowB[k][comp]); + dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); + } } } - else if (datatype == GL_UNSIGNED_BYTE && comps == 4) { + else if (datatype == GL_HALF_FLOAT_ARB && comps == 1) { GLuint i, j, k; - const GLubyte(*rowA)[4] = (const GLubyte(*)[4]) srcRowA; - const GLubyte(*rowB)[4] = (const GLubyte(*)[4]) srcRowB; - GLubyte(*dst)[4] = (GLubyte(*)[4]) dstRow; + const GLhalfARB *rowA = (const GLhalfARB *) srcRowA; + const GLhalfARB *rowB = (const GLhalfARB *) srcRowB; + GLhalfARB *dst = (GLhalfARB *) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; - dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; - dst[i][3] = (rowA[j][3] + rowA[k][3] + rowB[j][3] + rowB[k][3]) / 4; + GLfloat aj, ak, bj, bk; + aj = _mesa_half_to_float(rowA[j]); + ak = _mesa_half_to_float(rowA[k]); + bj = _mesa_half_to_float(rowB[j]); + bk = _mesa_half_to_float(rowB[k]); + dst[i] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); } } - else if (datatype == GL_UNSIGNED_BYTE && comps == 3) { + + else if (datatype == GL_UNSIGNED_INT && comps == 1) { GLuint i, j, k; - const GLubyte(*rowA)[3] = (const GLubyte(*)[3]) srcRowA; - const GLubyte(*rowB)[3] = (const GLubyte(*)[3]) srcRowB; - GLubyte(*dst)[3] = (GLubyte(*)[3]) dstRow; + const GLuint *rowA = (const GLuint *) srcRowA; + const GLuint *rowB = (const GLuint *) srcRowB; + GLfloat *dst = (GLfloat *) dstRow; for (i = j = 0, k = k0; i < (GLuint) dstWidth; i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) / 4; - dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) / 4; - dst[i][2] = (rowA[j][2] + rowA[k][2] + rowB[j][2] + rowB[k][2]) / 4; + dst[i] = rowA[j] / 4 + rowA[k] / 4 + rowB[j] / 4 + rowB[k] / 4; } } + else if (datatype == GL_UNSIGNED_SHORT_5_6_5 && comps == 3) { GLuint i, j, k; const GLushort *rowA = (const GLushort *) srcRowA; @@ -400,17 +547,6 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, dst[i] = (alpha << 15) | (blue << 10) | (green << 5) | red; } } - else if (datatype == GL_UNSIGNED_BYTE && comps == 2) { - GLuint i, j, k; - const GLubyte(*rowA)[2] = (const GLubyte(*)[2]) srcRowA; - const GLubyte(*rowB)[2] = (const GLubyte(*)[2]) srcRowB; - GLubyte(*dst)[2] = (GLubyte(*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + rowB[j][0] + rowB[k][0]) >> 2; - dst[i][1] = (rowA[j][1] + rowA[k][1] + rowB[j][1] + rowB[k][1]) >> 2; - } - } else if (datatype == GL_UNSIGNED_BYTE_3_3_2 && comps == 3) { GLuint i, j, k; const GLubyte *rowA = (const GLubyte *) srcRowA; @@ -436,137 +572,6 @@ do_row(GLenum datatype, GLuint comps, GLint srcWidth, dst[i] = (blue << 5) | (green << 2) | red; } } - else if (datatype == GL_UNSIGNED_BYTE && comps == 1) { - GLuint i, j, k; - const GLubyte *rowA = (const GLubyte *) srcRowA; - const GLubyte *rowB = (const GLubyte *) srcRowB; - GLubyte *dst = (GLubyte *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) >> 2; - } - } - else if (datatype == GL_FLOAT && comps == 4) { - GLuint i, j, k; - const GLfloat(*rowA)[4] = (const GLfloat(*)[4]) srcRowA; - const GLfloat(*rowB)[4] = (const GLfloat(*)[4]) srcRowB; - GLfloat(*dst)[4] = (GLfloat(*)[4]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) * 0.25F; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) * 0.25F; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) * 0.25F; - dst[i][3] = (rowA[j][3] + rowA[k][3] + - rowB[j][3] + rowB[k][3]) * 0.25F; - } - } - else if (datatype == GL_HALF_FLOAT_ARB && comps == 4) { - GLuint i, j, k, comp; - const GLhalfARB(*rowA)[4] = (const GLhalfARB(*)[4]) srcRowA; - const GLhalfARB(*rowB)[4] = (const GLhalfARB(*)[4]) srcRowB; - GLhalfARB(*dst)[4] = (GLhalfARB(*)[4]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - for (comp = 0; comp < 4; comp++) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j][comp]); - ak = _mesa_half_to_float(rowA[k][comp]); - bj = _mesa_half_to_float(rowB[j][comp]); - bk = _mesa_half_to_float(rowB[k][comp]); - dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } - } - } - else if (datatype == GL_FLOAT && comps == 3) { - GLuint i, j, k; - const GLfloat(*rowA)[3] = (const GLfloat(*)[3]) srcRowA; - const GLfloat(*rowB)[3] = (const GLfloat(*)[3]) srcRowB; - GLfloat(*dst)[3] = (GLfloat(*)[3]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) * 0.25F; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) * 0.25F; - dst[i][2] = (rowA[j][2] + rowA[k][2] + - rowB[j][2] + rowB[k][2]) * 0.25F; - } - } - else if (datatype == GL_HALF_FLOAT_ARB && comps == 3) { - GLuint i, j, k, comp; - const GLhalfARB(*rowA)[3] = (const GLhalfARB(*)[3]) srcRowA; - const GLhalfARB(*rowB)[3] = (const GLhalfARB(*)[3]) srcRowB; - GLhalfARB(*dst)[3] = (GLhalfARB(*)[3]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - for (comp = 0; comp < 3; comp++) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j][comp]); - ak = _mesa_half_to_float(rowA[k][comp]); - bj = _mesa_half_to_float(rowB[j][comp]); - bk = _mesa_half_to_float(rowB[k][comp]); - dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } - } - } - else if (datatype == GL_FLOAT && comps == 2) { - GLuint i, j, k; - const GLfloat(*rowA)[2] = (const GLfloat(*)[2]) srcRowA; - const GLfloat(*rowB)[2] = (const GLfloat(*)[2]) srcRowB; - GLfloat(*dst)[2] = (GLfloat(*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i][0] = (rowA[j][0] + rowA[k][0] + - rowB[j][0] + rowB[k][0]) * 0.25F; - dst[i][1] = (rowA[j][1] + rowA[k][1] + - rowB[j][1] + rowB[k][1]) * 0.25F; - } - } - else if (datatype == GL_HALF_FLOAT_ARB && comps == 2) { - GLuint i, j, k, comp; - const GLhalfARB(*rowA)[2] = (const GLhalfARB(*)[2]) srcRowA; - const GLhalfARB(*rowB)[2] = (const GLhalfARB(*)[2]) srcRowB; - GLhalfARB(*dst)[2] = (GLhalfARB(*)[2]) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - for (comp = 0; comp < 2; comp++) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j][comp]); - ak = _mesa_half_to_float(rowA[k][comp]); - bj = _mesa_half_to_float(rowB[j][comp]); - bk = _mesa_half_to_float(rowB[k][comp]); - dst[i][comp] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } - } - } - else if (datatype == GL_FLOAT && comps == 1) { - GLuint i, j, k; - const GLfloat *rowA = (const GLfloat *) srcRowA; - const GLfloat *rowB = (const GLfloat *) srcRowB; - GLfloat *dst = (GLfloat *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - dst[i] = (rowA[j] + rowA[k] + rowB[j] + rowB[k]) * 0.25F; - } - } - else if (datatype == GL_HALF_FLOAT_ARB && comps == 1) { - GLuint i, j, k; - const GLhalfARB *rowA = (const GLhalfARB *) srcRowA; - const GLhalfARB *rowB = (const GLhalfARB *) srcRowB; - GLhalfARB *dst = (GLhalfARB *) dstRow; - for (i = j = 0, k = k0; i < (GLuint) dstWidth; - i++, j += colStride, k += colStride) { - GLfloat aj, ak, bj, bk; - aj = _mesa_half_to_float(rowA[j]); - ak = _mesa_half_to_float(rowA[k]); - bj = _mesa_half_to_float(rowB[j]); - bk = _mesa_half_to_float(rowB[k]); - dst[i] = _mesa_float_to_half((aj + ak + bj + bk) * 0.25F); - } - } else { _mesa_problem(NULL, "bad format in do_row()"); } -- cgit v1.2.3 From a217d0c7d0d52871dbf4196f83fecefa87ef8a24 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 16:46:12 -0700 Subject: move _mesa_format_to_type_and_comps() to texformat.c --- src/mesa/main/mipmap.c | 168 +-------------------------------------------- src/mesa/main/texformat.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++ src/mesa/main/texformat.h | 6 ++ 3 files changed, 178 insertions(+), 167 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index db8ab65401f..981da5dd89e 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -45,172 +45,6 @@ bytes_per_pixel(GLenum datatype, GLuint comps) } -static void -mesa_format_to_type_and_comps(const struct gl_texture_format *format, - GLenum *datatype, GLuint *comps) -{ - switch (format->MesaFormat) { - case MESA_FORMAT_RGBA8888: - case MESA_FORMAT_RGBA8888_REV: - case MESA_FORMAT_ARGB8888: - case MESA_FORMAT_ARGB8888_REV: - *datatype = CHAN_TYPE; - *comps = 4; - return; - case MESA_FORMAT_RGB888: - case MESA_FORMAT_BGR888: - *datatype = GL_UNSIGNED_BYTE; - *comps = 3; - return; - case MESA_FORMAT_RGB565: - case MESA_FORMAT_RGB565_REV: - *datatype = GL_UNSIGNED_SHORT_5_6_5; - *comps = 3; - return; - - case MESA_FORMAT_ARGB4444: - case MESA_FORMAT_ARGB4444_REV: - *datatype = GL_UNSIGNED_SHORT_4_4_4_4; - *comps = 4; - return; - - case MESA_FORMAT_ARGB1555: - case MESA_FORMAT_ARGB1555_REV: - *datatype = GL_UNSIGNED_SHORT_1_5_5_5_REV; - *comps = 3; - return; - - case MESA_FORMAT_AL88: - case MESA_FORMAT_AL88_REV: - *datatype = GL_UNSIGNED_BYTE; - *comps = 2; - return; - case MESA_FORMAT_RGB332: - *datatype = GL_UNSIGNED_BYTE_3_3_2; - *comps = 3; - return; - - case MESA_FORMAT_A8: - case MESA_FORMAT_L8: - case MESA_FORMAT_I8: - case MESA_FORMAT_CI8: - *datatype = GL_UNSIGNED_BYTE; - *comps = 1; - return; - - case MESA_FORMAT_YCBCR: - case MESA_FORMAT_YCBCR_REV: - *datatype = GL_UNSIGNED_SHORT; - *comps = 2; - return; - - case MESA_FORMAT_Z24_S8: - *datatype = GL_UNSIGNED_INT; - *comps = 1; /* XXX OK? */ - return; - - case MESA_FORMAT_Z16: - *datatype = GL_UNSIGNED_SHORT; - *comps = 1; - return; - - case MESA_FORMAT_Z32: - *datatype = GL_UNSIGNED_INT; - *comps = 1; - return; - - case MESA_FORMAT_SRGB8: - *datatype = GL_UNSIGNED_BYTE; - *comps = 3; - return; - case MESA_FORMAT_SRGBA8: - *datatype = GL_UNSIGNED_BYTE; - *comps = 4; - return; - case MESA_FORMAT_SL8: - *datatype = GL_UNSIGNED_BYTE; - *comps = 1; - return; - case MESA_FORMAT_SLA8: - *datatype = GL_UNSIGNED_BYTE; - *comps = 2; - return; - - case MESA_FORMAT_RGB_FXT1: - case MESA_FORMAT_RGBA_FXT1: - case MESA_FORMAT_RGB_DXT1: - case MESA_FORMAT_RGBA_DXT1: - case MESA_FORMAT_RGBA_DXT3: - case MESA_FORMAT_RGBA_DXT5: - /* XXX generate error instead? */ - *datatype = GL_UNSIGNED_BYTE; - *comps = 0; - return; - - case MESA_FORMAT_RGBA: - *datatype = CHAN_TYPE; - *comps = 4; - return; - case MESA_FORMAT_RGB: - *datatype = CHAN_TYPE; - *comps = 3; - return; - case MESA_FORMAT_LUMINANCE_ALPHA: - *datatype = CHAN_TYPE; - *comps = 2; - return; - case MESA_FORMAT_ALPHA: - case MESA_FORMAT_LUMINANCE: - case MESA_FORMAT_INTENSITY: - *datatype = CHAN_TYPE; - *comps = 1; - return; - - case MESA_FORMAT_RGBA_FLOAT32: - *datatype = GL_FLOAT; - *comps = 4; - return; - case MESA_FORMAT_RGBA_FLOAT16: - *datatype = GL_HALF_FLOAT_ARB; - *comps = 4; - return; - case MESA_FORMAT_RGB_FLOAT32: - *datatype = GL_FLOAT; - *comps = 3; - return; - case MESA_FORMAT_RGB_FLOAT16: - *datatype = GL_HALF_FLOAT_ARB; - *comps = 3; - return; - case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32: - *datatype = GL_FLOAT; - *comps = 2; - return; - case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16: - *datatype = GL_HALF_FLOAT_ARB; - *comps = 2; - return; - case MESA_FORMAT_ALPHA_FLOAT32: - case MESA_FORMAT_LUMINANCE_FLOAT32: - case MESA_FORMAT_INTENSITY_FLOAT32: - *datatype = GL_FLOAT; - *comps = 1; - return; - case MESA_FORMAT_ALPHA_FLOAT16: - case MESA_FORMAT_LUMINANCE_FLOAT16: - case MESA_FORMAT_INTENSITY_FLOAT16: - *datatype = GL_HALF_FLOAT_ARB; - *comps = 1; - return; - - default: - _mesa_problem(NULL, "bad texture format in mesa_format_to_type_and_comps"); - *datatype = 0; - *comps = 1; - } -} - - /** * Average together two rows of a source image to produce a single new * row in the dest image. It's legal for the two source rows to point @@ -1085,7 +919,7 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, convertFormat = srcImage->TexFormat; } - mesa_format_to_type_and_comps(convertFormat, &datatype, &comps); + _mesa_format_to_type_and_comps(convertFormat, &datatype, &comps); for (level = texObj->BaseLevel; level < texObj->MaxLevel && level < maxLevels - 1; level++) { diff --git a/src/mesa/main/texformat.c b/src/mesa/main/texformat.c index acc268e622d..88fbd8f07c2 100644 --- a/src/mesa/main/texformat.c +++ b/src/mesa/main/texformat.c @@ -1569,3 +1569,174 @@ _mesa_choose_tex_format( GLcontext *ctx, GLint internalFormat, _mesa_problem(ctx, "unexpected format in _mesa_choose_tex_format()"); return NULL; } + + + +/** + * Return datatype and number of components per texel for the + * given gl_texture_format. + */ +void +_mesa_format_to_type_and_comps(const struct gl_texture_format *format, + GLenum *datatype, GLuint *comps) +{ + switch (format->MesaFormat) { + case MESA_FORMAT_RGBA8888: + case MESA_FORMAT_RGBA8888_REV: + case MESA_FORMAT_ARGB8888: + case MESA_FORMAT_ARGB8888_REV: + *datatype = CHAN_TYPE; + *comps = 4; + return; + case MESA_FORMAT_RGB888: + case MESA_FORMAT_BGR888: + *datatype = GL_UNSIGNED_BYTE; + *comps = 3; + return; + case MESA_FORMAT_RGB565: + case MESA_FORMAT_RGB565_REV: + *datatype = GL_UNSIGNED_SHORT_5_6_5; + *comps = 3; + return; + + case MESA_FORMAT_ARGB4444: + case MESA_FORMAT_ARGB4444_REV: + *datatype = GL_UNSIGNED_SHORT_4_4_4_4; + *comps = 4; + return; + + case MESA_FORMAT_ARGB1555: + case MESA_FORMAT_ARGB1555_REV: + *datatype = GL_UNSIGNED_SHORT_1_5_5_5_REV; + *comps = 3; + return; + + case MESA_FORMAT_AL88: + case MESA_FORMAT_AL88_REV: + *datatype = GL_UNSIGNED_BYTE; + *comps = 2; + return; + case MESA_FORMAT_RGB332: + *datatype = GL_UNSIGNED_BYTE_3_3_2; + *comps = 3; + return; + + case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: + case MESA_FORMAT_CI8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 1; + return; + + case MESA_FORMAT_YCBCR: + case MESA_FORMAT_YCBCR_REV: + *datatype = GL_UNSIGNED_SHORT; + *comps = 2; + return; + + case MESA_FORMAT_Z24_S8: + *datatype = GL_UNSIGNED_INT; + *comps = 1; /* XXX OK? */ + return; + + case MESA_FORMAT_Z16: + *datatype = GL_UNSIGNED_SHORT; + *comps = 1; + return; + + case MESA_FORMAT_Z32: + *datatype = GL_UNSIGNED_INT; + *comps = 1; + return; + + case MESA_FORMAT_SRGB8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 3; + return; + case MESA_FORMAT_SRGBA8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 4; + return; + case MESA_FORMAT_SL8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 1; + return; + case MESA_FORMAT_SLA8: + *datatype = GL_UNSIGNED_BYTE; + *comps = 2; + return; + + case MESA_FORMAT_RGB_FXT1: + case MESA_FORMAT_RGBA_FXT1: + case MESA_FORMAT_RGB_DXT1: + case MESA_FORMAT_RGBA_DXT1: + case MESA_FORMAT_RGBA_DXT3: + case MESA_FORMAT_RGBA_DXT5: + /* XXX generate error instead? */ + *datatype = GL_UNSIGNED_BYTE; + *comps = 0; + return; + + case MESA_FORMAT_RGBA: + *datatype = CHAN_TYPE; + *comps = 4; + return; + case MESA_FORMAT_RGB: + *datatype = CHAN_TYPE; + *comps = 3; + return; + case MESA_FORMAT_LUMINANCE_ALPHA: + *datatype = CHAN_TYPE; + *comps = 2; + return; + case MESA_FORMAT_ALPHA: + case MESA_FORMAT_LUMINANCE: + case MESA_FORMAT_INTENSITY: + *datatype = CHAN_TYPE; + *comps = 1; + return; + + case MESA_FORMAT_RGBA_FLOAT32: + *datatype = GL_FLOAT; + *comps = 4; + return; + case MESA_FORMAT_RGBA_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 4; + return; + case MESA_FORMAT_RGB_FLOAT32: + *datatype = GL_FLOAT; + *comps = 3; + return; + case MESA_FORMAT_RGB_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 3; + return; + case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32: + *datatype = GL_FLOAT; + *comps = 2; + return; + case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 2; + return; + case MESA_FORMAT_ALPHA_FLOAT32: + case MESA_FORMAT_LUMINANCE_FLOAT32: + case MESA_FORMAT_INTENSITY_FLOAT32: + *datatype = GL_FLOAT; + *comps = 1; + return; + case MESA_FORMAT_ALPHA_FLOAT16: + case MESA_FORMAT_LUMINANCE_FLOAT16: + case MESA_FORMAT_INTENSITY_FLOAT16: + *datatype = GL_HALF_FLOAT_ARB; + *comps = 1; + return; + + default: + _mesa_problem(NULL, "bad format in _mesa_format_to_type_and_comps"); + *datatype = 0; + *comps = 1; + } +} diff --git a/src/mesa/main/texformat.h b/src/mesa/main/texformat.h index 55851db7016..48f0fe99f27 100644 --- a/src/mesa/main/texformat.h +++ b/src/mesa/main/texformat.h @@ -239,4 +239,10 @@ extern const struct gl_texture_format * _mesa_choose_tex_format( GLcontext *ctx, GLint internalFormat, GLenum format, GLenum type ); + +extern void +_mesa_format_to_type_and_comps(const struct gl_texture_format *format, + GLenum *datatype, GLuint *comps); + + #endif -- cgit v1.2.3 From b81ef1c429bbd34536f30c1522f1915996476078 Mon Sep 17 00:00:00 2001 From: Brian Date: Fri, 8 Feb 2008 16:56:21 -0700 Subject: refactor code, export _mesa_generate_mipmap_level() --- src/mesa/main/mipmap.c | 98 ++++++++++++++++++++++++++++---------------------- src/mesa/main/mipmap.h | 11 ++++++ 2 files changed, 67 insertions(+), 42 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index 981da5dd89e..ed7795aef93 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -841,6 +841,59 @@ make_2d_stack_mipmap(GLenum datatype, GLuint comps, GLint border, } +/** + * Down-sample a texture image to produce the next lower mipmap level. + */ +void +_mesa_generate_mipmap_level(GLenum target, + GLenum datatype, GLuint comps, + GLint border, + GLint srcWidth, GLint srcHeight, GLint srcDepth, + const GLubyte *srcData, + GLint dstWidth, GLint dstHeight, GLint dstDepth, + GLubyte *dstData) +{ + switch (target) { + case GL_TEXTURE_1D: + make_1d_mipmap(datatype, comps, border, + srcWidth, srcData, + dstWidth, dstData); + break; + case GL_TEXTURE_2D: + case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: + make_2d_mipmap(datatype, comps, border, + srcWidth, srcHeight, srcData, + dstWidth, dstHeight, dstData); + break; + case GL_TEXTURE_3D: + make_3d_mipmap(datatype, comps, border, + srcWidth, srcHeight, srcDepth, srcData, + dstWidth, dstHeight, dstDepth, dstData); + break; + case GL_TEXTURE_1D_ARRAY_EXT: + make_1d_stack_mipmap(datatype, comps, border, + srcWidth, srcData, + dstWidth, dstHeight, dstData); + break; + case GL_TEXTURE_2D_ARRAY_EXT: + make_2d_stack_mipmap(datatype, comps, border, + srcWidth, srcHeight, srcData, + dstWidth, dstHeight, dstDepth, dstData); + break; + case GL_TEXTURE_RECTANGLE_NV: + /* no mipmaps, do nothing */ + break; + default: + _mesa_problem(NULL, "bad target in _mesa_generate_mipmap_level"); + } +} + + /** * For GL_SGIX_generate_mipmap: * Generate a complete set of mipmaps from texObj's base-level image. @@ -1032,48 +1085,9 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, dstData = (GLubyte *) dstImage->Data; } - /* - * We use simple 2x2 averaging to compute the next mipmap level. - */ - switch (target) { - case GL_TEXTURE_1D: - make_1d_mipmap(datatype, comps, border, - srcWidth, srcData, - dstWidth, dstData); - break; - case GL_TEXTURE_2D: - case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: - make_2d_mipmap(datatype, comps, border, - srcWidth, srcHeight, srcData, - dstWidth, dstHeight, dstData); - break; - case GL_TEXTURE_3D: - make_3d_mipmap(datatype, comps, border, - srcWidth, srcHeight, srcDepth, srcData, - dstWidth, dstHeight, dstDepth, dstData); - break; - case GL_TEXTURE_1D_ARRAY_EXT: - make_1d_stack_mipmap(datatype, comps, border, - srcWidth, srcData, - dstWidth, dstHeight, dstData); - break; - case GL_TEXTURE_2D_ARRAY_EXT: - make_2d_stack_mipmap(datatype, comps, border, - srcWidth, srcHeight, srcData, - dstWidth, dstHeight, dstDepth, dstData); - break; - case GL_TEXTURE_RECTANGLE_NV: - /* no mipmaps, do nothing */ - break; - default: - _mesa_problem(ctx, "bad dimensions in _mesa_generate_mipmaps"); - return; - } + _mesa_generate_mipmap_level(target, datatype, comps, border, + srcWidth, srcHeight, srcDepth, srcData, + dstWidth, dstHeight, dstDepth, dstData); if (dstImage->IsCompressed) { GLubyte *temp; diff --git a/src/mesa/main/mipmap.h b/src/mesa/main/mipmap.h index 46e16902c86..b6491f55073 100644 --- a/src/mesa/main/mipmap.h +++ b/src/mesa/main/mipmap.h @@ -28,6 +28,17 @@ #include "mtypes.h" + +extern void +_mesa_generate_mipmap_level(GLenum target, + GLenum datatype, GLuint comps, + GLint border, + GLint srcWidth, GLint srcHeight, GLint srcDepth, + const GLubyte *srcData, + GLint dstWidth, GLint dstHeight, GLint dstDepth, + GLubyte *dstData); + + extern void _mesa_generate_mipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj); -- cgit v1.2.3 From 1c7d7da30c9c99f663b9c29636e2854e0bf6af4e Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 08:18:43 -0700 Subject: gallium: fix fragment/vertex typo --- src/mesa/pipe/draw/draw_vertex_shader.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/draw/draw_vertex_shader.c b/src/mesa/pipe/draw/draw_vertex_shader.c index 574ac670572..9c31df1e3e1 100644 --- a/src/mesa/pipe/draw/draw_vertex_shader.c +++ b/src/mesa/pipe/draw/draw_vertex_shader.c @@ -304,7 +304,7 @@ draw_bind_vertex_shader(struct draw_context *draw, draw->vertex_shader = dvs; draw->num_vs_outputs = dvs->state->num_outputs; - /* specify the fragment program to interpret/execute */ + /* specify the vertex program to interpret/execute */ tgsi_exec_machine_init(&draw->machine, draw->vertex_shader->state->tokens, PIPE_MAX_SAMPLERS, -- cgit v1.2.3 From 23e6d1aebc4c667a24243c89ffa836bc1fc74252 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 08:56:18 -0700 Subject: gallium: update comment about buffer map flags --- src/mesa/pipe/p_winsys.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/pipe/p_winsys.h b/src/mesa/pipe/p_winsys.h index 95e3684008d..1e81eebd782 100644 --- a/src/mesa/pipe/p_winsys.h +++ b/src/mesa/pipe/p_winsys.h @@ -112,7 +112,7 @@ struct pipe_winsys /** * Map the entire data store of a buffer object into the client's address. - * flags is bitmask of PIPE_BUFFER_FLAG_READ/WRITE. + * flags is bitmask of PIPE_BUFFER_USAGE_CPU_READ/WRITE flags. */ void *(*buffer_map)( struct pipe_winsys *sws, struct pipe_buffer *buf, -- cgit v1.2.3 From 939aa5d3bdc4728e6b848c0bbf150f8e644d5e1b Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:33:28 -0700 Subject: added _mesa_tex_target_to_face() --- src/mesa/main/teximage.c | 20 +++++++++++++------- src/mesa/main/teximage.h | 5 +++++ 2 files changed, 18 insertions(+), 7 deletions(-) diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c index 09ec0d45533..5c96be92165 100644 --- a/src/mesa/main/teximage.c +++ b/src/mesa/main/teximage.c @@ -595,8 +595,12 @@ is_compressed_format(GLcontext *ctx, GLenum internalFormat) } -static GLuint -texture_face(GLenum target) +/** + * For cube map faces, return a face index in [0,5]. + * For other targets return 0; + */ +GLuint +_mesa_tex_target_to_face(GLenum target) { if (target >= GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB && target <= GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB) @@ -625,6 +629,7 @@ _mesa_set_tex_image(struct gl_texture_object *tObj, { ASSERT(tObj); ASSERT(texImage); + /* XXX simplify this with _mesa_tex_target_to_face() */ switch (target) { case GL_TEXTURE_1D: case GL_TEXTURE_2D: @@ -828,6 +833,7 @@ _mesa_select_tex_image(GLcontext *ctx, const struct gl_texture_object *texObj, if (level < 0 || level >= MAX_TEXTURE_LEVELS) return NULL; + /* XXX simplify this with _mesa_tex_target_to_face() */ switch (target) { case GL_TEXTURE_1D: case GL_PROXY_TEXTURE_1D: @@ -2424,7 +2430,7 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat, struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - const GLuint face = texture_face(target); + const GLuint face = _mesa_tex_target_to_face(target); if (texture_error_check(ctx, target, level, internalFormat, format, type, 1, postConvWidth, 1, 1, border)) { @@ -2527,7 +2533,7 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat, struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - const GLuint face = texture_face(target); + const GLuint face = _mesa_tex_target_to_face(target); if (texture_error_check(ctx, target, level, internalFormat, format, type, 2, postConvWidth, postConvHeight, @@ -2629,7 +2635,7 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat, struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - const GLuint face = texture_face(target); + const GLuint face = _mesa_tex_target_to_face(target); if (texture_error_check(ctx, target, level, (GLint) internalFormat, format, type, 3, width, height, depth, border)) { @@ -2897,7 +2903,7 @@ _mesa_CopyTexImage1D( GLenum target, GLint level, struct gl_texture_object *texObj; struct gl_texture_image *texImage; GLsizei postConvWidth = width; - const GLuint face = texture_face(target); + const GLuint face = _mesa_tex_target_to_face(target); GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); @@ -2960,7 +2966,7 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat, struct gl_texture_object *texObj; struct gl_texture_image *texImage; GLsizei postConvWidth = width, postConvHeight = height; - const GLuint face = texture_face(target); + const GLuint face = _mesa_tex_target_to_face(target); GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); diff --git a/src/mesa/main/teximage.h b/src/mesa/main/teximage.h index f2cad7eb2da..b718c0046d8 100644 --- a/src/mesa/main/teximage.h +++ b/src/mesa/main/teximage.h @@ -107,6 +107,11 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level, GLint width, GLint height, GLint depth, GLint border); +extern GLuint +_mesa_tex_target_to_face(GLenum target); + + + /** * Lock a texture for updating. See also _mesa_lock_context_textures(). */ -- cgit v1.2.3 From ba499584d624687e91c6436f8ea539ae77173cd4 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:35:28 -0700 Subject: gallium: use _mesa_tex_target_to_face() --- src/mesa/state_tracker/st_cb_texture.c | 22 +--------------------- 1 file changed, 1 insertion(+), 21 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 3350254654b..992723afba8 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -379,26 +379,6 @@ guess_and_alloc_texture(struct st_context *st, } - - -static GLuint -target_to_face(GLenum target) -{ - switch (target) { - case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: - return ((GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X); - default: - return 0; - } -} - - - /* There are actually quite a few combinations this will work for, * more than what I've listed here. */ @@ -512,7 +492,7 @@ st_TexImage(GLcontext * ctx, DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), level, width, height, depth, border); - stImage->face = target_to_face(target); + stImage->face = _mesa_tex_target_to_face(target); stImage->level = level; if (ctx->_ImageTransferState & IMAGE_CONVOLUTION_BIT) { -- cgit v1.2.3 From 59f0ce86b8696c6dadfaab3099ed2193b411e8d0 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:38:36 -0700 Subject: gallium: strip borders from textures passed to st_TexImage. Manipulate the unpack params to skip the border. Gallium drivers won't support texture borders. --- src/mesa/state_tracker/st_cb_texture.c | 53 ++++++++++++++++++++++++++++++++-- 1 file changed, 50 insertions(+), 3 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 992723afba8..7099ec33b97 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -467,6 +467,43 @@ try_pbo_upload(GLcontext *ctx, } +/** + * Adjust pixel unpack params and image dimensions to strip off the + * texture border. + * Gallium doesn't support texture borders. They've seldem been used + * and seldom been implemented correctly anyway. + * \param unpackNew returns the new pixel unpack parameters + */ +static void +strip_texture_border(GLint border, + GLint *width, GLint *height, GLint *depth, + const struct gl_pixelstore_attrib *unpack, + struct gl_pixelstore_attrib *unpackNew) +{ + assert(border > 0); /* sanity check */ + + *unpackNew = *unpack; + + if (unpackNew->RowLength == 0) + unpackNew->RowLength = *width; + + if (depth && unpackNew->ImageHeight == 0) + unpackNew->ImageHeight = *height; + + unpackNew->SkipPixels += border; + if (height) + unpackNew->SkipRows += border; + if (depth) + unpackNew->SkipImages += border; + + assert(*width >= 3); + *width = *width - 2 * border; + if (height && *height >= 3) + *height = *height - 2 * border; + if (depth && *depth >= 3) + *depth = *depth - 2 * border; +} + static void st_TexImage(GLcontext * ctx, @@ -483,15 +520,25 @@ st_TexImage(GLcontext * ctx, { struct st_texture_object *stObj = st_texture_object(texObj); struct st_texture_image *stImage = st_texture_image(texImage); - GLint postConvWidth = width; - GLint postConvHeight = height; + GLint postConvWidth, postConvHeight; GLint texelBytes, sizeInBytes; GLuint dstRowStride; - + struct gl_pixelstore_attrib unpackNB; DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), level, width, height, depth, border); + /* gallium does not support texture borders, strip it off */ + if (border) { + strip_texture_border(border, &width, &height, &depth, + unpack, &unpackNB); + unpack = &unpackNB; + border = 0; + } + + postConvWidth = width; + postConvHeight = height; + stImage->face = _mesa_tex_target_to_face(target); stImage->level = level; -- cgit v1.2.3 From 995924d566e3a5c06ee4728b846c18de39574966 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:42:02 -0700 Subject: gallium: implement software fallback for mipmap generation This is used when we can't render to the surface type of the texture (such as luminance/alpha). --- src/mesa/state_tracker/st_gen_mipmap.c | 68 ++++++++++++++++++++++++++++++---- 1 file changed, 60 insertions(+), 8 deletions(-) diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index a6ac9a55fb3..5c00392af7c 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -29,11 +29,14 @@ #include "main/imports.h" #include "main/mipmap.h" #include "main/teximage.h" +#include "main/texformat.h" #include "shader/prog_instruction.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_inlines.h" +#include "pipe/p_winsys.h" #include "pipe/cso_cache/cso_cache.h" #include "st_context.h" @@ -239,15 +242,18 @@ draw_quad(GLcontext *ctx) */ static boolean st_render_mipmap(struct st_context *st, + GLenum target, struct pipe_texture *pt, uint baseLevel, uint lastLevel) { struct pipe_context *pipe = st->pipe; struct pipe_framebuffer_state fb; - const uint face = 0, zslice = 0; + const uint face = _mesa_tex_target_to_face(target), zslice = 0; const uint first_level_save = pt->first_level; uint dstLevel; + assert(target != GL_TEXTURE_3D); /* not done yet */ + /* check if we can render in the texture's format */ if (!pipe->is_format_supported(pipe, pt->format, PIPE_SURFACE)) { return FALSE; @@ -307,6 +313,56 @@ st_render_mipmap(struct st_context *st, } +static void +fallback_generate_mipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) +{ + struct pipe_context *pipe = ctx->st->pipe; + struct pipe_winsys *ws = pipe->winsys; + struct pipe_texture *pt = st_get_texobj_texture(texObj); + const uint baseLevel = texObj->BaseLevel; + const uint lastLevel = pt->last_level; + const uint face = _mesa_tex_target_to_face(target), zslice = 0; + uint dstLevel; + GLenum datatype; + GLuint comps; + + assert(target != GL_TEXTURE_3D); /* not done yet */ + + _mesa_format_to_type_and_comps(texObj->Image[face][baseLevel]->TexFormat, + &datatype, &comps); + + for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) { + const uint srcLevel = dstLevel - 1; + struct pipe_surface *srcSurf, *dstSurf; + const ubyte *srcData; + ubyte *dstData; + + srcSurf = pipe->get_tex_surface(pipe, pt, face, srcLevel, zslice); + dstSurf = pipe->get_tex_surface(pipe, pt, face, dstLevel, zslice); + + srcData = (ubyte *) ws->buffer_map(ws, srcSurf->buffer, + PIPE_BUFFER_USAGE_CPU_READ) + + srcSurf->offset; + dstData = (ubyte *) ws->buffer_map(ws, dstSurf->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE) + + dstSurf->offset; + + _mesa_generate_mipmap_level(target, datatype, comps, + 0 /*border*/, + pt->width[srcLevel], pt->height[srcLevel], pt->depth[srcLevel], + srcData, + pt->width[dstLevel], pt->height[dstLevel], pt->depth[dstLevel], + dstData); + + ws->buffer_unmap(ws, srcSurf->buffer); + ws->buffer_unmap(ws, dstSurf->buffer); + + pipe_surface_reference(&srcSurf, NULL); + pipe_surface_reference(&dstSurf, NULL); + } +} + void st_generate_mipmap(GLcontext *ctx, GLenum target, @@ -318,13 +374,11 @@ st_generate_mipmap(GLcontext *ctx, GLenum target, const uint lastLevel = pt->last_level; uint dstLevel; - if (!st_render_mipmap(st, pt, baseLevel, lastLevel)) { - abort(); - /* XXX the following won't really work at this time */ - _mesa_generate_mipmap(ctx, target, texObj); - return; + if (!st_render_mipmap(st, target, pt, baseLevel, lastLevel)) { + fallback_generate_mipmap(ctx, target, texObj); } + /* Fill in the Mesa gl_texture_image fields */ for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) { const uint srcLevel = dstLevel - 1; const struct gl_texture_image *srcImage @@ -336,7 +390,6 @@ st_generate_mipmap(GLcontext *ctx, GLenum target, uint dstDepth = pt->depth[dstLevel]; uint border = srcImage->Border; - dstImage = _mesa_get_tex_image(ctx, texObj, target, dstLevel); if (!dstImage) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "generating mipmaps"); @@ -359,5 +412,4 @@ st_generate_mipmap(GLcontext *ctx, GLenum target, stImage = (struct st_texture_image *) dstImage; stImage->pt = pt; } - } -- cgit v1.2.3 From a705e157f2d14a7fcc81b292fcca9dab4f38c9b2 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:43:30 -0700 Subject: code refactoring, new next_mipmap_level_size() function --- src/mesa/main/mipmap.c | 75 +++++++++++++++++++++++++++++++++----------------- 1 file changed, 50 insertions(+), 25 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index ed7795aef93..721a227bb53 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -894,6 +894,51 @@ _mesa_generate_mipmap_level(GLenum target, } +/** + * compute next (level+1) image size + * \return GL_FALSE if no smaller size can be generated (eg. src is 1x1x1 size) + */ +static GLboolean +next_mipmap_level_size(GLenum target, GLint border, + GLint srcWidth, GLint srcHeight, GLint srcDepth, + GLint *dstWidth, GLint *dstHeight, GLint *dstDepth) +{ + if (srcWidth - 2 * border > 1) { + *dstWidth = (srcWidth - 2 * border) / 2 + 2 * border; + } + else { + *dstWidth = srcWidth; /* can't go smaller */ + } + + if ((srcHeight - 2 * border > 1) && + (target != GL_TEXTURE_1D_ARRAY_EXT)) { + *dstHeight = (srcHeight - 2 * border) / 2 + 2 * border; + } + else { + *dstHeight = srcHeight; /* can't go smaller */ + } + + if ((srcDepth - 2 * border > 1) && + (target != GL_TEXTURE_2D_ARRAY_EXT)) { + *dstDepth = (srcDepth - 2 * border) / 2 + 2 * border; + } + else { + *dstDepth = srcDepth; /* can't go smaller */ + } + + if (*dstWidth == srcWidth && + *dstHeight == srcHeight && + *dstDepth == srcDepth) { + return GL_FALSE; + } + else { + return GL_TRUE; + } +} + + + + /** * For GL_SGIX_generate_mipmap: * Generate a complete set of mipmaps from texObj's base-level image. @@ -982,6 +1027,7 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, GLint srcWidth, srcHeight, srcDepth; GLint dstWidth, dstHeight, dstDepth; GLint border, bytesPerTexel; + GLboolean nextLevel; /* get src image parameters */ srcImage = _mesa_select_tex_image(ctx, texObj, target, level); @@ -991,31 +1037,10 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, srcDepth = srcImage->Depth; border = srcImage->Border; - /* compute next (level+1) image size */ - if (srcWidth - 2 * border > 1) { - dstWidth = (srcWidth - 2 * border) / 2 + 2 * border; - } - else { - dstWidth = srcWidth; /* can't go smaller */ - } - if ((srcHeight - 2 * border > 1) && - (texObj->Target != GL_TEXTURE_1D_ARRAY_EXT)) { - dstHeight = (srcHeight - 2 * border) / 2 + 2 * border; - } - else { - dstHeight = srcHeight; /* can't go smaller */ - } - if ((srcDepth - 2 * border > 1) && - (texObj->Target != GL_TEXTURE_2D_ARRAY_EXT)) { - dstDepth = (srcDepth - 2 * border) / 2 + 2 * border; - } - else { - dstDepth = srcDepth; /* can't go smaller */ - } - - if (dstWidth == srcWidth && - dstHeight == srcHeight && - dstDepth == srcDepth) { + nextLevel = next_mipmap_level_size(target, border, + srcWidth, srcHeight, srcDepth, + &dstWidth, &dstHeight, &dstDepth); + if (!nextLevel) { /* all done */ if (srcImage->IsCompressed) { _mesa_free((void *) srcData); -- cgit v1.2.3 From 48c4a1ed12d30932c5a9d09424213a830efe2ef9 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:50:21 -0700 Subject: gallium: comments about mipmap gen --- src/mesa/state_tracker/st_gen_mipmap.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index 5c00392af7c..f6af37cfac3 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -273,6 +273,10 @@ st_render_mipmap(struct st_context *st, pipe->bind_fs_state(pipe, stfp->fs->data); pipe->bind_vs_state(pipe, stvp->cso->data); + /* + * XXX for small mipmap levels, it may be faster to use the software + * fallback path... + */ for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) { const uint srcLevel = dstLevel - 1; @@ -348,6 +352,7 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, PIPE_BUFFER_USAGE_CPU_WRITE) + dstSurf->offset; + /* XXX need to take stride/pitch info into account... */ _mesa_generate_mipmap_level(target, datatype, comps, 0 /*border*/, pt->width[srcLevel], pt->height[srcLevel], pt->depth[srcLevel], -- cgit v1.2.3 From 3812bba8391fbf6c6c32a778ce0e1081825d5c52 Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 10:59:40 -0700 Subject: gallium: take pitch/stride into account in mipmap generation --- src/mesa/main/mipmap.c | 85 +++++++++++++++++++++------------- src/mesa/main/mipmap.h | 2 + src/mesa/state_tracker/st_gen_mipmap.c | 3 +- 3 files changed, 57 insertions(+), 33 deletions(-) diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c index 721a227bb53..d3d19589518 100644 --- a/src/mesa/main/mipmap.c +++ b/src/mesa/main/mipmap.c @@ -447,23 +447,29 @@ make_1d_mipmap(GLenum datatype, GLuint comps, GLint border, /** - * XXX need to use the tex image's row stride! + * Strides are in bytes. If zero, it'll be computed as width * bpp. */ static void make_2d_mipmap(GLenum datatype, GLuint comps, GLint border, - GLint srcWidth, GLint srcHeight, const GLubyte *srcPtr, - GLint dstWidth, GLint dstHeight, GLubyte *dstPtr) + GLint srcWidth, GLint srcHeight, + GLint srcRowStride, const GLubyte *srcPtr, + GLint dstWidth, GLint dstHeight, + GLint dstRowStride, GLubyte *dstPtr) { const GLint bpt = bytes_per_pixel(datatype, comps); const GLint srcWidthNB = srcWidth - 2 * border; /* sizes w/out border */ const GLint dstWidthNB = dstWidth - 2 * border; const GLint dstHeightNB = dstHeight - 2 * border; - const GLint srcRowStride = bpt * srcWidth; - const GLint dstRowStride = bpt * dstWidth; const GLubyte *srcA, *srcB; GLubyte *dst; GLint row; + if (!srcRowStride) + srcRowStride = bpt * srcWidth; + + if (!dstRowStride) + dstRowStride = bpt * dstWidth; + /* Compute src and dst pointers, skipping any border */ srcA = srcPtr + border * ((srcWidth + 1) * bpt); if (srcHeight > 1) @@ -535,8 +541,10 @@ make_2d_mipmap(GLenum datatype, GLuint comps, GLint border, static void make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, GLint srcWidth, GLint srcHeight, GLint srcDepth, + GLint srcRowStride, const GLubyte *srcPtr, GLint dstWidth, GLint dstHeight, GLint dstDepth, + GLint dstRowStride, GLubyte *dstPtr) { const GLint bpt = bytes_per_pixel(datatype, comps); @@ -548,7 +556,6 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, GLvoid *tmpRowA, *tmpRowB; GLint img, row; GLint bytesPerSrcImage, bytesPerDstImage; - GLint bytesPerSrcRow, bytesPerDstRow; GLint srcImageOffset, srcRowOffset; (void) srcDepthNB; /* silence warnings */ @@ -566,8 +573,10 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, bytesPerSrcImage = srcWidth * srcHeight * bpt; bytesPerDstImage = dstWidth * dstHeight * bpt; - bytesPerSrcRow = srcWidth * bpt; - bytesPerDstRow = dstWidth * bpt; + if (!srcRowStride) + srcRowStride = srcWidth * bpt; + if (!dstRowStride) + dstRowStride = dstWidth * bpt; /* Offset between adjacent src images to be averaged together */ srcImageOffset = (srcDepth == dstDepth) ? 0 : bytesPerSrcImage; @@ -591,13 +600,13 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, for (img = 0; img < dstDepthNB; img++) { /* first source image pointer, skipping border */ const GLubyte *imgSrcA = srcPtr - + (bytesPerSrcImage + bytesPerSrcRow + border) * bpt * border + + (bytesPerSrcImage + srcRowStride + border) * bpt * border + img * (bytesPerSrcImage + srcImageOffset); /* second source image pointer, skipping border */ const GLubyte *imgSrcB = imgSrcA + srcImageOffset; /* address of the dest image, skipping border */ GLubyte *imgDst = dstPtr - + (bytesPerDstImage + bytesPerDstRow + border) * bpt * border + + (bytesPerDstImage + dstRowStride + border) * bpt * border + img * bytesPerDstImage; /* setup the four source row pointers and the dest row pointer */ @@ -618,11 +627,11 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, do_row(datatype, comps, srcWidthNB, tmpRowA, tmpRowB, dstWidthNB, dstImgRow); /* advance to next rows */ - srcImgARowA += bytesPerSrcRow + srcRowOffset; - srcImgARowB += bytesPerSrcRow + srcRowOffset; - srcImgBRowA += bytesPerSrcRow + srcRowOffset; - srcImgBRowB += bytesPerSrcRow + srcRowOffset; - dstImgRow += bytesPerDstRow; + srcImgARowA += srcRowStride + srcRowOffset; + srcImgARowB += srcRowStride + srcRowOffset; + srcImgBRowA += srcRowStride + srcRowOffset; + srcImgBRowB += srcRowStride + srcRowOffset; + dstImgRow += dstRowStride; } } @@ -632,12 +641,14 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, /* Luckily we can leverage the make_2d_mipmap() function here! */ if (border > 0) { /* do front border image */ - make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, srcPtr, - dstWidth, dstHeight, dstPtr); + make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, 0, srcPtr, + dstWidth, dstHeight, 0, dstPtr); /* do back border image */ make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, + 0, srcPtr + bytesPerSrcImage * (srcDepth - 1), dstWidth, dstHeight, + 0, dstPtr + bytesPerDstImage * (dstDepth - 1)); /* do four remaining border edges that span the image slices */ if (srcDepth == dstDepth) { @@ -653,9 +664,9 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, /* do border along [img][row=dstHeight-1][col=0] */ src = srcPtr + (img * 2 + 1) * bytesPerSrcImage - + (srcHeight - 1) * bytesPerSrcRow; + + (srcHeight - 1) * srcRowStride; dst = dstPtr + (img + 1) * bytesPerDstImage - + (dstHeight - 1) * bytesPerDstRow; + + (dstHeight - 1) * dstRowStride; MEMCPY(dst, src, bpt); /* do border along [img][row=0][col=dstWidth-1] */ @@ -687,9 +698,9 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border, /* do border along [img][row=dstHeight-1][col=0] */ src = srcPtr + (img * 2 + 1) * bytesPerSrcImage - + (srcHeight - 1) * bytesPerSrcRow; + + (srcHeight - 1) * srcRowStride; dst = dstPtr + (img + 1) * bytesPerDstImage - + (dstHeight - 1) * bytesPerDstRow; + + (dstHeight - 1) * dstRowStride; do_row(datatype, comps, 1, src, src + srcImageOffset, 1, dst); /* do border along [img][row=0][col=dstWidth-1] */ @@ -755,8 +766,11 @@ make_1d_stack_mipmap(GLenum datatype, GLuint comps, GLint border, */ static void make_2d_stack_mipmap(GLenum datatype, GLuint comps, GLint border, - GLint srcWidth, GLint srcHeight, const GLubyte *srcPtr, + GLint srcWidth, GLint srcHeight, + GLint srcRowStride, + const GLubyte *srcPtr, GLint dstWidth, GLint dstHeight, GLint dstDepth, + GLint dstRowStride, GLubyte *dstPtr) { const GLint bpt = bytes_per_pixel(datatype, comps); @@ -764,13 +778,17 @@ make_2d_stack_mipmap(GLenum datatype, GLuint comps, GLint border, const GLint dstWidthNB = dstWidth - 2 * border; const GLint dstHeightNB = dstHeight - 2 * border; const GLint dstDepthNB = dstDepth - 2 * border; - const GLint srcRowStride = bpt * srcWidth; - const GLint dstRowStride = bpt * dstWidth; const GLubyte *srcA, *srcB; GLubyte *dst; GLint layer; GLint row; + if (!srcRowStride) + srcRowStride = bpt * srcWidth; + + if (!dstRowStride) + dstRowStride = bpt * dstWidth; + /* Compute src and dst pointers, skipping any border */ srcA = srcPtr + border * ((srcWidth + 1) * bpt); if (srcHeight > 1) @@ -849,8 +867,10 @@ _mesa_generate_mipmap_level(GLenum target, GLenum datatype, GLuint comps, GLint border, GLint srcWidth, GLint srcHeight, GLint srcDepth, + GLint srcRowStride, const GLubyte *srcData, GLint dstWidth, GLint dstHeight, GLint dstDepth, + GLint dstRowStride, GLubyte *dstData) { switch (target) { @@ -867,13 +887,13 @@ _mesa_generate_mipmap_level(GLenum target, case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: make_2d_mipmap(datatype, comps, border, - srcWidth, srcHeight, srcData, - dstWidth, dstHeight, dstData); + srcWidth, srcHeight, srcRowStride, srcData, + dstWidth, dstHeight, dstRowStride, dstData); break; case GL_TEXTURE_3D: make_3d_mipmap(datatype, comps, border, - srcWidth, srcHeight, srcDepth, srcData, - dstWidth, dstHeight, dstDepth, dstData); + srcWidth, srcHeight, srcDepth, srcRowStride, srcData, + dstWidth, dstHeight, dstDepth, dstRowStride, dstData); break; case GL_TEXTURE_1D_ARRAY_EXT: make_1d_stack_mipmap(datatype, comps, border, @@ -882,8 +902,8 @@ _mesa_generate_mipmap_level(GLenum target, break; case GL_TEXTURE_2D_ARRAY_EXT: make_2d_stack_mipmap(datatype, comps, border, - srcWidth, srcHeight, srcData, - dstWidth, dstHeight, dstDepth, dstData); + srcWidth, srcHeight, srcRowStride, srcData, + dstWidth, dstHeight, dstDepth, dstRowStride, dstData); break; case GL_TEXTURE_RECTANGLE_NV: /* no mipmaps, do nothing */ @@ -1110,9 +1130,10 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target, dstData = (GLubyte *) dstImage->Data; } + /* Note, 0 indicates default row strides */ _mesa_generate_mipmap_level(target, datatype, comps, border, - srcWidth, srcHeight, srcDepth, srcData, - dstWidth, dstHeight, dstDepth, dstData); + srcWidth, srcHeight, srcDepth, 0, srcData, + dstWidth, dstHeight, dstDepth, 0, dstData); if (dstImage->IsCompressed) { GLubyte *temp; diff --git a/src/mesa/main/mipmap.h b/src/mesa/main/mipmap.h index b6491f55073..44ecdddb277 100644 --- a/src/mesa/main/mipmap.h +++ b/src/mesa/main/mipmap.h @@ -34,8 +34,10 @@ _mesa_generate_mipmap_level(GLenum target, GLenum datatype, GLuint comps, GLint border, GLint srcWidth, GLint srcHeight, GLint srcDepth, + GLint srcRowStride, const GLubyte *srcData, GLint dstWidth, GLint dstHeight, GLint dstDepth, + GLint dstRowStride, GLubyte *dstData); diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index f6af37cfac3..c152c599056 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -352,12 +352,13 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, PIPE_BUFFER_USAGE_CPU_WRITE) + dstSurf->offset; - /* XXX need to take stride/pitch info into account... */ _mesa_generate_mipmap_level(target, datatype, comps, 0 /*border*/, pt->width[srcLevel], pt->height[srcLevel], pt->depth[srcLevel], + srcSurf->pitch * srcSurf->cpp, /* stride in bytes */ srcData, pt->width[dstLevel], pt->height[dstLevel], pt->depth[dstLevel], + dstSurf->pitch * dstSurf->cpp, /* stride in bytes */ dstData); ws->buffer_unmap(ws, srcSurf->buffer); -- cgit v1.2.3 From 5d1af60edb1dbdf69fbf08b93fe0781f33f075dd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 12 Feb 2008 15:13:37 +1100 Subject: nv40: fix inline u08/u16 indices --- src/mesa/pipe/nv40/nv40_vbo.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 552058b3ae7..4e9cdb4585a 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -235,9 +235,9 @@ nv40_draw_elements_u08(struct nv40_context *nv40, void *ib, } while (count) { - push = MIN2(count, 2046); + push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push); + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); @@ -260,9 +260,9 @@ nv40_draw_elements_u16(struct nv40_context *nv40, void *ib, } while (count) { - push = MIN2(count, 2046); + push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push); + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); -- cgit v1.2.3 From 09e23e077b2bc3dc9ec0ecd97e1043ee7f32f2bb Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 14:53:25 -0700 Subject: gallium: clean-up, simplification of mipmapped textures Remove pipe_texture->first_level (always implicitly zero). This means there's never any unused mipmap levels at the top. In the state tracker, we no longer re-layout mipmapped textures if the MinLod/MaxLod texture parameters change. It's up to the driver to obey the pipe_sampler->min/max_lod clamps. --- src/mesa/pipe/i915simple/i915_state.c | 7 ++++ src/mesa/pipe/i915simple/i915_state_sampler.c | 2 +- src/mesa/pipe/i915simple/i915_texture.c | 30 +++++++-------- src/mesa/pipe/i965simple/brw_tex_layout.c | 8 ++-- src/mesa/pipe/i965simple/brw_wm_surface_state.c | 2 +- src/mesa/pipe/p_state.h | 14 ++----- src/mesa/pipe/softpipe/sp_state_sampler.c | 4 ++ src/mesa/pipe/softpipe/sp_tex_sample.c | 18 +++------ src/mesa/pipe/softpipe/sp_texture.c | 4 +- src/mesa/state_tracker/st_cb_texture.c | 51 +++++++++---------------- src/mesa/state_tracker/st_gen_mipmap.c | 43 ++++++++++++--------- src/mesa/state_tracker/st_texture.c | 35 ++++++++++++----- src/mesa/state_tracker/st_texture.h | 6 +-- 13 files changed, 115 insertions(+), 109 deletions(-) diff --git a/src/mesa/pipe/i915simple/i915_state.c b/src/mesa/pipe/i915simple/i915_state.c index 950ea52d601..abd5571b885 100644 --- a/src/mesa/pipe/i915simple/i915_state.c +++ b/src/mesa/pipe/i915simple/i915_state.c @@ -250,6 +250,13 @@ i915_create_sampler_state(struct pipe_context *pipe, if (sampler->normalized_coords) cso->state[1] |= SS3_NORMALIZED_COORDS; + if (0) /* XXX not tested yet */ + { + int minlod = (int) (16.0 * sampler->min_lod); + minlod = CLAMP(minlod, 0, 16 * 11); + cso->state[1] |= (minlod << SS3_MIN_LOD_SHIFT); + } + { ubyte r = float_to_ubyte(sampler->border_color[0]); ubyte g = float_to_ubyte(sampler->border_color[1]); diff --git a/src/mesa/pipe/i915simple/i915_state_sampler.c b/src/mesa/pipe/i915simple/i915_state_sampler.c index 0dbbc5241d5..9c1a5bbbd65 100644 --- a/src/mesa/pipe/i915simple/i915_state_sampler.c +++ b/src/mesa/pipe/i915simple/i915_state_sampler.c @@ -185,7 +185,7 @@ i915_update_texture(struct i915_context *i915, uint unit, const struct pipe_texture *pt = &tex->base; uint format, pitch; const uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; - const uint num_levels = pt->last_level - pt->first_level; + const uint num_levels = pt->last_level; assert(tex); assert(width); diff --git a/src/mesa/pipe/i915simple/i915_texture.c b/src/mesa/pipe/i915simple/i915_texture.c index 6faeab134ab..6d37ae3d747 100644 --- a/src/mesa/pipe/i915simple/i915_texture.c +++ b/src/mesa/pipe/i915simple/i915_texture.c @@ -118,11 +118,11 @@ i945_miptree_layout_2d( struct i915_texture *tex ) tex->pitch = pt->width[0]; /* May need to adjust pitch to accomodate the placement of - * the 2nd mipmap. This occurs when the alignment + * the 2nd mipmap level. This occurs when the alignment * constraints of mipmap placement push the right edge of the - * 2nd mipmap out past the width of its parent. + * 2nd mipmap level out past the width of its parent. */ - if (pt->first_level != pt->last_level) { + if (pt->last_level > 0) { unsigned mip1_width = align_int(minify(pt->width[0]), align_w) + minify(minify(pt->width[0])); @@ -136,7 +136,7 @@ i945_miptree_layout_2d( struct i915_texture *tex ) tex->pitch = align_int(tex->pitch * pt->cpp, 4) / pt->cpp; tex->total_height = 0; - for ( level = pt->first_level ; level <= pt->last_level ; level++ ) { + for (level = 0; level <= pt->last_level; level++) { unsigned img_height; i915_miptree_set_level_info(tex, level, 1, x, y, width, height, 1); @@ -152,9 +152,9 @@ i945_miptree_layout_2d( struct i915_texture *tex ) */ tex->total_height = MAX2(tex->total_height, y + img_height); - /* Layout_below: step right after second mipmap. + /* Layout_below: step right after second mipmap level. */ - if (level == pt->first_level + 1) { + if (level == 1) { x += align_int(width, align_w); } else { @@ -204,7 +204,7 @@ i915_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) tex->pitch = ((dim * pt->cpp * 2 + 3) & ~3) / pt->cpp; tex->total_height = dim * 4; - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { i915_miptree_set_level_info(tex, level, 6, 0, 0, /*OLD: tex->pitch, tex->total_height,*/ @@ -219,7 +219,7 @@ i915_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) unsigned y = initial_offsets[face][1] * dim; unsigned d = dim; - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { i915_miptree_set_image_offset(tex, level, face, x, y); d >>= 1; x += step_offsets[face][0] * d; @@ -240,7 +240,7 @@ i915_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) /* XXX: hardware expects/requires 9 levels at minimum. */ - for (level = pt->first_level; level <= MAX2(8, pt->last_level); + for (level = 0; level <= MAX2(8, pt->last_level); level++) { i915_miptree_set_level_info(tex, level, depth, 0, tex->total_height, width, height, depth); @@ -256,7 +256,7 @@ i915_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) /* Fixup depth image_offsets: */ depth = pt->depth[0]; - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { unsigned i; for (i = 0; i < depth; i++) i915_miptree_set_image_offset(tex, level, i, @@ -282,7 +282,7 @@ i915_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) tex->pitch = ((pt->width[0] * pt->cpp + 3) & ~3) / pt->cpp; tex->total_height = 0; - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { i915_miptree_set_level_info(tex, level, 1, 0, tex->total_height, width, height, 1); @@ -337,7 +337,7 @@ i945_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) /* Set all the levels to effectively occupy the whole rectangular region. */ - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { i915_miptree_set_level_info(tex, level, 6, 0, 0, lvlWidth, lvlHeight, 1); @@ -355,12 +355,12 @@ i945_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) y = tex->total_height - 4; x = (face - 4) * 8; } - else if (dim < 4 && (face > 0 || pt->first_level > 0)) { + else if (dim < 4 && (face > 0)) { y = tex->total_height - 4; x = face * 8; } - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { i915_miptree_set_image_offset(tex, level, face, x, y); d >>= 1; @@ -418,7 +418,7 @@ i945_miptree_layout(struct pipe_context *pipe, struct i915_texture * tex) pack_x_pitch = tex->pitch; pack_x_nr = 1; - for (level = pt->first_level; level <= pt->last_level; level++) { + for (level = 0; level <= pt->last_level; level++) { unsigned nr_images = pt->target == PIPE_TEXTURE_3D ? depth : 6; int x = 0; int y = 0; diff --git a/src/mesa/pipe/i965simple/brw_tex_layout.c b/src/mesa/pipe/i965simple/brw_tex_layout.c index 405fd1f7949..90561f1307d 100644 --- a/src/mesa/pipe/i965simple/brw_tex_layout.c +++ b/src/mesa/pipe/i965simple/brw_tex_layout.c @@ -146,7 +146,7 @@ static void i945_miptree_layout_2d(struct brw_texture *tex) * constraints of mipmap placement push the right edge of the * 2nd mipmap out past the width of its parent. */ - if (pt->first_level != pt->last_level) { + if (pt->last_level > 0) { unsigned mip1_width; if (pt->compressed) { @@ -168,7 +168,7 @@ static void i945_miptree_layout_2d(struct brw_texture *tex) tex->pitch = align(tex->pitch * pt->cpp, 4) / pt->cpp; tex->total_height = 0; - for ( level = pt->first_level ; level <= pt->last_level ; level++ ) { + for (level = 0; level <= pt->last_level; level++) { unsigned img_height; intel_miptree_set_level_info(tex, level, 1, x, y, width, @@ -187,7 +187,7 @@ static void i945_miptree_layout_2d(struct brw_texture *tex) /* Layout_below: step right after second mipmap. */ - if (level == pt->first_level + 1) { + if (level == 1) { x += align(width, align_w); } else { @@ -234,7 +234,7 @@ static boolean brw_miptree_layout(struct pipe_context *pipe, struct brw_texture pack_x_pitch = tex->pitch; pack_x_nr = 1; - for ( level = pt->first_level ; level <= pt->last_level ; level++ ) { + for (level = 0; level <= pt->last_level; level++) { unsigned nr_images = pt->target == PIPE_TEXTURE_3D ? depth : 6; int x = 0; int y = 0; diff --git a/src/mesa/pipe/i965simple/brw_wm_surface_state.c b/src/mesa/pipe/i965simple/brw_wm_surface_state.c index cbb4f2efd3b..d16d919bce9 100644 --- a/src/mesa/pipe/i965simple/brw_wm_surface_state.c +++ b/src/mesa/pipe/i965simple/brw_wm_surface_state.c @@ -154,7 +154,7 @@ void brw_update_texture_surface( struct brw_context *brw, /* Updated in emit_reloc */ surf.ss1.base_addr = brw_buffer_offset( brw, tObj->buffer ); - surf.ss2.mip_count = tObj->base.last_level - tObj->base.first_level; + surf.ss2.mip_count = tObj->base.last_level; surf.ss2.width = tObj->base.width[0] - 1; surf.ss2.height = tObj->base.height[0] - 1; diff --git a/src/mesa/pipe/p_state.h b/src/mesa/pipe/p_state.h index 83ca43f6785..4d3a6b2f413 100644 --- a/src/mesa/pipe/p_state.h +++ b/src/mesa/pipe/p_state.h @@ -234,14 +234,9 @@ struct pipe_sampler_state unsigned compare_mode:1; /**< PIPE_TEX_COMPARE_x */ unsigned compare_func:3; /**< PIPE_FUNC_x */ unsigned normalized_coords:1; /**< Are coords normalized to [0,1]? */ - float shadow_ambient; /**< shadow test fail color/intensity */ - float min_lod; - float max_lod; - float lod_bias; -#if 0 /* need these? */ - int BaseLevel; /**< min mipmap level, OpenGL 1.2 */ - int MaxLevel; /**< max mipmap level, OpenGL 1.2 */ -#endif + float shadow_ambient; /**< shadow test fail color/intensity */ + float lod_bias; /**< LOD/lambda bias */ + float min_lod, max_lod; /**< LOD clamp range, after bias */ float border_color[4]; float max_anisotropy; }; @@ -277,8 +272,7 @@ struct pipe_texture enum pipe_texture_target target; /**< PIPE_TEXTURE_x */ enum pipe_format format; /**< PIPE_FORMAT_x */ - unsigned first_level; - unsigned last_level; + unsigned last_level; /**< Index of last mipmap level present/defined */ unsigned width[PIPE_MAX_TEXTURE_LEVELS]; unsigned height[PIPE_MAX_TEXTURE_LEVELS]; diff --git a/src/mesa/pipe/softpipe/sp_state_sampler.c b/src/mesa/pipe/softpipe/sp_state_sampler.c index 291bbc40ad0..6a5a643c89e 100644 --- a/src/mesa/pipe/softpipe/sp_state_sampler.c +++ b/src/mesa/pipe/softpipe/sp_state_sampler.c @@ -34,6 +34,8 @@ #include "sp_state.h" #include "sp_texture.h" #include "sp_tile_cache.h" +#include "pipe/draw/draw_context.h" + void * @@ -73,6 +75,8 @@ softpipe_set_sampler_texture(struct pipe_context *pipe, { struct softpipe_context *softpipe = softpipe_context(pipe); + draw_flush(softpipe->draw); + assert(unit < PIPE_MAX_SAMPLERS); softpipe->texture[unit] = softpipe_texture(texture); /* ptr, not struct */ diff --git a/src/mesa/pipe/softpipe/sp_tex_sample.c b/src/mesa/pipe/softpipe/sp_tex_sample.c index 5e215c433ad..325bdb86da5 100644 --- a/src/mesa/pipe/softpipe/sp_tex_sample.c +++ b/src/mesa/pipe/softpipe/sp_tex_sample.c @@ -449,7 +449,6 @@ compute_lambda(struct tgsi_sampler *sampler, } lambda = LOG2(rho); - lambda += lodbias + sampler->state->lod_bias; lambda = CLAMP(lambda, sampler->state->min_lod, sampler->state->max_lod); @@ -457,7 +456,6 @@ compute_lambda(struct tgsi_sampler *sampler, } - /** * Do several things here: * 1. Compute lambda from the texcoords, if needed @@ -477,7 +475,7 @@ choose_mipmap_levels(struct tgsi_sampler *sampler, if (sampler->state->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) { /* no mipmap selection needed */ *imgFilter = sampler->state->mag_img_filter; - *level0 = *level1 = sampler->texture->first_level; + *level0 = *level1 = (int) sampler->state->min_lod; } else { float lambda; @@ -492,7 +490,7 @@ choose_mipmap_levels(struct tgsi_sampler *sampler, if (lambda < 0.0) { /* XXX threshold depends on the filter */ /* magnifying */ *imgFilter = sampler->state->mag_img_filter; - *level0 = *level1 = sampler->texture->first_level; + *level0 = *level1 = 0; } else { /* minifying */ @@ -503,19 +501,13 @@ choose_mipmap_levels(struct tgsi_sampler *sampler, /* Nearest mipmap level */ const int lvl = (int) (lambda + 0.5); *level0 = - *level1 = CLAMP(lvl, - (int) sampler->texture->first_level, - (int) sampler->texture->last_level); + *level1 = CLAMP(lvl, 0, (int) sampler->texture->last_level); } else { /* Linear interpolation between mipmap levels */ const int lvl = (int) lambda; - *level0 = CLAMP(lvl, - (int) sampler->texture->first_level, - (int) sampler->texture->last_level); - *level1 = CLAMP(lvl + 1, - (int) sampler->texture->first_level, - (int) sampler->texture->last_level); + *level0 = CLAMP(lvl, 0, (int) sampler->texture->last_level); + *level1 = CLAMP(lvl + 1, 0, (int) sampler->texture->last_level); *levelBlend = FRAC(lambda); /* blending weight between levels */ } } diff --git a/src/mesa/pipe/softpipe/sp_texture.c b/src/mesa/pipe/softpipe/sp_texture.c index fd2cc3dbbb4..6de7a9b543b 100644 --- a/src/mesa/pipe/softpipe/sp_texture.c +++ b/src/mesa/pipe/softpipe/sp_texture.c @@ -61,7 +61,7 @@ softpipe_texture_layout(struct softpipe_texture * spt) spt->buffer_size = 0; - for ( level = pt->first_level ; level <= pt->last_level ; level++ ) { + for (level = 0; level <= pt->last_level; level++) { pt->width[level] = width; pt->height[level] = height; pt->depth[level] = depth; @@ -139,6 +139,8 @@ softpipe_get_tex_surface(struct pipe_context *pipe, struct softpipe_texture *spt = softpipe_texture(pt); struct pipe_surface *ps; + assert(level <= pt->last_level); + ps = pipe->winsys->surface_alloc(pipe->winsys); if (ps) { assert(ps->refcount); diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 7099ec33b97..f012b2f1a05 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -57,15 +57,10 @@ struct st_texture_object { struct gl_texture_object base; /* The "parent" object */ - /* The texture must include at least these levels once validated: + /* The texture must include at levels [0..lastLevel] once validated: */ - GLuint firstLevel; GLuint lastLevel; - /* Offset for firstLevel image: - */ - GLuint textureOffset; - /* On validation any active images held in main memory or in other * textures will be copied to this texture and the old storage freed. */ @@ -585,12 +580,12 @@ st_TexImage(GLcontext * ctx, _mesa_align_free(texImage->Data); } - /* If this is the only texture image in the texture, could call + /* If this is the only mipmap level in the texture, could call * bmBufferData with NULL data to free the old block and avoid * waiting on any outstanding fences. */ if (stObj->pt && - stObj->pt->first_level == level && + /*stObj->pt->first_level == level &&*/ stObj->pt->last_level == level && stObj->pt->target != PIPE_TEXTURE_CUBE && !st_texture_match_image(stObj->pt, &stImage->base, @@ -1363,13 +1358,8 @@ calculate_first_last_level(struct st_texture_object *stObj) firstLevel = lastLevel = tObj->BaseLevel; } else { - firstLevel = tObj->BaseLevel + (GLint) (tObj->MinLod + 0.5); - firstLevel = MAX2(firstLevel, tObj->BaseLevel); - lastLevel = tObj->BaseLevel + (GLint) (tObj->MaxLod + 0.5); - lastLevel = MAX2(lastLevel, tObj->BaseLevel); - lastLevel = MIN2(lastLevel, tObj->BaseLevel + baseImage->MaxLog2); - lastLevel = MIN2(lastLevel, tObj->MaxLevel); - lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */ + firstLevel = 0; + lastLevel = MIN2(tObj->MaxLevel - tObj->BaseLevel, baseImage->MaxLog2); } break; case GL_TEXTURE_RECTANGLE_NV: @@ -1380,8 +1370,6 @@ calculate_first_last_level(struct st_texture_object *stObj) return; } - /* save these values */ - stObj->firstLevel = firstLevel; stObj->lastLevel = lastLevel; } @@ -1389,15 +1377,16 @@ calculate_first_last_level(struct st_texture_object *stObj) static void copy_image_data_to_texture(struct st_context *st, struct st_texture_object *stObj, + GLuint dstLevel, struct st_texture_image *stImage) { if (stImage->pt) { /* Copy potentially with the blitter: */ st_texture_image_copy(st->pipe, - stObj->pt, /* dest texture */ - stImage->face, stImage->level, - stImage->pt /* src texture */ + stObj->pt, dstLevel, /* dest texture, level */ + stImage->pt, /* src texture */ + stImage->face ); st->pipe->texture_release(st->pipe, &stImage->pt); @@ -1438,7 +1427,7 @@ st_finalize_texture(GLcontext *ctx, const GLuint nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1; int comp_byte = 0; int cpp; - GLuint face, i; + GLuint face; struct st_texture_image *firstImage; *needFlush = GL_FALSE; @@ -1450,7 +1439,7 @@ st_finalize_texture(GLcontext *ctx, /* What levels must the texture include at a minimum? */ calculate_first_last_level(stObj); - firstImage = st_texture_image(stObj->base.Image[0][stObj->firstLevel]); + firstImage = st_texture_image(stObj->base.Image[0][stObj->base.BaseLevel]); /* Fallback case: */ @@ -1469,7 +1458,6 @@ st_finalize_texture(GLcontext *ctx, */ if (firstImage->pt && firstImage->pt != stObj->pt && - firstImage->pt->first_level <= stObj->firstLevel && firstImage->pt->last_level >= stObj->lastLevel) { if (stObj->pt) @@ -1488,18 +1476,11 @@ st_finalize_texture(GLcontext *ctx, /* Check texture can hold all active levels. Check texture matches * target, imageFormat, etc. - * - * XXX: For some layouts (eg i945?), the test might have to be - * first_level == firstLevel, as the texture isn't valid except at the - * original start level. Hope to get around this by - * programming minLod, maxLod, baseLevel into the hardware and - * leaving the texture alone. */ if (stObj->pt && (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) || stObj->pt->format != st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat) || - stObj->pt->first_level != stObj->firstLevel || stObj->pt->last_level != stObj->lastLevel || stObj->pt->width[0] != firstImage->base.Width || stObj->pt->height[0] != firstImage->base.Height || @@ -1516,7 +1497,7 @@ st_finalize_texture(GLcontext *ctx, stObj->pt = st_texture_create(ctx->st, gl_target_to_pipe(stObj->base.Target), st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat), - stObj->firstLevel, + 0, /* first level */ stObj->lastLevel, firstImage->base.Width, firstImage->base.Height, @@ -1527,14 +1508,16 @@ st_finalize_texture(GLcontext *ctx, /* Pull in any images not in the object's texture: */ for (face = 0; face < nr_faces; face++) { - for (i = stObj->firstLevel; i <= stObj->lastLevel; i++) { + GLuint level; + for (level = 0; level <= stObj->lastLevel; level++) { struct st_texture_image *stImage = - st_texture_image(stObj->base.Image[face][i]); + //st_texture_image(stObj->base.Image[face][level]); + st_texture_image(stObj->base.Image[face][stObj->base.BaseLevel + level]); /* Need to import images in main memory or held in other textures. */ if (stObj->pt != stImage->pt) { - copy_image_data_to_texture(ctx->st, stObj, stImage); + copy_image_data_to_texture(ctx->st, stObj, level, stImage); *needFlush = GL_TRUE; } } diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index c152c599056..fd7d8cefea9 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -51,7 +51,6 @@ static void *blend_cso = NULL; static void *depthstencil_cso = NULL; static void *rasterizer_cso = NULL; -static void *sampler_cso = NULL; static struct st_fragment_program *stfp = NULL; static struct st_vertex_program *stvp = NULL; @@ -118,7 +117,6 @@ st_init_generate_mipmap(struct st_context *st) struct pipe_context *pipe = st->pipe; struct pipe_blend_state blend; struct pipe_rasterizer_state rasterizer; - struct pipe_sampler_state sampler; struct pipe_depth_stencil_alpha_state depthstencil; assert(!blend_cso); @@ -133,16 +131,6 @@ st_init_generate_mipmap(struct st_context *st) memset(&rasterizer, 0, sizeof(rasterizer)); rasterizer_cso = pipe->create_rasterizer_state(pipe, &rasterizer); - memset(&sampler, 0, sizeof(sampler)); - sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; - sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; - sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; - sampler.normalized_coords = 1; - sampler_cso = pipe->create_sampler_state(pipe, &sampler); - stfp = make_tex_fragment_program(st->ctx); stvp = st_make_passthrough_vertex_shader(st, GL_FALSE); } @@ -156,14 +144,12 @@ st_destroy_generate_mipmpap(struct st_context *st) pipe->delete_blend_state(pipe, blend_cso); pipe->delete_depth_stencil_alpha_state(pipe, depthstencil_cso); pipe->delete_rasterizer_state(pipe, rasterizer_cso); - pipe->delete_sampler_state(pipe, sampler_cso); /* XXX free stfp, stvp */ blend_cso = NULL; depthstencil_cso = NULL; rasterizer_cso = NULL; - sampler_cso = NULL; } @@ -248,8 +234,10 @@ st_render_mipmap(struct st_context *st, { struct pipe_context *pipe = st->pipe; struct pipe_framebuffer_state fb; + struct pipe_sampler_state sampler; + void *sampler_cso; const uint face = _mesa_tex_target_to_face(target), zslice = 0; - const uint first_level_save = pt->first_level; + /*const uint first_level_save = pt->first_level;*/ uint dstLevel; assert(target != GL_TEXTURE_3D); /* not done yet */ @@ -263,11 +251,21 @@ st_render_mipmap(struct st_context *st, memset(&fb, 0, sizeof(fb)); fb.num_cbufs = 1; + /* sampler state */ + memset(&sampler, 0, sizeof(sampler)); + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.normalized_coords = 1; + + /* bind CSOs */ pipe->bind_blend_state(pipe, blend_cso); pipe->bind_depth_stencil_alpha_state(pipe, depthstencil_cso); pipe->bind_rasterizer_state(pipe, rasterizer_cso); - pipe->bind_sampler_state(pipe, 0, sampler_cso); /* bind shaders */ pipe->bind_fs_state(pipe, stfp->fs->data); @@ -286,20 +284,29 @@ st_render_mipmap(struct st_context *st, fb.cbufs[0] = pipe->get_tex_surface(pipe, pt, face, dstLevel, zslice); pipe->set_framebuffer_state(pipe, &fb); + /* + * Setup sampler state + */ + sampler.min_lod = sampler.max_lod = srcLevel; + sampler_cso = pipe->create_sampler_state(pipe, &sampler); + pipe->bind_sampler_state(pipe, 0, sampler_cso); + simple_viewport(pipe, pt->width[dstLevel], pt->height[dstLevel]); /* * Setup src texture, override pt->first_level so we sample from * the right mipmap level. */ - pt->first_level = srcLevel; + /*pt->first_level = srcLevel;*/ pipe->set_sampler_texture(pipe, 0, pt); draw_quad(st->ctx); + + pipe->delete_sampler_state(pipe, sampler_cso); } /* restore first_level */ - pt->first_level = first_level_save; + /*pt->first_level = first_level_save;*/ /* restore pipe state */ if (st->state.rasterizer) diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 844a9f80d85..ff33da6f9e4 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -62,6 +62,10 @@ target_to_target(GLenum target) /** * Allocate a new pipe_texture object + * width0, height0, depth0 are the dimensions of the level 0 image + * (the highest resolution). last_level indicates how many mipmap levels + * to allocate storage for. For non-mipmapped textures, this will be zero. + * XXX first_level obsolete */ struct pipe_texture * st_texture_create(struct st_context *st, @@ -84,9 +88,9 @@ st_texture_create(struct st_context *st, assert(format); + memset(&pt, 0, sizeof(pt)); pt.target = target; pt.format = format; - pt.first_level = first_level; pt.last_level = last_level; pt.width[0] = width0; pt.height[0] = height0; @@ -266,23 +270,36 @@ st_texture_image_data(struct pipe_context *pipe, */ void st_texture_image_copy(struct pipe_context *pipe, - struct pipe_texture *dst, - GLuint face, GLuint level, - struct pipe_texture *src) + struct pipe_texture *dst, GLuint dstLevel, + struct pipe_texture *src, + GLuint face) { - GLuint width = src->width[level]; - GLuint height = src->height[level]; - GLuint depth = src->depth[level]; + GLuint width = dst->width[dstLevel]; + GLuint height = dst->height[dstLevel]; + GLuint depth = dst->depth[dstLevel]; struct pipe_surface *src_surface; struct pipe_surface *dst_surface; GLuint i; + /* XXX this is a hack */ if (dst->compressed) height /= 4; for (i = 0; i < depth; i++) { - dst_surface = pipe->get_tex_surface(pipe, dst, face, level, i); - src_surface = pipe->get_tex_surface(pipe, src, face, level, i); + GLuint srcLevel; + + /* find src texture level of needed size */ + for (srcLevel = 0; srcLevel <= src->last_level; srcLevel++) { + if (src->width[srcLevel] == width && + src->height[srcLevel] == height) { + break; + } + } + assert(src->width[srcLevel] == width); + assert(src->height[srcLevel] == height); + + dst_surface = pipe->get_tex_surface(pipe, dst, face, dstLevel, i); + src_surface = pipe->get_tex_surface(pipe, src, face, srcLevel, i); pipe->surface_copy(pipe, dst_surface, diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index 0b87a494c38..6c5f0930fa5 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -98,9 +98,9 @@ st_texture_image_data(struct pipe_context *pipe, */ extern void st_texture_image_copy(struct pipe_context *pipe, - struct pipe_texture *dst, - GLuint face, GLuint level, - struct pipe_texture *src); + struct pipe_texture *dst, GLuint dstLevel, + struct pipe_texture *src, + GLuint face); #endif -- cgit v1.2.3 From 7d99bac7d6e905b8851216f7d74a583e0f087e1b Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 14:55:51 -0700 Subject: gallium: remove unused first_level param from st_texture_create() --- src/mesa/state_tracker/st_cb_drawpixels.c | 6 +++--- src/mesa/state_tracker/st_cb_texture.c | 2 -- src/mesa/state_tracker/st_texture.c | 6 ++---- src/mesa/state_tracker/st_texture.h | 1 - 4 files changed, 5 insertions(+), 10 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 07886e79829..475e23653e8 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -471,7 +471,7 @@ make_texture(struct st_context *st, assert(pipeFormat); cpp = st_sizeof_format(pipeFormat); - pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, 0, width, height, + pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, width, height, 1, 0); if (!pt) return NULL; @@ -1017,7 +1017,7 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, /** * Create a texture. */ - pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0, 0, width, height, + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0, width, height, 1, 0); if (!pt) return NULL; @@ -1241,7 +1241,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, psRead = rbRead->surface; format = psRead->format; - pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0, 0, width, height, + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, format, 0, width, height, 1, 0); if (!pt) return; diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index f012b2f1a05..fab9eafc7f0 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -363,7 +363,6 @@ guess_and_alloc_texture(struct st_context *st, stObj->pt = st_texture_create(st, gl_target_to_pipe(stObj->base.Target), st_mesa_format_to_pipe_format(stImage->base.TexFormat->MesaFormat), - firstLevel, lastLevel, width, height, @@ -1497,7 +1496,6 @@ st_finalize_texture(GLcontext *ctx, stObj->pt = st_texture_create(ctx->st, gl_target_to_pipe(stObj->base.Target), st_mesa_format_to_pipe_format(firstImage->base.TexFormat->MesaFormat), - 0, /* first level */ stObj->lastLevel, firstImage->base.Width, firstImage->base.Height, diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index ff33da6f9e4..2622d009530 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -65,13 +65,11 @@ target_to_target(GLenum target) * width0, height0, depth0 are the dimensions of the level 0 image * (the highest resolution). last_level indicates how many mipmap levels * to allocate storage for. For non-mipmapped textures, this will be zero. - * XXX first_level obsolete */ struct pipe_texture * st_texture_create(struct st_context *st, enum pipe_texture_target target, enum pipe_format format, - GLuint first_level, GLuint last_level, GLuint width0, GLuint height0, @@ -82,9 +80,9 @@ st_texture_create(struct st_context *st, assert(target <= PIPE_TEXTURE_CUBE); - DBG("%s target %s format %s level %d..%d\n", __FUNCTION__, + DBG("%s target %s format %s last_level %d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), - _mesa_lookup_enum_by_nr(format), first_level, last_level); + _mesa_lookup_enum_by_nr(format), last_level); assert(format); diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index 6c5f0930fa5..72324cd9ab6 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -39,7 +39,6 @@ extern struct pipe_texture * st_texture_create(struct st_context *st, enum pipe_texture_target target, enum pipe_format format, - GLuint first_level, GLuint last_level, GLuint width0, GLuint height0, -- cgit v1.2.3 From 7eb2cd3427940c914d2bbc0c1e901b5c81ff50d5 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 14:59:07 -0700 Subject: tweak incr/decrements amounts for keyboard options --- progs/tests/mipmap_limits.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/progs/tests/mipmap_limits.c b/progs/tests/mipmap_limits.c index dc066cab1f0..b0a3be99ff3 100644 --- a/progs/tests/mipmap_limits.c +++ b/progs/tests/mipmap_limits.c @@ -194,22 +194,22 @@ key(unsigned char k, int x, int y) MaxLevel = 10; break; case 'l': - LodBias -= 0.02; + LodBias -= 0.25; break; case 'L': - LodBias += 0.02; + LodBias += 0.25; break; case 'n': - MinLod -= 0.02; + MinLod -= 0.25; break; case 'N': - MinLod += 0.02; + MinLod += 0.25; break; case 'x': - MaxLod -= 0.02; + MaxLod -= 0.25; break; case 'X': - MaxLod += 0.02; + MaxLod += 0.25; break; case 'f': NearestFilter = !NearestFilter; -- cgit v1.2.3 From 5b80529b6081bcff42ec20a096506c441729d39c Mon Sep 17 00:00:00 2001 From: Brian Date: Mon, 11 Feb 2008 09:46:10 -0700 Subject: press 0,1,2,etc keys for specific bias values --- progs/demos/lodbias.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/progs/demos/lodbias.c b/progs/demos/lodbias.c index a4db22e26e4..c5a2a1b4573 100644 --- a/progs/demos/lodbias.c +++ b/progs/demos/lodbias.c @@ -159,6 +159,18 @@ static void Key( unsigned char key, int x, int y ) case 'B': Bias += 10; break; + case '0': + case '1': + case '2': + case '3': + case '4': + case '5': + case '6': + case '7': + case '8': + case '9': + Bias = 100.0 * (key - '0'); + break; case 27: exit(0); break; -- cgit v1.2.3 From 7057a031f196f677366d0d397951aa87932ac887 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 15:53:37 -0700 Subject: Set Min/MaxLod with glTexParameterf, not glTexParameteri --- progs/tests/mipmap_limits.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/progs/tests/mipmap_limits.c b/progs/tests/mipmap_limits.c index b0a3be99ff3..d6d6e467b0f 100644 --- a/progs/tests/mipmap_limits.c +++ b/progs/tests/mipmap_limits.c @@ -131,8 +131,8 @@ static void display(void) glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, BaseLevel); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, MaxLevel); - glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_LOD, MinLod); - glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LOD, MaxLod); + glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_LOD, MinLod); + glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAX_LOD, MaxLod); if (NearestFilter) { glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); -- cgit v1.2.3 From aa31fe3b54592b1c017e0389de990040f2ad18f4 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 16:06:51 -0700 Subject: gallium: move gen-mipmap global/static vars into st_context This fixes potential problems with multi-context programs. --- src/mesa/state_tracker/st_context.h | 9 ++++++++ src/mesa/state_tracker/st_gen_mipmap.c | 41 +++++++++++----------------------- 2 files changed, 22 insertions(+), 28 deletions(-) diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index a756055898d..59d1590f053 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -166,6 +166,15 @@ struct st_context struct st_fragment_program *combined_prog; } bitmap; + /** For gen/render mipmap feature */ + struct { + void *blend_cso; + void *depthstencil_cso; + void *rasterizer_cso; + struct st_fragment_program *stfp; + struct st_vertex_program *stvp; + } gen_mipmap; + struct cso_cache *cache; }; diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index fd7d8cefea9..b4a21fd7e27 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -48,15 +48,6 @@ -static void *blend_cso = NULL; -static void *depthstencil_cso = NULL; -static void *rasterizer_cso = NULL; - -static struct st_fragment_program *stfp = NULL; -static struct st_vertex_program *stvp = NULL; - - - static struct st_fragment_program * make_tex_fragment_program(GLcontext *ctx) { @@ -119,20 +110,18 @@ st_init_generate_mipmap(struct st_context *st) struct pipe_rasterizer_state rasterizer; struct pipe_depth_stencil_alpha_state depthstencil; - assert(!blend_cso); - memset(&blend, 0, sizeof(blend)); blend.colormask = PIPE_MASK_RGBA; - blend_cso = pipe->create_blend_state(pipe, &blend); + st->gen_mipmap.blend_cso = pipe->create_blend_state(pipe, &blend); memset(&depthstencil, 0, sizeof(depthstencil)); - depthstencil_cso = pipe->create_depth_stencil_alpha_state(pipe, &depthstencil); + st->gen_mipmap.depthstencil_cso = pipe->create_depth_stencil_alpha_state(pipe, &depthstencil); memset(&rasterizer, 0, sizeof(rasterizer)); - rasterizer_cso = pipe->create_rasterizer_state(pipe, &rasterizer); + st->gen_mipmap.rasterizer_cso = pipe->create_rasterizer_state(pipe, &rasterizer); - stfp = make_tex_fragment_program(st->ctx); - stvp = st_make_passthrough_vertex_shader(st, GL_FALSE); + st->gen_mipmap.stfp = make_tex_fragment_program(st->ctx); + st->gen_mipmap.stvp = st_make_passthrough_vertex_shader(st, GL_FALSE); } @@ -141,15 +130,11 @@ st_destroy_generate_mipmpap(struct st_context *st) { struct pipe_context *pipe = st->pipe; - pipe->delete_blend_state(pipe, blend_cso); - pipe->delete_depth_stencil_alpha_state(pipe, depthstencil_cso); - pipe->delete_rasterizer_state(pipe, rasterizer_cso); + pipe->delete_blend_state(pipe, st->gen_mipmap.blend_cso); + pipe->delete_depth_stencil_alpha_state(pipe, st->gen_mipmap.depthstencil_cso); + pipe->delete_rasterizer_state(pipe, st->gen_mipmap.rasterizer_cso); /* XXX free stfp, stvp */ - - blend_cso = NULL; - depthstencil_cso = NULL; - rasterizer_cso = NULL; } @@ -263,13 +248,13 @@ st_render_mipmap(struct st_context *st, /* bind CSOs */ - pipe->bind_blend_state(pipe, blend_cso); - pipe->bind_depth_stencil_alpha_state(pipe, depthstencil_cso); - pipe->bind_rasterizer_state(pipe, rasterizer_cso); + pipe->bind_blend_state(pipe, st->gen_mipmap.blend_cso); + pipe->bind_depth_stencil_alpha_state(pipe, st->gen_mipmap.depthstencil_cso); + pipe->bind_rasterizer_state(pipe, st->gen_mipmap.rasterizer_cso); /* bind shaders */ - pipe->bind_fs_state(pipe, stfp->fs->data); - pipe->bind_vs_state(pipe, stvp->cso->data); + pipe->bind_fs_state(pipe, st->gen_mipmap.stfp->fs->data); + pipe->bind_vs_state(pipe, st->gen_mipmap.stvp->cso->data); /* * XXX for small mipmap levels, it may be faster to use the software -- cgit v1.2.3 From bbdbdaddb0b476ec347c100e20469b0c52c5d525 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 16:10:11 -0700 Subject: gallium: rename st_fragment_program's fs field to cso to match st_vertex_program --- src/mesa/state_tracker/st_atom_shader.c | 8 ++++---- src/mesa/state_tracker/st_cb_clear.c | 2 +- src/mesa/state_tracker/st_cb_drawpixels.c | 2 +- src/mesa/state_tracker/st_cb_program.c | 4 ++-- src/mesa/state_tracker/st_gen_mipmap.c | 2 +- src/mesa/state_tracker/st_program.c | 2 +- src/mesa/state_tracker/st_program.h | 2 +- 7 files changed, 11 insertions(+), 11 deletions(-) diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c index 91969185093..2c6ec8421b0 100644 --- a/src/mesa/state_tracker/st_atom_shader.c +++ b/src/mesa/state_tracker/st_atom_shader.c @@ -158,7 +158,7 @@ find_translated_vp(struct st_context *st, /* * Translate fragment program if needed. */ - if (!stfp->fs) { + if (!stfp->cso) { GLuint inAttr, numIn = 0; for (inAttr = 0; inAttr < FRAG_ATTRIB_MAX; inAttr++) { @@ -179,7 +179,7 @@ find_translated_vp(struct st_context *st, stfp->input_to_slot, stfp->tokens, ST_MAX_SHADER_TOKENS); - assert(stfp->fs); + assert(stfp->cso); } @@ -227,7 +227,7 @@ find_translated_vp(struct st_context *st, if (fpInAttrib >= 0) { GLuint fpInSlot = stfp->input_to_slot[fpInAttrib]; if (fpInSlot != ~0) { - GLuint vpOutSlot = stfp->fs->state.input_map[fpInSlot]; + GLuint vpOutSlot = stfp->cso->state.input_map[fpInSlot]; xvp->output_to_slot[outAttr] = vpOutSlot; numVpOuts++; } @@ -300,7 +300,7 @@ update_linkage( struct st_context *st ) st->pipe->bind_vs_state(st->pipe, st->state.vs->cso->data); st->fp = stfp; - st->state.fs = stfp->fs; + st->state.fs = stfp->cso; st->pipe->bind_fs_state(st->pipe, st->state.fs->data); st->vertex_result_to_slot = xvp->output_to_slot; diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index ab98b54bab4..410062e1e80 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -341,7 +341,7 @@ clear_with_quad(GLcontext *ctx, if (!stfp) { stfp = make_frag_shader(st); } - pipe->bind_fs_state(pipe, stfp->fs->data); + pipe->bind_fs_state(pipe, stfp->cso->data); } /* vertex shader state: color/position pass-through */ diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 475e23653e8..3245a7488bc 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -665,7 +665,7 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, } /* fragment shader state: TEX lookup program */ - pipe->bind_fs_state(pipe, stfp->fs->data); + pipe->bind_fs_state(pipe, stfp->cso->data); /* vertex shader state: position + texcoord pass-through */ pipe->bind_vs_state(pipe, stvp->cso->data); diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c index f1f33fb0dd8..af3ee655048 100644 --- a/src/mesa/state_tracker/st_cb_program.c +++ b/src/mesa/state_tracker/st_cb_program.c @@ -168,10 +168,10 @@ static void st_program_string_notify( GLcontext *ctx, stfp->serialNo++; - if (stfp->fs) { + if (stfp->cso) { /* free the TGSI code */ // cso_delete(stfp->vs); - stfp->fs = NULL; + stfp->cso = NULL; } stfp->param_state = stfp->Base.Base.Parameters->StateFlags; diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index b4a21fd7e27..459941cca87 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -253,7 +253,7 @@ st_render_mipmap(struct st_context *st, pipe->bind_rasterizer_state(pipe, st->gen_mipmap.rasterizer_cso); /* bind shaders */ - pipe->bind_fs_state(pipe, st->gen_mipmap.stfp->fs->data); + pipe->bind_fs_state(pipe, st->gen_mipmap.stfp->cso->data); pipe->bind_vs_state(pipe, st->gen_mipmap.stvp->cso->data); /* diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index 84a90940018..c8297badedd 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -420,7 +420,7 @@ st_translate_fragment_program(struct st_context *st, fs.tokens = tokensOut; cso = st_cached_fs_state(st, &fs); - stfp->fs = cso; + stfp->cso = cso; if (0) _mesa_print_program(&stfp->Base.Base); diff --git a/src/mesa/state_tracker/st_program.h b/src/mesa/state_tracker/st_program.h index de02c3185fd..ea1dde4a7a3 100644 --- a/src/mesa/state_tracker/st_program.h +++ b/src/mesa/state_tracker/st_program.h @@ -62,7 +62,7 @@ struct st_fragment_program struct tgsi_token tokens[ST_MAX_SHADER_TOKENS]; /** Pointer to the corresponding cached shader */ - const struct cso_fragment_shader *fs; + const struct cso_fragment_shader *cso; GLuint param_state; -- cgit v1.2.3 From c0a22da1570b104fb6d2ee5e620906b01d194165 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 16:37:49 -0700 Subject: gallium: added null ptr check --- src/mesa/state_tracker/st_cb_texture.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index fab9eafc7f0..c6fe928d471 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -1514,7 +1514,7 @@ st_finalize_texture(GLcontext *ctx, /* Need to import images in main memory or held in other textures. */ - if (stObj->pt != stImage->pt) { + if (stImage && stObj->pt != stImage->pt) { copy_image_data_to_texture(ctx->st, stObj, level, stImage); *needFlush = GL_TRUE; } -- cgit v1.2.3 From dee8e268f77d31e78fd76005d529ea3b61e41209 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 16:38:56 -0700 Subject: gallium: remove dead code --- src/mesa/state_tracker/st_cb_texture.c | 1 - 1 file changed, 1 deletion(-) diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index c6fe928d471..0ea367549b8 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -1509,7 +1509,6 @@ st_finalize_texture(GLcontext *ctx, GLuint level; for (level = 0; level <= stObj->lastLevel; level++) { struct st_texture_image *stImage = - //st_texture_image(stObj->base.Image[face][level]); st_texture_image(stObj->base.Image[face][stObj->base.BaseLevel + level]); /* Need to import images in main memory or held in other textures. -- cgit v1.2.3 From 2ef9df660c0ee06aa0ea13663d0706cc03fecbb7 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 16:56:18 -0700 Subject: Fix broken test. As-is, if the texture was too large for the target, an assertion would fail. Now check proxy texture first and if it works, then test non-proxy target. --- progs/tests/arbnpot.c | 57 ++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 40 insertions(+), 17 deletions(-) diff --git a/progs/tests/arbnpot.c b/progs/tests/arbnpot.c index 8107717e271..05ba85dad9e 100644 --- a/progs/tests/arbnpot.c +++ b/progs/tests/arbnpot.c @@ -113,44 +113,67 @@ static void Init( void ) minDim = imgWidth < imgHeight ? imgWidth : imgHeight; glPixelStorei(GL_UNPACK_ALIGNMENT, 1); - glTexImage1D(GL_TEXTURE_1D, 0, GL_RGB, imgWidth, 0, - imgFormat, GL_UNSIGNED_BYTE, image); - assert(glGetError() == GL_NO_ERROR); + /* + * 1D Texture. Test proxy first, if that works, test non-proxy target. + */ glTexImage1D(GL_PROXY_TEXTURE_1D, 0, GL_RGB, imgWidth, 0, imgFormat, GL_UNSIGNED_BYTE, image); glGetTexLevelParameteriv(GL_PROXY_TEXTURE_1D, 0, GL_TEXTURE_WIDTH, &w); - assert(w == imgWidth); + assert(w == imgWidth || w == 0); - glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, imgWidth, imgHeight, 0, - imgFormat, GL_UNSIGNED_BYTE, image); - assert(glGetError() == GL_NO_ERROR); + if (w) { + glTexImage1D(GL_TEXTURE_1D, 0, GL_RGB, imgWidth, 0, + imgFormat, GL_UNSIGNED_BYTE, image); + assert(glGetError() == GL_NO_ERROR); + } + + /* + * 2D Texture + */ glTexImage2D(GL_PROXY_TEXTURE_2D, 0, GL_RGB, imgWidth, imgHeight, 0, imgFormat, GL_UNSIGNED_BYTE, image); glGetTexLevelParameteriv(GL_PROXY_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &w); - assert(w == imgWidth); + assert(w == imgWidth || w == 0); - glTexImage3D(GL_TEXTURE_3D, 0, GL_RGB, imgWidth, imgHeight, 1, 0, - imgFormat, GL_UNSIGNED_BYTE, image); - assert(glGetError() == GL_NO_ERROR); + if (w) { + glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, imgWidth, imgHeight, 0, + imgFormat, GL_UNSIGNED_BYTE, image); + assert(glGetError() == GL_NO_ERROR); + } + + /* + * 3D Texture + */ glTexImage3D(GL_PROXY_TEXTURE_3D, 0, GL_RGB, imgWidth, imgHeight, 1, 0, imgFormat, GL_UNSIGNED_BYTE, image); glGetTexLevelParameteriv(GL_PROXY_TEXTURE_3D, 0, GL_TEXTURE_WIDTH, &w); - assert(w == imgWidth); + assert(w == imgWidth || w == 0); + + if (w) { + glTexImage3D(GL_TEXTURE_3D, 0, GL_RGB, imgWidth, imgHeight, 1, 0, + imgFormat, GL_UNSIGNED_BYTE, image); + assert(glGetError() == GL_NO_ERROR); + } - glTexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X, 0, GL_RGB, - minDim, minDim, 0, - imgFormat, GL_UNSIGNED_BYTE, image); - assert(glGetError() == GL_NO_ERROR); + /* + * Cube Texture + */ glTexImage2D(GL_PROXY_TEXTURE_CUBE_MAP, 0, GL_RGB, minDim, minDim, 0, imgFormat, GL_UNSIGNED_BYTE, image); glGetTexLevelParameteriv(GL_PROXY_TEXTURE_CUBE_MAP, 0, GL_TEXTURE_WIDTH, &w); - assert(w == minDim); + assert(w == minDim || w == 0); + if (w) { + glTexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X, 0, GL_RGB, + minDim, minDim, 0, + imgFormat, GL_UNSIGNED_BYTE, image); + assert(glGetError() == GL_NO_ERROR); + } glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); -- cgit v1.2.3 From b487e4b4fb3d9304feedf910a2f137703a0e260b Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 17:07:27 -0700 Subject: gallium: clamp min_lod so it's never negative --- src/mesa/state_tracker/st_atom_sampler.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/state_tracker/st_atom_sampler.c b/src/mesa/state_tracker/st_atom_sampler.c index 6241e70b550..92263cb6881 100644 --- a/src/mesa/state_tracker/st_atom_sampler.c +++ b/src/mesa/state_tracker/st_atom_sampler.c @@ -147,7 +147,7 @@ update_samplers(struct st_context *st) sampler.lod_bias = st->ctx->Texture.Unit[su].LodBias; #if 1 - sampler.min_lod = texobj->MinLod; + sampler.min_lod = (texobj->MinLod) < 0.0 ? 0.0 : texobj->MinLod; sampler.max_lod = texobj->MaxLod; #else /* min/max lod should really be as follows (untested). -- cgit v1.2.3 From 4ac85794b181fdc44f1d9727926c89c084ebb769 Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 17:44:57 -0700 Subject: gallium: fix/finish glReadPixels(GL_DEPTH_COMPONENT). This fixes demos/reflect ('d' key) and tests/zreaddraw.c --- src/mesa/state_tracker/st_cb_readpixels.c | 99 ++++++++++++++++++++++++------- 1 file changed, 78 insertions(+), 21 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index a1bbb3a8312..868c5f3c5f9 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -180,20 +180,13 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, if (!strb) return; + if (format == GL_RGBA && type == GL_FLOAT) { /* write tile(row) directly into user's buffer */ df = (GLfloat *) _mesa_image_address2d(&clippedPacking, dest, width, height, format, type, 0, 0); dfStride = width * 4; } -#if 0 - else if (format == GL_DEPTH_COMPONENT && type == GL_FLOAT) { - /* write tile(row) directly into user's buffer */ - df = (GLfloat *) _mesa_image_address2d(&clippedPacking, dest, width, - height, format, type, 0, 0); - dfStride = width; - } -#endif else { /* write tile(row) into temp row buffer */ df = (GLfloat *) temp; @@ -209,22 +202,86 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, yStep = 1; } - /* Do a row at a time to flip image data vertically */ - for (i = 0; i < height; i++) { - pipe_get_tile_rgba(pipe, strb->surface, x, y, width, 1, df); - y += yStep; - df += dfStride; - if (!dfStride) { - /* convert GLfloat to user's format/type */ - GLvoid *dst = _mesa_image_address2d(&clippedPacking, dest, width, - height, format, type, i, 0); + /* + * Copy pixels from pipe_surface to user memory + */ + { + /* dest of first pixel in client memory */ + GLubyte *dst = _mesa_image_address2d(&clippedPacking, dest, width, + height, format, type, 0, 0); + /* dest row stride */ + const GLint dstStride = _mesa_image_row_stride(&clippedPacking, width, + format, type); + + if (strb->surface->format == PIPE_FORMAT_S8Z24_UNORM) { if (format == GL_DEPTH_COMPONENT) { - _mesa_pack_depth_span(ctx, width, dst, type, - (GLfloat *) temp, &clippedPacking); + for (i = 0; i < height; i++) { + GLuint ztemp[MAX_WIDTH], j; + GLfloat zfloat[MAX_WIDTH]; + const double scale = 1.0 / ((1 << 24) - 1); + pipe_get_tile_raw(pipe, strb->surface, x, y, + width, 1, ztemp, 0); + y += yStep; + for (j = 0; j < width; j++) { + zfloat[j] = (float) (scale * (ztemp[j] & 0xffffff)); + } + _mesa_pack_depth_span(ctx, width, dst, type, + zfloat, &clippedPacking); + dst += dstStride; + } } else { - _mesa_pack_rgba_span_float(ctx, width, temp, format, type, dst, - &clippedPacking, transferOps); + /* untested, but simple: */ + assert(format == GL_DEPTH_STENCIL_EXT); + for (i = 0; i < height; i++) { + pipe_get_tile_raw(pipe, strb->surface, x, y, width, 1, dst, 0); + y += yStep; + dst += dstStride; + } + } + } + else if (strb->surface->format == PIPE_FORMAT_Z16_UNORM) { + for (i = 0; i < height; i++) { + GLshort ztemp[MAX_WIDTH], j; + GLfloat zfloat[MAX_WIDTH]; + const double scale = 1.0 / 0xffff; + pipe_get_tile_raw(pipe, strb->surface, x, y, width, 1, ztemp, 0); + y += yStep; + for (j = 0; j < width; j++) { + zfloat[j] = (float) (scale * ztemp[j]); + } + _mesa_pack_depth_span(ctx, width, dst, type, + zfloat, &clippedPacking); + dst += dstStride; + } + } + else if (strb->surface->format == PIPE_FORMAT_Z32_UNORM) { + for (i = 0; i < height; i++) { + GLuint ztemp[MAX_WIDTH], j; + GLfloat zfloat[MAX_WIDTH]; + const double scale = 1.0 / 0xffffffff; + pipe_get_tile_raw(pipe, strb->surface, x, y, width, 1, ztemp, 0); + y += yStep; + for (j = 0; j < width; j++) { + zfloat[j] = (float) (scale * ztemp[j]); + } + _mesa_pack_depth_span(ctx, width, dst, type, + zfloat, &clippedPacking); + dst += dstStride; + } + } + else { + /* RGBA format */ + /* Do a row at a time to flip image data vertically */ + for (i = 0; i < height; i++) { + pipe_get_tile_rgba(pipe, strb->surface, x, y, width, 1, df); + y += yStep; + df += dfStride; + if (!dfStride) { + _mesa_pack_rgba_span_float(ctx, width, temp, format, type, dst, + &clippedPacking, transferOps); + dst += dstStride; + } } } } -- cgit v1.2.3 From 67a483909f8999de1e0c40229b94f7dabab7403b Mon Sep 17 00:00:00 2001 From: Brian Date: Tue, 12 Feb 2008 19:14:58 -0700 Subject: gallium: initial code for wide/AA line drawing --- src/mesa/pipe/draw/draw_wide_prims.c | 75 +++++++++++++++++++++++++++++++++--- 1 file changed, 69 insertions(+), 6 deletions(-) diff --git a/src/mesa/pipe/draw/draw_wide_prims.c b/src/mesa/pipe/draw/draw_wide_prims.c index 9759e7e2e82..4c7e279b20f 100644 --- a/src/mesa/pipe/draw/draw_wide_prims.c +++ b/src/mesa/pipe/draw/draw_wide_prims.c @@ -76,7 +76,6 @@ static void passthrough_tri( struct draw_stage *stage, /** * Draw a wide line by drawing a quad (two triangles). - * XXX still need line stipple. * XXX need to disable polygon stipple. */ static void wide_line( struct draw_stage *stage, @@ -103,12 +102,9 @@ static void wide_line( struct draw_stage *stage, /* * Draw wide line as a quad (two tris) by "stretching" the line along * X or Y. - * XXX For AA lines, the quad corners have to be computed in a - * more sophisticated way. + * We need to tweak coords in several ways to be conformant here. */ - /* need to tweak coords in several ways to be conformant here */ - if (dx > dy) { /* x-major line */ pos0[1] = pos0[1] - half_width - 0.25f; @@ -165,6 +161,70 @@ static void wide_line( struct draw_stage *stage, } +/** + * Draw a wide line by drawing a quad, using geometry which will + * fullfill GL's antialiased line requirements. + */ +static void wide_line_aa(struct draw_stage *stage, + struct prim_header *header) +{ + const struct wide_stage *wide = wide_stage(stage); + const float half_width = wide->half_line_width; + struct prim_header tri; + struct vertex_header *v[4]; + float *pos; + float dx = header->v[1]->data[0][0] - header->v[0]->data[0][0]; + float dy = header->v[1]->data[0][1] - header->v[0]->data[0][1]; + const float len = sqrt(dx * dx + dy * dy); + uint i; + + dx = dx * half_width / len; + dy = dy * half_width / len; + + /* allocate/dup new verts */ + for (i = 0; i < 4; i++) { + v[i] = dup_vert(stage, header->v[i/2], i); + } + + /* + * Quad for line from v0 to v1: + * + * 1 3 + * +-------------------------+ + * | | + * *v0 v1* + * | | + * +-------------------------+ + * 0 2 + */ + + pos = v[0]->data[0]; + pos[0] += dy; + pos[1] -= dx; + + pos = v[1]->data[0]; + pos[0] -= dy; + pos[1] += dx; + + pos = v[2]->data[0]; + pos[0] += dy; + pos[1] -= dx; + + pos = v[3]->data[0]; + pos[0] -= dy; + pos[1] += dx; + + tri.det = header->det; /* only the sign matters */ + + tri.v[0] = v[2]; tri.v[1] = v[1]; tri.v[2] = v[0]; + stage->next->tri( stage->next, &tri ); + + tri.v[0] = v[3]; tri.v[1] = v[1]; tri.v[2] = v[2]; + stage->next->tri( stage->next, &tri ); + +} + + /** * Set the vertex texcoords for sprite mode. * Coords may be left untouched or set to a right-side-up or upside-down @@ -319,7 +379,10 @@ static void wide_first_line( struct draw_stage *stage, wide->half_line_width = 0.5f * draw->rasterizer->line_width; if (draw->rasterizer->line_width != 1.0) { - wide->stage.line = wide_line; + if (draw->rasterizer->line_smooth) + wide->stage.line = wide_line_aa; + else + wide->stage.line = wide_line; } else { wide->stage.line = passthrough_line; -- cgit v1.2.3 From 7bd15fd271ecaad0446632dd4e2190930abac487 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 13 Feb 2008 14:27:32 +1100 Subject: nv40: remove use of pt->first_level --- src/mesa/pipe/nv40/nv40_fragtex.c | 3 +-- src/mesa/pipe/nv40/nv40_miptree.c | 6 +++--- 2 files changed, 4 insertions(+), 5 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index d278ce18975..69c0fcb5e05 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -70,8 +70,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txf = ps->fmt; txf |= tf->format | 0x8000; - txf |= ((pt->last_level - pt->first_level + 1) << - NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); + txf |= ((pt->last_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); if (1) /* XXX */ txf |= NV40TCL_TEX_FORMAT_NO_BORDER; diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index df70feaa057..1b3c27dc45b 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -23,7 +23,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) nr_faces = 1; } - for (l = pt->first_level; l <= pt->last_level; l++) { + for (l = 0; l <= pt->last_level; l++) { pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; @@ -44,7 +44,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) } for (f = 0; f < nr_faces; f++) { - for (l = pt->first_level; l <= pt->last_level; l++) { + for (l = 0; l <= pt->last_level; l++) { nv40mt->level[l].image_offset[f] = offset; offset += nv40mt->level[l].pitch * pt->height[l]; } @@ -87,7 +87,7 @@ nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) int l; pipe_buffer_reference(ws, &nv40mt->buffer, NULL); - for (l = mt->first_level; l <= mt->last_level; l++) { + for (l = 0; l <= mt->last_level; l++) { if (nv40mt->level[l].image_offset) free(nv40mt->level[l].image_offset); } -- cgit v1.2.3 From f33fa253c66241724fe4ae6943b091e0bc0409d6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 13 Feb 2008 19:33:41 +1100 Subject: nv40: attempt at obeying sampler min_lod/max_lod/lod_bias --- src/mesa/pipe/nv40/nv40_context.c | 2 +- src/mesa/pipe/nv40/nv40_fragtex.c | 5 ++--- src/mesa/pipe/nv40/nv40_state.c | 14 ++++++++++++++ 3 files changed, 17 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index 302ad04c154..a8a2eaf2157 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -74,7 +74,7 @@ nv40_get_paramf(struct pipe_context *pipe, int param) case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; + return 16.0; case PIPE_CAP_BITMAP_TEXCOORD_BIAS: return 0.0; default: diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c index 69c0fcb5e05..5af5fbe7465 100644 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ b/src/mesa/pipe/nv40/nv40_fragtex.c @@ -108,10 +108,9 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); so_data (so, ps->wrap); - so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en | - (0x00078000) /* mipmap related? */); + so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en); so_data (so, txs); - so_data (so, ps->filt | 0x3fd6 /*voodoo*/); + so_data (so, ps->filt | 0x2000 /*voodoo*/); so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); so_data (so, ps->bcol); diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index 80e94737efb..bcd244528d5 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -193,6 +193,20 @@ nv40_sampler_state_create(struct pipe_context *pipe, ps->filt = filter; + { + float limit; + + limit = CLAMP(cso->lod_bias, -16.0, 15.0); + ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff; + + limit = CLAMP(cso->max_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 7; + + limit = CLAMP(cso->min_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 19; + } + + if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { switch (cso->compare_func) { case PIPE_FUNC_NEVER: -- cgit v1.2.3 From 4c239ec2d76a7cb1c4ce0a782a30639bb061705d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 15 Feb 2008 00:14:15 +1100 Subject: nv40: always copy 4 float immediates for now. --- src/mesa/pipe/nv40/nv40_fragprog.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index cb7b4a5e70d..cc637f5dae2 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -710,14 +710,15 @@ nv40_fragprog_translate(struct nv40_context *nv40, { struct tgsi_full_immediate *imm; float vals[4]; - int i; imm = &parse.FullToken.FullImmediate; assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); assert(fpc->nr_imm < MAX_IMM); - for (i = 0; i < (imm->Immediate.Size - 1); i++) - vals[i] = imm->u.ImmediateFloat32[i].Float; + vals[0] = imm->u.ImmediateFloat32[0].Float; + vals[1] = imm->u.ImmediateFloat32[1].Float; + vals[2] = imm->u.ImmediateFloat32[2].Float; + vals[3] = imm->u.ImmediateFloat32[3].Float; fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); } break; -- cgit v1.2.3 From 8d13f55d2c8483148f0a8786a50daf6e890690a3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 15 Feb 2008 13:54:07 +1100 Subject: nv30: fix build --- src/mesa/pipe/nv30/nv30_fragtex.c | 2 +- src/mesa/pipe/nv30/nv30_miptree.c | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/src/mesa/pipe/nv30/nv30_fragtex.c b/src/mesa/pipe/nv30/nv30_fragtex.c index e75b1f7f28b..575f968cc53 100644 --- a/src/mesa/pipe/nv30/nv30_fragtex.c +++ b/src/mesa/pipe/nv30/nv30_fragtex.c @@ -95,7 +95,7 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) } txf = tf->format << 8; - txf |= (pt->last_level - pt->first_level + 1) << 16; + txf |= (pt->last_level + 1) << 16; txf |= log2i(pt->width[0]) << 20; txf |= log2i(pt->height[0]) << 24; txf |= log2i(pt->depth[0]) << 28; diff --git a/src/mesa/pipe/nv30/nv30_miptree.c b/src/mesa/pipe/nv30/nv30_miptree.c index 75e9b993c16..475f1be8adf 100644 --- a/src/mesa/pipe/nv30/nv30_miptree.c +++ b/src/mesa/pipe/nv30/nv30_miptree.c @@ -23,7 +23,7 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) nr_faces = 1; } - for (l = pt->first_level; l <= pt->last_level; l++) { + for (l = 0; l <= pt->last_level; l++) { pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; @@ -44,7 +44,7 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) } for (f = 0; f < nr_faces; f++) { - for (l = pt->first_level; l <= pt->last_level; l++) { + for (l = 0; l <= pt->last_level; l++) { nv30mt->level[l].image_offset[f] = offset; offset += nv30mt->level[l].pitch * pt->height[l]; } @@ -88,7 +88,7 @@ nv30_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) int l; pipe_buffer_reference(ws, &nv30mt->buffer, NULL); - for (l = mt->first_level; l <= mt->last_level; l++) { + for (l = 0; l <= mt->last_level; l++) { if (nv30mt->level[l].image_offset) free(nv30mt->level[l].image_offset); } -- cgit v1.2.3 From 86bba420231766a908c6b204b0df036f6a5bf08d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 15 Feb 2008 13:55:55 +1100 Subject: nv30: more interface updates --- src/mesa/pipe/nv30/nv30_dma.h | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv30/nv30_dma.h b/src/mesa/pipe/nv30/nv30_dma.h index 6eff6b42903..f8bc6eca76a 100644 --- a/src/mesa/pipe/nv30/nv30_dma.h +++ b/src/mesa/pipe/nv30/nv30_dma.h @@ -35,10 +35,9 @@ #define OUT_RELOC(bo,data,flags,vor,tor) do { \ nv30->nvws->push_reloc(nv30->nvws->channel, \ - nv30->nvws->channel->pushbuf->cur, \ + nv30->nvws->channel->pushbuf->cur++, \ (struct nouveau_bo *)(bo), \ (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ } while(0) /* Raw data + flags depending on FB/TT buffer */ -- cgit v1.2.3 From 01fccea190cf07b41a675e95d82af8ab5228b176 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 15 Feb 2008 03:48:32 +0100 Subject: nouveau: nv30 fixes. --- src/mesa/pipe/nv30/nv30_context.c | 2 +- src/mesa/pipe/nv30/nv30_state.c | 5 ++++- src/mesa/pipe/nv30/nv30_vertprog.c | 8 +++----- 3 files changed, 8 insertions(+), 7 deletions(-) diff --git a/src/mesa/pipe/nv30/nv30_context.c b/src/mesa/pipe/nv30/nv30_context.c index c56f918ad9d..eef49fbcc2b 100644 --- a/src/mesa/pipe/nv30/nv30_context.c +++ b/src/mesa/pipe/nv30/nv30_context.c @@ -255,7 +255,7 @@ nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) int w=4096; int h=4096; int pitch=4096*4; - BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 5); + BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 5); OUT_RING (w<<16); OUT_RING (h<<16); OUT_RING (0x148); /* format */ diff --git a/src/mesa/pipe/nv30/nv30_state.c b/src/mesa/pipe/nv30/nv30_state.c index c29a6448093..abc22eacae9 100644 --- a/src/mesa/pipe/nv30/nv30_state.c +++ b/src/mesa/pipe/nv30/nv30_state.c @@ -604,7 +604,7 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); - OUT_RING (rt[0]->pitch * rt[0]->cpp); + OUT_RING ( (rt[0]->pitch * rt[0]->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); nv30->rt[0] = rt[0]->buffer; } @@ -615,6 +615,9 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, } if (zeta_format) { + /* XXX allocate LMA */ +/* BEGIN_RING(rankine, NV34TCL_LMA_DEPTH_OFFSET, 1); + OUT_RING(0);*/ BEGIN_RING(rankine, NV34TCL_ZETA_PITCH, 1); OUT_RING (zeta->pitch * zeta->cpp); nv30->zeta = zeta->buffer; diff --git a/src/mesa/pipe/nv30/nv30_vertprog.c b/src/mesa/pipe/nv30/nv30_vertprog.c index b712049fa7c..41957b67c4e 100644 --- a/src/mesa/pipe/nv30/nv30_vertprog.c +++ b/src/mesa/pipe/nv30/nv30_vertprog.c @@ -751,7 +751,7 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) } #endif BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (vp->exec->start); + OUT_RING (/*vp->exec->start*/0); for (i = 0; i < vp->nr_insns; i++) { BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); OUT_RINGp (vp->insns[i].data, 4); @@ -759,10 +759,8 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) } BEGIN_RING(rankine, NV34TCL_VP_START_FROM_ID, 1); - OUT_RING (vp->exec->start); - BEGIN_RING(rankine, NV34TCL_VP_ATTRIB_EN, 2); - OUT_RING (vp->ir); - OUT_RING (vp->or); +// OUT_RING (vp->exec->start); + OUT_RING (0); nv30->vertprog.active = vp; } -- cgit v1.2.3 From 0a653bef05fb3627fdd1857bfa8c3a1ebe08a4b7 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Fri, 15 Feb 2008 04:23:46 +0100 Subject: nouveau: more nv30 fixes, still doesn't work as well as before. --- configs/linux-dri | 2 +- src/mesa/pipe/nv30/nv30_context.c | 2 ++ src/mesa/pipe/nv30/nv30_state.c | 8 ++------ src/mesa/pipe/nv30/nv30_vbo.c | 8 ++++---- 4 files changed, 9 insertions(+), 11 deletions(-) diff --git a/configs/linux-dri b/configs/linux-dri index 494b0aab8e9..c7eb7112c0e 100644 --- a/configs/linux-dri +++ b/configs/linux-dri @@ -66,4 +66,4 @@ WINDOW_SYSTEM=dri # gamma are missing because they have not been converted to use the new # interface. -DRI_DIRS = intel_winsys nouveau_winsys +DRI_DIRS = nouveau_winsys diff --git a/src/mesa/pipe/nv30/nv30_context.c b/src/mesa/pipe/nv30/nv30_context.c index eef49fbcc2b..d12aab85d8c 100644 --- a/src/mesa/pipe/nv30/nv30_context.c +++ b/src/mesa/pipe/nv30/nv30_context.c @@ -71,6 +71,8 @@ nv30_get_paramf(struct pipe_context *pipe, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/mesa/pipe/nv30/nv30_state.c b/src/mesa/pipe/nv30/nv30_state.c index abc22eacae9..992afe033e9 100644 --- a/src/mesa/pipe/nv30/nv30_state.c +++ b/src/mesa/pipe/nv30/nv30_state.c @@ -614,12 +614,8 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, nv30->rt[1] = rt[1]->buffer; } - if (zeta_format) { - /* XXX allocate LMA */ -/* BEGIN_RING(rankine, NV34TCL_LMA_DEPTH_OFFSET, 1); - OUT_RING(0);*/ - BEGIN_RING(rankine, NV34TCL_ZETA_PITCH, 1); - OUT_RING (zeta->pitch * zeta->cpp); + if (zeta_format) + { nv30->zeta = zeta->buffer; } diff --git a/src/mesa/pipe/nv30/nv30_vbo.c b/src/mesa/pipe/nv30/nv30_vbo.c index e6c50d3820c..173a6e8fd7b 100644 --- a/src/mesa/pipe/nv30/nv30_vbo.c +++ b/src/mesa/pipe/nv30/nv30_vbo.c @@ -241,9 +241,9 @@ nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, } while (count) { - push = MIN2(count, 2046); + push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(rankine, NV40TCL_VB_ELEMENT_U16, push); + BEGIN_RING_NI(rankine, NV40TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); @@ -266,9 +266,9 @@ nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, } while (count) { - push = MIN2(count, 2046); + push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push); + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); -- cgit v1.2.3 From e82eabaf5e6cb91984476a991ec24e8105989dc4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 16 Feb 2008 02:30:56 +1100 Subject: nv40: fix non-debug builds + start on obeying portability guidelines. --- src/mesa/pipe/nouveau/nouveau_stateobj.h | 22 ++++++++++-------- src/mesa/pipe/nv40/nv40_context.c | 4 ++-- src/mesa/pipe/nv40/nv40_context.h | 2 +- src/mesa/pipe/nv40/nv40_fragprog.c | 2 +- src/mesa/pipe/nv40/nv40_miptree.c | 4 ++-- src/mesa/pipe/nv40/nv40_query.c | 4 ++-- src/mesa/pipe/nv40/nv40_state.c | 6 ++--- src/mesa/pipe/nv40/nv40_vbo.c | 39 ++++++++++++++++++++++++-------- src/mesa/pipe/nv40/nv40_vertprog.c | 4 ++-- 9 files changed, 54 insertions(+), 33 deletions(-) diff --git a/src/mesa/pipe/nouveau/nouveau_stateobj.h b/src/mesa/pipe/nouveau/nouveau_stateobj.h index 58167a24de7..07c31b014a5 100644 --- a/src/mesa/pipe/nouveau/nouveau_stateobj.h +++ b/src/mesa/pipe/nouveau/nouveau_stateobj.h @@ -1,6 +1,8 @@ #ifndef __NOUVEAU_STATEOBJ_H__ #define __NOUVEAU_STATEOBJ_H__ +#include "pipe/p_util.h" + struct nouveau_stateobj_reloc { struct pipe_buffer *bo; @@ -24,15 +26,15 @@ struct nouveau_stateobj { unsigned cur_reloc; }; -static inline struct nouveau_stateobj * +static INLINE struct nouveau_stateobj * so_new(unsigned push, unsigned reloc) { struct nouveau_stateobj *so; - so = malloc(sizeof(struct nouveau_stateobj)); + so = MALLOC(sizeof(struct nouveau_stateobj)); so->refcount = 1; - so->push = malloc(sizeof(unsigned) * push); - so->reloc = malloc(sizeof(struct nouveau_stateobj_reloc) * reloc); + so->push = MALLOC(sizeof(unsigned) * push); + so->reloc = MALLOC(sizeof(struct nouveau_stateobj_reloc) * reloc); so->cur = so->push; so->cur_reloc = so->cur_packet = 0; @@ -40,7 +42,7 @@ so_new(unsigned push, unsigned reloc) return so; } -static inline void +static INLINE void so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) { struct nouveau_stateobj *so; @@ -61,14 +63,14 @@ so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) } } -static inline void +static INLINE void so_data(struct nouveau_stateobj *so, unsigned data) { (*so->cur++) = (data); so->cur_packet += 4; } -static inline void +static INLINE void so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, unsigned mthd, unsigned size) { @@ -76,7 +78,7 @@ so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, so_data(so, (gr->subc << 13) | (size << 18) | mthd); } -static inline void +static INLINE void so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, unsigned data, unsigned flags, unsigned vor, unsigned tor) { @@ -92,7 +94,7 @@ so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, so_data(so, data); } -static inline void +static INLINE void so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) { struct nouveau_pushbuf *pb = nvws->channel->pushbuf; @@ -113,7 +115,7 @@ so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) pb->cur += nr; } -static inline void +static INLINE void so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) { struct nouveau_pushbuf *pb = nvws->channel->pushbuf; diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c index a8a2eaf2157..6e86ca00816 100644 --- a/src/mesa/pipe/nv40/nv40_context.c +++ b/src/mesa/pipe/nv40/nv40_context.c @@ -154,7 +154,7 @@ nv40_channel_init(struct pipe_winsys *ws, struct nouveau_winsys *nvws, return NULL; } - cnv40 = calloc(1, sizeof(struct nv40_channel_context)); + cnv40 = CALLOC(1, sizeof(struct nv40_channel_context)); if (!cnv40) return NULL; cnv40->chipset = chipset; @@ -274,7 +274,7 @@ nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, { struct nv40_context *nv40; - nv40 = calloc(1, sizeof(struct nv40_context)); + nv40 = CALLOC(1, sizeof(struct nv40_context)); if (!nv40) return NULL; diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h index d7c9ee7851e..cf2a14405a4 100644 --- a/src/mesa/pipe/nv40/nv40_context.h +++ b/src/mesa/pipe/nv40/nv40_context.h @@ -100,7 +100,7 @@ struct nv40_context { struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; -static inline struct nv40_context * +static INLINE struct nv40_context * nv40_context(struct pipe_context *pipe) { return (struct nv40_context *)pipe; diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c index cc637f5dae2..7487fb896f3 100644 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ b/src/mesa/pipe/nv40/nv40_fragprog.c @@ -675,7 +675,7 @@ nv40_fragprog_translate(struct nv40_context *nv40, struct tgsi_parse_context parse; struct nv40_fpc *fpc = NULL; - fpc = calloc(1, sizeof(struct nv40_fpc)); + fpc = CALLOC(1, sizeof(struct nv40_fpc)); if (!fpc) return; fpc->fp = fp; diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c index 1b3c27dc45b..92e6b3a43df 100644 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ b/src/mesa/pipe/nv40/nv40_miptree.c @@ -35,7 +35,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) nv40mt->level[l].pitch = (nv40mt->level[l].pitch + 63) & ~63; nv40mt->level[l].image_offset = - calloc(nr_faces, sizeof(unsigned)); + CALLOC(nr_faces, sizeof(unsigned)); width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); @@ -59,7 +59,7 @@ nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) struct pipe_winsys *ws = pipe->winsys; struct nv40_miptree *mt; - mt = malloc(sizeof(struct nv40_miptree)); + mt = MALLOC(sizeof(struct nv40_miptree)); if (!mt) return NULL; mt->base = *pt; diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c index eb305e64446..8bca2788b93 100644 --- a/src/mesa/pipe/nv40/nv40_query.c +++ b/src/mesa/pipe/nv40/nv40_query.c @@ -9,7 +9,7 @@ struct nv40_query { uint64_t result; }; -static inline struct nv40_query * +static INLINE struct nv40_query * nv40_query(struct pipe_query *pipe) { return (struct nv40_query *)pipe; @@ -20,7 +20,7 @@ nv40_query_create(struct pipe_context *pipe, unsigned query_type) { struct nv40_query *q; - q = calloc(1, sizeof(struct nv40_query)); + q = CALLOC(1, sizeof(struct nv40_query)); q->type = query_type; return (struct pipe_query *)q; diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c index bcd244528d5..713f31dbb12 100644 --- a/src/mesa/pipe/nv40/nv40_state.c +++ b/src/mesa/pipe/nv40/nv40_state.c @@ -112,7 +112,7 @@ nv40_sampler_state_create(struct pipe_context *pipe, struct nv40_sampler_state *ps; uint32_t filter = 0; - ps = malloc(sizeof(struct nv40_sampler_state)); + ps = MALLOC(sizeof(struct nv40_sampler_state)); ps->fmt = 0; if (!cso->normalized_coords) @@ -455,7 +455,7 @@ nv40_vp_state_create(struct pipe_context *pipe, { struct nv40_vertex_program *vp; - vp = calloc(1, sizeof(struct nv40_vertex_program)); + vp = CALLOC(1, sizeof(struct nv40_vertex_program)); vp->pipe = cso; return (void *)vp; @@ -487,7 +487,7 @@ nv40_fp_state_create(struct pipe_context *pipe, { struct nv40_fragment_program *fp; - fp = calloc(1, sizeof(struct nv40_fragment_program)); + fp = CALLOC(1, sizeof(struct nv40_fragment_program)); fp->pipe = cso; return (void *)fp; diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index 4e9cdb4585a..a18d0f9b251 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -30,7 +30,8 @@ nv40_vbo_type(uint format) case PIPE_FORMAT_TYPE_UNORM: return NV40TCL_VTXFMT_TYPE_UBYTE; default: - assert(0); + NOUVEAU_ERR("Unknown format 0x%08x\n", format); + return NV40TCL_VTXFMT_TYPE_FLOAT; } } @@ -188,8 +189,13 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, { struct nv40_context *nv40 = nv40_context(pipe); unsigned nr; + boolean ret; - assert(nv40_vbo_validate_state(nv40, NULL, 0)); + ret = nv40_vbo_validate_state(nv40, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -290,19 +296,26 @@ nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, } static boolean -nv40_draw_elements_inline(struct pipe_context *pipe, +nv40_draw_elements_INLINE(struct pipe_context *pipe, struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); struct pipe_winsys *ws = pipe->winsys; + boolean ret; void *map; - assert(nv40_vbo_validate_state(nv40, NULL, 0)); + ret = nv40_vbo_validate_state(nv40, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); - if (!ib) - assert(0); + if (!ib) { + NOUVEAU_ERR("failed mapping ib\n"); + return FALSE; + } BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -318,7 +331,7 @@ nv40_draw_elements_inline(struct pipe_context *pipe, nv40_draw_elements_u32(nv40, map, start, count); break; default: - assert(0); + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); break; } @@ -337,6 +350,7 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); unsigned nr, type; + boolean ret; switch (ib_size) { case 2: @@ -346,10 +360,15 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; break; default: - assert(0); + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); + return FALSE; } - assert(nv40_vbo_validate_state(nv40, ib, type)); + ret = nv40_vbo_validate_state(nv40, ib, type); + if (!ret) { + NOUVEAU_ERR("failed state validation\n"); + return FALSE; + } BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -391,7 +410,7 @@ nv40_draw_elements(struct pipe_context *pipe, * to be support on any chipset for 8-bit indices. */ if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { - nv40_draw_elements_inline(pipe, indexBuffer, indexSize, + nv40_draw_elements_INLINE(pipe, indexBuffer, indexSize, mode, start, count); } else { nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c index 415b3c70c79..d57e3ca350b 100644 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ b/src/mesa/pipe/nv40/nv40_vertprog.c @@ -551,7 +551,7 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) tgsi_parse_free(&p); if (nr_imm) { - vpc->imm = calloc(nr_imm, sizeof(struct nv40_sreg)); + vpc->imm = CALLOC(nr_imm, sizeof(struct nv40_sreg)); assert(vpc->imm); } @@ -565,7 +565,7 @@ nv40_vertprog_translate(struct nv40_context *nv40, struct tgsi_parse_context parse; struct nv40_vpc *vpc = NULL; - vpc = calloc(1, sizeof(struct nv40_vpc)); + vpc = CALLOC(1, sizeof(struct nv40_vpc)); if (!vpc) return; vpc->vp = vp; -- cgit v1.2.3 From 6e3e5ba3bb25183efafcf78d6794fc50236c2835 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 16 Feb 2008 02:42:06 +1100 Subject: nv40: over-zealous s/ --- src/mesa/pipe/nv40/nv40_vbo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c index a18d0f9b251..6b1ac65b49e 100644 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ b/src/mesa/pipe/nv40/nv40_vbo.c @@ -296,7 +296,7 @@ nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, } static boolean -nv40_draw_elements_INLINE(struct pipe_context *pipe, +nv40_draw_elements_inline(struct pipe_context *pipe, struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { @@ -410,7 +410,7 @@ nv40_draw_elements(struct pipe_context *pipe, * to be support on any chipset for 8-bit indices. */ if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { - nv40_draw_elements_INLINE(pipe, indexBuffer, indexSize, + nv40_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); } else { nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, -- cgit v1.2.3 From 4032ff3889021089debce1c43a0bb984b121cbf6 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sat, 16 Feb 2008 02:54:11 +0100 Subject: nv30: fixes. --- src/mesa/pipe/nv30/nv30_context.h | 2 +- src/mesa/pipe/nv30/nv30_dma.h | 65 -------------------------------------- src/mesa/pipe/nv30/nv30_fragprog.c | 5 +-- src/mesa/pipe/nv30/nv30_fragtex.c | 2 +- src/mesa/pipe/nv30/nv30_miptree.c | 2 +- src/mesa/pipe/nv30/nv30_query.c | 3 +- src/mesa/pipe/nv30/nv30_state.c | 8 ++--- src/mesa/pipe/nv30/nv30_vbo.c | 39 +++++++++++++++++------ src/mesa/pipe/nv30/nv30_vertprog.c | 5 +-- 9 files changed, 44 insertions(+), 87 deletions(-) delete mode 100644 src/mesa/pipe/nv30/nv30_dma.h diff --git a/src/mesa/pipe/nv30/nv30_context.h b/src/mesa/pipe/nv30/nv30_context.h index d2262c5065f..f6c69545991 100644 --- a/src/mesa/pipe/nv30/nv30_context.h +++ b/src/mesa/pipe/nv30/nv30_context.h @@ -83,7 +83,7 @@ struct nv30_context { struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; -static inline struct nv30_context * +static INLINE struct nv30_context * nv30_context(struct pipe_context *pipe) { return (struct nv30_context *)pipe; diff --git a/src/mesa/pipe/nv30/nv30_dma.h b/src/mesa/pipe/nv30/nv30_dma.h deleted file mode 100644 index f8bc6eca76a..00000000000 --- a/src/mesa/pipe/nv30/nv30_dma.h +++ /dev/null @@ -1,65 +0,0 @@ -#ifndef __NV30_DMA_H__ -#define __NV30_DMA_H__ - -#include "pipe/nouveau/nouveau_winsys.h" - -#define OUT_RING(data) do { \ - (*nv30->nvws->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - memcpy(nv30->nvws->channel->pushbuf->cur, (src), (size) * 4); \ - nv30->nvws->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - if (nv30->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - nv30->nvws->push_flush(nv30->nvws->channel, ((size) + 1)); \ - OUT_RING((nv30->obj->subc << 13) | ((size) << 18) | (mthd)); \ - nv30->nvws->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - -#define FIRE_RING() do { \ - nv30->nvws->push_flush(nv30->nvws->channel, 0); \ -} while(0) - -#define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nv30->nvws->push_reloc(nv30->nvws->channel, \ - nv30->nvws->channel->pushbuf->cur++, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ -} while(0) - -/* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) - -/* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - nv30->nvws->channel->vram->handle, \ - nv30->nvws->channel->gart->handle); \ -} while(0) - -/* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) - -/* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) - -#endif diff --git a/src/mesa/pipe/nv30/nv30_fragprog.c b/src/mesa/pipe/nv30/nv30_fragprog.c index 0233873d925..0db1ac868c7 100644 --- a/src/mesa/pipe/nv30/nv30_fragprog.c +++ b/src/mesa/pipe/nv30/nv30_fragprog.c @@ -1,6 +1,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "pipe/tgsi/util/tgsi_parse.h" @@ -675,7 +676,7 @@ nv30_fragprog_translate(struct nv30_context *nv30, struct tgsi_parse_context parse; struct nv30_fpc *fpc = NULL; - fpc = calloc(1, sizeof(struct nv30_fpc)); + fpc = CALLOC(1, sizeof(struct nv30_fpc)); if (!fpc) return; fpc->fp = fp; @@ -716,7 +717,7 @@ nv30_fragprog_translate(struct nv30_context *nv30, assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); assert(fpc->nr_imm < MAX_IMM); - for (i = 0; i < imm->Immediate.Size; i++) + for (i = 0; i < 4; i++) vals[i] = imm->u.ImmediateFloat32[i].Float; fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); } diff --git a/src/mesa/pipe/nv30/nv30_fragtex.c b/src/mesa/pipe/nv30/nv30_fragtex.c index 575f968cc53..45ee6db8d6a 100644 --- a/src/mesa/pipe/nv30/nv30_fragtex.c +++ b/src/mesa/pipe/nv30/nv30_fragtex.c @@ -1,6 +1,6 @@ #include "nv30_context.h" -static inline int log2i(int i) +static INLINE int log2i(int i) { int r = 0; diff --git a/src/mesa/pipe/nv30/nv30_miptree.c b/src/mesa/pipe/nv30/nv30_miptree.c index 475f1be8adf..5fb89f4cfdc 100644 --- a/src/mesa/pipe/nv30/nv30_miptree.c +++ b/src/mesa/pipe/nv30/nv30_miptree.c @@ -35,7 +35,7 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) nv30mt->level[l].pitch = (nv30mt->level[l].pitch + 63) & ~63; nv30mt->level[l].image_offset = - calloc(nr_faces, sizeof(unsigned)); + CALLOC(nr_faces, sizeof(unsigned)); width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); diff --git a/src/mesa/pipe/nv30/nv30_query.c b/src/mesa/pipe/nv30/nv30_query.c index ea74c0f5f1f..71fdcfa24df 100644 --- a/src/mesa/pipe/nv30/nv30_query.c +++ b/src/mesa/pipe/nv30/nv30_query.c @@ -1,4 +1,5 @@ #include "pipe/p_context.h" +#include "pipe/p_util.h" #include "nv30_context.h" @@ -20,7 +21,7 @@ nv30_query_create(struct pipe_context *pipe, unsigned query_type) { struct nv30_query *q; - q = calloc(1, sizeof(struct nv30_query)); + q = CALLOC(1, sizeof(struct nv30_query)); q->type = query_type; return (struct pipe_query *)q; diff --git a/src/mesa/pipe/nv30/nv30_state.c b/src/mesa/pipe/nv30/nv30_state.c index 992afe033e9..53368561e07 100644 --- a/src/mesa/pipe/nv30/nv30_state.c +++ b/src/mesa/pipe/nv30/nv30_state.c @@ -420,9 +420,9 @@ nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) BEGIN_RING(rankine, NV34TCL_DEPTH_FUNC, 3); OUT_RINGp ((uint32_t *)&hw->depth, 3); - BEGIN_RING(rankine, NV34TCL_STENCIL_FRONT_ENABLE, 16); - OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); + BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 16); OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); + OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3); OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); } @@ -439,7 +439,7 @@ nv30_vp_state_create(struct pipe_context *pipe, { struct nv30_vertex_program *vp; - vp = calloc(1, sizeof(struct nv30_vertex_program)); + vp = CALLOC(1, sizeof(struct nv30_vertex_program)); vp->pipe = cso; return (void *)vp; @@ -471,7 +471,7 @@ nv30_fp_state_create(struct pipe_context *pipe, { struct nv30_fragment_program *fp; - fp = calloc(1, sizeof(struct nv30_fragment_program)); + fp = CALLOC(1, sizeof(struct nv30_fragment_program)); fp->pipe = cso; return (void *)fp; diff --git a/src/mesa/pipe/nv30/nv30_vbo.c b/src/mesa/pipe/nv30/nv30_vbo.c index 173a6e8fd7b..414cf55ac8d 100644 --- a/src/mesa/pipe/nv30/nv30_vbo.c +++ b/src/mesa/pipe/nv30/nv30_vbo.c @@ -30,7 +30,8 @@ nv30_vbo_type(uint format) case PIPE_FORMAT_TYPE_UNORM: return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE; default: - assert(0); + NOUVEAU_ERR("Unknown format 0x%08x\n", format); + return NV40TCL_VTXFMT_TYPE_FLOAT; } } @@ -194,8 +195,13 @@ nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, { struct nv30_context *nv30 = nv30_context(pipe); unsigned nr; + boolean ret; - assert(nv30_vbo_validate_state(nv30, NULL, 0)); + ret = nv30_vbo_validate_state(nv30, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -302,13 +308,20 @@ nv30_draw_elements_inline(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); struct pipe_winsys *ws = pipe->winsys; + boolean ret; void *map; - assert(nv30_vbo_validate_state(nv30, NULL, 0)); + ret = nv30_vbo_validate_state(nv30, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); - if (!ib) - assert(0); + if (!ib) { + NOUVEAU_ERR("failed mapping ib\n"); + return FALSE; + } BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -324,7 +337,7 @@ nv30_draw_elements_inline(struct pipe_context *pipe, nv30_draw_elements_u32(nv30, map, start, count); break; default: - assert(0); + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); break; } @@ -343,6 +356,7 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); unsigned nr, type; + boolean ret; switch (ib_size) { case 2: @@ -352,10 +366,15 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; break; default: - assert(0); + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); + return FALSE; } - assert(nv30_vbo_validate_state(nv30, ib, type)); + ret = nv30_vbo_validate_state(nv30, ib, type); + if (!ret) { + NOUVEAU_ERR("failed state validation\n"); + return FALSE; + } BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -391,10 +410,10 @@ nv30_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { - if (indexSize != 1) { +/* if (indexSize != 1) { nv30_draw_elements_vbo(pipe, indexBuffer, indexSize, mode, start, count); - } else { + } else */{ nv30_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); } diff --git a/src/mesa/pipe/nv30/nv30_vertprog.c b/src/mesa/pipe/nv30/nv30_vertprog.c index 41957b67c4e..c96210d3fa7 100644 --- a/src/mesa/pipe/nv30/nv30_vertprog.c +++ b/src/mesa/pipe/nv30/nv30_vertprog.c @@ -1,6 +1,7 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "pipe/tgsi/util/tgsi_parse.h" @@ -539,7 +540,7 @@ nv30_vertprog_prepare(struct nv30_vpc *vpc) tgsi_parse_free(&p); if (nr_imm) { - vpc->imm = calloc(nr_imm, sizeof(struct nv30_sreg)); + vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); assert(vpc->imm); } @@ -553,7 +554,7 @@ nv30_vertprog_translate(struct nv30_context *nv30, struct tgsi_parse_context parse; struct nv30_vpc *vpc = NULL; - vpc = calloc(1, sizeof(struct nv30_vpc)); + vpc = CALLOC(1, sizeof(struct nv30_vpc)); if (!vpc) return; vpc->vp = vp; -- cgit v1.2.3 From 4d9c19d2f7eef263b49485b6e65be9afbe58363a Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sat, 16 Feb 2008 03:04:28 +0100 Subject: nv30: fixes. --- src/mesa/pipe/nv30/nv30_vbo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/mesa/pipe/nv30/nv30_vbo.c b/src/mesa/pipe/nv30/nv30_vbo.c index 414cf55ac8d..57fb9bc8a57 100644 --- a/src/mesa/pipe/nv30/nv30_vbo.c +++ b/src/mesa/pipe/nv30/nv30_vbo.c @@ -241,7 +241,7 @@ nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, int push, i; if (count & 1) { - BEGIN_RING(rankine, NV40TCL_VB_ELEMENT_U32, 1); + BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); OUT_RING (elts[0]); elts++; count--; } @@ -249,7 +249,7 @@ nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, while (count) { push = MIN2(count, 2047 * 2); - BEGIN_RING_NI(rankine, NV40TCL_VB_ELEMENT_U16, push >> 1); + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((elts[i+1] << 16) | elts[i]); -- cgit v1.2.3 From 26add9288c88108e3485ffc57c51ea9bdc0ee719 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 16 Feb 2008 17:23:12 +1100 Subject: nouveau: match gallium code reorginisation. That was... fun.. --- src/gallium/drivers/Makefile | 3 +- src/gallium/drivers/nouveau/nouveau_bo.h | 51 + src/gallium/drivers/nouveau/nouveau_channel.h | 39 + src/gallium/drivers/nouveau/nouveau_class.h | 6134 ++++++++++++++++++++ src/gallium/drivers/nouveau/nouveau_gldefs.h | 196 + src/gallium/drivers/nouveau/nouveau_grobj.h | 35 + src/gallium/drivers/nouveau/nouveau_notifier.h | 43 + src/gallium/drivers/nouveau/nouveau_push.h | 83 + src/gallium/drivers/nouveau/nouveau_pushbuf.h | 32 + src/gallium/drivers/nouveau/nouveau_resource.h | 37 + src/gallium/drivers/nouveau/nouveau_stateobj.h | 141 + src/gallium/drivers/nouveau/nouveau_winsys.h | 61 + src/gallium/drivers/nv30/Makefile | 29 + src/gallium/drivers/nv30/nv30_clear.c | 12 + src/gallium/drivers/nv30/nv30_context.c | 431 ++ src/gallium/drivers/nv30/nv30_context.h | 136 + src/gallium/drivers/nv30/nv30_draw.c | 62 + src/gallium/drivers/nv30/nv30_fragprog.c | 835 +++ src/gallium/drivers/nv30/nv30_fragtex.c | 160 + src/gallium/drivers/nv30/nv30_miptree.c | 105 + src/gallium/drivers/nv30/nv30_query.c | 113 + src/gallium/drivers/nv30/nv30_shader.h | 490 ++ src/gallium/drivers/nv30/nv30_state.c | 739 +++ src/gallium/drivers/nv30/nv30_state.h | 147 + src/gallium/drivers/nv30/nv30_state_emit.c | 83 + src/gallium/drivers/nv30/nv30_surface.c | 137 + src/gallium/drivers/nv30/nv30_vbo.c | 425 ++ src/gallium/drivers/nv30/nv30_vertprog.c | 777 +++ src/gallium/drivers/nv40/Makefile | 29 + src/gallium/drivers/nv40/nv40_clear.c | 12 + src/gallium/drivers/nv40/nv40_context.c | 312 + src/gallium/drivers/nv40/nv40_context.h | 153 + src/gallium/drivers/nv40/nv40_dma.h | 66 + src/gallium/drivers/nv40/nv40_draw.c | 62 + src/gallium/drivers/nv40/nv40_fragprog.c | 842 +++ src/gallium/drivers/nv40/nv40_fragtex.c | 151 + src/gallium/drivers/nv40/nv40_miptree.c | 104 + src/gallium/drivers/nv40/nv40_query.c | 113 + src/gallium/drivers/nv40/nv40_shader.h | 554 ++ src/gallium/drivers/nv40/nv40_state.c | 823 +++ src/gallium/drivers/nv40/nv40_state.h | 80 + src/gallium/drivers/nv40/nv40_state_emit.c | 77 + src/gallium/drivers/nv40/nv40_surface.c | 137 + src/gallium/drivers/nv40/nv40_vbo.c | 424 ++ src/gallium/drivers/nv40/nv40_vertprog.c | 790 +++ src/gallium/drivers/nv50/Makefile | 25 + src/gallium/drivers/nv50/nv50_clear.c | 12 + src/gallium/drivers/nv50/nv50_context.c | 202 + src/gallium/drivers/nv50/nv50_context.h | 57 + src/gallium/drivers/nv50/nv50_draw.c | 55 + src/gallium/drivers/nv50/nv50_miptree.c | 25 + src/gallium/drivers/nv50/nv50_query.c | 47 + src/gallium/drivers/nv50/nv50_state.c | 213 + src/gallium/drivers/nv50/nv50_state.h | 7 + src/gallium/drivers/nv50/nv50_surface.c | 75 + src/gallium/drivers/nv50/nv50_vbo.c | 24 + src/gallium/winsys/dri/nouveau/Makefile | 43 + src/gallium/winsys/dri/nouveau/nouveau_bo.c | 402 ++ src/gallium/winsys/dri/nouveau/nouveau_channel.c | 118 + src/gallium/winsys/dri/nouveau/nouveau_context.c | 206 + src/gallium/winsys/dri/nouveau/nouveau_context.h | 89 + src/gallium/winsys/dri/nouveau/nouveau_device.c | 146 + src/gallium/winsys/dri/nouveau/nouveau_device.h | 29 + src/gallium/winsys/dri/nouveau/nouveau_dma.c | 219 + src/gallium/winsys/dri/nouveau/nouveau_dma.h | 143 + src/gallium/winsys/dri/nouveau/nouveau_dri.h | 28 + src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 304 + src/gallium/winsys/dri/nouveau/nouveau_fence.c | 215 + src/gallium/winsys/dri/nouveau/nouveau_grobj.c | 107 + src/gallium/winsys/dri/nouveau/nouveau_local.h | 89 + src/gallium/winsys/dri/nouveau/nouveau_lock.c | 94 + src/gallium/winsys/dri/nouveau/nouveau_notifier.c | 137 + src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 261 + src/gallium/winsys/dri/nouveau/nouveau_resource.c | 111 + src/gallium/winsys/dri/nouveau/nouveau_screen.c | 308 + src/gallium/winsys/dri/nouveau/nouveau_screen.h | 19 + .../winsys/dri/nouveau/nouveau_swapbuffers.c | 86 + .../winsys/dri/nouveau/nouveau_swapbuffers.h | 10 + src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 124 + .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 196 + .../winsys/dri/nouveau/nouveau_winsys_pipe.h | 34 + .../winsys/dri/nouveau/nouveau_winsys_softpipe.c | 83 + src/gallium/winsys/dri/nouveau/nv04_surface.c | 226 + src/gallium/winsys/dri/nouveau/nv50_surface.c | 160 + src/mesa/drivers/dri/nouveau_winsys/Makefile | 43 - src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c | 402 -- .../drivers/dri/nouveau_winsys/nouveau_channel.c | 118 - .../drivers/dri/nouveau_winsys/nouveau_context.c | 206 - .../drivers/dri/nouveau_winsys/nouveau_context.h | 89 - .../drivers/dri/nouveau_winsys/nouveau_device.c | 146 - .../drivers/dri/nouveau_winsys/nouveau_device.h | 29 - src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c | 219 - src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h | 143 - src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h | 28 - .../drivers/dri/nouveau_winsys/nouveau_drmif.h | 304 - .../drivers/dri/nouveau_winsys/nouveau_fence.c | 215 - .../drivers/dri/nouveau_winsys/nouveau_grobj.c | 107 - .../drivers/dri/nouveau_winsys/nouveau_local.h | 89 - src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c | 94 - .../drivers/dri/nouveau_winsys/nouveau_notifier.c | 137 - .../drivers/dri/nouveau_winsys/nouveau_pushbuf.c | 261 - .../drivers/dri/nouveau_winsys/nouveau_resource.c | 111 - .../drivers/dri/nouveau_winsys/nouveau_screen.c | 308 - .../drivers/dri/nouveau_winsys/nouveau_screen.h | 19 - .../dri/nouveau_winsys/nouveau_swapbuffers.c | 86 - .../dri/nouveau_winsys/nouveau_swapbuffers.h | 10 - .../drivers/dri/nouveau_winsys/nouveau_winsys.c | 124 - .../dri/nouveau_winsys/nouveau_winsys_pipe.c | 196 - .../dri/nouveau_winsys/nouveau_winsys_pipe.h | 34 - .../dri/nouveau_winsys/nouveau_winsys_softpipe.c | 83 - src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c | 226 - src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c | 160 - src/mesa/pipe/nouveau/nouveau_bo.h | 51 - src/mesa/pipe/nouveau/nouveau_channel.h | 39 - src/mesa/pipe/nouveau/nouveau_class.h | 6134 -------------------- src/mesa/pipe/nouveau/nouveau_gldefs.h | 196 - src/mesa/pipe/nouveau/nouveau_grobj.h | 35 - src/mesa/pipe/nouveau/nouveau_notifier.h | 43 - src/mesa/pipe/nouveau/nouveau_push.h | 83 - src/mesa/pipe/nouveau/nouveau_pushbuf.h | 32 - src/mesa/pipe/nouveau/nouveau_resource.h | 37 - src/mesa/pipe/nouveau/nouveau_stateobj.h | 141 - src/mesa/pipe/nouveau/nouveau_winsys.h | 61 - src/mesa/pipe/nv30/Makefile | 29 - src/mesa/pipe/nv30/nv30_clear.c | 12 - src/mesa/pipe/nv30/nv30_context.c | 431 -- src/mesa/pipe/nv30/nv30_context.h | 136 - src/mesa/pipe/nv30/nv30_draw.c | 62 - src/mesa/pipe/nv30/nv30_fragprog.c | 835 --- src/mesa/pipe/nv30/nv30_fragtex.c | 160 - src/mesa/pipe/nv30/nv30_miptree.c | 105 - src/mesa/pipe/nv30/nv30_query.c | 113 - src/mesa/pipe/nv30/nv30_shader.h | 490 -- src/mesa/pipe/nv30/nv30_state.c | 739 --- src/mesa/pipe/nv30/nv30_state.h | 147 - src/mesa/pipe/nv30/nv30_state_emit.c | 83 - src/mesa/pipe/nv30/nv30_surface.c | 136 - src/mesa/pipe/nv30/nv30_vbo.c | 425 -- src/mesa/pipe/nv30/nv30_vertprog.c | 777 --- src/mesa/pipe/nv40/Makefile | 29 - src/mesa/pipe/nv40/nv40_clear.c | 12 - src/mesa/pipe/nv40/nv40_context.c | 312 - src/mesa/pipe/nv40/nv40_context.h | 153 - src/mesa/pipe/nv40/nv40_dma.h | 66 - src/mesa/pipe/nv40/nv40_draw.c | 62 - src/mesa/pipe/nv40/nv40_fragprog.c | 842 --- src/mesa/pipe/nv40/nv40_fragtex.c | 151 - src/mesa/pipe/nv40/nv40_miptree.c | 104 - src/mesa/pipe/nv40/nv40_query.c | 113 - src/mesa/pipe/nv40/nv40_shader.h | 554 -- src/mesa/pipe/nv40/nv40_state.c | 823 --- src/mesa/pipe/nv40/nv40_state.h | 80 - src/mesa/pipe/nv40/nv40_state_emit.c | 77 - src/mesa/pipe/nv40/nv40_surface.c | 136 - src/mesa/pipe/nv40/nv40_vbo.c | 424 -- src/mesa/pipe/nv40/nv40_vertprog.c | 790 --- src/mesa/pipe/nv50/Makefile | 25 - src/mesa/pipe/nv50/nv50_clear.c | 12 - src/mesa/pipe/nv50/nv50_context.c | 202 - src/mesa/pipe/nv50/nv50_context.h | 57 - src/mesa/pipe/nv50/nv50_draw.c | 55 - src/mesa/pipe/nv50/nv50_miptree.c | 25 - src/mesa/pipe/nv50/nv50_query.c | 47 - src/mesa/pipe/nv50/nv50_state.c | 213 - src/mesa/pipe/nv50/nv50_state.h | 7 - src/mesa/pipe/nv50/nv50_surface.c | 74 - src/mesa/pipe/nv50/nv50_vbo.c | 24 - 167 files changed, 20993 insertions(+), 20989 deletions(-) create mode 100644 src/gallium/drivers/nouveau/nouveau_bo.h create mode 100644 src/gallium/drivers/nouveau/nouveau_channel.h create mode 100644 src/gallium/drivers/nouveau/nouveau_class.h create mode 100644 src/gallium/drivers/nouveau/nouveau_gldefs.h create mode 100644 src/gallium/drivers/nouveau/nouveau_grobj.h create mode 100644 src/gallium/drivers/nouveau/nouveau_notifier.h create mode 100644 src/gallium/drivers/nouveau/nouveau_push.h create mode 100644 src/gallium/drivers/nouveau/nouveau_pushbuf.h create mode 100644 src/gallium/drivers/nouveau/nouveau_resource.h create mode 100644 src/gallium/drivers/nouveau/nouveau_stateobj.h create mode 100644 src/gallium/drivers/nouveau/nouveau_winsys.h create mode 100644 src/gallium/drivers/nv30/Makefile create mode 100644 src/gallium/drivers/nv30/nv30_clear.c create mode 100644 src/gallium/drivers/nv30/nv30_context.c create mode 100644 src/gallium/drivers/nv30/nv30_context.h create mode 100644 src/gallium/drivers/nv30/nv30_draw.c create mode 100644 src/gallium/drivers/nv30/nv30_fragprog.c create mode 100644 src/gallium/drivers/nv30/nv30_fragtex.c create mode 100644 src/gallium/drivers/nv30/nv30_miptree.c create mode 100644 src/gallium/drivers/nv30/nv30_query.c create mode 100644 src/gallium/drivers/nv30/nv30_shader.h create mode 100644 src/gallium/drivers/nv30/nv30_state.c create mode 100644 src/gallium/drivers/nv30/nv30_state.h create mode 100644 src/gallium/drivers/nv30/nv30_state_emit.c create mode 100644 src/gallium/drivers/nv30/nv30_surface.c create mode 100644 src/gallium/drivers/nv30/nv30_vbo.c create mode 100644 src/gallium/drivers/nv30/nv30_vertprog.c create mode 100644 src/gallium/drivers/nv40/Makefile create mode 100644 src/gallium/drivers/nv40/nv40_clear.c create mode 100644 src/gallium/drivers/nv40/nv40_context.c create mode 100644 src/gallium/drivers/nv40/nv40_context.h create mode 100644 src/gallium/drivers/nv40/nv40_dma.h create mode 100644 src/gallium/drivers/nv40/nv40_draw.c create mode 100644 src/gallium/drivers/nv40/nv40_fragprog.c create mode 100644 src/gallium/drivers/nv40/nv40_fragtex.c create mode 100644 src/gallium/drivers/nv40/nv40_miptree.c create mode 100644 src/gallium/drivers/nv40/nv40_query.c create mode 100644 src/gallium/drivers/nv40/nv40_shader.h create mode 100644 src/gallium/drivers/nv40/nv40_state.c create mode 100644 src/gallium/drivers/nv40/nv40_state.h create mode 100644 src/gallium/drivers/nv40/nv40_state_emit.c create mode 100644 src/gallium/drivers/nv40/nv40_surface.c create mode 100644 src/gallium/drivers/nv40/nv40_vbo.c create mode 100644 src/gallium/drivers/nv40/nv40_vertprog.c create mode 100644 src/gallium/drivers/nv50/Makefile create mode 100644 src/gallium/drivers/nv50/nv50_clear.c create mode 100644 src/gallium/drivers/nv50/nv50_context.c create mode 100644 src/gallium/drivers/nv50/nv50_context.h create mode 100644 src/gallium/drivers/nv50/nv50_draw.c create mode 100644 src/gallium/drivers/nv50/nv50_miptree.c create mode 100644 src/gallium/drivers/nv50/nv50_query.c create mode 100644 src/gallium/drivers/nv50/nv50_state.c create mode 100644 src/gallium/drivers/nv50/nv50_state.h create mode 100644 src/gallium/drivers/nv50/nv50_surface.c create mode 100644 src/gallium/drivers/nv50/nv50_vbo.c create mode 100644 src/gallium/winsys/dri/nouveau/Makefile create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_bo.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_channel.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_context.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_context.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_device.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_device.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_dma.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_dma.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_dri.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_drmif.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_fence.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_grobj.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_local.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_lock.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_notifier.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_resource.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_screen.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_screen.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h create mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c create mode 100644 src/gallium/winsys/dri/nouveau/nv04_surface.c create mode 100644 src/gallium/winsys/dri/nouveau/nv50_surface.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/Makefile delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c delete mode 100644 src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c delete mode 100644 src/mesa/pipe/nouveau/nouveau_bo.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_channel.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_class.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_gldefs.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_grobj.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_notifier.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_push.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_pushbuf.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_resource.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_stateobj.h delete mode 100644 src/mesa/pipe/nouveau/nouveau_winsys.h delete mode 100644 src/mesa/pipe/nv30/Makefile delete mode 100644 src/mesa/pipe/nv30/nv30_clear.c delete mode 100644 src/mesa/pipe/nv30/nv30_context.c delete mode 100644 src/mesa/pipe/nv30/nv30_context.h delete mode 100644 src/mesa/pipe/nv30/nv30_draw.c delete mode 100644 src/mesa/pipe/nv30/nv30_fragprog.c delete mode 100644 src/mesa/pipe/nv30/nv30_fragtex.c delete mode 100644 src/mesa/pipe/nv30/nv30_miptree.c delete mode 100644 src/mesa/pipe/nv30/nv30_query.c delete mode 100644 src/mesa/pipe/nv30/nv30_shader.h delete mode 100644 src/mesa/pipe/nv30/nv30_state.c delete mode 100644 src/mesa/pipe/nv30/nv30_state.h delete mode 100644 src/mesa/pipe/nv30/nv30_state_emit.c delete mode 100644 src/mesa/pipe/nv30/nv30_surface.c delete mode 100644 src/mesa/pipe/nv30/nv30_vbo.c delete mode 100644 src/mesa/pipe/nv30/nv30_vertprog.c delete mode 100644 src/mesa/pipe/nv40/Makefile delete mode 100644 src/mesa/pipe/nv40/nv40_clear.c delete mode 100644 src/mesa/pipe/nv40/nv40_context.c delete mode 100644 src/mesa/pipe/nv40/nv40_context.h delete mode 100644 src/mesa/pipe/nv40/nv40_dma.h delete mode 100644 src/mesa/pipe/nv40/nv40_draw.c delete mode 100644 src/mesa/pipe/nv40/nv40_fragprog.c delete mode 100644 src/mesa/pipe/nv40/nv40_fragtex.c delete mode 100644 src/mesa/pipe/nv40/nv40_miptree.c delete mode 100644 src/mesa/pipe/nv40/nv40_query.c delete mode 100644 src/mesa/pipe/nv40/nv40_shader.h delete mode 100644 src/mesa/pipe/nv40/nv40_state.c delete mode 100644 src/mesa/pipe/nv40/nv40_state.h delete mode 100644 src/mesa/pipe/nv40/nv40_state_emit.c delete mode 100644 src/mesa/pipe/nv40/nv40_surface.c delete mode 100644 src/mesa/pipe/nv40/nv40_vbo.c delete mode 100644 src/mesa/pipe/nv40/nv40_vertprog.c delete mode 100644 src/mesa/pipe/nv50/Makefile delete mode 100644 src/mesa/pipe/nv50/nv50_clear.c delete mode 100644 src/mesa/pipe/nv50/nv50_context.c delete mode 100644 src/mesa/pipe/nv50/nv50_context.h delete mode 100644 src/mesa/pipe/nv50/nv50_draw.c delete mode 100644 src/mesa/pipe/nv50/nv50_miptree.c delete mode 100644 src/mesa/pipe/nv50/nv50_query.c delete mode 100644 src/mesa/pipe/nv50/nv50_state.c delete mode 100644 src/mesa/pipe/nv50/nv50_state.h delete mode 100644 src/mesa/pipe/nv50/nv50_surface.c delete mode 100644 src/mesa/pipe/nv50/nv50_vbo.c diff --git a/src/gallium/drivers/Makefile b/src/gallium/drivers/Makefile index c0345a9cb54..58df6c50093 100644 --- a/src/gallium/drivers/Makefile +++ b/src/gallium/drivers/Makefile @@ -6,7 +6,8 @@ ifeq ($(CONFIG_NAME), linux-cell) CELL_DIR = cell endif -SUBDIRS = softpipe i915simple i965simple failover pipebuffer $(CELL_DIR) +SUBDIRS = softpipe i915simple i965simple nv30 nv40 nv50 \ + failover pipebuffer $(CELL_DIR) default: subdirs diff --git a/src/gallium/drivers/nouveau/nouveau_bo.h b/src/gallium/drivers/nouveau/nouveau_bo.h new file mode 100644 index 00000000000..18020e9c652 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_bo.h @@ -0,0 +1,51 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_BO_H__ +#define __NOUVEAU_BO_H__ + +/* Relocation/Buffer type flags */ +#define NOUVEAU_BO_VRAM (1 << 0) +#define NOUVEAU_BO_GART (1 << 1) +#define NOUVEAU_BO_RD (1 << 2) +#define NOUVEAU_BO_WR (1 << 3) +#define NOUVEAU_BO_RDWR (NOUVEAU_BO_RD | NOUVEAU_BO_WR) +#define NOUVEAU_BO_MAP (1 << 4) +#define NOUVEAU_BO_PIN (1 << 5) +#define NOUVEAU_BO_LOW (1 << 6) +#define NOUVEAU_BO_HIGH (1 << 7) +#define NOUVEAU_BO_OR (1 << 8) +#define NOUVEAU_BO_LOCAL (1 << 9) +#define NOUVEAU_BO_DUMMY (1 << 31) + +struct nouveau_bo { + struct nouveau_device *device; + uint64_t handle; + + uint64_t size; + void *map; + + uint32_t flags; + uint64_t offset; +}; + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_channel.h b/src/gallium/drivers/nouveau/nouveau_channel.h new file mode 100644 index 00000000000..b99de9add86 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_channel.h @@ -0,0 +1,39 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_CHANNEL_H__ +#define __NOUVEAU_CHANNEL_H__ + +struct nouveau_channel { + struct nouveau_device *device; + int id; + + struct nouveau_pushbuf *pushbuf; + + struct nouveau_grobj *vram; + struct nouveau_grobj *gart; + + void *user_private; + void (*hang_notify)(struct nouveau_channel *); +}; + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h new file mode 100644 index 00000000000..5998945677b --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -0,0 +1,6134 @@ +/************************************************************************* + + Autogenerated file, do not edit ! + +************************************************************************** + + Copyright (C) 2006-2007 : + Dmitry Baryshkov, + Laurent Carlier, + Matthieu Castet, + Dawid Gajownik, + Jeremy Kolb, + Stephane Loeuillet, + Patrice Mandin, + Stephane Marchesin, + Serge Martin, + Sylvain Munaut, + Simon Raffeiner, + Ben Skeggs, + Erik Waling, + koala_br, + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +*************************************************************************/ + + +#ifndef NOUVEAU_REG_H +#define NOUVEAU_REG_H 1 + + +#define NV01_ROOT 0x00000001 + + + +#define NV01_CONTEXT_DMA 0x00000002 + + + +#define NV01_DEVICE 0x00000003 + + + +#define NV01_TIMER 0x00000004 + +#define NV01_TIMER_SYNCHRONIZE 0x00000100 +#define NV01_TIMER_STOP_ALARM 0x00000104 +#define NV01_TIMER_DMA_NOTIFY 0x00000180 +#define NV01_TIMER_TIME(x) (0x00000300+((x)*4)) +#define NV01_TIMER_TIME__SIZE 0x00000002 +#define NV01_TIMER_ALARM_NOTIFY 0x00000308 + + +#define NV_IMAGE_STENCIL 0x00000010 + +#define NV_IMAGE_STENCIL_NOTIFY 0x00000104 +#define NV_IMAGE_STENCIL_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_STENCIL_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_STENCIL_IMAGE_INPUT(x) (0x00000204+((x)*4)) +#define NV_IMAGE_STENCIL_IMAGE_INPUT__SIZE 0x00000002 + + +#define NV_IMAGE_BLEND_AND 0x00000011 + +#define NV_IMAGE_BLEND_AND_NOP 0x00000100 +#define NV_IMAGE_BLEND_AND_NOTIFY 0x00000104 +#define NV_IMAGE_BLEND_AND_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_BLEND_AND_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_BLEND_AND_BETA_INPUT 0x00000204 +#define NV_IMAGE_BLEND_AND_IMAGE_INPUT 0x00000208 + + +#define NV01_CONTEXT_BETA1 0x00000012 + +#define NV01_CONTEXT_BETA1_NOP 0x00000100 +#define NV01_CONTEXT_BETA1_NOTIFY 0x00000104 +#define NV01_CONTEXT_BETA1_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_BETA1_BETA_1D31 0x00000300 + + +#define NV_IMAGE_ROP_AND 0x00000013 + +#define NV_IMAGE_ROP_AND_NOTIFY 0x00000104 +#define NV_IMAGE_ROP_AND_DMA_NOTIFY 0x00000180 +#define NV_IMAGE_ROP_AND_IMAGE_OUTPUT 0x00000200 +#define NV_IMAGE_ROP_AND_ROP_INPUT 0x00000204 +#define NV_IMAGE_ROP_AND_IMAGE_INPUT(x) (0x00000208+((x)*4)) +#define NV_IMAGE_ROP_AND_IMAGE_INPUT__SIZE 0x00000002 + + +#define NV_IMAGE_COLOR_KEY 0x00000015 + + + +#define NV01_CONTEXT_COLOR_KEY 0x00000017 + +#define NV01_CONTEXT_COLOR_KEY_NOP 0x00000100 +#define NV01_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 +#define NV01_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A8Y8 0x00000001 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X24Y8 0x00000002 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A1R5G5B5 0x00000003 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X17R5G5B5 0x00000004 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A8R8G8B8 0x00000005 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X8R8G8B8 0x00000006 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A16Y16 0x00000007 +#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16Y16 0x00000008 +#define NV01_CONTEXT_COLOR_KEY_COLOR 0x00000304 + + +#define NV01_CONTEXT_PATTERN 0x00000018 + +#define NV01_CONTEXT_PATTERN_NOP 0x00000100 +#define NV01_CONTEXT_PATTERN_NOTIFY 0x00000104 +#define NV01_CONTEXT_PATTERN_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_PATTERN_COLOR_FORMAT 0x00000300 +#define NV01_CONTEXT_PATTERN_MONOCHROME_FORMAT 0x00000304 +#define NV01_CONTEXT_PATTERN_SHAPE 0x00000308 +#define NV01_CONTEXT_PATTERN_COLOR(x) (0x00000310+((x)*4)) +#define NV01_CONTEXT_PATTERN_COLOR__SIZE 0x00000002 +#define NV01_CONTEXT_PATTERN_PATTERN(x) (0x00000318+((x)*4)) +#define NV01_CONTEXT_PATTERN_PATTERN__SIZE 0x00000002 + + +#define NV01_CONTEXT_CLIP_RECTANGLE 0x00000019 + +#define NV01_CONTEXT_CLIP_RECTANGLE_NOP 0x00000100 +#define NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY 0x00000104 +#define NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY 0x00000180 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT 0x00000300 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_SHIFT 0 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_MASK 0x0000ffff +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_SHIFT 16 +#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_MASK 0xffff0000 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE 0x00000304 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_SHIFT 0 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_MASK 0x0000ffff +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_SHIFT 16 +#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_LINE 0x0000001c + +#define NV01_RENDER_SOLID_LINE_NOP 0x00000100 +#define NV01_RENDER_SOLID_LINE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_LINE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_LINE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_LINE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_LINE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_LINE_ROP 0x0000018c +#define NV01_RENDER_SOLID_LINE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_LINE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_LINE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_LINE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY 0x00000003 +#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A8Y8 0x00000001 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X24Y8 0x00000002 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A1R5G5B5 0x00000003 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X17R5G5B5 0x00000004 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A8R8G8B8 0x00000005 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X8R8G8B8 0x00000006 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A16Y16 0x00000007 +#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16Y16 0x00000008 +#define NV01_RENDER_SOLID_LINE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0(x) (0x00000400+((x)*8)) +#define NV01_RENDER_SOLID_LINE_LINE_POINT0__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1(x) (0x00000404+((x)*8)) +#define NV01_RENDER_SOLID_LINE_LINE_POINT1__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X(x) (0x00000480+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y(x) (0x00000484+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X(x) (0x00000488+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y(x) (0x0000048c+((x)*16)) +#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_POLYLINE(x) (0x00000500+((x)*4)) +#define NV01_RENDER_SOLID_LINE_POLYLINE__SIZE 0x00000020 +#define NV01_RENDER_SOLID_LINE_POLYLINE_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_POLYLINE_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X(x) (0x00000580+((x)*8)) +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y(x) (0x00000584+((x)*8)) +#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR(x) (0x00000600+((x)*8)) +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT(x) (0x00000604+((x)*8)) +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_TRIANGLE 0x0000001d + +#define NV01_RENDER_SOLID_TRIANGLE_NOP 0x00000100 +#define NV01_RENDER_SOLID_TRIANGLE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_TRIANGLE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_TRIANGLE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_TRIANGLE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_TRIANGLE_ROP 0x0000018c +#define NV01_RENDER_SOLID_TRIANGLE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_TRIANGLE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY 0x00000003 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_RENDER_SOLID_TRIANGLE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_TRIANGLE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0 0x00000310 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1 0x00000314 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2 0x00000318 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_X 0x00000320 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_Y 0x00000324 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_X 0x00000328 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_Y 0x0000032c +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_X 0x00000330 +#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_Y 0x00000334 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH(x) (0x00000400+((x)*4)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH__SIZE 0x00000020 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X(x) (0x00000480+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y(x) (0x00000484+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR(x) (0x00000500+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0(x) (0x00000504+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1(x) (0x00000508+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2(x) (0x0000050c+((x)*16)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2__SIZE 0x00000008 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR(x) (0x00000580+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT(x) (0x00000584+((x)*8)) +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_MASK 0xffff0000 + + +#define NV01_RENDER_SOLID_RECTANGLE 0x0000001e + +#define NV01_RENDER_SOLID_RECTANGLE_NOP 0x00000100 +#define NV01_RENDER_SOLID_RECTANGLE_NOTIFY 0x00000104 +#define NV01_RENDER_SOLID_RECTANGLE_PATCH 0x0000010c +#define NV01_RENDER_SOLID_RECTANGLE_DMA_NOTIFY 0x00000180 +#define NV01_RENDER_SOLID_RECTANGLE_CLIP_RECTANGLE 0x00000184 +#define NV01_RENDER_SOLID_RECTANGLE_PATTERN 0x00000188 +#define NV01_RENDER_SOLID_RECTANGLE_ROP 0x0000018c +#define NV01_RENDER_SOLID_RECTANGLE_BETA1 0x00000190 +#define NV01_RENDER_SOLID_RECTANGLE_SURFACE 0x00000194 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION 0x000002fc +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_ROP_AND 0x00000001 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_AND 0x00000002 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY 0x00000003 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_RENDER_SOLID_RECTANGLE_COLOR_FORMAT 0x00000300 +#define NV01_RENDER_SOLID_RECTANGLE_COLOR 0x00000304 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT(x) (0x00000400+((x)*8)) +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT__SIZE 0x00000010 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_SHIFT 0 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_MASK 0x0000ffff +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_SHIFT 16 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_MASK 0xffff0000 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE__SIZE 0x00000010 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_SHIFT 0 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_MASK 0x0000ffff +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_SHIFT 16 +#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_MASK 0xffff0000 + + +#define NV01_IMAGE_BLIT 0x0000001f + +#define NV01_IMAGE_BLIT_NOP 0x00000100 +#define NV01_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV01_IMAGE_BLIT_PATCH 0x0000010c +#define NV01_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV01_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV01_IMAGE_BLIT_PATTERN 0x0000018c +#define NV01_IMAGE_BLIT_ROP 0x00000190 +#define NV01_IMAGE_BLIT_BETA1 0x00000194 +#define NV01_IMAGE_BLIT_SURFACE 0x0000019c +#define NV01_IMAGE_BLIT_OPERATION 0x000002fc +#define NV01_IMAGE_BLIT_IMAGE_INPUT 0x00000204 +#define NV01_IMAGE_BLIT_POINT_IN 0x00000300 +#define NV01_IMAGE_BLIT_POINT_IN_X_SHIFT 0 +#define NV01_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 +#define NV01_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 +#define NV01_IMAGE_BLIT_POINT_OUT 0x00000304 +#define NV01_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 +#define NV01_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 +#define NV01_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 +#define NV01_IMAGE_BLIT_SIZE 0x00000308 +#define NV01_IMAGE_BLIT_SIZE_W_SHIFT 0 +#define NV01_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff +#define NV01_IMAGE_BLIT_SIZE_H_SHIFT 16 +#define NV01_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 + + +#define NV01_IMAGE_FROM_CPU 0x00000021 + +#define NV01_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV01_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV01_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV01_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV01_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 +#define NV01_IMAGE_FROM_CPU_PATTERN 0x0000018c +#define NV01_IMAGE_FROM_CPU_ROP 0x00000190 +#define NV01_IMAGE_FROM_CPU_BETA1 0x00000194 +#define NV01_IMAGE_FROM_CPU_SURFACE 0x00000198 +#define NV01_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_AND 0x00000000 +#define NV01_IMAGE_FROM_CPU_OPERATION_ROP_AND 0x00000001 +#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_AND 0x00000002 +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY 0x00000003 +#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_PREMULT 0x00000005 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_Y8 0x00000001 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A1R5G5B5 0x00000002 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X1R5G5B5 0x00000003 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A8R8G8B8 0x00000004 +#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X8R8G8B8 0x00000005 +#define NV01_IMAGE_FROM_CPU_POINT 0x00000304 +#define NV01_IMAGE_FROM_CPU_POINT_X_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_POINT_X_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_POINT_Y_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_POINT_Y_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_SIZE_IN 0x0000030c +#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 +#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff +#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 +#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 +#define NV01_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV01_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 + + +#define NV01_NULL 0x00000030 + + + +#define NV03_STRETCHED_IMAGE_FROM_CPU 0x00000036 + +#define NV03_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV03_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV03_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV03_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 +#define NV03_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 +#define NV03_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c +#define NV03_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000194 +#define NV03_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_DX_DU 0x00000308 +#define NV03_STRETCHED_IMAGE_FROM_CPU_DY_DV 0x0000030c +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT 0x00000310 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE 0x00000314 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_SHIFT 0 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_MASK 0x0000ffff +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_SHIFT 16 +#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_MASK 0xffff0000 +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 + + +#define NV03_SCALED_IMAGE_FROM_MEMORY 0x00000037 + +#define NV03_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 +#define NV03_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 +#define NV03_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 +#define NV03_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c +#define NV03_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 +#define NV03_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000194 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 +#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY 0x00000003 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000005 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT 0x00000310 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE 0x00000314 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DU_DX 0x00000318 +#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DV_DY 0x0000031c +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE 0x00000400 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_MASK 0xffff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT 0x00000404 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_MASK 0x00ff0000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CENTER 0x00010000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CORNER 0x00020000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_SHIFT 24 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_MASK 0xff000000 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_OFFSET 0x00000408 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT 0x0000040c +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_SHIFT 0 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_MASK 0x0000ffff +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_SHIFT 16 +#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_MASK 0xffff0000 + + +#define NV04_DVD_SUBPICTURE 0x00000038 + +#define NV04_DVD_SUBPICTURE_NOP 0x00000100 +#define NV04_DVD_SUBPICTURE_NOTIFY 0x00000104 +#define NV04_DVD_SUBPICTURE_WAIT_FOR_IDLE 0x00000108 +#define NV04_DVD_SUBPICTURE_DMA_NOTIFY 0x00000180 +#define NV04_DVD_SUBPICTURE_DMA_OVERLAY 0x00000184 +#define NV04_DVD_SUBPICTURE_DMA_IMAGEIN 0x00000188 +#define NV04_DVD_SUBPICTURE_DMA_IMAGEOUT 0x0000018c +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT 0x00000300 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE 0x00000304 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT 0x00000308 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEOUT_OFFSET 0x0000030c +#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DU_DX 0x00000310 +#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DV_DY 0x00000314 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE 0x00000318 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT 0x0000031c +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_IMAGEIN_OFFSET 0x00000320 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT 0x00000324 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_SHIFT 0 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_SHIFT 16 +#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DU_DX 0x00000328 +#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DV_DY 0x0000032c +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE 0x00000330 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT 0x00000334 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_MASK 0xffff0000 +#define NV04_DVD_SUBPICTURE_OVERLAY_OFFSET 0x00000338 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT 0x0000033c +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_SHIFT 0 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_MASK 0x0000ffff +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_SHIFT 16 +#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK 0xffff0000 + + +#define NV04_MEMORY_TO_MEMORY_FORMAT 0x00000039 + +#define NV04_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 +#define NV04_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 +#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 +#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 +#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 +#define NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c +#define NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 +#define NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 +#define NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 +#define NV04_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c +#define NV04_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 +#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 +#define NV04_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 + + +#define NV01_MEMORY_LOCAL_BANKED 0x0000003d + + + +#define NV01_MAPPING_SYSTEM 0x0000003e + + + +#define NV03_MEMORY_LOCAL_CURSOR 0x0000003f + + + +#define NV01_MEMORY_LOCAL_LINEAR 0x00000040 + + + +#define NV01_MAPPING_LOCAL 0x00000041 + + + +#define NV04_CONTEXT_SURFACES_2D 0x00000042 + +#define NV04_CONTEXT_SURFACES_2D_NOP 0x00000100 +#define NV04_CONTEXT_SURFACES_2D_NOTIFY 0x00000104 +#define NV04_CONTEXT_SURFACES_2D_PM_TRIGGER 0x00000140 +#define NV04_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE 0x00000184 +#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_DESTIN 0x00000188 +#define NV04_CONTEXT_SURFACES_2D_FORMAT 0x00000300 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y8 0x00000001 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_Z1R5G5B5 0x00000002 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_X1R5G5B5 0x00000003 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5 0x00000004 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y16 0x00000005 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_Z8R8G8B8 0x00000006 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_X8R8G8B8 0x00000007 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_X1A7R8G8B8 0x00000009 +#define NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8 0x0000000a +#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y32 0x0000000b +#define NV04_CONTEXT_SURFACES_2D_PITCH 0x00000304 +#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 +#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 +#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 +#define NV04_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c + + +#define NV03_CONTEXT_ROP 0x00000043 + +#define NV03_CONTEXT_ROP_NOP 0x00000100 +#define NV03_CONTEXT_ROP_NOTIFY 0x00000104 +#define NV03_CONTEXT_ROP_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_ROP_ROP 0x00000300 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SHIFT 0 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_MASK 0x0000000f +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_CLEAR 0x00000000 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOR 0x00000001 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_INVERTED 0x00000002 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY_INVERTED 0x00000003 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_REVERSE 0x00000004 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_INVERT 0x00000005 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_XOR 0x00000006 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NAND 0x00000007 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND 0x00000008 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_EQUI 0x00000009 +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOOP 0x0000000a +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_INVERTED 0x0000000b +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY 0x0000000c +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_REVERSE 0x0000000d +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR 0x0000000e +#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SET 0x0000000f +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SHIFT 4 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_MASK 0x000000f0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_CLEAR 0x00000000 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOR 0x00000010 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_INVERTED 0x00000020 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY_INVERTED 0x00000030 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_REVERSE 0x00000040 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_INVERT 0x00000050 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_XOR 0x00000060 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NAND 0x00000070 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND 0x00000080 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_EQUI 0x00000090 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOOP 0x000000a0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_INVERTED 0x000000b0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY 0x000000c0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_REVERSE 0x000000d0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR 0x000000e0 +#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SET 0x000000f0 + + +#define NV04_IMAGE_PATTERN 0x00000044 + +#define NV04_IMAGE_PATTERN_NOP 0x00000100 +#define NV04_IMAGE_PATTERN_NOTIFY 0x00000104 +#define NV04_IMAGE_PATTERN_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT 0x00000300 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A16R5G6B5 0x00000001 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_X16A1R5G5B5 0x00000002 +#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT 0x00000304 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_CGA6 0x00000001 +#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_LE 0x00000002 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE 0x00000308 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_8X8 0x00000000 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_64X1 0x00000001 +#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_1X64 0x00000002 +#define NV04_IMAGE_PATTERN_PATTERN_SELECT 0x0000030c +#define NV04_IMAGE_PATTERN_PATTERN_SELECT_MONO 0x00000001 +#define NV04_IMAGE_PATTERN_PATTERN_SELECT_COLOR 0x00000002 +#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR0 0x00000310 +#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR1 0x00000314 +#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0 0x00000318 +#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1 0x0000031c +#define NV04_IMAGE_PATTERN_PATTERN_Y8(x) (0x00000400+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_Y8__SIZE 0x00000010 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_MASK 0x000000ff +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_SHIFT 8 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_MASK 0x0000ff00 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_MASK 0x00ff0000 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_SHIFT 24 +#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_MASK 0xff000000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5(x) (0x00000500+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5__SIZE 0x00000020 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_MASK 0x0000001f +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_SHIFT 5 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_MASK 0x000007e0 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_SHIFT 11 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_MASK 0x0000f800 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_MASK 0x001f0000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_SHIFT 21 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_MASK 0x07e00000 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_SHIFT 27 +#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_MASK 0xf8000000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5(x) (0x00000600+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5__SIZE 0x00000020 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_MASK 0x0000001f +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_SHIFT 5 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_MASK 0x000003e0 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_SHIFT 10 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_MASK 0x00007c00 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_MASK 0x001f0000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_SHIFT 21 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_MASK 0x03e00000 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_SHIFT 26 +#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_MASK 0x7c000000 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8(x) (0x00000700+((x)*4)) +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8__SIZE 0x00000040 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_SHIFT 0 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_MASK 0x000000ff +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_SHIFT 8 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_MASK 0x0000ff00 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_SHIFT 16 +#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_MASK 0x00ff0000 + + +#define NV03_VIDEO_LUT_CURSOR_DAC 0x00000046 + +#define NV03_VIDEO_LUT_CURSOR_DAC_SYNCHRONIZE 0x00000100 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_IMAGE 0x00000104 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_CURSOR 0x00000108 +#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_DAC 0x0000010c +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_NOTIFY 0x00000180 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE(x) (0x00000184+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT(x) (0x0000018c+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR(x) (0x00000194+((x)*4)) +#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_GET 0x000002fc +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET(x) (0x00000300+((x)*8)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT(x) (0x00000304+((x)*8)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET(x) (0x00000340+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT(x) (0x00000344+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT(x) (0x00000348+((x)*12)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A 0x00000358 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE(x) (0x00000380+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_MASK 0xffff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC(x) (0x00000384+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC(x) (0x00000388+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE(x) (0x0000038c+((x)*16)) +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE__SIZE 0x00000002 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_SHIFT 0 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_MASK 0x0000ffff +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_SHIFT 16 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_MASK 0x0fff0000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_SHIFT 28 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_MASK 0xf0000000 +#define NV03_VIDEO_LUT_CURSOR_DAC_SET_PIXEL_CLOCK 0x000003a0 + + +#define NV03_DX3_TEXTURED_TRIANGLE 0x00000048 + +#define NV03_DX3_TEXTURED_TRIANGLE_NOP 0x00000100 +#define NV03_DX3_TEXTURED_TRIANGLE_NOTIFY 0x00000104 +#define NV03_DX3_TEXTURED_TRIANGLE_PATCH 0x0000010c +#define NV03_DX3_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV03_DX3_TEXTURED_TRIANGLE_DMA_TEXTURE 0x00000184 +#define NV03_DX3_TEXTURED_TRIANGLE_CLIP_RECTANGLE 0x00000188 +#define NV03_DX3_TEXTURED_TRIANGLE_SURFACE 0x0000018c +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_OFFSET 0x00000304 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT 0x00000308 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_MASK 0x0000ffff +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_MASK 0x0f000000 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_SHIFT 28 +#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_MASK 0xf0000000 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER 0x0000030c +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_MASK 0x0000001f +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_MASK 0x00001f00 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_MASK 0x00ff0000 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR 0x00000310 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_MASK 0x000000ff +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_MASK 0x0000ff00 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_MASK 0x00ff0000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT 0x00000314 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_MASK 0x0000000f +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_SHIFT 4 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_MASK 0x00000030 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_SHIFT 6 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_MASK 0x000000c0 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_MASK 0x00000f00 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_SHIFT 12 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_MASK 0x00007000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_PERSPECTIVE_ENABLE (1 << 15) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_MASK 0x07000000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_SHIFT 27 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_MASK 0x18000000 +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_BETA (1 << 29) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_DST_BLEND (1 << 30) +#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SRC_BLEND (1 << 31) +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL 0x00000318 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_MASK 0x000000ff +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_MASK 0xffffff00 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR(x) (0x00001000+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_SHIFT 0 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_MASK 0x0000000f +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_SHIFT 4 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_MASK 0x000000f0 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_SHIFT 8 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_MASK 0x00000f00 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_SHIFT 12 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_MASK 0x0000f000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_SHIFT 16 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_MASK 0x000f0000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_SHIFT 20 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_MASK 0x00f00000 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_SHIFT 24 +#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_MASK 0xff000000 +#define NV03_DX3_TEXTURED_TRIANGLE_COLOR(x) (0x00001004+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_COLOR__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_X(x) (0x00001008+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_X__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_Y(x) (0x0000100c+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_Y__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_Z(x) (0x00001010+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_Z__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_M(x) (0x00001014+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_M__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_U(x) (0x00001018+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_U__SIZE 0x00000040 +#define NV03_DX3_TEXTURED_TRIANGLE_V(x) (0x0000101c+((x)*32)) +#define NV03_DX3_TEXTURED_TRIANGLE_V__SIZE 0x00000040 + + +#define NV04_GDI_RECTANGLE_TEXT 0x0000004a + +#define NV04_GDI_RECTANGLE_TEXT_NOP 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 +#define NV04_GDI_RECTANGLE_TEXT_PATCH 0x0000010c +#define NV04_GDI_RECTANGLE_TEXT_PM_TRIGGER 0x00000140 +#define NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 +#define NV04_GDI_RECTANGLE_TEXT_DMA_FONTS 0x00000184 +#define NV04_GDI_RECTANGLE_TEXT_PATTERN 0x00000188 +#define NV04_GDI_RECTANGLE_TEXT_ROP 0x0000018c +#define NV04_GDI_RECTANGLE_TEXT_BETA1 0x00000190 +#define NV04_GDI_RECTANGLE_TEXT_BETA4 0x00000194 +#define NV04_GDI_RECTANGLE_TEXT_SURFACE 0x00000198 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_AND 0x00000000 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_ROP_AND 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_AND 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY 0x00000003 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_PREMULT 0x00000005 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_X16A1R5G5B5 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_CGA6 0x00000001 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE 0x00000002 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(x) (0x00000400+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0 0x000005f4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1 0x000005f8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_B 0x000005fc +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0(x) (0x00000600+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1(x) (0x00000604+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1__SIZE 0x00000020 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x000007ec +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x000007f0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_C 0x000007f4 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C 0x000007f8 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C 0x000007fc +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000800+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000080 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x00000be4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x00000be8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR0_E 0x00000bec +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_E 0x00000bf0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x00000bf4 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x00000bf8 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E 0x00000bfc +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00000c00+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000080 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F 0x00000ff0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_MASK 0x0fffffff +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_SHIFT 28 +#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_MASK 0xf0000000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0 0x00000ff4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1 0x00000ff8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_F 0x00000ffc +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F(x) (0x00001000+((x)*4)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F__SIZE 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_MASK 0x000000ff +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_SHIFT 8 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_MASK 0x000fff00 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_SHIFT 20 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_MASK 0xfff00000 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G 0x000017f0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_MASK 0x0fffffff +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_SHIFT 28 +#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_MASK 0xf0000000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0 0x000017f4 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1 0x000017f8 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_COLOR1_G 0x000017fc +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT(x) (0x00001800+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT__SIZE 0x00000100 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_SHIFT 0 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_MASK 0x0000ffff +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_SHIFT 16 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_MASK 0xffff0000 +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX(x) (0x00001804+((x)*8)) +#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX__SIZE 0x00000100 + + +#define NV03_GDI_RECTANGLE_TEXT 0x0000004b + +#define NV03_GDI_RECTANGLE_TEXT_NOP 0x00000100 +#define NV03_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 +#define NV03_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 +#define NV03_GDI_RECTANGLE_TEXT_PATTERN 0x00000184 +#define NV03_GDI_RECTANGLE_TEXT_ROP 0x00000188 +#define NV03_GDI_RECTANGLE_TEXT_BETA1 0x0000018c +#define NV03_GDI_RECTANGLE_TEXT_SURFACE 0x00000190 +#define NV03_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc +#define NV03_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT 0x00000400 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE 0x00000404 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B 0x000007f4 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B 0x000007f8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_B 0x000007fc +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0 0x00000800 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1 0x00000804 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x00000bec +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x00000bf0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_C 0x00000bf4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C 0x00000bf8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C 0x00000bfc +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000c00+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000020 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0 0x00000fe8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1 0x00000fec +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_D 0x00000ff0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D 0x00000ff4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D 0x00000ff8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D 0x00000ffc +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D(x) (0x00001000+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D__SIZE 0x00000020 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x000013e4 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x000013e8 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_COLOR0_E 0x000013ec +#define NV03_GDI_RECTANGLE_TEXT_COLOR1_E 0x000013f0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x000013f4 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x000013f8 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E 0x000013fc +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 +#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00001400+((x)*4)) +#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000020 + + +#define NV04_SWIZZLED_SURFACE 0x00000052 + +#define NV04_SWIZZLED_SURFACE_NOP 0x00000100 +#define NV04_SWIZZLED_SURFACE_NOTIFY 0x00000104 +#define NV04_SWIZZLED_SURFACE_DMA_NOTIFY 0x00000180 +#define NV04_SWIZZLED_SURFACE_DMA_IMAGE 0x00000184 +#define NV04_SWIZZLED_SURFACE_FORMAT 0x00000300 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_SHIFT 0 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK 0x000000ff +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y8 0x00000001 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000002 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000003 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_R5G6B5 0x00000004 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y16 0x00000005 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000006 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000007 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000009 +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8 0x0000000a +#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y32 0x0000000b +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_MASK 0xff000000 +#define NV04_SWIZZLED_SURFACE_OFFSET 0x00000304 + + +#define NV04_CONTEXT_SURFACES_3D 0x00000053 + +#define NV04_CONTEXT_SURFACES_3D_NOP 0x00000100 +#define NV04_CONTEXT_SURFACES_3D_NOTIFY 0x00000104 +#define NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 +#define NV04_CONTEXT_SURFACES_3D_DMA_COLOR 0x00000184 +#define NV04_CONTEXT_SURFACES_3D_DMA_ZETA 0x00000188 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL 0x000002f8 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL 0x000002fc +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_FORMAT 0x00000300 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK 0x000000ff +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000001 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000002 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000004 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000005 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000006 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000007 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT 8 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK 0x0000ff00 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_PITCH 0x00000100 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SWIZZLE 0x00000200 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_MASK 0xff000000 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE 0x00000304 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_PITCH 0x00000308 +#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_SHIFT 0 +#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK 0x0000ffff +#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT 16 +#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK 0xffff0000 +#define NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x0000030c +#define NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000310 + + +#define NV04_DX5_TEXTURED_TRIANGLE 0x00000054 + +#define NV04_DX5_TEXTURED_TRIANGLE_NOP 0x00000100 +#define NV04_DX5_TEXTURED_TRIANGLE_NOTIFY 0x00000104 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_A 0x00000184 +#define NV04_DX5_TEXTURED_TRIANGLE_DMA_B 0x00000188 +#define NV04_DX5_TEXTURED_TRIANGLE_SURFACE 0x0000018c +#define NV04_DX5_TEXTURED_TRIANGLE_COLORKEY 0x00000300 +#define NV04_DX5_TEXTURED_TRIANGLE_OFFSET 0x00000304 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT 0x00000308 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_MASK 0x00000003 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_SHIFT 2 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_MASK 0x0000000c +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CENTER 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER 0x00000020 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CENTER 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER 0x00000080 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_Y8 0x00000100 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A1R5G5B5 0x00000200 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X1R5G5B5 0x00000300 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A4R4G4B4 0x00000400 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_R5G6B5 0x00000500 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A8R8G8B8 0x00000600 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X8R8G8B8 0x00000700 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT 0x01000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT 0x02000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE 0x03000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER 0x04000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP 0x05000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPU (1 << 27) +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_REPEAT 0x10000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MIRRORED_REPEAT 0x20000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE 0x30000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_BORDER 0x40000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP 0x50000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPV (1 << 31) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER 0x0000030c +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST 0x01000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR 0x02000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST 0x10000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR 0x20000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND 0x00000310 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_MASK 0x0000000f +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD 0x00000080 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_PHONG 0x000000c0 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK 0x0f000000 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT 28 +#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK 0xf0000000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL 0x00000314 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE (1 << 12) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN (1 << 13) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT 14 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_MASK 0x0000c000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_MASK 0x00300000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE (1 << 22) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE (1 << 23) +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_MASK 0x3f000000 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT 30 +#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_MASK 0xc0000000 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR 0x00000318 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(x) (0x00000400+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY(x) (0x00000404+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ(x) (0x00000408+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW(x) (0x0000040c+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR(x) (0x00000410+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR(x) (0x00000414+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_MASK 0x000000ff +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_MASK 0x0000ff00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_MASK 0x00ff0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_SHIFT 24 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_MASK 0xff000000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU(x) (0x00000418+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV(x) (0x0000041c+((x)*32)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV__SIZE 0x00000010 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(x) (0x00000600+((x)*4)) +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE__SIZE 0x00000040 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 +#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 + + +#define NV04_DX6_MULTITEX_TRIANGLE 0x00000055 + +#define NV04_DX6_MULTITEX_TRIANGLE_NOP 0x00000100 +#define NV04_DX6_MULTITEX_TRIANGLE_NOTIFY 0x00000104 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY 0x00000180 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_A 0x00000184 +#define NV04_DX6_MULTITEX_TRIANGLE_DMA_B 0x00000188 +#define NV04_DX6_MULTITEX_TRIANGLE_SURFACE 0x0000018c +#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET(x) (0x00000308+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT(x) (0x00000310+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPU (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPV (1 << 31) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER(x) (0x00000318+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER__SIZE 0x00000002 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 +#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA 0x00000320 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR 0x00000324 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA 0x0000032c +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR 0x00000330 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE0 (1 << 0) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA0 (1 << 1) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_SHIFT 2 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_MASK 0x000000fc +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE1 (1 << 8) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA1 (1 << 9) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_SHIFT 10 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_MASK 0x0000fc00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE2 (1 << 16) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA2 (1 << 17) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_SHIFT 18 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_MASK 0x00fc0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE3 (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA3 (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_SHIFT 26 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_MASK 0x1c000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_SHIFT 29 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_MASK 0xe0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR 0x00000334 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND 0x00000338 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK 0x0f000000 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT 28 +#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK 0xf0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0 0x0000033c +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_TEST_ENABLE (1 << 12) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ORIGIN (1 << 13) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_SHIFT 14 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_MASK 0x0000c000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_MASK 0x00300000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_DITHER_ENABLE (1 << 22) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_PERSPECTIVE_ENABLE (1 << 23) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_WRITE_ENABLE (1 << 24) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_ENABLE (1 << 25) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_ENABLE (1 << 26) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_ENABLE (1 << 27) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_ENABLE (1 << 28) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_ENABLE (1 << 29) +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_SHIFT 30 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_MASK 0xc0000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1 0x00000340 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2 0x00000344 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR 0x00000348 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX(x) (0x00000400+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY(x) (0x00000404+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ(x) (0x00000408+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW(x) (0x0000040c+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR(x) (0x00000410+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR(x) (0x00000414+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_MASK 0x000000ff +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_MASK 0x0000ff00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_MASK 0x00ff0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_SHIFT 24 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_MASK 0xff000000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0(x) (0x00000418+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0(x) (0x0000041c+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1(x) (0x00000420+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1(x) (0x00000424+((x)*40)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1__SIZE 0x00000008 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE(x) (0x00000540+((x)*4)) +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE__SIZE 0x00000030 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 +#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 + + +#define NV10TCL 0x00000056 + +#define NV10TCL_NOP 0x00000100 +#define NV10TCL_NOTIFY 0x00000104 +#define NV10TCL_DMA_NOTIFY 0x00000180 +#define NV10TCL_DMA_IN_MEMORY0 0x00000184 +#define NV10TCL_DMA_IN_MEMORY1 0x00000188 +#define NV10TCL_DISPLAY_LIST 0x0000018c +#define NV10TCL_DMA_IN_MEMORY2 0x00000194 +#define NV10TCL_DMA_IN_MEMORY3 0x00000198 +#define NV10TCL_VIEWPORT_HORIZ 0x00000200 +#define NV10TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV10TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV10TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV10TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV10TCL_VIEWPORT_VERT 0x00000204 +#define NV10TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV10TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV10TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV10TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV10TCL_BUFFER_FORMAT 0x00000208 +#define NV10TCL_BUFFER_PITCH 0x0000020c +#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 +#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff +#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 +#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 +#define NV10TCL_COLOR_OFFSET 0x00000210 +#define NV10TCL_ZETA_OFFSET 0x00000214 +#define NV10TCL_TX_OFFSET(x) (0x00000218+((x)*4)) +#define NV10TCL_TX_OFFSET__SIZE 0x00000002 +#define NV10TCL_TX_FORMAT(x) (0x00000220+((x)*4)) +#define NV10TCL_TX_FORMAT__SIZE 0x00000002 +#define NV10TCL_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV10TCL_TX_FORMAT_FORMAT_SHIFT 7 +#define NV10TCL_TX_FORMAT_FORMAT_MASK 0x00000780 +#define NV10TCL_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV10TCL_TX_FORMAT_FORMAT_A8 0x00000080 +#define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000100 +#define NV10TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000180 +#define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000200 +#define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000300 +#define NV10TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000380 +#define NV10TCL_TX_FORMAT_FORMAT_INDEX8 0x00000580 +#define NV10TCL_TX_FORMAT_FORMAT_DXT1 0x00000600 +#define NV10TCL_TX_FORMAT_FORMAT_DXT3 0x00000700 +#define NV10TCL_TX_FORMAT_FORMAT_DXT5 0x00000780 +#define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00000800 +#define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00000900 +#define NV10TCL_TX_FORMAT_FORMAT_L8_RECT 0x00000980 +#define NV10TCL_TX_FORMAT_FORMAT_A8L8 0x00000d00 +#define NV10TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00000d80 +#define NV10TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 +#define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00000e80 +#define NV10TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00001000 +#define NV10TCL_TX_FORMAT_FORMAT_A16 0x00001900 +#define NV10TCL_TX_FORMAT_FORMAT_A16_RECT 0x00001a80 +#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00002500 +#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00002580 +#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00002600 +#define NV10TCL_TX_FORMAT_NPOT (1 << 11) +#define NV10TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 +#define NV10TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 +#define NV10TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 16 +#define NV10TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 +#define NV10TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 20 +#define NV10TCL_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 +#define NV10TCL_TX_FORMAT_WRAP_S_SHIFT 24 +#define NV10TCL_TX_FORMAT_WRAP_S_MASK 0x0f000000 +#define NV10TCL_TX_FORMAT_WRAP_S_REPEAT 0x01000000 +#define NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x02000000 +#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x03000000 +#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x04000000 +#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP 0x05000000 +#define NV10TCL_TX_FORMAT_WRAP_T_SHIFT 28 +#define NV10TCL_TX_FORMAT_WRAP_T_MASK 0xf0000000 +#define NV10TCL_TX_FORMAT_WRAP_T_REPEAT 0x10000000 +#define NV10TCL_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x20000000 +#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x30000000 +#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x40000000 +#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP 0x50000000 +#define NV10TCL_TX_ENABLE(x) (0x00000228+((x)*4)) +#define NV10TCL_TX_ENABLE__SIZE 0x00000002 +#define NV10TCL_TX_ENABLE_ANISOTROPY_SHIFT 4 +#define NV10TCL_TX_ENABLE_ANISOTROPY_MASK 0x00000030 +#define NV10TCL_TX_ENABLE_ENABLE (1 << 30) +#define NV10TCL_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) +#define NV10TCL_TX_NPOT_PITCH__SIZE 0x00000002 +#define NV10TCL_TX_NPOT_PITCH_PITCH_SHIFT 16 +#define NV10TCL_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 +#define NV10TCL_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) +#define NV10TCL_TX_NPOT_SIZE__SIZE 0x00000002 +#define NV10TCL_TX_NPOT_SIZE_H_SHIFT 0 +#define NV10TCL_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV10TCL_TX_NPOT_SIZE_W_SHIFT 16 +#define NV10TCL_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV10TCL_TX_FILTER(x) (0x00000248+((x)*4)) +#define NV10TCL_TX_FILTER__SIZE 0x00000002 +#define NV10TCL_TX_FILTER_MINIFY_SHIFT 24 +#define NV10TCL_TX_FILTER_MINIFY_MASK 0x0f000000 +#define NV10TCL_TX_FILTER_MINIFY_NEAREST 0x01000000 +#define NV10TCL_TX_FILTER_MINIFY_LINEAR 0x02000000 +#define NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 +#define NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 +#define NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 +#define NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 +#define NV10TCL_TX_FILTER_MAGNIFY_SHIFT 28 +#define NV10TCL_TX_FILTER_MAGNIFY_MASK 0xf0000000 +#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST 0x10000000 +#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR 0x20000000 +#define NV10TCL_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) +#define NV10TCL_TX_PALETTE_OFFSET__SIZE 0x00000002 +#define NV10TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV10TCL_RC_IN_ALPHA__SIZE 0x00000002 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SHIFT 0 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f +#define NV10TCL_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) +#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SHIFT 5 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SHIFT 8 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_FOG 0x00000300 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SHIFT 13 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SHIFT 16 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_FOG 0x00030000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SHIFT 21 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SHIFT 24 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_FOG 0x03000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 +#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SHIFT 29 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_RC_IN_RGB(x) (0x00000268+((x)*4)) +#define NV10TCL_RC_IN_RGB__SIZE 0x00000002 +#define NV10TCL_RC_IN_RGB_D_INPUT_SHIFT 0 +#define NV10TCL_RC_IN_RGB_D_INPUT_MASK 0x0000000f +#define NV10TCL_RC_IN_RGB_D_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_IN_RGB_D_INPUT_FOG 0x00000003 +#define NV10TCL_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) +#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10TCL_RC_IN_RGB_D_MAPPING_SHIFT 5 +#define NV10TCL_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 +#define NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10TCL_RC_IN_RGB_C_INPUT_SHIFT 8 +#define NV10TCL_RC_IN_RGB_C_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_IN_RGB_C_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_IN_RGB_C_INPUT_FOG 0x00000300 +#define NV10TCL_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_SHIFT 13 +#define NV10TCL_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SHIFT 16 +#define NV10TCL_RC_IN_RGB_B_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_IN_RGB_B_INPUT_FOG 0x00030000 +#define NV10TCL_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_SHIFT 21 +#define NV10TCL_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SHIFT 24 +#define NV10TCL_RC_IN_RGB_A_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_FOG 0x03000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_SHIFT 29 +#define NV10TCL_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_RC_COLOR(x) (0x00000270+((x)*4)) +#define NV10TCL_RC_COLOR__SIZE 0x00000002 +#define NV10TCL_RC_COLOR_B_SHIFT 0 +#define NV10TCL_RC_COLOR_B_MASK 0x000000ff +#define NV10TCL_RC_COLOR_G_SHIFT 8 +#define NV10TCL_RC_COLOR_G_MASK 0x0000ff00 +#define NV10TCL_RC_COLOR_R_SHIFT 16 +#define NV10TCL_RC_COLOR_R_MASK 0x00ff0000 +#define NV10TCL_RC_COLOR_A_SHIFT 24 +#define NV10TCL_RC_COLOR_A_MASK 0xff000000 +#define NV10TCL_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) +#define NV10TCL_RC_OUT_ALPHA__SIZE 0x00000002 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000030 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000080 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000090 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x000000c0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x000000d0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 +#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000300 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) +#define NV10TCL_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) +#define NV10TCL_RC_OUT_ALPHA_MUX_SUM (1 << 14) +#define NV10TCL_RC_OUT_ALPHA_BIAS (1 << 15) +#define NV10TCL_RC_OUT_ALPHA_BIAS_NONE 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SHIFT 17 +#define NV10TCL_RC_OUT_ALPHA_SCALE_MASK 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_NONE 0x00000000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00020000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00040000 +#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 +#define NV10TCL_RC_OUT_RGB(x) (0x00000280+((x)*4)) +#define NV10TCL_RC_OUT_RGB__SIZE 0x00000002 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000030 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000080 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000090 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x000000c0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x000000d0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 +#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000300 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) +#define NV10TCL_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) +#define NV10TCL_RC_OUT_RGB_MUX_SUM (1 << 14) +#define NV10TCL_RC_OUT_RGB_BIAS (1 << 15) +#define NV10TCL_RC_OUT_RGB_BIAS_NONE 0x00000000 +#define NV10TCL_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 +#define NV10TCL_RC_OUT_RGB_SCALE_SHIFT 17 +#define NV10TCL_RC_OUT_RGB_SCALE_MASK 0x00000000 +#define NV10TCL_RC_OUT_RGB_SCALE_NONE 0x00000000 +#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00020000 +#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00040000 +#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 +#define NV10TCL_RC_OUT_RGB_OPERATION_SHIFT 27 +#define NV10TCL_RC_OUT_RGB_OPERATION_MASK 0x38000000 +#define NV10TCL_RC_FINAL0 0x00000288 +#define NV10TCL_RC_FINAL0_D_INPUT_SHIFT 0 +#define NV10TCL_RC_FINAL0_D_INPUT_MASK 0x0000000f +#define NV10TCL_RC_FINAL0_D_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 +#define NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 +#define NV10TCL_RC_FINAL0_D_INPUT_FOG 0x00000003 +#define NV10TCL_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 +#define NV10TCL_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 +#define NV10TCL_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000008 +#define NV10TCL_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000009 +#define NV10TCL_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c +#define NV10TCL_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d +#define NV10TCL_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e +#define NV10TCL_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f +#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) +#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000010 +#define NV10TCL_RC_FINAL0_D_MAPPING_SHIFT 5 +#define NV10TCL_RC_FINAL0_D_MAPPING_MASK 0x000000e0 +#define NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 +#define NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 +#define NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 +#define NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 +#define NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 +#define NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 +#define NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 +#define NV10TCL_RC_FINAL0_C_INPUT_SHIFT 8 +#define NV10TCL_RC_FINAL0_C_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_FINAL0_C_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_FINAL0_C_INPUT_FOG 0x00000300 +#define NV10TCL_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_FINAL0_C_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_FINAL0_C_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_FINAL0_C_MAPPING_SHIFT 13 +#define NV10TCL_RC_FINAL0_C_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_FINAL0_B_INPUT_SHIFT 16 +#define NV10TCL_RC_FINAL0_B_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_FINAL0_B_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_FINAL0_B_INPUT_FOG 0x00030000 +#define NV10TCL_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_FINAL0_B_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_FINAL0_B_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_FINAL0_B_MAPPING_SHIFT 21 +#define NV10TCL_RC_FINAL0_B_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_FINAL0_A_INPUT_SHIFT 24 +#define NV10TCL_RC_FINAL0_A_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_FINAL0_A_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_FINAL0_A_INPUT_FOG 0x03000000 +#define NV10TCL_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_SHIFT 29 +#define NV10TCL_RC_FINAL0_A_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_RC_FINAL1 0x0000028c +#define NV10TCL_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) +#define NV10TCL_RC_FINAL1_G_INPUT_SHIFT 8 +#define NV10TCL_RC_FINAL1_G_INPUT_MASK 0x00000f00 +#define NV10TCL_RC_FINAL1_G_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000100 +#define NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000200 +#define NV10TCL_RC_FINAL1_G_INPUT_FOG 0x00000300 +#define NV10TCL_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000400 +#define NV10TCL_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000500 +#define NV10TCL_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000800 +#define NV10TCL_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000900 +#define NV10TCL_RC_FINAL1_G_INPUT_SPARE0_NV 0x00000c00 +#define NV10TCL_RC_FINAL1_G_INPUT_SPARE1_NV 0x00000d00 +#define NV10TCL_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 +#define NV10TCL_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x00000f00 +#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) +#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00001000 +#define NV10TCL_RC_FINAL1_G_MAPPING_SHIFT 13 +#define NV10TCL_RC_FINAL1_G_MAPPING_MASK 0x0000e000 +#define NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00002000 +#define NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00004000 +#define NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00006000 +#define NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 +#define NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 +#define NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 +#define NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x0000e000 +#define NV10TCL_RC_FINAL1_F_INPUT_SHIFT 16 +#define NV10TCL_RC_FINAL1_F_INPUT_MASK 0x000f0000 +#define NV10TCL_RC_FINAL1_F_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00010000 +#define NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00020000 +#define NV10TCL_RC_FINAL1_F_INPUT_FOG 0x00030000 +#define NV10TCL_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00040000 +#define NV10TCL_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00050000 +#define NV10TCL_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00080000 +#define NV10TCL_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00090000 +#define NV10TCL_RC_FINAL1_F_INPUT_SPARE0_NV 0x000c0000 +#define NV10TCL_RC_FINAL1_F_INPUT_SPARE1_NV 0x000d0000 +#define NV10TCL_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 +#define NV10TCL_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x000f0000 +#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) +#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00100000 +#define NV10TCL_RC_FINAL1_F_MAPPING_SHIFT 21 +#define NV10TCL_RC_FINAL1_F_MAPPING_MASK 0x00e00000 +#define NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00200000 +#define NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00400000 +#define NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00600000 +#define NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 +#define NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 +#define NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 +#define NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00e00000 +#define NV10TCL_RC_FINAL1_E_INPUT_SHIFT 24 +#define NV10TCL_RC_FINAL1_E_INPUT_MASK 0x0f000000 +#define NV10TCL_RC_FINAL1_E_INPUT_ZERO 0x00000000 +#define NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x01000000 +#define NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x02000000 +#define NV10TCL_RC_FINAL1_E_INPUT_FOG 0x03000000 +#define NV10TCL_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x04000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x05000000 +#define NV10TCL_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x08000000 +#define NV10TCL_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x09000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SPARE0_NV 0x0c000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SPARE1_NV 0x0d000000 +#define NV10TCL_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 +#define NV10TCL_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0f000000 +#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) +#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 +#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x10000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_SHIFT 29 +#define NV10TCL_RC_FINAL1_E_MAPPING_MASK 0xe0000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x20000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x40000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x60000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 +#define NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0xe0000000 +#define NV10TCL_LIGHT_MODEL 0x00000294 +#define NV10TCL_LIGHT_MODEL_COLOR_CONTROL (1 << 1) +#define NV10TCL_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) +#define NV10TCL_COLOR_MATERIAL_ENABLE 0x00000298 +#define NV10TCL_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) +#define NV10TCL_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) +#define NV10TCL_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) +#define NV10TCL_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) +#define NV10TCL_FOG_MODE 0x0000029c +#define NV10TCL_FOG_MODE_EXP 0x00000800 +#define NV10TCL_FOG_MODE_EXP_2 0x00000802 +#define NV10TCL_FOG_MODE_EXP2 0x00000803 +#define NV10TCL_FOG_MODE_LINEAR 0x00000804 +#define NV10TCL_FOG_MODE_LINEAR_2 0x00002601 +#define NV10TCL_FOG_COORD_DIST 0x000002a0 +#define NV10TCL_FOG_ENABLE 0x000002a4 +#define NV10TCL_FOG_COLOR 0x000002a8 +#define NV10TCL_FOG_COLOR_R_SHIFT 0 +#define NV10TCL_FOG_COLOR_R_MASK 0x000000ff +#define NV10TCL_FOG_COLOR_G_SHIFT 8 +#define NV10TCL_FOG_COLOR_G_MASK 0x0000ff00 +#define NV10TCL_FOG_COLOR_B_SHIFT 16 +#define NV10TCL_FOG_COLOR_B_MASK 0x00ff0000 +#define NV10TCL_FOG_COLOR_A_SHIFT 24 +#define NV10TCL_FOG_COLOR_A_MASK 0xff000000 +#define NV10TCL_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV10TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV10TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 +#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) +#define NV10TCL_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV10TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 +#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) +#define NV10TCL_ALPHA_FUNC_ENABLE 0x00000300 +#define NV10TCL_BLEND_FUNC_ENABLE 0x00000304 +#define NV10TCL_CULL_FACE_ENABLE 0x00000308 +#define NV10TCL_DEPTH_TEST_ENABLE 0x0000030c +#define NV10TCL_DITHER_ENABLE 0x00000310 +#define NV10TCL_LIGHTING_ENABLE 0x00000314 +#define NV10TCL_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV10TCL_POINT_SMOOTH_ENABLE 0x0000031c +#define NV10TCL_LINE_SMOOTH_ENABLE 0x00000320 +#define NV10TCL_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV10TCL_VERTEX_WEIGHT_ENABLE 0x00000328 +#define NV10TCL_STENCIL_ENABLE 0x0000032c +#define NV10TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV10TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV10TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV10TCL_ALPHA_FUNC_FUNC 0x0000033c +#define NV10TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV10TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV10TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV10TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10TCL_ALPHA_FUNC_REF 0x00000340 +#define NV10TCL_BLEND_FUNC_SRC 0x00000344 +#define NV10TCL_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV10TCL_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV10TCL_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10TCL_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10TCL_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV10TCL_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10TCL_BLEND_FUNC_DST 0x00000348 +#define NV10TCL_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV10TCL_BLEND_FUNC_DST_ONE 0x00000001 +#define NV10TCL_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV10TCL_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV10TCL_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV10TCL_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV10TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV10TCL_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV10TCL_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV10TCL_BLEND_COLOR 0x0000034c +#define NV10TCL_BLEND_COLOR_B_SHIFT 0 +#define NV10TCL_BLEND_COLOR_B_MASK 0x000000ff +#define NV10TCL_BLEND_COLOR_G_SHIFT 8 +#define NV10TCL_BLEND_COLOR_G_MASK 0x0000ff00 +#define NV10TCL_BLEND_COLOR_R_SHIFT 16 +#define NV10TCL_BLEND_COLOR_R_MASK 0x00ff0000 +#define NV10TCL_BLEND_COLOR_A_SHIFT 24 +#define NV10TCL_BLEND_COLOR_A_MASK 0xff000000 +#define NV10TCL_BLEND_EQUATION 0x00000350 +#define NV10TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV10TCL_BLEND_EQUATION_MIN 0x00008007 +#define NV10TCL_BLEND_EQUATION_MAX 0x00008008 +#define NV10TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV10TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV10TCL_DEPTH_FUNC 0x00000354 +#define NV10TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV10TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV10TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV10TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV10TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV10TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV10TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV10TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV10TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV10TCL_COLOR_MASK 0x00000358 +#define NV10TCL_COLOR_MASK_B (1 << 0) +#define NV10TCL_COLOR_MASK_G (1 << 8) +#define NV10TCL_COLOR_MASK_R (1 << 16) +#define NV10TCL_COLOR_MASK_A (1 << 24) +#define NV10TCL_DEPTH_WRITE_ENABLE 0x0000035c +#define NV10TCL_STENCIL_MASK 0x00000360 +#define NV10TCL_STENCIL_FUNC_FUNC 0x00000364 +#define NV10TCL_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV10TCL_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV10TCL_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV10TCL_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV10TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV10TCL_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV10TCL_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV10TCL_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV10TCL_STENCIL_FUNC_REF 0x00000368 +#define NV10TCL_STENCIL_FUNC_MASK 0x0000036c +#define NV10TCL_STENCIL_OP_FAIL 0x00000370 +#define NV10TCL_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV10TCL_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV10TCL_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV10TCL_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV10TCL_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV10TCL_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV10TCL_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV10TCL_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV10TCL_STENCIL_OP_ZFAIL 0x00000374 +#define NV10TCL_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV10TCL_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV10TCL_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV10TCL_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV10TCL_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV10TCL_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV10TCL_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV10TCL_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV10TCL_STENCIL_OP_ZPASS 0x00000378 +#define NV10TCL_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV10TCL_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV10TCL_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV10TCL_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV10TCL_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV10TCL_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV10TCL_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV10TCL_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV10TCL_SHADE_MODEL 0x0000037c +#define NV10TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV10TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV10TCL_LINE_WIDTH 0x00000380 +#define NV10TCL_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV10TCL_POLYGON_OFFSET_UNITS 0x00000388 +#define NV10TCL_POLYGON_MODE_FRONT 0x0000038c +#define NV10TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV10TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV10TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV10TCL_POLYGON_MODE_BACK 0x00000390 +#define NV10TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV10TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV10TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV10TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV10TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV10TCL_CULL_FACE 0x0000039c +#define NV10TCL_CULL_FACE_FRONT 0x00000404 +#define NV10TCL_CULL_FACE_BACK 0x00000405 +#define NV10TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV10TCL_FRONT_FACE 0x000003a0 +#define NV10TCL_FRONT_FACE_CW 0x00000900 +#define NV10TCL_FRONT_FACE_CCW 0x00000901 +#define NV10TCL_NORMALIZE_ENABLE 0x000003a4 +#define NV10TCL_COLOR_MATERIAL_R 0x000003a8 +#define NV10TCL_COLOR_MATERIAL_G 0x000003ac +#define NV10TCL_COLOR_MATERIAL_B 0x000003b0 +#define NV10TCL_COLOR_MATERIAL_A 0x000003b4 +#define NV10TCL_COLOR_CONTROL 0x000003b8 +#define NV10TCL_ENABLED_LIGHTS 0x000003bc +#define NV10TCL_ENABLED_LIGHTS_LIGHT0 (1 << 0) +#define NV10TCL_ENABLED_LIGHTS_LIGHT1 (1 << 2) +#define NV10TCL_ENABLED_LIGHTS_LIGHT2 (1 << 4) +#define NV10TCL_ENABLED_LIGHTS_LIGHT3 (1 << 6) +#define NV10TCL_ENABLED_LIGHTS_LIGHT4 (1 << 8) +#define NV10TCL_ENABLED_LIGHTS_LIGHT5 (1 << 10) +#define NV10TCL_ENABLED_LIGHTS_LIGHT6 (1 << 12) +#define NV10TCL_ENABLED_LIGHTS_LIGHT7 (1 << 14) +#define NV10TCL_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV10TCL_CLIP_PLANE_ENABLE__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_ENABLE_FALSE 0x00000000 +#define NV10TCL_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 +#define NV10TCL_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 +#define NV10TCL_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) +#define NV10TCL_TX_MATRIX_ENABLE__SIZE 0x00000002 +#define NV10TCL_VIEW_MATRIX_ENABLE 0x000003e8 +#define NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) +#define NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) +#define NV10TCL_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) +#define NV10TCL_POINT_SIZE 0x000003ec +#define NV10TCL_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) +#define NV10TCL_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10TCL_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) +#define NV10TCL_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10TCL_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) +#define NV10TCL_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 +#define NV10TCL_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) +#define NV10TCL_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 +#define NV10TCL_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) +#define NV10TCL_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV10TCL_TX0_MATRIX(x) (0x00000540+((x)*4)) +#define NV10TCL_TX0_MATRIX__SIZE 0x00000010 +#define NV10TCL_TX1_MATRIX(x) (0x00000580+((x)*4)) +#define NV10TCL_TX1_MATRIX__SIZE 0x00000010 +#define NV10TCL_CLIP_PLANE_A(x) (0x00000600+((x)*16)) +#define NV10TCL_CLIP_PLANE_A__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_B(x) (0x00000604+((x)*16)) +#define NV10TCL_CLIP_PLANE_B__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_C(x) (0x00000608+((x)*16)) +#define NV10TCL_CLIP_PLANE_C__SIZE 0x00000008 +#define NV10TCL_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) +#define NV10TCL_CLIP_PLANE_D__SIZE 0x00000008 +#define NV10TCL_FOG_EQUATION_CONSTANT 0x00000680 +#define NV10TCL_FOG_EQUATION_LINEAR 0x00000684 +#define NV10TCL_FOG_EQUATION_QUADRATIC 0x00000688 +#define NV10TCL_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) +#define NV10TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000006c4 +#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000006c8 +#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000006cc +#define NV10TCL_VIEWPORT_SCALE_X 0x000006e8 +#define NV10TCL_VIEWPORT_SCALE_Y 0x000006ec +#define NV10TCL_VIEWPORT_SCALE_Z 0x000006f0 +#define NV10TCL_VIEWPORT_SCALE_W 0x000006f4 +#define NV10TCL_POINT_PARAMETER(x) (0x000006f8+((x)*4)) +#define NV10TCL_POINT_PARAMETER__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) +#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) +#define NV10TCL_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) +#define NV10TCL_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) +#define NV10TCL_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) +#define NV10TCL_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) +#define NV10TCL_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) +#define NV10TCL_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) +#define NV10TCL_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) +#define NV10TCL_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) +#define NV10TCL_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) +#define NV10TCL_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV10TCL_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) +#define NV10TCL_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV10TCL_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) +#define NV10TCL_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV10TCL_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) +#define NV10TCL_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV10TCL_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) +#define NV10TCL_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV10TCL_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) +#define NV10TCL_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV10TCL_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) +#define NV10TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV10TCL_VERTEX_POS_3F_X 0x00000c00 +#define NV10TCL_VERTEX_POS_3F_Y 0x00000c04 +#define NV10TCL_VERTEX_POS_3F_Z 0x00000c08 +#define NV10TCL_VERTEX_POS_4F_X 0x00000c18 +#define NV10TCL_VERTEX_POS_4F_Y 0x00000c1c +#define NV10TCL_VERTEX_POS_4F_Z 0x00000c20 +#define NV10TCL_VERTEX_POS_4F_W 0x00000c24 +#define NV10TCL_VERTEX_NOR_3F_X 0x00000c30 +#define NV10TCL_VERTEX_NOR_3F_Y 0x00000c34 +#define NV10TCL_VERTEX_NOR_3F_Z 0x00000c38 +#define NV10TCL_VERTEX_NOR_3I_XY 0x00000c40 +#define NV10TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV10TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV10TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV10TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV10TCL_VERTEX_NOR_3I_Z 0x00000c44 +#define NV10TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV10TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV10TCL_VERTEX_COL_4F_R 0x00000c50 +#define NV10TCL_VERTEX_COL_4F_G 0x00000c54 +#define NV10TCL_VERTEX_COL_4F_B 0x00000c58 +#define NV10TCL_VERTEX_COL_4F_A 0x00000c5c +#define NV10TCL_VERTEX_COL_3F_R 0x00000c60 +#define NV10TCL_VERTEX_COL_3F_G 0x00000c64 +#define NV10TCL_VERTEX_COL_3F_B 0x00000c68 +#define NV10TCL_VERTEX_COL_4I 0x00000c6c +#define NV10TCL_VERTEX_COL_4I_R_SHIFT 0 +#define NV10TCL_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV10TCL_VERTEX_COL_4I_G_SHIFT 8 +#define NV10TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV10TCL_VERTEX_COL_4I_B_SHIFT 16 +#define NV10TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV10TCL_VERTEX_COL_4I_A_SHIFT 24 +#define NV10TCL_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV10TCL_VERTEX_COL2_3F_R 0x00000c80 +#define NV10TCL_VERTEX_COL2_3F_G 0x00000c84 +#define NV10TCL_VERTEX_COL2_3F_B 0x00000c88 +#define NV10TCL_VERTEX_COL2_3I 0x00000c8c +#define NV10TCL_VERTEX_COL2_3I_R_SHIFT 0 +#define NV10TCL_VERTEX_COL2_3I_R_MASK 0x000000ff +#define NV10TCL_VERTEX_COL2_3I_G_SHIFT 8 +#define NV10TCL_VERTEX_COL2_3I_G_MASK 0x0000ff00 +#define NV10TCL_VERTEX_COL2_3I_B_SHIFT 16 +#define NV10TCL_VERTEX_COL2_3I_B_MASK 0x00ff0000 +#define NV10TCL_VERTEX_TX0_2F_S 0x00000c90 +#define NV10TCL_VERTEX_TX0_2F_T 0x00000c94 +#define NV10TCL_VERTEX_TX0_2I 0x00000c98 +#define NV10TCL_VERTEX_TX0_2I_S_SHIFT 0 +#define NV10TCL_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX0_2I_T_SHIFT 16 +#define NV10TCL_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX0_4F_S 0x00000ca0 +#define NV10TCL_VERTEX_TX0_4F_T 0x00000ca4 +#define NV10TCL_VERTEX_TX0_4F_R 0x00000ca8 +#define NV10TCL_VERTEX_TX0_4F_Q 0x00000cac +#define NV10TCL_VERTEX_TX0_4I_ST 0x00000cb0 +#define NV10TCL_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV10TCL_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV10TCL_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX0_4I_RQ 0x00000cb4 +#define NV10TCL_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV10TCL_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV10TCL_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX1_2F_S 0x00000cb8 +#define NV10TCL_VERTEX_TX1_2F_T 0x00000cbc +#define NV10TCL_VERTEX_TX1_2I 0x00000cc0 +#define NV10TCL_VERTEX_TX1_2I_S_SHIFT 0 +#define NV10TCL_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX1_2I_T_SHIFT 16 +#define NV10TCL_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX1_4F_S 0x00000cc8 +#define NV10TCL_VERTEX_TX1_4F_T 0x00000ccc +#define NV10TCL_VERTEX_TX1_4F_R 0x00000cd0 +#define NV10TCL_VERTEX_TX1_4F_Q 0x00000cd4 +#define NV10TCL_VERTEX_TX1_4I_ST 0x00000cd8 +#define NV10TCL_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV10TCL_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV10TCL_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV10TCL_VERTEX_TX1_4I_RQ 0x00000cdc +#define NV10TCL_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV10TCL_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV10TCL_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV10TCL_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV10TCL_VERTEX_FOG_1F 0x00000ce0 +#define NV10TCL_VERTEX_WGH_1F 0x00000ce4 +#define NV10TCL_EDGEFLAG_ENABLE 0x00000cec +#define NV10TCL_VERTEX_ARRAY_VALIDATE 0x00000cf0 +#define NV10TCL_VERTEX_ARRAY_OFFSET_POS 0x00000d00 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS 0x00000d04 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_COL 0x00000d08 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL 0x00000d0c +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 +#define NV10TCL_VERTEX_BEGIN_END 0x00000dfc +#define NV10TCL_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV10TCL_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV10TCL_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV10TCL_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10TCL_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10TCL_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV10TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10TCL_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10TCL_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV10TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV10TCL_DRAW_INDEX 0x00000e00 +#define NV10TCL_DRAW_INDEX_I0_SHIFT 0 +#define NV10TCL_DRAW_INDEX_I0_MASK 0x0000ffff +#define NV10TCL_DRAW_INDEX_I1_SHIFT 24 +#define NV10TCL_DRAW_INDEX_I1_MASK 0xff000000 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END 0x000013fc +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV10TCL_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 +#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 +#define NV10TCL_VERTEX_ARRAY_DATA 0x00001800 + + +#define NV04_CONTEXT_COLOR_KEY 0x00000057 + + + +#define NV03_CONTEXT_SURFACES_2D 0x00000058 + +#define NV03_CONTEXT_SURFACES_2D_SYNCHRONIZE 0x00000100 +#define NV03_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_SURFACES_2D_DMA_SOURCE 0x00000184 +#define NV03_CONTEXT_SURFACES_2D_DMA_DESTIN 0x00000188 +#define NV03_CONTEXT_SURFACES_2D_COLOR_FORMAT 0x00000300 +#define NV03_CONTEXT_SURFACES_2D_PITCH 0x00000304 +#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 +#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff +#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 +#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 +#define NV03_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 +#define NV03_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c + + +#define NV03_CONTEXT_SURFACES_3D 0x0000005a + +#define NV03_CONTEXT_SURFACES_3D_SYNCHRONIZE 0x00000100 +#define NV03_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 +#define NV03_CONTEXT_SURFACES_3D_DMA_SURFACE 0x00000184 +#define NV03_CONTEXT_SURFACES_3D_PITCH 0x00000300 +#define NV03_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x00000304 +#define NV03_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000308 + + +#define NV04_RENDER_SOLID_LINE 0x0000005c + +#define NV04_RENDER_SOLID_LINE_SURFACE 0x00000198 + + +#define NV04_RENDER_SOLID_TRIANGLE 0x0000005d + + + +#define NV04_RENDER_SOLID_RECTANGLE 0x0000005e + +#define NV04_RENDER_SOLID_RECTANGLE_SURFACE 0x00000198 + + +#define NV04_IMAGE_BLIT 0x0000005f + +#define NV04_IMAGE_BLIT_NOP 0x00000100 +#define NV04_IMAGE_BLIT_NOTIFY 0x00000104 +#define NV04_IMAGE_BLIT_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_BLIT_COLOR_KEY 0x00000184 +#define NV04_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 +#define NV04_IMAGE_BLIT_PATTERN 0x0000018c +#define NV04_IMAGE_BLIT_ROP 0x00000190 +#define NV04_IMAGE_BLIT_BETA4 0x00000198 +#define NV04_IMAGE_BLIT_SURFACE 0x0000019c +#define NV04_IMAGE_BLIT_OPERATION 0x000002fc +#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 +#define NV04_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 +#define NV04_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 +#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY 0x00000003 +#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV04_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000005 + + +#define NV04_INDEXED_IMAGE_FROM_CPU 0x00000060 + +#define NV04_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 +#define NV04_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 +#define NV04_INDEXED_IMAGE_FROM_CPU_PATCH 0x0000010c +#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 +#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 +#define NV04_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec +#define NV04_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 +#define NV04_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 +#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 +#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc +#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR 0x00000400 + + +#define NV04_IMAGE_FROM_CPU 0x00000061 + +#define NV04_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV04_IMAGE_FROM_CPU_SURFACE 0x0000019c + + +#define NV10_CONTEXT_SURFACES_2D 0x00000062 + + + +#define NV05_SCALED_IMAGE_FROM_MEMORY 0x00000063 + +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 +#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 + + +#define NV01_IMAGE_SRCCOPY_AND 0x00000064 + +#define NV01_IMAGE_SRCCOPY_AND_NOTIFY 0x00000104 +#define NV01_IMAGE_SRCCOPY_AND_DMA_NOTIFY 0x00000180 +#define NV01_IMAGE_SRCCOPY_AND_IMAGE_OUTPUT 0x00000200 +#define NV01_IMAGE_SRCCOPY_AND_IMAGE_INPUT 0x00000204 + + +#define NV05_INDEXED_IMAGE_FROM_CPU 0x00000064 + +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_KEY 0x00000188 +#define NV05_INDEXED_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x0000018c +#define NV05_INDEXED_IMAGE_FROM_CPU_PATTERN 0x00000190 +#define NV05_INDEXED_IMAGE_FROM_CPU_ROP 0x00000194 +#define NV05_INDEXED_IMAGE_FROM_CPU_BETA1 0x00000198 +#define NV05_INDEXED_IMAGE_FROM_CPU_BETA4 0x0000019c +#define NV05_INDEXED_IMAGE_FROM_CPU_SURFACE 0x000001a0 +#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000003e0 +#define NV05_INDEXED_IMAGE_FROM_CPU_OPERATION 0x000003e4 +#define NV05_INDEXED_IMAGE_FROM_CPU_INDICES 0x00000400 + + +#define NV05_IMAGE_FROM_CPU 0x00000065 + +#define NV05_IMAGE_FROM_CPU_BETA4 0x00000198 +#define NV05_IMAGE_FROM_CPU_SURFACE 0x0000019c + + +#define NV05_STRETCHED_IMAGE_FROM_CPU 0x00000066 + +#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA4 0x00000194 +#define NV05_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000198 +#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 + + +#define NV04_IMAGE_BLEND_PREMULT 0x00000067 + +#define NV04_IMAGE_BLEND_PREMULT_NOP 0x00000100 +#define NV04_IMAGE_BLEND_PREMULT_NOTIFY 0x00000104 +#define NV04_IMAGE_BLEND_PREMULT_DMA_NOTIFY 0x00000180 +#define NV04_IMAGE_BLEND_PREMULT_IMAGE_OUTPUT 0x00000200 +#define NV04_IMAGE_BLEND_PREMULT_BETA_INPUT 0x00000204 +#define NV04_IMAGE_BLEND_PREMULT_IMAGE_INPUT 0x00000208 + + +#define NV03_CHANNEL_PIO 0x0000006a + + + +#define NV03_CHANNEL_DMA 0x0000006b + + + +#define NV04_BETA_SOLID 0x00000072 + +#define NV04_BETA_SOLID_NOP 0x00000100 +#define NV04_BETA_SOLID_NOTIFY 0x00000104 +#define NV04_BETA_SOLID_DMA_NOTIFY 0x00000180 +#define NV04_BETA_SOLID_BETA_OUTPUT 0x00000200 +#define NV04_BETA_SOLID_BETA_FACTOR 0x00000300 + + +#define NV04_STRETCHED_IMAGE_FROM_CPU 0x00000076 + + + +#define NV04_SCALED_IMAGE_FROM_MEMORY 0x00000077 + +#define NV04_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 +#define NV04_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 +#define NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 +#define NV04_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c +#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 +#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA4 0x00000194 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000198 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 +#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY 0x00000003 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000005 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT 0x00000310 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE 0x00000314 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX 0x00000318 +#define NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY 0x0000031c +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE 0x00000400 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_MASK 0xffff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT 0x00000404 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_MASK 0x00ff0000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER 0x00010000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CORNER 0x00020000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_SHIFT 24 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_MASK 0xff000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE 0x00000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_BILINEAR 0x01000000 +#define NV04_SCALED_IMAGE_FROM_MEMORY_ADDRESS 0x00000408 +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT 0x0000040c +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_SHIFT 0 +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_MASK 0x0000ffff +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_SHIFT 16 +#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_MASK 0xffff0000 + + +#define NV10_TEXTURE_FROM_CPU 0x0000007b + +#define NV10_TEXTURE_FROM_CPU_NOP 0x00000100 +#define NV10_TEXTURE_FROM_CPU_NOTIFY 0x00000104 +#define NV10_TEXTURE_FROM_CPU_WAIT_FOR_IDLE 0x00000108 +#define NV10_TEXTURE_FROM_CPU_PM_TRIGGER 0x00000140 +#define NV10_TEXTURE_FROM_CPU_DMA_NOTIFY 0x00000180 +#define NV10_TEXTURE_FROM_CPU_SURFACE 0x00000184 +#define NV10_TEXTURE_FROM_CPU_COLOR_FORMAT 0x00000300 +#define NV10_TEXTURE_FROM_CPU_POINT 0x00000304 +#define NV10_TEXTURE_FROM_CPU_POINT_X_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_POINT_X_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_POINT_Y_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_POINT_Y_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_SIZE 0x00000308 +#define NV10_TEXTURE_FROM_CPU_SIZE_W_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_SIZE_W_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_SIZE_H_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_SIZE_H_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL 0x0000030c +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL 0x00000310 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_SHIFT 0 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_MASK 0x0000ffff +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_SHIFT 16 +#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_MASK 0xffff0000 +#define NV10_TEXTURE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) +#define NV10_TEXTURE_FROM_CPU_COLOR__SIZE 0x00000700 + + +#define NV10_VIDEO_DISPLAY 0x0000007c + + + +#define NV10_DVD_SUBPICTURE 0x00000088 + + + +#define NV10_SCALED_IMAGE_FROM_MEMORY 0x00000089 + +#define NV10_SCALED_IMAGE_FROM_MEMORY_WAIT_FOR_IDLE 0x00000108 + + +#define NV10_IMAGE_FROM_CPU 0x0000008a + +#define NV10_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 + + +#define NV10_CONTEXT_SURFACES_3D 0x00000093 + + + +#define NV10_DX5_TEXTURE_TRIANGLE 0x00000094 + + + +#define NV10_DX6_MULTI_TEXTURE_TRIANGLE 0x00000095 + + + +#define NV11TCL 0x00000096 + +#define NV11TCL_COLOR_LOGIC_OP_ENABLE 0x00000d40 +#define NV11TCL_COLOR_LOGIC_OP_OP 0x00000d44 +#define NV11TCL_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV11TCL_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV11TCL_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV11TCL_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV11TCL_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV11TCL_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV11TCL_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV11TCL_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV11TCL_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV11TCL_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV11TCL_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV11TCL_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV11TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV11TCL_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV11TCL_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV11TCL_COLOR_LOGIC_OP_OP_SET 0x0000150f + + +#define NV20TCL 0x00000097 + +#define NV20TCL_NOP 0x00000100 +#define NV20TCL_NOTIFY 0x00000104 +#define NV20TCL_DMA_NOTIFY 0x00000180 +#define NV20TCL_DMA_IN_MEMORY0 0x00000184 +#define NV20TCL_DMA_IN_MEMORY1 0x00000188 +#define NV20TCL_DMA_IN_MEMORY2 0x00000194 +#define NV20TCL_DMA_IN_MEMORY3 0x00000198 +#define NV20TCL_DMA_IN_MEMORY6 0x000001a4 +#define NV20TCL_DMA_IN_MEMORY7 0x000001a8 +#define NV20TCL_VIEWPORT_HORIZ 0x00000200 +#define NV20TCL_VIEWPORT_VERT 0x00000204 +#define NV20TCL_BUFFER_FORMAT 0x00000208 +#define NV20TCL_BUFFER_PITCH 0x0000020c +#define NV20TCL_COLOR_OFFSET 0x00000210 +#define NV20TCL_ZETA_OFFSET 0x00000214 +#define NV20TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) +#define NV20TCL_RC_IN_ALPHA__SIZE 0x00000008 +#define NV20TCL_RC_FINAL0 0x00000288 +#define NV20TCL_RC_FINAL1 0x0000028c +#define NV20TCL_LIGHT_CONTROL 0x00000294 +#define NV20TCL_FOG_MODE 0x0000029c +#define NV20TCL_FOG_COORD_DIST 0x000002a0 +#define NV20TCL_FOG_ENABLE 0x000002a4 +#define NV20TCL_FOG_COLOR 0x000002a8 +#define NV20TCL_VIEWPORT_CLIP_MODE 0x000002b4 +#define NV20TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) +#define NV20TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV20TCL_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) +#define NV20TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV20TCL_ALPHA_FUNC_ENABLE 0x00000300 +#define NV20TCL_BLEND_FUNC_ENABLE 0x00000304 +#define NV20TCL_CULL_FACE_ENABLE 0x00000308 +#define NV20TCL_DEPTH_TEST_ENABLE 0x0000030c +#define NV20TCL_DITHER_ENABLE 0x00000310 +#define NV20TCL_LIGHTING_ENABLE 0x00000314 +#define NV20TCL_POINT_PARAMETERS_ENABLE 0x00000318 +#define NV20TCL_LINE_SMOOTH_ENABLE 0x00000320 +#define NV20TCL_POLYGON_SMOOTH_ENABLE 0x00000324 +#define NV20TCL_STENCIL_ENABLE 0x0000032c +#define NV20TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000330 +#define NV20TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000334 +#define NV20TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000338 +#define NV20TCL_ALPHA_FUNC_FUNC 0x0000033c +#define NV20TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV20TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV20TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV20TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20TCL_ALPHA_FUNC_REF 0x00000340 +#define NV20TCL_BLEND_FUNC_SRC 0x00000344 +#define NV20TCL_BLEND_FUNC_SRC_ZERO 0x00000000 +#define NV20TCL_BLEND_FUNC_SRC_ONE 0x00000001 +#define NV20TCL_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20TCL_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20TCL_BLEND_FUNC_SRC_DST_COLOR 0x00000306 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 +#define NV20TCL_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 +#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20TCL_BLEND_FUNC_DST 0x00000348 +#define NV20TCL_BLEND_FUNC_DST_ZERO 0x00000000 +#define NV20TCL_BLEND_FUNC_DST_ONE 0x00000001 +#define NV20TCL_BLEND_FUNC_DST_SRC_COLOR 0x00000300 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV20TCL_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV20TCL_BLEND_FUNC_DST_DST_ALPHA 0x00000304 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV20TCL_BLEND_FUNC_DST_DST_COLOR 0x00000306 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 +#define NV20TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 +#define NV20TCL_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV20TCL_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 +#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV20TCL_BLEND_COLOR 0x0000034c +#define NV20TCL_BLEND_EQUATION 0x00000350 +#define NV20TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV20TCL_BLEND_EQUATION_MIN 0x00008007 +#define NV20TCL_BLEND_EQUATION_MAX 0x00008008 +#define NV20TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV20TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV20TCL_DEPTH_FUNC 0x00000354 +#define NV20TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV20TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV20TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV20TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV20TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV20TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV20TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV20TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV20TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV20TCL_COLOR_MASK 0x00000358 +#define NV20TCL_DEPTH_WRITE_ENABLE 0x0000035c +#define NV20TCL_STENCIL_MASK 0x00000360 +#define NV20TCL_STENCIL_FUNC_FUNC 0x00000364 +#define NV20TCL_STENCIL_FUNC_FUNC_NEVER 0x00000200 +#define NV20TCL_STENCIL_FUNC_FUNC_LESS 0x00000201 +#define NV20TCL_STENCIL_FUNC_FUNC_EQUAL 0x00000202 +#define NV20TCL_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 +#define NV20TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 +#define NV20TCL_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV20TCL_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 +#define NV20TCL_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 +#define NV20TCL_STENCIL_FUNC_REF 0x00000368 +#define NV20TCL_STENCIL_FUNC_MASK 0x0000036c +#define NV20TCL_STENCIL_OP_FAIL 0x00000370 +#define NV20TCL_STENCIL_OP_FAIL_ZERO 0x00000000 +#define NV20TCL_STENCIL_OP_FAIL_INVERT 0x0000150a +#define NV20TCL_STENCIL_OP_FAIL_KEEP 0x00001e00 +#define NV20TCL_STENCIL_OP_FAIL_REPLACE 0x00001e01 +#define NV20TCL_STENCIL_OP_FAIL_INCR 0x00001e02 +#define NV20TCL_STENCIL_OP_FAIL_DECR 0x00001e03 +#define NV20TCL_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 +#define NV20TCL_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 +#define NV20TCL_STENCIL_OP_ZFAIL 0x00000374 +#define NV20TCL_STENCIL_OP_ZFAIL_ZERO 0x00000000 +#define NV20TCL_STENCIL_OP_ZFAIL_INVERT 0x0000150a +#define NV20TCL_STENCIL_OP_ZFAIL_KEEP 0x00001e00 +#define NV20TCL_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 +#define NV20TCL_STENCIL_OP_ZFAIL_INCR 0x00001e02 +#define NV20TCL_STENCIL_OP_ZFAIL_DECR 0x00001e03 +#define NV20TCL_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV20TCL_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV20TCL_STENCIL_OP_ZPASS 0x00000378 +#define NV20TCL_STENCIL_OP_ZPASS_ZERO 0x00000000 +#define NV20TCL_STENCIL_OP_ZPASS_INVERT 0x0000150a +#define NV20TCL_STENCIL_OP_ZPASS_KEEP 0x00001e00 +#define NV20TCL_STENCIL_OP_ZPASS_REPLACE 0x00001e01 +#define NV20TCL_STENCIL_OP_ZPASS_INCR 0x00001e02 +#define NV20TCL_STENCIL_OP_ZPASS_DECR 0x00001e03 +#define NV20TCL_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV20TCL_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV20TCL_SHADE_MODEL 0x0000037c +#define NV20TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV20TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV20TCL_LINE_WIDTH 0x00000380 +#define NV20TCL_POLYGON_OFFSET_FACTOR 0x00000384 +#define NV20TCL_POLYGON_OFFSET_UNITS 0x00000388 +#define NV20TCL_POLYGON_MODE_FRONT 0x0000038c +#define NV20TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV20TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV20TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV20TCL_POLYGON_MODE_BACK 0x00000390 +#define NV20TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV20TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV20TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV20TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV20TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV20TCL_CULL_FACE 0x0000039c +#define NV20TCL_CULL_FACE_FRONT 0x00000404 +#define NV20TCL_CULL_FACE_BACK 0x00000405 +#define NV20TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV20TCL_FRONT_FACE 0x000003a0 +#define NV20TCL_FRONT_FACE_CW 0x00000900 +#define NV20TCL_FRONT_FACE_CCW 0x00000901 +#define NV20TCL_NORMALIZE_ENABLE 0x000003a4 +#define NV20TCL_SEPARATE_SPECULAR_ENABLE 0x000003b8 +#define NV20TCL_ENABLED_LIGHTS 0x000003bc +#define NV20TCL_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) +#define NV20TCL_CLIP_PLANE_ENABLE__SIZE 0x00000010 +#define NV20TCL_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) +#define NV20TCL_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV20TCL_POINT_SIZE 0x0000043c +#define NV20TCL_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV20TCL_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV20TCL_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV20TCL_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV20TCL_TX0_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV20TCL_TX1_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV20TCL_TX2_MATRIX__SIZE 0x00000010 +#define NV20TCL_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV20TCL_TX3_MATRIX__SIZE 0x00000010 +#define NV20TCL_FOG_EQUATION_CONSTANT 0x000009c0 +#define NV20TCL_FOG_EQUATION_LINEAR 0x000009c4 +#define NV20TCL_FOG_EQUATION_QUADRATIC 0x000009c8 +#define NV20TCL_VIEWPORT_SCALE0_X 0x00000a20 +#define NV20TCL_VIEWPORT_SCALE0_Y 0x00000a24 +#define NV20TCL_VIEWPORT_SCALE0_Z 0x00000a28 +#define NV20TCL_VIEWPORT_SCALE0_W 0x00000a2c +#define NV20TCL_POINT_PARAMETER(x) (0x00000a30+((x)*4)) +#define NV20TCL_POINT_PARAMETER__SIZE 0x00000007 +#define NV20TCL_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) +#define NV20TCL_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV20TCL_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) +#define NV20TCL_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV20TCL_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) +#define NV20TCL_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV20TCL_RC_IN_RGB(x) (0x00000ac0+((x)*4)) +#define NV20TCL_RC_IN_RGB__SIZE 0x00000008 +#define NV20TCL_VIEWPORT_SCALE1_X 0x00000af0 +#define NV20TCL_VIEWPORT_SCALE1_Y 0x00000af4 +#define NV20TCL_VIEWPORT_SCALE1_Z 0x00000af8 +#define NV20TCL_VIEWPORT_SCALE1_W 0x00000afc +#define NV20TCL_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) +#define NV20TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV20TCL_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) +#define NV20TCL_VP_UPLOAD_CONST__SIZE 0x00000004 +#define NV20TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV20TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV20TCL_VERTEX_POS_3F_X 0x00001500 +#define NV20TCL_VERTEX_POS_3F_Y 0x00001504 +#define NV20TCL_VERTEX_POS_3F_Z 0x00001508 +#define NV20TCL_VERTEX_POS_4F_X 0x00001518 +#define NV20TCL_VERTEX_POS_4F_Y 0x0000151c +#define NV20TCL_VERTEX_POS_4F_Z 0x00001520 +#define NV20TCL_VERTEX_POS_3I_XY 0x00001528 +#define NV20TCL_VERTEX_POS_3I_XY_X_SHIFT 0 +#define NV20TCL_VERTEX_POS_3I_XY_X_MASK 0x0000ffff +#define NV20TCL_VERTEX_POS_3I_XY_Y_SHIFT 16 +#define NV20TCL_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 +#define NV20TCL_VERTEX_POS_3I_Z 0x0000152c +#define NV20TCL_VERTEX_POS_3I_Z_Z_SHIFT 0 +#define NV20TCL_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff +#define NV20TCL_VERTEX_NOR_3F_X 0x00001530 +#define NV20TCL_VERTEX_NOR_3F_Y 0x00001534 +#define NV20TCL_VERTEX_NOR_3F_Z 0x00001538 +#define NV20TCL_VERTEX_NOR_3I_XY 0x00001540 +#define NV20TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV20TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV20TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV20TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV20TCL_VERTEX_NOR_3I_Z 0x00001544 +#define NV20TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV20TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV20TCL_VERTEX_COL_4F_X 0x00001550 +#define NV20TCL_VERTEX_COL_4F_Y 0x00001554 +#define NV20TCL_VERTEX_COL_4F_Z 0x00001558 +#define NV20TCL_VERTEX_COL_4F_W 0x0000155c +#define NV20TCL_VERTEX_COL_3F_X 0x00001560 +#define NV20TCL_VERTEX_COL_3F_Y 0x00001564 +#define NV20TCL_VERTEX_COL_3F_Z 0x00001568 +#define NV20TCL_VERTEX_COL_4I 0x0000156c +#define NV20TCL_VERTEX_COL_4I_R_SHIFT 0 +#define NV20TCL_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV20TCL_VERTEX_COL_4I_G_SHIFT 8 +#define NV20TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV20TCL_VERTEX_COL_4I_B_SHIFT 16 +#define NV20TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV20TCL_VERTEX_COL_4I_A_SHIFT 24 +#define NV20TCL_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV20TCL_VERTEX_COL2_3F_X 0x00001580 +#define NV20TCL_VERTEX_COL2_3F_Y 0x00001584 +#define NV20TCL_VERTEX_COL2_3F_Z 0x00001588 +#define NV20TCL_VERTEX_COL2_4I 0x0000158c +#define NV20TCL_VERTEX_COL2_4I_R_SHIFT 0 +#define NV20TCL_VERTEX_COL2_4I_R_MASK 0x000000ff +#define NV20TCL_VERTEX_COL2_4I_G_SHIFT 8 +#define NV20TCL_VERTEX_COL2_4I_G_MASK 0x0000ff00 +#define NV20TCL_VERTEX_COL2_4I_B_SHIFT 16 +#define NV20TCL_VERTEX_COL2_4I_B_MASK 0x00ff0000 +#define NV20TCL_VERTEX_COL2_4I_A_SHIFT 24 +#define NV20TCL_VERTEX_COL2_4I_A_MASK 0xff000000 +#define NV20TCL_VERTEX_TX0_2F_S 0x00001590 +#define NV20TCL_VERTEX_TX0_2F_T 0x00001594 +#define NV20TCL_VERTEX_TX0_2I 0x00001598 +#define NV20TCL_VERTEX_TX0_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX0_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX0_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX0_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX0_4F_S 0x000015a0 +#define NV20TCL_VERTEX_TX0_4F_T 0x000015a4 +#define NV20TCL_VERTEX_TX0_4F_R 0x000015a8 +#define NV20TCL_VERTEX_TX0_4F_Q 0x000015ac +#define NV20TCL_VERTEX_TX0_4I_ST 0x000015b0 +#define NV20TCL_VERTEX_TX0_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX0_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX0_4I_RQ 0x000015b4 +#define NV20TCL_VERTEX_TX0_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX0_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX1_2F_S 0x000015b8 +#define NV20TCL_VERTEX_TX1_2F_T 0x000015bc +#define NV20TCL_VERTEX_TX1_2I 0x000015c0 +#define NV20TCL_VERTEX_TX1_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX1_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX1_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX1_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX1_4F_S 0x000015c8 +#define NV20TCL_VERTEX_TX1_4F_T 0x000015cc +#define NV20TCL_VERTEX_TX1_4F_R 0x000015d0 +#define NV20TCL_VERTEX_TX1_4F_Q 0x000015d4 +#define NV20TCL_VERTEX_TX1_4I_ST 0x000015d8 +#define NV20TCL_VERTEX_TX1_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX1_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX1_4I_RQ 0x000015dc +#define NV20TCL_VERTEX_TX1_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX1_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX2_2F_S 0x000015e0 +#define NV20TCL_VERTEX_TX2_2F_T 0x000015e4 +#define NV20TCL_VERTEX_TX2_2I 0x000015e8 +#define NV20TCL_VERTEX_TX2_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX2_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX2_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX2_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX2_4F_S 0x000015f0 +#define NV20TCL_VERTEX_TX2_4F_T 0x000015f4 +#define NV20TCL_VERTEX_TX2_4F_R 0x000015f8 +#define NV20TCL_VERTEX_TX2_4F_Q 0x000015fc +#define NV20TCL_VERTEX_TX2_4I_ST 0x00001600 +#define NV20TCL_VERTEX_TX2_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX2_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX2_4I_RQ 0x00001604 +#define NV20TCL_VERTEX_TX2_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX2_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX3_2F_S 0x00001608 +#define NV20TCL_VERTEX_TX3_2F_T 0x0000160c +#define NV20TCL_VERTEX_TX3_2I 0x00001610 +#define NV20TCL_VERTEX_TX3_2I_S_SHIFT 0 +#define NV20TCL_VERTEX_TX3_2I_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX3_2I_T_SHIFT 16 +#define NV20TCL_VERTEX_TX3_2I_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX3_4F_S 0x00001620 +#define NV20TCL_VERTEX_TX3_4F_T 0x00001624 +#define NV20TCL_VERTEX_TX3_4F_R 0x00001628 +#define NV20TCL_VERTEX_TX3_4F_Q 0x0000162c +#define NV20TCL_VERTEX_TX3_4I_ST 0x00001630 +#define NV20TCL_VERTEX_TX3_4I_ST_S_SHIFT 0 +#define NV20TCL_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX3_4I_ST_T_SHIFT 16 +#define NV20TCL_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 +#define NV20TCL_VERTEX_TX3_4I_RQ 0x00001634 +#define NV20TCL_VERTEX_TX3_4I_RQ_R_SHIFT 0 +#define NV20TCL_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff +#define NV20TCL_VERTEX_TX3_4I_RQ_Q_SHIFT 16 +#define NV20TCL_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 +#define NV20TCL_EDGEFLAG_ENABLE 0x000016bc +#define NV20TCL_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) +#define NV20TCL_VERTEX_ATTR_OFFSET__SIZE 0x00000010 +#define NV20TCL_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) +#define NV20TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV20TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV20TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV20TCL_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 +#define NV20TCL_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 +#define NV20TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV20TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV20TCL_COLOR_LOGIC_OP_ENABLE 0x000017bc +#define NV20TCL_COLOR_LOGIC_OP_OP 0x000017c0 +#define NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 +#define NV20TCL_TX_SHADER_CULL_MODE 0x000017f8 +#define NV20TCL_VERTEX_BEGIN_END 0x000017fc +#define NV20TCL_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 +#define NV20TCL_VERTEX_ARRAY_DATA 0x00001818 +#define NV20TCL_TX_OFFSET(x) (0x00001b00+((x)*64)) +#define NV20TCL_TX_OFFSET__SIZE 0x00000004 +#define NV20TCL_TX_FORMAT(x) (0x00001b04+((x)*64)) +#define NV20TCL_TX_FORMAT__SIZE 0x00000004 +#define NV20TCL_TX_WRAP(x) (0x00001b08+((x)*64)) +#define NV20TCL_TX_WRAP__SIZE 0x00000004 +#define NV20TCL_TX_ENABLE(x) (0x00001b0c+((x)*64)) +#define NV20TCL_TX_ENABLE__SIZE 0x00000004 +#define NV20TCL_TX_FILTER(x) (0x00001b14+((x)*64)) +#define NV20TCL_TX_FILTER__SIZE 0x00000004 +#define NV20TCL_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) +#define NV20TCL_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV20TCL_SCISSOR_HORIZ 0x00001c30 +#define NV20TCL_SCISSOR_VERT 0x00001c50 +#define NV20TCL_MULTISAMPLE_CONTROL 0x00001d7c +#define NV20TCL_CLEAR_VALUE 0x00001d90 +#define NV20TCL_CLEAR_BUFFERS 0x00001d94 +#define NV20TCL_RC_COLOR0 0x00001e20 +#define NV20TCL_RC_COLOR1 0x00001e24 +#define NV20TCL_RC_OUT_RGB(x) (0x00001e40+((x)*4)) +#define NV20TCL_RC_OUT_RGB__SIZE 0x00000008 +#define NV20TCL_RC_ENABLE 0x00001e60 +#define NV20TCL_TX_SHADER_OP 0x00001e70 +#define NV20TCL_VP_UPLOAD_CONST_ID 0x00001ea4 + + +#define NV17TCL 0x00000099 + +#define NV17TCL_DMA_IN_MEMORY4 0x000001ac +#define NV17TCL_DMA_IN_MEMORY5 0x000001b0 +#define NV17TCL_COLOR_MASK_ENABLE 0x000002bc +#define NV17TCL_LMA_DEPTH_BUFFER_PITCH 0x00000d5c +#define NV17TCL_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 +#define NV17TCL_LMA_DEPTH_FILL_VALUE 0x00000d68 +#define NV17TCL_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c +#define NV17TCL_LMA_DEPTH_ENABLE 0x00001658 + + +#define NV20_SWIZZLED_SURFACE 0x0000009e + + + +#define NV12_IMAGE_BLIT 0x0000009f + + + +#define NV30_CONTEXT_SURFACES_2D 0x00000362 + + + +#define NV30_STRETCHED_IMAGE_FROM_CPU 0x00000366 + + + +#define NV30_TEXTURE_FROM_CPU 0x0000037b + + + +#define NV30_SCALED_IMAGE_FROM_MEMORY 0x00000389 + + + +#define NV30_IMAGE_FROM_CPU 0x0000038a + + + +#define NV30TCL 0x00000397 + + + +#define NV30_SWIZZLED_SURFACE 0x0000039e + + + +#define NV35TCL 0x00000497 + + + +#define NV25TCL 0x00000597 + +#define NV25TCL_DMA_IN_MEMORY4 0x0000019c +#define NV25TCL_DMA_IN_MEMORY5 0x000001a0 +#define NV25TCL_DMA_IN_MEMORY8 0x000001ac +#define NV25TCL_DMA_IN_MEMORY9 0x000001b0 + + +#define NV34TCL 0x00000697 + +#define NV34TCL_NOP 0x00000100 +#define NV34TCL_NOTIFY 0x00000104 +#define NV34TCL_DMA_NOTIFY 0x00000180 +#define NV34TCL_DMA_TEXTURE0 0x00000184 +#define NV34TCL_DMA_TEXTURE1 0x00000188 +#define NV34TCL_DMA_COLOR1 0x0000018c +#define NV34TCL_DMA_COLOR0 0x00000194 +#define NV34TCL_DMA_ZETA 0x00000198 +#define NV34TCL_DMA_VTXBUF0 0x0000019c +#define NV34TCL_DMA_VTXBUF1 0x000001a0 +#define NV34TCL_DMA_FENCE 0x000001a4 +#define NV34TCL_DMA_QUERY 0x000001a8 +#define NV34TCL_DMA_IN_MEMORY7 0x000001ac +#define NV34TCL_DMA_IN_MEMORY8 0x000001b0 +#define NV34TCL_RT_HORIZ 0x00000200 +#define NV34TCL_RT_HORIZ_X_SHIFT 0 +#define NV34TCL_RT_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_RT_HORIZ_W_SHIFT 16 +#define NV34TCL_RT_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_RT_VERT 0x00000204 +#define NV34TCL_RT_VERT_Y_SHIFT 0 +#define NV34TCL_RT_VERT_Y_MASK 0x0000ffff +#define NV34TCL_RT_VERT_H_SHIFT 16 +#define NV34TCL_RT_VERT_H_MASK 0xffff0000 +#define NV34TCL_RT_FORMAT 0x00000208 +#define NV34TCL_RT_FORMAT_TYPE_SHIFT 8 +#define NV34TCL_RT_FORMAT_TYPE_MASK 0x00000f00 +#define NV34TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 +#define NV34TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 +#define NV34TCL_RT_FORMAT_ZETA_SHIFT 5 +#define NV34TCL_RT_FORMAT_ZETA_MASK 0x000000e0 +#define NV34TCL_RT_FORMAT_ZETA_Z16 0x00000020 +#define NV34TCL_RT_FORMAT_ZETA_Z24S8 0x00000040 +#define NV34TCL_RT_FORMAT_COLOR_SHIFT 0 +#define NV34TCL_RT_FORMAT_COLOR_MASK 0x0000001f +#define NV34TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV34TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 +#define NV34TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV34TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV34TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f +#define NV34TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 +#define NV34TCL_COLOR0_PITCH 0x0000020c +#define NV34TCL_COLOR0_PITCH_COLOR0_SHIFT 0 +#define NV34TCL_COLOR0_PITCH_COLOR0_MASK 0x0000ffff +#define NV34TCL_COLOR0_PITCH_ZETA_SHIFT 16 +#define NV34TCL_COLOR0_PITCH_ZETA_MASK 0xffff0000 +#define NV34TCL_COLOR0_OFFSET 0x00000210 +#define NV34TCL_ZETA_OFFSET 0x00000214 +#define NV34TCL_COLOR1_OFFSET 0x00000218 +#define NV34TCL_COLOR1_PITCH 0x0000021c +#define NV34TCL_RT_ENABLE 0x00000220 +#define NV34TCL_RT_ENABLE_MRT (1 << 4) +#define NV34TCL_RT_ENABLE_COLOR3 (1 << 3) +#define NV34TCL_RT_ENABLE_COLOR2 (1 << 2) +#define NV34TCL_RT_ENABLE_COLOR1 (1 << 1) +#define NV34TCL_RT_ENABLE_COLOR0 (1 << 0) +#define NV34TCL_ZETA_PITCH 0x0000022c +#define NV34TCL_LMA_DEPTH_OFFSET 0x00000230 +#define NV34TCL_TX_UNITS_ENABLE 0x0000023c +#define NV34TCL_TX_UNITS_ENABLE_TX0 (1 << 0) +#define NV34TCL_TX_UNITS_ENABLE_TX1 (1 << 1) +#define NV34TCL_TX_UNITS_ENABLE_TX2 (1 << 2) +#define NV34TCL_TX_UNITS_ENABLE_TX3 (1 << 3) +#define NV34TCL_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) +#define NV34TCL_TX_MATRIX_ENABLE__SIZE 0x00000004 +#define NV34TCL_UNK0250(x) (0x00000250+((x)*4)) +#define NV34TCL_UNK0250__SIZE 0x00000004 +#define NV34TCL_VIEWPORT_TX_ORIGIN 0x000002b8 +#define NV34TCL_VIEWPORT_TX_ORIGIN_X_SHIFT 0 +#define NV34TCL_VIEWPORT_TX_ORIGIN_X_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_TX_ORIGIN_Y_SHIFT 16 +#define NV34TCL_VIEWPORT_TX_ORIGIN_Y_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV34TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV34TCL_VIEWPORT_CLIP_HORIZ_L_SHIFT 0 +#define NV34TCL_VIEWPORT_CLIP_HORIZ_L_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_CLIP_HORIZ_R_SHIFT 16 +#define NV34TCL_VIEWPORT_CLIP_HORIZ_R_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV34TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV34TCL_VIEWPORT_CLIP_VERT_T_SHIFT 0 +#define NV34TCL_VIEWPORT_CLIP_VERT_T_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_CLIP_VERT_D_SHIFT 16 +#define NV34TCL_VIEWPORT_CLIP_VERT_D_MASK 0xffff0000 +#define NV34TCL_DITHER_ENABLE 0x00000300 +#define NV34TCL_ALPHA_FUNC_ENABLE 0x00000304 +#define NV34TCL_ALPHA_FUNC_FUNC 0x00000308 +#define NV34TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 +#define NV34TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 +#define NV34TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 +#define NV34TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34TCL_ALPHA_FUNC_REF 0x0000030c +#define NV34TCL_BLEND_FUNC_ENABLE 0x00000310 +#define NV34TCL_BLEND_FUNC_SRC 0x00000314 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV34TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV34TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV34TCL_BLEND_FUNC_DST 0x00000318 +#define NV34TCL_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV34TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV34TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV34TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV34TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV34TCL_BLEND_FUNC_COLOR 0x0000031c +#define NV34TCL_BLEND_FUNC_EQUATION 0x00000320 +#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 +#define NV34TCL_BLEND_FUNC_EQUATION_MIN 0x00008007 +#define NV34TCL_BLEND_FUNC_EQUATION_MAX 0x00008008 +#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV34TCL_COLOR_MASK 0x00000324 +#define NV34TCL_COLOR_MASK_B_SHIFT 0 +#define NV34TCL_COLOR_MASK_B_MASK 0x000000ff +#define NV34TCL_COLOR_MASK_G_SHIFT 8 +#define NV34TCL_COLOR_MASK_G_MASK 0x0000ff00 +#define NV34TCL_COLOR_MASK_R_SHIFT 16 +#define NV34TCL_COLOR_MASK_R_MASK 0x00ff0000 +#define NV34TCL_COLOR_MASK_A_SHIFT 24 +#define NV34TCL_COLOR_MASK_A_MASK 0xff000000 +#define NV34TCL_STENCIL_BACK_ENABLE 0x00000328 +#define NV34TCL_STENCIL_BACK_MASK 0x0000032c +#define NV34TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34TCL_STENCIL_BACK_FUNC_REF 0x00000334 +#define NV34TCL_STENCIL_BACK_FUNC_MASK 0x00000338 +#define NV34TCL_STENCIL_BACK_OP_FAIL 0x0000033c +#define NV34TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_BACK_OP_ZPASS 0x00000344 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV34TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_FRONT_ENABLE 0x00000348 +#define NV34TCL_STENCIL_FRONT_MASK 0x0000034c +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV34TCL_STENCIL_FRONT_FUNC_REF 0x00000354 +#define NV34TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 +#define NV34TCL_STENCIL_FRONT_OP_FAIL 0x0000035c +#define NV34TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV34TCL_SHADE_MODEL 0x00000368 +#define NV34TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV34TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV34TCL_FOG_ENABLE 0x0000036c +#define NV34TCL_FOG_COLOR 0x00000370 +#define NV34TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV34TCL_COLOR_LOGIC_OP_OP 0x00000378 +#define NV34TCL_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 +#define NV34TCL_COLOR_LOGIC_OP_OP_AND 0x00001501 +#define NV34TCL_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 +#define NV34TCL_COLOR_LOGIC_OP_OP_COPY 0x00001503 +#define NV34TCL_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 +#define NV34TCL_COLOR_LOGIC_OP_OP_NOOP 0x00001505 +#define NV34TCL_COLOR_LOGIC_OP_OP_XOR 0x00001506 +#define NV34TCL_COLOR_LOGIC_OP_OP_OR 0x00001507 +#define NV34TCL_COLOR_LOGIC_OP_OP_NOR 0x00001508 +#define NV34TCL_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 +#define NV34TCL_COLOR_LOGIC_OP_OP_INVERT 0x0000150a +#define NV34TCL_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b +#define NV34TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c +#define NV34TCL_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d +#define NV34TCL_COLOR_LOGIC_OP_OP_NAND 0x0000150e +#define NV34TCL_COLOR_LOGIC_OP_OP_SET 0x0000150f +#define NV34TCL_NORMALIZE_ENABLE 0x0000037c +#define NV34TCL_COLOR_MATERIAL 0x00000390 +#define NV34TCL_COLOR_MATERIAL_FRONT_EMISSION_ENABLE (1 << 0) +#define NV34TCL_COLOR_MATERIAL_FRONT_AMBIENT_ENABLE (1 << 2) +#define NV34TCL_COLOR_MATERIAL_FRONT_DIFFUSE_ENABLE (1 << 4) +#define NV34TCL_COLOR_MATERIAL_FRONT_SPECULAR_ENABLE (1 << 6) +#define NV34TCL_COLOR_MATERIAL_BACK_EMISSION_ENABLE (1 << 8) +#define NV34TCL_COLOR_MATERIAL_BACK_AMBIENT_ENABLE (1 << 10) +#define NV34TCL_COLOR_MATERIAL_BACK_DIFFUSE_ENABLE (1 << 12) +#define NV34TCL_COLOR_MATERIAL_BACK_SPECULAR_ENABLE (1 << 14) +#define NV34TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV34TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV34TCL_COLOR_MATERIAL_FRONT_R 0x000003a0 +#define NV34TCL_COLOR_MATERIAL_FRONT_G 0x000003a4 +#define NV34TCL_COLOR_MATERIAL_FRONT_B 0x000003a8 +#define NV34TCL_COLOR_MATERIAL_FRONT_A 0x000003b4 +#define NV34TCL_LINE_WIDTH 0x000003b8 +#define NV34TCL_LINE_SMOOTH_ENABLE 0x000003bc +#define NV34TCL_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) +#define NV34TCL_CLIP_PLANE_ENABLE__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_ENABLE_FALSE 0x00000000 +#define NV34TCL_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 +#define NV34TCL_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 +#define NV34TCL_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) +#define NV34TCL_MODELVIEW_MATRIX__SIZE 0x00000010 +#define NV34TCL_INVERSE_MODELVIEW_MATRIX(x) (0x00000580+((x)*4)) +#define NV34TCL_INVERSE_MODELVIEW_MATRIX__SIZE 0x0000000c +#define NV34TCL_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) +#define NV34TCL_PROJECTION_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX0_MATRIX(x) (0x000006c0+((x)*4)) +#define NV34TCL_TX0_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX1_MATRIX(x) (0x00000700+((x)*4)) +#define NV34TCL_TX1_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX2_MATRIX(x) (0x00000740+((x)*4)) +#define NV34TCL_TX2_MATRIX__SIZE 0x00000010 +#define NV34TCL_TX3_MATRIX(x) (0x00000780+((x)*4)) +#define NV34TCL_TX3_MATRIX__SIZE 0x00000010 +#define NV34TCL_SCISSOR_HORIZ 0x000008c0 +#define NV34TCL_SCISSOR_HORIZ_X_SHIFT 0 +#define NV34TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_SCISSOR_HORIZ_W_SHIFT 16 +#define NV34TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_SCISSOR_VERT 0x000008c4 +#define NV34TCL_SCISSOR_VERT_Y_SHIFT 0 +#define NV34TCL_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV34TCL_SCISSOR_VERT_H_SHIFT 16 +#define NV34TCL_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV34TCL_FOG_COORD_DIST 0x000008c8 +#define NV34TCL_FOG_MODE 0x000008cc +#define NV34TCL_FOG_MODE_EXP 0x00000800 +#define NV34TCL_FOG_MODE_EXP_2 0x00000802 +#define NV34TCL_FOG_MODE_EXP2 0x00000803 +#define NV34TCL_FOG_MODE_LINEAR 0x00000804 +#define NV34TCL_FOG_MODE_LINEAR_2 0x00002601 +#define NV34TCL_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV34TCL_FOG_EQUATION_LINEAR 0x000008d4 +#define NV34TCL_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV34TCL_FP_ACTIVE_PROGRAM 0x000008e4 +#define NV34TCL_FP_ACTIVE_PROGRAM_DMA0 (1 << 0) +#define NV34TCL_FP_ACTIVE_PROGRAM_DMA1 (1 << 1) +#define NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_SHIFT 2 +#define NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_MASK 0xfffffffc +#define NV34TCL_RC_COLOR0 0x000008ec +#define NV34TCL_RC_COLOR0_B_SHIFT 0 +#define NV34TCL_RC_COLOR0_B_MASK 0x000000ff +#define NV34TCL_RC_COLOR0_G_SHIFT 8 +#define NV34TCL_RC_COLOR0_G_MASK 0x0000ff00 +#define NV34TCL_RC_COLOR0_R_SHIFT 16 +#define NV34TCL_RC_COLOR0_R_MASK 0x00ff0000 +#define NV34TCL_RC_COLOR0_A_SHIFT 24 +#define NV34TCL_RC_COLOR0_A_MASK 0xff000000 +#define NV34TCL_RC_COLOR1 0x000008f0 +#define NV34TCL_RC_COLOR1_B_SHIFT 0 +#define NV34TCL_RC_COLOR1_B_MASK 0x000000ff +#define NV34TCL_RC_COLOR1_G_SHIFT 8 +#define NV34TCL_RC_COLOR1_G_MASK 0x0000ff00 +#define NV34TCL_RC_COLOR1_R_SHIFT 16 +#define NV34TCL_RC_COLOR1_R_MASK 0x00ff0000 +#define NV34TCL_RC_COLOR1_A_SHIFT 24 +#define NV34TCL_RC_COLOR1_A_MASK 0xff000000 +#define NV34TCL_RC_FINAL0 0x000008f4 +#define NV34TCL_RC_FINAL1 0x000008f8 +#define NV34TCL_RC_ENABLE 0x000008fc +#define NV34TCL_RC_IN_ALPHA(x) (0x00000900+((x)*32)) +#define NV34TCL_RC_IN_ALPHA__SIZE 0x00000008 +#define NV34TCL_RC_IN_RGB(x) (0x00000904+((x)*32)) +#define NV34TCL_RC_IN_RGB__SIZE 0x00000008 +#define NV34TCL_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) +#define NV34TCL_RC_CONSTANT_COLOR0__SIZE 0x00000008 +#define NV34TCL_RC_CONSTANT_COLOR0_B_SHIFT 0 +#define NV34TCL_RC_CONSTANT_COLOR0_B_MASK 0x000000ff +#define NV34TCL_RC_CONSTANT_COLOR0_G_SHIFT 8 +#define NV34TCL_RC_CONSTANT_COLOR0_G_MASK 0x0000ff00 +#define NV34TCL_RC_CONSTANT_COLOR0_R_SHIFT 16 +#define NV34TCL_RC_CONSTANT_COLOR0_R_MASK 0x00ff0000 +#define NV34TCL_RC_CONSTANT_COLOR0_A_SHIFT 24 +#define NV34TCL_RC_CONSTANT_COLOR0_A_MASK 0xff000000 +#define NV34TCL_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) +#define NV34TCL_RC_CONSTANT_COLOR1__SIZE 0x00000008 +#define NV34TCL_RC_CONSTANT_COLOR1_B_SHIFT 0 +#define NV34TCL_RC_CONSTANT_COLOR1_B_MASK 0x000000ff +#define NV34TCL_RC_CONSTANT_COLOR1_G_SHIFT 8 +#define NV34TCL_RC_CONSTANT_COLOR1_G_MASK 0x0000ff00 +#define NV34TCL_RC_CONSTANT_COLOR1_R_SHIFT 16 +#define NV34TCL_RC_CONSTANT_COLOR1_R_MASK 0x00ff0000 +#define NV34TCL_RC_CONSTANT_COLOR1_A_SHIFT 24 +#define NV34TCL_RC_CONSTANT_COLOR1_A_MASK 0xff000000 +#define NV34TCL_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) +#define NV34TCL_RC_OUT_ALPHA__SIZE 0x00000008 +#define NV34TCL_RC_OUT_RGB(x) (0x00000914+((x)*32)) +#define NV34TCL_RC_OUT_RGB__SIZE 0x00000008 +#define NV34TCL_VIEWPORT_HORIZ 0x00000a00 +#define NV34TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV34TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV34TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV34TCL_VIEWPORT_VERT 0x00000a04 +#define NV34TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV34TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV34TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV34TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x00000a10 +#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x00000a14 +#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x00000a18 +#define NV34TCL_VIEWPORT_TRANSLATE_X 0x00000a20 +#define NV34TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 +#define NV34TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 +#define NV34TCL_VIEWPORT_TRANSLATE_W 0x00000a2c +#define NV34TCL_VIEWPORT_SCALE_X 0x00000a30 +#define NV34TCL_VIEWPORT_SCALE_Y 0x00000a34 +#define NV34TCL_VIEWPORT_SCALE_Z 0x00000a38 +#define NV34TCL_VIEWPORT_SCALE_W 0x00000a3c +#define NV34TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV34TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV34TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV34TCL_DEPTH_FUNC 0x00000a6c +#define NV34TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV34TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV34TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV34TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV34TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV34TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV34TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV34TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV34TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV34TCL_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV34TCL_DEPTH_TEST_ENABLE 0x00000a74 +#define NV34TCL_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV34TCL_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV34TCL_VERTEX_NOR_3I_XY 0x00000a90 +#define NV34TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 +#define NV34TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff +#define NV34TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 +#define NV34TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 +#define NV34TCL_VERTEX_NOR_3I_Z 0x00000a94 +#define NV34TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 +#define NV34TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV34TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV34TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV34TCL_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) +#define NV34TCL_CLIP_PLANE_A__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) +#define NV34TCL_CLIP_PLANE_B__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) +#define NV34TCL_CLIP_PLANE_C__SIZE 0x00000020 +#define NV34TCL_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) +#define NV34TCL_CLIP_PLANE_D__SIZE 0x00000020 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) +#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) +#define NV34TCL_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) +#define NV34TCL_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) +#define NV34TCL_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) +#define NV34TCL_LIGHT_DIRECTION_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) +#define NV34TCL_LIGHT_DIRECTION_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) +#define NV34TCL_LIGHT_DIRECTION_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) +#define NV34TCL_LIGHT_SPOT_DIR_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) +#define NV34TCL_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) +#define NV34TCL_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) +#define NV34TCL_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 +#define NV34TCL_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) +#define NV34TCL_LIGHT_POSITION_X__SIZE 0x00000008 +#define NV34TCL_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) +#define NV34TCL_LIGHT_POSITION_Y__SIZE 0x00000008 +#define NV34TCL_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) +#define NV34TCL_LIGHT_POSITION_Z__SIZE 0x00000008 +#define NV34TCL_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) +#define NV34TCL_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 +#define NV34TCL_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) +#define NV34TCL_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 +#define NV34TCL_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) +#define NV34TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 +#define NV34TCL_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) +#define NV34TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34TCL_FP_REG_CONTROL 0x00001450 +#define NV34TCL_FP_REG_CONTROL_UNK1_SHIFT 16 +#define NV34TCL_FP_REG_CONTROL_UNK1_MASK 0xffff0000 +#define NV34TCL_FP_REG_CONTROL_UNK0_SHIFT 0 +#define NV34TCL_FP_REG_CONTROL_UNK0_MASK 0x0000ffff +#define NV34TCL_VP_CLIP_PLANES_ENABLE 0x00001478 +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE0 (1 << 1) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE1 (1 << 5) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE2 (1 << 9) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE3 (1 << 13) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE4 (1 << 17) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE5 (1 << 21) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE6 (1 << 25) +#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE7 (1 << 29) +#define NV34TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV34TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV34TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV34TCL_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV34TCL_VERTEX_ATTR_3F_X__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV34TCL_VERTEX_ATTR_3F_Y__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV34TCL_VERTEX_ATTR_3F_Z__SIZE 0x00000010 +#define NV34TCL_VP_CLIP_PLANE_A(x) (0x00001600+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_A__SIZE 0x00000006 +#define NV34TCL_VP_CLIP_PLANE_B(x) (0x00001604+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_B__SIZE 0x00000006 +#define NV34TCL_VP_CLIP_PLANE_C(x) (0x00001608+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_C__SIZE 0x00000006 +#define NV34TCL_VP_CLIP_PLANE_D(x) (0x0000160c+((x)*16)) +#define NV34TCL_VP_CLIP_PLANE_D__SIZE 0x00000006 +#define NV34TCL_VERTEX_BUFFER_ADDRESS(x) (0x00001680+((x)*4)) +#define NV34TCL_VERTEX_BUFFER_ADDRESS__SIZE 0x00000010 +#define NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1 (1 << 31) +#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_SHIFT 0 +#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_MASK 0x0fffffff +#define NV34TCL_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) +#define NV34TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT 0x00000002 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE 0x00000004 +#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT 4 +#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_MASK 0x000000f0 +#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 +#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000017a0 +#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000017a4 +#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000017a8 +#define NV34TCL_COLOR_MATERIAL_BACK_R 0x000017b0 +#define NV34TCL_COLOR_MATERIAL_BACK_G 0x000017b4 +#define NV34TCL_COLOR_MATERIAL_BACK_B 0x000017b8 +#define NV34TCL_COLOR_MATERIAL_BACK_A 0x000017c0 +#define NV34TCL_QUERY_RESET 0x000017c8 +#define NV34TCL_QUERY_UNK17CC 0x000017cc +#define NV34TCL_QUERY_GET 0x00001800 +#define NV34TCL_QUERY_GET_UNK24_SHIFT 24 +#define NV34TCL_QUERY_GET_UNK24_MASK 0xff000000 +#define NV34TCL_QUERY_GET_OFFSET_SHIFT 0 +#define NV34TCL_QUERY_GET_OFFSET_MASK 0x00ffffff +#define NV34TCL_VERTEX_BEGIN_END 0x00001808 +#define NV34TCL_VERTEX_BEGIN_END_STOP 0x00000000 +#define NV34TCL_VERTEX_BEGIN_END_POINTS 0x00000001 +#define NV34TCL_VERTEX_BEGIN_END_LINES 0x00000002 +#define NV34TCL_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 +#define NV34TCL_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 +#define NV34TCL_VERTEX_BEGIN_END_TRIANGLES 0x00000005 +#define NV34TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV34TCL_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV34TCL_VERTEX_BEGIN_END_QUADS 0x00000008 +#define NV34TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV34TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a +#define NV34TCL_VB_ELEMENT_U16 0x0000180c +#define NV34TCL_VB_ELEMENT_U16_I0_SHIFT 0 +#define NV34TCL_VB_ELEMENT_U16_I0_MASK 0x0000ffff +#define NV34TCL_VB_ELEMENT_U16_I1_SHIFT 16 +#define NV34TCL_VB_ELEMENT_U16_I1_MASK 0xffff0000 +#define NV34TCL_VB_ELEMENT_U32 0x00001810 +#define NV34TCL_VB_VERTEX_BATCH 0x00001814 +#define NV34TCL_VB_VERTEX_BATCH_OFFSET_SHIFT 0 +#define NV34TCL_VB_VERTEX_BATCH_OFFSET_MASK 0x00ffffff +#define NV34TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 +#define NV34TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 +#define NV34TCL_VERTEX_DATA 0x00001818 +#define NV34TCL_POLYGON_MODE_FRONT 0x00001828 +#define NV34TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV34TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV34TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV34TCL_POLYGON_MODE_BACK 0x0000182c +#define NV34TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV34TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV34TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV34TCL_CULL_FACE 0x00001830 +#define NV34TCL_CULL_FACE_FRONT 0x00000404 +#define NV34TCL_CULL_FACE_BACK 0x00000405 +#define NV34TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV34TCL_FRONT_FACE 0x00001834 +#define NV34TCL_FRONT_FACE_CW 0x00000900 +#define NV34TCL_FRONT_FACE_CCW 0x00000901 +#define NV34TCL_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV34TCL_CULL_FACE_ENABLE 0x0000183c +#define NV34TCL_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV34TCL_VERTEX_ATTR_2F_X__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV34TCL_VERTEX_ATTR_2F_Y__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV34TCL_VERTEX_ATTR_2I__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_2I_Y_SHIFT 16 +#define NV34TCL_VERTEX_ATTR_2I_Y_MASK 0xffff0000 +#define NV34TCL_VERTEX_ATTR_2I_X_SHIFT 0 +#define NV34TCL_VERTEX_ATTR_2I_X_MASK 0x0000ffff +#define NV34TCL_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) +#define NV34TCL_VERTEX_COL_4I__SIZE 0x00000002 +#define NV34TCL_VERTEX_COL_4I_R_SHIFT 0 +#define NV34TCL_VERTEX_COL_4I_R_MASK 0x000000ff +#define NV34TCL_VERTEX_COL_4I_G_SHIFT 8 +#define NV34TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 +#define NV34TCL_VERTEX_COL_4I_B_SHIFT 16 +#define NV34TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 +#define NV34TCL_VERTEX_COL_4I_A_SHIFT 24 +#define NV34TCL_VERTEX_COL_4I_A_MASK 0xff000000 +#define NV34TCL_VERTEX_POS_4I_XY 0x00001980 +#define NV34TCL_VERTEX_POS_4I_XY_X_SHIFT 0 +#define NV34TCL_VERTEX_POS_4I_XY_X_MASK 0x0000ffff +#define NV34TCL_VERTEX_POS_4I_XY_Y_SHIFT 16 +#define NV34TCL_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 +#define NV34TCL_VERTEX_POS_4I_ZW 0x00001984 +#define NV34TCL_VERTEX_POS_4I_ZW_Z_SHIFT 0 +#define NV34TCL_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff +#define NV34TCL_VERTEX_POS_4I_ZW_W_SHIFT 16 +#define NV34TCL_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 +#define NV34TCL_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) +#define NV34TCL_VERTEX_TX_4I_ST__SIZE 0x00000004 +#define NV34TCL_VERTEX_TX_4I_ST_S_SHIFT 0 +#define NV34TCL_VERTEX_TX_4I_ST_S_MASK 0x0000ffff +#define NV34TCL_VERTEX_TX_4I_ST_T_SHIFT 16 +#define NV34TCL_VERTEX_TX_4I_ST_T_MASK 0xffff0000 +#define NV34TCL_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) +#define NV34TCL_VERTEX_TX_4I_RQ__SIZE 0x00000004 +#define NV34TCL_VERTEX_TX_4I_RQ_R_SHIFT 0 +#define NV34TCL_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff +#define NV34TCL_VERTEX_TX_4I_RQ_Q_SHIFT 16 +#define NV34TCL_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 +#define NV34TCL_TX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV34TCL_TX_OFFSET__SIZE 0x00000004 +#define NV34TCL_TX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV34TCL_TX_FORMAT__SIZE 0x00000004 +#define NV34TCL_TX_FORMAT_DMA0 (1 << 0) +#define NV34TCL_TX_FORMAT_DMA1 (1 << 1) +#define NV34TCL_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV34TCL_TX_FORMAT_DIMS_SHIFT 4 +#define NV34TCL_TX_FORMAT_DIMS_MASK 0x000000f0 +#define NV34TCL_TX_FORMAT_DIMS_1D 0x00000010 +#define NV34TCL_TX_FORMAT_DIMS_2D 0x00000020 +#define NV34TCL_TX_FORMAT_DIMS_3D 0x00000030 +#define NV34TCL_TX_FORMAT_FORMAT_SHIFT 8 +#define NV34TCL_TX_FORMAT_FORMAT_MASK 0x0000ff00 +#define NV34TCL_TX_FORMAT_FORMAT_L8 0x00000000 +#define NV34TCL_TX_FORMAT_FORMAT_A8 0x00000100 +#define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000200 +#define NV34TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000300 +#define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000400 +#define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000600 +#define NV34TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000700 +#define NV34TCL_TX_FORMAT_FORMAT_INDEX8 0x00000b00 +#define NV34TCL_TX_FORMAT_FORMAT_DXT1 0x00000c00 +#define NV34TCL_TX_FORMAT_FORMAT_DXT3 0x00000e00 +#define NV34TCL_TX_FORMAT_FORMAT_DXT5 0x00000f00 +#define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00001000 +#define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00001200 +#define NV34TCL_TX_FORMAT_FORMAT_L8_RECT 0x00001300 +#define NV34TCL_TX_FORMAT_FORMAT_A8L8 0x00001a00 +#define NV34TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00001b00 +#define NV34TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 +#define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00001d00 +#define NV34TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00002000 +#define NV34TCL_TX_FORMAT_FORMAT_A16 0x00003200 +#define NV34TCL_TX_FORMAT_FORMAT_A16_RECT 0x00003500 +#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00004a00 +#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00004b00 +#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00004c00 +#define NV34TCL_TX_FORMAT_NPOT (1 << 12) +#define NV34TCL_TX_FORMAT_RECT (1 << 14) +#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 +#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 +#define NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 20 +#define NV34TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 +#define NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 24 +#define NV34TCL_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 +#define NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT 28 +#define NV34TCL_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 +#define NV34TCL_TX_WRAP(x) (0x00001a08+((x)*32)) +#define NV34TCL_TX_WRAP__SIZE 0x00000004 +#define NV34TCL_TX_WRAP_S_SHIFT 0 +#define NV34TCL_TX_WRAP_S_MASK 0x000000ff +#define NV34TCL_TX_WRAP_S_REPEAT 0x00000001 +#define NV34TCL_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV34TCL_TX_WRAP_S_CLAMP 0x00000005 +#define NV34TCL_TX_WRAP_T_SHIFT 8 +#define NV34TCL_TX_WRAP_T_MASK 0x0000ff00 +#define NV34TCL_TX_WRAP_T_REPEAT 0x00000100 +#define NV34TCL_TX_WRAP_T_MIRRORED_REPEAT 0x00000200 +#define NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE 0x00000300 +#define NV34TCL_TX_WRAP_T_CLAMP_TO_BORDER 0x00000400 +#define NV34TCL_TX_WRAP_T_CLAMP 0x00000500 +#define NV34TCL_TX_WRAP_R_SHIFT 16 +#define NV34TCL_TX_WRAP_R_MASK 0x00ff0000 +#define NV34TCL_TX_WRAP_R_REPEAT 0x00010000 +#define NV34TCL_TX_WRAP_R_MIRRORED_REPEAT 0x00020000 +#define NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE 0x00030000 +#define NV34TCL_TX_WRAP_R_CLAMP_TO_BORDER 0x00040000 +#define NV34TCL_TX_WRAP_R_CLAMP 0x00050000 +#define NV34TCL_TX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV34TCL_TX_ENABLE__SIZE 0x00000004 +#define NV34TCL_TX_ENABLE_ENABLE (1 << 30) +#define NV34TCL_TX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV34TCL_TX_SWIZZLE__SIZE 0x00000004 +#define NV34TCL_TX_SWIZZLE_S0_X_SHIFT 14 +#define NV34TCL_TX_SWIZZLE_S0_X_MASK 0x0000c000 +#define NV34TCL_TX_SWIZZLE_S0_X_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_X_ONE 0x00004000 +#define NV34TCL_TX_SWIZZLE_S0_X_S1 0x00008000 +#define NV34TCL_TX_SWIZZLE_S0_Y_SHIFT 12 +#define NV34TCL_TX_SWIZZLE_S0_Y_MASK 0x00003000 +#define NV34TCL_TX_SWIZZLE_S0_Y_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_Y_ONE 0x00001000 +#define NV34TCL_TX_SWIZZLE_S0_Y_S1 0x00002000 +#define NV34TCL_TX_SWIZZLE_S0_Z_SHIFT 10 +#define NV34TCL_TX_SWIZZLE_S0_Z_MASK 0x00000c00 +#define NV34TCL_TX_SWIZZLE_S0_Z_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_Z_ONE 0x00000400 +#define NV34TCL_TX_SWIZZLE_S0_Z_S1 0x00000800 +#define NV34TCL_TX_SWIZZLE_S0_W_SHIFT 8 +#define NV34TCL_TX_SWIZZLE_S0_W_MASK 0x00000300 +#define NV34TCL_TX_SWIZZLE_S0_W_ZERO 0x00000000 +#define NV34TCL_TX_SWIZZLE_S0_W_ONE 0x00000100 +#define NV34TCL_TX_SWIZZLE_S0_W_S1 0x00000200 +#define NV34TCL_TX_SWIZZLE_S1_X_SHIFT 6 +#define NV34TCL_TX_SWIZZLE_S1_X_MASK 0x000000c0 +#define NV34TCL_TX_SWIZZLE_S1_X_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_X_Z 0x00000040 +#define NV34TCL_TX_SWIZZLE_S1_X_Y 0x00000080 +#define NV34TCL_TX_SWIZZLE_S1_X_X 0x000000c0 +#define NV34TCL_TX_SWIZZLE_S1_Y_SHIFT 4 +#define NV34TCL_TX_SWIZZLE_S1_Y_MASK 0x00000030 +#define NV34TCL_TX_SWIZZLE_S1_Y_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_Y_Z 0x00000010 +#define NV34TCL_TX_SWIZZLE_S1_Y_Y 0x00000020 +#define NV34TCL_TX_SWIZZLE_S1_Y_X 0x00000030 +#define NV34TCL_TX_SWIZZLE_S1_Z_SHIFT 2 +#define NV34TCL_TX_SWIZZLE_S1_Z_MASK 0x0000000c +#define NV34TCL_TX_SWIZZLE_S1_Z_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_Z_Z 0x00000004 +#define NV34TCL_TX_SWIZZLE_S1_Z_Y 0x00000008 +#define NV34TCL_TX_SWIZZLE_S1_Z_X 0x0000000c +#define NV34TCL_TX_SWIZZLE_S1_W_SHIFT 0 +#define NV34TCL_TX_SWIZZLE_S1_W_MASK 0x00000003 +#define NV34TCL_TX_SWIZZLE_S1_W_W 0x00000000 +#define NV34TCL_TX_SWIZZLE_S1_W_Z 0x00000001 +#define NV34TCL_TX_SWIZZLE_S1_W_Y 0x00000002 +#define NV34TCL_TX_SWIZZLE_S1_W_X 0x00000003 +#define NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT 16 +#define NV34TCL_TX_SWIZZLE_RECT_PITCH_MASK 0xffff0000 +#define NV34TCL_TX_FILTER(x) (0x00001a14+((x)*32)) +#define NV34TCL_TX_FILTER__SIZE 0x00000004 +#define NV34TCL_TX_FILTER_MINIFY_SHIFT 16 +#define NV34TCL_TX_FILTER_MINIFY_MASK 0x000f0000 +#define NV34TCL_TX_FILTER_MINIFY_NEAREST 0x00010000 +#define NV34TCL_TX_FILTER_MINIFY_LINEAR 0x00020000 +#define NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00030000 +#define NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00040000 +#define NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00050000 +#define NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00060000 +#define NV34TCL_TX_FILTER_MAGNIFY_SHIFT 24 +#define NV34TCL_TX_FILTER_MAGNIFY_MASK 0x0f000000 +#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST 0x01000000 +#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR 0x02000000 +#define NV34TCL_TX_FILTER_SIGNED_BLUE (1 << 28) +#define NV34TCL_TX_FILTER_SIGNED_GREEN (1 << 29) +#define NV34TCL_TX_FILTER_SIGNED_RED (1 << 30) +#define NV34TCL_TX_FILTER_SIGNED_ALPHA (1 << 31) +#define NV34TCL_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) +#define NV34TCL_TX_NPOT_SIZE__SIZE 0x00000004 +#define NV34TCL_TX_NPOT_SIZE_H_SHIFT 0 +#define NV34TCL_TX_NPOT_SIZE_H_MASK 0x0000ffff +#define NV34TCL_TX_NPOT_SIZE_W_SHIFT 16 +#define NV34TCL_TX_NPOT_SIZE_W_MASK 0xffff0000 +#define NV34TCL_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV34TCL_TX_BORDER_COLOR__SIZE 0x00000004 +#define NV34TCL_TX_BORDER_COLOR_B_SHIFT 0 +#define NV34TCL_TX_BORDER_COLOR_B_MASK 0x000000ff +#define NV34TCL_TX_BORDER_COLOR_G_SHIFT 8 +#define NV34TCL_TX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV34TCL_TX_BORDER_COLOR_R_SHIFT 16 +#define NV34TCL_TX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV34TCL_TX_BORDER_COLOR_A_SHIFT 24 +#define NV34TCL_TX_BORDER_COLOR_A_MASK 0xff000000 +#define NV34TCL_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_X__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_Y__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_Z__SIZE 0x00000010 +#define NV34TCL_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV34TCL_VERTEX_ATTR_4F_W__SIZE 0x00000010 +#define NV34TCL_FP_CONTROL 0x00001d60 +#define NV34TCL_FP_CONTROL_USES_KIL (1 << 7) +#define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_SHIFT 0 +#define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_MASK 0x0000000f +#define NV34TCL_MULTISAMPLE_CONTROL 0x00001d7c +#define NV34TCL_CLEAR_DEPTH_VALUE 0x00001d8c +#define NV34TCL_CLEAR_COLOR_VALUE 0x00001d90 +#define NV34TCL_CLEAR_COLOR_VALUE_B_SHIFT 0 +#define NV34TCL_CLEAR_COLOR_VALUE_B_MASK 0x000000ff +#define NV34TCL_CLEAR_COLOR_VALUE_G_SHIFT 8 +#define NV34TCL_CLEAR_COLOR_VALUE_G_MASK 0x0000ff00 +#define NV34TCL_CLEAR_COLOR_VALUE_R_SHIFT 16 +#define NV34TCL_CLEAR_COLOR_VALUE_R_MASK 0x00ff0000 +#define NV34TCL_CLEAR_COLOR_VALUE_A_SHIFT 24 +#define NV34TCL_CLEAR_COLOR_VALUE_A_MASK 0xff000000 +#define NV34TCL_CLEAR_BUFFERS 0x00001d94 +#define NV34TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) +#define NV34TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) +#define NV34TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) +#define NV34TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) +#define NV34TCL_CLEAR_BUFFERS_STENCIL (1 << 1) +#define NV34TCL_CLEAR_BUFFERS_DEPTH (1 << 0) +#define NV34TCL_DO_VERTICES 0x00001dac +#define NV34TCL_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV34TCL_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV34TCL_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) +#define NV34TCL_BACK_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34TCL_VERTEX_FOG_1F 0x00001e54 +#define NV34TCL_VP_UPLOAD_FROM_ID 0x00001e9c +#define NV34TCL_VP_START_FROM_ID 0x00001ea0 +#define NV34TCL_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) +#define NV34TCL_POINT_PARAMETERS__SIZE 0x00000008 +#define NV34TCL_POINT_SIZE 0x00001ee0 +#define NV34TCL_POINT_PARAMETERS_ENABLE 0x00001ee4 +#define NV34TCL_POINT_SPRITE 0x00001ee8 +#define NV34TCL_POINT_SPRITE_ENABLE (1 << 0) +#define NV34TCL_POINT_SPRITE_R_MODE_SHIFT 1 +#define NV34TCL_POINT_SPRITE_R_MODE_MASK 0x00000006 +#define NV34TCL_POINT_SPRITE_R_MODE_GL_ZERO 0x00000000 +#define NV34TCL_POINT_SPRITE_R_MODE_GL_R 0x00000002 +#define NV34TCL_POINT_SPRITE_R_MODE_GL_S 0x00000004 +#define NV34TCL_POINT_SPRITE_COORD_REPLACE (1 << 11) +#define NV34TCL_VP_UPLOAD_CONST_ID 0x00001efc +#define NV34TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 +#define NV34TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 +#define NV34TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 +#define NV34TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) +#define NV34TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 +#define NV34TCL_UNK1f80(x) (0x00001f80+((x)*4)) +#define NV34TCL_UNK1f80__SIZE 0x00000010 +#define NV34TCL_VP_ATTRIB_EN 0x00001ff0 +#define NV34TCL_VP_RESULT_EN 0x00001ff4 + + +#define NV40_CONTEXT_SURFACES_2D 0x00003062 + + + +#define NV40_STRETCHED_IMAGE_FROM_CPU 0x00003066 + + + +#define NV40_TEXTURE_FROM_CPU 0x0000307b + + + +#define NV40_SCALED_IMAGE_FROM_MEMORY 0x00003089 + + + +#define NV40_IMAGE_FROM_CPU 0x0000308a + + + +#define NV40_SWIZZLED_SURFACE 0x0000309e + + + +#define NV40TCL 0x00004097 + +#define NV40TCL_REF_CNT 0x00000050 +#define NV40TCL_NOP 0x00000100 +#define NV40TCL_NOTIFY 0x00000104 +#define NV40TCL_DMA_NOTIFY 0x00000180 +#define NV40TCL_DMA_TEXTURE0 0x00000184 +#define NV40TCL_DMA_TEXTURE1 0x00000188 +#define NV40TCL_DMA_COLOR1 0x0000018c +#define NV40TCL_DMA_COLOR0 0x00000194 +#define NV40TCL_DMA_ZETA 0x00000198 +#define NV40TCL_DMA_VTXBUF0 0x0000019c +#define NV40TCL_DMA_VTXBUF1 0x000001a0 +#define NV40TCL_DMA_FENCE 0x000001a4 +#define NV40TCL_DMA_QUERY 0x000001a8 +#define NV40TCL_DMA_UNK01AC 0x000001ac +#define NV40TCL_DMA_UNK01B0 0x000001b0 +#define NV40TCL_DMA_COLOR2 0x000001b4 +#define NV40TCL_DMA_COLOR3 0x000001b8 +#define NV40TCL_RT_HORIZ 0x00000200 +#define NV40TCL_RT_HORIZ_W_SHIFT 16 +#define NV40TCL_RT_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_RT_HORIZ_X_SHIFT 0 +#define NV40TCL_RT_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_RT_VERT 0x00000204 +#define NV40TCL_RT_VERT_H_SHIFT 16 +#define NV40TCL_RT_VERT_H_MASK 0xffff0000 +#define NV40TCL_RT_VERT_Y_SHIFT 0 +#define NV40TCL_RT_VERT_Y_MASK 0x0000ffff +#define NV40TCL_RT_FORMAT 0x00000208 +#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT 24 +#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_MASK 0xff000000 +#define NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT 16 +#define NV40TCL_RT_FORMAT_LOG2_WIDTH_MASK 0x00ff0000 +#define NV40TCL_RT_FORMAT_TYPE_SHIFT 8 +#define NV40TCL_RT_FORMAT_TYPE_MASK 0x00000f00 +#define NV40TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 +#define NV40TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 +#define NV40TCL_RT_FORMAT_ZETA_SHIFT 5 +#define NV40TCL_RT_FORMAT_ZETA_MASK 0x000000e0 +#define NV40TCL_RT_FORMAT_ZETA_Z16 0x00000020 +#define NV40TCL_RT_FORMAT_ZETA_Z24S8 0x00000040 +#define NV40TCL_RT_FORMAT_COLOR_SHIFT 0 +#define NV40TCL_RT_FORMAT_COLOR_MASK 0x0000001f +#define NV40TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV40TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 +#define NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV40TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV40TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f +#define NV40TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 +#define NV40TCL_COLOR0_PITCH 0x0000020c +#define NV40TCL_COLOR0_OFFSET 0x00000210 +#define NV40TCL_ZETA_OFFSET 0x00000214 +#define NV40TCL_COLOR1_OFFSET 0x00000218 +#define NV40TCL_COLOR1_PITCH 0x0000021c +#define NV40TCL_RT_ENABLE 0x00000220 +#define NV40TCL_RT_ENABLE_MRT (1 << 4) +#define NV40TCL_RT_ENABLE_COLOR3 (1 << 3) +#define NV40TCL_RT_ENABLE_COLOR2 (1 << 2) +#define NV40TCL_RT_ENABLE_COLOR1 (1 << 1) +#define NV40TCL_RT_ENABLE_COLOR0 (1 << 0) +#define NV40TCL_ZETA_PITCH 0x0000022c +#define NV40TCL_COLOR2_PITCH 0x00000280 +#define NV40TCL_COLOR3_PITCH 0x00000284 +#define NV40TCL_COLOR2_OFFSET 0x00000288 +#define NV40TCL_COLOR3_OFFSET 0x0000028c +#define NV40TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) +#define NV40TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV40TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) +#define NV40TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV40TCL_DITHER_ENABLE 0x00000300 +#define NV40TCL_ALPHA_TEST_ENABLE 0x00000304 +#define NV40TCL_ALPHA_TEST_FUNC 0x00000308 +#define NV40TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 +#define NV40TCL_ALPHA_TEST_FUNC_LESS 0x00000201 +#define NV40TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 +#define NV40TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 +#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV40TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 +#define NV40TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 +#define NV40TCL_ALPHA_TEST_REF 0x0000030c +#define NV40TCL_BLEND_ENABLE 0x00000310 +#define NV40TCL_BLEND_FUNC_SRC 0x00000314 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 +#define NV40TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV40TCL_BLEND_FUNC_DST 0x00000318 +#define NV40TCL_BLEND_FUNC_DST_RGB_SHIFT 0 +#define NV40TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV40TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV40TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 +#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 +#define NV40TCL_BLEND_COLOR 0x0000031c +#define NV40TCL_BLEND_EQUATION 0x00000320 +#define NV40TCL_BLEND_EQUATION_RGB_SHIFT 0 +#define NV40TCL_BLEND_EQUATION_RGB_MASK 0x0000ffff +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 +#define NV40TCL_BLEND_EQUATION_RGB_MIN 0x00008007 +#define NV40TCL_BLEND_EQUATION_RGB_MAX 0x00008008 +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a +#define NV40TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV40TCL_BLEND_EQUATION_ALPHA_SHIFT 16 +#define NV40TCL_BLEND_EQUATION_ALPHA_MASK 0xffff0000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x80060000 +#define NV40TCL_BLEND_EQUATION_ALPHA_MIN 0x80070000 +#define NV40TCL_BLEND_EQUATION_ALPHA_MAX 0x80080000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x800a0000 +#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x800b0000 +#define NV40TCL_COLOR_MASK 0x00000324 +#define NV40TCL_COLOR_MASK_BUFFER0_B_SHIFT 0 +#define NV40TCL_COLOR_MASK_BUFFER0_B_MASK 0x000000ff +#define NV40TCL_COLOR_MASK_BUFFER0_G_SHIFT 8 +#define NV40TCL_COLOR_MASK_BUFFER0_G_MASK 0x0000ff00 +#define NV40TCL_COLOR_MASK_BUFFER0_R_SHIFT 16 +#define NV40TCL_COLOR_MASK_BUFFER0_R_MASK 0x00ff0000 +#define NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT 24 +#define NV40TCL_COLOR_MASK_BUFFER0_A_MASK 0xff000000 +#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000328 +#define NV40TCL_STENCIL_FRONT_MASK 0x0000032c +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000330 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000334 +#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000338 +#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000033c +#define NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000340 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000344 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_ENABLE 0x00000348 +#define NV40TCL_STENCIL_BACK_MASK 0x0000034c +#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000350 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000354 +#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000358 +#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000035c +#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000360 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000364 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV40TCL_SHADE_MODEL 0x00000368 +#define NV40TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV40TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV40TCL_MRT_COLOR_MASK 0x00000370 +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_A (1 << 4) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_R (1 << 5) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_G (1 << 6) +#define NV40TCL_MRT_COLOR_MASK_BUFFER1_B (1 << 7) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_A (1 << 8) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_R (1 << 9) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_G (1 << 10) +#define NV40TCL_MRT_COLOR_MASK_BUFFER2_B (1 << 11) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_A (1 << 12) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_R (1 << 13) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_G (1 << 14) +#define NV40TCL_MRT_COLOR_MASK_BUFFER3_B (1 << 15) +#define NV40TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 +#define NV40TCL_COLOR_LOGIC_OP 0x00000378 +#define NV40TCL_COLOR_LOGIC_OP_CLEAR 0x00001500 +#define NV40TCL_COLOR_LOGIC_OP_AND 0x00001501 +#define NV40TCL_COLOR_LOGIC_OP_AND_REVERSE 0x00001502 +#define NV40TCL_COLOR_LOGIC_OP_COPY 0x00001503 +#define NV40TCL_COLOR_LOGIC_OP_AND_INVERTED 0x00001504 +#define NV40TCL_COLOR_LOGIC_OP_NOOP 0x00001505 +#define NV40TCL_COLOR_LOGIC_OP_XOR 0x00001506 +#define NV40TCL_COLOR_LOGIC_OP_OR 0x00001507 +#define NV40TCL_COLOR_LOGIC_OP_NOR 0x00001508 +#define NV40TCL_COLOR_LOGIC_OP_EQUIV 0x00001509 +#define NV40TCL_COLOR_LOGIC_OP_INVERT 0x0000150a +#define NV40TCL_COLOR_LOGIC_OP_OR_REVERSE 0x0000150b +#define NV40TCL_COLOR_LOGIC_OP_COPY_INVERTED 0x0000150c +#define NV40TCL_COLOR_LOGIC_OP_OR_INVERTED 0x0000150d +#define NV40TCL_COLOR_LOGIC_OP_NAND 0x0000150e +#define NV40TCL_COLOR_LOGIC_OP_SET 0x0000150f +#define NV40TCL_DEPTH_RANGE_NEAR 0x00000394 +#define NV40TCL_DEPTH_RANGE_FAR 0x00000398 +#define NV40TCL_LINE_WIDTH 0x000003b8 +#define NV40TCL_LINE_SMOOTH_ENABLE 0x000003bc +#define NV40TCL_UNK03C0(x) (0x000003c0+((x)*4)) +#define NV40TCL_UNK03C0__SIZE 0x00000010 +#define NV40TCL_UNK0400(x) (0x00000400+((x)*4)) +#define NV40TCL_UNK0400__SIZE 0x00000010 +#define NV40TCL_UNK0440(x) (0x00000440+((x)*4)) +#define NV40TCL_UNK0440__SIZE 0x00000020 +#define NV40TCL_SCISSOR_HORIZ 0x000008c0 +#define NV40TCL_SCISSOR_HORIZ_X_SHIFT 0 +#define NV40TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_SCISSOR_HORIZ_W_SHIFT 16 +#define NV40TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_SCISSOR_VERT 0x000008c4 +#define NV40TCL_SCISSOR_VERT_Y_SHIFT 0 +#define NV40TCL_SCISSOR_VERT_Y_MASK 0x0000ffff +#define NV40TCL_SCISSOR_VERT_H_SHIFT 16 +#define NV40TCL_SCISSOR_VERT_H_MASK 0xffff0000 +#define NV40TCL_FOG_MODE 0x000008cc +#define NV40TCL_FOG_EQUATION_CONSTANT 0x000008d0 +#define NV40TCL_FOG_EQUATION_LINEAR 0x000008d4 +#define NV40TCL_FOG_EQUATION_QUADRATIC 0x000008d8 +#define NV40TCL_FP_ADDRESS 0x000008e4 +#define NV40TCL_FP_ADDRESS_OFFSET_SHIFT 8 +#define NV40TCL_FP_ADDRESS_OFFSET_MASK 0xffffff00 +#define NV40TCL_FP_ADDRESS_DMA1 (1 << 1) +#define NV40TCL_FP_ADDRESS_DMA0 (1 << 0) +#define NV40TCL_VIEWPORT_HORIZ 0x00000a00 +#define NV40TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV40TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV40TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV40TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV40TCL_VIEWPORT_VERT 0x00000a04 +#define NV40TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV40TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV40TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV40TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV40TCL_VIEWPORT_TRANSLATE_X 0x00000a20 +#define NV40TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 +#define NV40TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 +#define NV40TCL_VIEWPORT_TRANSLATE_W 0x00000a2c +#define NV40TCL_VIEWPORT_SCALE_X 0x00000a30 +#define NV40TCL_VIEWPORT_SCALE_Y 0x00000a34 +#define NV40TCL_VIEWPORT_SCALE_Z 0x00000a38 +#define NV40TCL_VIEWPORT_SCALE_W 0x00000a3c +#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV40TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 +#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV40TCL_DEPTH_FUNC 0x00000a6c +#define NV40TCL_DEPTH_FUNC_NEVER 0x00000200 +#define NV40TCL_DEPTH_FUNC_LESS 0x00000201 +#define NV40TCL_DEPTH_FUNC_EQUAL 0x00000202 +#define NV40TCL_DEPTH_FUNC_LEQUAL 0x00000203 +#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 +#define NV40TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 +#define NV40TCL_DEPTH_FUNC_GEQUAL 0x00000206 +#define NV40TCL_DEPTH_FUNC_ALWAYS 0x00000207 +#define NV40TCL_DEPTH_WRITE_ENABLE 0x00000a70 +#define NV40TCL_DEPTH_TEST_ENABLE 0x00000a74 +#define NV40TCL_POLYGON_OFFSET_FACTOR 0x00000a78 +#define NV40TCL_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV40TCL_UNK0B40(x) (0x00000b40+((x)*4)) +#define NV40TCL_UNK0B40__SIZE 0x00000008 +#define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) +#define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 +#define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) +#define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c +#define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) +#define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV40TCL_VTX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV40TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 +#define NV40TCL_VTXBUF_ADDRESS(x) (0x00001680+((x)*4)) +#define NV40TCL_VTXBUF_ADDRESS__SIZE 0x00000010 +#define NV40TCL_VTXBUF_ADDRESS_DMA1 (1 << 31) +#define NV40TCL_VTXBUF_ADDRESS_OFFSET_SHIFT 0 +#define NV40TCL_VTXBUF_ADDRESS_OFFSET_MASK 0x0fffffff +#define NV40TCL_VTX_CACHE_INVALIDATE 0x00001714 +#define NV40TCL_VTXFMT(x) (0x00001740+((x)*4)) +#define NV40TCL_VTXFMT__SIZE 0x00000010 +#define NV40TCL_VTXFMT_TYPE_SHIFT 0 +#define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f +#define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 +#define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 +#define NV40TCL_VTXFMT_SIZE_SHIFT 4 +#define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 +#define NV40TCL_VTXFMT_STRIDE_SHIFT 8 +#define NV40TCL_VTXFMT_STRIDE_MASK 0x0000ff00 +#define NV40TCL_QUERY_RESET 0x000017c8 +#define NV40TCL_QUERY_UNK17CC 0x000017cc +#define NV40TCL_QUERY_GET 0x00001800 +#define NV40TCL_QUERY_GET_UNK24_SHIFT 24 +#define NV40TCL_QUERY_GET_UNK24_MASK 0xff000000 +#define NV40TCL_QUERY_GET_OFFSET_SHIFT 0 +#define NV40TCL_QUERY_GET_OFFSET_MASK 0x00ffffff +#define NV40TCL_BEGIN_END 0x00001808 +#define NV40TCL_BEGIN_END_STOP 0x00000000 +#define NV40TCL_BEGIN_END_POINTS 0x00000001 +#define NV40TCL_BEGIN_END_LINES 0x00000002 +#define NV40TCL_BEGIN_END_LINE_LOOP 0x00000003 +#define NV40TCL_BEGIN_END_LINE_STRIP 0x00000004 +#define NV40TCL_BEGIN_END_TRIANGLES 0x00000005 +#define NV40TCL_BEGIN_END_TRIANGLE_STRIP 0x00000006 +#define NV40TCL_BEGIN_END_TRIANGLE_FAN 0x00000007 +#define NV40TCL_BEGIN_END_QUADS 0x00000008 +#define NV40TCL_BEGIN_END_QUAD_STRIP 0x00000009 +#define NV40TCL_BEGIN_END_POLYGON 0x0000000a +#define NV40TCL_VB_ELEMENT_U16 0x0000180c +#define NV40TCL_VB_ELEMENT_U16_1_SHIFT 16 +#define NV40TCL_VB_ELEMENT_U16_1_MASK 0xffff0000 +#define NV40TCL_VB_ELEMENT_U16_0_SHIFT 0 +#define NV40TCL_VB_ELEMENT_U16_0_MASK 0x0000ffff +#define NV40TCL_VB_ELEMENT_U32 0x00001810 +#define NV40TCL_VB_VERTEX_BATCH 0x00001814 +#define NV40TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 +#define NV40TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 +#define NV40TCL_VB_VERTEX_BATCH_START_SHIFT 0 +#define NV40TCL_VB_VERTEX_BATCH_START_MASK 0x00ffffff +#define NV40TCL_VERTEX_DATA 0x00001818 +#define NV40TCL_IDXBUF_ADDRESS 0x0000181c +#define NV40TCL_IDXBUF_FORMAT 0x00001820 +#define NV40TCL_IDXBUF_FORMAT_TYPE_SHIFT 4 +#define NV40TCL_IDXBUF_FORMAT_TYPE_MASK 0x000000f0 +#define NV40TCL_IDXBUF_FORMAT_TYPE_U32 0x00000000 +#define NV40TCL_IDXBUF_FORMAT_TYPE_U16 0x00000010 +#define NV40TCL_IDXBUF_FORMAT_DMA1 (1 << 0) +#define NV40TCL_VB_INDEX_BATCH 0x00001824 +#define NV40TCL_VB_INDEX_BATCH_COUNT_SHIFT 24 +#define NV40TCL_VB_INDEX_BATCH_COUNT_MASK 0xff000000 +#define NV40TCL_VB_INDEX_BATCH_START_SHIFT 0 +#define NV40TCL_VB_INDEX_BATCH_START_MASK 0x00ffffff +#define NV40TCL_POLYGON_MODE_FRONT 0x00001828 +#define NV40TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV40TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV40TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV40TCL_POLYGON_MODE_BACK 0x0000182c +#define NV40TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV40TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV40TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV40TCL_CULL_FACE 0x00001830 +#define NV40TCL_CULL_FACE_FRONT 0x00000404 +#define NV40TCL_CULL_FACE_BACK 0x00000405 +#define NV40TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV40TCL_FRONT_FACE 0x00001834 +#define NV40TCL_FRONT_FACE_CW 0x00000900 +#define NV40TCL_FRONT_FACE_CCW 0x00000901 +#define NV40TCL_POLYGON_SMOOTH_ENABLE 0x00001838 +#define NV40TCL_CULL_FACE_ENABLE 0x0000183c +#define NV40TCL_TEX_SIZE1(x) (0x00001840+((x)*4)) +#define NV40TCL_TEX_SIZE1__SIZE 0x00000008 +#define NV40TCL_TEX_SIZE1_DEPTH_SHIFT 20 +#define NV40TCL_TEX_SIZE1_DEPTH_MASK 0xfff00000 +#define NV40TCL_TEX_SIZE1_PITCH_SHIFT 0 +#define NV40TCL_TEX_SIZE1_PITCH_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV40TCL_VTX_ATTR_2F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV40TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV40TCL_VTX_ATTR_2I__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_2I_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_2I_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_0(x) (0x00001900+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_0__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_0_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_0_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_1(x) (0x00001904+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_1__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_1_W_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_1_Z_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff +#define NV40TCL_TEX_OFFSET(x) (0x00001a00+((x)*32)) +#define NV40TCL_TEX_OFFSET__SIZE 0x00000010 +#define NV40TCL_TEX_FORMAT(x) (0x00001a04+((x)*32)) +#define NV40TCL_TEX_FORMAT__SIZE 0x00000010 +#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT 16 +#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_MASK 0x000f0000 +#define NV40TCL_TEX_FORMAT_RECT (1 << 14) +#define NV40TCL_TEX_FORMAT_LINEAR (1 << 13) +#define NV40TCL_TEX_FORMAT_FORMAT_SHIFT 8 +#define NV40TCL_TEX_FORMAT_FORMAT_MASK 0x00001f00 +#define NV40TCL_TEX_FORMAT_FORMAT_L8 0x00000100 +#define NV40TCL_TEX_FORMAT_FORMAT_A1R5G5B5 0x00000200 +#define NV40TCL_TEX_FORMAT_FORMAT_A4R4G4B4 0x00000300 +#define NV40TCL_TEX_FORMAT_FORMAT_R5G6B5 0x00000400 +#define NV40TCL_TEX_FORMAT_FORMAT_A8R8G8B8 0x00000500 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT1 0x00000600 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT3 0x00000700 +#define NV40TCL_TEX_FORMAT_FORMAT_DXT5 0x00000800 +#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00000b00 +#define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x00001000 +#define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x00001200 +#define NV40TCL_TEX_FORMAT_FORMAT_HILO8 0x00001800 +#define NV40TCL_TEX_FORMAT_FORMAT_RGBA16F 0x00001a00 +#define NV40TCL_TEX_FORMAT_FORMAT_RGBA32F 0x00001b00 +#define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 +#define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 +#define NV40TCL_TEX_FORMAT_DIMS_1D 0x00000010 +#define NV40TCL_TEX_FORMAT_DIMS_2D 0x00000020 +#define NV40TCL_TEX_FORMAT_DIMS_3D 0x00000030 +#define NV40TCL_TEX_FORMAT_NO_BORDER (1 << 3) +#define NV40TCL_TEX_FORMAT_CUBIC (1 << 2) +#define NV40TCL_TEX_FORMAT_DMA1 (1 << 1) +#define NV40TCL_TEX_FORMAT_DMA0 (1 << 0) +#define NV40TCL_TEX_WRAP(x) (0x00001a08+((x)*32)) +#define NV40TCL_TEX_WRAP__SIZE 0x00000010 +#define NV40TCL_TEX_WRAP_S_SHIFT 0 +#define NV40TCL_TEX_WRAP_S_MASK 0x000000ff +#define NV40TCL_TEX_WRAP_S_REPEAT 0x00000001 +#define NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT 0x00000002 +#define NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE 0x00000003 +#define NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER 0x00000004 +#define NV40TCL_TEX_WRAP_S_CLAMP 0x00000005 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE 0x00000006 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER 0x00000007 +#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP 0x00000008 +#define NV40TCL_TEX_WRAP_T_SHIFT 8 +#define NV40TCL_TEX_WRAP_T_MASK 0x0000ff00 +#define NV40TCL_TEX_WRAP_T_REPEAT 0x00000100 +#define NV40TCL_TEX_WRAP_T_MIRRORED_REPEAT 0x00000200 +#define NV40TCL_TEX_WRAP_T_CLAMP_TO_EDGE 0x00000300 +#define NV40TCL_TEX_WRAP_T_CLAMP_TO_BORDER 0x00000400 +#define NV40TCL_TEX_WRAP_T_CLAMP 0x00000500 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_EDGE 0x00000600 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_BORDER 0x00000700 +#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP 0x00000800 +#define NV40TCL_TEX_WRAP_R_SHIFT 16 +#define NV40TCL_TEX_WRAP_R_MASK 0x00ff0000 +#define NV40TCL_TEX_WRAP_R_REPEAT 0x00010000 +#define NV40TCL_TEX_WRAP_R_MIRRORED_REPEAT 0x00020000 +#define NV40TCL_TEX_WRAP_R_CLAMP_TO_EDGE 0x00030000 +#define NV40TCL_TEX_WRAP_R_CLAMP_TO_BORDER 0x00040000 +#define NV40TCL_TEX_WRAP_R_CLAMP 0x00050000 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE 0x00060000 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER 0x00070000 +#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP 0x00080000 +#define NV40TCL_TEX_WRAP_RCOMP_SHIFT 28 +#define NV40TCL_TEX_WRAP_RCOMP_MASK 0xf0000000 +#define NV40TCL_TEX_WRAP_RCOMP_NEVER 0x00000000 +#define NV40TCL_TEX_WRAP_RCOMP_GREATER 0x10000000 +#define NV40TCL_TEX_WRAP_RCOMP_EQUAL 0x20000000 +#define NV40TCL_TEX_WRAP_RCOMP_GEQUAL 0x30000000 +#define NV40TCL_TEX_WRAP_RCOMP_LESS 0x40000000 +#define NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL 0x50000000 +#define NV40TCL_TEX_WRAP_RCOMP_LEQUAL 0x60000000 +#define NV40TCL_TEX_WRAP_RCOMP_ALWAYS 0x70000000 +#define NV40TCL_TEX_ENABLE(x) (0x00001a0c+((x)*32)) +#define NV40TCL_TEX_ENABLE__SIZE 0x00000010 +#define NV40TCL_TEX_ENABLE_ENABLE (1 << 31) +#define NV40TCL_TEX_ENABLE_ANISO_SHIFT 4 +#define NV40TCL_TEX_ENABLE_ANISO_MASK 0x000000f0 +#define NV40TCL_TEX_ENABLE_ANISO_NONE 0x00000000 +#define NV40TCL_TEX_ENABLE_ANISO_2X 0x00000010 +#define NV40TCL_TEX_ENABLE_ANISO_4X 0x00000020 +#define NV40TCL_TEX_ENABLE_ANISO_6X 0x00000030 +#define NV40TCL_TEX_ENABLE_ANISO_8X 0x00000040 +#define NV40TCL_TEX_ENABLE_ANISO_10X 0x00000050 +#define NV40TCL_TEX_ENABLE_ANISO_12X 0x00000060 +#define NV40TCL_TEX_ENABLE_ANISO_16X 0x00000070 +#define NV40TCL_TEX_SWIZZLE(x) (0x00001a10+((x)*32)) +#define NV40TCL_TEX_SWIZZLE__SIZE 0x00000010 +#define NV40TCL_TEX_SWIZZLE_S0_X_SHIFT 14 +#define NV40TCL_TEX_SWIZZLE_S0_X_MASK 0x0000c000 +#define NV40TCL_TEX_SWIZZLE_S0_X_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_X_ONE 0x00004000 +#define NV40TCL_TEX_SWIZZLE_S0_X_S1 0x00008000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT 12 +#define NV40TCL_TEX_SWIZZLE_S0_Y_MASK 0x00003000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_ONE 0x00001000 +#define NV40TCL_TEX_SWIZZLE_S0_Y_S1 0x00002000 +#define NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT 10 +#define NV40TCL_TEX_SWIZZLE_S0_Z_MASK 0x00000c00 +#define NV40TCL_TEX_SWIZZLE_S0_Z_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_Z_ONE 0x00000400 +#define NV40TCL_TEX_SWIZZLE_S0_Z_S1 0x00000800 +#define NV40TCL_TEX_SWIZZLE_S0_W_SHIFT 8 +#define NV40TCL_TEX_SWIZZLE_S0_W_MASK 0x00000300 +#define NV40TCL_TEX_SWIZZLE_S0_W_ZERO 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S0_W_ONE 0x00000100 +#define NV40TCL_TEX_SWIZZLE_S0_W_S1 0x00000200 +#define NV40TCL_TEX_SWIZZLE_S1_X_SHIFT 6 +#define NV40TCL_TEX_SWIZZLE_S1_X_MASK 0x000000c0 +#define NV40TCL_TEX_SWIZZLE_S1_X_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_X_Z 0x00000040 +#define NV40TCL_TEX_SWIZZLE_S1_X_Y 0x00000080 +#define NV40TCL_TEX_SWIZZLE_S1_X_X 0x000000c0 +#define NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT 4 +#define NV40TCL_TEX_SWIZZLE_S1_Y_MASK 0x00000030 +#define NV40TCL_TEX_SWIZZLE_S1_Y_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_Y_Z 0x00000010 +#define NV40TCL_TEX_SWIZZLE_S1_Y_Y 0x00000020 +#define NV40TCL_TEX_SWIZZLE_S1_Y_X 0x00000030 +#define NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT 2 +#define NV40TCL_TEX_SWIZZLE_S1_Z_MASK 0x0000000c +#define NV40TCL_TEX_SWIZZLE_S1_Z_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_Z_Z 0x00000004 +#define NV40TCL_TEX_SWIZZLE_S1_Z_Y 0x00000008 +#define NV40TCL_TEX_SWIZZLE_S1_Z_X 0x0000000c +#define NV40TCL_TEX_SWIZZLE_S1_W_SHIFT 0 +#define NV40TCL_TEX_SWIZZLE_S1_W_MASK 0x00000003 +#define NV40TCL_TEX_SWIZZLE_S1_W_W 0x00000000 +#define NV40TCL_TEX_SWIZZLE_S1_W_Z 0x00000001 +#define NV40TCL_TEX_SWIZZLE_S1_W_Y 0x00000002 +#define NV40TCL_TEX_SWIZZLE_S1_W_X 0x00000003 +#define NV40TCL_TEX_FILTER(x) (0x00001a14+((x)*32)) +#define NV40TCL_TEX_FILTER__SIZE 0x00000010 +#define NV40TCL_TEX_FILTER_SIGNED_ALPHA (1 << 31) +#define NV40TCL_TEX_FILTER_SIGNED_RED (1 << 30) +#define NV40TCL_TEX_FILTER_SIGNED_GREEN (1 << 29) +#define NV40TCL_TEX_FILTER_SIGNED_BLUE (1 << 28) +#define NV40TCL_TEX_FILTER_MIN_SHIFT 16 +#define NV40TCL_TEX_FILTER_MIN_MASK 0x000f0000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST 0x00010000 +#define NV40TCL_TEX_FILTER_MIN_LINEAR 0x00020000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST 0x00030000 +#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST 0x00040000 +#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR 0x00050000 +#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR 0x00060000 +#define NV40TCL_TEX_FILTER_MAG_SHIFT 24 +#define NV40TCL_TEX_FILTER_MAG_MASK 0x0f000000 +#define NV40TCL_TEX_FILTER_MAG_NEAREST 0x01000000 +#define NV40TCL_TEX_FILTER_MAG_LINEAR 0x02000000 +#define NV40TCL_TEX_SIZE0(x) (0x00001a18+((x)*32)) +#define NV40TCL_TEX_SIZE0__SIZE 0x00000010 +#define NV40TCL_TEX_SIZE0_H_SHIFT 0 +#define NV40TCL_TEX_SIZE0_H_MASK 0x0000ffff +#define NV40TCL_TEX_SIZE0_W_SHIFT 16 +#define NV40TCL_TEX_SIZE0_W_MASK 0xffff0000 +#define NV40TCL_TEX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) +#define NV40TCL_TEX_BORDER_COLOR__SIZE 0x00000010 +#define NV40TCL_TEX_BORDER_COLOR_B_SHIFT 0 +#define NV40TCL_TEX_BORDER_COLOR_B_MASK 0x000000ff +#define NV40TCL_TEX_BORDER_COLOR_G_SHIFT 8 +#define NV40TCL_TEX_BORDER_COLOR_G_MASK 0x0000ff00 +#define NV40TCL_TEX_BORDER_COLOR_R_SHIFT 16 +#define NV40TCL_TEX_BORDER_COLOR_R_MASK 0x00ff0000 +#define NV40TCL_TEX_BORDER_COLOR_A_SHIFT 24 +#define NV40TCL_TEX_BORDER_COLOR_A_MASK 0xff000000 +#define NV40TCL_VTX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_X__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV40TCL_VTX_ATTR_4F_W__SIZE 0x00000010 +#define NV40TCL_FP_CONTROL 0x00001d60 +#define NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT 24 +#define NV40TCL_FP_CONTROL_TEMP_COUNT_MASK 0xff000000 +#define NV40TCL_FP_CONTROL_KIL (1 << 7) +#define NV40TCL_MULTISAMPLE_CONTROL 0x00001d7c +#define NV40TCL_CLEAR_VALUE_DEPTH 0x00001d8c +#define NV40TCL_CLEAR_VALUE_COLOR 0x00001d90 +#define NV40TCL_CLEAR_BUFFERS 0x00001d94 +#define NV40TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) +#define NV40TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) +#define NV40TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) +#define NV40TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) +#define NV40TCL_CLEAR_BUFFERS_STENCIL (1 << 1) +#define NV40TCL_CLEAR_BUFFERS_DEPTH (1 << 0) +#define NV40TCL_LINE_STIPPLE_ENABLE 0x00001db4 +#define NV40TCL_LINE_STIPPLE_PATTERN 0x00001db8 +#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 +#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff +#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 +#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV40TCL_VP_UPLOAD_FROM_ID 0x00001e9c +#define NV40TCL_VP_START_FROM_ID 0x00001ea0 +#define NV40TCL_POINT_SIZE 0x00001ee0 +#define NV40TCL_POINT_SPRITE 0x00001ee8 +#define NV40TCL_VP_UPLOAD_CONST_ID 0x00001efc +#define NV40TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 +#define NV40TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) +#define NV40TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 +#define NV40TCL_TEX_CACHE_CTL 0x00001fd8 +#define NV40TCL_VP_ATTRIB_EN 0x00001ff0 +#define NV40TCL_VP_RESULT_EN 0x00001ff4 + + +#define NV44TCL 0x00004497 + + + +#define NV50_2D 0x0000502d + +#define NV50_2D_NOP 0x00000100 +#define NV50_2D_NOTIFY 0x00000104 +#define NV50_2D_DMA_NOTIFY 0x00000180 +#define NV50_2D_DMA_IN_MEMORY0 0x00000184 +#define NV50_2D_DMA_IN_MEMORY1 0x00000188 +#define NV50_2D_DMA_IN_MEMORY2 0x0000018c +#define NV50_2D_DST_FORMAT 0x00000200 +#define NV50_2D_DST_FORMAT_32BPP 0x000000cf +#define NV50_2D_DST_FORMAT_24BPP 0x000000e6 +#define NV50_2D_DST_FORMAT_16BPP 0x000000e8 +#define NV50_2D_DST_FORMAT_8BPP 0x000000f3 +#define NV50_2D_DST_FORMAT_15BPP 0x000000f8 +#define NV50_2D_DST_PITCH 0x00000214 +#define NV50_2D_DST_WIDTH 0x00000218 +#define NV50_2D_DST_HEIGHT 0x0000021c +#define NV50_2D_DST_ADDRESS_HIGH 0x00000220 +#define NV50_2D_DST_ADDRESS_LOW 0x00000224 +#define NV50_2D_SRC_FORMAT 0x00000230 +#define NV50_2D_SRC_FORMAT_32BPP 0x000000cf +#define NV50_2D_SRC_FORMAT_24BPP 0x000000e6 +#define NV50_2D_SRC_FORMAT_16BPP 0x000000e8 +#define NV50_2D_SRC_FORMAT_8BPP 0x000000f3 +#define NV50_2D_SRC_FORMAT_15BPP 0x000000f8 +#define NV50_2D_SRC_PITCH 0x00000244 +#define NV50_2D_SRC_WIDTH 0x00000248 +#define NV50_2D_SRC_HEIGHT 0x0000024c +#define NV50_2D_SRC_ADDRESS_HIGH 0x00000250 +#define NV50_2D_SRC_ADDRESS_LOW 0x00000254 +#define NV50_2D_CLIP_X 0x00000280 +#define NV50_2D_CLIP_Y 0x00000284 +#define NV50_2D_CLIP_Z 0x00000288 +#define NV50_2D_CLIP_W 0x0000028c +#define NV50_2D_ROP 0x000002a0 +#define NV50_2D_OPERATION 0x000002ac +#define NV50_2D_OPERATION_SRCCOPY_AND 0x00000000 +#define NV50_2D_OPERATION_ROP_AND 0x00000001 +#define NV50_2D_OPERATION_BLEND_AND 0x00000002 +#define NV50_2D_OPERATION_SRCCOPY 0x00000003 +#define NV50_2D_OPERATION_SRCCOPY_PREMULT 0x00000004 +#define NV50_2D_OPERATION_BLEND_PREMULT 0x00000005 +#define NV50_2D_PATTERN_FORMAT 0x000002e8 +#define NV50_2D_PATTERN_FORMAT_16BPP 0x00000000 +#define NV50_2D_PATTERN_FORMAT_15BPP 0x00000001 +#define NV50_2D_PATTERN_FORMAT_32BPP 0x00000002 +#define NV50_2D_PATTERN_FORMAT_8BPP 0x00000003 +#define NV50_2D_PATTERN_COLOR(x) (0x000002f0+((x)*4)) +#define NV50_2D_PATTERN_COLOR__SIZE 0x00000002 +#define NV50_2D_PATTERN_BITMAP(x) (0x000002f8+((x)*4)) +#define NV50_2D_PATTERN_BITMAP__SIZE 0x00000002 +#define NV50_2D_RECT_FORMAT 0x00000584 +#define NV50_2D_RECT_FORMAT_32BPP 0x000000cf +#define NV50_2D_RECT_FORMAT_24BPP 0x000000e6 +#define NV50_2D_RECT_FORMAT_16BPP 0x000000e8 +#define NV50_2D_RECT_FORMAT_8BPP 0x000000f3 +#define NV50_2D_RECT_FORMAT_15BPP 0x000000f8 +#define NV50_2D_RECT_COLOR 0x00000588 +#define NV50_2D_RECT_X1 0x00000600 +#define NV50_2D_RECT_Y1 0x00000604 +#define NV50_2D_RECT_X2 0x00000608 +#define NV50_2D_RECT_Y2 0x0000060c +#define NV50_2D_BLIT_DST_X 0x000008b0 +#define NV50_2D_BLIT_DST_Y 0x000008b4 +#define NV50_2D_BLIT_DST_W 0x000008b8 +#define NV50_2D_BLIT_DST_H 0x000008bc +#define NV50_2D_BLIT_SRC_X 0x000008d4 +#define NV50_2D_BLIT_SRC_Y 0x000008dc +#define NV50_2D_SIFC_UNK0800 0x00000800 +#define NV50_2D_SIFC_FORMAT 0x00000804 +#define NV50_2D_SIFC_FORMAT_32BPP 0x000000cf +#define NV50_2D_SIFC_FORMAT_24BPP 0x000000e6 +#define NV50_2D_SIFC_FORMAT_16BPP 0x000000e8 +#define NV50_2D_SIFC_FORMAT_8BPP 0x000000f3 +#define NV50_2D_SIFC_FORMAT_15BPP 0x000000f8 +#define NV50_2D_SIFC_WIDTH 0x00000838 +#define NV50_2D_SIFC_HEIGHT 0x0000083c +#define NV50_2D_SIFC_SCALE_UNK0840 0x00000840 +#define NV50_2D_SIFC_SCALE_UNK0844 0x00000844 +#define NV50_2D_SIFC_SCALE_UNK0848 0x00000848 +#define NV50_2D_SIFC_SCALE_UNK084C 0x0000084c +#define NV50_2D_SIFC_UNK0850 0x00000850 +#define NV50_2D_SIFC_DST_X 0x00000854 +#define NV50_2D_SIFC_UNK0858 0x00000858 +#define NV50_2D_SIFC_DST_Y 0x0000085c +#define NV50_2D_SIFC_DATA 0x00000860 + + +#define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 + +#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH 0x00000238 +#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT_HIGH 0x0000023c + + +#define NV50TCL 0x00005097 + +#define NV50TCL_NOP 0x00000100 +#define NV50TCL_NOTIFY 0x00000104 +#define NV50TCL_DMA_NOTIFY 0x00000180 +#define NV50TCL_DMA_IN_MEMORY0(x) (0x00000184+((x)*4)) +#define NV50TCL_DMA_IN_MEMORY0__SIZE 0x0000000b +#define NV50TCL_DMA_IN_MEMORY1(x) (0x000001c0+((x)*4)) +#define NV50TCL_DMA_IN_MEMORY1__SIZE 0x00000008 +#define NV50TCL_RT_ADDRESS_HIGH(x) (0x00000200+((x)*32)) +#define NV50TCL_RT_ADDRESS_HIGH__SIZE 0x00000008 +#define NV50TCL_RT_ADDRESS_LOW(x) (0x00000204+((x)*32)) +#define NV50TCL_RT_ADDRESS_LOW__SIZE 0x00000008 +#define NV50TCL_RT_FORMAT(x) (0x00000208+((x)*32)) +#define NV50TCL_RT_FORMAT__SIZE 0x00000008 +#define NV50TCL_RT_UNK3(x) (0x0000020c+((x)*32)) +#define NV50TCL_RT_UNK3__SIZE 0x00000008 +#define NV50TCL_RT_UNK4(x) (0x00000210+((x)*32)) +#define NV50TCL_RT_UNK4__SIZE 0x00000008 +#define NV50TCL_VTX_ATTR_1F(x) (0x00000300+((x)*4)) +#define NV50TCL_VTX_ATTR_1F__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2F_X(x) (0x00000380+((x)*8)) +#define NV50TCL_VTX_ATTR_2F_X__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2F_Y(x) (0x00000384+((x)*8)) +#define NV50TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_X(x) (0x00000400+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_X__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_Y(x) (0x00000404+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_Z(x) (0x00000408+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_3F_W(x) (0x0000040c+((x)*16)) +#define NV50TCL_VTX_ATTR_3F_W__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_X(x) (0x00000500+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_X__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_Y(x) (0x00000504+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_Z(x) (0x00000508+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4F_W(x) (0x0000050c+((x)*16)) +#define NV50TCL_VTX_ATTR_4F_W__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2I(x) (0x00000680+((x)*4)) +#define NV50TCL_VTX_ATTR_2I__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_2I_X_SHIFT 0 +#define NV50TCL_VTX_ATTR_2I_X_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV50TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4I_0(x) (0x00000700+((x)*8)) +#define NV50TCL_VTX_ATTR_4I_0__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4I_0_X_SHIFT 0 +#define NV50TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4I_0_Y_SHIFT 16 +#define NV50TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4I_1(x) (0x00000704+((x)*8)) +#define NV50TCL_VTX_ATTR_4I_1__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4I_1_Z_SHIFT 0 +#define NV50TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4I_1_W_SHIFT 16 +#define NV50TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4NI_0(x) (0x00000780+((x)*8)) +#define NV50TCL_VTX_ATTR_4NI_0__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4NI_0_X_SHIFT 0 +#define NV50TCL_VTX_ATTR_4NI_0_X_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4NI_0_Y_SHIFT 16 +#define NV50TCL_VTX_ATTR_4NI_0_Y_MASK 0xffff0000 +#define NV50TCL_VTX_ATTR_4NI_1(x) (0x00000784+((x)*8)) +#define NV50TCL_VTX_ATTR_4NI_1__SIZE 0x00000010 +#define NV50TCL_VTX_ATTR_4NI_1_Z_SHIFT 0 +#define NV50TCL_VTX_ATTR_4NI_1_Z_MASK 0x0000ffff +#define NV50TCL_VTX_ATTR_4NI_1_W_SHIFT 16 +#define NV50TCL_VTX_ATTR_4NI_1_W_MASK 0xffff0000 +#define NV50TCL_VERTEX_ARRAY_FORMAT(x) (0x00000900+((x)*16)) +#define NV50TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 +#define NV50TCL_VIEWPORT_UNK0(x) (0x00000a00+((x)*4)) +#define NV50TCL_VIEWPORT_UNK0__SIZE 0x00000003 +#define NV50TCL_VIEWPORT_UNK1(x) (0x00000a0c+((x)*4)) +#define NV50TCL_VIEWPORT_UNK1__SIZE 0x00000003 +#define NV50TCL_VIEWPORT_HORIZ 0x00000c00 +#define NV50TCL_VIEWPORT_HORIZ_X_SHIFT 0 +#define NV50TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff +#define NV50TCL_VIEWPORT_HORIZ_W_SHIFT 16 +#define NV50TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 +#define NV50TCL_VIEWPORT_VERT 0x00000c04 +#define NV50TCL_VIEWPORT_VERT_Y_SHIFT 0 +#define NV50TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff +#define NV50TCL_VIEWPORT_VERT_H_SHIFT 16 +#define NV50TCL_VIEWPORT_VERT_H_MASK 0xffff0000 +#define NV50TCL_DEPTH_RANGE_NEAR 0x00000c08 +#define NV50TCL_DEPTH_RANGE_FAR 0x00000c0c +#define NV50TCL_VIEWPORT_CLIP_HORIZ(x) (0x00000d00+((x)*8)) +#define NV50TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 +#define NV50TCL_VIEWPORT_CLIP_VERT(x) (0x00000d04+((x)*8)) +#define NV50TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 +#define NV50TCL_VERTEX_BUFFER_FIRST 0x00000d74 +#define NV50TCL_VERTEX_BUFFER_COUNT 0x00000d78 +#define NV50TCL_CLEAR_COLOR(x) (0x00000d80+((x)*4)) +#define NV50TCL_CLEAR_COLOR__SIZE 0x00000004 +#define NV50TCL_CLEAR_DEPTH 0x00000d90 +#define NV50TCL_CLEAR_STENCIL 0x00000da0 +#define NV50TCL_POLYGON_MODE_FRONT 0x00000dac +#define NV50TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 +#define NV50TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 +#define NV50TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 +#define NV50TCL_POLYGON_MODE_BACK 0x00000db0 +#define NV50TCL_POLYGON_MODE_BACK_POINT 0x00001b00 +#define NV50TCL_POLYGON_MODE_BACK_LINE 0x00001b01 +#define NV50TCL_POLYGON_MODE_BACK_FILL 0x00001b02 +#define NV50TCL_POLYGON_SMOOTH_ENABLE 0x00000db4 +#define NV50TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000dc0 +#define NV50TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000dc4 +#define NV50TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000dc8 +#define NV50TCL_SCISSOR_HORIZ 0x00000e04 +#define NV50TCL_SCISSOR_HORIZ_L_SHIFT 0 +#define NV50TCL_SCISSOR_HORIZ_L_MASK 0x0000ffff +#define NV50TCL_SCISSOR_HORIZ_R_SHIFT 16 +#define NV50TCL_SCISSOR_HORIZ_R_MASK 0xffff0000 +#define NV50TCL_SCISSOR_VERT 0x00000e08 +#define NV50TCL_SCISSOR_VERT_T_SHIFT 0 +#define NV50TCL_SCISSOR_VERT_T_MASK 0x0000ffff +#define NV50TCL_SCISSOR_VERT_B_SHIFT 16 +#define NV50TCL_SCISSOR_VERT_B_MASK 0xffff0000 +#define NV50TCL_VP_UPLOAD_CONST_ID 0x00000f00 +#define NV50TCL_VP_UPLOAD_CONST(x) (0x00000f04+((x)*4)) +#define NV50TCL_VP_UPLOAD_CONST__SIZE 0x00000010 +#define NV50TCL_STENCIL_FRONT_FUNC_REF 0x00000f54 +#define NV50TCL_STENCIL_FRONT_MASK 0x00000f58 +#define NV50TCL_STENCIL_FRONT_FUNC_MASK 0x00000f5c +#define NV50TCL_GP_ADDRESS_HIGH 0x00000f70 +#define NV50TCL_GP_ADDRESS_LOW 0x00000f74 +#define NV50TCL_VP_ADDRESS_HIGH 0x00000f7c +#define NV50TCL_VP_ADDRESS_LOW 0x00000f80 +#define NV50TCL_FP_ADDRESS_HIGH 0x00000fa4 +#define NV50TCL_FP_ADDRESS_LOW 0x00000fa8 +#define NV50TCL_ZETA_ADDRESS_HIGH 0x00000fe0 +#define NV50TCL_ZETA_ADDRESS_LOW 0x00000fe4 +#define NV50TCL_UNKFF4 0x00000ff4 +#define NV50TCL_UNKFF4_W_SHIFT 16 +#define NV50TCL_UNKFF4_W_MASK 0xffff0000 +#define NV50TCL_UNKFF8 0x00000ff8 +#define NV50TCL_UNKFF8_H_SHIFT 16 +#define NV50TCL_UNKFF8_H_MASK 0xffff0000 +#define NV50TCL_RT_HORIZ(x) (0x00001240+((x)*8)) +#define NV50TCL_RT_HORIZ__SIZE 0x00000008 +#define NV50TCL_RT_VERT(x) (0x00001244+((x)*8)) +#define NV50TCL_RT_VERT__SIZE 0x00000008 +#define NV50TCL_DEPTH_TEST_ENABLE 0x000012cc +#define NV50TCL_SHADE_MODEL 0x000012d4 +#define NV50TCL_SHADE_MODEL_FLAT 0x00001d00 +#define NV50TCL_SHADE_MODEL_SMOOTH 0x00001d01 +#define NV50TCL_DEPTH_WRITE_ENABLE 0x000012e8 +#define NV50TCL_ALPHA_TEST_ENABLE 0x000012ec +#define NV50TCL_DEPTH_TEST_FUNC 0x0000130c +#define NV50TCL_DEPTH_TEST_FUNC_NEVER 0x00000200 +#define NV50TCL_DEPTH_TEST_FUNC_LESS 0x00000201 +#define NV50TCL_DEPTH_TEST_FUNC_EQUAL 0x00000202 +#define NV50TCL_DEPTH_TEST_FUNC_LEQUAL 0x00000203 +#define NV50TCL_DEPTH_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_DEPTH_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_DEPTH_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_DEPTH_TEST_FUNC_GEQUAL 0x00000206 +#define NV50TCL_DEPTH_TEST_FUNC_ALWAYS 0x00000207 +#define NV50TCL_ALPHA_TEST_REF 0x00001310 +#define NV50TCL_ALPHA_TEST_FUNC 0x00001314 +#define NV50TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 +#define NV50TCL_ALPHA_TEST_FUNC_LESS 0x00000201 +#define NV50TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 +#define NV50TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 +#define NV50TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 +#define NV50TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 +#define NV50TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 +#define NV50TCL_BLEND_COLOR(x) (0x0000131c+((x)*4)) +#define NV50TCL_BLEND_COLOR__SIZE 0x00000004 +#define NV50TCL_BLEND_EQUATION_RGB 0x00001340 +#define NV50TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 +#define NV50TCL_BLEND_EQUATION_RGB_MIN 0x00008007 +#define NV50TCL_BLEND_EQUATION_RGB_MAX 0x00008008 +#define NV50TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a +#define NV50TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV50TCL_BLEND_FUNC_SRC_RGB 0x00001344 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_FUNC_DST_RGB 0x00001348 +#define NV50TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_EQUATION_ALPHA 0x0000134c +#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 +#define NV50TCL_BLEND_EQUATION_ALPHA_MIN 0x00008007 +#define NV50TCL_BLEND_EQUATION_ALPHA_MAX 0x00008008 +#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a +#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV50TCL_BLEND_FUNC_SRC_ALPHA 0x00001350 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_FUNC_DST_ALPHA 0x00001358 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 +#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 +#define NV50TCL_BLEND_ENABLE(x) (0x00001360+((x)*4)) +#define NV50TCL_BLEND_ENABLE__SIZE 0x00000008 +#define NV50TCL_STENCIL_BACK_ENABLE 0x00001380 +#define NV50TCL_STENCIL_BACK_OP_FAIL 0x00001384 +#define NV50TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL 0x00001388 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_BACK_OP_ZPASS 0x0000138c +#define NV50TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a +#define NV50TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC 0x00001390 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50TCL_STENCIL_BACK_FUNC_REF 0x00001394 +#define NV50TCL_STENCIL_BACK_MASK 0x00001398 +#define NV50TCL_STENCIL_BACK_FUNC_MASK 0x0000139c +#define NV50TCL_LINE_WIDTH 0x000013b0 +#define NV50TCL_VP_START_ID 0x0000140c +#define NV50TCL_GP_START_ID 0x00001410 +#define NV50TCL_FP_START_ID 0x00001414 +#define NV50TCL_POINT_SIZE 0x00001518 +#define NV50TCL_TEX_CB0_ADDRESS_HIGH 0x0000155c +#define NV50TCL_TEX_CB0_ADDRESS_LOW 0x00001560 +#define NV50TCL_POLYGON_OFFSET_FACTOR 0x0000156c +#define NV50TCL_LINE_SMOOTH_ENABLE 0x00001570 +#define NV50TCL_TEX_CB1_ADDRESS_HIGH 0x00001574 +#define NV50TCL_TEX_CB1_ADDRESS_LOW 0x00001578 +#define NV50TCL_STENCIL_FRONT_ENABLE 0x00001594 +#define NV50TCL_STENCIL_FRONT_OP_FAIL 0x00001598 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL 0x0000159c +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS 0x000015a0 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 +#define NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC 0x000015a4 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 +#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 +#define NV50TCL_POLYGON_OFFSET_UNITS 0x000015bc +#define NV50TCL_VERTEX_BEGIN 0x000015dc +#define NV50TCL_VERTEX_BEGIN_POINTS 0x00000000 +#define NV50TCL_VERTEX_BEGIN_LINES 0x00000001 +#define NV50TCL_VERTEX_BEGIN_LINE_LOOP 0x00000002 +#define NV50TCL_VERTEX_BEGIN_LINE_STRIP 0x00000003 +#define NV50TCL_VERTEX_BEGIN_TRIANGLES 0x00000004 +#define NV50TCL_VERTEX_BEGIN_TRIANGLE_STRIP 0x00000005 +#define NV50TCL_VERTEX_BEGIN_TRIANGLE_FAN 0x00000006 +#define NV50TCL_VERTEX_BEGIN_QUADS 0x00000007 +#define NV50TCL_VERTEX_BEGIN_QUAD_STRIP 0x00000008 +#define NV50TCL_VERTEX_BEGIN_POLYGON 0x00000009 +#define NV50TCL_VERTEX_END 0x000015e0 +#define NV50TCL_VERTEX_DATA 0x00001640 +#define NV50TCL_VP_ATTR_EN_0 0x00001650 +#define NV50TCL_VP_ATTR_EN_0_7_SHIFT 28 +#define NV50TCL_VP_ATTR_EN_0_7_MASK 0xf0000000 +#define NV50TCL_VP_ATTR_EN_0_7_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNNN 0x10000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYNN 0x20000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYNN 0x30000000 +#define NV50TCL_VP_ATTR_EN_0_7_NNZN 0x40000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNZN 0x50000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYZN 0x60000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYZN 0x70000000 +#define NV50TCL_VP_ATTR_EN_0_7_NNNW 0x80000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNNW 0x90000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYNW 0xa0000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYNW 0xb0000000 +#define NV50TCL_VP_ATTR_EN_0_7_NNZW 0xc0000000 +#define NV50TCL_VP_ATTR_EN_0_7_XNZW 0xd0000000 +#define NV50TCL_VP_ATTR_EN_0_7_NYZW 0xe0000000 +#define NV50TCL_VP_ATTR_EN_0_7_XYZW 0xf0000000 +#define NV50TCL_VP_ATTR_EN_0_6_SHIFT 24 +#define NV50TCL_VP_ATTR_EN_0_6_MASK 0x0f000000 +#define NV50TCL_VP_ATTR_EN_0_6_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNNN 0x01000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYNN 0x02000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYNN 0x03000000 +#define NV50TCL_VP_ATTR_EN_0_6_NNZN 0x04000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNZN 0x05000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYZN 0x06000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYZN 0x07000000 +#define NV50TCL_VP_ATTR_EN_0_6_NNNW 0x08000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNNW 0x09000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYNW 0x0a000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYNW 0x0b000000 +#define NV50TCL_VP_ATTR_EN_0_6_NNZW 0x0c000000 +#define NV50TCL_VP_ATTR_EN_0_6_XNZW 0x0d000000 +#define NV50TCL_VP_ATTR_EN_0_6_NYZW 0x0e000000 +#define NV50TCL_VP_ATTR_EN_0_6_XYZW 0x0f000000 +#define NV50TCL_VP_ATTR_EN_0_5_SHIFT 20 +#define NV50TCL_VP_ATTR_EN_0_5_MASK 0x00f00000 +#define NV50TCL_VP_ATTR_EN_0_5_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_5_XNNN 0x00100000 +#define NV50TCL_VP_ATTR_EN_0_5_NYNN 0x00200000 +#define NV50TCL_VP_ATTR_EN_0_5_XYNN 0x00300000 +#define NV50TCL_VP_ATTR_EN_0_5_NNZN 0x00400000 +#define NV50TCL_VP_ATTR_EN_0_5_XNZN 0x00500000 +#define NV50TCL_VP_ATTR_EN_0_5_NYZN 0x00600000 +#define NV50TCL_VP_ATTR_EN_0_5_XYZN 0x00700000 +#define NV50TCL_VP_ATTR_EN_0_5_NNNW 0x00800000 +#define NV50TCL_VP_ATTR_EN_0_5_XNNW 0x00900000 +#define NV50TCL_VP_ATTR_EN_0_5_NYNW 0x00a00000 +#define NV50TCL_VP_ATTR_EN_0_5_XYNW 0x00b00000 +#define NV50TCL_VP_ATTR_EN_0_5_NNZW 0x00c00000 +#define NV50TCL_VP_ATTR_EN_0_5_XNZW 0x00d00000 +#define NV50TCL_VP_ATTR_EN_0_5_NYZW 0x00e00000 +#define NV50TCL_VP_ATTR_EN_0_5_XYZW 0x00f00000 +#define NV50TCL_VP_ATTR_EN_0_4_SHIFT 16 +#define NV50TCL_VP_ATTR_EN_0_4_MASK 0x000f0000 +#define NV50TCL_VP_ATTR_EN_0_4_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_4_XNNN 0x00010000 +#define NV50TCL_VP_ATTR_EN_0_4_NYNN 0x00020000 +#define NV50TCL_VP_ATTR_EN_0_4_XYNN 0x00030000 +#define NV50TCL_VP_ATTR_EN_0_4_NNZN 0x00040000 +#define NV50TCL_VP_ATTR_EN_0_4_XNZN 0x00050000 +#define NV50TCL_VP_ATTR_EN_0_4_NYZN 0x00060000 +#define NV50TCL_VP_ATTR_EN_0_4_XYZN 0x00070000 +#define NV50TCL_VP_ATTR_EN_0_4_NNNW 0x00080000 +#define NV50TCL_VP_ATTR_EN_0_4_XNNW 0x00090000 +#define NV50TCL_VP_ATTR_EN_0_4_NYNW 0x000a0000 +#define NV50TCL_VP_ATTR_EN_0_4_XYNW 0x000b0000 +#define NV50TCL_VP_ATTR_EN_0_4_NNZW 0x000c0000 +#define NV50TCL_VP_ATTR_EN_0_4_XNZW 0x000d0000 +#define NV50TCL_VP_ATTR_EN_0_4_NYZW 0x000e0000 +#define NV50TCL_VP_ATTR_EN_0_4_XYZW 0x000f0000 +#define NV50TCL_VP_ATTR_EN_0_3_SHIFT 12 +#define NV50TCL_VP_ATTR_EN_0_3_MASK 0x0000f000 +#define NV50TCL_VP_ATTR_EN_0_3_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_3_XNNN 0x00001000 +#define NV50TCL_VP_ATTR_EN_0_3_NYNN 0x00002000 +#define NV50TCL_VP_ATTR_EN_0_3_XYNN 0x00003000 +#define NV50TCL_VP_ATTR_EN_0_3_NNZN 0x00004000 +#define NV50TCL_VP_ATTR_EN_0_3_XNZN 0x00005000 +#define NV50TCL_VP_ATTR_EN_0_3_NYZN 0x00006000 +#define NV50TCL_VP_ATTR_EN_0_3_XYZN 0x00007000 +#define NV50TCL_VP_ATTR_EN_0_3_NNNW 0x00008000 +#define NV50TCL_VP_ATTR_EN_0_3_XNNW 0x00009000 +#define NV50TCL_VP_ATTR_EN_0_3_NYNW 0x0000a000 +#define NV50TCL_VP_ATTR_EN_0_3_XYNW 0x0000b000 +#define NV50TCL_VP_ATTR_EN_0_3_NNZW 0x0000c000 +#define NV50TCL_VP_ATTR_EN_0_3_XNZW 0x0000d000 +#define NV50TCL_VP_ATTR_EN_0_3_NYZW 0x0000e000 +#define NV50TCL_VP_ATTR_EN_0_3_XYZW 0x0000f000 +#define NV50TCL_VP_ATTR_EN_0_2_SHIFT 8 +#define NV50TCL_VP_ATTR_EN_0_2_MASK 0x00000f00 +#define NV50TCL_VP_ATTR_EN_0_2_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_2_XNNN 0x00000100 +#define NV50TCL_VP_ATTR_EN_0_2_NYNN 0x00000200 +#define NV50TCL_VP_ATTR_EN_0_2_XYNN 0x00000300 +#define NV50TCL_VP_ATTR_EN_0_2_NNZN 0x00000400 +#define NV50TCL_VP_ATTR_EN_0_2_XNZN 0x00000500 +#define NV50TCL_VP_ATTR_EN_0_2_NYZN 0x00000600 +#define NV50TCL_VP_ATTR_EN_0_2_XYZN 0x00000700 +#define NV50TCL_VP_ATTR_EN_0_2_NNNW 0x00000800 +#define NV50TCL_VP_ATTR_EN_0_2_XNNW 0x00000900 +#define NV50TCL_VP_ATTR_EN_0_2_NYNW 0x00000a00 +#define NV50TCL_VP_ATTR_EN_0_2_XYNW 0x00000b00 +#define NV50TCL_VP_ATTR_EN_0_2_NNZW 0x00000c00 +#define NV50TCL_VP_ATTR_EN_0_2_XNZW 0x00000d00 +#define NV50TCL_VP_ATTR_EN_0_2_NYZW 0x00000e00 +#define NV50TCL_VP_ATTR_EN_0_2_XYZW 0x00000f00 +#define NV50TCL_VP_ATTR_EN_0_1_SHIFT 4 +#define NV50TCL_VP_ATTR_EN_0_1_MASK 0x000000f0 +#define NV50TCL_VP_ATTR_EN_0_1_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_1_XNNN 0x00000010 +#define NV50TCL_VP_ATTR_EN_0_1_NYNN 0x00000020 +#define NV50TCL_VP_ATTR_EN_0_1_XYNN 0x00000030 +#define NV50TCL_VP_ATTR_EN_0_1_NNZN 0x00000040 +#define NV50TCL_VP_ATTR_EN_0_1_XNZN 0x00000050 +#define NV50TCL_VP_ATTR_EN_0_1_NYZN 0x00000060 +#define NV50TCL_VP_ATTR_EN_0_1_XYZN 0x00000070 +#define NV50TCL_VP_ATTR_EN_0_1_NNNW 0x00000080 +#define NV50TCL_VP_ATTR_EN_0_1_XNNW 0x00000090 +#define NV50TCL_VP_ATTR_EN_0_1_NYNW 0x000000a0 +#define NV50TCL_VP_ATTR_EN_0_1_XYNW 0x000000b0 +#define NV50TCL_VP_ATTR_EN_0_1_NNZW 0x000000c0 +#define NV50TCL_VP_ATTR_EN_0_1_XNZW 0x000000d0 +#define NV50TCL_VP_ATTR_EN_0_1_NYZW 0x000000e0 +#define NV50TCL_VP_ATTR_EN_0_1_XYZW 0x000000f0 +#define NV50TCL_VP_ATTR_EN_0_0_SHIFT 0 +#define NV50TCL_VP_ATTR_EN_0_0_MASK 0x0000000f +#define NV50TCL_VP_ATTR_EN_0_0_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_0_0_XNNN 0x00000001 +#define NV50TCL_VP_ATTR_EN_0_0_NYNN 0x00000002 +#define NV50TCL_VP_ATTR_EN_0_0_XYNN 0x00000003 +#define NV50TCL_VP_ATTR_EN_0_0_NNZN 0x00000004 +#define NV50TCL_VP_ATTR_EN_0_0_XNZN 0x00000005 +#define NV50TCL_VP_ATTR_EN_0_0_NYZN 0x00000006 +#define NV50TCL_VP_ATTR_EN_0_0_XYZN 0x00000007 +#define NV50TCL_VP_ATTR_EN_0_0_NNNW 0x00000008 +#define NV50TCL_VP_ATTR_EN_0_0_XNNW 0x00000009 +#define NV50TCL_VP_ATTR_EN_0_0_NYNW 0x0000000a +#define NV50TCL_VP_ATTR_EN_0_0_XYNW 0x0000000b +#define NV50TCL_VP_ATTR_EN_0_0_NNZW 0x0000000c +#define NV50TCL_VP_ATTR_EN_0_0_XNZW 0x0000000d +#define NV50TCL_VP_ATTR_EN_0_0_NYZW 0x0000000e +#define NV50TCL_VP_ATTR_EN_0_0_XYZW 0x0000000f +#define NV50TCL_VP_ATTR_EN_1 0x00001654 +#define NV50TCL_VP_ATTR_EN_1_15_SHIFT 28 +#define NV50TCL_VP_ATTR_EN_1_15_MASK 0xf0000000 +#define NV50TCL_VP_ATTR_EN_1_15_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNNN 0x10000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYNN 0x20000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYNN 0x30000000 +#define NV50TCL_VP_ATTR_EN_1_15_NNZN 0x40000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNZN 0x50000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYZN 0x60000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYZN 0x70000000 +#define NV50TCL_VP_ATTR_EN_1_15_NNNW 0x80000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNNW 0x90000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYNW 0xa0000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYNW 0xb0000000 +#define NV50TCL_VP_ATTR_EN_1_15_NNZW 0xc0000000 +#define NV50TCL_VP_ATTR_EN_1_15_XNZW 0xd0000000 +#define NV50TCL_VP_ATTR_EN_1_15_NYZW 0xe0000000 +#define NV50TCL_VP_ATTR_EN_1_15_XYZW 0xf0000000 +#define NV50TCL_VP_ATTR_EN_1_14_SHIFT 24 +#define NV50TCL_VP_ATTR_EN_1_14_MASK 0x0f000000 +#define NV50TCL_VP_ATTR_EN_1_14_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNNN 0x01000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYNN 0x02000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYNN 0x03000000 +#define NV50TCL_VP_ATTR_EN_1_14_NNZN 0x04000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNZN 0x05000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYZN 0x06000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYZN 0x07000000 +#define NV50TCL_VP_ATTR_EN_1_14_NNNW 0x08000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNNW 0x09000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYNW 0x0a000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYNW 0x0b000000 +#define NV50TCL_VP_ATTR_EN_1_14_NNZW 0x0c000000 +#define NV50TCL_VP_ATTR_EN_1_14_XNZW 0x0d000000 +#define NV50TCL_VP_ATTR_EN_1_14_NYZW 0x0e000000 +#define NV50TCL_VP_ATTR_EN_1_14_XYZW 0x0f000000 +#define NV50TCL_VP_ATTR_EN_1_13_SHIFT 20 +#define NV50TCL_VP_ATTR_EN_1_13_MASK 0x00f00000 +#define NV50TCL_VP_ATTR_EN_1_13_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_13_XNNN 0x00100000 +#define NV50TCL_VP_ATTR_EN_1_13_NYNN 0x00200000 +#define NV50TCL_VP_ATTR_EN_1_13_XYNN 0x00300000 +#define NV50TCL_VP_ATTR_EN_1_13_NNZN 0x00400000 +#define NV50TCL_VP_ATTR_EN_1_13_XNZN 0x00500000 +#define NV50TCL_VP_ATTR_EN_1_13_NYZN 0x00600000 +#define NV50TCL_VP_ATTR_EN_1_13_XYZN 0x00700000 +#define NV50TCL_VP_ATTR_EN_1_13_NNNW 0x00800000 +#define NV50TCL_VP_ATTR_EN_1_13_XNNW 0x00900000 +#define NV50TCL_VP_ATTR_EN_1_13_NYNW 0x00a00000 +#define NV50TCL_VP_ATTR_EN_1_13_XYNW 0x00b00000 +#define NV50TCL_VP_ATTR_EN_1_13_NNZW 0x00c00000 +#define NV50TCL_VP_ATTR_EN_1_13_XNZW 0x00d00000 +#define NV50TCL_VP_ATTR_EN_1_13_NYZW 0x00e00000 +#define NV50TCL_VP_ATTR_EN_1_13_XYZW 0x00f00000 +#define NV50TCL_VP_ATTR_EN_1_12_SHIFT 16 +#define NV50TCL_VP_ATTR_EN_1_12_MASK 0x000f0000 +#define NV50TCL_VP_ATTR_EN_1_12_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_12_XNNN 0x00010000 +#define NV50TCL_VP_ATTR_EN_1_12_NYNN 0x00020000 +#define NV50TCL_VP_ATTR_EN_1_12_XYNN 0x00030000 +#define NV50TCL_VP_ATTR_EN_1_12_NNZN 0x00040000 +#define NV50TCL_VP_ATTR_EN_1_12_XNZN 0x00050000 +#define NV50TCL_VP_ATTR_EN_1_12_NYZN 0x00060000 +#define NV50TCL_VP_ATTR_EN_1_12_XYZN 0x00070000 +#define NV50TCL_VP_ATTR_EN_1_12_NNNW 0x00080000 +#define NV50TCL_VP_ATTR_EN_1_12_XNNW 0x00090000 +#define NV50TCL_VP_ATTR_EN_1_12_NYNW 0x000a0000 +#define NV50TCL_VP_ATTR_EN_1_12_XYNW 0x000b0000 +#define NV50TCL_VP_ATTR_EN_1_12_NNZW 0x000c0000 +#define NV50TCL_VP_ATTR_EN_1_12_XNZW 0x000d0000 +#define NV50TCL_VP_ATTR_EN_1_12_NYZW 0x000e0000 +#define NV50TCL_VP_ATTR_EN_1_12_XYZW 0x000f0000 +#define NV50TCL_VP_ATTR_EN_1_11_SHIFT 12 +#define NV50TCL_VP_ATTR_EN_1_11_MASK 0x0000f000 +#define NV50TCL_VP_ATTR_EN_1_11_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_11_XNNN 0x00001000 +#define NV50TCL_VP_ATTR_EN_1_11_NYNN 0x00002000 +#define NV50TCL_VP_ATTR_EN_1_11_XYNN 0x00003000 +#define NV50TCL_VP_ATTR_EN_1_11_NNZN 0x00004000 +#define NV50TCL_VP_ATTR_EN_1_11_XNZN 0x00005000 +#define NV50TCL_VP_ATTR_EN_1_11_NYZN 0x00006000 +#define NV50TCL_VP_ATTR_EN_1_11_XYZN 0x00007000 +#define NV50TCL_VP_ATTR_EN_1_11_NNNW 0x00008000 +#define NV50TCL_VP_ATTR_EN_1_11_XNNW 0x00009000 +#define NV50TCL_VP_ATTR_EN_1_11_NYNW 0x0000a000 +#define NV50TCL_VP_ATTR_EN_1_11_XYNW 0x0000b000 +#define NV50TCL_VP_ATTR_EN_1_11_NNZW 0x0000c000 +#define NV50TCL_VP_ATTR_EN_1_11_XNZW 0x0000d000 +#define NV50TCL_VP_ATTR_EN_1_11_NYZW 0x0000e000 +#define NV50TCL_VP_ATTR_EN_1_11_XYZW 0x0000f000 +#define NV50TCL_VP_ATTR_EN_1_10_SHIFT 8 +#define NV50TCL_VP_ATTR_EN_1_10_MASK 0x00000f00 +#define NV50TCL_VP_ATTR_EN_1_10_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_10_XNNN 0x00000100 +#define NV50TCL_VP_ATTR_EN_1_10_NYNN 0x00000200 +#define NV50TCL_VP_ATTR_EN_1_10_XYNN 0x00000300 +#define NV50TCL_VP_ATTR_EN_1_10_NNZN 0x00000400 +#define NV50TCL_VP_ATTR_EN_1_10_XNZN 0x00000500 +#define NV50TCL_VP_ATTR_EN_1_10_NYZN 0x00000600 +#define NV50TCL_VP_ATTR_EN_1_10_XYZN 0x00000700 +#define NV50TCL_VP_ATTR_EN_1_10_NNNW 0x00000800 +#define NV50TCL_VP_ATTR_EN_1_10_XNNW 0x00000900 +#define NV50TCL_VP_ATTR_EN_1_10_NYNW 0x00000a00 +#define NV50TCL_VP_ATTR_EN_1_10_XYNW 0x00000b00 +#define NV50TCL_VP_ATTR_EN_1_10_NNZW 0x00000c00 +#define NV50TCL_VP_ATTR_EN_1_10_XNZW 0x00000d00 +#define NV50TCL_VP_ATTR_EN_1_10_NYZW 0x00000e00 +#define NV50TCL_VP_ATTR_EN_1_10_XYZW 0x00000f00 +#define NV50TCL_VP_ATTR_EN_1_9_SHIFT 4 +#define NV50TCL_VP_ATTR_EN_1_9_MASK 0x000000f0 +#define NV50TCL_VP_ATTR_EN_1_9_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_9_XNNN 0x00000010 +#define NV50TCL_VP_ATTR_EN_1_9_NYNN 0x00000020 +#define NV50TCL_VP_ATTR_EN_1_9_XYNN 0x00000030 +#define NV50TCL_VP_ATTR_EN_1_9_NNZN 0x00000040 +#define NV50TCL_VP_ATTR_EN_1_9_XNZN 0x00000050 +#define NV50TCL_VP_ATTR_EN_1_9_NYZN 0x00000060 +#define NV50TCL_VP_ATTR_EN_1_9_XYZN 0x00000070 +#define NV50TCL_VP_ATTR_EN_1_9_NNNW 0x00000080 +#define NV50TCL_VP_ATTR_EN_1_9_XNNW 0x00000090 +#define NV50TCL_VP_ATTR_EN_1_9_NYNW 0x000000a0 +#define NV50TCL_VP_ATTR_EN_1_9_XYNW 0x000000b0 +#define NV50TCL_VP_ATTR_EN_1_9_NNZW 0x000000c0 +#define NV50TCL_VP_ATTR_EN_1_9_XNZW 0x000000d0 +#define NV50TCL_VP_ATTR_EN_1_9_NYZW 0x000000e0 +#define NV50TCL_VP_ATTR_EN_1_9_XYZW 0x000000f0 +#define NV50TCL_VP_ATTR_EN_1_8_SHIFT 0 +#define NV50TCL_VP_ATTR_EN_1_8_MASK 0x0000000f +#define NV50TCL_VP_ATTR_EN_1_8_NONE 0x00000000 +#define NV50TCL_VP_ATTR_EN_1_8_XNNN 0x00000001 +#define NV50TCL_VP_ATTR_EN_1_8_NYNN 0x00000002 +#define NV50TCL_VP_ATTR_EN_1_8_XYNN 0x00000003 +#define NV50TCL_VP_ATTR_EN_1_8_NNZN 0x00000004 +#define NV50TCL_VP_ATTR_EN_1_8_XNZN 0x00000005 +#define NV50TCL_VP_ATTR_EN_1_8_NYZN 0x00000006 +#define NV50TCL_VP_ATTR_EN_1_8_XYZN 0x00000007 +#define NV50TCL_VP_ATTR_EN_1_8_NNNW 0x00000008 +#define NV50TCL_VP_ATTR_EN_1_8_XNNW 0x00000009 +#define NV50TCL_VP_ATTR_EN_1_8_NYNW 0x0000000a +#define NV50TCL_VP_ATTR_EN_1_8_XYNW 0x0000000b +#define NV50TCL_VP_ATTR_EN_1_8_NNZW 0x0000000c +#define NV50TCL_VP_ATTR_EN_1_8_XNZW 0x0000000d +#define NV50TCL_VP_ATTR_EN_1_8_NYZW 0x0000000e +#define NV50TCL_VP_ATTR_EN_1_8_XYZW 0x0000000f +#define NV50TCL_LINE_STIPPLE_ENABLE 0x0000166c +#define NV50TCL_LINE_STIPPLE_PATTERN 0x00001680 +#define NV50TCL_POLYGON_STIPPLE_ENABLE 0x0000168c +#define NV50TCL_VP_REG_HPOS 0x000016bc +#define NV50TCL_VP_REG_HPOS_X_SHIFT 0 +#define NV50TCL_VP_REG_HPOS_X_MASK 0x000000ff +#define NV50TCL_VP_REG_HPOS_Y_SHIFT 8 +#define NV50TCL_VP_REG_HPOS_Y_MASK 0x0000ff00 +#define NV50TCL_VP_REG_HPOS_Z_SHIFT 16 +#define NV50TCL_VP_REG_HPOS_Z_MASK 0x00ff0000 +#define NV50TCL_VP_REG_HPOS_W_SHIFT 24 +#define NV50TCL_VP_REG_HPOS_W_MASK 0xff000000 +#define NV50TCL_VP_REG_COL0 0x000016c0 +#define NV50TCL_VP_REG_COL0_X_SHIFT 0 +#define NV50TCL_VP_REG_COL0_X_MASK 0x000000ff +#define NV50TCL_VP_REG_COL0_Y_SHIFT 8 +#define NV50TCL_VP_REG_COL0_Y_MASK 0x0000ff00 +#define NV50TCL_VP_REG_COL0_Z_SHIFT 16 +#define NV50TCL_VP_REG_COL0_Z_MASK 0x00ff0000 +#define NV50TCL_VP_REG_COL0_W_SHIFT 24 +#define NV50TCL_VP_REG_COL0_W_MASK 0xff000000 +#define NV50TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001700+((x)*4)) +#define NV50TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 +#define NV50TCL_CULL_FACE_ENABLE 0x00001918 +#define NV50TCL_FRONT_FACE 0x0000191c +#define NV50TCL_FRONT_FACE_CW 0x00000900 +#define NV50TCL_FRONT_FACE_CCW 0x00000901 +#define NV50TCL_CULL_FACE 0x00001920 +#define NV50TCL_CULL_FACE_FRONT 0x00000404 +#define NV50TCL_CULL_FACE_BACK 0x00000405 +#define NV50TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 +#define NV50TCL_LOGIC_OP_ENABLE 0x000019c4 +#define NV50TCL_LOGIC_OP 0x000019c8 +#define NV50TCL_LOGIC_OP_CLEAR 0x00001500 +#define NV50TCL_LOGIC_OP_AND 0x00001501 +#define NV50TCL_LOGIC_OP_AND_REVERSE 0x00001502 +#define NV50TCL_LOGIC_OP_COPY 0x00001503 +#define NV50TCL_LOGIC_OP_AND_INVERTED 0x00001504 +#define NV50TCL_LOGIC_OP_NOOP 0x00001505 +#define NV50TCL_LOGIC_OP_XOR 0x00001506 +#define NV50TCL_LOGIC_OP_OR 0x00001507 +#define NV50TCL_LOGIC_OP_NOR 0x00001508 +#define NV50TCL_LOGIC_OP_EQUIV 0x00001509 +#define NV50TCL_LOGIC_OP_INVERT 0x0000150a +#define NV50TCL_LOGIC_OP_OR_REVERSE 0x0000150b +#define NV50TCL_LOGIC_OP_COPY_INVERTED 0x0000150c +#define NV50TCL_LOGIC_OP_OR_INVERTED 0x0000150d +#define NV50TCL_LOGIC_OP_NAND 0x0000150e +#define NV50TCL_LOGIC_OP_SET 0x0000150f +#define NV50TCL_CLEAR_BUFFERS 0x000019d0 +#define NV50TCL_COLOR_MASK(x) (0x00001a00+((x)*4)) +#define NV50TCL_COLOR_MASK__SIZE 0x00000008 +#define NV50TCL_COLOR_MASK_R_SHIFT 0 +#define NV50TCL_COLOR_MASK_R_MASK 0x0000000f +#define NV50TCL_COLOR_MASK_G_SHIFT 4 +#define NV50TCL_COLOR_MASK_G_MASK 0x000000f0 +#define NV50TCL_COLOR_MASK_B_SHIFT 8 +#define NV50TCL_COLOR_MASK_B_MASK 0x00000f00 +#define NV50TCL_COLOR_MASK_A_SHIFT 12 +#define NV50TCL_COLOR_MASK_A_MASK 0x0000f000 + + +#define NV50_COMPUTE 0x000050c0 + +#define NV50_COMPUTE_DMA_UNK0 0x000001a0 +#define NV50_COMPUTE_DMA_STATUS 0x000001a4 +#define NV50_COMPUTE_DMA_UNK1 0x000001b8 +#define NV50_COMPUTE_DMA_UNK2 0x000001bc +#define NV50_COMPUTE_DMA_UNK3 0x000001c0 +#define NV50_COMPUTE_UNK4_HIGH 0x00000210 +#define NV50_COMPUTE_UNK4_LOW 0x00000214 +#define NV50_COMPUTE_UNK5_HIGH 0x00000218 +#define NV50_COMPUTE_UNK5_LOW 0x0000021c +#define NV50_COMPUTE_UNK6_HIGH 0x00000294 +#define NV50_COMPUTE_UNK6_LOW 0x00000298 +#define NV50_COMPUTE_CONST_BASE_HIGH 0x000002a4 +#define NV50_COMPUTE_CONST_BASE_LO 0x000002a8 +#define NV50_COMPUTE_CONST_SIZE_SEG 0x000002ac +#define NV50_COMPUTE_REG_COUNT 0x000002c0 +#define NV50_COMPUTE_STATUS_HIGH 0x00000310 +#define NV50_COMPUTE_STATUS_LOW 0x00000314 +#define NV50_COMPUTE_EXECUTE 0x0000031c +#define NV50_COMPUTE_USER_PARAM_COUNT 0x00000374 +#define NV50_COMPUTE_GRIDDIM_YX 0x000003a4 +#define NV50_COMPUTE_SHARED_SIZE 0x000003a8 +#define NV50_COMPUTE_BLOCKDIM_YX 0x000003ac +#define NV50_COMPUTE_BLOCKDIM_Z 0x000003b0 +#define NV50_COMPUTE_CALL_ADDRESS 0x000003b4 +#define NV50_COMPUTE_GLOBAL_BASE_HIGH(x) (0x00000400+((x)*32)) +#define NV50_COMPUTE_GLOBAL_BASE_HIGH__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_BASE_LOW(x) (0x00000404+((x)*32)) +#define NV50_COMPUTE_GLOBAL_BASE_LOW__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_LIMIT_HIGH(x) (0x00000408+((x)*32)) +#define NV50_COMPUTE_GLOBAL_LIMIT_HIGH__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_LIMIT_LOW(x) (0x0000040c+((x)*32)) +#define NV50_COMPUTE_GLOBAL_LIMIT_LOW__SIZE 0x00000010 +#define NV50_COMPUTE_GLOBAL_UNK(x) (0x00000410+((x)*32)) +#define NV50_COMPUTE_GLOBAL_UNK__SIZE 0x00000010 +#define NV50_COMPUTE_USER_PARAM(x) (0x00000600+((x)*4)) +#define NV50_COMPUTE_USER_PARAM__SIZE 0x00000040 + + +#define NV54TCL 0x00008297 + + + +#endif /* NOUVEAU_REG_H */ diff --git a/src/gallium/drivers/nouveau/nouveau_gldefs.h b/src/gallium/drivers/nouveau/nouveau_gldefs.h new file mode 100644 index 00000000000..e1015c93a27 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_gldefs.h @@ -0,0 +1,196 @@ +#ifndef __NOUVEAU_GLDEFS_H__ +#define __NOUVEAU_GLDEFS_H__ + +static INLINE unsigned +nvgl_blend_func(unsigned factor) +{ + switch (factor) { + case PIPE_BLENDFACTOR_ZERO: + return 0x0000; + case PIPE_BLENDFACTOR_ONE: + return 0x0001; + case PIPE_BLENDFACTOR_SRC_COLOR: + return 0x0300; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + return 0x0301; + case PIPE_BLENDFACTOR_SRC_ALPHA: + return 0x0302; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + return 0x0303; + case PIPE_BLENDFACTOR_DST_ALPHA: + return 0x0304; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + return 0x0305; + case PIPE_BLENDFACTOR_DST_COLOR: + return 0x0306; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + return 0x0307; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + return 0x0308; + case PIPE_BLENDFACTOR_CONST_COLOR: + return 0x8001; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + return 0x8002; + case PIPE_BLENDFACTOR_CONST_ALPHA: + return 0x8003; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + return 0x8004; + default: + return 0x0000; + } +} + +static INLINE unsigned +nvgl_blend_eqn(unsigned func) +{ + switch (func) { + case PIPE_BLEND_ADD: + return 0x8006; + case PIPE_BLEND_MIN: + return 0x8007; + case PIPE_BLEND_MAX: + return 0x8008; + case PIPE_BLEND_SUBTRACT: + return 0x800a; + case PIPE_BLEND_REVERSE_SUBTRACT: + return 0x800b; + default: + return 0x8006; + } +} + +static INLINE unsigned +nvgl_logicop_func(unsigned func) +{ + switch (func) { + case PIPE_LOGICOP_CLEAR: + return 0x1500; + case PIPE_LOGICOP_NOR: + return 0x1508; + case PIPE_LOGICOP_AND_INVERTED: + return 0x1504; + case PIPE_LOGICOP_COPY_INVERTED: + return 0x150c; + case PIPE_LOGICOP_AND_REVERSE: + return 0x1502; + case PIPE_LOGICOP_INVERT: + return 0x150a; + case PIPE_LOGICOP_XOR: + return 0x1506; + case PIPE_LOGICOP_NAND: + return 0x150e; + case PIPE_LOGICOP_AND: + return 0x1501; + case PIPE_LOGICOP_EQUIV: + return 0x1509; + case PIPE_LOGICOP_NOOP: + return 0x1505; + case PIPE_LOGICOP_OR_INVERTED: + return 0x150d; + case PIPE_LOGICOP_COPY: + return 0x1503; + case PIPE_LOGICOP_OR_REVERSE: + return 0x150b; + case PIPE_LOGICOP_OR: + return 0x1507; + case PIPE_LOGICOP_SET: + return 0x150f; + default: + return 0x1505; + } +} + +static INLINE unsigned +nvgl_comparison_op(unsigned op) +{ + switch (op) { + case PIPE_FUNC_NEVER: + return 0x0200; + case PIPE_FUNC_LESS: + return 0x0201; + case PIPE_FUNC_EQUAL: + return 0x0202; + case PIPE_FUNC_LEQUAL: + return 0x0203; + case PIPE_FUNC_GREATER: + return 0x0204; + case PIPE_FUNC_NOTEQUAL: + return 0x0205; + case PIPE_FUNC_GEQUAL: + return 0x0206; + case PIPE_FUNC_ALWAYS: + return 0x0207; + default: + return 0x0207; + } +} + +static INLINE unsigned +nvgl_polygon_mode(unsigned mode) +{ + switch (mode) { + case PIPE_POLYGON_MODE_POINT: + return 0x1b00; + case PIPE_POLYGON_MODE_LINE: + return 0x1b01; + case PIPE_POLYGON_MODE_FILL: + return 0x1b02; + default: + return 0x1b02; + } +} + +static INLINE unsigned +nvgl_stencil_op(unsigned op) +{ + switch (op) { + case PIPE_STENCIL_OP_ZERO: + return 0x0000; + case PIPE_STENCIL_OP_INVERT: + return 0x150a; + case PIPE_STENCIL_OP_KEEP: + return 0x1e00; + case PIPE_STENCIL_OP_REPLACE: + return 0x1e01; + case PIPE_STENCIL_OP_INCR: + return 0x1e02; + case PIPE_STENCIL_OP_DECR: + return 0x1e03; + case PIPE_STENCIL_OP_INCR_WRAP: + return 0x8507; + case PIPE_STENCIL_OP_DECR_WRAP: + return 0x8508; + default: + return 0x1e00; + } +} + +static INLINE unsigned +nvgl_primitive(unsigned prim) { + switch (prim) { + case PIPE_PRIM_POINTS: + return 0x0001; + case PIPE_PRIM_LINES: + return 0x0002; + case PIPE_PRIM_LINE_LOOP: + return 0x0003; + case PIPE_PRIM_LINE_STRIP: + return 0x0004; + case PIPE_PRIM_TRIANGLES: + return 0x0005; + case PIPE_PRIM_TRIANGLE_STRIP: + return 0x0006; + case PIPE_PRIM_TRIANGLE_FAN: + return 0x0007; + case PIPE_PRIM_QUADS: + return 0x0008; + case PIPE_PRIM_QUAD_STRIP: + return 0x0009; + case PIPE_PRIM_POLYGON: + return 0x000a; + default: + return 0x0001; + } +} + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_grobj.h b/src/gallium/drivers/nouveau/nouveau_grobj.h new file mode 100644 index 00000000000..8f5abf90514 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_grobj.h @@ -0,0 +1,35 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_GROBJ_H__ +#define __NOUVEAU_GROBJ_H__ + +#include "nouveau_channel.h" + +struct nouveau_grobj { + struct nouveau_channel *channel; + int grclass; + uint32_t handle; + int subc; +}; + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_notifier.h b/src/gallium/drivers/nouveau/nouveau_notifier.h new file mode 100644 index 00000000000..35adde1e324 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_notifier.h @@ -0,0 +1,43 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_NOTIFIER_H__ +#define __NOUVEAU_NOTIFIER_H__ + +#define NV_NOTIFIER_SIZE 32 +#define NV_NOTIFY_TIME_0 0x00000000 +#define NV_NOTIFY_TIME_1 0x00000004 +#define NV_NOTIFY_RETURN_VALUE 0x00000008 +#define NV_NOTIFY_STATE 0x0000000C +#define NV_NOTIFY_STATE_STATUS_MASK 0xFF000000 +#define NV_NOTIFY_STATE_STATUS_SHIFT 24 +#define NV_NOTIFY_STATE_STATUS_COMPLETED 0x00 +#define NV_NOTIFY_STATE_STATUS_IN_PROCESS 0x01 +#define NV_NOTIFY_STATE_ERROR_CODE_MASK 0x0000FFFF +#define NV_NOTIFY_STATE_ERROR_CODE_SHIFT 0 + +struct nouveau_notifier { + struct nouveau_channel *channel; + uint32_t handle; +}; + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_push.h b/src/gallium/drivers/nouveau/nouveau_push.h new file mode 100644 index 00000000000..c5c5d988d58 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_push.h @@ -0,0 +1,83 @@ +#ifndef __NOUVEAU_PUSH_H__ +#define __NOUVEAU_PUSH_H__ + +#include "nouveau/nouveau_winsys.h" + +#ifndef NOUVEAU_PUSH_CONTEXT +#error undefined push context +#endif + +#define OUT_RING(data) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + (*pc->nvws->channel->pushbuf->cur++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + memcpy(pc->nvws->channel->pushbuf->cur, (src), (size) * 4); \ + pc->nvws->channel->pushbuf->cur += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define BEGIN_RING(obj,mthd,size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + OUT_RING((pc->obj->subc << 13) | ((size) << 18) | (mthd)); \ + pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + +#define FIRE_RING() do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + pc->nvws->push_flush(pc->nvws->channel, 0); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + pc->nvws->push_reloc(pc->nvws->channel, \ + pc->nvws->channel->pushbuf->cur++, \ + (bo), (data), (flags), (vor), (tor)); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + pc->nvws->channel->vram->handle, \ + pc->nvws->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +/* A reloc which'll recombine into a NV_DMA_METHOD packet header */ +#define OUT_RELOCm(bo, flags, obj, mthd, size) do { \ + NOUVEAU_PUSH_CONTEXT(pc); \ + if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + OUT_RELOCd((bo), (pc->obj->subc << 13) | ((size) << 18) | (mthd), \ + (flags), 0, 0); \ + pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_pushbuf.h b/src/gallium/drivers/nouveau/nouveau_pushbuf.h new file mode 100644 index 00000000000..19097650982 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_pushbuf.h @@ -0,0 +1,32 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_PUSHBUF_H__ +#define __NOUVEAU_PUSHBUF_H__ + +struct nouveau_pushbuf { + struct nouveau_channel *channel; + unsigned remaining; + uint32_t *cur; +}; + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_resource.h b/src/gallium/drivers/nouveau/nouveau_resource.h new file mode 100644 index 00000000000..1af7961d301 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_resource.h @@ -0,0 +1,37 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_RESOURCE_H__ +#define __NOUVEAU_RESOURCE_H__ + +struct nouveau_resource { + struct nouveau_resource *prev; + struct nouveau_resource *next; + + int in_use; + void *priv; + + unsigned int start; + unsigned int size; +}; + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h new file mode 100644 index 00000000000..07c31b014a5 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -0,0 +1,141 @@ +#ifndef __NOUVEAU_STATEOBJ_H__ +#define __NOUVEAU_STATEOBJ_H__ + +#include "pipe/p_util.h" + +struct nouveau_stateobj_reloc { + struct pipe_buffer *bo; + + unsigned offset; + unsigned packet; + + unsigned data; + unsigned flags; + unsigned vor; + unsigned tor; +}; + +struct nouveau_stateobj { + int refcount; + + unsigned *push; + struct nouveau_stateobj_reloc *reloc; + + unsigned *cur; + unsigned cur_packet; + unsigned cur_reloc; +}; + +static INLINE struct nouveau_stateobj * +so_new(unsigned push, unsigned reloc) +{ + struct nouveau_stateobj *so; + + so = MALLOC(sizeof(struct nouveau_stateobj)); + so->refcount = 1; + so->push = MALLOC(sizeof(unsigned) * push); + so->reloc = MALLOC(sizeof(struct nouveau_stateobj_reloc) * reloc); + + so->cur = so->push; + so->cur_reloc = so->cur_packet = 0; + + return so; +} + +static INLINE void +so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) +{ + struct nouveau_stateobj *so; + + so = *pso; + if (so) { + if (--so->refcount <= 0) { + free(so->push); + free(so->reloc); + free(so); + } + *pso = NULL; + } + + if (ref) { + ref->refcount++; + *pso = ref; + } +} + +static INLINE void +so_data(struct nouveau_stateobj *so, unsigned data) +{ + (*so->cur++) = (data); + so->cur_packet += 4; +} + +static INLINE void +so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, + unsigned mthd, unsigned size) +{ + so->cur_packet = (gr->subc << 13) | (1 << 18) | (mthd - 4); + so_data(so, (gr->subc << 13) | (size << 18) | mthd); +} + +static INLINE void +so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + struct nouveau_stateobj_reloc *r = &so->reloc[so->cur_reloc++]; + + r->bo = bo; + r->offset = so->cur - so->push; + r->packet = so->cur_packet; + r->data = data; + r->flags = flags; + r->vor = vor; + r->tor = tor; + so_data(so, data); +} + +static INLINE void +so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) +{ + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + unsigned nr, i; + + nr = so->cur - so->push; + if (pb->remaining < nr) + nvws->push_flush(nvws->channel, nr); + pb->remaining -= nr; + + memcpy(pb->cur, so->push, nr * 4); + for (i = 0; i < so->cur_reloc; i++) { + struct nouveau_stateobj_reloc *r = &so->reloc[i]; + + nvws->push_reloc(nvws->channel, pb->cur + r->offset, r->bo, + r->data, r->flags, r->vor, r->tor); + } + pb->cur += nr; +} + +static INLINE void +so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) +{ + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + unsigned i; + + i = so->cur_reloc << 1; + if (nvws->channel->pushbuf->remaining < i) + nvws->push_flush(nvws->channel, i); + nvws->channel->pushbuf->remaining -= i; + + for (i = 0; i < so->cur_reloc; i++) { + struct nouveau_stateobj_reloc *r = &so->reloc[i]; + + nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->packet, + (r->flags & + (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART)) | + NOUVEAU_BO_DUMMY, 0, 0); + nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->data, + r->flags | NOUVEAU_BO_DUMMY, r->vor, r->tor); + } +} + +#endif diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h new file mode 100644 index 00000000000..b5e470cfaa4 --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -0,0 +1,61 @@ +#ifndef NOUVEAU_WINSYS_H +#define NOUVEAU_WINSYS_H + +#include +#include "pipe/p_winsys.h" +#include "pipe/p_defines.h" + +#include "nouveau/nouveau_bo.h" +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_class.h" +#include "nouveau/nouveau_grobj.h" +#include "nouveau/nouveau_notifier.h" +#include "nouveau/nouveau_resource.h" +#include "nouveau/nouveau_pushbuf.h" + +struct nouveau_winsys { + struct nouveau_context *nv; + + struct nouveau_channel *channel; + + int (*res_init)(struct nouveau_resource **heap, unsigned start, + unsigned size); + int (*res_alloc)(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **); + void (*res_free)(struct nouveau_resource **); + + int (*push_reloc)(struct nouveau_channel *, void *ptr, + struct pipe_buffer *, uint32_t data, + uint32_t flags, uint32_t vor, uint32_t tor); + int (*push_flush)(struct nouveau_channel *, unsigned size); + + int (*grobj_alloc)(struct nouveau_winsys *, int grclass, + struct nouveau_grobj **); + void (*grobj_free)(struct nouveau_grobj **); + + int (*notifier_alloc)(struct nouveau_winsys *, int count, + struct nouveau_notifier **); + void (*notifier_free)(struct nouveau_notifier **); + void (*notifier_reset)(struct nouveau_notifier *, int id); + uint32_t (*notifier_status)(struct nouveau_notifier *, int id); + uint32_t (*notifier_retval)(struct nouveau_notifier *, int id); + int (*notifier_wait)(struct nouveau_notifier *, int id, + int status, int timeout); + + int (*surface_copy)(struct nouveau_winsys *, struct pipe_surface *, + unsigned, unsigned, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned); + int (*surface_fill)(struct nouveau_winsys *, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned, unsigned); +}; + +extern struct pipe_context * +nv30_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); + +extern struct pipe_context * +nv40_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); + +extern struct pipe_context * +nv50_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); + +#endif diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile new file mode 100644 index 00000000000..b7c252fc986 --- /dev/null +++ b/src/gallium/drivers/nv30/Makefile @@ -0,0 +1,29 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv30 + +DRIVER_SOURCES = \ + nv30_clear.c \ + nv30_context.c \ + nv30_draw.c \ + nv30_fragprog.c \ + nv30_fragtex.c \ + nv30_miptree.c \ + nv30_query.c \ + nv30_state.c \ + nv30_state_emit.c \ + nv30_surface.c \ + nv30_vbo.c \ + nv30_vertprog.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/drivers/nv30/nv30_clear.c b/src/gallium/drivers/nv30/nv30_clear.c new file mode 100644 index 00000000000..71f413588ee --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv30_context.h" + +void +nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c new file mode 100644 index 00000000000..e9afeb80176 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -0,0 +1,431 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" + +static const char * +nv30_get_name(struct pipe_context *pipe) +{ + struct nv30_context *nv30 = nv30_context(pipe); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv30->chipset); + return buffer; +} + +static const char * +nv30_get_vendor(struct pipe_context *pipe) +{ + return "nouveau"; +} + +static int +nv30_get_param(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 16; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 1; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 2; + case PIPE_CAP_OCCLUSION_QUERY: + return 1; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 1; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv30_get_paramf(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv30_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + if (flags & PIPE_FLUSH_TEXTURE_CACHE) { + BEGIN_RING(rankine, 0x1fd8, 1); + OUT_RING (2); + BEGIN_RING(rankine, 0x1fd8, 1); + OUT_RING (1); + } + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv30->sync, 0); + BEGIN_RING(rankine, 0x104, 1); + OUT_RING (0); + BEGIN_RING(rankine, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv30->sync, 0, 0, 2000); +} + +static void +nv30_destroy(struct pipe_context *pipe) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + if (nv30->draw) + draw_destroy(nv30->draw); + + nvws->res_free(&nv30->vertprog.exec_heap); + nvws->res_free(&nv30->vertprog.data_heap); + + nvws->res_free(&nv30->query_heap); + nvws->notifier_free(&nv30->query); + + nvws->notifier_free(&nv30->sync); + + nvws->grobj_free(&nv30->rankine); + + free(nv30); +} + +static boolean +nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) +{ + struct nouveau_winsys *nvws = nv30->nvws; + int ret; + int i; + + ret = nvws->grobj_alloc(nvws, rankine_class, &nv30->rankine); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + BEGIN_RING(rankine, NV34TCL_DMA_NOTIFY, 1); + OUT_RING (nv30->sync->handle); + BEGIN_RING(rankine, NV34TCL_DMA_TEXTURE0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(rankine, NV34TCL_DMA_VTXBUF0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); +/* BEGIN_RING(rankine, NV34TCL_DMA_FENCE, 2); + OUT_RING (0); + OUT_RING (nv30->query->handle);*/ + BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY7, 1); + OUT_RING (nvws->channel->vram->handle); + BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY8, 1); + OUT_RING (nvws->channel->vram->handle); + + for (i=1; i<8; i++) { + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(i), 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_VERT(i), 1); + OUT_RING (0); + } + + BEGIN_RING(rankine, 0x220, 1); + OUT_RING (1); + + BEGIN_RING(rankine, 0x03b0, 1); + OUT_RING (0x00100000); + BEGIN_RING(rankine, 0x1454, 1); + OUT_RING (0); + BEGIN_RING(rankine, 0x1d80, 1); + OUT_RING (3); + BEGIN_RING(rankine, 0x1450, 1); + OUT_RING (0x00030004); + + /* NEW */ + BEGIN_RING(rankine, 0x1e98, 1); + OUT_RING (0); + BEGIN_RING(rankine, 0x17e0, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x3f800000); + BEGIN_RING(rankine, 0x1f80, 16); + OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); + OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); + OUT_RING (0x0000ffff); + OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); + OUT_RING (0); OUT_RING (0); OUT_RING (0); + + BEGIN_RING(rankine, 0x120, 3); + OUT_RING (0); + OUT_RING (1); + OUT_RING (2); + + BEGIN_RING(rankine, 0x1d88, 1); + OUT_RING (0x00001200); + + BEGIN_RING(rankine, NV34TCL_RC_ENABLE, 1); + OUT_RING (0); + + /* Attempt to setup a known state.. Probably missing a heap of + * stuff here.. + */ + BEGIN_RING(rankine, NV34TCL_STENCIL_FRONT_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_DEPTH_WRITE_ENABLE, 2); + OUT_RING (0); /* wr disable */ + OUT_RING (0); /* test disable */ + BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); + OUT_RING (0x01010101); /* TR,TR,TR,TR */ + BEGIN_RING(rankine, NV34TCL_CULL_FACE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 5); + OUT_RING (0); /* Blend enable */ + OUT_RING (0); /* Blend src */ + OUT_RING (0); /* Blend dst */ + OUT_RING (0x00000000); /* Blend colour */ + OUT_RING (0x8006); /* FUNC_ADD */ + BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); + OUT_RING (0); + OUT_RING (0x1503 /*GL_COPY*/); + BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); + OUT_RING (1); + BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); + OUT_RING (0x1d01 /*GL_SMOOTH*/); + BEGIN_RING(rankine, NV34TCL_POLYGON_OFFSET_FACTOR,2); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (0x1b02 /*GL_FILL*/); + OUT_RING (0x1b02 /*GL_FILL*/); + /* - Disable texture units + * - Set fragprog to MOVR result.color, fragment.color */ + for (i=0;i<16;i++) { + BEGIN_RING(rankine, + NV34TCL_TX_ENABLE(i), 1); + OUT_RING (0); + } + /* Polygon stipple */ + BEGIN_RING(rankine, + NV34TCL_POLYGON_STIPPLE_PATTERN(0), 0x20); + for (i=0;i<0x20;i++) + OUT_RING (0xFFFFFFFF); + + int w=4096; + int h=4096; + int pitch=4096*4; + BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 5); + OUT_RING (w<<16); + OUT_RING (h<<16); + OUT_RING (0x148); /* format */ + OUT_RING (pitch << 16 | pitch); + OUT_RING (0x0); + BEGIN_RING(rankine, 0x0a00, 2); + OUT_RING ((w<<16) | 0); + OUT_RING ((h<<16) | 0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING ((w-1)<<16); + OUT_RING ((h-1)<<16); + BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); + OUT_RING (w<<16); + OUT_RING (h<<16); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); + OUT_RING (w<<16); + OUT_RING (h<<16); + + BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + + BEGIN_RING(rankine, NV34TCL_MODELVIEW_MATRIX(0), 16); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + + BEGIN_RING(rankine, NV34TCL_PROJECTION_MATRIX(0), 16); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + + BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); + OUT_RING (4096<<16); + OUT_RING (4096<<16); + + BEGIN_RING(rankine, NV34TCL_MULTISAMPLE_CONTROL, 1); + OUT_RING (0xffff0000); + + FIRE_RING (); + return TRUE; +} + +#define NV30TCL_CHIPSET_3X_MASK 0x00000003 +#define NV34TCL_CHIPSET_3X_MASK 0x00000010 +#define NV35TCL_CHIPSET_3X_MASK 0x000001e0 + +struct pipe_context * +nv30_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv30_context *nv30; + int rankine_class = 0, ret; + + if ((chipset & 0xf0) != 0x30) { + NOUVEAU_ERR("Not a NV3X chipset\n"); + return NULL; + } + + if (NV30TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { + rankine_class = 0x0397; + } else if (NV34TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { + rankine_class = 0x0697; + } else if (NV35TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { + rankine_class = 0x0497; + } else { + NOUVEAU_ERR("Unknown NV3X chipset: NV%02x\n", chipset); + return NULL; + } + + nv30 = CALLOC_STRUCT(nv30_context); + if (!nv30) + return NULL; + nv30->chipset = chipset; + nv30->nvws = nvws; + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &nv30->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Query objects */ + ret = nvws->notifier_alloc(nvws, 32, &nv30->query); + if (ret) { + NOUVEAU_ERR("Error initialising query objects: %d\n", ret); + nv30_destroy(&nv30->pipe); + return NULL; + } + + ret = nvws->res_init(&nv30->query_heap, 0, 32); + if (ret) { + NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Vtxprog resources */ + if (nvws->res_init(&nv30->vertprog.exec_heap, 0, 512) || + nvws->res_init(&nv30->vertprog.data_heap, 0, 256)) { + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Static rankine initialisation */ + if (!nv30_init_hwctx(nv30, rankine_class)) { + nv30_destroy(&nv30->pipe); + return NULL; + } + + /* Pipe context setup */ + nv30->pipe.winsys = pipe_winsys; + + nv30->pipe.destroy = nv30_destroy; + nv30->pipe.get_name = nv30_get_name; + nv30->pipe.get_vendor = nv30_get_vendor; + nv30->pipe.get_param = nv30_get_param; + nv30->pipe.get_paramf = nv30_get_paramf; + + nv30->pipe.draw_arrays = nv30_draw_arrays; + nv30->pipe.draw_elements = nv30_draw_elements; + nv30->pipe.clear = nv30_clear; + + nv30->pipe.flush = nv30_flush; + + nv30_init_query_functions(nv30); + nv30_init_surface_functions(nv30); + nv30_init_state_functions(nv30); + nv30_init_miptree_functions(nv30); + + nv30->draw = draw_create(); + assert(nv30->draw); + draw_set_rasterize_stage(nv30->draw, nv30_draw_render_stage(nv30)); + + return &nv30->pipe; +} + diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h new file mode 100644 index 00000000000..d6d16ee8686 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -0,0 +1,136 @@ +#ifndef __NV30_CONTEXT_H__ +#define __NV30_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv30_context *ctx = nv30 +#include "nouveau/nouveau_push.h" + +#include "nv30_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV30_NEW_VERTPROG (1 << 1) +#define NV30_NEW_FRAGPROG (1 << 2) +#define NV30_NEW_ARRAYS (1 << 3) + +struct nv30_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct draw_context *draw; + + int chipset; + struct nouveau_grobj *rankine; + struct nouveau_notifier *sync; + + /* query objects */ + struct nouveau_notifier *query; + struct nouveau_resource *query_heap; + + uint32_t dirty; + + struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + uint32_t rt_enable; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; + + struct { + struct pipe_buffer *buffer; + uint32_t format; + } tex[16]; + + unsigned vb_enable; + struct { + struct pipe_buffer *buffer; + unsigned delta; + } vb[16]; + + struct { + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + + struct nv30_vertex_program *active; + + struct nv30_vertex_program *current; + struct pipe_buffer *constant_buf; + } vertprog; + + struct { + struct nv30_fragment_program *active; + + struct nv30_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; +}; + +static INLINE struct nv30_context * +nv30_context(struct pipe_context *pipe) +{ + return (struct nv30_context *)pipe; +} + +extern void nv30_init_state_functions(struct nv30_context *nv30); +extern void nv30_init_surface_functions(struct nv30_context *nv30); +extern void nv30_init_miptree_functions(struct nv30_context *nv30); +extern void nv30_init_query_functions(struct nv30_context *nv30); + +/* nv30_draw.c */ +extern struct draw_stage *nv30_draw_render_stage(struct nv30_context *nv30); + +/* nv30_vertprog.c */ +extern void nv30_vertprog_translate(struct nv30_context *, + struct nv30_vertex_program *); +extern void nv30_vertprog_bind(struct nv30_context *, + struct nv30_vertex_program *); +extern void nv30_vertprog_destroy(struct nv30_context *, + struct nv30_vertex_program *); + +/* nv30_fragprog.c */ +extern void nv30_fragprog_translate(struct nv30_context *, + struct nv30_fragment_program *); +extern void nv30_fragprog_bind(struct nv30_context *, + struct nv30_fragment_program *); +extern void nv30_fragprog_destroy(struct nv30_context *, + struct nv30_fragment_program *); + +/* nv30_fragtex.c */ +extern void nv30_fragtex_bind(struct nv30_context *); + +/* nv30_state.c and friends */ +extern void nv30_emit_hw_state(struct nv30_context *nv30); +extern void nv30_state_tex_update(struct nv30_context *nv30); + +/* nv30_vbo.c */ +extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv30_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned mode, unsigned start, + unsigned count); + +/* nv30_clear.c */ +extern void nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +#endif diff --git a/src/gallium/drivers/nv30/nv30_draw.c b/src/gallium/drivers/nv30/nv30_draw.c new file mode 100644 index 00000000000..59a72657996 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_draw.c @@ -0,0 +1,62 @@ +#include "draw/draw_private.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" + +struct nv30_draw_stage { + struct draw_stage draw; + struct nv30_context *nv30; +}; + +static void +nv30_draw_point(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_line(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_tri(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_flush(struct draw_stage *draw, unsigned flags) +{ +} + +static void +nv30_draw_reset_stipple_counter(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv30_draw_destroy(struct draw_stage *draw) +{ + free(draw); +} + +struct draw_stage * +nv30_draw_render_stage(struct nv30_context *nv30) +{ + struct nv30_draw_stage *nv30draw = CALLOC_STRUCT(nv30_draw_stage); + + nv30draw->nv30 = nv30; + nv30draw->draw.draw = nv30->draw; + nv30draw->draw.point = nv30_draw_point; + nv30draw->draw.line = nv30_draw_line; + nv30draw->draw.tri = nv30_draw_tri; + nv30draw->draw.flush = nv30_draw_flush; + nv30draw->draw.reset_stipple_counter = nv30_draw_reset_stipple_counter; + nv30draw->draw.destroy = nv30_draw_destroy; + + return &nv30draw->draw; +} + diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c new file mode 100644 index 00000000000..09ad555c324 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -0,0 +1,835 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_util.h" + +#include "nv30_context.h" + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 1 +#define MASK_Y 2 +#define MASK_Z 4 +#define MASK_W 8 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE NV30_FP_OP_DST_SCALE_1X +#define DEF_CTEST NV30_FP_OP_COND_TR +#include "nv30_shader.h" + +#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv30_sr_neg((s)) +#define abs(s) nv30_sr_abs((s)) +#define scale(s,v) nv30_sr_scale((s), NV30_FP_OP_DST_SCALE_##v) + +#define MAX_CONSTS 128 +#define MAX_IMM 32 +struct nv30_fpc { + struct nv30_fragment_program *fp; + + uint attrib_map[PIPE_MAX_SHADER_INPUTS]; + + int high_temp; + int temp_temp_count; + int num_regs; + + uint depth_id; + uint colour_id; + + unsigned inst_offset; + + struct { + int pipe; + float vals[4]; + } consts[MAX_CONSTS]; + int nr_consts; + + struct nv30_sreg imm[MAX_IMM]; + unsigned nr_imm; +}; + +static INLINE struct nv30_sreg +temp(struct nv30_fpc *fpc) +{ + int idx; + + idx = fpc->temp_temp_count++; + idx += fpc->high_temp + 1; + return nv30_sr(NV30SR_TEMP, idx); +} + +static INLINE struct nv30_sreg +constant(struct nv30_fpc *fpc, int pipe, float vals[4]) +{ + int idx; + + if (fpc->nr_consts == MAX_CONSTS) + assert(0); + idx = fpc->nr_consts++; + + fpc->consts[idx].pipe = pipe; + if (pipe == -1) + memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float)); + return nv30_sr(NV30SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv30_fp_arith((cc), (s), NV30_FP_OP_OPCODE_##o, \ + (d), (m), (s0), (s1), (s2)) +#define tex(cc,s,o,u,d,m,s0,s1,s2) \ + nv30_fp_tex((cc), (s), NV30_FP_OP_OPCODE_##o, (u), \ + (d), (m), (s0), none, none) + +static void +grow_insns(struct nv30_fpc *fpc, int size) +{ + struct nv30_fragment_program *fp = fpc->fp; + + fp->insn_len += size; + fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len); +} + +static void +emit_src(struct nv30_fpc *fpc, int pos, struct nv30_sreg src) +{ + struct nv30_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; + uint32_t sr = 0; + + switch (src.type) { + case NV30SR_INPUT: + sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); + hw[0] |= (src.index << NV30_FP_OP_INPUT_SRC_SHIFT); + break; + case NV30SR_OUTPUT: + sr |= NV30_FP_REG_SRC_HALF; + /* fall-through */ + case NV30SR_TEMP: + sr |= (NV30_FP_REG_TYPE_TEMP << NV30_FP_REG_TYPE_SHIFT); + sr |= (src.index << NV30_FP_REG_SRC_SHIFT); + break; + case NV30SR_CONST: + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + if (fpc->consts[src.index].pipe >= 0) { + struct nv30_fragment_program_data *fpd; + + fp->consts = realloc(fp->consts, ++fp->nr_consts * + sizeof(*fpd)); + fpd = &fp->consts[fp->nr_consts - 1]; + fpd->offset = fpc->inst_offset + 4; + fpd->index = fpc->consts[src.index].pipe; + memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4); + } else { + memcpy(&fp->insn[fpc->inst_offset + 4], + fpc->consts[src.index].vals, + sizeof(uint32_t) * 4); + } + + sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT); + break; + case NV30SR_NONE: + sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV30_FP_REG_NEGATE; + + if (src.abs) + hw[1] |= (1 << (29 + pos)); + + sr |= ((src.swz[0] << NV30_FP_REG_SWZ_X_SHIFT) | + (src.swz[1] << NV30_FP_REG_SWZ_Y_SHIFT) | + (src.swz[2] << NV30_FP_REG_SWZ_Z_SHIFT) | + (src.swz[3] << NV30_FP_REG_SWZ_W_SHIFT)); + + hw[pos + 1] |= sr; +} + +static void +emit_dst(struct nv30_fpc *fpc, struct nv30_sreg dst) +{ + struct nv30_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; + + switch (dst.type) { + case NV30SR_TEMP: + if (fpc->num_regs < (dst.index + 1)) + fpc->num_regs = dst.index + 1; + break; + case NV30SR_OUTPUT: + if (dst.index == 1) { + fp->fp_control |= 0xe; + } else { + hw[0] |= NV30_FP_OP_OUT_REG_HALF; + } + break; + case NV30SR_NONE: + hw[0] |= (1 << 30); + break; + default: + assert(0); + } + + hw[0] |= (dst.index << NV30_FP_OP_OUT_REG_SHIFT); +} + +static void +nv30_fp_arith(struct nv30_fpc *fpc, int sat, int op, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2) +{ + struct nv30_fragment_program *fp = fpc->fp; + uint32_t *hw; + + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + memset(hw, 0, sizeof(uint32_t) * 4); + + if (op == NV30_FP_OP_OPCODE_KIL) + fp->fp_control |= NV34TCL_FP_CONTROL_USES_KIL; + hw[0] |= (op << NV30_FP_OP_OPCODE_SHIFT); + hw[0] |= (mask << NV30_FP_OP_OUTMASK_SHIFT); + hw[2] |= (dst.dst_scale << NV30_FP_OP_DST_SCALE_SHIFT); + + if (sat) + hw[0] |= NV30_FP_OP_OUT_SAT; + + if (dst.cc_update) + hw[0] |= NV30_FP_OP_COND_WRITE_ENABLE; + hw[1] |= (dst.cc_test << NV30_FP_OP_COND_SHIFT); + hw[1] |= ((dst.cc_swz[0] << NV30_FP_OP_COND_SWZ_X_SHIFT) | + (dst.cc_swz[1] << NV30_FP_OP_COND_SWZ_Y_SHIFT) | + (dst.cc_swz[2] << NV30_FP_OP_COND_SWZ_Z_SHIFT) | + (dst.cc_swz[3] << NV30_FP_OP_COND_SWZ_W_SHIFT)); + + emit_dst(fpc, dst); + emit_src(fpc, 0, s0); + emit_src(fpc, 1, s1); + emit_src(fpc, 2, s2); +} + +static void +nv30_fp_tex(struct nv30_fpc *fpc, int sat, int op, int unit, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2) +{ + struct nv30_fragment_program *fp = fpc->fp; + + nv30_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); + + fp->insn[fpc->inst_offset] |= (unit << NV30_FP_OP_TEX_UNIT_SHIFT); + fp->samplers |= (1 << unit); +} + +static INLINE struct nv30_sreg +tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc) +{ + struct nv30_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv30_sr(NV30SR_INPUT, + fpc->attrib_map[fsrc->SrcRegister.Index]); + break; + case TGSI_FILE_CONSTANT: + src = constant(fpc, fsrc->SrcRegister.Index, NULL); + break; + case TGSI_FILE_IMMEDIATE: + assert(fsrc->SrcRegister.Index < fpc->nr_imm); + src = fpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index + 1); + if (fpc->high_temp < src.index) + fpc->high_temp = src.index; + break; + /* This is clearly insane, but gallium hands us shaders like this. + * Luckily fragprog results are just temp regs.. + */ + case TGSI_FILE_OUTPUT: + if (fsrc->SrcRegister.Index == fpc->colour_id) + return nv30_sr(NV30SR_OUTPUT, 0); + else + return nv30_sr(NV30SR_OUTPUT, 1); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv30_sreg +tgsi_dst(struct nv30_fpc *fpc, const struct tgsi_full_dst_register *fdst) { + int idx; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + if (fdst->DstRegister.Index == fpc->colour_id) + return nv30_sr(NV30SR_OUTPUT, 0); + else + return nv30_sr(NV30SR_OUTPUT, 1); + break; + case TGSI_FILE_TEMPORARY: + idx = fdst->DstRegister.Index + 1; + if (fpc->high_temp < idx) + fpc->high_temp = idx; + return nv30_sr(NV30SR_TEMP, idx); + case TGSI_FILE_NULL: + return nv30_sr(NV30SR_NONE, 0); + default: + NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); + return nv30_sr(NV30SR_NONE, 0); + } +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +src_native_swz(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc, + struct nv30_sreg *src) +{ + const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + struct nv30_sreg tgsi = tgsi_src(fpc, fsrc); + uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; + uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, + fsrc->SrcRegisterExtSwz.NegateY, + fsrc->SrcRegisterExtSwz.NegateZ, + fsrc->SrcRegisterExtSwz.NegateW }; + uint c; + + for (c = 0; c < 4; c++) { + switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + mask |= (1 << c); + break; + case TGSI_EXTSWIZZLE_ZERO: + zero_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + case TGSI_EXTSWIZZLE_ONE: + one_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + default: + assert(0); + } + + if (!tgsi.negate && neg[c]) + neg_mask |= (1 << c); + } + + if (mask == MASK_ALL && !neg_mask) + return TRUE; + + *src = temp(fpc); + + if (mask) + arith(fpc, 0, MOV, *src, mask, tgsi, none, none); + + if (zero_mask) + arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); + + if (one_mask) + arith(fpc, 0, STR, *src, one_mask, *src, none, none); + + if (neg_mask) { + struct nv30_sreg one = temp(fpc); + arith(fpc, 0, STR, one, neg_mask, one, none, none); + arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none); + } + + return FALSE; +} + +static boolean +nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, + const struct tgsi_full_instruction *finst) +{ + const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + struct nv30_sreg src[3], dst, tmp; + int mask, sat, unit; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + fpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(fpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + case TGSI_FILE_CONSTANT: + case TGSI_FILE_TEMPORARY: + if (!src_native_swz(fpc, fsrc, &src[i])) + continue; + break; + default: + break; + } + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + NOUVEAU_MSG("extra src attr %d\n", + fsrc->SrcRegister.Index); + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + case TGSI_FILE_SAMPLER: + unit = fsrc->SrcRegister.Index; + break; + case TGSI_FILE_OUTPUT: + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_CMP: + tmp = temp(fpc); + arith(fpc, sat, MOV, dst, mask, src[2], none, none); + tmp.cc_update = 1; + arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); + dst.cc_test = NV30_VP_INST_COND_LT; + arith(fpc, sat, MOV, dst, mask, src[1], none, none); + break; + case TGSI_OPCODE_COS: + arith(fpc, sat, COS, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(fpc, sat, DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none); + arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X), + swz(src[1], W, W, W, W), none); + break; + case TGSI_OPCODE_DST: + arith(fpc, sat, DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(fpc, sat, EX2, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FLR: + arith(fpc, sat, FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(fpc, sat, FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_KIL: + arith(fpc, 0, KIL, none, 0, none, none, none); + break; + case TGSI_OPCODE_KILP: + dst = nv30_sr(NV30SR_NONE, 0); + dst.cc_update = 1; + arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); + dst.cc_update = 0; dst.cc_test = NV30_FP_OP_COND_LT; + arith(fpc, 0, KIL, dst, 0, none, none, none); + break; + case TGSI_OPCODE_LG2: + arith(fpc, sat, LG2, dst, mask, src[0], none, none); + break; +// case TGSI_OPCODE_LIT: + case TGSI_OPCODE_LRP: + tmp = temp(fpc); + arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); + arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); + break; + case TGSI_OPCODE_MAD: + arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(fpc, sat, MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(fpc, sat, MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(fpc, sat, MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(fpc); + arith(fpc, 0, LG2, tmp, MASK_X, + swz(src[0], X, X, X, X), none, none); + arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(fpc, sat, EX2, dst, mask, + swz(tmp, X, X, X, X), none, none); + break; + case TGSI_OPCODE_RCP: + arith(fpc, sat, RCP, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_RET: + assert(0); + break; + case TGSI_OPCODE_RFL: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); + arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); + arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, + swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); + arith(fpc, sat, MAD, dst, mask, + swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); + break; + case TGSI_OPCODE_RSQ: + tmp = temp(fpc); + arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, + abs(swz(src[0], X, X, X, X)), none, none); + arith(fpc, sat, EX2, dst, mask, + neg(swz(tmp, X, X, X, X)), none, none); + break; + case TGSI_OPCODE_SCS: + if (mask & MASK_X) { + arith(fpc, sat, COS, dst, MASK_X, + swz(src[0], X, X, X, X), none, none); + } + if (mask & MASK_Y) { + arith(fpc, sat, SIN, dst, MASK_Y, + swz(src[0], X, X, X, X), none, none); + } + break; + case TGSI_OPCODE_SIN: + arith(fpc, sat, SIN, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_SGE: + arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); + break; + case TGSI_OPCODE_TEX: + if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == + TGSI_EXTSWIZZLE_W) { + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + } else + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_TXB: + tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_XPD: + tmp = temp(fpc); + arith(fpc, 0, MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(fpc, sat, MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV30_FP_OP_INPUT_SRC_POSITION; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_FP_OP_INPUT_SRC_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_FP_OP_INPUT_SRC_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV30_FP_OP_INPUT_SRC_FOGC; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV30_FP_OP_INPUT_SRC_TC(fdec->Semantic. + SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad input semantic\n"); + return FALSE; + } + + fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + fpc->depth_id = fdec->u.DeclarationRange.First; + break; + case TGSI_SEMANTIC_COLOR: + fpc->colour_id = fdec->u.DeclarationRange.First; + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + return TRUE; +} + +void +nv30_fragprog_translate(struct nv30_context *nv30, + struct nv30_fragment_program *fp) +{ + struct tgsi_parse_context parse; + struct nv30_fpc *fpc = NULL; + + fpc = CALLOC(1, sizeof(struct nv30_fpc)); + if (!fpc) + return; + fpc->fp = fp; + fpc->high_temp = -1; + fpc->num_regs = 2; + + tgsi_parse_init(&parse, fp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_INPUT: + if (!nv30_fragprog_parse_decl_attrib(fpc, fdec)) + goto out_err; + break; + case TGSI_FILE_OUTPUT: + if (!nv30_fragprog_parse_decl_output(fpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + struct tgsi_full_immediate *imm; + float vals[4]; + int i; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); + assert(fpc->nr_imm < MAX_IMM); + + for (i = 0; i < 4; i++) + vals[i] = imm->u.ImmediateFloat32[i].Float; + fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + + finst = &parse.FullToken.FullInstruction; + if (!nv30_fragprog_parse_instruction(fpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + fp->fp_control |= (fpc->num_regs-1)/2; + fp->fp_reg_control = (1<<16)|0x4; + + /* Terminate final instruction */ + fp->insn[fpc->inst_offset] |= 0x00000001; + + /* Append NOP + END instruction, may or may not be necessary. */ + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + fp->insn[fpc->inst_offset + 0] = 0x00000001; + fp->insn[fpc->inst_offset + 1] = 0x00000000; + fp->insn[fpc->inst_offset + 2] = 0x00000000; + fp->insn[fpc->inst_offset + 3] = 0x00000000; + + fp->translated = TRUE; + fp->on_hw = FALSE; +out_err: + tgsi_parse_free(&parse); + free(fpc); +} + +void +nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) +{ + struct pipe_winsys *ws = nv30->pipe.winsys; + int i; + + if (!fp->translated) { + nv30_fragprog_translate(nv30, fp); + if (!fp->translated) + assert(0); + } + + if (fp->nr_consts) { + float *map = ws->buffer_map(ws, nv30->fragprog.constant_buf, + PIPE_BUFFER_USAGE_CPU_READ); + for (i = 0; i < fp->nr_consts; i++) { + struct nv30_fragment_program_data *fpd = &fp->consts[i]; + uint32_t *p = &fp->insn[fpd->offset]; + uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; + + if (!memcmp(p, cb, 4 * sizeof(float))) + continue; + memcpy(p, cb, 4 * sizeof(float)); + fp->on_hw = 0; + } + ws->buffer_unmap(ws, nv30->fragprog.constant_buf); + } + + if (!fp->on_hw) { + const uint32_t le = 1; + uint32_t *map; + + if (!fp->buffer) + fp->buffer = ws->buffer_create(ws, 0x100, 0, + fp->insn_len * 4); + map = ws->buffer_map(ws, fp->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); + +#if 0 + for (i = 0; i < fp->insn_len; i++) { + NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + } +#endif + + if ((*(const uint8_t *)&le)) { + for (i = 0; i < fp->insn_len; i++) { + map[i] = fp->insn[i]; + } + } else { + /* Weird swapping for big-endian chips */ + for (i = 0; i < fp->insn_len; i++) { + map[i] = ((fp->insn[i] & 0xffff) << 16) | + ((fp->insn[i] >> 16) & 0xffff); + } + } + + ws->buffer_unmap(ws, fp->buffer); + fp->on_hw = TRUE; + } + + BEGIN_RING(rankine, NV34TCL_FP_CONTROL, 1); + OUT_RING (fp->fp_control); + BEGIN_RING(rankine, NV34TCL_FP_REG_CONTROL, 1); + OUT_RING (fp->fp_reg_control); + + nv30->fragprog.active = fp; +} + +void +nv30_fragprog_destroy(struct nv30_context *nv30, + struct nv30_fragment_program *fp) +{ + if (fp->insn_len) + free(fp->insn); +} + diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c new file mode 100644 index 00000000000..45ee6db8d6a --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -0,0 +1,160 @@ +#include "nv30_context.h" + +static INLINE int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV34TCL_TX_FORMAT_FORMAT_##tf, \ + (NV34TCL_TX_SWIZZLE_S0_X_##ts0x | NV34TCL_TX_SWIZZLE_S0_Y_##ts0y | \ + NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w | \ + NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y | \ + NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w), \ +} + +struct nv30_texture_format { + boolean defined; + uint pipe; + int format; + int swizzle; +}; + +static struct nv30_texture_format +nv30_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), +// _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), + _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), +// _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), +// _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), +// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + {}, +}; + +static struct nv30_texture_format * +nv30_fragtex_format(uint pipe_format) +{ + struct nv30_texture_format *tf = nv30_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv30_fragtex_build(struct nv30_context *nv30, int unit) +{ + struct nv30_sampler_state *ps = nv30->tex_sampler[unit]; + struct nv30_miptree *nv30mt = nv30->tex_miptree[unit]; + struct pipe_texture *pt = &nv30mt->base; + struct nv30_texture_format *tf; + uint32_t txf, txs, txp; + int swizzled = 0; /*XXX: implement in region code? */ + + tf = nv30_fragtex_format(pt->format); + if (!tf || !tf->defined) { + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); + return; + } + + txf = tf->format << 8; + txf |= (pt->last_level + 1) << 16; + txf |= log2i(pt->width[0]) << 20; + txf |= log2i(pt->height[0]) << 24; + txf |= log2i(pt->depth[0]) << 28; + txf |= 8; + + switch (pt->target) { +/* case PIPE_TEXTURE_CUBE: + txf |= NV34TCL_TEX_FORMAT_CUBIC;*/ + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= (2<<4); + break; + case PIPE_TEXTURE_3D: + txf |= (3<<4); + break; + case PIPE_TEXTURE_1D: + txf |= (1<<4); + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + txs = tf->swizzle; + + BEGIN_RING(rankine, NV34TCL_TX_OFFSET(unit), 8); + OUT_RELOCl(nv30mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv30mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RING (ps->wrap); + OUT_RING (0x40000000); /* enable */ + OUT_RING (txs); + OUT_RING (ps->filt | 0x2000 /* magic */); + OUT_RING ((pt->width[0] << 16) | pt->height[0]); + OUT_RING (ps->bcol); +} + +void +nv30_fragtex_bind(struct nv30_context *nv30) +{ + struct nv30_fragment_program *fp = nv30->fragprog.active; + unsigned samplers, unit; + + samplers = nv30->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(rankine, NV34TCL_TX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv30->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv30_fragtex_build(nv30, unit); + } + + nv30->fp_samplers = fp->samplers; +} + diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c new file mode 100644 index 00000000000..5fb89f4cfdc --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -0,0 +1,105 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nv30_context.h" + +static void +nv30_miptree_layout(struct nv30_miptree *nv30mt) +{ + struct pipe_texture *pt = &nv30mt->base; + boolean swizzled = FALSE; + uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint offset = 0; + int nr_faces, l, f; + + if (pt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else + if (pt->target == PIPE_TEXTURE_3D) { + nr_faces = pt->depth[0]; + } else { + nr_faces = 1; + } + + for (l = 0; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->depth[l] = depth; + + if (swizzled) + nv30mt->level[l].pitch = pt->width[l] * pt->cpp; + else + nv30mt->level[l].pitch = pt->width[0] * pt->cpp; + nv30mt->level[l].pitch = (nv30mt->level[l].pitch + 63) & ~63; + + nv30mt->level[l].image_offset = + CALLOC(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + depth = MAX2(1, depth >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = 0; l <= pt->last_level; l++) { + nv30mt->level[l].image_offset[f] = offset; + offset += nv30mt->level[l].pitch * pt->height[l]; + } + } + + nv30mt->total_size = offset; +} + +static void +nv30_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv30_miptree *nv30mt; + + nv30mt = realloc(*pt, sizeof(struct nv30_miptree)); + if (!nv30mt) + return; + *pt = NULL; + + nv30_miptree_layout(nv30mt); + + nv30mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + nv30mt->total_size); + if (!nv30mt->buffer) { + free(nv30mt); + return; + } + + *pt = &nv30mt->base; +} + +static void +nv30_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv30_miptree *nv30mt = (struct nv30_miptree *)mt; + int l; + + pipe_buffer_reference(ws, &nv30mt->buffer, NULL); + for (l = 0; l <= mt->last_level; l++) { + if (nv30mt->level[l].image_offset) + free(nv30mt->level[l].image_offset); + } + free(nv30mt); + } +} + +void +nv30_init_miptree_functions(struct nv30_context *nv30) +{ + nv30->pipe.texture_create = nv30_miptree_create; + nv30->pipe.texture_release = nv30_miptree_release; +} + diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c new file mode 100644 index 00000000000..71fdcfa24df --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_query.c @@ -0,0 +1,113 @@ +#include "pipe/p_context.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" + +struct nv30_query { + struct nouveau_resource *object; + unsigned type; + boolean ready; + uint64_t result; +}; + +static inline struct nv30_query * +nv30_query(struct pipe_query *pipe) +{ + return (struct nv30_query *)pipe; +} + +static struct pipe_query * +nv30_query_create(struct pipe_context *pipe, unsigned query_type) +{ + struct nv30_query *q; + + q = CALLOC(1, sizeof(struct nv30_query)); + q->type = query_type; + + return (struct pipe_query *)q; +} + +static void +nv30_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + + if (q->object) + nv30->nvws->res_free(&q->object); + free(q); +} + +static void +nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + + assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (nv30->nvws->res_alloc(nv30->query_heap, 1, NULL, &q->object)) + assert(0); + nv30->nvws->notifier_reset(nv30->query, q->object->start); + + BEGIN_RING(rankine, NV34TCL_QUERY_RESET, 1); + OUT_RING (1); + BEGIN_RING(rankine, NV34TCL_QUERY_UNK17CC, 1); + OUT_RING (1); + + q->ready = FALSE; +} + +static void +nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + + BEGIN_RING(rankine, NV34TCL_QUERY_GET, 1); + OUT_RING ((0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) | + ((q->object->start * 32) << NV34TCL_QUERY_GET_OFFSET_SHIFT)); + FIRE_RING(); +} + +static boolean +nv30_query_result(struct pipe_context *pipe, struct pipe_query *pq, + boolean wait, uint64 *result) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_query *q = nv30_query(pq); + struct nouveau_winsys *nvws = nv30->nvws; + + assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (!q->ready) { + unsigned status; + + status = nvws->notifier_status(nv30->query, q->object->start); + if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { + if (wait == FALSE) + return FALSE; + nvws->notifier_wait(nv30->query, q->object->start, + NV_NOTIFY_STATE_STATUS_COMPLETED, + 0); + } + + q->result = nvws->notifier_retval(nv30->query, + q->object->start); + q->ready = TRUE; + nvws->res_free(&q->object); + } + + *result = q->result; + return TRUE; +} + +void +nv30_init_query_functions(struct nv30_context *nv30) +{ + nv30->pipe.create_query = nv30_query_create; + nv30->pipe.destroy_query = nv30_query_destroy; + nv30->pipe.begin_query = nv30_query_begin; + nv30->pipe.end_query = nv30_query_end; + nv30->pipe.get_query_result = nv30_query_result; +} diff --git a/src/gallium/drivers/nv30/nv30_shader.h b/src/gallium/drivers/nv30/nv30_shader.h new file mode 100644 index 00000000000..dd3a36f78f3 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_shader.h @@ -0,0 +1,490 @@ +#ifndef __NV30_SHADER_H__ +#define __NV30_SHADER_H__ + +/* Vertex programs instruction set + * + * 128bit opcodes, split into 4 32-bit ones for ease of use. + * + * Non-native instructions + * ABS - MOV + NV40_VP_INST0_DEST_ABS + * POW - EX2 + MUL + LG2 + * SUB - ADD, second source negated + * SWZ - MOV + * XPD - + * + * Register access + * - Only one INPUT can be accessed per-instruction (move extras into TEMPs) + * - Only one CONST can be accessed per-instruction (move extras into TEMPs) + * + * Relative Addressing + * According to the value returned for + * MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB + * + * there are only two address registers available. The destination in the + * ARL instruction is set to TEMP (The temp isn't actually written). + * + * When using vanilla ARB_v_p, the proprietary driver will squish both the + * available ADDRESS regs into the first hardware reg in the X and Y + * components. + * + * To use an address reg as an index into consts, the CONST_SRC is set to + * (const_base + offset) and INDEX_CONST is set. + * + * To access the second address reg use ADDR_REG_SELECT_1. A particular + * component of the address regs is selected with ADDR_SWZ. + * + * Only one address register can be accessed per instruction. + * + * Conditional execution (see NV_vertex_program{2,3} for details) Conditional + * execution of an instruction is enabled by setting COND_TEST_ENABLE, and + * selecting the condition which will allow the test to pass with + * COND_{FL,LT,...}. It is possible to swizzle the values in the condition + * register, which allows for testing against an individual component. + * + * Branching: + * + * The BRA/CAL instructions seem to follow a slightly different opcode + * layout. The destination instruction ID (IADDR) overlaps a source field. + * Instruction ID's seem to be numbered based on the UPLOAD_FROM_ID FIFO + * command, and is incremented automatically on each UPLOAD_INST FIFO + * command. + * + * Conditional branching is achieved by using the condition tests described + * above. There doesn't appear to be dedicated looping instructions, but + * this can be done using a temp reg + conditional branching. + * + * Subroutines may be uploaded before the main program itself, but the first + * executed instruction is determined by the PROGRAM_START_ID FIFO command. + * + */ + +/* DWORD 0 */ + +#define NV30_VP_INST_ADDR_REG_SELECT_1 (1 << 24) +#define NV30_VP_INST_SRC2_ABS (1 << 23) /* guess */ +#define NV30_VP_INST_SRC1_ABS (1 << 22) /* guess */ +#define NV30_VP_INST_SRC0_ABS (1 << 21) /* guess */ +#define NV30_VP_INST_VEC_RESULT (1 << 20) +#define NV30_VP_INST_DEST_TEMP_ID_SHIFT 16 +#define NV30_VP_INST_DEST_TEMP_ID_MASK (0x0F << 16) +#define NV30_VP_INST_COND_UPDATE_ENABLE (1<<15) +#define NV30_VP_INST_VEC_DEST_TEMP_MASK (0xF << 16) +#define NV30_VP_INST_COND_TEST_ENABLE (1<<14) +#define NV30_VP_INST_COND_SHIFT 11 +#define NV30_VP_INST_COND_MASK (0x07 << 11) +# define NV30_VP_INST_COND_FL 0 /* guess */ +# define NV30_VP_INST_COND_LT 1 +# define NV30_VP_INST_COND_EQ 2 +# define NV30_VP_INST_COND_LE 3 +# define NV30_VP_INST_COND_GT 4 +# define NV30_VP_INST_COND_NE 5 +# define NV30_VP_INST_COND_GE 6 +# define NV30_VP_INST_COND_TR 7 /* guess */ +#define NV30_VP_INST_COND_SWZ_X_SHIFT 9 +#define NV30_VP_INST_COND_SWZ_X_MASK (0x03 << 9) +#define NV30_VP_INST_COND_SWZ_Y_SHIFT 7 +#define NV30_VP_INST_COND_SWZ_Y_MASK (0x03 << 7) +#define NV30_VP_INST_COND_SWZ_Z_SHIFT 5 +#define NV30_VP_INST_COND_SWZ_Z_MASK (0x03 << 5) +#define NV30_VP_INST_COND_SWZ_W_SHIFT 3 +#define NV30_VP_INST_COND_SWZ_W_MASK (0x03 << 3) +#define NV30_VP_INST_COND_SWZ_ALL_SHIFT 3 +#define NV30_VP_INST_COND_SWZ_ALL_MASK (0xFF << 3) +#define NV30_VP_INST_ADDR_SWZ_SHIFT 1 +#define NV30_VP_INST_ADDR_SWZ_MASK (0x03 << 1) +#define NV30_VP_INST_SCA_OPCODEH_SHIFT 0 +#define NV30_VP_INST_SCA_OPCODEH_MASK (0x01 << 0) + +/* DWORD 1 */ +#define NV30_VP_INST_SCA_OPCODEL_SHIFT 28 +#define NV30_VP_INST_SCA_OPCODEL_MASK (0x0F << 28) +# define NV30_VP_INST_OP_NOP 0x00 +# define NV30_VP_INST_OP_RCP 0x02 +# define NV30_VP_INST_OP_RCC 0x03 +# define NV30_VP_INST_OP_RSQ 0x04 +# define NV30_VP_INST_OP_EXP 0x05 +# define NV30_VP_INST_OP_LOG 0x06 +# define NV30_VP_INST_OP_LIT 0x07 +# define NV30_VP_INST_OP_BRA 0x09 +# define NV30_VP_INST_OP_CAL 0x0B +# define NV30_VP_INST_OP_RET 0x0C +# define NV30_VP_INST_OP_LG2 0x0D +# define NV30_VP_INST_OP_EX2 0x0E +# define NV30_VP_INST_OP_SIN 0x0F +# define NV30_VP_INST_OP_COS 0x10 +#define NV30_VP_INST_VEC_OPCODE_SHIFT 23 +#define NV30_VP_INST_VEC_OPCODE_MASK (0x1F << 23) +# define NV30_VP_INST_OP_NOPV 0x00 +# define NV30_VP_INST_OP_MOV 0x01 +# define NV30_VP_INST_OP_MUL 0x02 +# define NV30_VP_INST_OP_ADD 0x03 +# define NV30_VP_INST_OP_MAD 0x04 +# define NV30_VP_INST_OP_DP3 0x05 +# define NV30_VP_INST_OP_DP4 0x07 +# define NV30_VP_INST_OP_DPH 0x06 +# define NV30_VP_INST_OP_DST 0x08 +# define NV30_VP_INST_OP_MIN 0x09 +# define NV30_VP_INST_OP_MAX 0x0A +# define NV30_VP_INST_OP_SLT 0x0B +# define NV30_VP_INST_OP_SGE 0x0C +# define NV30_VP_INST_OP_ARL 0x0D +# define NV30_VP_INST_OP_FRC 0x0E +# define NV30_VP_INST_OP_FLR 0x0F +# define NV30_VP_INST_OP_SEQ 0x10 +# define NV30_VP_INST_OP_SFL 0x11 +# define NV30_VP_INST_OP_SGT 0x12 +# define NV30_VP_INST_OP_SLE 0x13 +# define NV30_VP_INST_OP_SNE 0x14 +# define NV30_VP_INST_OP_STR 0x15 +# define NV30_VP_INST_OP_SSG 0x16 +# define NV30_VP_INST_OP_ARR 0x17 +# define NV30_VP_INST_OP_ARA 0x18 +#define NV30_VP_INST_CONST_SRC_SHIFT 14 +#define NV30_VP_INST_CONST_SRC_MASK (0xFF << 14) +#define NV30_VP_INST_INPUT_SRC_SHIFT 9 /*NV20*/ +#define NV30_VP_INST_INPUT_SRC_MASK (0x0F << 9) /*NV20*/ +# define NV30_VP_INST_IN_POS 0 /* These seem to match the bindings specified in */ +# define NV30_VP_INST_IN_WEIGHT 1 /* the ARB_v_p spec (2.14.3.1) */ +# define NV30_VP_INST_IN_NORMAL 2 +# define NV30_VP_INST_IN_COL0 3 /* Should probably confirm them all though */ +# define NV30_VP_INST_IN_COL1 4 +# define NV30_VP_INST_IN_FOGC 5 +# define NV30_VP_INST_IN_TC0 8 +# define NV30_VP_INST_IN_TC(n) (8+n) +#define NV30_VP_INST_SRC0H_SHIFT 0 /*NV20*/ +#define NV30_VP_INST_SRC0H_MASK (0x1FF << 0) /*NV20*/ + +/* Please note: the IADDR fields overlap other fields because they are used + * only for branch instructions. See Branching: label above + * + * DWORD 2 + */ +#define NV30_VP_INST_SRC0L_SHIFT 26 /*NV20*/ +#define NV30_VP_INST_SRC0L_MASK (0x3F <<26) /* NV30_VP_SRC0_LOW_MASK << 26 */ +#define NV30_VP_INST_SRC1_SHIFT 11 /*NV20*/ +#define NV30_VP_INST_SRC1_MASK (0x7FFF<<11) /*NV20*/ +#define NV30_VP_INST_SRC2H_SHIFT 0 /*NV20*/ +#define NV30_VP_INST_SRC2H_MASK (0x7FF << 0) /* NV30_VP_SRC2_HIGH_MASK >> 4*/ +#define NV30_VP_INST_IADDR_SHIFT 2 +#define NV30_VP_INST_IADDR_MASK (0xF << 28) /* NV30_VP_SRC2_LOW_MASK << 28 */ + +/* DWORD 3 */ +#define NV30_VP_INST_SRC2L_SHIFT 28 /*NV20*/ +#define NV30_VP_INST_SRC2L_MASK (0x0F <<28) /*NV20*/ +#define NV30_VP_INST_STEMP_WRITEMASK_SHIFT 24 +#define NV30_VP_INST_STEMP_WRITEMASK_MASK (0x0F << 24) +#define NV30_VP_INST_VTEMP_WRITEMASK_SHIFT 20 +#define NV30_VP_INST_VTEMP_WRITEMASK_MASK (0x0F << 20) +#define NV30_VP_INST_SDEST_WRITEMASK_SHIFT 16 +#define NV30_VP_INST_SDEST_WRITEMASK_MASK (0x0F << 16) +#define NV30_VP_INST_VDEST_WRITEMASK_SHIFT 12 /*NV20*/ +#define NV30_VP_INST_VDEST_WRITEMASK_MASK (0x0F << 12) /*NV20*/ +#define NV30_VP_INST_DEST_SHIFT 2 +#define NV30_VP_INST_DEST_MASK (0x0F << 2) +# define NV30_VP_INST_DEST_POS 0 +# define NV30_VP_INST_DEST_BFC0 1 +# define NV30_VP_INST_DEST_BFC1 2 +# define NV30_VP_INST_DEST_COL0 3 +# define NV30_VP_INST_DEST_COL1 4 +# define NV30_VP_INST_DEST_FOGC 5 +# define NV30_VP_INST_DEST_PSZ 6 +# define NV30_VP_INST_DEST_TC(n) (8+n) + +#define NV30_VP_INST_LAST (1 << 0) + +/* Useful to split the source selection regs into their pieces */ +#define NV30_VP_SRC0_HIGH_SHIFT 6 +#define NV30_VP_SRC0_HIGH_MASK 0x00007FC0 +#define NV30_VP_SRC0_LOW_MASK 0x0000003F +#define NV30_VP_SRC2_HIGH_SHIFT 4 +#define NV30_VP_SRC2_HIGH_MASK 0x00007FF0 +#define NV30_VP_SRC2_LOW_MASK 0x0000000F + + +/* Source-register definition - matches NV20 exactly */ +#define NV30_VP_SRC_NEGATE (1<<14) +#define NV30_VP_SRC_SWZ_X_SHIFT 12 +#define NV30_VP_SRC_REG_SWZ_X_MASK (0x03 <<12) +#define NV30_VP_SRC_SWZ_Y_SHIFT 10 +#define NV30_VP_SRC_REG_SWZ_Y_MASK (0x03 <<10) +#define NV30_VP_SRC_SWZ_Z_SHIFT 8 +#define NV30_VP_SRC_REG_SWZ_Z_MASK (0x03 << 8) +#define NV30_VP_SRC_SWZ_W_SHIFT 6 +#define NV30_VP_SRC_REG_SWZ_W_MASK (0x03 << 6) +#define NV30_VP_SRC_REG_SWZ_ALL_SHIFT 6 +#define NV30_VP_SRC_REG_SWZ_ALL_MASK (0xFF << 6) +#define NV30_VP_SRC_TEMP_SRC_SHIFT 2 +#define NV30_VP_SRC_REG_TEMP_ID_MASK (0x0F << 0) +#define NV30_VP_SRC_REG_TYPE_SHIFT 0 +#define NV30_VP_SRC_REG_TYPE_MASK (0x03 << 0) +#define NV30_VP_SRC_REG_TYPE_TEMP 1 +#define NV30_VP_SRC_REG_TYPE_INPUT 2 +#define NV30_VP_SRC_REG_TYPE_CONST 3 /* guess */ + +/* + * Each fragment program opcode appears to be comprised of 4 32-bit values. + * + * 0 - Opcode, output reg/mask, ATTRIB source + * 1 - Source 0 + * 2 - Source 1 + * 3 - Source 2 + * + * There appears to be no special difference between result regs and temp regs. + * result.color == R0.xyzw + * result.depth == R1.z + * When the fragprog contains instructions to write depth, NV30_TCL_PRIMITIVE_3D_UNK1D78=0 + * otherwise it is set to 1. + * + * Constants are inserted directly after the instruction that uses them. + * + * It appears that it's not possible to use two input registers in one + * instruction as the input sourcing is done in the instruction dword + * and not the source selection dwords. As such instructions such as: + * + * ADD result.color, fragment.color, fragment.texcoord[0]; + * + * must be split into two MOV's and then an ADD (nvidia does this) but + * I'm not sure why it's not just one MOV and then source the second input + * in the ADD instruction.. + * + * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary + * negation requires multiplication with a const. + * + * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO/SWIZZLE_ONE + * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as SWIZZLE_ZERO + * is implemented simply by not writing to the relevant components of the destination. + * + * Conditional execution + * TODO + * + * Non-native instructions: + * LIT + * LRP - MAD+MAD + * SUB - ADD, negate second source + * RSQ - LG2 + EX2 + * POW - LG2 + MUL + EX2 + * SCS - COS + SIN + * XPD + */ + +//== Opcode / Destination selection == +#define NV30_FP_OP_PROGRAM_END (1 << 0) +#define NV30_FP_OP_OUT_REG_SHIFT 1 +#define NV30_FP_OP_OUT_REG_MASK (31 << 1) /* uncertain */ +/* Needs to be set when writing outputs to get expected result.. */ +#define NV30_FP_OP_OUT_REG_HALF (1 << 7) +#define NV30_FP_OP_COND_WRITE_ENABLE (1 << 8) +#define NV30_FP_OP_OUTMASK_SHIFT 9 +#define NV30_FP_OP_OUTMASK_MASK (0xF << 9) +# define NV30_FP_OP_OUT_X (1<<9) +# define NV30_FP_OP_OUT_Y (1<<10) +# define NV30_FP_OP_OUT_Z (1<<11) +# define NV30_FP_OP_OUT_W (1<<12) +/* Uncertain about these, especially the input_src values.. it's possible that + * they can be dynamically changed. + */ +#define NV30_FP_OP_INPUT_SRC_SHIFT 13 +#define NV30_FP_OP_INPUT_SRC_MASK (15 << 13) +# define NV30_FP_OP_INPUT_SRC_POSITION 0x0 +# define NV30_FP_OP_INPUT_SRC_COL0 0x1 +# define NV30_FP_OP_INPUT_SRC_COL1 0x2 +# define NV30_FP_OP_INPUT_SRC_FOGC 0x3 +# define NV30_FP_OP_INPUT_SRC_TC0 0x4 +# define NV30_FP_OP_INPUT_SRC_TC(n) (0x4 + n) +#define NV30_FP_OP_TEX_UNIT_SHIFT 17 +#define NV30_FP_OP_TEX_UNIT_MASK (0xF << 17) /* guess */ +#define NV30_FP_OP_PRECISION_SHIFT 22 +#define NV30_FP_OP_PRECISION_MASK (3 << 22) +# define NV30_FP_PRECISION_FP32 0 +# define NV30_FP_PRECISION_FP16 1 +# define NV30_FP_PRECISION_FX12 2 +#define NV30_FP_OP_OPCODE_SHIFT 24 +#define NV30_FP_OP_OPCODE_MASK (0x3F << 24) +# define NV30_FP_OP_OPCODE_NOP 0x00 +# define NV30_FP_OP_OPCODE_MOV 0x01 +# define NV30_FP_OP_OPCODE_MUL 0x02 +# define NV30_FP_OP_OPCODE_ADD 0x03 +# define NV30_FP_OP_OPCODE_MAD 0x04 +# define NV30_FP_OP_OPCODE_DP3 0x05 +# define NV30_FP_OP_OPCODE_DP4 0x06 +# define NV30_FP_OP_OPCODE_DST 0x07 +# define NV30_FP_OP_OPCODE_MIN 0x08 +# define NV30_FP_OP_OPCODE_MAX 0x09 +# define NV30_FP_OP_OPCODE_SLT 0x0A +# define NV30_FP_OP_OPCODE_SGE 0x0B +# define NV30_FP_OP_OPCODE_SLE 0x0C +# define NV30_FP_OP_OPCODE_SGT 0x0D +# define NV30_FP_OP_OPCODE_SNE 0x0E +# define NV30_FP_OP_OPCODE_SEQ 0x0F +# define NV30_FP_OP_OPCODE_FRC 0x10 +# define NV30_FP_OP_OPCODE_FLR 0x11 +# define NV30_FP_OP_OPCODE_KIL 0x12 +# define NV30_FP_OP_OPCODE_PK4B 0x13 +# define NV30_FP_OP_OPCODE_UP4B 0x14 +# define NV30_FP_OP_OPCODE_DDX 0x15 /* can only write XY */ +# define NV30_FP_OP_OPCODE_DDY 0x16 /* can only write XY */ +# define NV30_FP_OP_OPCODE_TEX 0x17 +# define NV30_FP_OP_OPCODE_TXP 0x18 +# define NV30_FP_OP_OPCODE_TXD 0x19 +# define NV30_FP_OP_OPCODE_RCP 0x1A +# define NV30_FP_OP_OPCODE_RSQ 0x1B +# define NV30_FP_OP_OPCODE_EX2 0x1C +# define NV30_FP_OP_OPCODE_LG2 0x1D +# define NV30_FP_OP_OPCODE_LIT 0x1E +# define NV30_FP_OP_OPCODE_LRP 0x1F +# define NV30_FP_OP_OPCODE_STR 0x20 +# define NV30_FP_OP_OPCODE_SFL 0x21 +# define NV30_FP_OP_OPCODE_COS 0x22 +# define NV30_FP_OP_OPCODE_SIN 0x23 +# define NV30_FP_OP_OPCODE_PK2H 0x24 +# define NV30_FP_OP_OPCODE_UP2H 0x25 +# define NV30_FP_OP_OPCODE_POW 0x26 +# define NV30_FP_OP_OPCODE_PK4UB 0x27 +# define NV30_FP_OP_OPCODE_UP4UB 0x28 +# define NV30_FP_OP_OPCODE_PK2US 0x29 +# define NV30_FP_OP_OPCODE_UP2US 0x2A +# define NV30_FP_OP_OPCODE_DP2A 0x2E +# define NV30_FP_OP_OPCODE_TXB 0x31 +# define NV30_FP_OP_OPCODE_RFL 0x36 +# define NV30_FP_OP_OPCODE_DIV 0x3A +#define NV30_FP_OP_OUT_SAT (1 << 31) + +/* high order bits of SRC0 */ +#define NV30_FP_OP_OUT_ABS (1 << 29) +#define NV30_FP_OP_COND_SWZ_W_SHIFT 27 +#define NV30_FP_OP_COND_SWZ_W_MASK (3 << 27) +#define NV30_FP_OP_COND_SWZ_Z_SHIFT 25 +#define NV30_FP_OP_COND_SWZ_Z_MASK (3 << 25) +#define NV30_FP_OP_COND_SWZ_Y_SHIFT 23 +#define NV30_FP_OP_COND_SWZ_Y_MASK (3 << 23) +#define NV30_FP_OP_COND_SWZ_X_SHIFT 21 +#define NV30_FP_OP_COND_SWZ_X_MASK (3 << 21) +#define NV30_FP_OP_COND_SWZ_ALL_SHIFT 21 +#define NV30_FP_OP_COND_SWZ_ALL_MASK (0xFF << 21) +#define NV30_FP_OP_COND_SHIFT 18 +#define NV30_FP_OP_COND_MASK (0x07 << 18) +# define NV30_FP_OP_COND_FL 0 +# define NV30_FP_OP_COND_LT 1 +# define NV30_FP_OP_COND_EQ 2 +# define NV30_FP_OP_COND_LE 3 +# define NV30_FP_OP_COND_GT 4 +# define NV30_FP_OP_COND_NE 5 +# define NV30_FP_OP_COND_GE 6 +# define NV30_FP_OP_COND_TR 7 + +/* high order bits of SRC1 */ +#define NV30_FP_OP_DST_SCALE_SHIFT 28 +#define NV30_FP_OP_DST_SCALE_MASK (3 << 28) +#define NV30_FP_OP_DST_SCALE_1X 0 +#define NV30_FP_OP_DST_SCALE_2X 1 +#define NV30_FP_OP_DST_SCALE_4X 2 +#define NV30_FP_OP_DST_SCALE_8X 3 +#define NV30_FP_OP_DST_SCALE_INV_2X 5 +#define NV30_FP_OP_DST_SCALE_INV_4X 6 +#define NV30_FP_OP_DST_SCALE_INV_8X 7 + + +/* high order bits of SRC2 */ +#define NV30_FP_OP_INDEX_INPUT (1 << 30) + +//== Register selection == +#define NV30_FP_REG_TYPE_SHIFT 0 +#define NV30_FP_REG_TYPE_MASK (3 << 0) +# define NV30_FP_REG_TYPE_TEMP 0 +# define NV30_FP_REG_TYPE_INPUT 1 +# define NV30_FP_REG_TYPE_CONST 2 +#define NV30_FP_REG_SRC_SHIFT 2 /* uncertain */ +#define NV30_FP_REG_SRC_MASK (31 << 2) +#define NV30_FP_REG_SRC_HALF (1 << 8) +#define NV30_FP_REG_SWZ_ALL_SHIFT 9 +#define NV30_FP_REG_SWZ_ALL_MASK (255 << 9) +#define NV30_FP_REG_SWZ_X_SHIFT 9 +#define NV30_FP_REG_SWZ_X_MASK (3 << 9) +#define NV30_FP_REG_SWZ_Y_SHIFT 11 +#define NV30_FP_REG_SWZ_Y_MASK (3 << 11) +#define NV30_FP_REG_SWZ_Z_SHIFT 13 +#define NV30_FP_REG_SWZ_Z_MASK (3 << 13) +#define NV30_FP_REG_SWZ_W_SHIFT 15 +#define NV30_FP_REG_SWZ_W_MASK (3 << 15) +# define NV30_FP_SWIZZLE_X 0 +# define NV30_FP_SWIZZLE_Y 1 +# define NV30_FP_SWIZZLE_Z 2 +# define NV30_FP_SWIZZLE_W 3 +#define NV30_FP_REG_NEGATE (1 << 17) + +#define NV30SR_NONE 0 +#define NV30SR_OUTPUT 1 +#define NV30SR_INPUT 2 +#define NV30SR_TEMP 3 +#define NV30SR_CONST 4 + +struct nv30_sreg { + int type; + int index; + + int dst_scale; + + int negate; + int abs; + int swz[4]; + + int cc_update; + int cc_update_reg; + int cc_test; + int cc_test_reg; + int cc_swz[4]; +}; + +static INLINE struct nv30_sreg +nv30_sr(int type, int index) +{ + struct nv30_sreg temp = { + .type = type, + .index = index, + .dst_scale = DEF_SCALE, + .abs = 0, + .negate = 0, + .swz = { 0, 1, 2, 3 }, + .cc_update = 0, + .cc_update_reg = 0, + .cc_test = DEF_CTEST, + .cc_test_reg = 0, + .cc_swz = { 0, 1, 2, 3 }, + }; + return temp; +} + +static INLINE struct nv30_sreg +nv30_sr_swz(struct nv30_sreg src, int x, int y, int z, int w) +{ + struct nv30_sreg dst = src; + + dst.swz[SWZ_X] = src.swz[x]; + dst.swz[SWZ_Y] = src.swz[y]; + dst.swz[SWZ_Z] = src.swz[z]; + dst.swz[SWZ_W] = src.swz[w]; + return dst; +} + +static INLINE struct nv30_sreg +nv30_sr_neg(struct nv30_sreg src) +{ + src.negate = !src.negate; + return src; +} + +static INLINE struct nv30_sreg +nv30_sr_abs(struct nv30_sreg src) +{ + src.abs = 1; + return src; +} + +static INLINE struct nv30_sreg +nv30_sr_scale(struct nv30_sreg src, int scale) +{ + src.dst_scale = scale; + return src; +} + +#endif diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c new file mode 100644 index 00000000000..53368561e07 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -0,0 +1,739 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +static void * +nv30_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv30_blend_state *cb; + + cb = malloc(sizeof(struct nv30_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | + (nvgl_blend_eqn(cso->rgb_func))); + + cb->l_enable = cso->logicop_enable ? 1 : 0; + cb->l_op = nvgl_logicop_func(cso->logicop_func); + + cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + + cb->d_enable = cso->dither ? 1 : 0; + + return (void *)cb; +} + +static void +nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_blend_state *cb = hwcso; + + BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); + OUT_RING (cb->d_enable); + + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 3); + OUT_RING (cb->b_enable); + OUT_RING (cb->b_srcfunc); + OUT_RING (cb->b_dstfunc); + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_EQUATION, 1); + OUT_RING (cb->b_eqn); + + BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); + OUT_RING (cb->c_mask); + + BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); + OUT_RING (cb->l_enable); + OUT_RING (cb->l_op); +} + +static void +nv30_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV34TCL_TX_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV34TCL_TX_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV34TCL_TX_WRAP_S_CLAMP; + break; +/* case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP; + break;*/ + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV34TCL_TX_WRAP_S_REPEAT; + break; + } + + return ret >> NV34TCL_TX_WRAP_S_SHIFT; +} + +static void * +nv30_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv30_sampler_state *ps; + uint32_t filter = 0; + + ps = malloc(sizeof(struct nv30_sampler_state)); + + ps->fmt = 0; + if (!cso->normalized_coords) + ps->fmt |= NV34TCL_TX_FORMAT_RECT; + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV34TCL_TX_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV34TCL_TX_WRAP_T_SHIFT) | + (wrap_mode(cso->wrap_r) << NV34TCL_TX_WRAP_R_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy >= 2.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + +/* if (cso->max_anisotropy >= 16.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_16X; + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_12X; + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_10X; + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_6X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_4X; + } else { + ps->en |= NV34TCL_TX_ENABLE_ANISO_2X; + }*/ + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV34TCL_TX_FILTER_MAGNIFY_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV34TCL_TX_FILTER_MAGNIFY_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST; + break; + } + break; + } + + ps->filt = filter; + +/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NEVER; + break; + case PIPE_FUNC_GREATER: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GREATER; + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_EQUAL; + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GEQUAL; + break; + case PIPE_FUNC_LESS: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LESS; + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NOTEQUAL; + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LEQUAL; + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= NV34TCL_TX_WRAP_RCOMP_ALWAYS; + break; + default: + break; + } + }*/ + + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv30_sampler_state_bind(struct pipe_context *pipe, unsigned unit, + void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_sampler_state *ps = hwcso; + + nv30->tex_sampler[unit] = ps; + nv30->dirty_samplers |= (1 << unit); +} + +static void +nv30_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv30_set_sampler_texture(struct pipe_context *pipe, unsigned unit, + struct pipe_texture *miptree) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->tex_miptree[unit] = (struct nv30_miptree *)miptree; + nv30->dirty_samplers |= (1 << unit); +} + +static void * +nv30_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv30_rasterizer_state *rs; + int i; + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + rs = malloc(sizeof(struct nv30_rasterizer_state)); + + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; + + rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; + rs->line_smooth_en = cso->line_smooth ? 1 : 0; + rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; + rs->line_stipple = (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor; + + rs->point_size = *(uint32_t*)&cso->point_size; + + rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; + rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; + + if (cso->front_winding == PIPE_WINDING_CCW) { + rs->front_face = NV34TCL_FRONT_FACE_CCW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + } else { + rs->front_face = NV34TCL_FRONT_FACE_CW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + } + + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CCW) + rs->cull_face = NV34TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV34TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_CW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CW) + rs->cull_face = NV34TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV34TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_BOTH: + rs->cull_face_en = 1; + rs->cull_face = NV34TCL_CULL_FACE_FRONT_AND_BACK; + break; + case PIPE_WINDING_NONE: + default: + rs->cull_face_en = 0; + rs->cull_face = 0; + break; + } + + if (cso->point_sprite) { + rs->point_sprite = (1 << 0); + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + rs->point_sprite |= (1 << (8 + i)); + } + } else { + rs->point_sprite = 0; + } + + return (void *)rs; +} + +static void +nv30_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_rasterizer_state *rs = hwcso; + + BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); + OUT_RING (rs->shade_model); + + BEGIN_RING(rankine, NV34TCL_LINE_WIDTH, 2); + OUT_RING (rs->line_width); + OUT_RING (rs->line_smooth_en); + BEGIN_RING(rankine, NV34TCL_LINE_STIPPLE_ENABLE, 2); + OUT_RING (rs->line_stipple_en); + OUT_RING (rs->line_stipple); + + BEGIN_RING(rankine, NV34TCL_POINT_SIZE, 1); + OUT_RING (rs->point_size); + + BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 6); + OUT_RING (rs->poly_mode_front); + OUT_RING (rs->poly_mode_back); + OUT_RING (rs->cull_face); + OUT_RING (rs->front_face); + OUT_RING (rs->poly_smooth_en); + OUT_RING (rs->cull_face_en); + + BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); + OUT_RING (rs->poly_stipple_en); + + BEGIN_RING(rankine, NV34TCL_POINT_SPRITE, 1); + OUT_RING (rs->point_sprite); +} + +static void +nv30_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv30_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, + unsigned idx, struct nv30_stencil_push *hw) +{ + hw->enable = cso->stencil[idx].enabled ? 1 : 0; + hw->wmask = cso->stencil[idx].write_mask; + hw->func = nvgl_comparison_op(cso->stencil[idx].func); + hw->ref = cso->stencil[idx].ref_value; + hw->vmask = cso->stencil[idx].value_mask; + hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); + hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); + hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); +} + +static void * +nv30_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv30_depth_stencil_alpha_state *hw; + + hw = malloc(sizeof(struct nv30_depth_stencil_alpha_state)); + + hw->depth.func = nvgl_comparison_op(cso->depth.func); + hw->depth.write_enable = cso->depth.writemask ? 1 : 0; + hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + + nv30_translate_stencil(cso, 0, &hw->stencil.front); + nv30_translate_stencil(cso, 1, &hw->stencil.back); + + hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; + hw->alpha.func = nvgl_comparison_op(cso->alpha.func); + hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + + return (void *)hw; +} + +static void +nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_depth_stencil_alpha_state *hw = hwcso; + + BEGIN_RING(rankine, NV34TCL_DEPTH_FUNC, 3); + OUT_RINGp ((uint32_t *)&hw->depth, 3); + BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 16); + OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); + OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); + BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3); + OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); +} + +static void +nv30_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv30_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv30_vertex_program *vp; + + vp = CALLOC(1, sizeof(struct nv30_vertex_program)); + vp->pipe = cso; + + return (void *)vp; +} + +static void +nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_vertex_program *vp = hwcso; + + nv30->vertprog.current = vp; + nv30->dirty |= NV30_NEW_VERTPROG; +} + +static void +nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_vertex_program *vp = hwcso; + + nv30_vertprog_destroy(nv30, vp); + free(vp); +} + +static void * +nv30_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv30_fragment_program *fp; + + fp = CALLOC(1, sizeof(struct nv30_fragment_program)); + fp->pipe = cso; + + return (void *)fp; +} + +static void +nv30_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_fragment_program *fp = hwcso; + + nv30->fragprog.current = fp; + nv30->dirty |= NV30_NEW_FRAGPROG; +} + +static void +nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_fragment_program *fp = hwcso; + + nv30_fragprog_destroy(nv30, fp); + free(fp); +} + +static void +nv30_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_COLOR, 1); + OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0)); +} + +static void +nv30_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv30_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + if (shader == PIPE_SHADER_VERTEX) { + nv30->vertprog.constant_buf = buf->buffer; + nv30->dirty |= NV30_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv30->fragprog.constant_buf = buf->buffer; + nv30->dirty |= NV30_NEW_FRAGPROG; + } +} + +static void +nv30_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct pipe_surface *rt[4], *zeta; + uint32_t rt_enable, rt_format, w, h; + int i, colour_format = 0, zeta_format = 0; + + rt_enable = 0; + for (i = 0; i < 4; i++) { + if (!fb->cbufs[i]) + continue; + + if (colour_format) { + assert(w == fb->cbufs[i]->width); + assert(h == fb->cbufs[i]->height); + assert(colour_format == fb->cbufs[i]->format); + } else { + w = fb->cbufs[i]->width; + h = fb->cbufs[i]->height; + colour_format = fb->cbufs[i]->format; + rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i); + rt[i] = fb->cbufs[i]; + } + } + + if (rt_enable & (NV34TCL_RT_ENABLE_COLOR1 | NV34TCL_RT_ENABLE_COLOR2 | + NV34TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV34TCL_RT_ENABLE_MRT; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV34TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + switch (zeta_format) { + case PIPE_FORMAT_Z16_UNORM: + rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16; + break; + case PIPE_FORMAT_Z24S8_UNORM: + case 0: + rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8; + break; + default: + assert(0); + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { + BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); + OUT_RING ( (rt[0]->pitch * rt[0]->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + nv30->rt[0] = rt[0]->buffer; + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { + BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 2); + OUT_RING (rt[1]->pitch * rt[1]->cpp); + nv30->rt[1] = rt[1]->buffer; + } + + if (zeta_format) + { + nv30->zeta = zeta->buffer; + } + + nv30->rt_enable = rt_enable; + BEGIN_RING(rankine, NV34TCL_RT_ENABLE, 1); + OUT_RING (rt_enable); + BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 3); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((w - 1) << 16) | 0); + OUT_RING (((h - 1) << 16) | 0); +} + +static void +nv30_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32); + OUT_RINGp ((uint32_t *)stipple->stipple, 32); +} + +static void +nv30_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny); +} + +static void +nv30_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); + OUT_RINGf (vpt->translate[0]); + OUT_RINGf (vpt->translate[1]); + OUT_RINGf (vpt->translate[2]); + OUT_RINGf (vpt->translate[3]); + OUT_RINGf (vpt->scale[0]); + OUT_RINGf (vpt->scale[1]); + OUT_RINGf (vpt->scale[2]); + OUT_RINGf (vpt->scale[3]); +} + +static void +nv30_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->vtxbuf[index] = *vb; + + nv30->dirty |= NV30_NEW_ARRAYS; +} + +static void +nv30_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->vtxelt[index] = *ve; + + nv30->dirty |= NV30_NEW_ARRAYS; +} + +void +nv30_init_state_functions(struct nv30_context *nv30) +{ + nv30->pipe.create_blend_state = nv30_blend_state_create; + nv30->pipe.bind_blend_state = nv30_blend_state_bind; + nv30->pipe.delete_blend_state = nv30_blend_state_delete; + + nv30->pipe.create_sampler_state = nv30_sampler_state_create; + nv30->pipe.bind_sampler_state = nv30_sampler_state_bind; + nv30->pipe.delete_sampler_state = nv30_sampler_state_delete; + nv30->pipe.set_sampler_texture = nv30_set_sampler_texture; + + nv30->pipe.create_rasterizer_state = nv30_rasterizer_state_create; + nv30->pipe.bind_rasterizer_state = nv30_rasterizer_state_bind; + nv30->pipe.delete_rasterizer_state = nv30_rasterizer_state_delete; + + nv30->pipe.create_depth_stencil_alpha_state = + nv30_depth_stencil_alpha_state_create; + nv30->pipe.bind_depth_stencil_alpha_state = + nv30_depth_stencil_alpha_state_bind; + nv30->pipe.delete_depth_stencil_alpha_state = + nv30_depth_stencil_alpha_state_delete; + + nv30->pipe.create_vs_state = nv30_vp_state_create; + nv30->pipe.bind_vs_state = nv30_vp_state_bind; + nv30->pipe.delete_vs_state = nv30_vp_state_delete; + + nv30->pipe.create_fs_state = nv30_fp_state_create; + nv30->pipe.bind_fs_state = nv30_fp_state_bind; + nv30->pipe.delete_fs_state = nv30_fp_state_delete; + + nv30->pipe.set_blend_color = nv30_set_blend_color; + nv30->pipe.set_clip_state = nv30_set_clip_state; + nv30->pipe.set_constant_buffer = nv30_set_constant_buffer; + nv30->pipe.set_framebuffer_state = nv30_set_framebuffer_state; + nv30->pipe.set_polygon_stipple = nv30_set_polygon_stipple; + nv30->pipe.set_scissor_state = nv30_set_scissor_state; + nv30->pipe.set_viewport_state = nv30_set_viewport_state; + + nv30->pipe.set_vertex_buffer = nv30_set_vertex_buffer; + nv30->pipe.set_vertex_element = nv30_set_vertex_element; +} + diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h new file mode 100644 index 00000000000..233600f69ab --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -0,0 +1,147 @@ +#ifndef __NV30_STATE_H__ +#define __NV30_STATE_H__ + +#include "pipe/p_state.h" + +struct nv30_blend_state { + uint32_t b_enable; + uint32_t b_srcfunc; + uint32_t b_dstfunc; + uint32_t b_eqn; + + uint32_t l_enable; + uint32_t l_op; + + uint32_t c_mask; + + uint32_t d_enable; +}; + +struct nv30_sampler_state { + uint32_t fmt; + uint32_t wrap; + uint32_t en; + uint32_t filt; + uint32_t bcol; +}; + +struct nv30_rasterizer_state { + uint32_t shade_model; + + uint32_t line_width; + uint32_t line_smooth_en; + uint32_t line_stipple_en; + uint32_t line_stipple; + + uint32_t point_size; + + uint32_t poly_smooth_en; + uint32_t poly_stipple_en; + + uint32_t poly_mode_front; + uint32_t poly_mode_back; + + uint32_t front_face; + uint32_t cull_face; + uint32_t cull_face_en; + + uint32_t point_sprite; +}; + +struct nv30_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv30_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + +struct nv30_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + struct nv30_vertex_program_exec *insns; + unsigned nr_insns; + struct nv30_vertex_program_data *consts; + unsigned nr_consts; + + struct nouveau_resource *exec; + unsigned exec_start; + struct nouveau_resource *data; + unsigned data_start; + unsigned data_start_min; + + uint32_t ir; + uint32_t or; +}; + +struct nv30_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv30_fragment_program { + const struct pipe_shader_state *pipe; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv30_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + uint32_t fp_reg_control; +}; + +struct nv30_stencil_push { + uint32_t enable; + uint32_t wmask; + uint32_t func; + uint32_t ref; + uint32_t vmask; + uint32_t fail; + uint32_t zfail; + uint32_t zpass; +}; + +struct nv30_depth_stencil_alpha_state { + struct { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; + } depth; + + struct { + struct nv30_stencil_push back; + struct nv30_stencil_push front; + } stencil; + + struct { + uint32_t enabled; + uint32_t func; + uint32_t ref; + } alpha; +}; + +struct nv30_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +#endif diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c new file mode 100644 index 00000000000..70b98836f0f --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -0,0 +1,83 @@ +#include "nv30_context.h" +#include "nv30_state.h" + +void +nv30_emit_hw_state(struct nv30_context *nv30) +{ + int i; + + if (nv30->dirty & NV30_NEW_FRAGPROG) { + nv30_fragprog_bind(nv30, nv30->fragprog.current); + /*XXX: clear NV30_NEW_FRAGPROG if no new program uploaded */ + } + + if (nv30->dirty_samplers || (nv30->dirty & NV30_NEW_FRAGPROG)) { + nv30_fragtex_bind(nv30); +/* + BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); + OUT_RING (1);*/ + nv30->dirty &= ~NV30_NEW_FRAGPROG; + } + + if (nv30->dirty & NV30_NEW_VERTPROG) { + nv30_vertprog_bind(nv30, nv30->vertprog.current); + nv30->dirty &= ~NV30_NEW_VERTPROG; + } + + nv30->dirty_samplers = 0; + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv30_vbo.c + */ + + /* Render targets */ + if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR0) { + BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 1); + OUT_RELOCo(nv30->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(rankine, NV34TCL_COLOR0_OFFSET, 1); + OUT_RELOCl(nv30->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR1) { + BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); + OUT_RELOCo(nv30->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(rankine, NV34TCL_COLOR1_OFFSET, 1); + OUT_RELOCl(nv30->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } + + if (nv30->zeta) { + BEGIN_RING(rankine, NV34TCL_DMA_ZETA, 1); + OUT_RELOCo(nv30->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(rankine, NV34TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv30->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + /* XXX allocate LMA */ +/* BEGIN_RING(rankine, NV34TCL_LMA_DEPTH_OFFSET, 1); + OUT_RING(0);*/ + } + + /* Texture images */ + for (i = 0; i < 16; i++) { + if (!(nv30->fp_samplers & (1 << i))) + continue; + BEGIN_RING(rankine, NV34TCL_TX_OFFSET(i), 2); + OUT_RELOCl(nv30->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv30->tex[i].buffer, nv30->tex[i].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV34TCL_TX_FORMAT_DMA0, + NV34TCL_TX_FORMAT_DMA1); + } + + /* Fragment program */ + BEGIN_RING(rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); + OUT_RELOC (nv30->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR, NV34TCL_FP_ACTIVE_PROGRAM_DMA0, + NV34TCL_FP_ACTIVE_PROGRAM_DMA1); +} + diff --git a/src/gallium/drivers/nv30/nv30_surface.c b/src/gallium/drivers/nv30/nv30_surface.c new file mode 100644 index 00000000000..974965679f5 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_surface.c @@ -0,0 +1,137 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv30_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/p_tile.h" + +static boolean +nv30_surface_format_supported(struct pipe_context *pipe, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static struct pipe_surface * +nv30_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv30mt->level[level].pitch / ps->cpp; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv30mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv30mt->level[level].image_offset[zslice]; + } else { + ps->offset = nv30mt->level[level].image_offset[0]; + } + + return ps; +} + +static void +nv30_surface_copy(struct pipe_context *pipe, unsigned do_flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv30_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_winsys *nvws = nv30->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv30_init_surface_functions(struct nv30_context *nv30) +{ + nv30->pipe.is_format_supported = nv30_surface_format_supported; + nv30->pipe.get_tex_surface = nv30_get_tex_surface; + nv30->pipe.surface_copy = nv30_surface_copy; + nv30->pipe.surface_fill = nv30_surface_fill; +} diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c new file mode 100644 index 00000000000..9e00cdac3fb --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -0,0 +1,425 @@ +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_pushbuf.h" + +static INLINE int +nv30_vbo_ncomp(uint format) +{ + int ncomp = 0; + + if (pf_size_x(format)) ncomp++; + if (pf_size_y(format)) ncomp++; + if (pf_size_z(format)) ncomp++; + if (pf_size_w(format)) ncomp++; + + return ncomp; +} + +static INLINE int +nv30_vbo_type(uint format) +{ + switch (pf_type(format)) { + case PIPE_FORMAT_TYPE_FLOAT: + return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + case PIPE_FORMAT_TYPE_UNORM: + return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE; + default: + NOUVEAU_ERR("Unknown format 0x%08x\n", format); + return NV40TCL_VTXFMT_TYPE_FLOAT; + } +} + +static boolean +nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, + struct pipe_vertex_element *ve, + struct pipe_vertex_buffer *vb) +{ + struct pipe_winsys *ws = nv30->pipe.winsys; + int type, ncomp; + void *map; + + type = nv30_vbo_type(ve->src_format); + ncomp = nv30_vbo_ncomp(ve->src_format); + + map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); + map += vb->buffer_offset + ve->src_offset; + + switch (type) { + case NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT: + { + float *v = map; + + BEGIN_RING(rankine, NV34TCL_VERTEX_ATTR_4F_X(attrib), 4); + switch (ncomp) { + case 4: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(v[3]); + break; + case 3: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(1.0); + break; + case 2: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + case 1: + OUT_RINGf(v[0]); + OUT_RINGf(0.0); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + } + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + + ws->buffer_unmap(ws, vb->buffer); + + return TRUE; +} + +static void +nv30_vbo_arrays_update(struct nv30_context *nv30) +{ + struct nv30_vertex_program *vp = nv30->vertprog.active; + uint32_t inputs, vtxfmt[16]; + int hw, num_hw; + + nv30->vb_enable = 0; + + inputs = vp->ir; + for (hw = 0; hw < 16 && inputs; hw++) { + if (inputs & (1 << hw)) { + num_hw = hw; + inputs &= ~(1 << hw); + } + } + num_hw++; + + inputs = vp->ir; + for (hw = 0; hw < num_hw; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + + if (!(inputs & (1 << hw))) { + vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + continue; + } + + ve = &nv30->vtxelt[hw]; + vb = &nv30->vtxbuf[ve->vertex_buffer_index]; + + if (vb->pitch == 0) { + vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + if (nv30_vbo_static_attrib(nv30, hw, ve, vb) == TRUE) + continue; + } + + nv30->vb_enable |= (1 << hw); + nv30->vb[hw].delta = vb->buffer_offset + ve->src_offset; + nv30->vb[hw].buffer = vb->buffer; + + vtxfmt[hw] = ((vb->pitch << NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT) | + (nv30_vbo_ncomp(ve->src_format) << + NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT) | + nv30_vbo_type(ve->src_format)); + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_ARRAY_FORMAT(0), num_hw); + OUT_RINGp (vtxfmt, num_hw); +} + +static boolean +nv30_vbo_validate_state(struct nv30_context *nv30, + struct pipe_buffer *ib, unsigned ib_format) +{ + unsigned inputs; + + nv30_emit_hw_state(nv30); + + if (nv30->dirty & NV30_NEW_ARRAYS) { + nv30_vbo_arrays_update(nv30); + nv30->dirty &= ~NV30_NEW_ARRAYS; + } + + inputs = nv30->vb_enable; + while (inputs) { + unsigned a = ffs(inputs) - 1; + + inputs &= ~(1 << a); + + BEGIN_RING(rankine, NV34TCL_VERTEX_BUFFER_ADDRESS(a), 1); + OUT_RELOC (nv30->vb[a].buffer, nv30->vb[a].delta, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | + NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, + NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1); + } + + if (ib) { + BEGIN_RING(rankine, NV40TCL_IDXBUF_ADDRESS, 2); + OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD); + OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); + } + + BEGIN_RING(rankine, 0x1710, 1); + OUT_RING (0); /* vtx cache flush */ + + return TRUE; +} + +boolean +nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, + unsigned count) +{ + struct nv30_context *nv30 = nv30_context(pipe); + unsigned nr; + boolean ret; + + ret = nv30_vbo_validate_state(nv30, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + pipe->flush(pipe, 0); + return TRUE; +} + +static INLINE void +nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, + unsigned start, unsigned count) +{ + uint8_t *elts = (uint8_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2047 * 2); + + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, + unsigned start, unsigned count) +{ + uint16_t *elts = (uint16_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2047 * 2); + + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, + unsigned start, unsigned count) +{ + uint32_t *elts = (uint32_t *)ib + start; + int push; + + while (count) { + push = MIN2(count, 2047); + + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push); + OUT_RINGp (elts, push); + + count -= push; + elts += push; + } +} + +static boolean +nv30_draw_elements_inline(struct pipe_context *pipe, + struct pipe_buffer *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv30_context *nv30 = nv30_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + boolean ret; + void *map; + + ret = nv30_vbo_validate_state(nv30, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } + + map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); + if (!ib) { + NOUVEAU_ERR("failed mapping ib\n"); + return FALSE; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + switch (ib_size) { + case 1: + nv30_draw_elements_u08(nv30, map, start, count); + break; + case 2: + nv30_draw_elements_u16(nv30, map, start, count); + break; + case 4: + nv30_draw_elements_u32(nv30, map, start, count); + break; + default: + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); + break; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + ws->buffer_unmap(ws, ib); + + return TRUE; +} + +static boolean +nv30_draw_elements_vbo(struct pipe_context *pipe, + struct pipe_buffer *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv30_context *nv30 = nv30_context(pipe); + unsigned nr, type; + boolean ret; + + switch (ib_size) { + case 2: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; + break; + case 4: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; + break; + default: + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); + return FALSE; + } + + ret = nv30_vbo_validate_state(nv30, ib, type); + if (!ret) { + NOUVEAU_ERR("failed state validation\n"); + return FALSE; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(rankine, NV40TCL_VB_INDEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(rankine, NV40TCL_VB_INDEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + return TRUE; +} + +boolean +nv30_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ +/* if (indexSize != 1) { + nv30_draw_elements_vbo(pipe, indexBuffer, indexSize, + mode, start, count); + } else */{ + nv30_draw_elements_inline(pipe, indexBuffer, indexSize, + mode, start, count); + } + + pipe->flush(pipe, 0); + return TRUE; +} + + diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c new file mode 100644 index 00000000000..4a8269d5dd7 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -0,0 +1,777 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +/* TODO (at least...): + * 1. Indexed consts + ARL + * 2. Arb. swz/negation + * 3. NV_vp11, NV_vp2, NV_vp3 features + * - extra arith opcodes + * - branching + * - texture sampling + * - indexed attribs + * - indexed results + * 4. bugs + */ + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 8 +#define MASK_Y 4 +#define MASK_Z 2 +#define MASK_W 1 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE 0 +#define DEF_CTEST 0 +#include "nv30_shader.h" + +#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv30_sr_neg((s)) +#define abs(s) nv30_sr_abs((s)) + +struct nv30_vpc { + struct nv30_vertex_program *vp; + + struct nv30_vertex_program_exec *vpi; + + unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; + + int high_temp; + int temp_temp_count; + + struct nv30_sreg *imm; + unsigned nr_imm; +}; + +static struct nv30_sreg +temp(struct nv30_vpc *vpc) +{ + int idx; + + idx = vpc->temp_temp_count++; + idx += vpc->high_temp + 1; + return nv30_sr(NV30SR_TEMP, idx); +} + +static struct nv30_sreg +constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) +{ + struct nv30_vertex_program *vp = vpc->vp; + struct nv30_vertex_program_data *vpd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < vp->nr_consts; idx++) { + if (vp->consts[idx].index == pipe) + return nv30_sr(NV30SR_CONST, idx); + } + } + + idx = vp->nr_consts++; + vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); + vpd = &vp->consts[idx]; + + vpd->index = pipe; + vpd->value[0] = x; + vpd->value[1] = y; + vpd->value[2] = z; + vpd->value[3] = w; + return nv30_sr(NV30SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) + +static void +emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t sr = 0; + + switch (src.type) { + case NV30SR_TEMP: + sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT); + sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV30SR_INPUT: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + vp->ir |= (1 << src.index); + hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT); + break; + case NV30SR_CONST: + sr |= (NV30_VP_SRC_REG_TYPE_CONST << + NV30_VP_SRC_REG_TYPE_SHIFT); + assert(vpc->vpi->const_index == -1 || + vpc->vpi->const_index == src.index); + vpc->vpi->const_index = src.index; + break; + case NV30SR_NONE: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV30_VP_SRC_NEGATE; + + if (src.abs) + hw[0] |= (1 << (21 + pos)); + + sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) | + (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) | + (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) | + (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT)); + +/* + * |VVV| + * d°.°b + * \u/ + * + */ + + switch (pos) { + case 0: + hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >> + NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT; + hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) << + NV30_VP_INST_SRC0L_SHIFT; + break; + case 1: + hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT; + break; + case 2: + hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >> + NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT; + hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) << + NV30_VP_INST_SRC2L_SHIFT; + break; + default: + assert(0); + } +} + +static void +emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) +{ + struct nv30_vertex_program *vp = vpc->vp; + + switch (dst.type) { + case NV30SR_TEMP: + hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT); + break; + case NV30SR_OUTPUT: + switch (dst.index) { + case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; + case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; + case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; + case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; + case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; + case NV30_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; + case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; + case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; + case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; + case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; + case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; + case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; + case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; + case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + default: + break; + } + + hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); + hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); + break; + default: + assert(0); + } +} + +static void +nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, + struct nv30_sreg s2) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t *hw; + + vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); + vpc->vpi = &vp->insns[vp->nr_insns - 1]; + memset(vpc->vpi, 0, sizeof(*vpc->vpi)); + vpc->vpi->const_index = -1; + + hw = vpc->vpi->data; + + hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT); + hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) | + (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) | + (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) | + (3 << NV30_VP_INST_COND_SWZ_W_SHIFT)); + + hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT); +// hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; +// hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); + + emit_dst(vpc, hw, slot, dst); + emit_src(vpc, hw, 0, s0); + emit_src(vpc, hw, 1, s1); + emit_src(vpc, hw, 2, s2); +} + +static INLINE struct nv30_sreg +tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv30_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); + break; + case TGSI_FILE_CONSTANT: + src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + break; + case TGSI_FILE_IMMEDIATE: + src = vpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + if (vpc->high_temp < fsrc->SrcRegister.Index) + vpc->high_temp = fsrc->SrcRegister.Index; + src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv30_sreg +tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + struct nv30_sreg dst; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + dst = nv30_sr(NV30SR_OUTPUT, + vpc->output_map[fdst->DstRegister.Index]); + + break; + case TGSI_FILE_TEMPORARY: + dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); + if (vpc->high_temp < dst.index) + vpc->high_temp = dst.index; + break; + default: + NOUVEAU_ERR("bad dst file\n"); + break; + } + + return dst; +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, + const struct tgsi_full_instruction *finst) +{ + struct nv30_sreg src[3], dst, tmp; + struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + int mask; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + vpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(vpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + /*XXX: index comparison is broken now that consts come from + * two different register files. + */ + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); + break; + case TGSI_OPCODE_ARL: + arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DST: + arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_EXP: + arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_FLR: + arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_LG2: + arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LIT: + arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LOG: + arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_MAD: + arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(vpc); + arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, + swz(src[0], X, X, X, X)); + arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(vpc, 1, OP_EX2, dst, mask, none, none, + swz(tmp, X, X, X, X)); + break; + case TGSI_OPCODE_RCP: + arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_RET: + break; + case TGSI_OPCODE_RSQ: + arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_SGE: + arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); + break; + case TGSI_OPCODE_XPD: + tmp = temp(vpc); + arith(vpc, 0, OP_MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV30_VP_INST_DEST_POS; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_BCOLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_BFC0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_BFC1; + } else { + NOUVEAU_ERR("bad bcolour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV30_VP_INST_DEST_FOGC; + break; + case TGSI_SEMANTIC_PSIZE: + hw = NV30_VP_INST_DEST_PSZ; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + vpc->output_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv30_vertprog_prepare(struct nv30_vpc *vpc) +{ + struct tgsi_parse_context p; + int nr_imm = 0; + + tgsi_parse_init(&p, vpc->vp->pipe->tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch(tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + nr_imm++; + break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (nr_imm) { + vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); + assert(vpc->imm); + } + + return TRUE; +} + +void +nv30_vertprog_translate(struct nv30_context *nv30, + struct nv30_vertex_program *vp) +{ + struct tgsi_parse_context parse; + struct nv30_vpc *vpc = NULL; + + vpc = CALLOC(1, sizeof(struct nv30_vpc)); + if (!vpc) + return; + vpc->vp = vp; + vpc->high_temp = -1; + + if (!nv30_vertprog_prepare(vpc)) { + free(vpc); + return; + } + + tgsi_parse_init(&parse, vp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_OUTPUT: + if (!nv30_vertprog_parse_decl_output(vpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); +// assert(imm->Immediate.Size == 4); + vpc->imm[vpc->nr_imm++] = + constant(vpc, -1, + imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + finst = &parse.FullToken.FullInstruction; + if (!nv30_vertprog_parse_instruction(vpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST; + vp->translated = TRUE; +out_err: + tgsi_parse_free(&parse); + free(vpc); +} + +void +nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) +{ + struct nouveau_winsys *nvws = nv30->nvws; + struct pipe_winsys *ws = nv30->pipe.winsys; + boolean upload_code = FALSE, upload_data = FALSE; + int i; + + /* Translate TGSI shader into hw bytecode */ + if (!vp->translated) { + nv30_vertprog_translate(nv30, vp); + if (!vp->translated) + assert(0); + } + + /* Allocate hw vtxprog exec slots */ + if (!vp->exec) { + struct nouveau_resource *heap = nv30->vertprog.exec_heap; + uint vplen = vp->nr_insns; + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { + while (heap->next && heap->size < vplen) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->exec); + } + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) + assert(0); + } + + upload_code = TRUE; + } + + /* Allocate hw vtxprog const slots */ + if (vp->nr_consts && !vp->data) { + struct nouveau_resource *heap = nv30->vertprog.data_heap; + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { + while (heap->next && heap->size < vp->nr_consts) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) + assert(0); + } + + /*XXX: handle this some day */ + assert(vp->data->start >= vp->data_start_min); + + upload_data = TRUE; + if (vp->data_start != vp->data->start) + upload_code = TRUE; + } + + /* If exec or data segments moved we need to patch the program to + * fixup offsets and register IDs. + */ + if (vp->exec_start != vp->exec->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->has_branch_offset) { + assert(0); + } + } + + vp->exec_start = vp->exec->start; + } + + if (vp->nr_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->const_index >= 0) { + vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; + vpi->data[1] |= + (vpi->const_index + vp->data->start) << + NV30_VP_INST_CONST_SRC_SHIFT; + + } + } + + vp->data_start = vp->data->start; + } + + /* Update + Upload constant values */ + if (vp->nr_consts) { + float *map = NULL; + + if (nv30->vertprog.constant_buf) { + map = ws->buffer_map(ws, nv30->vertprog.constant_buf, + PIPE_BUFFER_USAGE_CPU_READ); + } + + for (i = 0; i < vp->nr_consts; i++) { + struct nv30_vertex_program_data *vpd = &vp->consts[i]; + + if (vpd->index >= 0) { + if (!upload_data && + !memcmp(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float))) + continue; + memcpy(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float)); + } + + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (i + vp->data->start); + OUT_RINGp ((uint32_t *)vpd->value, 4); + } + + if (map) { + ws->buffer_unmap(ws, nv30->vertprog.constant_buf); + } + } + + /* Upload vtxprog */ + if (upload_code) { +#if 0 + for (i = 0; i < vp->nr_insns; i++) { + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); + } +#endif + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (/*vp->exec->start*/0); + for (i = 0; i < vp->nr_insns; i++) { + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (vp->insns[i].data, 4); + } + } + + BEGIN_RING(rankine, NV34TCL_VP_START_FROM_ID, 1); +// OUT_RING (vp->exec->start); + OUT_RING (0); + + nv30->vertprog.active = vp; +} + +void +nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) +{ + if (vp->nr_consts) + free(vp->consts); + if (vp->nr_insns) + free(vp->insns); +} + diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile new file mode 100644 index 00000000000..2a9de4a2dcb --- /dev/null +++ b/src/gallium/drivers/nv40/Makefile @@ -0,0 +1,29 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv40 + +DRIVER_SOURCES = \ + nv40_clear.c \ + nv40_context.c \ + nv40_draw.c \ + nv40_fragprog.c \ + nv40_fragtex.c \ + nv40_miptree.c \ + nv40_query.c \ + nv40_state.c \ + nv40_state_emit.c \ + nv40_surface.c \ + nv40_vbo.c \ + nv40_vertprog.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/drivers/nv40/nv40_clear.c b/src/gallium/drivers/nv40/nv40_clear.c new file mode 100644 index 00000000000..2c4e8f01fda --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv40_context.h" + +void +nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c new file mode 100644 index 00000000000..8b5cc693de0 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -0,0 +1,312 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" + +#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf +#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 +#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 + +static const char * +nv40_get_name(struct pipe_context *pipe) +{ + struct nv40_context *nv40 = nv40_context(pipe); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv40->chipset); + return buffer; +} + +static const char * +nv40_get_vendor(struct pipe_context *pipe) +{ + return "nouveau"; +} + +static int +nv40_get_param(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 16; + case PIPE_CAP_NPOT_TEXTURES: + return 1; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 1; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 4; + case PIPE_CAP_OCCLUSION_QUERY: + return 1; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 1; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv40_get_paramf(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 16.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv40_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_winsys *nvws = nv40->nvws; + + if (flags & PIPE_FLUSH_TEXTURE_CACHE) { + BEGIN_RING(curie, 0x1fd8, 1); + OUT_RING (2); + BEGIN_RING(curie, 0x1fd8, 1); + OUT_RING (1); + } + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv40->hw->sync, 0); + BEGIN_RING(curie, 0x104, 1); + OUT_RING (0); + BEGIN_RING(curie, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv40->hw->sync, 0, 0, 2000); +} + +static void +nv40_channel_takedown(struct nv40_channel_context *cnv40) +{ + struct nouveau_winsys *nvws = cnv40->nvws; + + nvws->res_free(&cnv40->vp_exec_heap); + nvws->res_free(&cnv40->vp_data_heap); + nvws->res_free(&cnv40->query_heap); + nvws->notifier_free(&cnv40->query); + nvws->notifier_free(&cnv40->sync); + nvws->grobj_free(&cnv40->curie); + free(cnv40); +} + +static struct nv40_channel_context * +nv40_channel_init(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv40_channel_context *cnv40 = NULL; + struct nouveau_stateobj *so; + unsigned curie_class = 0; + int ret; + + switch (chipset & 0xf0) { + case 0x40: + if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV40TCL; + else + if (NV4X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV44TCL; + break; + case 0x60: + if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV44TCL; + break; + default: + break; + } + + if (!curie_class) { + NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset); + return NULL; + } + + cnv40 = CALLOC(1, sizeof(struct nv40_channel_context)); + if (!cnv40) + return NULL; + cnv40->chipset = chipset; + cnv40->nvws = nvws; + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &cnv40->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv40_channel_takedown(cnv40); + return NULL; + } + + /* Query objects */ + ret = nvws->notifier_alloc(nvws, 32, &cnv40->query); + if (ret) { + NOUVEAU_ERR("Error initialising query objects: %d\n", ret); + nv40_channel_takedown(cnv40); + return NULL; + } + + ret = nvws->res_init(&cnv40->query_heap, 0, 32); + if (ret) { + NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); + nv40_channel_takedown(cnv40); + return NULL; + } + + /* Vtxprog resources */ + if (nvws->res_init(&cnv40->vp_exec_heap, 0, 512) || + nvws->res_init(&cnv40->vp_data_heap, 0, 256)) { + nv40_channel_takedown(cnv40); + return NULL; + } + + /* 3D object */ + ret = nvws->grobj_alloc(nvws, curie_class, &cnv40->curie); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + /* Static curie initialisation */ + so = so_new(128, 0); + so_method(so, cnv40->curie, NV40TCL_DMA_NOTIFY, 1); + so_data (so, cnv40->sync->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_TEXTURE0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR1, 1); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_VTXBUF0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_FENCE, 2); + so_data (so, 0); + so_data (so, cnv40->query->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_UNK01AC, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, cnv40->curie, NV40TCL_DMA_COLOR2, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + + so_method(so, cnv40->curie, 0x1ea4, 3); + so_data (so, 0x00000010); + so_data (so, 0x01000100); + so_data (so, 0xff800006); + + /* vtxprog output routing */ + so_method(so, cnv40->curie, 0x1fc4, 1); + so_data (so, 0x06144321); + so_method(so, cnv40->curie, 0x1fc8, 2); + so_data (so, 0xedcba987); + so_data (so, 0x00000021); + so_method(so, cnv40->curie, 0x1fd0, 1); + so_data (so, 0x00171615); + so_method(so, cnv40->curie, 0x1fd4, 1); + so_data (so, 0x001b1a19); + + so_method(so, cnv40->curie, 0x1ef8, 1); + so_data (so, 0x0020ffff); + so_method(so, cnv40->curie, 0x1d64, 1); + so_data (so, 0x00d30000); + so_method(so, cnv40->curie, 0x1e94, 1); + so_data (so, 0x00000001); + + so_emit(nvws, so); + so_ref(NULL, &so); + nvws->push_flush(nvws->channel, 0); + + return cnv40; +} + +static void +nv40_destroy(struct pipe_context *pipe) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + if (nv40->draw) + draw_destroy(nv40->draw); + + if (nv40->hw) { + if (--nv40->hw->refcount == 0) + nv40_channel_takedown(nv40->hw); + } + + free(nv40); +} + +struct pipe_context * +nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv40_context *nv40; + + nv40 = CALLOC(1, sizeof(struct nv40_context)); + if (!nv40) + return NULL; + + nv40->hw = nv40_channel_init(ws, nvws, chipset); + if (!nv40->hw) { + nv40_destroy(&nv40->pipe); + return NULL; + } + + nv40->chipset = chipset; + nv40->nvws = nvws; + + nv40->pipe.winsys = ws; + nv40->pipe.destroy = nv40_destroy; + nv40->pipe.get_name = nv40_get_name; + nv40->pipe.get_vendor = nv40_get_vendor; + nv40->pipe.get_param = nv40_get_param; + nv40->pipe.get_paramf = nv40_get_paramf; + nv40->pipe.draw_arrays = nv40_draw_arrays; + nv40->pipe.draw_elements = nv40_draw_elements; + nv40->pipe.clear = nv40_clear; + nv40->pipe.flush = nv40_flush; + + nv40_init_query_functions(nv40); + nv40_init_surface_functions(nv40); + nv40_init_state_functions(nv40); + nv40_init_miptree_functions(nv40); + + nv40->draw = draw_create(); + assert(nv40->draw); + draw_set_rasterize_stage(nv40->draw, nv40_draw_render_stage(nv40)); + + return &nv40->pipe; +} + diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h new file mode 100644 index 00000000000..f511759e3be --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -0,0 +1,153 @@ +#ifndef __NV40_CONTEXT_H__ +#define __NV40_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv40_channel_context *ctx = nv40->hw +#include "nouveau/nouveau_push.h" +#include "nouveau/nouveau_stateobj.h" + +#include "nv40_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV40_NEW_BLEND (1 << 0) +#define NV40_NEW_RAST (1 << 1) +#define NV40_NEW_ZSA (1 << 2) +#define NV40_NEW_SAMPLER (1 << 3) +#define NV40_NEW_FB (1 << 4) +#define NV40_NEW_STIPPLE (1 << 5) +#define NV40_NEW_SCISSOR (1 << 6) +#define NV40_NEW_VIEWPORT (1 << 7) +#define NV40_NEW_BCOL (1 << 8) +#define NV40_NEW_VERTPROG (1 << 9) +#define NV40_NEW_FRAGPROG (1 << 10) +#define NV40_NEW_ARRAYS (1 << 11) + +struct nv40_channel_context { + struct nouveau_winsys *nvws; + unsigned refcount; + + unsigned chipset; + + /* HW graphics objects */ + struct nouveau_grobj *curie; + struct nouveau_notifier *sync; + + /* Query object resources */ + struct nouveau_notifier *query; + struct nouveau_resource *query_heap; + + /* Vtxprog resources */ + struct nouveau_resource *vp_exec_heap; + struct nouveau_resource *vp_data_heap; +}; + +struct nv40_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct nv40_channel_context *hw; + struct draw_context *draw; + + int chipset; + + uint32_t dirty; + + struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + struct nouveau_stateobj *so_framebuffer; + struct nouveau_stateobj *so_fragtex[16]; + struct nouveau_stateobj *so_vtxbuf; + struct nouveau_stateobj *so_blend; + struct nouveau_stateobj *so_rast; + struct nouveau_stateobj *so_zsa; + struct nouveau_stateobj *so_bcol; + struct nouveau_stateobj *so_scissor; + struct nouveau_stateobj *so_viewport; + struct nouveau_stateobj *so_stipple; + + struct { + struct nv40_vertex_program *active; + + struct nv40_vertex_program *current; + struct pipe_buffer *constant_buf; + } vertprog; + + struct { + struct nv40_fragment_program *active; + + struct nv40_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; +}; + +static INLINE struct nv40_context * +nv40_context(struct pipe_context *pipe) +{ + return (struct nv40_context *)pipe; +} + +extern void nv40_init_state_functions(struct nv40_context *nv40); +extern void nv40_init_surface_functions(struct nv40_context *nv40); +extern void nv40_init_miptree_functions(struct nv40_context *nv40); +extern void nv40_init_query_functions(struct nv40_context *nv40); + +/* nv40_draw.c */ +extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); + +/* nv40_vertprog.c */ +extern void nv40_vertprog_translate(struct nv40_context *, + struct nv40_vertex_program *); +extern void nv40_vertprog_bind(struct nv40_context *, + struct nv40_vertex_program *); +extern void nv40_vertprog_destroy(struct nv40_context *, + struct nv40_vertex_program *); + +/* nv40_fragprog.c */ +extern void nv40_fragprog_translate(struct nv40_context *, + struct nv40_fragment_program *); +extern void nv40_fragprog_bind(struct nv40_context *, + struct nv40_fragment_program *); +extern void nv40_fragprog_destroy(struct nv40_context *, + struct nv40_fragment_program *); + +/* nv40_fragtex.c */ +extern void nv40_fragtex_bind(struct nv40_context *); + +/* nv40_state.c and friends */ +extern void nv40_emit_hw_state(struct nv40_context *nv40); +extern void nv40_state_tex_update(struct nv40_context *nv40); + +/* nv40_vbo.c */ +extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv40_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned mode, unsigned start, + unsigned count); + +/* nv40_clear.c */ +extern void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +#endif diff --git a/src/gallium/drivers/nv40/nv40_dma.h b/src/gallium/drivers/nv40/nv40_dma.h new file mode 100644 index 00000000000..1fb82677689 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_dma.h @@ -0,0 +1,66 @@ +#ifndef __NV40_DMA_H__ +#define __NV40_DMA_H__ + +#include "pipe/nouveau/nouveau_winsys.h" + +#define OUT_RING(data) do { \ + (*nv40->nvws->channel->pushbuf->cur++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + memcpy(nv40->nvws->channel->pushbuf->cur, (src), (size) * 4); \ + nv40->nvws->channel->pushbuf->cur += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define BEGIN_RING(obj,mthd,size) do { \ + if (nv40->nvws->channel->pushbuf->remaining < ((size) + 1)) \ + nv40->nvws->push_flush(nv40->nvws->channel, ((size) + 1)); \ + OUT_RING((nv40->obj->subc << 13) | ((size) << 18) | (mthd)); \ + nv40->nvws->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING_NI(obj,mthd,size) do { \ + BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ +} while(0) + +#define FIRE_RING() do { \ + nv40->nvws->push_flush(nv40->nvws->channel, 0); \ +} while(0) + +#define OUT_RELOC(bo,data,flags,vor,tor) do { \ + nv40->nvws->push_reloc(nv40->nvws->channel, \ + nv40->nvws->channel->pushbuf->cur, \ + (struct nouveau_bo *)(bo), \ + (data), (flags), (vor), (tor)); \ + OUT_RING(0); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + nv40->nvws->channel->vram->handle, \ + nv40->nvws->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c new file mode 100644 index 00000000000..a39bb85e99b --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -0,0 +1,62 @@ +#include "draw/draw_private.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" + +struct nv40_draw_stage { + struct draw_stage draw; + struct nv40_context *nv40; +}; + +static void +nv40_draw_point(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_line(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_tri(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_flush(struct draw_stage *draw, unsigned flags) +{ +} + +static void +nv40_draw_reset_stipple_counter(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv40_draw_destroy(struct draw_stage *draw) +{ + free(draw); +} + +struct draw_stage * +nv40_draw_render_stage(struct nv40_context *nv40) +{ + struct nv40_draw_stage *nv40draw = CALLOC_STRUCT(nv40_draw_stage); + + nv40draw->nv40 = nv40; + nv40draw->draw.draw = nv40->draw; + nv40draw->draw.point = nv40_draw_point; + nv40draw->draw.line = nv40_draw_line; + nv40draw->draw.tri = nv40_draw_tri; + nv40draw->draw.flush = nv40_draw_flush; + nv40draw->draw.reset_stipple_counter = nv40_draw_reset_stipple_counter; + nv40draw->draw.destroy = nv40_draw_destroy; + + return &nv40draw->draw; +} + diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c new file mode 100644 index 00000000000..07a418c1e9f --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -0,0 +1,842 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_util.h" + +#include "nv40_context.h" + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 1 +#define MASK_Y 2 +#define MASK_Z 4 +#define MASK_W 8 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE NV40_FP_OP_DST_SCALE_1X +#define DEF_CTEST NV40_FP_OP_COND_TR +#include "nv40_shader.h" + +#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv40_sr_neg((s)) +#define abs(s) nv40_sr_abs((s)) +#define scale(s,v) nv40_sr_scale((s), NV40_FP_OP_DST_SCALE_##v) + +#define MAX_CONSTS 128 +#define MAX_IMM 32 +struct nv40_fpc { + struct nv40_fragment_program *fp; + + uint attrib_map[PIPE_MAX_SHADER_INPUTS]; + + int high_temp; + int temp_temp_count; + int num_regs; + + uint depth_id; + uint colour_id; + + unsigned inst_offset; + + struct { + int pipe; + float vals[4]; + } consts[MAX_CONSTS]; + int nr_consts; + + struct nv40_sreg imm[MAX_IMM]; + unsigned nr_imm; +}; + +static INLINE struct nv40_sreg +temp(struct nv40_fpc *fpc) +{ + int idx; + + idx = fpc->temp_temp_count++; + idx += fpc->high_temp + 1; + return nv40_sr(NV40SR_TEMP, idx); +} + +static INLINE struct nv40_sreg +constant(struct nv40_fpc *fpc, int pipe, float vals[4]) +{ + int idx; + + if (fpc->nr_consts == MAX_CONSTS) + assert(0); + idx = fpc->nr_consts++; + + fpc->consts[idx].pipe = pipe; + if (pipe == -1) + memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float)); + return nv40_sr(NV40SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv40_fp_arith((cc), (s), NV40_FP_OP_OPCODE_##o, \ + (d), (m), (s0), (s1), (s2)) +#define tex(cc,s,o,u,d,m,s0,s1,s2) \ + nv40_fp_tex((cc), (s), NV40_FP_OP_OPCODE_##o, (u), \ + (d), (m), (s0), none, none) + +static void +grow_insns(struct nv40_fpc *fpc, int size) +{ + struct nv40_fragment_program *fp = fpc->fp; + + fp->insn_len += size; + fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len); +} + +static void +emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) +{ + struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; + uint32_t sr = 0; + + switch (src.type) { + case NV40SR_INPUT: + sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); + hw[0] |= (src.index << NV40_FP_OP_INPUT_SRC_SHIFT); + break; + case NV40SR_OUTPUT: + sr |= NV40_FP_REG_SRC_HALF; + /* fall-through */ + case NV40SR_TEMP: + sr |= (NV40_FP_REG_TYPE_TEMP << NV40_FP_REG_TYPE_SHIFT); + sr |= (src.index << NV40_FP_REG_SRC_SHIFT); + break; + case NV40SR_CONST: + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + if (fpc->consts[src.index].pipe >= 0) { + struct nv40_fragment_program_data *fpd; + + fp->consts = realloc(fp->consts, ++fp->nr_consts * + sizeof(*fpd)); + fpd = &fp->consts[fp->nr_consts - 1]; + fpd->offset = fpc->inst_offset + 4; + fpd->index = fpc->consts[src.index].pipe; + memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4); + } else { + memcpy(&fp->insn[fpc->inst_offset + 4], + fpc->consts[src.index].vals, + sizeof(uint32_t) * 4); + } + + sr |= (NV40_FP_REG_TYPE_CONST << NV40_FP_REG_TYPE_SHIFT); + break; + case NV40SR_NONE: + sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV40_FP_REG_NEGATE; + + if (src.abs) + hw[1] |= (1 << (29 + pos)); + + sr |= ((src.swz[0] << NV40_FP_REG_SWZ_X_SHIFT) | + (src.swz[1] << NV40_FP_REG_SWZ_Y_SHIFT) | + (src.swz[2] << NV40_FP_REG_SWZ_Z_SHIFT) | + (src.swz[3] << NV40_FP_REG_SWZ_W_SHIFT)); + + hw[pos + 1] |= sr; +} + +static void +emit_dst(struct nv40_fpc *fpc, struct nv40_sreg dst) +{ + struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw = &fp->insn[fpc->inst_offset]; + + switch (dst.type) { + case NV40SR_TEMP: + if (fpc->num_regs < (dst.index + 1)) + fpc->num_regs = dst.index + 1; + break; + case NV40SR_OUTPUT: + if (dst.index == 1) { + fp->fp_control |= 0xe; + } else { + hw[0] |= NV40_FP_OP_OUT_REG_HALF; + } + break; + case NV40SR_NONE: + hw[0] |= (1 << 30); + break; + default: + assert(0); + } + + hw[0] |= (dst.index << NV40_FP_OP_OUT_REG_SHIFT); +} + +static void +nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, + struct nv40_sreg dst, int mask, + struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) +{ + struct nv40_fragment_program *fp = fpc->fp; + uint32_t *hw; + + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + hw = &fp->insn[fpc->inst_offset]; + memset(hw, 0, sizeof(uint32_t) * 4); + + if (op == NV40_FP_OP_OPCODE_KIL) + fp->fp_control |= NV40TCL_FP_CONTROL_KIL; + hw[0] |= (op << NV40_FP_OP_OPCODE_SHIFT); + hw[0] |= (mask << NV40_FP_OP_OUTMASK_SHIFT); + hw[2] |= (dst.dst_scale << NV40_FP_OP_DST_SCALE_SHIFT); + + if (sat) + hw[0] |= NV40_FP_OP_OUT_SAT; + + if (dst.cc_update) + hw[0] |= NV40_FP_OP_COND_WRITE_ENABLE; + hw[1] |= (dst.cc_test << NV40_FP_OP_COND_SHIFT); + hw[1] |= ((dst.cc_swz[0] << NV40_FP_OP_COND_SWZ_X_SHIFT) | + (dst.cc_swz[1] << NV40_FP_OP_COND_SWZ_Y_SHIFT) | + (dst.cc_swz[2] << NV40_FP_OP_COND_SWZ_Z_SHIFT) | + (dst.cc_swz[3] << NV40_FP_OP_COND_SWZ_W_SHIFT)); + + emit_dst(fpc, dst); + emit_src(fpc, 0, s0); + emit_src(fpc, 1, s1); + emit_src(fpc, 2, s2); +} + +static void +nv40_fp_tex(struct nv40_fpc *fpc, int sat, int op, int unit, + struct nv40_sreg dst, int mask, + struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) +{ + struct nv40_fragment_program *fp = fpc->fp; + + nv40_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); + + fp->insn[fpc->inst_offset] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); + fp->samplers |= (1 << unit); +} + +static INLINE struct nv40_sreg +tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) +{ + struct nv40_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv40_sr(NV40SR_INPUT, + fpc->attrib_map[fsrc->SrcRegister.Index]); + break; + case TGSI_FILE_CONSTANT: + src = constant(fpc, fsrc->SrcRegister.Index, NULL); + break; + case TGSI_FILE_IMMEDIATE: + assert(fsrc->SrcRegister.Index < fpc->nr_imm); + src = fpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index + 1); + if (fpc->high_temp < src.index) + fpc->high_temp = src.index; + break; + /* This is clearly insane, but gallium hands us shaders like this. + * Luckily fragprog results are just temp regs.. + */ + case TGSI_FILE_OUTPUT: + if (fsrc->SrcRegister.Index == fpc->colour_id) + return nv40_sr(NV40SR_OUTPUT, 0); + else + return nv40_sr(NV40SR_OUTPUT, 1); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv40_sreg +tgsi_dst(struct nv40_fpc *fpc, const struct tgsi_full_dst_register *fdst) { + int idx; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + if (fdst->DstRegister.Index == fpc->colour_id) + return nv40_sr(NV40SR_OUTPUT, 0); + else + return nv40_sr(NV40SR_OUTPUT, 1); + break; + case TGSI_FILE_TEMPORARY: + idx = fdst->DstRegister.Index + 1; + if (fpc->high_temp < idx) + fpc->high_temp = idx; + return nv40_sr(NV40SR_TEMP, idx); + case TGSI_FILE_NULL: + return nv40_sr(NV40SR_NONE, 0); + default: + NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); + return nv40_sr(NV40SR_NONE, 0); + } +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +src_native_swz(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc, + struct nv40_sreg *src) +{ + const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); + struct nv40_sreg tgsi = tgsi_src(fpc, fsrc); + uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; + uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, + fsrc->SrcRegisterExtSwz.NegateY, + fsrc->SrcRegisterExtSwz.NegateZ, + fsrc->SrcRegisterExtSwz.NegateW }; + uint c; + + for (c = 0; c < 4; c++) { + switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + mask |= (1 << c); + break; + case TGSI_EXTSWIZZLE_ZERO: + zero_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + case TGSI_EXTSWIZZLE_ONE: + one_mask |= (1 << c); + tgsi.swz[c] = SWZ_X; + break; + default: + assert(0); + } + + if (!tgsi.negate && neg[c]) + neg_mask |= (1 << c); + } + + if (mask == MASK_ALL && !neg_mask) + return TRUE; + + *src = temp(fpc); + + if (mask) + arith(fpc, 0, MOV, *src, mask, tgsi, none, none); + + if (zero_mask) + arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); + + if (one_mask) + arith(fpc, 0, STR, *src, one_mask, *src, none, none); + + if (neg_mask) { + struct nv40_sreg one = temp(fpc); + arith(fpc, 0, STR, one, neg_mask, one, none, none); + arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none); + } + + return FALSE; +} + +static boolean +nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, + const struct tgsi_full_instruction *finst) +{ + const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); + struct nv40_sreg src[3], dst, tmp; + int mask, sat, unit; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + fpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(fpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + case TGSI_FILE_CONSTANT: + case TGSI_FILE_TEMPORARY: + if (!src_native_swz(fpc, fsrc, &src[i])) + continue; + break; + default: + break; + } + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + NOUVEAU_MSG("extra src attr %d\n", + fsrc->SrcRegister.Index); + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + case TGSI_FILE_SAMPLER: + unit = fsrc->SrcRegister.Index; + break; + case TGSI_FILE_OUTPUT: + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_CMP: + tmp = temp(fpc); + arith(fpc, sat, MOV, dst, mask, src[2], none, none); + tmp.cc_update = 1; + arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); + dst.cc_test = NV40_VP_INST_COND_LT; + arith(fpc, sat, MOV, dst, mask, src[1], none, none); + break; + case TGSI_OPCODE_COS: + arith(fpc, sat, COS, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(fpc, sat, DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none); + arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X), + swz(src[1], W, W, W, W), none); + break; + case TGSI_OPCODE_DST: + arith(fpc, sat, DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(fpc, sat, EX2, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FLR: + arith(fpc, sat, FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(fpc, sat, FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_KIL: + arith(fpc, 0, KIL, none, 0, none, none, none); + break; + case TGSI_OPCODE_KILP: + dst = nv40_sr(NV40SR_NONE, 0); + dst.cc_update = 1; + arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); + dst.cc_update = 0; dst.cc_test = NV40_FP_OP_COND_LT; + arith(fpc, 0, KIL, dst, 0, none, none, none); + break; + case TGSI_OPCODE_LG2: + arith(fpc, sat, LG2, dst, mask, src[0], none, none); + break; +// case TGSI_OPCODE_LIT: + case TGSI_OPCODE_LRP: + tmp = temp(fpc); + arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); + arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); + break; + case TGSI_OPCODE_MAD: + arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(fpc, sat, MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(fpc, sat, MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(fpc, sat, MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(fpc); + arith(fpc, 0, LG2, tmp, MASK_X, + swz(src[0], X, X, X, X), none, none); + arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(fpc, sat, EX2, dst, mask, + swz(tmp, X, X, X, X), none, none); + break; + case TGSI_OPCODE_RCP: + arith(fpc, sat, RCP, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_RET: + assert(0); + break; + case TGSI_OPCODE_RFL: + tmp = temp(fpc); + arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); + arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); + arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, + swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); + arith(fpc, sat, MAD, dst, mask, + swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); + break; + case TGSI_OPCODE_RSQ: + tmp = temp(fpc); + arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, + abs(swz(src[0], X, X, X, X)), none, none); + arith(fpc, sat, EX2, dst, mask, + neg(swz(tmp, X, X, X, X)), none, none); + break; + case TGSI_OPCODE_SCS: + if (mask & MASK_X) { + arith(fpc, sat, COS, dst, MASK_X, + swz(src[0], X, X, X, X), none, none); + } + if (mask & MASK_Y) { + arith(fpc, sat, SIN, dst, MASK_Y, + swz(src[0], X, X, X, X), none, none); + } + break; + case TGSI_OPCODE_SIN: + arith(fpc, sat, SIN, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_SGE: + arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); + break; + case TGSI_OPCODE_TEX: + if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == + TGSI_EXTSWIZZLE_W) { + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + } else + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_TXB: + tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_XPD: + tmp = temp(fpc); + arith(fpc, 0, MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(fpc, sat, MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv40_fragprog_parse_decl_attrib(struct nv40_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV40_FP_OP_INPUT_SRC_POSITION; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV40_FP_OP_INPUT_SRC_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV40_FP_OP_INPUT_SRC_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV40_FP_OP_INPUT_SRC_FOGC; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV40_FP_OP_INPUT_SRC_TC(fdec->Semantic. + SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad input semantic\n"); + return FALSE; + } + + fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, + const struct tgsi_full_declaration *fdec) +{ + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + fpc->depth_id = fdec->u.DeclarationRange.First; + break; + case TGSI_SEMANTIC_COLOR: + fpc->colour_id = fdec->u.DeclarationRange.First; + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + return TRUE; +} + +void +nv40_fragprog_translate(struct nv40_context *nv40, + struct nv40_fragment_program *fp) +{ + struct tgsi_parse_context parse; + struct nv40_fpc *fpc = NULL; + + fpc = CALLOC(1, sizeof(struct nv40_fpc)); + if (!fpc) + return; + fpc->fp = fp; + fpc->high_temp = -1; + fpc->num_regs = 2; + + tgsi_parse_init(&parse, fp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_INPUT: + if (!nv40_fragprog_parse_decl_attrib(fpc, fdec)) + goto out_err; + break; + case TGSI_FILE_OUTPUT: + if (!nv40_fragprog_parse_decl_output(fpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + struct tgsi_full_immediate *imm; + float vals[4]; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); + assert(fpc->nr_imm < MAX_IMM); + + vals[0] = imm->u.ImmediateFloat32[0].Float; + vals[1] = imm->u.ImmediateFloat32[1].Float; + vals[2] = imm->u.ImmediateFloat32[2].Float; + vals[3] = imm->u.ImmediateFloat32[3].Float; + fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + + finst = &parse.FullToken.FullInstruction; + if (!nv40_fragprog_parse_instruction(fpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + fp->fp_control |= fpc->num_regs << NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT; + + /* Terminate final instruction */ + fp->insn[fpc->inst_offset] |= 0x00000001; + + /* Append NOP + END instruction, may or may not be necessary. */ + fpc->inst_offset = fp->insn_len; + grow_insns(fpc, 4); + fp->insn[fpc->inst_offset + 0] = 0x00000001; + fp->insn[fpc->inst_offset + 1] = 0x00000000; + fp->insn[fpc->inst_offset + 2] = 0x00000000; + fp->insn[fpc->inst_offset + 3] = 0x00000000; + + fp->translated = TRUE; + fp->on_hw = FALSE; +out_err: + tgsi_parse_free(&parse); + free(fpc); +} + +void +nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) +{ + struct pipe_winsys *ws = nv40->pipe.winsys; + struct nouveau_stateobj *so; + int i; + + if (!fp->translated) { + nv40_fragprog_translate(nv40, fp); + if (!fp->translated) + assert(0); + } + + if (fp->nr_consts) { + float *map = ws->buffer_map(ws, nv40->fragprog.constant_buf, + PIPE_BUFFER_USAGE_CPU_READ); + for (i = 0; i < fp->nr_consts; i++) { + struct nv40_fragment_program_data *fpd = &fp->consts[i]; + uint32_t *p = &fp->insn[fpd->offset]; + uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; + + if (!memcmp(p, cb, 4 * sizeof(float))) + continue; + memcpy(p, cb, 4 * sizeof(float)); + fp->on_hw = 0; + } + ws->buffer_unmap(ws, nv40->fragprog.constant_buf); + } + + if (!fp->on_hw) { + const uint32_t le = 1; + uint32_t *map; + + if (!fp->buffer) + fp->buffer = ws->buffer_create(ws, 0x100, 0, + fp->insn_len * 4); + map = ws->buffer_map(ws, fp->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); + +#if 0 + for (i = 0; i < fp->insn_len; i++) { + NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + } +#endif + + if ((*(const uint8_t *)&le)) { + for (i = 0; i < fp->insn_len; i++) { + map[i] = fp->insn[i]; + } + } else { + /* Weird swapping for big-endian chips */ + for (i = 0; i < fp->insn_len; i++) { + map[i] = ((fp->insn[i] & 0xffff) << 16) | + ((fp->insn[i] >> 16) & 0xffff); + } + } + + ws->buffer_unmap(ws, fp->buffer); + fp->on_hw = TRUE; + } + + so = so_new(4, 1); + so_method(so, nv40->hw->curie, NV40TCL_FP_ADDRESS, 1); + so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); + so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); + so_data (so, fp->fp_control); + + so_emit(nv40->nvws, so); + so_ref(so, &fp->so); + so_ref(NULL, &so); + + nv40->fragprog.active = fp; +} + +void +nv40_fragprog_destroy(struct nv40_context *nv40, + struct nv40_fragment_program *fp) +{ + if (fp->insn_len) + free(fp->insn); +} + diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c new file mode 100644 index 00000000000..5af5fbe7465 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -0,0 +1,151 @@ +#include "nv40_context.h" + +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV40TCL_TEX_FORMAT_FORMAT_##tf, \ + (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x | NV40TCL_TEX_SWIZZLE_S0_Y_##ts0y | \ + NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w | \ + NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y | \ + NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ +} + +struct nv40_texture_format { + boolean defined; + uint pipe; + int format; + int swizzle; +}; + +static struct nv40_texture_format +nv40_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), + _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), + _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), + _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), + _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), + _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), +// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), +// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + {}, +}; + +static struct nv40_texture_format * +nv40_fragtex_format(uint pipe_format) +{ + struct nv40_texture_format *tf = nv40_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv40_fragtex_build(struct nv40_context *nv40, int unit) +{ + struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; + struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; + struct pipe_texture *pt = &nv40mt->base; + struct nv40_texture_format *tf; + struct nouveau_stateobj *so; + uint32_t txf, txs, txp; + int swizzled = 0; /*XXX: implement in region code? */ + unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; + + tf = nv40_fragtex_format(pt->format); + if (!tf) + assert(0); + + txf = ps->fmt; + txf |= tf->format | 0x8000; + txf |= ((pt->last_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); + + if (1) /* XXX */ + txf |= NV40TCL_TEX_FORMAT_NO_BORDER; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE: + txf |= NV40TCL_TEX_FORMAT_CUBIC; + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= NV40TCL_TEX_FORMAT_DIMS_2D; + break; + case PIPE_TEXTURE_3D: + txf |= NV40TCL_TEX_FORMAT_DIMS_3D; + break; + case PIPE_TEXTURE_1D: + txf |= NV40TCL_TEX_FORMAT_DIMS_1D; + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + if (swizzled) { + txp = 0; + } else { + txp = nv40mt->level[0].pitch; + txf |= NV40TCL_TEX_FORMAT_LINEAR; + } + + txs = tf->swizzle; + + so = so_new(16, 2); + so_method(so, nv40->hw->curie, NV40TCL_TEX_OFFSET(unit), 8); + so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, + NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); + so_data (so, ps->wrap); + so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en); + so_data (so, txs); + so_data (so, ps->filt | 0x2000 /*voodoo*/); + so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | + pt->height[0]); + so_data (so, ps->bcol); + so_method(so, nv40->hw->curie, NV40TCL_TEX_SIZE1(unit), 1); + so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); + + so_emit(nv40->nvws, so); + so_ref (so, &nv40->so_fragtex[unit]); + so_ref (NULL, &so); +} + +void +nv40_fragtex_bind(struct nv40_context *nv40) +{ + struct nv40_fragment_program *fp = nv40->fragprog.active; + unsigned samplers, unit; + + samplers = nv40->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + so_ref(NULL, &nv40->so_fragtex[unit]); + BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv40->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv40_fragtex_build(nv40, unit); + } + + nv40->fp_samplers = fp->samplers; +} + diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c new file mode 100644 index 00000000000..92e6b3a43df --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -0,0 +1,104 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nv40_context.h" + +static void +nv40_miptree_layout(struct nv40_miptree *nv40mt) +{ + struct pipe_texture *pt = &nv40mt->base; + boolean swizzled = FALSE; + uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint offset = 0; + int nr_faces, l, f; + + if (pt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else + if (pt->target == PIPE_TEXTURE_3D) { + nr_faces = pt->depth[0]; + } else { + nr_faces = 1; + } + + for (l = 0; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->depth[l] = depth; + + if (swizzled) + nv40mt->level[l].pitch = pt->width[l] * pt->cpp; + else + nv40mt->level[l].pitch = pt->width[0] * pt->cpp; + nv40mt->level[l].pitch = (nv40mt->level[l].pitch + 63) & ~63; + + nv40mt->level[l].image_offset = + CALLOC(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + depth = MAX2(1, depth >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = 0; l <= pt->last_level; l++) { + nv40mt->level[l].image_offset[f] = offset; + offset += nv40mt->level[l].pitch * pt->height[l]; + } + } + + nv40mt->total_size = offset; +} + +static struct pipe_texture * +nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv40_miptree *mt; + + mt = MALLOC(sizeof(struct nv40_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + nv40_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + free(mt); + return NULL; + } + + return &mt->base; +} + +static void +nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = pipe->winsys; + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv40_miptree *nv40mt = (struct nv40_miptree *)mt; + int l; + + pipe_buffer_reference(ws, &nv40mt->buffer, NULL); + for (l = 0; l <= mt->last_level; l++) { + if (nv40mt->level[l].image_offset) + free(nv40mt->level[l].image_offset); + } + free(nv40mt); + } +} + +void +nv40_init_miptree_functions(struct nv40_context *nv40) +{ + nv40->pipe.texture_create = nv40_miptree_create; + nv40->pipe.texture_release = nv40_miptree_release; +} + diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c new file mode 100644 index 00000000000..8bca2788b93 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_query.c @@ -0,0 +1,113 @@ +#include "pipe/p_context.h" + +#include "nv40_context.h" + +struct nv40_query { + struct nouveau_resource *object; + unsigned type; + boolean ready; + uint64_t result; +}; + +static INLINE struct nv40_query * +nv40_query(struct pipe_query *pipe) +{ + return (struct nv40_query *)pipe; +} + +static struct pipe_query * +nv40_query_create(struct pipe_context *pipe, unsigned query_type) +{ + struct nv40_query *q; + + q = CALLOC(1, sizeof(struct nv40_query)); + q->type = query_type; + + return (struct pipe_query *)q; +} + +static void +nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *q = nv40_query(pq); + + if (q->object) + nv40->nvws->res_free(&q->object); + free(q); +} + +static void +nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *q = nv40_query(pq); + + assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (nv40->nvws->res_alloc(nv40->hw->query_heap, 1, NULL, &q->object)) + assert(0); + nv40->nvws->notifier_reset(nv40->hw->query, q->object->start); + + BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); + OUT_RING (1); + BEGIN_RING(curie, NV40TCL_QUERY_UNK17CC, 1); + OUT_RING (1); + + q->ready = FALSE; +} + +static void +nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *q = nv40_query(pq); + + BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); + OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | + ((q->object->start * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); + FIRE_RING(); +} + +static boolean +nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, + boolean wait, uint64 *result) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_query *q = nv40_query(pq); + struct nouveau_winsys *nvws = nv40->nvws; + + assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER); + + if (!q->ready) { + unsigned status; + + status = nvws->notifier_status(nv40->hw->query, + q->object->start); + if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { + if (wait == FALSE) + return FALSE; + nvws->notifier_wait(nv40->hw->query, q->object->start, + NV_NOTIFY_STATE_STATUS_COMPLETED, + 0); + } + + q->result = nvws->notifier_retval(nv40->hw->query, + q->object->start); + q->ready = TRUE; + nvws->res_free(&q->object); + } + + *result = q->result; + return TRUE; +} + +void +nv40_init_query_functions(struct nv40_context *nv40) +{ + nv40->pipe.create_query = nv40_query_create; + nv40->pipe.destroy_query = nv40_query_destroy; + nv40->pipe.begin_query = nv40_query_begin; + nv40->pipe.end_query = nv40_query_end; + nv40->pipe.get_query_result = nv40_query_result; +} diff --git a/src/gallium/drivers/nv40/nv40_shader.h b/src/gallium/drivers/nv40/nv40_shader.h new file mode 100644 index 00000000000..5909c70713c --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_shader.h @@ -0,0 +1,554 @@ +#ifndef __NV40_SHADER_H__ +#define __NV40_SHADER_H__ + +/* Vertex programs instruction set + * + * The NV40 instruction set is very similar to NV30. Most fields are in + * a slightly different position in the instruction however. + * + * Merged instructions + * In some cases it is possible to put two instructions into one opcode + * slot. The rules for when this is OK is not entirely clear to me yet. + * + * There are separate writemasks and dest temp register fields for each + * grouping of instructions. There is however only one field with the + * ID of a result register. Writing to temp/result regs is selected by + * setting VEC_RESULT/SCA_RESULT. + * + * Temporary registers + * The source/dest temp register fields have been extended by 1 bit, to + * give a total of 32 temporary registers. + * + * Relative Addressing + * NV40 can use an address register to index into vertex attribute regs. + * This is done by putting the offset value into INPUT_SRC and setting + * the INDEX_INPUT flag. + * + * Conditional execution (see NV_vertex_program{2,3} for details) + * There is a second condition code register on NV40, it's use is enabled + * by setting the COND_REG_SELECT_1 flag. + * + * Texture lookup + * TODO + */ + +/* ---- OPCODE BITS 127:96 / data DWORD 0 --- */ +#define NV40_VP_INST_VEC_RESULT (1 << 30) +/* uncertain.. */ +#define NV40_VP_INST_COND_UPDATE_ENABLE ((1 << 14)|1<<29) +/* use address reg as index into attribs */ +#define NV40_VP_INST_INDEX_INPUT (1 << 27) +#define NV40_VP_INST_COND_REG_SELECT_1 (1 << 25) +#define NV40_VP_INST_ADDR_REG_SELECT_1 (1 << 24) +#define NV40_VP_INST_SRC2_ABS (1 << 23) +#define NV40_VP_INST_SRC1_ABS (1 << 22) +#define NV40_VP_INST_SRC0_ABS (1 << 21) +#define NV40_VP_INST_VEC_DEST_TEMP_SHIFT 15 +#define NV40_VP_INST_VEC_DEST_TEMP_MASK (0x1F << 15) +#define NV40_VP_INST_COND_TEST_ENABLE (1 << 13) +#define NV40_VP_INST_COND_SHIFT 10 +#define NV40_VP_INST_COND_MASK (0x7 << 10) +# define NV40_VP_INST_COND_FL 0 +# define NV40_VP_INST_COND_LT 1 +# define NV40_VP_INST_COND_EQ 2 +# define NV40_VP_INST_COND_LE 3 +# define NV40_VP_INST_COND_GT 4 +# define NV40_VP_INST_COND_NE 5 +# define NV40_VP_INST_COND_GE 6 +# define NV40_VP_INST_COND_TR 7 +#define NV40_VP_INST_COND_SWZ_X_SHIFT 8 +#define NV40_VP_INST_COND_SWZ_X_MASK (3 << 8) +#define NV40_VP_INST_COND_SWZ_Y_SHIFT 6 +#define NV40_VP_INST_COND_SWZ_Y_MASK (3 << 6) +#define NV40_VP_INST_COND_SWZ_Z_SHIFT 4 +#define NV40_VP_INST_COND_SWZ_Z_MASK (3 << 4) +#define NV40_VP_INST_COND_SWZ_W_SHIFT 2 +#define NV40_VP_INST_COND_SWZ_W_MASK (3 << 2) +#define NV40_VP_INST_COND_SWZ_ALL_SHIFT 2 +#define NV40_VP_INST_COND_SWZ_ALL_MASK (0xFF << 2) +#define NV40_VP_INST_ADDR_SWZ_SHIFT 0 +#define NV40_VP_INST_ADDR_SWZ_MASK (0x03 << 0) +#define NV40_VP_INST0_KNOWN ( \ + NV40_VP_INST_INDEX_INPUT | \ + NV40_VP_INST_COND_REG_SELECT_1 | \ + NV40_VP_INST_ADDR_REG_SELECT_1 | \ + NV40_VP_INST_SRC2_ABS | \ + NV40_VP_INST_SRC1_ABS | \ + NV40_VP_INST_SRC0_ABS | \ + NV40_VP_INST_VEC_DEST_TEMP_MASK | \ + NV40_VP_INST_COND_TEST_ENABLE | \ + NV40_VP_INST_COND_MASK | \ + NV40_VP_INST_COND_SWZ_ALL_MASK | \ + NV40_VP_INST_ADDR_SWZ_MASK) + +/* ---- OPCODE BITS 95:64 / data DWORD 1 --- */ +#define NV40_VP_INST_VEC_OPCODE_SHIFT 22 +#define NV40_VP_INST_VEC_OPCODE_MASK (0x1F << 22) +# define NV40_VP_INST_OP_NOP 0x00 +# define NV40_VP_INST_OP_MOV 0x01 +# define NV40_VP_INST_OP_MUL 0x02 +# define NV40_VP_INST_OP_ADD 0x03 +# define NV40_VP_INST_OP_MAD 0x04 +# define NV40_VP_INST_OP_DP3 0x05 +# define NV40_VP_INST_OP_DPH 0x06 +# define NV40_VP_INST_OP_DP4 0x07 +# define NV40_VP_INST_OP_DST 0x08 +# define NV40_VP_INST_OP_MIN 0x09 +# define NV40_VP_INST_OP_MAX 0x0A +# define NV40_VP_INST_OP_SLT 0x0B +# define NV40_VP_INST_OP_SGE 0x0C +# define NV40_VP_INST_OP_ARL 0x0D +# define NV40_VP_INST_OP_FRC 0x0E +# define NV40_VP_INST_OP_FLR 0x0F +# define NV40_VP_INST_OP_SEQ 0x10 +# define NV40_VP_INST_OP_SFL 0x11 +# define NV40_VP_INST_OP_SGT 0x12 +# define NV40_VP_INST_OP_SLE 0x13 +# define NV40_VP_INST_OP_SNE 0x14 +# define NV40_VP_INST_OP_STR 0x15 +# define NV40_VP_INST_OP_SSG 0x16 +# define NV40_VP_INST_OP_ARR 0x17 +# define NV40_VP_INST_OP_ARA 0x18 +# define NV40_VP_INST_OP_TXL 0x19 +#define NV40_VP_INST_SCA_OPCODE_SHIFT 27 +#define NV40_VP_INST_SCA_OPCODE_MASK (0x1F << 27) +# define NV40_VP_INST_OP_NOP 0x00 +# define NV40_VP_INST_OP_MOV 0x01 +# define NV40_VP_INST_OP_RCP 0x02 +# define NV40_VP_INST_OP_RCC 0x03 +# define NV40_VP_INST_OP_RSQ 0x04 +# define NV40_VP_INST_OP_EXP 0x05 +# define NV40_VP_INST_OP_LOG 0x06 +# define NV40_VP_INST_OP_LIT 0x07 +# define NV40_VP_INST_OP_BRA 0x09 +# define NV40_VP_INST_OP_CAL 0x0B +# define NV40_VP_INST_OP_RET 0x0C +# define NV40_VP_INST_OP_LG2 0x0D +# define NV40_VP_INST_OP_EX2 0x0E +# define NV40_VP_INST_OP_SIN 0x0F +# define NV40_VP_INST_OP_COS 0x10 +# define NV40_VP_INST_OP_PUSHA 0x13 +# define NV40_VP_INST_OP_POPA 0x14 +#define NV40_VP_INST_CONST_SRC_SHIFT 12 +#define NV40_VP_INST_CONST_SRC_MASK (0xFF << 12) +#define NV40_VP_INST_INPUT_SRC_SHIFT 8 +#define NV40_VP_INST_INPUT_SRC_MASK (0x0F << 8) +# define NV40_VP_INST_IN_POS 0 +# define NV40_VP_INST_IN_WEIGHT 1 +# define NV40_VP_INST_IN_NORMAL 2 +# define NV40_VP_INST_IN_COL0 3 +# define NV40_VP_INST_IN_COL1 4 +# define NV40_VP_INST_IN_FOGC 5 +# define NV40_VP_INST_IN_TC0 8 +# define NV40_VP_INST_IN_TC(n) (8+n) +#define NV40_VP_INST_SRC0H_SHIFT 0 +#define NV40_VP_INST_SRC0H_MASK (0xFF << 0) +#define NV40_VP_INST1_KNOWN ( \ + NV40_VP_INST_VEC_OPCODE_MASK | \ + NV40_VP_INST_SCA_OPCODE_MASK | \ + NV40_VP_INST_CONST_SRC_MASK | \ + NV40_VP_INST_INPUT_SRC_MASK | \ + NV40_VP_INST_SRC0H_MASK \ + ) + +/* ---- OPCODE BITS 63:32 / data DWORD 2 --- */ +#define NV40_VP_INST_SRC0L_SHIFT 23 +#define NV40_VP_INST_SRC0L_MASK (0x1FF << 23) +#define NV40_VP_INST_SRC1_SHIFT 6 +#define NV40_VP_INST_SRC1_MASK (0x1FFFF << 6) +#define NV40_VP_INST_SRC2H_SHIFT 0 +#define NV40_VP_INST_SRC2H_MASK (0x3F << 0) +#define NV40_VP_INST_IADDRH_SHIFT 0 +#define NV40_VP_INST_IADDRH_MASK (0x1F << 0) + +/* ---- OPCODE BITS 31:0 / data DWORD 3 --- */ +#define NV40_VP_INST_IADDRL_SHIFT 29 +#define NV40_VP_INST_IADDRL_MASK (7 << 29) +#define NV40_VP_INST_SRC2L_SHIFT 21 +#define NV40_VP_INST_SRC2L_MASK (0x7FF << 21) +#define NV40_VP_INST_SCA_WRITEMASK_SHIFT 17 +#define NV40_VP_INST_SCA_WRITEMASK_MASK (0xF << 17) +# define NV40_VP_INST_SCA_WRITEMASK_X (1 << 20) +# define NV40_VP_INST_SCA_WRITEMASK_Y (1 << 19) +# define NV40_VP_INST_SCA_WRITEMASK_Z (1 << 18) +# define NV40_VP_INST_SCA_WRITEMASK_W (1 << 17) +#define NV40_VP_INST_VEC_WRITEMASK_SHIFT 13 +#define NV40_VP_INST_VEC_WRITEMASK_MASK (0xF << 13) +# define NV40_VP_INST_VEC_WRITEMASK_X (1 << 16) +# define NV40_VP_INST_VEC_WRITEMASK_Y (1 << 15) +# define NV40_VP_INST_VEC_WRITEMASK_Z (1 << 14) +# define NV40_VP_INST_VEC_WRITEMASK_W (1 << 13) +#define NV40_VP_INST_SCA_RESULT (1 << 12) +#define NV40_VP_INST_SCA_DEST_TEMP_SHIFT 7 +#define NV40_VP_INST_SCA_DEST_TEMP_MASK (0x1F << 7) +#define NV40_VP_INST_DEST_SHIFT 2 +#define NV40_VP_INST_DEST_MASK (31 << 2) +# define NV40_VP_INST_DEST_POS 0 +# define NV40_VP_INST_DEST_COL0 1 +# define NV40_VP_INST_DEST_COL1 2 +# define NV40_VP_INST_DEST_BFC0 3 +# define NV40_VP_INST_DEST_BFC1 4 +# define NV40_VP_INST_DEST_FOGC 5 +# define NV40_VP_INST_DEST_PSZ 6 +# define NV40_VP_INST_DEST_TC0 7 +# define NV40_VP_INST_DEST_TC(n) (7+n) +# define NV40_VP_INST_DEST_TEMP 0x1F +#define NV40_VP_INST_INDEX_CONST (1 << 1) +#define NV40_VP_INST_LAST (1 << 0) +#define NV40_VP_INST3_KNOWN ( \ + NV40_VP_INST_SRC2L_MASK |\ + NV40_VP_INST_SCA_WRITEMASK_MASK |\ + NV40_VP_INST_VEC_WRITEMASK_MASK |\ + NV40_VP_INST_SCA_DEST_TEMP_MASK |\ + NV40_VP_INST_DEST_MASK |\ + NV40_VP_INST_INDEX_CONST) + +/* Useful to split the source selection regs into their pieces */ +#define NV40_VP_SRC0_HIGH_SHIFT 9 +#define NV40_VP_SRC0_HIGH_MASK 0x0001FE00 +#define NV40_VP_SRC0_LOW_MASK 0x000001FF +#define NV40_VP_SRC2_HIGH_SHIFT 11 +#define NV40_VP_SRC2_HIGH_MASK 0x0001F800 +#define NV40_VP_SRC2_LOW_MASK 0x000007FF + +/* Source selection - these are the bits you fill NV40_VP_INST_SRCn with */ +#define NV40_VP_SRC_NEGATE (1 << 16) +#define NV40_VP_SRC_SWZ_X_SHIFT 14 +#define NV40_VP_SRC_SWZ_X_MASK (3 << 14) +#define NV40_VP_SRC_SWZ_Y_SHIFT 12 +#define NV40_VP_SRC_SWZ_Y_MASK (3 << 12) +#define NV40_VP_SRC_SWZ_Z_SHIFT 10 +#define NV40_VP_SRC_SWZ_Z_MASK (3 << 10) +#define NV40_VP_SRC_SWZ_W_SHIFT 8 +#define NV40_VP_SRC_SWZ_W_MASK (3 << 8) +#define NV40_VP_SRC_SWZ_ALL_SHIFT 8 +#define NV40_VP_SRC_SWZ_ALL_MASK (0xFF << 8) +#define NV40_VP_SRC_TEMP_SRC_SHIFT 2 +#define NV40_VP_SRC_TEMP_SRC_MASK (0x1F << 2) +#define NV40_VP_SRC_REG_TYPE_SHIFT 0 +#define NV40_VP_SRC_REG_TYPE_MASK (3 << 0) +# define NV40_VP_SRC_REG_TYPE_UNK0 0 +# define NV40_VP_SRC_REG_TYPE_TEMP 1 +# define NV40_VP_SRC_REG_TYPE_INPUT 2 +# define NV40_VP_SRC_REG_TYPE_CONST 3 + + +/* + * Each fragment program opcode appears to be comprised of 4 32-bit values. + * + * 0 - Opcode, output reg/mask, ATTRIB source + * 1 - Source 0 + * 2 - Source 1 + * 3 - Source 2 + * + * There appears to be no special difference between result regs and temp regs. + * result.color == R0.xyzw + * result.depth == R1.z + * When the fragprog contains instructions to write depth, + * NV30_TCL_PRIMITIVE_3D_UNK1D78=0 otherwise it is set to 1. + * + * Constants are inserted directly after the instruction that uses them. + * + * It appears that it's not possible to use two input registers in one + * instruction as the input sourcing is done in the instruction dword + * and not the source selection dwords. As such instructions such as: + * + * ADD result.color, fragment.color, fragment.texcoord[0]; + * + * must be split into two MOV's and then an ADD (nvidia does this) but + * I'm not sure why it's not just one MOV and then source the second input + * in the ADD instruction.. + * + * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary + * negation requires multiplication with a const. + * + * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO and + * SWIZZLE_ONE. + * + * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as + * SWIZZLE_ZERO is implemented simply by not writing to the relevant components + * of the destination. + * + * Looping + * Loops appear to be fairly expensive on NV40 at least, the proprietary + * driver goes to a lot of effort to avoid using the native looping + * instructions. If the total number of *executed* instructions between + * REP/ENDREP or LOOP/ENDLOOP is <=500, the driver will unroll the loop. + * The maximum loop count is 255. + * + * Conditional execution + * TODO + * + * Non-native instructions: + * LIT + * LRP - MAD+MAD + * SUB - ADD, negate second source + * RSQ - LG2 + EX2 + * POW - LG2 + MUL + EX2 + * SCS - COS + SIN + * XPD + * DP2 - MUL + ADD + * NRM + */ + +//== Opcode / Destination selection == +#define NV40_FP_OP_PROGRAM_END (1 << 0) +#define NV40_FP_OP_OUT_REG_SHIFT 1 +#define NV40_FP_OP_OUT_REG_MASK (63 << 1) +/* Needs to be set when writing outputs to get expected result.. */ +#define NV40_FP_OP_OUT_REG_HALF (1 << 7) +#define NV40_FP_OP_COND_WRITE_ENABLE (1 << 8) +#define NV40_FP_OP_OUTMASK_SHIFT 9 +#define NV40_FP_OP_OUTMASK_MASK (0xF << 9) +# define NV40_FP_OP_OUT_X (1 << 9) +# define NV40_FP_OP_OUT_Y (1 <<10) +# define NV40_FP_OP_OUT_Z (1 <<11) +# define NV40_FP_OP_OUT_W (1 <<12) +/* Uncertain about these, especially the input_src values.. it's possible that + * they can be dynamically changed. + */ +#define NV40_FP_OP_INPUT_SRC_SHIFT 13 +#define NV40_FP_OP_INPUT_SRC_MASK (15 << 13) +# define NV40_FP_OP_INPUT_SRC_POSITION 0x0 +# define NV40_FP_OP_INPUT_SRC_COL0 0x1 +# define NV40_FP_OP_INPUT_SRC_COL1 0x2 +# define NV40_FP_OP_INPUT_SRC_FOGC 0x3 +# define NV40_FP_OP_INPUT_SRC_TC0 0x4 +# define NV40_FP_OP_INPUT_SRC_TC(n) (0x4 + n) +# define NV40_FP_OP_INPUT_SRC_FACING 0xE +#define NV40_FP_OP_TEX_UNIT_SHIFT 17 +#define NV40_FP_OP_TEX_UNIT_MASK (0xF << 17) +#define NV40_FP_OP_PRECISION_SHIFT 22 +#define NV40_FP_OP_PRECISION_MASK (3 << 22) +# define NV40_FP_PRECISION_FP32 0 +# define NV40_FP_PRECISION_FP16 1 +# define NV40_FP_PRECISION_FX12 2 +#define NV40_FP_OP_OPCODE_SHIFT 24 +#define NV40_FP_OP_OPCODE_MASK (0x3F << 24) +# define NV40_FP_OP_OPCODE_NOP 0x00 +# define NV40_FP_OP_OPCODE_MOV 0x01 +# define NV40_FP_OP_OPCODE_MUL 0x02 +# define NV40_FP_OP_OPCODE_ADD 0x03 +# define NV40_FP_OP_OPCODE_MAD 0x04 +# define NV40_FP_OP_OPCODE_DP3 0x05 +# define NV40_FP_OP_OPCODE_DP4 0x06 +# define NV40_FP_OP_OPCODE_DST 0x07 +# define NV40_FP_OP_OPCODE_MIN 0x08 +# define NV40_FP_OP_OPCODE_MAX 0x09 +# define NV40_FP_OP_OPCODE_SLT 0x0A +# define NV40_FP_OP_OPCODE_SGE 0x0B +# define NV40_FP_OP_OPCODE_SLE 0x0C +# define NV40_FP_OP_OPCODE_SGT 0x0D +# define NV40_FP_OP_OPCODE_SNE 0x0E +# define NV40_FP_OP_OPCODE_SEQ 0x0F +# define NV40_FP_OP_OPCODE_FRC 0x10 +# define NV40_FP_OP_OPCODE_FLR 0x11 +# define NV40_FP_OP_OPCODE_KIL 0x12 +# define NV40_FP_OP_OPCODE_PK4B 0x13 +# define NV40_FP_OP_OPCODE_UP4B 0x14 +/* DDX/DDY can only write to XY */ +# define NV40_FP_OP_OPCODE_DDX 0x15 +# define NV40_FP_OP_OPCODE_DDY 0x16 +# define NV40_FP_OP_OPCODE_TEX 0x17 +# define NV40_FP_OP_OPCODE_TXP 0x18 +# define NV40_FP_OP_OPCODE_TXD 0x19 +# define NV40_FP_OP_OPCODE_RCP 0x1A +# define NV40_FP_OP_OPCODE_EX2 0x1C +# define NV40_FP_OP_OPCODE_LG2 0x1D +# define NV40_FP_OP_OPCODE_STR 0x20 +# define NV40_FP_OP_OPCODE_SFL 0x21 +# define NV40_FP_OP_OPCODE_COS 0x22 +# define NV40_FP_OP_OPCODE_SIN 0x23 +# define NV40_FP_OP_OPCODE_PK2H 0x24 +# define NV40_FP_OP_OPCODE_UP2H 0x25 +# define NV40_FP_OP_OPCODE_PK4UB 0x27 +# define NV40_FP_OP_OPCODE_UP4UB 0x28 +# define NV40_FP_OP_OPCODE_PK2US 0x29 +# define NV40_FP_OP_OPCODE_UP2US 0x2A +# define NV40_FP_OP_OPCODE_DP2A 0x2E +# define NV40_FP_OP_OPCODE_TXL 0x2F +# define NV40_FP_OP_OPCODE_TXB 0x31 +# define NV40_FP_OP_OPCODE_DIV 0x3A +# define NV40_FP_OP_OPCODE_UNK_LIT 0x3C +/* The use of these instructions appears to be indicated by bit 31 of DWORD 2.*/ +# define NV40_FP_OP_BRA_OPCODE_BRK 0x0 +# define NV40_FP_OP_BRA_OPCODE_CAL 0x1 +# define NV40_FP_OP_BRA_OPCODE_IF 0x2 +# define NV40_FP_OP_BRA_OPCODE_LOOP 0x3 +# define NV40_FP_OP_BRA_OPCODE_REP 0x4 +# define NV40_FP_OP_BRA_OPCODE_RET 0x5 +#define NV40_FP_OP_OUT_SAT (1 << 31) + +/* high order bits of SRC0 */ +#define NV40_FP_OP_OUT_ABS (1 << 29) +#define NV40_FP_OP_COND_SWZ_W_SHIFT 27 +#define NV40_FP_OP_COND_SWZ_W_MASK (3 << 27) +#define NV40_FP_OP_COND_SWZ_Z_SHIFT 25 +#define NV40_FP_OP_COND_SWZ_Z_MASK (3 << 25) +#define NV40_FP_OP_COND_SWZ_Y_SHIFT 23 +#define NV40_FP_OP_COND_SWZ_Y_MASK (3 << 23) +#define NV40_FP_OP_COND_SWZ_X_SHIFT 21 +#define NV40_FP_OP_COND_SWZ_X_MASK (3 << 21) +#define NV40_FP_OP_COND_SWZ_ALL_SHIFT 21 +#define NV40_FP_OP_COND_SWZ_ALL_MASK (0xFF << 21) +#define NV40_FP_OP_COND_SHIFT 18 +#define NV40_FP_OP_COND_MASK (0x07 << 18) +# define NV40_FP_OP_COND_FL 0 +# define NV40_FP_OP_COND_LT 1 +# define NV40_FP_OP_COND_EQ 2 +# define NV40_FP_OP_COND_LE 3 +# define NV40_FP_OP_COND_GT 4 +# define NV40_FP_OP_COND_NE 5 +# define NV40_FP_OP_COND_GE 6 +# define NV40_FP_OP_COND_TR 7 + +/* high order bits of SRC1 */ +#define NV40_FP_OP_OPCODE_IS_BRANCH (1<<31) +#define NV40_FP_OP_DST_SCALE_SHIFT 28 +#define NV40_FP_OP_DST_SCALE_MASK (3 << 28) +#define NV40_FP_OP_DST_SCALE_1X 0 +#define NV40_FP_OP_DST_SCALE_2X 1 +#define NV40_FP_OP_DST_SCALE_4X 2 +#define NV40_FP_OP_DST_SCALE_8X 3 +#define NV40_FP_OP_DST_SCALE_INV_2X 5 +#define NV40_FP_OP_DST_SCALE_INV_4X 6 +#define NV40_FP_OP_DST_SCALE_INV_8X 7 + +/* SRC1 LOOP */ +#define NV40_FP_OP_LOOP_INCR_SHIFT 19 +#define NV40_FP_OP_LOOP_INCR_MASK (0xFF << 19) +#define NV40_FP_OP_LOOP_INDEX_SHIFT 10 +#define NV40_FP_OP_LOOP_INDEX_MASK (0xFF << 10) +#define NV40_FP_OP_LOOP_COUNT_SHIFT 2 +#define NV40_FP_OP_LOOP_COUNT_MASK (0xFF << 2) + +/* SRC1 IF */ +#define NV40_FP_OP_ELSE_ID_SHIFT 2 +#define NV40_FP_OP_ELSE_ID_MASK (0xFF << 2) + +/* SRC1 CAL */ +#define NV40_FP_OP_IADDR_SHIFT 2 +#define NV40_FP_OP_IADDR_MASK (0xFF << 2) + +/* SRC1 REP + * I have no idea why there are 3 count values here.. but they + * have always been filled with the same value in my tests so + * far.. + */ +#define NV40_FP_OP_REP_COUNT1_SHIFT 2 +#define NV40_FP_OP_REP_COUNT1_MASK (0xFF << 2) +#define NV40_FP_OP_REP_COUNT2_SHIFT 10 +#define NV40_FP_OP_REP_COUNT2_MASK (0xFF << 10) +#define NV40_FP_OP_REP_COUNT3_SHIFT 19 +#define NV40_FP_OP_REP_COUNT3_MASK (0xFF << 19) + +/* SRC2 REP/IF */ +#define NV40_FP_OP_END_ID_SHIFT 2 +#define NV40_FP_OP_END_ID_MASK (0xFF << 2) + +// SRC2 high-order +#define NV40_FP_OP_INDEX_INPUT (1 << 30) +#define NV40_FP_OP_ADDR_INDEX_SHIFT 19 +#define NV40_FP_OP_ADDR_INDEX_MASK (0xF << 19) + +//== Register selection == +#define NV40_FP_REG_TYPE_SHIFT 0 +#define NV40_FP_REG_TYPE_MASK (3 << 0) +# define NV40_FP_REG_TYPE_TEMP 0 +# define NV40_FP_REG_TYPE_INPUT 1 +# define NV40_FP_REG_TYPE_CONST 2 +#define NV40_FP_REG_SRC_SHIFT 2 +#define NV40_FP_REG_SRC_MASK (63 << 2) +#define NV40_FP_REG_SRC_HALF (1 << 8) +#define NV40_FP_REG_SWZ_ALL_SHIFT 9 +#define NV40_FP_REG_SWZ_ALL_MASK (255 << 9) +#define NV40_FP_REG_SWZ_X_SHIFT 9 +#define NV40_FP_REG_SWZ_X_MASK (3 << 9) +#define NV40_FP_REG_SWZ_Y_SHIFT 11 +#define NV40_FP_REG_SWZ_Y_MASK (3 << 11) +#define NV40_FP_REG_SWZ_Z_SHIFT 13 +#define NV40_FP_REG_SWZ_Z_MASK (3 << 13) +#define NV40_FP_REG_SWZ_W_SHIFT 15 +#define NV40_FP_REG_SWZ_W_MASK (3 << 15) +# define NV40_FP_SWIZZLE_X 0 +# define NV40_FP_SWIZZLE_Y 1 +# define NV40_FP_SWIZZLE_Z 2 +# define NV40_FP_SWIZZLE_W 3 +#define NV40_FP_REG_NEGATE (1 << 17) + +#define NV40SR_NONE 0 +#define NV40SR_OUTPUT 1 +#define NV40SR_INPUT 2 +#define NV40SR_TEMP 3 +#define NV40SR_CONST 4 + +struct nv40_sreg { + int type; + int index; + + int dst_scale; + + int negate; + int abs; + int swz[4]; + + int cc_update; + int cc_update_reg; + int cc_test; + int cc_test_reg; + int cc_swz[4]; +}; + +static INLINE struct nv40_sreg +nv40_sr(int type, int index) +{ + struct nv40_sreg temp = { + .type = type, + .index = index, + .dst_scale = DEF_SCALE, + .abs = 0, + .negate = 0, + .swz = { 0, 1, 2, 3 }, + .cc_update = 0, + .cc_update_reg = 0, + .cc_test = DEF_CTEST, + .cc_test_reg = 0, + .cc_swz = { 0, 1, 2, 3 }, + }; + return temp; +} + +static INLINE struct nv40_sreg +nv40_sr_swz(struct nv40_sreg src, int x, int y, int z, int w) +{ + struct nv40_sreg dst = src; + + dst.swz[SWZ_X] = src.swz[x]; + dst.swz[SWZ_Y] = src.swz[y]; + dst.swz[SWZ_Z] = src.swz[z]; + dst.swz[SWZ_W] = src.swz[w]; + return dst; +} + +static INLINE struct nv40_sreg +nv40_sr_neg(struct nv40_sreg src) +{ + src.negate = !src.negate; + return src; +} + +static INLINE struct nv40_sreg +nv40_sr_abs(struct nv40_sreg src) +{ + src.abs = 1; + return src; +} + +static INLINE struct nv40_sreg +nv40_sr_scale(struct nv40_sreg src, int scale) +{ + src.dst_scale = scale; + return src; +} + +#endif diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c new file mode 100644 index 00000000000..713f31dbb12 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -0,0 +1,823 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" +#include "nv40_state.h" + +static void * +nv40_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_grobj *curie = nv40->hw->curie; + struct nouveau_stateobj *so = so_new(16, 0); + + if (cso->blend_enable) { + so_method(so, curie, NV40TCL_BLEND_ENABLE, 3); + so_data (so, 1); + so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | + nvgl_blend_func(cso->rgb_src_factor)); + so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | + nvgl_blend_func(cso->rgb_dst_factor)); + so_method(so, curie, NV40TCL_BLEND_EQUATION, 1); + so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | + nvgl_blend_eqn(cso->rgb_func)); + } else { + so_method(so, curie, NV40TCL_BLEND_ENABLE, 1); + so_data (so, 0); + } + + so_method(so, curie, NV40TCL_COLOR_MASK, 1); + so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); + + if (cso->logicop_enable) { + so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); + so_data (so, 1); + so_data (so, nvgl_logicop_func(cso->logicop_func)); + } else { + so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); + so_data (so, 0); + } + + so_method(so, curie, NV40TCL_DITHER_ENABLE, 1); + so_data (so, cso->dither ? 1 : 0); + + return (void *)so; +} + +static void +nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + so_ref(hwcso, &nv40->so_blend); + nv40->dirty |= NV40_NEW_BLEND; +} + +static void +nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV40TCL_TEX_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV40TCL_TEX_WRAP_S_CLAMP; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP; + break; + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV40TCL_TEX_WRAP_S_REPEAT; + break; + } + + return ret >> NV40TCL_TEX_WRAP_S_SHIFT; +} + +static void * +nv40_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv40_sampler_state *ps; + uint32_t filter = 0; + + ps = MALLOC(sizeof(struct nv40_sampler_state)); + + ps->fmt = 0; + if (!cso->normalized_coords) + ps->fmt |= NV40TCL_TEX_FORMAT_RECT; + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) | + (wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_R_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy >= 2.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + + if (cso->max_anisotropy >= 16.0) { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_16X; + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_12X; + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_10X; + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_6X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_4X; + } else { + ps->en |= NV40TCL_TEX_ENABLE_ANISO_2X; + } + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV40TCL_TEX_FILTER_MAG_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV40TCL_TEX_FILTER_MAG_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV40TCL_TEX_FILTER_MIN_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV40TCL_TEX_FILTER_MIN_NEAREST; + break; + } + break; + } + + ps->filt = filter; + + { + float limit; + + limit = CLAMP(cso->lod_bias, -16.0, 15.0); + ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff; + + limit = CLAMP(cso->max_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 7; + + limit = CLAMP(cso->min_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 19; + } + + + if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NEVER; + break; + case PIPE_FUNC_GREATER: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GREATER; + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_EQUAL; + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GEQUAL; + break; + case PIPE_FUNC_LESS: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LESS; + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL; + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LEQUAL; + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_ALWAYS; + break; + default: + break; + } + } + + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, + void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_sampler_state *ps = hwcso; + + nv40->tex_sampler[unit] = ps; + nv40->dirty_samplers |= (1 << unit); +} + +static void +nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, + struct pipe_texture *miptree) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + nv40->tex_miptree[unit] = (struct nv40_miptree *)miptree; + nv40->dirty_samplers |= (1 << unit); +} + +static void * +nv40_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(32, 0); + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + + so_method(so, nv40->hw->curie, NV40TCL_SHADE_MODEL, 1); + so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : + NV40TCL_SHADE_MODEL_SMOOTH); + + so_method(so, nv40->hw->curie, NV40TCL_LINE_WIDTH, 2); + so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); + so_data (so, cso->line_smooth ? 1 : 0); + so_method(so, nv40->hw->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + so_data (so, cso->line_stipple_enable ? 1 : 0); + so_data (so, (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor); + + so_method(so, nv40->hw->curie, NV40TCL_POINT_SIZE, 1); + so_data (so, fui(cso->point_size)); + + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_MODE_FRONT, 6); + if (cso->front_winding == PIPE_WINDING_CCW) { + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV40TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_CW: + so_data(so, NV40TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, 0); + break; + } + so_data(so, NV40TCL_FRONT_FACE_CCW); + } else { + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV40TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_CW: + so_data(so, NV40TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, 0); + break; + } + so_data(so, NV40TCL_FRONT_FACE_CW); + } + so_data(so, cso->poly_smooth ? 1 : 0); + so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); + + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, cso->poly_stipple_enable ? 1 : 0); + + so_method(so, nv40->hw->curie, NV40TCL_POINT_SPRITE, 1); + if (cso->point_sprite) { + unsigned psctl = (1 << 0), i; + + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + psctl |= (1 << (8 + i)); + } + + so_data(so, psctl); + } else { + so_data(so, 0); + } + + return (void *)so; +} + +static void +nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + so_ref(hwcso, &nv40->so_rast); + nv40->dirty |= NV40_NEW_RAST; +} + +static void +nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); +} + +static void * +nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(32, 0); + + so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); + so_data (so, nvgl_comparison_op(cso->depth.func)); + so_data (so, cso->depth.writemask ? 1 : 0); + so_data (so, cso->depth.enabled ? 1 : 0); + + so_method(so, nv40->hw->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_data (so, cso->alpha.enabled ? 1 : 0); + so_data (so, nvgl_comparison_op(cso->alpha.func)); + so_data (so, float_to_ubyte(cso->alpha.ref)); + + if (cso->stencil[0].enabled) { + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_data (so, cso->stencil[0].enabled ? 1 : 0); + so_data (so, cso->stencil[0].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[0].func)); + so_data (so, cso->stencil[0].ref_value); + so_data (so, cso->stencil[0].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); + } else { + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_data (so, 0); + } + + if (cso->stencil[1].enabled) { + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_data (so, cso->stencil[1].enabled ? 1 : 0); + so_data (so, cso->stencil[1].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[1].func)); + so_data (so, cso->stencil[1].ref_value); + so_data (so, cso->stencil[1].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); + } else { + so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_data (so, 0); + } + + return (void *)so; +} + +static void +nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + so_ref(hwcso, &nv40->so_zsa); + nv40->dirty |= NV40_NEW_ZSA; +} + +static void +nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nouveau_stateobj *so = hwcso; + + so_ref(NULL, &so); +} + +static void * +nv40_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv40_vertex_program *vp; + + vp = CALLOC(1, sizeof(struct nv40_vertex_program)); + vp->pipe = cso; + + return (void *)vp; +} + +static void +nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_vertex_program *vp = hwcso; + + nv40->vertprog.current = vp; + nv40->dirty |= NV40_NEW_VERTPROG; +} + +static void +nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_vertex_program *vp = hwcso; + + nv40_vertprog_destroy(nv40, vp); + free(vp); +} + +static void * +nv40_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv40_fragment_program *fp; + + fp = CALLOC(1, sizeof(struct nv40_fragment_program)); + fp->pipe = cso; + + return (void *)fp; +} + +static void +nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_fragment_program *fp = hwcso; + + nv40->fragprog.current = fp; + nv40->dirty |= NV40_NEW_FRAGPROG; +} + +static void +nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_fragment_program *fp = hwcso; + + nv40_fragprog_destroy(nv40, fp); + free(fp); +} + +static void +nv40_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(2, 0); + + so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); + so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0))); + + so_ref(so, &nv40->so_bcol); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_BCOL; +} + +static void +nv40_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + if (shader == PIPE_SHADER_VERTEX) { + nv40->vertprog.constant_buf = buf->buffer; + nv40->dirty |= NV40_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv40->fragprog.constant_buf = buf->buffer; + nv40->dirty |= NV40_NEW_FRAGPROG; + } +} + +static void +nv40_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct pipe_surface *rt[4], *zeta; + uint32_t rt_enable, rt_format, w, h; + int i, colour_format = 0, zeta_format = 0; + struct nouveau_stateobj *so = so_new(64, 10); + unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; + + rt_enable = 0; + for (i = 0; i < 4; i++) { + if (!fb->cbufs[i]) + continue; + + if (colour_format) { + assert(w == fb->cbufs[i]->width); + assert(h == fb->cbufs[i]->height); + assert(colour_format == fb->cbufs[i]->format); + } else { + w = fb->cbufs[i]->width; + h = fb->cbufs[i]->height; + colour_format = fb->cbufs[i]->format; + rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); + rt[i] = fb->cbufs[i]; + } + } + + if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | + NV40TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV40TCL_RT_ENABLE_MRT; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + switch (zeta_format) { + case PIPE_FORMAT_Z16_UNORM: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; + break; + case PIPE_FORMAT_Z24S8_UNORM: + case 0: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; + break; + default: + assert(0); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); + so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); + so_data (so, rt[0]->pitch * rt[0]->cpp); + so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); + so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); + so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, rt[1]->pitch * rt[1]->cpp); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); + so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); + so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); + so_data (so, rt[2]->pitch * rt[2]->cpp); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); + so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); + so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); + so_data (so, rt[3]->pitch * rt[3]->cpp); + } + + if (zeta_format) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); + so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); + so_reloc (so, zeta->buffer, zeta->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); + so_data (so, zeta->pitch * zeta->cpp); + } + + so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); + so_data (so, rt_enable); + so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_data (so, rt_format); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_data (so, ((w - 1) << 16) | 0); + so_data (so, ((h - 1) << 16) | 0); + + so_ref(so, &nv40->so_framebuffer); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_FB; +} + +static void +nv40_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(33, 0); + unsigned i; + + so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, stipple->stipple[i]); + + so_ref(so, &nv40->so_stipple); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_STIPPLE; +} + +static void +nv40_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(3, 0); + + so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); + so_data (so, ((s->maxx - s->minx) << 16) | s->minx); + so_data (so, ((s->maxy - s->miny) << 16) | s->miny); + + so_ref(so, &nv40->so_scissor); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_SCISSOR; +} + +static void +nv40_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_stateobj *so = so_new(9, 0); + + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(vpt->translate[0])); + so_data (so, fui(vpt->translate[1])); + so_data (so, fui(vpt->translate[2])); + so_data (so, fui(vpt->translate[3])); + so_data (so, fui(vpt->scale[0])); + so_data (so, fui(vpt->scale[1])); + so_data (so, fui(vpt->scale[2])); + so_data (so, fui(vpt->scale[3])); + + so_ref(so, &nv40->so_viewport); + so_ref(NULL, &so); + nv40->dirty |= NV40_NEW_VIEWPORT; +} + +static void +nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + nv40->vtxbuf[index] = *vb; + + nv40->dirty |= NV40_NEW_ARRAYS; +} + +static void +nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + nv40->vtxelt[index] = *ve; + + nv40->dirty |= NV40_NEW_ARRAYS; +} + +void +nv40_init_state_functions(struct nv40_context *nv40) +{ + nv40->pipe.create_blend_state = nv40_blend_state_create; + nv40->pipe.bind_blend_state = nv40_blend_state_bind; + nv40->pipe.delete_blend_state = nv40_blend_state_delete; + + nv40->pipe.create_sampler_state = nv40_sampler_state_create; + nv40->pipe.bind_sampler_state = nv40_sampler_state_bind; + nv40->pipe.delete_sampler_state = nv40_sampler_state_delete; + nv40->pipe.set_sampler_texture = nv40_set_sampler_texture; + + nv40->pipe.create_rasterizer_state = nv40_rasterizer_state_create; + nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind; + nv40->pipe.delete_rasterizer_state = nv40_rasterizer_state_delete; + + nv40->pipe.create_depth_stencil_alpha_state = + nv40_depth_stencil_alpha_state_create; + nv40->pipe.bind_depth_stencil_alpha_state = + nv40_depth_stencil_alpha_state_bind; + nv40->pipe.delete_depth_stencil_alpha_state = + nv40_depth_stencil_alpha_state_delete; + + nv40->pipe.create_vs_state = nv40_vp_state_create; + nv40->pipe.bind_vs_state = nv40_vp_state_bind; + nv40->pipe.delete_vs_state = nv40_vp_state_delete; + + nv40->pipe.create_fs_state = nv40_fp_state_create; + nv40->pipe.bind_fs_state = nv40_fp_state_bind; + nv40->pipe.delete_fs_state = nv40_fp_state_delete; + + nv40->pipe.set_blend_color = nv40_set_blend_color; + nv40->pipe.set_clip_state = nv40_set_clip_state; + nv40->pipe.set_constant_buffer = nv40_set_constant_buffer; + nv40->pipe.set_framebuffer_state = nv40_set_framebuffer_state; + nv40->pipe.set_polygon_stipple = nv40_set_polygon_stipple; + nv40->pipe.set_scissor_state = nv40_set_scissor_state; + nv40->pipe.set_viewport_state = nv40_set_viewport_state; + + nv40->pipe.set_vertex_buffer = nv40_set_vertex_buffer; + nv40->pipe.set_vertex_element = nv40_set_vertex_element; +} + diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h new file mode 100644 index 00000000000..e82ab9de98a --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -0,0 +1,80 @@ +#ifndef __NV40_STATE_H__ +#define __NV40_STATE_H__ + +#include "pipe/p_state.h" + +struct nv40_sampler_state { + uint32_t fmt; + uint32_t wrap; + uint32_t en; + uint32_t filt; + uint32_t bcol; +}; + +struct nv40_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv40_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + +struct nv40_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + struct nv40_vertex_program_exec *insns; + unsigned nr_insns; + struct nv40_vertex_program_data *consts; + unsigned nr_consts; + + struct nouveau_resource *exec; + unsigned exec_start; + struct nouveau_resource *data; + unsigned data_start; + unsigned data_start_min; + + uint32_t ir; + uint32_t or; +}; + +struct nv40_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv40_fragment_program { + const struct pipe_shader_state *pipe; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv40_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + struct nouveau_stateobj *so; +}; + +struct nv40_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +#endif diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c new file mode 100644 index 00000000000..a10c9955480 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -0,0 +1,77 @@ +#include "nv40_context.h" +#include "nv40_state.h" + +/* Emit relocs for every referenced buffer. + * + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. These relocs appear in the push buffer as + * NOPs, and will only be turned into state changes if a buffer + * actually moves. + */ +static void +nv40_state_emit_dummy_relocs(struct nv40_context *nv40) +{ + unsigned i; + + so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); + for (i = 0; i < 16; i++) { + if (!(nv40->fp_samplers & (1 << i))) + continue; + so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); + } + so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); +} + +void +nv40_emit_hw_state(struct nv40_context *nv40) +{ + if (nv40->dirty & NV40_NEW_FB) + so_emit(nv40->nvws, nv40->so_framebuffer); + + if (nv40->dirty & NV40_NEW_BLEND) + so_emit(nv40->nvws, nv40->so_blend); + + if (nv40->dirty & NV40_NEW_RAST) + so_emit(nv40->nvws, nv40->so_rast); + + if (nv40->dirty & NV40_NEW_ZSA) + so_emit(nv40->nvws, nv40->so_zsa); + + if (nv40->dirty & NV40_NEW_BCOL) + so_emit(nv40->nvws, nv40->so_bcol); + + if (nv40->dirty & NV40_NEW_SCISSOR) + so_emit(nv40->nvws, nv40->so_scissor); + + if (nv40->dirty & NV40_NEW_VIEWPORT) + so_emit(nv40->nvws, nv40->so_viewport); + + if (nv40->dirty & NV40_NEW_STIPPLE) + so_emit(nv40->nvws, nv40->so_stipple); + + if (nv40->dirty & NV40_NEW_FRAGPROG) { + nv40_fragprog_bind(nv40, nv40->fragprog.current); + /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ + } + + if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { + nv40_fragtex_bind(nv40); + + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (1); + nv40->dirty &= ~NV40_NEW_FRAGPROG; + } + + if (nv40->dirty & NV40_NEW_VERTPROG) { + nv40_vertprog_bind(nv40, nv40->vertprog.current); + nv40->dirty &= ~NV40_NEW_VERTPROG; + } + + nv40->dirty_samplers = 0; + nv40->dirty = 0; + + nv40_state_emit_dummy_relocs(nv40); +} + diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c new file mode 100644 index 00000000000..9726ab4e4dc --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_surface.c @@ -0,0 +1,137 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv40_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/p_tile.h" + +static boolean +nv40_surface_format_supported(struct pipe_context *pipe, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static struct pipe_surface * +nv40_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv40mt->level[level].pitch / ps->cpp; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv40mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv40mt->level[level].image_offset[zslice]; + } else { + ps->offset = nv40mt->level[level].image_offset[0]; + } + + return ps; +} + +static void +nv40_surface_copy(struct pipe_context *pipe, unsigned do_flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct nouveau_winsys *nvws = nv40->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv40_init_surface_functions(struct nv40_context *nv40) +{ + nv40->pipe.is_format_supported = nv40_surface_format_supported; + nv40->pipe.get_tex_surface = nv40_get_tex_surface; + nv40->pipe.surface_copy = nv40_surface_copy; + nv40->pipe.surface_fill = nv40_surface_fill; +} diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c new file mode 100644 index 00000000000..fa827ef0c5d --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -0,0 +1,424 @@ +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" +#include "nv40_state.h" + +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_pushbuf.h" + +static INLINE int +nv40_vbo_ncomp(uint format) +{ + int ncomp = 0; + + if (pf_size_x(format)) ncomp++; + if (pf_size_y(format)) ncomp++; + if (pf_size_z(format)) ncomp++; + if (pf_size_w(format)) ncomp++; + + return ncomp; +} + +static INLINE int +nv40_vbo_type(uint format) +{ + switch (pf_type(format)) { + case PIPE_FORMAT_TYPE_FLOAT: + return NV40TCL_VTXFMT_TYPE_FLOAT; + case PIPE_FORMAT_TYPE_UNORM: + return NV40TCL_VTXFMT_TYPE_UBYTE; + default: + NOUVEAU_ERR("Unknown format 0x%08x\n", format); + return NV40TCL_VTXFMT_TYPE_FLOAT; + } +} + +static boolean +nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, + struct pipe_vertex_element *ve, + struct pipe_vertex_buffer *vb) +{ + struct pipe_winsys *ws = nv40->pipe.winsys; + int type, ncomp; + void *map; + + type = nv40_vbo_type(ve->src_format); + ncomp = nv40_vbo_ncomp(ve->src_format); + + map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); + map += vb->buffer_offset + ve->src_offset; + + switch (type) { + case NV40TCL_VTXFMT_TYPE_FLOAT: + { + float *v = map; + + BEGIN_RING(curie, NV40TCL_VTX_ATTR_4F_X(attrib), 4); + switch (ncomp) { + case 4: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(v[3]); + break; + case 3: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(v[2]); + OUT_RINGf(1.0); + break; + case 2: + OUT_RINGf(v[0]); + OUT_RINGf(v[1]); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + case 1: + OUT_RINGf(v[0]); + OUT_RINGf(0.0); + OUT_RINGf(0.0); + OUT_RINGf(1.0); + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + } + break; + default: + ws->buffer_unmap(ws, vb->buffer); + return FALSE; + } + + ws->buffer_unmap(ws, vb->buffer); + + return TRUE; +} + +static void +nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, + unsigned ib_format) +{ + struct nv40_vertex_program *vp = nv40->vertprog.active; + struct nouveau_stateobj *vtxbuf, *vtxfmt; + unsigned inputs, hw, num_hw; + unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; + + inputs = vp->ir; + for (hw = 0; hw < 16 && inputs; hw++) { + if (inputs & (1 << hw)) { + num_hw = hw; + inputs &= ~(1 << hw); + } + } + num_hw++; + + vtxbuf = so_new(20, 18); + so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + vtxfmt = so_new(17, 0); + so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); + + inputs = vp->ir; + for (hw = 0; hw < num_hw; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + + if (!(inputs & (1 << hw))) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; + } + + ve = &nv40->vtxelt[hw]; + vb = &nv40->vtxbuf[ve->vertex_buffer_index]; + + if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; + } + + so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, + vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + 0, NV40TCL_VTXBUF_ADDRESS_DMA1); + so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | + (nv40_vbo_ncomp(ve->src_format) << + NV40TCL_VTXFMT_SIZE_SHIFT) | + nv40_vbo_type(ve->src_format))); + } + + if (ib) { + so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); + } + + so_emit(nv40->nvws, vtxfmt); + so_emit(nv40->nvws, vtxbuf); + so_ref (vtxbuf, &nv40->so_vtxbuf); + so_ref (NULL, &vtxbuf); + so_ref (NULL, &vtxfmt); +} + +static boolean +nv40_vbo_validate_state(struct nv40_context *nv40, + struct pipe_buffer *ib, unsigned ib_format) +{ + unsigned vdn = nv40->dirty & NV40_NEW_ARRAYS; + + nv40_emit_hw_state(nv40); + if (vdn || ib) { + nv40_vbo_arrays_update(nv40, ib, ib_format); + nv40->dirty &= ~NV40_NEW_ARRAYS; + } + + so_emit_reloc_markers(nv40->nvws, nv40->so_vtxbuf); + + BEGIN_RING(curie, 0x1710, 1); + OUT_RING (0); /* vtx cache flush */ + + return TRUE; +} + +boolean +nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, + unsigned count) +{ + struct nv40_context *nv40 = nv40_context(pipe); + unsigned nr; + boolean ret; + + ret = nv40_vbo_validate_state(nv40, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(curie, NV40TCL_VB_VERTEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + pipe->flush(pipe, 0); + return TRUE; +} + +static INLINE void +nv40_draw_elements_u08(struct nv40_context *nv40, void *ib, + unsigned start, unsigned count) +{ + uint8_t *elts = (uint8_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2047 * 2); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv40_draw_elements_u16(struct nv40_context *nv40, void *ib, + unsigned start, unsigned count) +{ + uint16_t *elts = (uint16_t *)ib + start; + int push, i; + + if (count & 1) { + BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; count--; + } + + while (count) { + push = MIN2(count, 2047 * 2); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + count -= push; + elts += push; + } +} + +static INLINE void +nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, + unsigned start, unsigned count) +{ + uint32_t *elts = (uint32_t *)ib + start; + int push; + + while (count) { + push = MIN2(count, 2047); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U32, push); + OUT_RINGp (elts, push); + + count -= push; + elts += push; + } +} + +static boolean +nv40_draw_elements_inline(struct pipe_context *pipe, + struct pipe_buffer *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + boolean ret; + void *map; + + ret = nv40_vbo_validate_state(nv40, NULL, 0); + if (!ret) { + NOUVEAU_ERR("state validate failed\n"); + return FALSE; + } + + map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); + if (!ib) { + NOUVEAU_ERR("failed mapping ib\n"); + return FALSE; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + switch (ib_size) { + case 1: + nv40_draw_elements_u08(nv40, map, start, count); + break; + case 2: + nv40_draw_elements_u16(nv40, map, start, count); + break; + case 4: + nv40_draw_elements_u32(nv40, map, start, count); + break; + default: + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); + break; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + ws->buffer_unmap(ws, ib); + + return TRUE; +} + +static boolean +nv40_draw_elements_vbo(struct pipe_context *pipe, + struct pipe_buffer *ib, unsigned ib_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv40_context *nv40 = nv40_context(pipe); + unsigned nr, type; + boolean ret; + + switch (ib_size) { + case 2: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; + break; + case 4: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; + break; + default: + NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); + return FALSE; + } + + ret = nv40_vbo_validate_state(nv40, ib, type); + if (!ret) { + NOUVEAU_ERR("failed state validation\n"); + return FALSE; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (count & 0xff); + if (nr) { + BEGIN_RING(curie, NV40TCL_VB_INDEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } + + nr = count >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(curie, NV40TCL_VB_INDEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + return TRUE; +} + +boolean +nv40_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + /* 0x4497 doesn't support real index buffers, and there doesn't appear + * to be support on any chipset for 8-bit indices. + */ + if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { + nv40_draw_elements_inline(pipe, indexBuffer, indexSize, + mode, start, count); + } else { + nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, + mode, start, count); + } + + pipe->flush(pipe, 0); + return TRUE; +} + + diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c new file mode 100644 index 00000000000..9f4738b8308 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -0,0 +1,790 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" + +#include "nv40_context.h" +#include "nv40_state.h" + +/* TODO (at least...): + * 1. Indexed consts + ARL + * 2. Arb. swz/negation + * 3. NV_vp11, NV_vp2, NV_vp3 features + * - extra arith opcodes + * - branching + * - texture sampling + * - indexed attribs + * - indexed results + * 4. bugs + */ + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 8 +#define MASK_Y 4 +#define MASK_Z 2 +#define MASK_W 1 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE 0 +#define DEF_CTEST 0 +#include "nv40_shader.h" + +#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv40_sr_neg((s)) +#define abs(s) nv40_sr_abs((s)) + +struct nv40_vpc { + struct nv40_vertex_program *vp; + + struct nv40_vertex_program_exec *vpi; + + unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; + + int high_temp; + int temp_temp_count; + + struct nv40_sreg *imm; + unsigned nr_imm; +}; + +static struct nv40_sreg +temp(struct nv40_vpc *vpc) +{ + int idx; + + idx = vpc->temp_temp_count++; + idx += vpc->high_temp + 1; + return nv40_sr(NV40SR_TEMP, idx); +} + +static struct nv40_sreg +constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w) +{ + struct nv40_vertex_program *vp = vpc->vp; + struct nv40_vertex_program_data *vpd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < vp->nr_consts; idx++) { + if (vp->consts[idx].index == pipe) + return nv40_sr(NV40SR_CONST, idx); + } + } + + idx = vp->nr_consts++; + vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); + vpd = &vp->consts[idx]; + + vpd->index = pipe; + vpd->value[0] = x; + vpd->value[1] = y; + vpd->value[2] = z; + vpd->value[3] = w; + return nv40_sr(NV40SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv40_vp_arith((cc), (s), NV40_VP_INST_##o, (d), (m), (s0), (s1), (s2)) + +static void +emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src) +{ + struct nv40_vertex_program *vp = vpc->vp; + uint32_t sr = 0; + + switch (src.type) { + case NV40SR_TEMP: + sr |= (NV40_VP_SRC_REG_TYPE_TEMP << NV40_VP_SRC_REG_TYPE_SHIFT); + sr |= (src.index << NV40_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV40SR_INPUT: + sr |= (NV40_VP_SRC_REG_TYPE_INPUT << + NV40_VP_SRC_REG_TYPE_SHIFT); + vp->ir |= (1 << src.index); + hw[1] |= (src.index << NV40_VP_INST_INPUT_SRC_SHIFT); + break; + case NV40SR_CONST: + sr |= (NV40_VP_SRC_REG_TYPE_CONST << + NV40_VP_SRC_REG_TYPE_SHIFT); + assert(vpc->vpi->const_index == -1 || + vpc->vpi->const_index == src.index); + vpc->vpi->const_index = src.index; + break; + case NV40SR_NONE: + sr |= (NV40_VP_SRC_REG_TYPE_INPUT << + NV40_VP_SRC_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV40_VP_SRC_NEGATE; + + if (src.abs) + hw[0] |= (1 << (21 + pos)); + + sr |= ((src.swz[0] << NV40_VP_SRC_SWZ_X_SHIFT) | + (src.swz[1] << NV40_VP_SRC_SWZ_Y_SHIFT) | + (src.swz[2] << NV40_VP_SRC_SWZ_Z_SHIFT) | + (src.swz[3] << NV40_VP_SRC_SWZ_W_SHIFT)); + + switch (pos) { + case 0: + hw[1] |= ((sr & NV40_VP_SRC0_HIGH_MASK) >> + NV40_VP_SRC0_HIGH_SHIFT) << NV40_VP_INST_SRC0H_SHIFT; + hw[2] |= (sr & NV40_VP_SRC0_LOW_MASK) << + NV40_VP_INST_SRC0L_SHIFT; + break; + case 1: + hw[2] |= sr << NV40_VP_INST_SRC1_SHIFT; + break; + case 2: + hw[2] |= ((sr & NV40_VP_SRC2_HIGH_MASK) >> + NV40_VP_SRC2_HIGH_SHIFT) << NV40_VP_INST_SRC2H_SHIFT; + hw[3] |= (sr & NV40_VP_SRC2_LOW_MASK) << + NV40_VP_INST_SRC2L_SHIFT; + break; + default: + assert(0); + } +} + +static void +emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) +{ + struct nv40_vertex_program *vp = vpc->vp; + + switch (dst.type) { + case NV40SR_TEMP: + hw[3] |= NV40_VP_INST_DEST_MASK; + if (slot == 0) { + hw[0] |= (dst.index << + NV40_VP_INST_VEC_DEST_TEMP_SHIFT); + } else { + hw[3] |= (dst.index << + NV40_VP_INST_SCA_DEST_TEMP_SHIFT); + } + break; + case NV40SR_OUTPUT: + switch (dst.index) { + case NV40_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; + case NV40_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; + case NV40_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; + case NV40_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; + case NV40_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; + case NV40_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; + case NV40_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; + case NV40_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; + case NV40_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; + case NV40_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; + case NV40_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; + case NV40_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; + case NV40_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; + case NV40_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + default: + break; + } + + hw[3] |= (dst.index << NV40_VP_INST_DEST_SHIFT); + if (slot == 0) { + hw[0] |= NV40_VP_INST_VEC_RESULT; + hw[0] |= NV40_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); + } else { + hw[3] |= NV40_VP_INST_SCA_RESULT; + hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; + } + break; + default: + assert(0); + } +} + +static void +nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op, + struct nv40_sreg dst, int mask, + struct nv40_sreg s0, struct nv40_sreg s1, + struct nv40_sreg s2) +{ + struct nv40_vertex_program *vp = vpc->vp; + uint32_t *hw; + + vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); + vpc->vpi = &vp->insns[vp->nr_insns - 1]; + memset(vpc->vpi, 0, sizeof(*vpc->vpi)); + vpc->vpi->const_index = -1; + + hw = vpc->vpi->data; + + hw[0] |= (NV40_VP_INST_COND_TR << NV40_VP_INST_COND_SHIFT); + hw[0] |= ((0 << NV40_VP_INST_COND_SWZ_X_SHIFT) | + (1 << NV40_VP_INST_COND_SWZ_Y_SHIFT) | + (2 << NV40_VP_INST_COND_SWZ_Z_SHIFT) | + (3 << NV40_VP_INST_COND_SWZ_W_SHIFT)); + + if (slot == 0) { + hw[1] |= (op << NV40_VP_INST_VEC_OPCODE_SHIFT); + hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; + hw[3] |= (mask << NV40_VP_INST_VEC_WRITEMASK_SHIFT); + } else { + hw[1] |= (op << NV40_VP_INST_SCA_OPCODE_SHIFT); + hw[0] |= (NV40_VP_INST_VEC_DEST_TEMP_MASK | (1 << 20)); + hw[3] |= (mask << NV40_VP_INST_SCA_WRITEMASK_SHIFT); + } + + emit_dst(vpc, hw, slot, dst); + emit_src(vpc, hw, 0, s0); + emit_src(vpc, hw, 1, s1); + emit_src(vpc, hw, 2, s2); +} + +static INLINE struct nv40_sreg +tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv40_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv40_sr(NV40SR_INPUT, fsrc->SrcRegister.Index); + break; + case TGSI_FILE_CONSTANT: + src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + break; + case TGSI_FILE_IMMEDIATE: + src = vpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + if (vpc->high_temp < fsrc->SrcRegister.Index) + vpc->high_temp = fsrc->SrcRegister.Index; + src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv40_sreg +tgsi_dst(struct nv40_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + struct nv40_sreg dst; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + dst = nv40_sr(NV40SR_OUTPUT, + vpc->output_map[fdst->DstRegister.Index]); + + break; + case TGSI_FILE_TEMPORARY: + dst = nv40_sr(NV40SR_TEMP, fdst->DstRegister.Index); + if (vpc->high_temp < dst.index) + vpc->high_temp = dst.index; + break; + default: + NOUVEAU_ERR("bad dst file\n"); + break; + } + + return dst; +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, + const struct tgsi_full_instruction *finst) +{ + struct nv40_sreg src[3], dst, tmp; + struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); + int mask; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + vpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(vpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + /*XXX: index comparison is broken now that consts come from + * two different register files. + */ + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); + break; + case TGSI_OPCODE_ARL: + arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DST: + arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_EXP: + arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_FLR: + arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_LG2: + arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LIT: + arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LOG: + arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_MAD: + arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(vpc); + arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, + swz(src[0], X, X, X, X)); + arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(vpc, 1, OP_EX2, dst, mask, none, none, + swz(tmp, X, X, X, X)); + break; + case TGSI_OPCODE_RCP: + arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_RET: + break; + case TGSI_OPCODE_RSQ: + arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_SGE: + arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); + break; + case TGSI_OPCODE_XPD: + tmp = temp(vpc); + arith(vpc, 0, OP_MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV40_VP_INST_DEST_POS; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV40_VP_INST_DEST_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV40_VP_INST_DEST_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_BCOLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV40_VP_INST_DEST_BFC0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV40_VP_INST_DEST_BFC1; + } else { + NOUVEAU_ERR("bad bcolour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV40_VP_INST_DEST_FOGC; + break; + case TGSI_SEMANTIC_PSIZE: + hw = NV40_VP_INST_DEST_PSZ; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV40_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + vpc->output_map[fdec->u.DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv40_vertprog_prepare(struct nv40_vpc *vpc) +{ + struct tgsi_parse_context p; + int nr_imm = 0; + + tgsi_parse_init(&p, vpc->vp->pipe->tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch(tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + nr_imm++; + break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (nr_imm) { + vpc->imm = CALLOC(nr_imm, sizeof(struct nv40_sreg)); + assert(vpc->imm); + } + + return TRUE; +} + +void +nv40_vertprog_translate(struct nv40_context *nv40, + struct nv40_vertex_program *vp) +{ + struct tgsi_parse_context parse; + struct nv40_vpc *vpc = NULL; + + vpc = CALLOC(1, sizeof(struct nv40_vpc)); + if (!vpc) + return; + vpc->vp = vp; + vpc->high_temp = -1; + + if (!nv40_vertprog_prepare(vpc)) { + free(vpc); + return; + } + + tgsi_parse_init(&parse, vp->pipe->tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_OUTPUT: + if (!nv40_vertprog_parse_decl_output(vpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); +// assert(imm->Immediate.Size == 4); + vpc->imm[vpc->nr_imm++] = + constant(vpc, -1, + imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + finst = &parse.FullToken.FullInstruction; + if (!nv40_vertprog_parse_instruction(vpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + vp->insns[vp->nr_insns - 1].data[3] |= NV40_VP_INST_LAST; + vp->translated = TRUE; +out_err: + tgsi_parse_free(&parse); + free(vpc); +} + +void +nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) +{ + struct nouveau_winsys *nvws = nv40->nvws; + struct pipe_winsys *ws = nv40->pipe.winsys; + boolean upload_code = FALSE, upload_data = FALSE; + int i; + + /* Translate TGSI shader into hw bytecode */ + if (!vp->translated) { + nv40_vertprog_translate(nv40, vp); + if (!vp->translated) + assert(0); + } + + /* Allocate hw vtxprog exec slots */ + if (!vp->exec) { + struct nouveau_resource *heap = nv40->hw->vp_exec_heap; + uint vplen = vp->nr_insns; + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { + while (heap->next && heap->size < vplen) { + struct nv40_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->exec); + } + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) + assert(0); + } + + upload_code = TRUE; + } + + /* Allocate hw vtxprog const slots */ + if (vp->nr_consts && !vp->data) { + struct nouveau_resource *heap = nv40->hw->vp_data_heap; + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { + while (heap->next && heap->size < vp->nr_consts) { + struct nv40_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) + assert(0); + } + + /*XXX: handle this some day */ + assert(vp->data->start >= vp->data_start_min); + + upload_data = TRUE; + if (vp->data_start != vp->data->start) + upload_code = TRUE; + } + + /* If exec or data segments moved we need to patch the program to + * fixup offsets and register IDs. + */ + if (vp->exec_start != vp->exec->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv40_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->has_branch_offset) { + assert(0); + } + } + + vp->exec_start = vp->exec->start; + } + + if (vp->nr_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv40_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->const_index >= 0) { + vpi->data[1] &= ~NV40_VP_INST_CONST_SRC_MASK; + vpi->data[1] |= + (vpi->const_index + vp->data->start) << + NV40_VP_INST_CONST_SRC_SHIFT; + + } + } + + vp->data_start = vp->data->start; + } + + /* Update + Upload constant values */ + if (vp->nr_consts) { + float *map = NULL; + + if (nv40->vertprog.constant_buf) { + map = ws->buffer_map(ws, nv40->vertprog.constant_buf, + PIPE_BUFFER_USAGE_CPU_READ); + } + + for (i = 0; i < vp->nr_consts; i++) { + struct nv40_vertex_program_data *vpd = &vp->consts[i]; + + if (vpd->index >= 0) { + if (!upload_data && + !memcmp(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float))) + continue; + memcpy(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float)); + } + + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (i + vp->data->start); + OUT_RINGp ((uint32_t *)vpd->value, 4); + } + + if (map) { + ws->buffer_unmap(ws, nv40->vertprog.constant_buf); + } + } + + /* Upload vtxprog */ + if (upload_code) { +#if 0 + for (i = 0; i < vp->nr_insns; i++) { + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); + NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); + } +#endif + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (vp->exec->start); + for (i = 0; i < vp->nr_insns; i++) { + BEGIN_RING(curie, NV40TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (vp->insns[i].data, 4); + } + } + + BEGIN_RING(curie, NV40TCL_VP_START_FROM_ID, 1); + OUT_RING (vp->exec->start); + BEGIN_RING(curie, NV40TCL_VP_ATTRIB_EN, 2); + OUT_RING (vp->ir); + OUT_RING (vp->or); + + nv40->vertprog.active = vp; +} + +void +nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp) +{ + if (vp->nr_consts) + free(vp->consts); + if (vp->nr_insns) + free(vp->insns); +} + diff --git a/src/gallium/drivers/nv50/Makefile b/src/gallium/drivers/nv50/Makefile new file mode 100644 index 00000000000..68eb49ff2a3 --- /dev/null +++ b/src/gallium/drivers/nv50/Makefile @@ -0,0 +1,25 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv50 + +DRIVER_SOURCES = \ + nv50_clear.c \ + nv50_context.c \ + nv50_draw.c \ + nv50_miptree.c \ + nv50_query.c \ + nv50_state.c \ + nv50_surface.c \ + nv50_vbo.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/drivers/nv50/nv50_clear.c b/src/gallium/drivers/nv50/nv50_clear.c new file mode 100644 index 00000000000..552b92f72e2 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv50_context.h" + +void +nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c new file mode 100644 index 00000000000..3c5a54bfd38 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -0,0 +1,202 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" + +static boolean +nv50_is_format_supported(struct pipe_context *pipe, enum pipe_format format, + uint type) +{ + return FALSE; +} + +static const char * +nv50_get_name(struct pipe_context *pipe) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv50->chipset); + return buffer; +} + +static const char * +nv50_get_vendor(struct pipe_context *pipe) +{ + return "nouveau"; +} + +static int +nv50_get_param(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 32; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 0; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 8; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv50_get_paramf(struct pipe_context *pipe, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv50_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv50->sync, 0); + BEGIN_RING(tesla, 0x104, 1); + OUT_RING (0); + BEGIN_RING(tesla, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv50->sync, 0, 0, 2000); +} + +static void +nv50_destroy(struct pipe_context *pipe) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + + draw_destroy(nv50->draw); + free(nv50); +} + +static boolean +nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) +{ + struct nouveau_winsys *nvws = nv50->nvws; + int ret; + + if ((ret = nvws->grobj_alloc(nvws, tesla_class, &nv50->tesla))) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + BEGIN_RING(tesla, NV50TCL_DMA_NOTIFY, 1); + OUT_RING (nv50->sync->handle); + + FIRE_RING (); + return TRUE; +} + +#define GRCLASS5097_CHIPSETS 0x00000000 +#define GRCLASS8297_CHIPSETS 0x00000010 +struct pipe_context * +nv50_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv50_context *nv50; + int tesla_class, ret; + + if ((chipset & 0xf0) != 0x50 && (chipset & 0xf0) != 0x80) { + NOUVEAU_ERR("Not a G8x chipset\n"); + return NULL; + } + + if (GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) { + tesla_class = 0x5097; + } else + if (GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) { + tesla_class = 0x8297; + } else { + NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset); + return NULL; + } + + nv50 = CALLOC_STRUCT(nv50_context); + if (!nv50) + return NULL; + nv50->chipset = chipset; + nv50->nvws = nvws; + + if ((ret = nvws->notifier_alloc(nvws, 1, &nv50->sync))) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + free(nv50); + return NULL; + } + + if (!nv50_init_hwctx(nv50, tesla_class)) { + free(nv50); + return NULL; + } + + nv50->pipe.winsys = pipe_winsys; + + nv50->pipe.destroy = nv50_destroy; + nv50->pipe.is_format_supported = nv50_is_format_supported; + nv50->pipe.get_name = nv50_get_name; + nv50->pipe.get_vendor = nv50_get_vendor; + nv50->pipe.get_param = nv50_get_param; + nv50->pipe.get_paramf = nv50_get_paramf; + + nv50->pipe.draw_arrays = nv50_draw_arrays; + nv50->pipe.draw_elements = nv50_draw_elements; + nv50->pipe.clear = nv50_clear; + + nv50->pipe.flush = nv50_flush; + + nv50_init_miptree_functions(nv50); + nv50_init_surface_functions(nv50); + nv50_init_state_functions(nv50); + nv50_init_query_functions(nv50); + + nv50->draw = draw_create(); + assert(nv50->draw); + draw_set_rasterize_stage(nv50->draw, nv50_draw_render_stage(nv50)); + + return &nv50->pipe; +} + + diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h new file mode 100644 index 00000000000..b99254f6191 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -0,0 +1,57 @@ +#ifndef __NV50_CONTEXT_H__ +#define __NV50_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv50_context *ctx = nv50 +#include "nouveau/nouveau_push.h" + +#include "nv50_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +struct nv50_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct draw_context *draw; + + int chipset; + struct nouveau_grobj *tesla; + struct nouveau_notifier *sync; +}; + + +extern void nv50_init_miptree_functions(struct nv50_context *nv50); +extern void nv50_init_surface_functions(struct nv50_context *nv50); +extern void nv50_init_state_functions(struct nv50_context *nv50); +extern void nv50_init_query_functions(struct nv50_context *nv50); + +/* nv50_draw.c */ +extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50); + +/* nv50_vbo.c */ +extern boolean nv50_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv50_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned mode, unsigned start, + unsigned count); + +/* nv50_clear.c */ +extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +#endif diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c new file mode 100644 index 00000000000..c6ed6838c6a --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_draw.c @@ -0,0 +1,55 @@ +#include "draw/draw_private.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" + +struct nv50_draw_stage { + struct draw_stage draw; + struct nv50_context *nv50; +}; + +static void +nv50_draw_point(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_line(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_tri(struct draw_stage *draw, struct prim_header *prim) +{ + NOUVEAU_ERR("\n"); +} + +static void +nv50_draw_flush(struct draw_stage *draw, unsigned flags) +{ +} + +static void +nv50_draw_reset_stipple_counter(struct draw_stage *draw) +{ + NOUVEAU_ERR("\n"); +} + +struct draw_stage * +nv50_draw_render_stage(struct nv50_context *nv50) +{ + struct nv50_draw_stage *nv50draw = CALLOC_STRUCT(nv50_draw_stage); + + nv50draw->nv50 = nv50; + nv50draw->draw.draw = nv50->draw; + nv50draw->draw.point = nv50_draw_point; + nv50draw->draw.line = nv50_draw_line; + nv50draw->draw.tri = nv50_draw_tri; + nv50draw->draw.flush = nv50_draw_flush; + nv50draw->draw.reset_stipple_counter = nv50_draw_reset_stipple_counter; + + return &nv50draw->draw; +} + diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c new file mode 100644 index 00000000000..0c034ed4387 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -0,0 +1,25 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" + +static struct pipe_texture * +nv50_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) +{ + NOUVEAU_ERR("unimplemented\n"); + return NULL; +} + +static void +nv50_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +void +nv50_init_miptree_functions(struct nv50_context *nv50) +{ + nv50->pipe.texture_create = nv50_miptree_create; + nv50->pipe.texture_release = nv50_miptree_release; +} diff --git a/src/gallium/drivers/nv50/nv50_query.c b/src/gallium/drivers/nv50/nv50_query.c new file mode 100644 index 00000000000..d8c3491c2c5 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_query.c @@ -0,0 +1,47 @@ +#include "pipe/p_context.h" + +#include "nv50_context.h" + +static struct pipe_query * +nv50_query_create(struct pipe_context *pipe, unsigned type) +{ + NOUVEAU_ERR("unimplemented\n"); + return NULL; +} + +static void +nv50_query_destroy(struct pipe_context *pipe, struct pipe_query *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +static void +nv50_query_begin(struct pipe_context *pipe, struct pipe_query *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +static void +nv50_query_end(struct pipe_context *pipe, struct pipe_query *q) +{ + NOUVEAU_ERR("unimplemented\n"); +} + +static boolean +nv50_query_result(struct pipe_context *pipe, struct pipe_query *q, + boolean wait, uint64 *result) +{ + NOUVEAU_ERR("unimplemented\n"); + *result = 0xdeadcafe; + return TRUE; +} + +void +nv50_init_query_functions(struct nv50_context *nv50) +{ + nv50->pipe.create_query = nv50_query_create; + nv50->pipe.destroy_query = nv50_query_destroy; + nv50->pipe.begin_query = nv50_query_begin; + nv50->pipe.end_query = nv50_query_end; + nv50->pipe.get_query_result = nv50_query_result; +} diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c new file mode 100644 index 00000000000..99dcab51b26 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -0,0 +1,213 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" +#include "nv50_state.h" + +static void * +nv50_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + return NULL; +} + +static void +nv50_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + return NULL; +} + +static void +nv50_sampler_state_bind(struct pipe_context *pipe, unsigned unit, + void *hwcso) +{ +} + +static void +nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_set_sampler_texture(struct pipe_context *pipe, unsigned unit, + struct pipe_texture *pt) +{ +} + +static void * +nv50_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + return NULL; +} + +static void +nv50_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + return NULL; +} + +static void +nv50_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + return NULL; +} + +static void +nv50_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void * +nv50_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + return NULL; +} + +static void +nv50_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ +} + +static void +nv50_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ +} + +static void +nv50_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ +} + +static void +nv50_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ +} + +static void +nv50_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ +} + +static void +nv50_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ +} + +static void +nv50_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ +} + +static void +nv50_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ +} + +static void +nv50_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ +} + +void +nv50_init_state_functions(struct nv50_context *nv50) +{ + nv50->pipe.create_blend_state = nv50_blend_state_create; + nv50->pipe.bind_blend_state = nv50_blend_state_bind; + nv50->pipe.delete_blend_state = nv50_blend_state_delete; + + nv50->pipe.create_sampler_state = nv50_sampler_state_create; + nv50->pipe.bind_sampler_state = nv50_sampler_state_bind; + nv50->pipe.delete_sampler_state = nv50_sampler_state_delete; + nv50->pipe.set_sampler_texture = nv50_set_sampler_texture; + + nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create; + nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; + nv50->pipe.delete_rasterizer_state = nv50_rasterizer_state_delete; + + nv50->pipe.create_depth_stencil_alpha_state = + nv50_depth_stencil_alpha_state_create; + nv50->pipe.bind_depth_stencil_alpha_state = + nv50_depth_stencil_alpha_state_bind; + nv50->pipe.delete_depth_stencil_alpha_state = + nv50_depth_stencil_alpha_state_delete; + + nv50->pipe.create_vs_state = nv50_vp_state_create; + nv50->pipe.bind_vs_state = nv50_vp_state_bind; + nv50->pipe.delete_vs_state = nv50_vp_state_delete; + + nv50->pipe.create_fs_state = nv50_fp_state_create; + nv50->pipe.bind_fs_state = nv50_fp_state_bind; + nv50->pipe.delete_fs_state = nv50_fp_state_delete; + + nv50->pipe.set_blend_color = nv50_set_blend_color; + nv50->pipe.set_clip_state = nv50_set_clip_state; + nv50->pipe.set_constant_buffer = nv50_set_constant_buffer; + nv50->pipe.set_framebuffer_state = nv50_set_framebuffer_state; + nv50->pipe.set_polygon_stipple = nv50_set_polygon_stipple; + nv50->pipe.set_scissor_state = nv50_set_scissor_state; + nv50->pipe.set_viewport_state = nv50_set_viewport_state; + + nv50->pipe.set_vertex_buffer = nv50_set_vertex_buffer; + nv50->pipe.set_vertex_element = nv50_set_vertex_element; +} + diff --git a/src/gallium/drivers/nv50/nv50_state.h b/src/gallium/drivers/nv50/nv50_state.h new file mode 100644 index 00000000000..a3b781d4c61 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_state.h @@ -0,0 +1,7 @@ +#ifndef __NV50_STATE_H__ +#define __NV50_STATE_H__ + +#include "pipe/p_state.h" + + +#endif diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c new file mode 100644 index 00000000000..ca92ff02b89 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -0,0 +1,75 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv50_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/p_tile.h" + +static struct pipe_surface * +nv50_get_tex_surface(struct pipe_context *pipe, + struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + NOUVEAU_ERR("unimplemented\n"); + return NULL; +} + +static void +nv50_surface_copy(struct pipe_context *pipe, unsigned flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv50_context *nv50 = (struct nv50_context *)pipe; + struct nouveau_winsys *nvws = nv50->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv50_init_surface_functions(struct nv50_context *nv50) +{ + nv50->pipe.get_tex_surface = nv50_get_tex_surface; + nv50->pipe.surface_copy = nv50_surface_copy; + nv50->pipe.surface_fill = nv50_surface_fill; +} diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c new file mode 100644 index 00000000000..6c0dc23a439 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -0,0 +1,24 @@ +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" +#include "nv50_state.h" + +boolean +nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, + unsigned count) +{ + NOUVEAU_ERR("unimplemented\n"); + return TRUE; +} + +boolean +nv50_draw_elements(struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, unsigned indexSize, + unsigned mode, unsigned start, unsigned count) +{ + NOUVEAU_ERR("unimplemented\n"); + return TRUE; +} + diff --git a/src/gallium/winsys/dri/nouveau/Makefile b/src/gallium/winsys/dri/nouveau/Makefile new file mode 100644 index 00000000000..b463f218fd5 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/Makefile @@ -0,0 +1,43 @@ + +TOP = ../../../../.. +include $(TOP)/configs/current + +LIBNAME = nouveau_dri.so + +MINIGLX_SOURCES = + +PIPE_DRIVERS = \ + $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ + $(TOP)/src/gallium/drivers/nv30/libnv30.a \ + $(TOP)/src/gallium/drivers/nv40/libnv40.a \ + $(TOP)/src/gallium/drivers/nv50/libnv50.a + +DRIVER_SOURCES = \ + nouveau_bo.c \ + nouveau_channel.c \ + nouveau_context.c \ + nouveau_device.c \ + nouveau_dma.c \ + nouveau_fence.c \ + nouveau_grobj.c \ + nouveau_lock.c \ + nouveau_notifier.c \ + nouveau_pushbuf.c \ + nouveau_resource.c \ + nouveau_screen.c \ + nouveau_swapbuffers.c \ + nouveau_winsys.c \ + nouveau_winsys_pipe.c \ + nouveau_winsys_softpipe.c \ + nv04_surface.c \ + nv50_surface.c + +C_SOURCES = \ + $(COMMON_GALLIUM_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../Makefile.template + +symlinks: diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c new file mode 100644 index 00000000000..6887ffa6886 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -0,0 +1,402 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +static void +nouveau_mem_free(struct nouveau_device *dev, struct drm_nouveau_mem_alloc *ma, + void **map) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_mem_free mf; + + if (map && *map) { + drmUnmap(*map, ma->size); + *map = NULL; + } + + if (ma->size) { + mf.offset = ma->offset; + mf.flags = ma->flags; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, + &mf, sizeof(mf)); + ma->size = 0; + } +} + +static int +nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, + uint32_t flags, struct drm_nouveau_mem_alloc *ma, void **map) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + int ret; + + ma->alignment = align; + ma->size = size; + ma->flags = flags; + if (map) + ma->flags |= NOUVEAU_MEM_MAPPED; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, ma, + sizeof(struct drm_nouveau_mem_alloc)); + if (ret) + return ret; + + if (map) { + ret = drmMap(nvdev->fd, ma->map_handle, ma->size, map); + if (ret) { + *map = NULL; + nouveau_mem_free(dev, ma, map); + return ret; + } + } + + return 0; +} + +static void +nouveau_bo_tmp_del(void *priv) +{ + struct nouveau_resource *r = priv; + + nouveau_fence_ref(NULL, (struct nouveau_fence **)&r->priv); + nouveau_resource_free(&r); +} + +static struct nouveau_resource * +nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, + struct nouveau_fence *fence) +{ + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_resource *r = NULL; + struct nouveau_fence *ref = NULL; + + if (fence) + nouveau_fence_ref(fence, &ref); + else + nouveau_fence_new(chan, &ref); + assert(ref); + + while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) { + nouveau_fence_flush(chan); + } + nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r); + + return r; +} + +int +nouveau_bo_init(struct nouveau_device *dev) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + int ret; + + ret = nouveau_mem_alloc(dev, 128*1024, 0, NOUVEAU_MEM_AGP | + NOUVEAU_MEM_PCI, &nvdev->sa, &nvdev->sa_map); + if (ret) + return ret; + + ret = nouveau_resource_init(&nvdev->sa_heap, 0, nvdev->sa.size); + if (ret) { + nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); + return ret; + } + + return 0; +} + +void +nouveau_bo_takedown(struct nouveau_device *dev) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); +} + +int +nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, + int size, struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo; + int ret; + + if (!dev || !bo || *bo) + return -EINVAL; + + nvbo = calloc(1, sizeof(struct nouveau_bo_priv)); + if (!nvbo) + return -ENOMEM; + nvbo->base.device = dev; + nvbo->base.size = size; + nvbo->base.handle = bo_to_ptr(nvbo); + nvbo->drm.alignment = align; + nvbo->refcount = 1; + + ret = nouveau_bo_set_status(&nvbo->base, flags); + if (ret) { + free(nvbo); + return ret; + } + + *bo = &nvbo->base; + return 0; +} + +int +nouveau_bo_user(struct nouveau_device *dev, void *ptr, int size, + struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo; + + if (!dev || !bo || *bo) + return -EINVAL; + + nvbo = calloc(1, sizeof(*nvbo)); + if (!nvbo) + return -ENOMEM; + nvbo->base.device = dev; + + nvbo->sysmem = ptr; + nvbo->user = 1; + + nvbo->base.size = size; + nvbo->base.offset = nvbo->drm.offset; + nvbo->base.handle = bo_to_ptr(nvbo); + nvbo->refcount = 1; + *bo = &nvbo->base; + return 0; +} + +int +nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, + struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo = ptr_to_bo(handle); + + if (!dev || !bo || *bo) + return -EINVAL; + + nvbo->refcount++; + *bo = &nvbo->base; + return 0; +} + +void +nouveau_bo_del(struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo; + + if (!bo || !*bo) + return; + nvbo = nouveau_bo(*bo); + *bo = NULL; + + if (--nvbo->refcount) + return; + + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem && !nvbo->user) + free(nvbo->sysmem); + free(nvbo); +} + +int +nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + + if (!nvbo) + return -EINVAL; + + if (flags & NOUVEAU_BO_WR) + nouveau_fence_wait(&nvbo->fence); + else + nouveau_fence_wait(&nvbo->wr_fence); + + if (nvbo->sysmem) + bo->map = nvbo->sysmem; + else + bo->map = nvbo->map; + return 0; +} + +void +nouveau_bo_unmap(struct nouveau_bo *bo) +{ + bo->map = NULL; +} + +static int +nouveau_bo_upload(struct nouveau_bo_priv *nvbo) +{ + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + memcpy(nvbo->map, nvbo->sysmem, nvbo->drm.size); + return 0; +} + +int +nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct drm_nouveau_mem_alloc new; + void *new_map = NULL, *new_sysmem = NULL; + unsigned new_flags = 0, ret; + + assert(!bo->map); + + /* Check current memtype vs requested, if they match do nothing */ + if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) + return 0; + if ((nvbo->drm.flags & NOUVEAU_MEM_AGP) && (flags & NOUVEAU_BO_GART)) + return 0; + if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) + return 0; + + memset(&new, 0x00, sizeof(new)); + + /* Allocate new memory */ + if (flags & NOUVEAU_BO_VRAM) + new_flags |= NOUVEAU_MEM_FB; + else + if (flags & NOUVEAU_BO_GART) + new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); + + if (new_flags) { + ret = nouveau_mem_alloc(bo->device, bo->size, + nvbo->drm.alignment, new_flags, + &new, &new_map); + if (ret) + return ret; + } else { + new_sysmem = malloc(bo->size); + } + + /* Copy old -> new */ + /*XXX: use M2MF */ + if (nvbo->sysmem || nvbo->map) { + nouveau_bo_map(bo, NOUVEAU_BO_RD); + memcpy(new_map, bo->map, bo->size); + nouveau_bo_unmap(bo); + } + + /* Free old memory */ + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem) + free(nvbo->sysmem); + + nvbo->drm = new; + nvbo->map = new_map; + nvbo->sysmem = new_sysmem; + bo->flags = flags; + bo->offset = nvbo->drm.offset; + return 0; +} + +static int +nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_resource *r; + + if (nvchan->user_charge + bo->size > nvdev->sa.size) + return 1; + nvchan->user_charge += bo->size; + + if (!(flags & NOUVEAU_BO_GART)) + return 1; + + r = nouveau_bo_tmp(chan, bo->size, fence); + if (!r) + return 1; + + memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); + + nvbo->offset = nvdev->sa.offset + r->start; + nvbo->flags = NOUVEAU_BO_GART; + return 0; +} + +static int +nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + int ret; + + ret = nouveau_bo_set_status(bo, flags); + if (ret) + return ret; + + if (nvbo->user) + nouveau_bo_upload(nvbo); + + nvbo->offset = nvbo->drm.offset; + if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) + nvbo->flags = NOUVEAU_BO_GART; + else + nvbo->flags = NOUVEAU_BO_VRAM; + + return 0; +} + +int +nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + int ret; + + assert(bo->map == NULL); + + if (nvbo->user) { + ret = nouveau_bo_validate_user(chan, bo, fence, flags); + if (ret) { + ret = nouveau_bo_validate_bo(chan, bo, fence, flags); + if (ret) + return ret; + } + } else { + ret = nouveau_bo_validate_bo(chan, bo, fence, flags); + if (ret) + return ret; + } + + if (flags & NOUVEAU_BO_WR) + nouveau_fence_ref(fence, &nvbo->wr_fence); + nouveau_fence_ref(fence, &nvbo->fence); + return 0; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_channel.c b/src/gallium/winsys/dri/nouveau/nouveau_channel.c new file mode 100644 index 00000000000..df80d04add5 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_channel.c @@ -0,0 +1,118 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +int +nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma, + uint32_t tt_ctxdma, struct nouveau_channel **chan) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct nouveau_channel_priv *nvchan; + int ret; + + if (!nvdev || !chan || *chan) + return -EINVAL; + + nvchan = calloc(1, sizeof(struct nouveau_channel_priv)); + if (!nvchan) + return -ENOMEM; + nvchan->base.device = dev; + + nvchan->drm.fb_ctxdma_handle = fb_ctxdma; + nvchan->drm.tt_ctxdma_handle = tt_ctxdma; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_CHANNEL_ALLOC, + &nvchan->drm, sizeof(nvchan->drm)); + if (ret) { + free(nvchan); + return ret; + } + + nvchan->base.id = nvchan->drm.channel; + if (nouveau_grobj_ref(&nvchan->base, nvchan->drm.fb_ctxdma_handle, + &nvchan->base.vram) || + nouveau_grobj_ref(&nvchan->base, nvchan->drm.tt_ctxdma_handle, + &nvchan->base.gart)) { + nouveau_channel_free((void *)&nvchan); + return -EINVAL; + } + + ret = drmMap(nvdev->fd, nvchan->drm.ctrl, nvchan->drm.ctrl_size, + (void*)&nvchan->user); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + nvchan->put = &nvchan->user[0x40/4]; + nvchan->get = &nvchan->user[0x44/4]; + nvchan->ref_cnt = &nvchan->user[0x48/4]; + + ret = drmMap(nvdev->fd, nvchan->drm.notifier, nvchan->drm.notifier_size, + (drmAddressPtr)&nvchan->notifier_block); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + + ret = drmMap(nvdev->fd, nvchan->drm.cmdbuf, nvchan->drm.cmdbuf_size, + (void*)&nvchan->pushbuf); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + + nouveau_dma_channel_init(&nvchan->base); + nouveau_pushbuf_init(&nvchan->base); + + *chan = &nvchan->base; + return 0; +} + +void +nouveau_channel_free(struct nouveau_channel **chan) +{ + struct nouveau_channel_priv *nvchan; + struct nouveau_device_priv *nvdev; + struct drm_nouveau_channel_free cf; + + if (!chan || !*chan) + return; + nvchan = nouveau_channel(*chan); + *chan = NULL; + nvdev = nouveau_device(nvchan->base.device); + + FIRE_RING_CH(&nvchan->base); + + nouveau_grobj_free(&nvchan->base.vram); + nouveau_grobj_free(&nvchan->base.gart); + + cf.channel = nvchan->drm.channel; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf)); + free(nvchan); +} + + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c new file mode 100644 index 00000000000..01fada5b89b --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -0,0 +1,206 @@ +#include "main/glheader.h" +#include "glapi/glthread.h" +#include +#include "utils.h" + +#include "state_tracker/st_public.h" +#include "pipe/p_defines.h" +#include "pipe/p_context.h" + +#include "nouveau_context.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_winsys_pipe.h" + +#ifdef DEBUG +static const struct dri_debug_control debug_control[] = { + { "bo", DEBUG_BO }, + { NULL, 0 } +}; +int __nouveau_debug = 0; +#endif + +GLboolean +nouveau_context_create(const __GLcontextModes *glVis, + __DRIcontextPrivate *driContextPriv, + void *sharedContextPrivate) +{ + __DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv; + struct nouveau_screen *nv_screen = driScrnPriv->private; + struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); + struct nouveau_device_priv *nvdev; + struct pipe_context *pipe = NULL; + struct st_context *st_share = NULL; + int ret; + + if (sharedContextPrivate) { + st_share = ((struct nouveau_context *)sharedContextPrivate)->st; + } + + if ((ret = nouveau_device_get_param(nv_screen->device, + NOUVEAU_GETPARAM_CHIPSET_ID, + &nv->chipset))) { + NOUVEAU_ERR("Error determining chipset id: %d\n", ret); + return GL_FALSE; + } + + if ((ret = nouveau_channel_alloc(nv_screen->device, + 0x8003d001, 0x8003d002, + &nv->channel))) { + NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); + return GL_FALSE; + } + + driContextPriv->driverPrivate = (void *)nv; + nv->nv_screen = nv_screen; + nv->dri_screen = driScrnPriv; + + nvdev = nouveau_device(nv_screen->device); + nvdev->ctx = driContextPriv->hHWContext; + nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; + + driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, + nv->dri_screen->myNum, "nouveau"); +#ifdef DEBUG + __nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"), + debug_control); +#endif + + /*XXX: Hack up a fake region and buffer object for front buffer. + * This will go away with TTM, replaced with a simple reference + * of the front buffer handle passed to us by the DDX. + */ + { + struct pipe_surface *fb_surf; + struct nouveau_pipe_buffer *fb_buf; + struct nouveau_bo_priv *fb_bo; + + fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); + fb_bo->drm.offset = nv_screen->front_offset; + fb_bo->drm.flags = NOUVEAU_MEM_FB; + fb_bo->drm.size = nv_screen->front_pitch * + nv_screen->front_height; + fb_bo->refcount = 1; + fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; + fb_bo->base.offset = fb_bo->drm.offset; + fb_bo->base.handle = (unsigned long)fb_bo; + fb_bo->base.size = fb_bo->drm.size; + fb_bo->base.device = nv_screen->device; + + fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer)); + fb_buf->bo = &fb_bo->base; + + fb_surf = calloc(1, sizeof(struct pipe_surface)); + fb_surf->cpp = nv_screen->front_cpp; + fb_surf->pitch = nv_screen->front_pitch / fb_surf->cpp; + fb_surf->height = nv_screen->front_height; + fb_surf->refcount = 1; + fb_surf->buffer = &fb_buf->base; + + nv->frontbuffer = fb_surf; + } + + if ((ret = nouveau_grobj_alloc(nv->channel, 0x00000000, 0x30, + &nv->NvNull))) { + NOUVEAU_ERR("Error creating NULL object: %d\n", ret); + return GL_FALSE; + } + nv->next_handle = 0x80000000; + + if ((ret = nouveau_notifier_alloc(nv->channel, nv->next_handle++, 1, + &nv->sync_notifier))) { + NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); + return GL_FALSE; + } + + if (nv->chipset < 0x50) + ret = nouveau_surface_init_nv04(nv); + else + ret = nouveau_surface_init_nv50(nv); + if (ret) { + return GL_FALSE; + } + + if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { + pipe = nouveau_pipe_create(nv); + if (!pipe) + NOUVEAU_ERR("Couldn't create hw pipe\n"); + } + + if (!pipe) { + NOUVEAU_MSG("Using softpipe\n"); + pipe = nouveau_create_softpipe(nv); + if (!pipe) { + NOUVEAU_ERR("Error creating pipe, bailing\n"); + return GL_FALSE; + } + } + + pipe->priv = nv; + nv->st = st_create_context(pipe, glVis, st_share); + return GL_TRUE; +} + +void +nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) +{ + struct nouveau_context *nv = driContextPriv->driverPrivate; + + assert(nv); + + st_flush(nv->st, PIPE_FLUSH_WAIT); + st_destroy_context(nv->st); + + nouveau_grobj_free(&nv->NvCtxSurf2D); + nouveau_grobj_free(&nv->NvImageBlit); + nouveau_channel_free(&nv->channel); + + free(nv); +} + +GLboolean +nouveau_context_bind(__DRIcontextPrivate *driContextPriv, + __DRIdrawablePrivate *driDrawPriv, + __DRIdrawablePrivate *driReadPriv) +{ + struct nouveau_context *nv; + struct nouveau_framebuffer *draw, *read; + + if (!driContextPriv) { + st_make_current(NULL, NULL, NULL); + return GL_TRUE; + } + + nv = driContextPriv->driverPrivate; + draw = driDrawPriv->driverPrivate; + read = driReadPriv->driverPrivate; + + st_make_current(nv->st, draw->stfb, read->stfb); + + if ((nv->dri_drawable != driDrawPriv) || + (nv->last_stamp != driDrawPriv->lastStamp)) { + nv->dri_drawable = driDrawPriv; + st_resize_framebuffer(draw->stfb, driDrawPriv->w, + driDrawPriv->h); + nv->last_stamp = driDrawPriv->lastStamp; + } + + if (driDrawPriv != driReadPriv) { + st_resize_framebuffer(read->stfb, driReadPriv->w, + driReadPriv->h); + } + + return GL_TRUE; +} + +GLboolean +nouveau_context_unbind(__DRIcontextPrivate *driContextPriv) +{ + struct nouveau_context *nv = driContextPriv->driverPrivate; + (void)nv; + + st_flush(nv->st, 0); + return GL_TRUE; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h new file mode 100644 index 00000000000..5805f969ba8 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -0,0 +1,89 @@ +#ifndef __NOUVEAU_CONTEXT_H__ +#define __NOUVEAU_CONTEXT_H__ + +#include "dri_util.h" +#include "xmlconfig.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau_device.h" +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +struct nouveau_framebuffer { + struct st_framebuffer *stfb; +}; + +struct nouveau_context { + struct st_context *st; + + /* Misc HW info */ + uint64_t chipset; + + /* DRI stuff */ + __DRIscreenPrivate *dri_screen; + __DRIdrawablePrivate *dri_drawable; + unsigned int last_stamp; + driOptionCache dri_option_cache; + drm_context_t drm_context; + drmLock drm_lock; + GLboolean locked; + struct nouveau_screen *nv_screen; + struct pipe_surface *frontbuffer; + + /* Hardware context */ + struct nouveau_channel *channel; + struct nouveau_notifier *sync_notifier; + struct nouveau_grobj *NvNull; + struct nouveau_grobj *NvCtxSurf2D; + struct nouveau_grobj *NvImageBlit; + struct nouveau_grobj *NvGdiRect; + struct nouveau_grobj *NvM2MF; + struct nouveau_grobj *Nv2D; + uint32_t next_handle; + uint32_t next_subchannel; + uint32_t next_sequence; + + /* pipe_surface accel */ + struct pipe_surface *surf_src, *surf_dst; + unsigned surf_src_offset, surf_dst_offset; + int (*surface_copy_prep)(struct nouveau_context *, + struct pipe_surface *dst, + struct pipe_surface *src); + void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h); + void (*surface_copy_done)(struct nouveau_context *); + int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned, unsigned); +}; + +extern GLboolean nouveau_context_create(const __GLcontextModes *, + __DRIcontextPrivate *, void *); +extern void nouveau_context_destroy(__DRIcontextPrivate *); +extern GLboolean nouveau_context_bind(__DRIcontextPrivate *, + __DRIdrawablePrivate *draw, + __DRIdrawablePrivate *read); +extern GLboolean nouveau_context_unbind(__DRIcontextPrivate *); + +#ifdef DEBUG +extern int __nouveau_debug; + +#define DEBUG_BO (1 << 0) + +#define DBG(flag, ...) do { \ + if (__nouveau_debug & (DEBUG_##flag)) \ + NOUVEAU_ERR(__VA_ARGS__); \ +} while(0) +#else +#define DBG(flag, ...) +#endif + +extern void LOCK_HARDWARE(struct nouveau_context *); +extern void UNLOCK_HARDWARE(struct nouveau_context *); + +extern int nouveau_surface_init_nv04(struct nouveau_context *); +extern int nouveau_surface_init_nv50(struct nouveau_context *); + +extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int); +extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *); + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_device.c b/src/gallium/winsys/dri/nouveau/nouveau_device.c new file mode 100644 index 00000000000..409e4415f76 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_device.c @@ -0,0 +1,146 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" + +int +nouveau_device_open_existing(struct nouveau_device **dev, int close, + int fd, drm_context_t ctx) +{ + struct nouveau_device_priv *nvdev; + int ret; + + if (!dev || *dev) + return -EINVAL; + + nvdev = calloc(1, sizeof(*nvdev)); + if (!nvdev) + return -ENOMEM; + nvdev->fd = fd; + nvdev->ctx = ctx; + nvdev->needs_close = close; + + drmCommandNone(nvdev->fd, DRM_NOUVEAU_CARD_INIT); + + if ((ret = nouveau_bo_init(&nvdev->base))) { + nouveau_device_close((void *)&nvdev); + return ret; + } + + *dev = &nvdev->base; + return 0; +} + +int +nouveau_device_open(struct nouveau_device **dev, const char *busid) +{ + drm_context_t ctx; + int fd, ret; + + if (!dev || *dev) + return -EINVAL; + + fd = drmOpen("nouveau", busid); + if (fd < 0) + return -EINVAL; + + ret = drmCreateContext(fd, &ctx); + if (ret) { + drmClose(fd); + return ret; + } + + ret = nouveau_device_open_existing(dev, 1, fd, ctx); + if (ret) { + drmDestroyContext(fd, ctx); + drmClose(fd); + return ret; + } + + return 0; +} + +void +nouveau_device_close(struct nouveau_device **dev) +{ + struct nouveau_device_priv *nvdev; + + if (dev || !*dev) + return; + nvdev = nouveau_device(*dev); + *dev = NULL; + + nouveau_bo_takedown(&nvdev->base); + + if (nvdev->needs_close) { + drmDestroyContext(nvdev->fd, nvdev->ctx); + drmClose(nvdev->fd); + } + free(nvdev); +} + +int +nouveau_device_get_param(struct nouveau_device *dev, + uint64_t param, uint64_t *value) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_getparam g; + int ret; + + if (!nvdev || !value) + return -EINVAL; + + g.param = param; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_GETPARAM, + &g, sizeof(g)); + if (ret) + return ret; + + *value = g.value; + return 0; +} + +int +nouveau_device_set_param(struct nouveau_device *dev, + uint64_t param, uint64_t value) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_setparam s; + int ret; + + if (!nvdev) + return -EINVAL; + + s.param = param; + s.value = value; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_SETPARAM, + &s, sizeof(s)); + if (ret) + return ret; + + return 0; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_device.h b/src/gallium/winsys/dri/nouveau/nouveau_device.h new file mode 100644 index 00000000000..744a89f74bf --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_device.h @@ -0,0 +1,29 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DEVICE_H__ +#define __NOUVEAU_DEVICE_H__ + +struct nouveau_device { +}; + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_dma.c b/src/gallium/winsys/dri/nouveau/nouveau_dma.c new file mode 100644 index 00000000000..f8a8ba04f6d --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_dma.c @@ -0,0 +1,219 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +static inline uint32_t +READ_GET(struct nouveau_channel_priv *nvchan) +{ + return *nvchan->get; +} + +static inline void +WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) +{ + uint32_t put = ((val << 2) + nvchan->dma->base); + volatile int dum; + + NOUVEAU_DMA_BARRIER; + dum = READ_GET(nvchan); + + *nvchan->put = put; + nvchan->dma->put = val; +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put); +#endif + + NOUVEAU_DMA_BARRIER; +} + +static inline int +LOCAL_GET(struct nouveau_dma_priv *dma, uint32_t *val) +{ + uint32_t get = *val; + + if (get >= dma->base && get <= (dma->base + (dma->max << 2))) { + *val = (get - dma->base) >> 2; + return 1; + } + + return 0; +} + +void +nouveau_dma_channel_init(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + int i; + + nvchan->dma = &nvchan->dma_master; + nvchan->dma->base = nvchan->drm.put_base; + nvchan->dma->cur = nvchan->dma->put = 0; + nvchan->dma->max = (nvchan->drm.cmdbuf_size >> 2) - 2; + nvchan->dma->free = nvchan->dma->max - nvchan->dma->cur; + + RING_SPACE_CH(chan, RING_SKIPS); + for (i = 0; i < RING_SKIPS; i++) + OUT_RING_CH(chan, 0); +} + +#define CHECK_TIMEOUT() do { \ + if ((NOUVEAU_TIME_MSEC() - t_start) > NOUVEAU_DMA_TIMEOUT) \ + return - EBUSY; \ +} while(0) + +int +nouveau_dma_wait(struct nouveau_channel *chan, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + uint32_t get, t_start; + + FIRE_RING_CH(chan); + + t_start = NOUVEAU_TIME_MSEC(); + while (dma->free < size) { + CHECK_TIMEOUT(); + + get = READ_GET(nvchan); + if (!LOCAL_GET(dma, &get)) + continue; + + if (dma->put >= get) { + dma->free = dma->max - dma->cur; + + if (dma->free < size) { +#ifdef NOUVEAU_DMA_DEBUG + dma->push_free = 1; +#endif + OUT_RING_CH(chan, 0x20000000 | dma->base); + if (get <= RING_SKIPS) { + /*corner case - will be idle*/ + if (dma->put <= RING_SKIPS) + WRITE_PUT(nvchan, + RING_SKIPS + 1); + + do { + CHECK_TIMEOUT(); + get = READ_GET(nvchan); + if (!LOCAL_GET(dma, &get)) + get = 0; + } while (get <= RING_SKIPS); + } + + WRITE_PUT(nvchan, RING_SKIPS); + dma->cur = dma->put = RING_SKIPS; + dma->free = get - (RING_SKIPS + 1); + } + } else { + dma->free = get - dma->cur - 1; + } + } + + return 0; +} + +#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF +static void +nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + unsigned mthd_count = 0; + + while (get != put) { + uint32_t gpuget = (get << 2) + nvchan->drm.put_base; + uint32_t data; + + if (get < 0 || get >= nvchan->drm.cmdbuf_size) { + NOUVEAU_ERR("DMA_PT 0x%08x\n", gpuget); + assert(0); + } + data = nvchan->pushbuf[get++]; + + if (mthd_count) { + NOUVEAU_MSG("0x%08x 0x%08x\n", gpuget, data); + mthd_count--; + continue; + } + + switch (data & 0x60000000) { + case 0x00000000: + mthd_count = (data >> 18) & 0x7ff; + NOUVEAU_MSG("0x%08x 0x%08x MTHD " + "Sc %d Mthd 0x%04x Size %d\n", + gpuget, data, (data>>13) & 7, data & 0x1ffc, + mthd_count); + break; + case 0x20000000: + get = (data & 0x1ffffffc) >> 2; + NOUVEAU_MSG("0x%08x 0x%08x JUMP 0x%08x\n", + gpuget, data, data & 0x1ffffffc); + continue; + case 0x40000000: + mthd_count = (data >> 18) & 0x7ff; + NOUVEAU_MSG("0x%08x 0x%08x NINC " + "Sc %d Mthd 0x%04x Size %d\n", + gpuget, data, (data>>13) & 7, data & 0x1ffc, + mthd_count); + break; + case 0x60000000: + /* DMA_OPCODE_CALL apparently, doesn't seem to work on + * my NV40 at least.. + */ + /* fall-through */ + default: + NOUVEAU_MSG("DMA_PUSHER 0x%08x 0x%08x\n", + gpuget, data); + assert(0); + } + } +} +#endif + +void +nouveau_dma_kickoff(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + + if (dma->cur == dma->put) + return; + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free) { + NOUVEAU_ERR("Packet incomplete: %d left\n", dma->push_free); + return; + } +#endif + +#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF + nouveau_dma_parse_pushbuf(chan, dma->put, dma->cur); +#endif + + WRITE_PUT(nvchan, dma->cur); +} diff --git a/src/gallium/winsys/dri/nouveau/nouveau_dma.h b/src/gallium/winsys/dri/nouveau/nouveau_dma.h new file mode 100644 index 00000000000..cfa6d26e828 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_dma.h @@ -0,0 +1,143 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DMA_H__ +#define __NOUVEAU_DMA_H__ + +#include +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +#define RING_SKIPS 8 + +extern int nouveau_dma_wait(struct nouveau_channel *chan, int size); +extern void nouveau_dma_subc_bind(struct nouveau_grobj *); +extern void nouveau_dma_channel_init(struct nouveau_channel *); +extern void nouveau_dma_kickoff(struct nouveau_channel *); + +#ifdef NOUVEAU_DMA_DEBUG +static char faulty[1024]; +#endif + +static inline void +nouveau_dma_out(struct nouveau_channel *chan, uint32_t data) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free == 0) { + NOUVEAU_ERR("No space left in packet at %s\n", faulty); + return; + } + dma->push_free--; +#endif +#ifdef NOUVEAU_DMA_TRACE + { + uint32_t offset = (dma->cur << 2) + dma->base; + NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", + nvchan->drm.channel, offset, data); + } +#endif + nvchan->pushbuf[dma->cur + (dma->base - nvchan->drm.put_base)/4] = data; + dma->cur++; +} + +static inline void +nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + (void)dma; + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free < size) { + NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", + dma->push_free, size); + return; + } +#endif +#ifdef NOUVEAU_DMA_TRACE + while (size--) { + nouveau_dma_out(chan, *ptr); + ptr++; + } +#else + memcpy(&nvchan->pushbuf[dma->cur], ptr, size << 2); +#ifdef NOUVEAU_DMA_DEBUG + dma->push_free -= size; +#endif + dma->cur += size; +#endif +} + +static inline void +nouveau_dma_space(struct nouveau_channel *chan, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + + if (dma->free < size) { + if (nouveau_dma_wait(chan, size) && chan->hang_notify) + chan->hang_notify(chan); + } + dma->free -= size; +#ifdef NOUVEAU_DMA_DEBUG + dma->push_free = size; +#endif +} + +static inline void +nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, + int method, int size, const char* file, int line) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + (void)dma; + +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, + grobj->handle, grobj->subc, method, size); +#endif + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free) { + NOUVEAU_ERR("Previous packet incomplete: %d left at %s\n", + dma->push_free, faulty); + return; + } + sprintf(faulty,"%s:%d",file,line); +#endif + + nouveau_dma_space(chan, (size + 1)); + nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); +} + +#define RING_SPACE_CH(ch,sz) nouveau_dma_space((ch), (sz)) +#define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) +#define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) +#define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ + (dwords)) +#define FIRE_RING_CH(ch) nouveau_dma_kickoff((ch)) +#define WAIT_RING_CH(ch,sz) nouveau_dma_wait((ch), (sz)) + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_dri.h b/src/gallium/winsys/dri/nouveau/nouveau_dri.h new file mode 100644 index 00000000000..1207c2d609c --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_dri.h @@ -0,0 +1,28 @@ +#ifndef _NOUVEAU_DRI_ +#define _NOUVEAU_DRI_ + +#include "xf86drm.h" +#include "drm.h" +#include "nouveau_drm.h" + +struct nouveau_dri { + uint32_t device_id; /**< \brief PCI device ID */ + uint32_t width; /**< \brief width in pixels of display */ + uint32_t height; /**< \brief height in scanlines of display */ + uint32_t depth; /**< \brief depth of display (8, 15, 16, 24) */ + uint32_t bpp; /**< \brief bit depth of display (8, 16, 24, 32) */ + + uint32_t bus_type; /**< \brief ths bus type */ + uint32_t bus_mode; /**< \brief bus mode (used for AGP, maybe also for PCI-E ?) */ + + uint32_t front_offset; /**< \brief front buffer offset */ + uint32_t front_pitch; /**< \brief front buffer pitch */ + uint32_t back_offset; /**< \brief private back buffer offset */ + uint32_t back_pitch; /**< \brief private back buffer pitch */ + uint32_t depth_offset; /**< \brief private depth buffer offset */ + uint32_t depth_pitch; /**< \brief private depth buffer pitch */ + +}; + +#endif + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h new file mode 100644 index 00000000000..37e404fc6cb --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h @@ -0,0 +1,304 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DRMIF_H__ +#define __NOUVEAU_DRMIF_H__ + +#include +#include +#include + +#include "nouveau_device.h" +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_grobj.h" +#include "nouveau/nouveau_notifier.h" +#include "nouveau/nouveau_bo.h" +#include "nouveau/nouveau_resource.h" +#include "nouveau/nouveau_pushbuf.h" + +struct nouveau_device_priv { + struct nouveau_device base; + + int fd; + drm_context_t ctx; + drmLock *lock; + int needs_close; + + struct drm_nouveau_mem_alloc sa; + void *sa_map; + struct nouveau_resource *sa_heap; +}; +#define nouveau_device(n) ((struct nouveau_device_priv *)(n)) + +extern int +nouveau_device_open_existing(struct nouveau_device **, int close, + int fd, drm_context_t ctx); + +extern int +nouveau_device_open(struct nouveau_device **, const char *busid); + +extern void +nouveau_device_close(struct nouveau_device **); + +extern int +nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v); + +extern int +nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val); + +struct nouveau_fence { + struct nouveau_channel *channel; +}; + +struct nouveau_fence_cb { + struct nouveau_fence_cb *next; + void (*func)(void *); + void *priv; +}; + +struct nouveau_fence_priv { + struct nouveau_fence base; + int refcount; + + struct nouveau_fence *next; + struct nouveau_fence_cb *signal_cb; + + uint32_t sequence; + int emitted; + int signalled; +}; +#define nouveau_fence(n) ((struct nouveau_fence_priv *)(n)) + +extern int +nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **); + +extern int +nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); + +extern int +nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *); + +extern void +nouveau_fence_emit(struct nouveau_fence *); + +extern int +nouveau_fence_wait(struct nouveau_fence **); + +extern void +nouveau_fence_flush(struct nouveau_channel *); + +struct nouveau_pushbuf_reloc { + uint64_t next; + uint64_t handle; + uint32_t *ptr; + uint32_t flags; + uint32_t data; + uint32_t vor; + uint32_t tor; +}; + +struct nouveau_pushbuf_bo { + uint64_t next; + uint64_t handle; + uint64_t flags; + uint64_t relocs; + int nr_relocs; +}; + +struct nouveau_pushbuf_priv { + struct nouveau_pushbuf base; + + unsigned nop_jump; + unsigned start; + unsigned size; + + uint64_t buffers; + int nr_buffers; +}; +#define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) + +#define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h)) +#define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_pbrel(h) ((struct nouveau_pushbuf_reloc *)(unsigned long)(h)) +#define bo_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_bo(h) ((struct nouveau_bo_priv *)(unsigned long)(h)) + +extern int +nouveau_pushbuf_init(struct nouveau_channel *); + +extern int +nouveau_pushbuf_flush(struct nouveau_channel *, unsigned min); + +extern int +nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr, + struct nouveau_bo *, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor); + +struct nouveau_dma_priv { + uint32_t base; + uint32_t max; + uint32_t cur; + uint32_t put; + uint32_t free; + + int push_free; +} dma; + +struct nouveau_channel_priv { + struct nouveau_channel base; + + struct drm_nouveau_channel_alloc drm; + + uint32_t *pushbuf; + void *notifier_block; + + volatile uint32_t *user; + volatile uint32_t *put; + volatile uint32_t *get; + volatile uint32_t *ref_cnt; + + struct nouveau_dma_priv dma_master; + struct nouveau_dma_priv dma_bufmgr; + struct nouveau_dma_priv *dma; + + struct nouveau_fence *fence_head; + struct nouveau_fence *fence_tail; + uint32_t fence_sequence; + + struct nouveau_pushbuf_priv pb; + + unsigned user_charge; +}; +#define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) + +extern int +nouveau_channel_alloc(struct nouveau_device *, uint32_t fb, uint32_t tt, + struct nouveau_channel **); + +extern void +nouveau_channel_free(struct nouveau_channel **); + +struct nouveau_grobj_priv { + struct nouveau_grobj base; +}; +#define nouveau_grobj(n) ((struct nouveau_grobj_priv *)(n)) + +extern int nouveau_grobj_alloc(struct nouveau_channel *, uint32_t handle, + int class, struct nouveau_grobj **); +extern int nouveau_grobj_ref(struct nouveau_channel *, uint32_t handle, + struct nouveau_grobj **); +extern void nouveau_grobj_free(struct nouveau_grobj **); + + +struct nouveau_notifier_priv { + struct nouveau_notifier base; + + struct drm_nouveau_notifierobj_alloc drm; + volatile void *map; +}; +#define nouveau_notifier(n) ((struct nouveau_notifier_priv *)(n)) + +extern int +nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, int count, + struct nouveau_notifier **); + +extern void +nouveau_notifier_free(struct nouveau_notifier **); + +extern void +nouveau_notifier_reset(struct nouveau_notifier *, int id); + +extern uint32_t +nouveau_notifier_status(struct nouveau_notifier *, int id); + +extern uint32_t +nouveau_notifier_return_val(struct nouveau_notifier *, int id); + +extern int +nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, + int timeout); + +struct nouveau_bo_priv { + struct nouveau_bo base; + + struct nouveau_fence *fence; + struct nouveau_fence *wr_fence; + + struct drm_nouveau_mem_alloc drm; + void *map; + + void *sysmem; + int user; + + int refcount; + + uint64_t offset; + uint64_t flags; +}; +#define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) + +extern int +nouveau_bo_init(struct nouveau_device *); + +extern void +nouveau_bo_takedown(struct nouveau_device *); + +extern int +nouveau_bo_new(struct nouveau_device *, uint32_t flags, int align, int size, + struct nouveau_bo **); + +extern int +nouveau_bo_user(struct nouveau_device *, void *ptr, int size, + struct nouveau_bo **); + +extern int +nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); + +extern int +nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags); + +extern void +nouveau_bo_del(struct nouveau_bo **); + +extern int +nouveau_bo_map(struct nouveau_bo *, uint32_t flags); + +extern void +nouveau_bo_unmap(struct nouveau_bo *); + +extern int +nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, + struct nouveau_fence *fence, uint32_t flags); + +extern int +nouveau_resource_init(struct nouveau_resource **heap, unsigned start, + unsigned size); + +extern int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **); + +extern void +nouveau_resource_free(struct nouveau_resource **); + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_fence.c b/src/gallium/winsys/dri/nouveau/nouveau_fence.c new file mode 100644 index 00000000000..7714e6f2485 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_fence.c @@ -0,0 +1,215 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +static void +nouveau_fence_del_unsignalled(struct nouveau_fence *fence) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); + struct nouveau_fence *le; + + if (nvchan->fence_head == fence) { + nvchan->fence_head = nouveau_fence(fence)->next; + if (nvchan->fence_head == NULL) + nvchan->fence_tail = NULL; + return; + } + + le = nvchan->fence_head; + while (le && nouveau_fence(le)->next != fence) + le = nouveau_fence(le)->next; + assert(le && nouveau_fence(le)->next == fence); + nouveau_fence(le)->next = nouveau_fence(fence)->next; + if (nvchan->fence_tail == fence) + nvchan->fence_tail = le; +} + +static void +nouveau_fence_del(struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence || !*fence) + return; + nvfence = nouveau_fence(*fence); + *fence = NULL; + + if (--nvfence->refcount) + return; + + if (nvfence->emitted && !nvfence->signalled) { + if (nvfence->signal_cb) { + nvfence->refcount++; + nouveau_fence_wait((void *)&nvfence); + return; + } + + nouveau_fence_del_unsignalled(&nvfence->base); + } + free(nvfence); +} + +int +nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!chan || !fence || *fence) + return -EINVAL; + + nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); + if (!nvfence) + return -ENOMEM; + nvfence->base.channel = chan; + nvfence->refcount = 1; + + *fence = &nvfence->base; + return 0; +} + +int +nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence) + return -EINVAL; + + if (*fence) { + nouveau_fence_del(fence); + *fence = NULL; + } + + if (ref) { + nvfence = nouveau_fence(ref); + nvfence->refcount++; + *fence = &nvfence->base; + } + + return 0; +} + +int +nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *), + void *priv) +{ + struct nouveau_fence_priv *nvfence = nouveau_fence(fence); + struct nouveau_fence_cb *cb; + + if (!nvfence || !func) + return -EINVAL; + + cb = malloc(sizeof(struct nouveau_fence_cb)); + if (!cb) + return -ENOMEM; + + cb->func = func; + cb->priv = priv; + cb->next = nvfence->signal_cb; + nvfence->signal_cb = cb; + return 0; +} + +void +nouveau_fence_emit(struct nouveau_fence *fence) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); + struct nouveau_fence_priv *nvfence = nouveau_fence(fence); + + nvfence->emitted = 1; + nvfence->sequence = ++nvchan->fence_sequence; + if (nvfence->sequence == 0xffffffff) + NOUVEAU_ERR("AII wrap unhandled\n"); + + /*XXX: assumes subc 0 is populated */ + RING_SPACE_CH(fence->channel, 2); + OUT_RING_CH (fence->channel, 0x00040050); + OUT_RING_CH (fence->channel, nvfence->sequence); + + if (nvchan->fence_tail) { + nouveau_fence(nvchan->fence_tail)->next = fence; + } else { + nvchan->fence_head = fence; + } + nvchan->fence_tail = fence; +} + +void +nouveau_fence_flush(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + uint32_t sequence = *nvchan->ref_cnt; + + while (nvchan->fence_head) { + struct nouveau_fence_priv *nvfence; + + nvfence = nouveau_fence(nvchan->fence_head); + if (nvfence->sequence > sequence) + break; + + nouveau_fence_del_unsignalled(&nvfence->base); + nvfence->signalled = 1; + + if (nvfence->signal_cb) { + struct nouveau_fence *fence = NULL; + + nouveau_fence_ref(nvchan->fence_head, &fence); + + while (nvfence->signal_cb) { + struct nouveau_fence_cb *cb; + + cb = nvfence->signal_cb; + nvfence->signal_cb = cb->next; + cb->func(cb->priv); + free(cb); + } + + nouveau_fence_ref(NULL, &fence); + } + } +} + +int +nouveau_fence_wait(struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence || !*fence) + return -EINVAL; + nvfence = nouveau_fence(*fence); + + if (nvfence->emitted) { + while (!nvfence->signalled) + nouveau_fence_flush(nvfence->base.channel); + } + nouveau_fence_ref(NULL, fence); + + return 0; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_grobj.c b/src/gallium/winsys/dri/nouveau/nouveau_grobj.c new file mode 100644 index 00000000000..55dfeb99aa7 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_grobj.c @@ -0,0 +1,107 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" + +int +nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle, + int class, struct nouveau_grobj **grobj) +{ + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_grobj_priv *nvgrobj; + struct drm_nouveau_grobj_alloc g; + int ret; + + if (!nvdev || !grobj || *grobj) + return -EINVAL; + + nvgrobj = calloc(1, sizeof(*nvgrobj)); + if (!nvgrobj) + return -ENOMEM; + nvgrobj->base.channel = chan; + nvgrobj->base.handle = handle; + nvgrobj->base.grclass = class; + + g.channel = chan->id; + g.handle = handle; + g.class = class; + ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC, + &g, sizeof(g)); + if (ret) { + nouveau_grobj_free((void *)&grobj); + return ret; + } + + *grobj = &nvgrobj->base; + return 0; +} + +int +nouveau_grobj_ref(struct nouveau_channel *chan, uint32_t handle, + struct nouveau_grobj **grobj) +{ + struct nouveau_grobj_priv *nvgrobj; + + if (!chan || !grobj || *grobj) + return -EINVAL; + + nvgrobj = calloc(1, sizeof(struct nouveau_grobj_priv)); + if (!nvgrobj) + return -ENOMEM; + nvgrobj->base.channel = chan; + nvgrobj->base.handle = handle; + nvgrobj->base.grclass = 0; + + *grobj = &nvgrobj->base; + return 0; +} + +void +nouveau_grobj_free(struct nouveau_grobj **grobj) +{ + struct nouveau_device_priv *nvdev; + struct nouveau_channel_priv *chan; + struct nouveau_grobj_priv *nvgrobj; + + if (!grobj || !*grobj) + return; + nvgrobj = nouveau_grobj(*grobj); + *grobj = NULL; + + + chan = nouveau_channel(nvgrobj->base.channel); + nvdev = nouveau_device(chan->base.device); + + if (nvgrobj->base.grclass) { + struct drm_nouveau_gpuobj_free f; + + f.channel = chan->drm.channel; + f.handle = nvgrobj->base.handle; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, + &f, sizeof(f)); + } + free(nvgrobj); +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_local.h b/src/gallium/winsys/dri/nouveau/nouveau_local.h new file mode 100644 index 00000000000..59febca2929 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_local.h @@ -0,0 +1,89 @@ +#ifndef __NOUVEAU_LOCAL_H__ +#define __NOUVEAU_LOCAL_H__ + +#include + +/* Debug output */ +#define NOUVEAU_MSG(fmt, args...) do { \ + fprintf(stdout, "nouveau: "fmt, ##args); \ + fflush(stdout); \ +} while(0) + +#define NOUVEAU_ERR(fmt, args...) do { \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); \ + fflush(stderr); \ +} while(0) + +#define NOUVEAU_TIME_MSEC() 0 + +/* User FIFO control */ +//#define NOUVEAU_DMA_TRACE +//#define NOUVEAU_DMA_DEBUG +//#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF +#define NOUVEAU_DMA_BARRIER +#define NOUVEAU_DMA_TIMEOUT 2000 + +/* Push buffer access macros */ +#define OUT_RING(data) do { \ + (*nv->channel->pushbuf->cur++) = (data); \ +} while(0) + +#define OUT_RINGp(src,size) do { \ + memcpy(nv->channel->pushbuf->cur, (src), (size)<<2); \ + nv->channel->pushbuf->cur += (size); \ +} while(0) + +#define OUT_RINGf(data) do { \ + union { float v; uint32_t u; } c; \ + c.v = (data); \ + OUT_RING(c.u); \ +} while(0) + +#define FIRE_RING() do { \ + nouveau_pushbuf_flush(nv->channel, 0); \ +} while(0) + +#define BEGIN_RING_GR(obj,mthd,size) do { \ + if (nv->channel->pushbuf->remaining < ((size) + 1)) \ + nouveau_pushbuf_flush(nv->channel, ((size) + 1)); \ + OUT_RING(((obj)->subc << 13) | ((size) << 18) | (mthd)); \ + nv->channel->pushbuf->remaining -= ((size) + 1); \ +} while(0) + +#define BEGIN_RING(obj,mthd,size) do { \ + BEGIN_RING_GR(nv->obj, (mthd), (size)); \ +} while(0) + +#define BIND_RING(o,s) do { \ + nv->o->subc = (s); \ + BEGIN_RING(o, 0x0000, 1); \ + OUT_RING (nv->o->handle); \ +} while(0) + +#define OUT_RELOC(buf,data,flags,vor,tor) do { \ + nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur++, \ + buf, (data), (flags), (vor), (tor)); \ +} while(0) + +/* Raw data + flags depending on FB/TT buffer */ +#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ + OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ +} while(0) + +/* FB/TT object handle */ +#define OUT_RELOCo(bo,flags) do { \ + OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ + nv->channel->vram->handle, nv->channel->gart->handle); \ +} while(0) + +/* Low 32-bits of offset */ +#define OUT_RELOCl(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ +} while(0) + +/* High 32-bits of offset */ +#define OUT_RELOCh(bo,delta,flags) do { \ + OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ +} while(0) + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_lock.c b/src/gallium/winsys/dri/nouveau/nouveau_lock.c new file mode 100644 index 00000000000..9adb9ac8547 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_lock.c @@ -0,0 +1,94 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "main/glheader.h" +#include "glapi/glthread.h" +#include + +#include "nouveau_context.h" +#include "nouveau_screen.h" + +_glthread_DECLARE_STATIC_MUTEX( lockMutex ); + +static void +nouveau_contended_lock(struct nouveau_context *nv, GLuint flags) +{ + __DRIdrawablePrivate *dPriv = nv->dri_drawable; + __DRIscreenPrivate *sPriv = nv->dri_screen; + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + drmGetLock(nvdev->fd, nvdev->ctx, flags); + + /* If the window moved, may need to set a new cliprect now. + * + * NOTE: This releases and regains the hw lock, so all state + * checking must be done *after* this call: + */ + if (dPriv) + DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv); +} + +/* Lock the hardware and validate our state. + */ +void +LOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + char __ret=0; + + _glthread_LOCK_MUTEX(lockMutex); + assert(!nv->locked); + + DRM_CAS(nvdev->lock, nvdev->ctx, + (DRM_LOCK_HELD | nvdev->ctx), __ret); + + if (__ret) + nouveau_contended_lock(nv, 0); + nv->locked = GL_TRUE; +} + + + /* Unlock the hardware using the global current context + */ +void +UNLOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + assert(nv->locked); + nv->locked = GL_FALSE; + + DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx); + + _glthread_UNLOCK_MUTEX(lockMutex); +} diff --git a/src/gallium/winsys/dri/nouveau/nouveau_notifier.c b/src/gallium/winsys/dri/nouveau/nouveau_notifier.c new file mode 100644 index 00000000000..01e8f38440e --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_notifier.c @@ -0,0 +1,137 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +#define NOTIFIER(__v) \ + struct nouveau_notifier_priv *nvnotify = nouveau_notifier(notifier); \ + volatile uint32_t *__v = (void*)nvnotify->map + (id * 32) + +int +nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, + int count, struct nouveau_notifier **notifier) +{ + struct nouveau_notifier_priv *nvnotify; + int ret; + + if (!chan || !notifier || *notifier) + return -EINVAL; + + nvnotify = calloc(1, sizeof(struct nouveau_notifier_priv)); + if (!nvnotify) + return -ENOMEM; + nvnotify->base.channel = chan; + nvnotify->base.handle = handle; + + nvnotify->drm.channel = chan->id; + nvnotify->drm.handle = handle; + nvnotify->drm.count = count; + if ((ret = drmCommandWriteRead(nouveau_device(chan->device)->fd, + DRM_NOUVEAU_NOTIFIEROBJ_ALLOC, + &nvnotify->drm, + sizeof(nvnotify->drm)))) { + nouveau_notifier_free((void *)&nvnotify); + return ret; + } + + nvnotify->map = (void *)nouveau_channel(chan)->notifier_block + + nvnotify->drm.offset; + *notifier = &nvnotify->base; + return 0; +} + +void +nouveau_notifier_free(struct nouveau_notifier **notifier) +{ + + struct nouveau_notifier_priv *nvnotify; + struct nouveau_channel_priv *nvchan; + struct nouveau_device_priv *nvdev; + struct drm_nouveau_gpuobj_free f; + + if (!notifier || !*notifier) + return; + nvnotify = nouveau_notifier(*notifier); + *notifier = NULL; + + nvchan = nouveau_channel(nvnotify->base.channel); + nvdev = nouveau_device(nvchan->base.device); + + f.channel = nvchan->drm.channel; + f.handle = nvnotify->base.handle; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); + free(nvnotify); +} + +void +nouveau_notifier_reset(struct nouveau_notifier *notifier, int id) +{ + NOTIFIER(n); + + n[NV_NOTIFY_TIME_0 /4] = 0x00000000; + n[NV_NOTIFY_TIME_1 /4] = 0x00000000; + n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000; + n[NV_NOTIFY_STATE /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS << + NV_NOTIFY_STATE_STATUS_SHIFT); +} + +uint32_t +nouveau_notifier_status(struct nouveau_notifier *notifier, int id) +{ + NOTIFIER(n); + + return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; +} + +uint32_t +nouveau_notifier_return_val(struct nouveau_notifier *notifier, int id) +{ + NOTIFIER(n); + + return n[NV_NOTIFY_RETURN_VALUE/4]; +} + +int +nouveau_notifier_wait_status(struct nouveau_notifier *notifier, int id, + int status, int timeout) +{ + NOTIFIER(n); + uint32_t time = 0, t_start = NOUVEAU_TIME_MSEC(); + + while (time <= timeout) { + uint32_t v; + + v = n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; + if (v == status) + return 0; + + if (timeout) + time = NOUVEAU_TIME_MSEC() - t_start; + } + + return -EBUSY; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c new file mode 100644 index 00000000000..7d5eddb92ff --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c @@ -0,0 +1,261 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +#define PB_BUFMGR_DWORDS (4096 / 2) +#define PB_MIN_USER_DWORDS 2048 + +static int +nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; + + assert((min + 1) <= nvchan->dma->max); + + /* Wait for enough space in push buffer */ + min = min < PB_MIN_USER_DWORDS ? PB_MIN_USER_DWORDS : min; + min += 1; /* a bit extra for the NOP */ + if (nvchan->dma->free < min) + WAIT_RING_CH(chan, min); + + /* Insert NOP, may turn into a jump later */ + RING_SPACE_CH(chan, 1); + nvpb->nop_jump = nvchan->dma->cur; + OUT_RING_CH(chan, 0); + + /* Any remaining space is available to the user */ + nvpb->start = nvchan->dma->cur; + nvpb->size = nvchan->dma->free; + nvpb->base.channel = chan; + nvpb->base.remaining = nvpb->size; + nvpb->base.cur = &nvchan->pushbuf[nvpb->start]; + + return 0; +} + +int +nouveau_pushbuf_init(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *m = &nvchan->dma_master; + struct nouveau_dma_priv *b = &nvchan->dma_bufmgr; + int i; + + if (!nvchan) + return -EINVAL; + + /* Reassign last bit of push buffer for a "separate" bufmgr + * ring buffer + */ + m->max -= PB_BUFMGR_DWORDS; + m->free -= PB_BUFMGR_DWORDS; + + b->base = m->base + ((m->max + 2) << 2); + b->max = PB_BUFMGR_DWORDS - 2; + b->cur = b->put = 0; + b->free = b->max - b->cur; + + /* Some NOPs just to be safe + *XXX: RING_SKIPS + */ + nvchan->dma = b; + RING_SPACE_CH(chan, 8); + for (i = 0; i < 8; i++) + OUT_RING_CH(chan, 0); + nvchan->dma = m; + + nouveau_pushbuf_space(chan, 0); + chan->pushbuf = &nvchan->pb.base; + + return 0; +} + +static uint32_t +nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo, + struct nouveau_pushbuf_reloc *r) +{ + uint32_t push; + + if (r->flags & NOUVEAU_BO_LOW) { + push = bo->offset + r->data; + } else + if (r->flags & NOUVEAU_BO_HIGH) { + push = (bo->offset + r->data) >> 32; + } else { + push = r->data; + } + + if (r->flags & NOUVEAU_BO_OR) { + if (bo->flags & NOUVEAU_BO_VRAM) + push |= r->vor; + else + push |= r->tor; + } + + return push; +} + +/* This would be our TTM "superioctl" */ +int +nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; + struct nouveau_pushbuf_bo *pbbo; + struct nouveau_fence *fence = NULL; + int ret; + + if (nvpb->base.remaining == nvpb->size) + return 0; + + nvpb->size -= nvpb->base.remaining; + nvchan->dma->cur += nvpb->size; + nvchan->dma->free -= nvpb->size; + assert(nvchan->dma->cur <= nvchan->dma->max); + + ret = nouveau_fence_new(chan, &fence); + if (ret) + return ret; + + nvchan->dma = &nvchan->dma_bufmgr; + nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 | + (nvchan->dma->base + (nvchan->dma->cur << 2)); + + /* Validate buffers + apply relocations */ + nvchan->user_charge = 0; + while ((pbbo = ptr_to_pbbo(nvpb->buffers))) { + struct nouveau_pushbuf_reloc *r; + struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; + + ret = nouveau_bo_validate(chan, bo, fence, pbbo->flags); + assert (ret == 0); + + if (bo->offset == nouveau_bo(bo)->offset && + bo->flags == nouveau_bo(bo)->flags) { + while ((r = ptr_to_pbrel(pbbo->relocs))) { + pbbo->relocs = r->next; + free(r); + } + + nvpb->buffers = pbbo->next; + free(pbbo); + continue; + } + bo->offset = nouveau_bo(bo)->offset; + bo->flags = nouveau_bo(bo)->flags; + + while ((r = ptr_to_pbrel(pbbo->relocs))) { + *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); + pbbo->relocs = r->next; + free(r); + } + + nvpb->buffers = pbbo->next; + free(pbbo); + } + nvpb->nr_buffers = 0; + + /* Switch back to user's ring */ + RING_SPACE_CH(chan, 1); + OUT_RING_CH(chan, 0x20000000 | ((nvpb->start << 2) + + nvchan->dma_master.base)); + nvchan->dma = &nvchan->dma_master; + + /* Fence + kickoff */ + nouveau_fence_emit(fence); + FIRE_RING_CH(chan); + nouveau_fence_ref(NULL, &fence); + + /* Allocate space for next push buffer */ + assert(!nouveau_pushbuf_space(chan, min)); + + return 0; +} + +static struct nouveau_pushbuf_bo * +nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) +{ + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); + struct nouveau_pushbuf_bo *pbbo = ptr_to_pbbo(nvpb->buffers); + + while (pbbo) { + if (pbbo->handle == bo->handle) + return pbbo; + pbbo = ptr_to_pbbo(pbbo->next); + } + + pbbo = malloc(sizeof(struct nouveau_pushbuf_bo)); + pbbo->next = nvpb->buffers; + nvpb->buffers = pbbo_to_ptr(pbbo); + nvpb->nr_buffers++; + + pbbo->handle = bo_to_ptr(bo); + pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; + pbbo->relocs = 0; + pbbo->nr_relocs = 0; + return pbbo; +} + +int +nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, + struct nouveau_bo *bo, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor) +{ + struct nouveau_pushbuf_bo *pbbo; + struct nouveau_pushbuf_reloc *r; + + if (!chan) + return -EINVAL; + + pbbo = nouveau_pushbuf_emit_buffer(chan, bo); + if (!pbbo) + return -EFAULT; + + r = malloc(sizeof(struct nouveau_pushbuf_reloc)); + r->next = pbbo->relocs; + pbbo->relocs = pbrel_to_ptr(r); + pbbo->nr_relocs++; + + pbbo->flags |= (flags & NOUVEAU_BO_RDWR); + pbbo->flags &= (flags | NOUVEAU_BO_RDWR); + + r->handle = bo_to_ptr(r); + r->ptr = ptr; + r->flags = flags; + r->data = data; + r->vor = vor; + r->tor = tor; + + if (flags & NOUVEAU_BO_DUMMY) + *(uint32_t *)ptr = 0; + else + *(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r); + return 0; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_resource.c b/src/gallium/winsys/dri/nouveau/nouveau_resource.c new file mode 100644 index 00000000000..5d9d578b4fe --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_resource.c @@ -0,0 +1,111 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +int +nouveau_resource_init(struct nouveau_resource **heap, + unsigned start, unsigned size) +{ + struct nouveau_resource *r; + + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = start; + r->size = size; + *heap = r; + return 0; +} + +int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!heap || !size || !res || *res) + return 1; + + while (heap) { + if (!heap->in_use && heap->size >= size) { + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = (heap->start + heap->size) - size; + r->size = size; + r->in_use = 1; + r->priv = priv; + + heap->size -= size; + + r->next = heap->next; + if (heap->next) + heap->next->prev = r; + r->prev = heap; + heap->next = r; + + *res = r; + return 0; + } + + heap = heap->next; + } + + return 1; +} + +void +nouveau_resource_free(struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!res || !*res) + return; + r = *res; + + if (r->prev && !r->prev->in_use) { + r->prev->next = r->next; + if (r->next) + r->next->prev = r->prev; + r->prev->size += r->size; + free(r); + } else + if (r->next && !r->next->in_use) { + r->next->prev = r->prev; + if (r->prev) + r->prev->next = r->next; + r->next->size += r->size; + r->next->start = r->start; + free(r); + } else { + r->in_use = 0; + } + + *res = NULL; +} diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c new file mode 100644 index 00000000000..f06e1784831 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.c @@ -0,0 +1,308 @@ +#include "utils.h" +#include "vblank.h" +#include "xmlpool.h" + +#include "pipe/p_context.h" +#include "state_tracker/st_public.h" +#include "state_tracker/st_cb_fbo.h" + +#include "nouveau_context.h" +#include "nouveau_device.h" +#include "nouveau_drm.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10 +#error nouveau_drm.h version does not match expected version +#endif + +/* Extension stuff, enabling of extensions handled by Gallium's GL state + * tracker. But, we still need to define the entry points we want. + */ +#define need_GL_ARB_fragment_program +#define need_GL_ARB_multisample +#define need_GL_ARB_occlusion_query +#define need_GL_ARB_point_parameters +#define need_GL_ARB_shader_objects +#define need_GL_ARB_texture_compression +#define need_GL_ARB_vertex_program +#define need_GL_ARB_vertex_shader +#define need_GL_ARB_vertex_buffer_object +#define need_GL_EXT_compiled_vertex_array +#define need_GL_EXT_fog_coord +#define need_GL_EXT_secondary_color +#define need_GL_EXT_framebuffer_object +#define need_GL_VERSION_2_0 +#define need_GL_VERSION_2_1 +#include "extension_helper.h" + +const struct dri_extension card_extensions[] = +{ + { "GL_ARB_multisample", GL_ARB_multisample_functions }, + { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, + { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, + { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, + { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, + { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, + { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, + { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, + { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, + { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, + { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, + { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, + { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, + { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, + { NULL, 0 } +}; + +PUBLIC const char __driConfigOptions[] = +DRI_CONF_BEGIN +DRI_CONF_END; +static const GLuint __driNConfigOptions = 0; + +extern const struct dri_extension common_extensions[]; +extern const struct dri_extension nv40_extensions[]; + +static GLboolean +nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_dri *nv_dri = driScrnPriv->pDevPriv; + struct nouveau_screen *nv_screen; + int ret; + + if (driScrnPriv->devPrivSize != sizeof(struct nouveau_dri)) { + NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); + return GL_FALSE; + } + + nv_screen = CALLOC_STRUCT(nouveau_screen); + if (!nv_screen) + return GL_FALSE; + nv_screen->driScrnPriv = driScrnPriv; + driScrnPriv->private = (void *)nv_screen; + + driParseOptionInfo(&nv_screen->option_cache, + __driConfigOptions, __driNConfigOptions); + + if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, + driScrnPriv->fd, 0))) { + NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); + return GL_FALSE; + } + + nv_screen->front_offset = nv_dri->front_offset; + nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); + nv_screen->front_cpp = nv_dri->bpp / 8; + nv_screen->front_height = nv_dri->height; + + return GL_TRUE; +} + +static void +nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_screen *nv_screen = driScrnPriv->private; + + driScrnPriv->private = NULL; + FREE(nv_screen); +} + +static GLboolean +nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, + __DRIdrawablePrivate * driDrawPriv, + const __GLcontextModes *glVis, GLboolean pixmapBuffer) +{ + struct nouveau_framebuffer *nvfb; + enum pipe_format colour, depth, stencil; + + if (pixmapBuffer) + return GL_FALSE; + + nvfb = CALLOC_STRUCT(nouveau_framebuffer); + if (!nvfb) + return GL_FALSE; + + if (glVis->redBits == 5) + colour = PIPE_FORMAT_R5G6B5_UNORM; + else + colour = PIPE_FORMAT_A8R8G8B8_UNORM; + + if (glVis->depthBits == 16) + depth = PIPE_FORMAT_Z16_UNORM; + else if (glVis->depthBits == 24) + depth = PIPE_FORMAT_Z24S8_UNORM; + else + depth = PIPE_FORMAT_NONE; + + if (glVis->stencilBits == 8) + stencil = PIPE_FORMAT_Z24S8_UNORM; + else + stencil = PIPE_FORMAT_NONE; + + nvfb->stfb = st_create_framebuffer(glVis, colour, depth, stencil, + driDrawPriv->w, driDrawPriv->h, + (void*)nvfb); + if (!nvfb->stfb) { + free(nvfb); + return GL_FALSE; + } + + driDrawPriv->driverPrivate = (void *)nvfb; + return GL_TRUE; +} + +static void +nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) +{ + struct nouveau_framebuffer *nvfb; + + nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate; + st_unreference_framebuffer(&nvfb->stfb); + free(nvfb); +} + +static struct __DriverAPIRec +nouveau_api = { + .InitDriver = nouveau_screen_create, + .DestroyScreen = nouveau_screen_destroy, + .CreateContext = nouveau_context_create, + .DestroyContext = nouveau_context_destroy, + .CreateBuffer = nouveau_create_buffer, + .DestroyBuffer = nouveau_destroy_buffer, + .SwapBuffers = nouveau_swap_buffers, + .MakeCurrent = nouveau_context_bind, + .UnbindContext = nouveau_context_unbind, + .GetSwapInfo = NULL, + .GetMSC = NULL, + .WaitForMSC = NULL, + .WaitForSBC = NULL, + .SwapBuffersMSC = NULL, + .CopySubBuffer = nouveau_copy_sub_buffer, + .setTexOffset = NULL +}; + +static __GLcontextModes * +nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, + unsigned stencil_bits, GLboolean have_back_buffer) +{ + __GLcontextModes * modes; + __GLcontextModes * m; + unsigned num_modes; + unsigned depth_buffer_factor; + unsigned back_buffer_factor; + int i; + + static const struct { + GLenum format; + GLenum type; + } fb_format_array[] = { + { GL_RGB , GL_UNSIGNED_SHORT_5_6_5 }, + { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV }, + { GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV }, + }; + + /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't + * support pageflipping at all. + */ + static const GLenum back_buffer_modes[] = { + GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML + }; + + u_int8_t depth_bits_array[4] = { 0, 16, 24, 24 }; + u_int8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; + + depth_buffer_factor = 4; + back_buffer_factor = (have_back_buffer) ? 3 : 1; + + num_modes = ((pixel_bits==16) ? 1 : 2) * + depth_buffer_factor * back_buffer_factor * 4; + modes = (*dri_interface->createContextModes)(num_modes, + sizeof(__GLcontextModes)); + m = modes; + + for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) { + if (!driFillInModes(&m, fb_format_array[i].format, + fb_format_array[i].type, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + GLX_TRUE_COLOR)) { + fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__ ); + return NULL; + } + + if (!driFillInModes(&m, fb_format_array[i].format, + fb_format_array[i].type, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + GLX_DIRECT_COLOR)) { + fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__ ); + return NULL; + } + } + + return modes; +} +PUBLIC void * +__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, + __DRIscreen *psc, const __GLcontextModes * modes, + const __DRIversion * ddx_version, + const __DRIversion * dri_version, + const __DRIversion * drm_version, + const __DRIframebuffer * frame_buffer, + void * pSAREA, int fd, int internal_api_version, + const __DRIinterfaceMethods * interface, + __GLcontextModes ** driver_modes) +{ + __DRIscreenPrivate *psp; + static const __DRIversion ddx_expected = + { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; + static const __DRIversion dri_expected = { 4, 0, 0 }; + static const __DRIversion drm_expected = + { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; + struct nouveau_dri *nv_dri = NULL; + + dri_interface = interface; + + if (!driCheckDriDdxDrmVersions2("nouveau", + dri_version, &dri_expected, + ddx_version, &ddx_expected, + drm_version, &drm_expected)) { + return NULL; + } + + if (drm_expected.patch != drm_version->patch) { + fprintf(stderr, "Incompatible DRM patch level.\n" + "Expected: %d\n" "Current : %d\n", + drm_expected.patch, drm_version->patch); + return NULL; + } + + psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, + ddx_version, dri_version, drm_version, + frame_buffer, pSAREA, fd, + internal_api_version, + &nouveau_api); + if (psp == NULL) + return NULL; + nv_dri = psp->pDevPriv; + + *driver_modes = nouveau_fill_in_modes(nv_dri->bpp, + (nv_dri->bpp == 16) ? 16 : 24, + (nv_dri->bpp == 16) ? 0 : 8, + 1); + + driInitExtensions(NULL, card_extensions, GL_FALSE); + + return (void *)psp; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.h b/src/gallium/winsys/dri/nouveau/nouveau_screen.h new file mode 100644 index 00000000000..019823bd44d --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.h @@ -0,0 +1,19 @@ +#ifndef __NOUVEAU_SCREEN_H__ +#define __NOUVEAU_SCREEN_H__ + +#include "xmlconfig.h" +#include "nouveau_device.h" + +struct nouveau_screen { + __DRIscreenPrivate *driScrnPriv; + driOptionCache option_cache; + + struct nouveau_device *device; + + uint32_t front_offset; + uint32_t front_pitch; + uint32_t front_cpp; + uint32_t front_height; +}; + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c new file mode 100644 index 00000000000..91bf243f424 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c @@ -0,0 +1,86 @@ +#include "main/glheader.h" +#include "glapi/glthread.h" +#include + +#include "pipe/p_context.h" +#include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" +#include "state_tracker/st_cb_fbo.h" + +#include "nouveau_context.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +void +nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, + const drm_clip_rect_t *rect) +{ + struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; + drm_clip_rect_t *pbox; + int nbox, i; + + LOCK_HARDWARE(nv); + if (!dPriv->numClipRects) { + UNLOCK_HARDWARE(nv); + return; + } + pbox = dPriv->pClipRects; + nbox = dPriv->numClipRects; + + nv->surface_copy_prep(nv, nv->frontbuffer, surf); + for (i = 0; i < nbox; i++, pbox++) { + int sx, sy, dx, dy, w, h; + + sx = pbox->x1 - dPriv->x; + sy = pbox->y1 - dPriv->y; + dx = pbox->x1; + dy = pbox->y1; + w = pbox->x2 - pbox->x1; + h = pbox->y2 - pbox->y1; + + nv->surface_copy(nv, dx, dy, sx, sy, w, h); + } + + FIRE_RING(); + UNLOCK_HARDWARE(nv); + + if (nv->last_stamp != dPriv->lastStamp) { + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + st_resize_framebuffer(nvfb->stfb, dPriv->w, dPriv->h); + nv->last_stamp = dPriv->lastStamp; + } +} + +void +nouveau_copy_sub_buffer(__DRIdrawablePrivate *dPriv, int x, int y, int w, int h) +{ + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + struct pipe_surface *surf; + + surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + drm_clip_rect_t rect; + rect.x1 = x; + rect.y1 = y; + rect.x2 = x + w; + rect.y2 = y + h; + + st_notify_swapbuffers(nvfb->stfb); + nouveau_copy_buffer(dPriv, surf, &rect); + } +} + +void +nouveau_swap_buffers(__DRIdrawablePrivate *dPriv) +{ + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + struct pipe_surface *surf; + + surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + st_notify_swapbuffers(nvfb->stfb); + nouveau_copy_buffer(dPriv, surf, NULL); + } +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h new file mode 100644 index 00000000000..825d3da6da5 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h @@ -0,0 +1,10 @@ +#ifndef __NOUVEAU_SWAPBUFFERS_H__ +#define __NOUVEAU_SWAPBUFFERS_H__ + +extern void nouveau_copy_buffer(__DRIdrawablePrivate *, struct pipe_surface *, + const drm_clip_rect_t *); +extern void nouveau_copy_sub_buffer(__DRIdrawablePrivate *, + int x, int y, int w, int h); +extern void nouveau_swap_buffers(__DRIdrawablePrivate *); + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c new file mode 100644 index 00000000000..2ca05d84c60 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -0,0 +1,124 @@ +#include "pipe/p_util.h" + +#include "nouveau_context.h" +#include "nouveau_winsys_pipe.h" + +#include "nouveau/nouveau_winsys.h" + +static int +nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, + struct nouveau_notifier **notify) +{ + struct nouveau_context *nv = nvws->nv; + + return nouveau_notifier_alloc(nv->channel, nv->next_handle++, + count, notify); +} + +static int +nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, + struct nouveau_grobj **grobj) +{ + struct nouveau_context *nv = nvws->nv; + int ret; + + ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, + grclass, grobj); + if (ret) + return ret; + + (*grobj)->subc = nv->next_subchannel++; + assert((*grobj)->subc <= 7); + BEGIN_RING_GR(*grobj, 0x0000, 1); + OUT_RING ((*grobj)->handle); + return 0; +} + +static int +nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, struct pipe_surface *src, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct nouveau_context *nv = nvws->nv; + + if (nv->surface_copy_prep(nv, dst, src)) + return 1; + nv->surface_copy(nv, dx, dy, sx, sy, w, h); + nv->surface_copy_done(nv); + + return 0; +} + +static int +nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + if (nvws->nv->surface_fill(nvws->nv, dst, dx, dy, w, h, value)) + return 1; + return 0; +} + +int +nouveau_pipe_emit_reloc(struct nouveau_channel *chan, void *ptr, + struct pipe_buffer *buf, uint32_t data, + uint32_t flags, uint32_t vor, uint32_t tor) +{ + return nouveau_pushbuf_emit_reloc(chan, ptr, nouveau_buffer(buf)->bo, + data, flags, vor, tor); +} + +struct pipe_context * +nouveau_pipe_create(struct nouveau_context *nv) +{ + struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); + struct pipe_context *(*hw_create)(struct pipe_winsys *, + struct nouveau_winsys *, + unsigned); + + if (!nvws) + return NULL; + + switch (nv->chipset & 0xf0) { + case 0x30: + hw_create = nv30_create; + break; + case 0x40: + case 0x60: + hw_create = nv40_create; + break; + case 0x50: + case 0x80: + hw_create = nv50_create; + break; + default: + NOUVEAU_ERR("Unknown chipset NV%02x\n", (int)nv->chipset); + return NULL; + } + + nvws->nv = nv; + nvws->channel = nv->channel; + + nvws->res_init = nouveau_resource_init; + nvws->res_alloc = nouveau_resource_alloc; + nvws->res_free = nouveau_resource_free; + + nvws->push_reloc = nouveau_pipe_emit_reloc; + nvws->push_flush = nouveau_pushbuf_flush; + + nvws->grobj_alloc = nouveau_pipe_grobj_alloc; + nvws->grobj_free = nouveau_grobj_free; + + nvws->notifier_alloc = nouveau_pipe_notifier_alloc; + nvws->notifier_free = nouveau_notifier_free; + nvws->notifier_reset = nouveau_notifier_reset; + nvws->notifier_status = nouveau_notifier_status; + nvws->notifier_retval = nouveau_notifier_return_val; + nvws->notifier_wait = nouveau_notifier_wait_status; + + nvws->surface_copy = nouveau_pipe_surface_copy; + nvws->surface_fill = nouveau_pipe_surface_fill; + + return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c new file mode 100644 index 00000000000..e1a9271395b --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -0,0 +1,196 @@ +#include "pipe/p_winsys.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nouveau_context.h" +#include "nouveau_device.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" +#include "nouveau_winsys_pipe.h" + +static void +nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, + void *context_private) +{ + struct nouveau_context *nv = context_private; + __DRIdrawablePrivate *dPriv = nv->dri_drawable; + + nouveau_copy_buffer(dPriv, surf, NULL); +} + +static void +nouveau_printf(struct pipe_winsys *pws, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + vfprintf(stderr, fmt, args); + va_end(args); +} + +static const char * +nouveau_get_name(struct pipe_winsys *pws) +{ + return "Nouveau/DRI"; +} + +static struct pipe_surface * +nouveau_surface_alloc(struct pipe_winsys *ws) +{ + struct pipe_surface *surf; + + surf = CALLOC_STRUCT(pipe_surface); + if (!surf) + return NULL; + + surf->refcount = 1; + surf->winsys = ws; + return surf; +} + +static int +nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, + unsigned width, unsigned height, + enum pipe_format format, unsigned flags) +{ + unsigned pitch = ((width * pf_get_size(format)) + 63) & ~63; + + surf->format = format; + surf->width = width; + surf->height = height; + surf->cpp = pf_get_size(format); + surf->pitch = pitch / surf->cpp; + + surf->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + pitch * height); + if (!surf->buffer) + return 1; + + return 0; +} + +static void +nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) +{ + struct pipe_surface *surf = *s; + + *s = NULL; + if (--surf->refcount <= 0) { + if (surf->buffer) + pipe_buffer_reference(ws, &surf->buffer, NULL); + free(surf); + } +} + +static struct pipe_buffer * +nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, + unsigned usage, unsigned size) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_pipe_buffer *nvbuf; + uint32_t flags = 0; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.alignment = alignment; + nvbuf->base.usage = usage; + nvbuf->base.size = size; + + flags = NOUVEAU_BO_LOCAL; + if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { + free(nvbuf); + return NULL; + } + + return &nvbuf->base; +} + +static struct pipe_buffer * +nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_pipe_buffer *nvbuf; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.size = bytes; + + if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) { + free(nvbuf); + return NULL; + } + + return &nvbuf->base; +} + +static void +nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + + nouveau_bo_del(&nvbuf->bo); + free(nvbuf); +} + +static void * +nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf, + unsigned flags) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + uint32_t map_flags = 0; + + if (flags & PIPE_BUFFER_USAGE_CPU_READ) + map_flags |= NOUVEAU_BO_RD; + if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) + map_flags |= NOUVEAU_BO_WR; + + if (nouveau_bo_map(nvbuf->bo, map_flags)) + return NULL; + return nvbuf->bo->map; +} + +static void +nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + + nouveau_bo_unmap(nvbuf->bo); +} + +struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv) +{ + struct nouveau_pipe_winsys *nvpws; + struct pipe_winsys *pws; + + nvpws = CALLOC_STRUCT(nouveau_pipe_winsys); + if (!nvpws) + return NULL; + nvpws->nv = nv; + pws = &nvpws->pws; + + pws->flush_frontbuffer = nouveau_flush_frontbuffer; + pws->printf = nouveau_printf; + + pws->surface_alloc = nouveau_surface_alloc; + pws->surface_alloc_storage = nouveau_surface_alloc_storage; + pws->surface_release = nouveau_surface_release; + + pws->buffer_create = nouveau_pipe_bo_create; + pws->buffer_destroy = nouveau_pipe_bo_del; + pws->user_buffer_create = nouveau_pipe_bo_user_create; + pws->buffer_map = nouveau_pipe_bo_map; + pws->buffer_unmap = nouveau_pipe_bo_unmap; + + pws->get_name = nouveau_get_name; + + return &nvpws->pws; +} + diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h new file mode 100644 index 00000000000..6a03ac0d773 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h @@ -0,0 +1,34 @@ +#ifndef NOUVEAU_PIPE_WINSYS_H +#define NOUVEAU_PIPE_WINSYS_H + +#include "pipe/p_context.h" +#include "pipe/p_winsys.h" +#include "nouveau_context.h" + +struct nouveau_pipe_buffer { + struct pipe_buffer base; + struct nouveau_bo *bo; +}; + +static inline struct nouveau_pipe_buffer * +nouveau_buffer(struct pipe_buffer *buf) +{ + return (struct nouveau_pipe_buffer *)buf; +} + +struct nouveau_pipe_winsys { + struct pipe_winsys pws; + + struct nouveau_context *nv; +}; + +extern struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv); + +struct pipe_context * +nouveau_create_softpipe(struct nouveau_context *nv); + +struct pipe_context * +nouveau_pipe_create(struct nouveau_context *nv); + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c new file mode 100644 index 00000000000..0e1b4273d1e --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c @@ -0,0 +1,83 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * + **************************************************************************/ +/* + * Authors: Keith Whitwell + */ + +#include "imports.h" + +#include "pipe/p_defines.h" +#include "pipe/p_format.h" +#include "softpipe/sp_winsys.h" + +#include "nouveau_context.h" +#include "nouveau_winsys_pipe.h" + +struct nouveau_softpipe_winsys { + struct softpipe_winsys sws; + struct nouveau_context *nv; +}; + +/** + * Return list of surface formats supported by this driver. + */ +static boolean +nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) +{ + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + }; + + return FALSE; +} + + + +struct pipe_context * +nouveau_create_softpipe(struct nouveau_context *nv) +{ + struct nouveau_softpipe_winsys *nvsws; + + nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); + + /* Fill in this struct with callbacks that softpipe will need to + * communicate with the window system, buffer manager, etc. + */ + nvsws->sws.is_format_supported = nouveau_is_format_supported; + nvsws->nv = nv; + + /* Create the softpipe context: + */ + return softpipe_create(nouveau_create_pipe_winsys(nv), &nvsws->sws); +} + diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c new file mode 100644 index 00000000000..fe1ea4ed70f --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -0,0 +1,226 @@ +#include "pipe/p_context.h" + +#include "nouveau_context.h" + +static INLINE int +nv04_surface_format(int cpp) +{ + switch (cpp) { + case 1: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; + case 2: return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; + case 4: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; + default: + return -1; + } +} + +static INLINE int +nv04_rect_format(int cpp) +{ + switch (cpp) { + case 1: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + case 2: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; + case 4: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + default: + return -1; + } +} + +static void +nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct pipe_surface *dst = nv->surf_dst; + struct pipe_surface *src = nv->surf_src; + unsigned dst_offset, src_offset; + + dst_offset = dst->offset + (dy * dst->pitch + dx) * dst->cpp; + src_offset = src->offset + (sy * src->pitch + sx) * src->cpp; + + while (h) { + int count = (h > 2047) ? 2047 : h; + + BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_WR); + OUT_RING (src->pitch * src->cpp); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (w * src->cpp); + OUT_RING (count); + OUT_RING (0x0101); + OUT_RING (0); + + h -= count; + src_offset += src->pitch * src->cpp * count; + dst_offset += dst->pitch * dst->cpp * count; + } +} + +static void +nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + BEGIN_RING(NvImageBlit, 0x0300, 3); + OUT_RING ((sy << 16) | sx); + OUT_RING ((dy << 16) | dx); + OUT_RING (( h << 16) | w); +} + +static int +nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, + struct pipe_surface *src) +{ + int format; + + if (src->cpp != dst->cpp) + return 1; + + /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback + * to NV_MEMORY_TO_MEMORY_FORMAT in this case. + */ + if ((src->offset & 63) || (dst->offset & 63)) { + BEGIN_RING(NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | + NOUVEAU_BO_WR); + + nv->surface_copy = nv04_surface_copy_m2mf; + nv->surf_dst = dst; + nv->surf_src = src; + return 0; + + } + + if ((format = nv04_surface_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + nv->surface_copy = nv04_surface_copy_blit; + + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (format); + OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); + OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + return 0; +} + +static void +nv04_surface_copy_done(struct nouveau_context *nv) +{ + FIRE_RING(); +} + +static int +nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + int cs2d_format, gdirect_format; + + if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + if ((gdirect_format = nv04_rect_format(dst->cpp)) < 0) { + NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + return 1; + } + + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (cs2d_format); + OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); + OUT_RING (gdirect_format); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); + OUT_RING (value); + BEGIN_RING(NvGdiRect, + NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); + OUT_RING ((dx << 16) | dy); + OUT_RING (( w << 16) | h); + + FIRE_RING(); + return 0; +} + +int +nouveau_surface_init_nv04(struct nouveau_context *nv) +{ + unsigned class; + int ret; + + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, + &nv->NvM2MF))) { + NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); + return 1; + } + BIND_RING (NvM2MF, nv->next_subchannel++); + BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + + class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : + NV10_CONTEXT_SURFACES_2D; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, + &nv->NvCtxSurf2D))) { + NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); + return 1; + } + BIND_RING (NvCtxSurf2D, nv->next_subchannel++); + BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RING (nv->channel->vram->handle); + OUT_RING (nv->channel->vram->handle); + + class = nv->chipset < 0x10 ? NV04_IMAGE_BLIT : + NV12_IMAGE_BLIT; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, + &nv->NvImageBlit))) { + NOUVEAU_ERR("Error creating blit object: %d\n", ret); + return 1; + } + BIND_RING (NvImageBlit, nv->next_subchannel++); + BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); + OUT_RING (nv->NvCtxSurf2D->handle); + BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); + OUT_RING (NV04_IMAGE_BLIT_OPERATION_SRCCOPY); + + class = NV04_GDI_RECTANGLE_TEXT; + if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, + &nv->NvGdiRect))) { + NOUVEAU_ERR("Error creating rect object: %d\n", ret); + return 1; + } + BIND_RING (NvGdiRect, nv->next_subchannel++); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); + OUT_RING (nv->NvCtxSurf2D->handle); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); + OUT_RING (NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); + BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); + OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); + + nv->surface_copy_prep = nv04_surface_copy_prep; + nv->surface_copy = nv04_surface_copy_blit; + nv->surface_copy_done = nv04_surface_copy_done; + nv->surface_fill = nv04_surface_fill; + return 0; +} + diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c new file mode 100644 index 00000000000..15a10028613 --- /dev/null +++ b/src/gallium/winsys/dri/nouveau/nv50_surface.c @@ -0,0 +1,160 @@ +#include "pipe/p_context.h" + +#include "nouveau_context.h" + +static INLINE int +nv50_format(int cpp) +{ + switch (cpp) { + case 4: return NV50_2D_DST_FORMAT_32BPP; + case 3: return NV50_2D_DST_FORMAT_24BPP; + case 2: return NV50_2D_DST_FORMAT_16BPP; + case 1: return NV50_2D_DST_FORMAT_8BPP; + default: + return -1; + } +} + +static int +nv50_surface_copy_prep(struct nouveau_context *nv, + struct pipe_surface *dst, struct pipe_surface *src) +{ + int surf_format; + + assert(src->cpp == dst->cpp); + + surf_format = nv50_format(dst->cpp); + assert(surf_format >= 0); + + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); + OUT_RING (0); + OUT_RING (0); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + + BEGIN_RING(Nv2D, NV50_2D_SRC_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_SRC_PITCH, 5); + OUT_RING (src->pitch * src->cpp); + OUT_RING (src->pitch); + OUT_RING (src->height); + OUT_RELOCh(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + + return 0; +} + +static void +nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + BEGIN_RING(Nv2D, 0x0110, 1); + OUT_RING (0); + BEGIN_RING(Nv2D, NV50_2D_BLIT_DST_X, 12); + OUT_RING (dx); + OUT_RING (dy); + OUT_RING (w); + OUT_RING (h); + OUT_RING (0); + OUT_RING (1); + OUT_RING (0); + OUT_RING (1); + OUT_RING (0); + OUT_RING (sx); + OUT_RING (0); + OUT_RING (sy); +} + +static void +nv50_surface_copy_done(struct nouveau_context *nv) +{ + FIRE_RING(); +} + +static int +nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + int surf_format, rect_format; + + surf_format = nv50_format(dst->cpp); + if (surf_format < 0) + return 1; + + rect_format = nv50_format(dst->cpp); + if (rect_format < 0) + return 1; + + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY1, 1); + OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); + OUT_RING (surf_format); + OUT_RING (1); + BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); + OUT_RING (dst->pitch * dst->cpp); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); + OUT_RING (0); + OUT_RING (0); + OUT_RING (dst->pitch); + OUT_RING (dst->height); + + BEGIN_RING(Nv2D, 0x0580, 3); + OUT_RING (4); + OUT_RING (rect_format); + OUT_RING (value); + + BEGIN_RING(Nv2D, NV50_2D_RECT_X1, 4); + OUT_RING (dx); + OUT_RING (dy); + OUT_RING (dx + w); + OUT_RING (dy + h); + + FIRE_RING(); + + return 0; +} + +int +nouveau_surface_init_nv50(struct nouveau_context *nv) +{ + int ret; + + ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, NV50_2D, + &nv->Nv2D); + if (ret) + return ret; + BIND_RING (Nv2D, 0); + BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); + OUT_RING (nv->sync_notifier->handle); + BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RING (nv->channel->vram->handle); + OUT_RING (nv->channel->vram->handle); + BEGIN_RING(Nv2D, NV50_2D_OPERATION, 1); + OUT_RING (NV50_2D_OPERATION_SRCCOPY); + + nv->surface_copy_prep = nv50_surface_copy_prep; + nv->surface_copy = nv50_surface_copy; + nv->surface_copy_done = nv50_surface_copy_done; + nv->surface_fill = nv50_surface_fill; + return 0; +} + diff --git a/src/mesa/drivers/dri/nouveau_winsys/Makefile b/src/mesa/drivers/dri/nouveau_winsys/Makefile deleted file mode 100644 index 98ec5a79f5d..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/Makefile +++ /dev/null @@ -1,43 +0,0 @@ - -TOP = ../../../../.. -include $(TOP)/configs/current - -LIBNAME = nouveau_dri.so - -MINIGLX_SOURCES = - -PIPE_DRIVERS = \ - $(TOP)/src/mesa/pipe/softpipe/libsoftpipe.a \ - $(TOP)/src/mesa/pipe/nv30/libnv30.a \ - $(TOP)/src/mesa/pipe/nv40/libnv40.a \ - $(TOP)/src/mesa/pipe/nv50/libnv50.a - -DRIVER_SOURCES = \ - nouveau_bo.c \ - nouveau_channel.c \ - nouveau_context.c \ - nouveau_device.c \ - nouveau_dma.c \ - nouveau_fence.c \ - nouveau_grobj.c \ - nouveau_lock.c \ - nouveau_notifier.c \ - nouveau_pushbuf.c \ - nouveau_resource.c \ - nouveau_screen.c \ - nouveau_swapbuffers.c \ - nouveau_winsys.c \ - nouveau_winsys_pipe.c \ - nouveau_winsys_softpipe.c \ - nv04_surface.c \ - nv50_surface.c - -C_SOURCES = \ - $(COMMON_GALLIUM_SOURCES) \ - $(DRIVER_SOURCES) - -ASM_SOURCES = - -include ../Makefile.template - -symlinks: diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c deleted file mode 100644 index 6887ffa6886..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_bo.c +++ /dev/null @@ -1,402 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" -#include "nouveau_local.h" - -static void -nouveau_mem_free(struct nouveau_device *dev, struct drm_nouveau_mem_alloc *ma, - void **map) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct drm_nouveau_mem_free mf; - - if (map && *map) { - drmUnmap(*map, ma->size); - *map = NULL; - } - - if (ma->size) { - mf.offset = ma->offset; - mf.flags = ma->flags; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, - &mf, sizeof(mf)); - ma->size = 0; - } -} - -static int -nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, - uint32_t flags, struct drm_nouveau_mem_alloc *ma, void **map) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - int ret; - - ma->alignment = align; - ma->size = size; - ma->flags = flags; - if (map) - ma->flags |= NOUVEAU_MEM_MAPPED; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, ma, - sizeof(struct drm_nouveau_mem_alloc)); - if (ret) - return ret; - - if (map) { - ret = drmMap(nvdev->fd, ma->map_handle, ma->size, map); - if (ret) { - *map = NULL; - nouveau_mem_free(dev, ma, map); - return ret; - } - } - - return 0; -} - -static void -nouveau_bo_tmp_del(void *priv) -{ - struct nouveau_resource *r = priv; - - nouveau_fence_ref(NULL, (struct nouveau_fence **)&r->priv); - nouveau_resource_free(&r); -} - -static struct nouveau_resource * -nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, - struct nouveau_fence *fence) -{ - struct nouveau_device_priv *nvdev = nouveau_device(chan->device); - struct nouveau_resource *r = NULL; - struct nouveau_fence *ref = NULL; - - if (fence) - nouveau_fence_ref(fence, &ref); - else - nouveau_fence_new(chan, &ref); - assert(ref); - - while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) { - nouveau_fence_flush(chan); - } - nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r); - - return r; -} - -int -nouveau_bo_init(struct nouveau_device *dev) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - int ret; - - ret = nouveau_mem_alloc(dev, 128*1024, 0, NOUVEAU_MEM_AGP | - NOUVEAU_MEM_PCI, &nvdev->sa, &nvdev->sa_map); - if (ret) - return ret; - - ret = nouveau_resource_init(&nvdev->sa_heap, 0, nvdev->sa.size); - if (ret) { - nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); - return ret; - } - - return 0; -} - -void -nouveau_bo_takedown(struct nouveau_device *dev) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); -} - -int -nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, - int size, struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo; - int ret; - - if (!dev || !bo || *bo) - return -EINVAL; - - nvbo = calloc(1, sizeof(struct nouveau_bo_priv)); - if (!nvbo) - return -ENOMEM; - nvbo->base.device = dev; - nvbo->base.size = size; - nvbo->base.handle = bo_to_ptr(nvbo); - nvbo->drm.alignment = align; - nvbo->refcount = 1; - - ret = nouveau_bo_set_status(&nvbo->base, flags); - if (ret) { - free(nvbo); - return ret; - } - - *bo = &nvbo->base; - return 0; -} - -int -nouveau_bo_user(struct nouveau_device *dev, void *ptr, int size, - struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo; - - if (!dev || !bo || *bo) - return -EINVAL; - - nvbo = calloc(1, sizeof(*nvbo)); - if (!nvbo) - return -ENOMEM; - nvbo->base.device = dev; - - nvbo->sysmem = ptr; - nvbo->user = 1; - - nvbo->base.size = size; - nvbo->base.offset = nvbo->drm.offset; - nvbo->base.handle = bo_to_ptr(nvbo); - nvbo->refcount = 1; - *bo = &nvbo->base; - return 0; -} - -int -nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, - struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo = ptr_to_bo(handle); - - if (!dev || !bo || *bo) - return -EINVAL; - - nvbo->refcount++; - *bo = &nvbo->base; - return 0; -} - -void -nouveau_bo_del(struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo; - - if (!bo || !*bo) - return; - nvbo = nouveau_bo(*bo); - *bo = NULL; - - if (--nvbo->refcount) - return; - - if (nvbo->fence) - nouveau_fence_wait(&nvbo->fence); - nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); - if (nvbo->sysmem && !nvbo->user) - free(nvbo->sysmem); - free(nvbo); -} - -int -nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - - if (!nvbo) - return -EINVAL; - - if (flags & NOUVEAU_BO_WR) - nouveau_fence_wait(&nvbo->fence); - else - nouveau_fence_wait(&nvbo->wr_fence); - - if (nvbo->sysmem) - bo->map = nvbo->sysmem; - else - bo->map = nvbo->map; - return 0; -} - -void -nouveau_bo_unmap(struct nouveau_bo *bo) -{ - bo->map = NULL; -} - -static int -nouveau_bo_upload(struct nouveau_bo_priv *nvbo) -{ - if (nvbo->fence) - nouveau_fence_wait(&nvbo->fence); - memcpy(nvbo->map, nvbo->sysmem, nvbo->drm.size); - return 0; -} - -int -nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - struct drm_nouveau_mem_alloc new; - void *new_map = NULL, *new_sysmem = NULL; - unsigned new_flags = 0, ret; - - assert(!bo->map); - - /* Check current memtype vs requested, if they match do nothing */ - if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) - return 0; - if ((nvbo->drm.flags & NOUVEAU_MEM_AGP) && (flags & NOUVEAU_BO_GART)) - return 0; - if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) - return 0; - - memset(&new, 0x00, sizeof(new)); - - /* Allocate new memory */ - if (flags & NOUVEAU_BO_VRAM) - new_flags |= NOUVEAU_MEM_FB; - else - if (flags & NOUVEAU_BO_GART) - new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); - - if (new_flags) { - ret = nouveau_mem_alloc(bo->device, bo->size, - nvbo->drm.alignment, new_flags, - &new, &new_map); - if (ret) - return ret; - } else { - new_sysmem = malloc(bo->size); - } - - /* Copy old -> new */ - /*XXX: use M2MF */ - if (nvbo->sysmem || nvbo->map) { - nouveau_bo_map(bo, NOUVEAU_BO_RD); - memcpy(new_map, bo->map, bo->size); - nouveau_bo_unmap(bo); - } - - /* Free old memory */ - if (nvbo->fence) - nouveau_fence_wait(&nvbo->fence); - nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); - if (nvbo->sysmem) - free(nvbo->sysmem); - - nvbo->drm = new; - nvbo->map = new_map; - nvbo->sysmem = new_sysmem; - bo->flags = flags; - bo->offset = nvbo->drm.offset; - return 0; -} - -static int -nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_device_priv *nvdev = nouveau_device(chan->device); - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - struct nouveau_resource *r; - - if (nvchan->user_charge + bo->size > nvdev->sa.size) - return 1; - nvchan->user_charge += bo->size; - - if (!(flags & NOUVEAU_BO_GART)) - return 1; - - r = nouveau_bo_tmp(chan, bo->size, fence); - if (!r) - return 1; - - memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); - - nvbo->offset = nvdev->sa.offset + r->start; - nvbo->flags = NOUVEAU_BO_GART; - return 0; -} - -static int -nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - int ret; - - ret = nouveau_bo_set_status(bo, flags); - if (ret) - return ret; - - if (nvbo->user) - nouveau_bo_upload(nvbo); - - nvbo->offset = nvbo->drm.offset; - if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) - nvbo->flags = NOUVEAU_BO_GART; - else - nvbo->flags = NOUVEAU_BO_VRAM; - - return 0; -} - -int -nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - int ret; - - assert(bo->map == NULL); - - if (nvbo->user) { - ret = nouveau_bo_validate_user(chan, bo, fence, flags); - if (ret) { - ret = nouveau_bo_validate_bo(chan, bo, fence, flags); - if (ret) - return ret; - } - } else { - ret = nouveau_bo_validate_bo(chan, bo, fence, flags); - if (ret) - return ret; - } - - if (flags & NOUVEAU_BO_WR) - nouveau_fence_ref(fence, &nvbo->wr_fence); - nouveau_fence_ref(fence, &nvbo->fence); - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c deleted file mode 100644 index df80d04add5..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_channel.c +++ /dev/null @@ -1,118 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" - -int -nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma, - uint32_t tt_ctxdma, struct nouveau_channel **chan) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct nouveau_channel_priv *nvchan; - int ret; - - if (!nvdev || !chan || *chan) - return -EINVAL; - - nvchan = calloc(1, sizeof(struct nouveau_channel_priv)); - if (!nvchan) - return -ENOMEM; - nvchan->base.device = dev; - - nvchan->drm.fb_ctxdma_handle = fb_ctxdma; - nvchan->drm.tt_ctxdma_handle = tt_ctxdma; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_CHANNEL_ALLOC, - &nvchan->drm, sizeof(nvchan->drm)); - if (ret) { - free(nvchan); - return ret; - } - - nvchan->base.id = nvchan->drm.channel; - if (nouveau_grobj_ref(&nvchan->base, nvchan->drm.fb_ctxdma_handle, - &nvchan->base.vram) || - nouveau_grobj_ref(&nvchan->base, nvchan->drm.tt_ctxdma_handle, - &nvchan->base.gart)) { - nouveau_channel_free((void *)&nvchan); - return -EINVAL; - } - - ret = drmMap(nvdev->fd, nvchan->drm.ctrl, nvchan->drm.ctrl_size, - (void*)&nvchan->user); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - nvchan->put = &nvchan->user[0x40/4]; - nvchan->get = &nvchan->user[0x44/4]; - nvchan->ref_cnt = &nvchan->user[0x48/4]; - - ret = drmMap(nvdev->fd, nvchan->drm.notifier, nvchan->drm.notifier_size, - (drmAddressPtr)&nvchan->notifier_block); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - - ret = drmMap(nvdev->fd, nvchan->drm.cmdbuf, nvchan->drm.cmdbuf_size, - (void*)&nvchan->pushbuf); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - - nouveau_dma_channel_init(&nvchan->base); - nouveau_pushbuf_init(&nvchan->base); - - *chan = &nvchan->base; - return 0; -} - -void -nouveau_channel_free(struct nouveau_channel **chan) -{ - struct nouveau_channel_priv *nvchan; - struct nouveau_device_priv *nvdev; - struct drm_nouveau_channel_free cf; - - if (!chan || !*chan) - return; - nvchan = nouveau_channel(*chan); - *chan = NULL; - nvdev = nouveau_device(nvchan->base.device); - - FIRE_RING_CH(&nvchan->base); - - nouveau_grobj_free(&nvchan->base.vram); - nouveau_grobj_free(&nvchan->base.gart); - - cf.channel = nvchan->drm.channel; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf)); - free(nvchan); -} - - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c deleted file mode 100644 index 01fada5b89b..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.c +++ /dev/null @@ -1,206 +0,0 @@ -#include "main/glheader.h" -#include "glapi/glthread.h" -#include -#include "utils.h" - -#include "state_tracker/st_public.h" -#include "pipe/p_defines.h" -#include "pipe/p_context.h" - -#include "nouveau_context.h" -#include "nouveau_dri.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_winsys_pipe.h" - -#ifdef DEBUG -static const struct dri_debug_control debug_control[] = { - { "bo", DEBUG_BO }, - { NULL, 0 } -}; -int __nouveau_debug = 0; -#endif - -GLboolean -nouveau_context_create(const __GLcontextModes *glVis, - __DRIcontextPrivate *driContextPriv, - void *sharedContextPrivate) -{ - __DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv; - struct nouveau_screen *nv_screen = driScrnPriv->private; - struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); - struct nouveau_device_priv *nvdev; - struct pipe_context *pipe = NULL; - struct st_context *st_share = NULL; - int ret; - - if (sharedContextPrivate) { - st_share = ((struct nouveau_context *)sharedContextPrivate)->st; - } - - if ((ret = nouveau_device_get_param(nv_screen->device, - NOUVEAU_GETPARAM_CHIPSET_ID, - &nv->chipset))) { - NOUVEAU_ERR("Error determining chipset id: %d\n", ret); - return GL_FALSE; - } - - if ((ret = nouveau_channel_alloc(nv_screen->device, - 0x8003d001, 0x8003d002, - &nv->channel))) { - NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); - return GL_FALSE; - } - - driContextPriv->driverPrivate = (void *)nv; - nv->nv_screen = nv_screen; - nv->dri_screen = driScrnPriv; - - nvdev = nouveau_device(nv_screen->device); - nvdev->ctx = driContextPriv->hHWContext; - nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; - - driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, - nv->dri_screen->myNum, "nouveau"); -#ifdef DEBUG - __nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"), - debug_control); -#endif - - /*XXX: Hack up a fake region and buffer object for front buffer. - * This will go away with TTM, replaced with a simple reference - * of the front buffer handle passed to us by the DDX. - */ - { - struct pipe_surface *fb_surf; - struct nouveau_pipe_buffer *fb_buf; - struct nouveau_bo_priv *fb_bo; - - fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); - fb_bo->drm.offset = nv_screen->front_offset; - fb_bo->drm.flags = NOUVEAU_MEM_FB; - fb_bo->drm.size = nv_screen->front_pitch * - nv_screen->front_height; - fb_bo->refcount = 1; - fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; - fb_bo->base.offset = fb_bo->drm.offset; - fb_bo->base.handle = (unsigned long)fb_bo; - fb_bo->base.size = fb_bo->drm.size; - fb_bo->base.device = nv_screen->device; - - fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer)); - fb_buf->bo = &fb_bo->base; - - fb_surf = calloc(1, sizeof(struct pipe_surface)); - fb_surf->cpp = nv_screen->front_cpp; - fb_surf->pitch = nv_screen->front_pitch / fb_surf->cpp; - fb_surf->height = nv_screen->front_height; - fb_surf->refcount = 1; - fb_surf->buffer = &fb_buf->base; - - nv->frontbuffer = fb_surf; - } - - if ((ret = nouveau_grobj_alloc(nv->channel, 0x00000000, 0x30, - &nv->NvNull))) { - NOUVEAU_ERR("Error creating NULL object: %d\n", ret); - return GL_FALSE; - } - nv->next_handle = 0x80000000; - - if ((ret = nouveau_notifier_alloc(nv->channel, nv->next_handle++, 1, - &nv->sync_notifier))) { - NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); - return GL_FALSE; - } - - if (nv->chipset < 0x50) - ret = nouveau_surface_init_nv04(nv); - else - ret = nouveau_surface_init_nv50(nv); - if (ret) { - return GL_FALSE; - } - - if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { - pipe = nouveau_pipe_create(nv); - if (!pipe) - NOUVEAU_ERR("Couldn't create hw pipe\n"); - } - - if (!pipe) { - NOUVEAU_MSG("Using softpipe\n"); - pipe = nouveau_create_softpipe(nv); - if (!pipe) { - NOUVEAU_ERR("Error creating pipe, bailing\n"); - return GL_FALSE; - } - } - - pipe->priv = nv; - nv->st = st_create_context(pipe, glVis, st_share); - return GL_TRUE; -} - -void -nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) -{ - struct nouveau_context *nv = driContextPriv->driverPrivate; - - assert(nv); - - st_flush(nv->st, PIPE_FLUSH_WAIT); - st_destroy_context(nv->st); - - nouveau_grobj_free(&nv->NvCtxSurf2D); - nouveau_grobj_free(&nv->NvImageBlit); - nouveau_channel_free(&nv->channel); - - free(nv); -} - -GLboolean -nouveau_context_bind(__DRIcontextPrivate *driContextPriv, - __DRIdrawablePrivate *driDrawPriv, - __DRIdrawablePrivate *driReadPriv) -{ - struct nouveau_context *nv; - struct nouveau_framebuffer *draw, *read; - - if (!driContextPriv) { - st_make_current(NULL, NULL, NULL); - return GL_TRUE; - } - - nv = driContextPriv->driverPrivate; - draw = driDrawPriv->driverPrivate; - read = driReadPriv->driverPrivate; - - st_make_current(nv->st, draw->stfb, read->stfb); - - if ((nv->dri_drawable != driDrawPriv) || - (nv->last_stamp != driDrawPriv->lastStamp)) { - nv->dri_drawable = driDrawPriv; - st_resize_framebuffer(draw->stfb, driDrawPriv->w, - driDrawPriv->h); - nv->last_stamp = driDrawPriv->lastStamp; - } - - if (driDrawPriv != driReadPriv) { - st_resize_framebuffer(read->stfb, driReadPriv->w, - driReadPriv->h); - } - - return GL_TRUE; -} - -GLboolean -nouveau_context_unbind(__DRIcontextPrivate *driContextPriv) -{ - struct nouveau_context *nv = driContextPriv->driverPrivate; - (void)nv; - - st_flush(nv->st, 0); - return GL_TRUE; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h deleted file mode 100644 index 7a74f7deecd..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_context.h +++ /dev/null @@ -1,89 +0,0 @@ -#ifndef __NOUVEAU_CONTEXT_H__ -#define __NOUVEAU_CONTEXT_H__ - -#include "dri_util.h" -#include "xmlconfig.h" - -#include "pipe/nouveau/nouveau_winsys.h" -#include "nouveau_device.h" -#include "nouveau_drmif.h" -#include "nouveau_dma.h" - -struct nouveau_framebuffer { - struct st_framebuffer *stfb; -}; - -struct nouveau_context { - struct st_context *st; - - /* Misc HW info */ - uint64_t chipset; - - /* DRI stuff */ - __DRIscreenPrivate *dri_screen; - __DRIdrawablePrivate *dri_drawable; - unsigned int last_stamp; - driOptionCache dri_option_cache; - drm_context_t drm_context; - drmLock drm_lock; - GLboolean locked; - struct nouveau_screen *nv_screen; - struct pipe_surface *frontbuffer; - - /* Hardware context */ - struct nouveau_channel *channel; - struct nouveau_notifier *sync_notifier; - struct nouveau_grobj *NvNull; - struct nouveau_grobj *NvCtxSurf2D; - struct nouveau_grobj *NvImageBlit; - struct nouveau_grobj *NvGdiRect; - struct nouveau_grobj *NvM2MF; - struct nouveau_grobj *Nv2D; - uint32_t next_handle; - uint32_t next_subchannel; - uint32_t next_sequence; - - /* pipe_surface accel */ - struct pipe_surface *surf_src, *surf_dst; - unsigned surf_src_offset, surf_dst_offset; - int (*surface_copy_prep)(struct nouveau_context *, - struct pipe_surface *dst, - struct pipe_surface *src); - void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h); - void (*surface_copy_done)(struct nouveau_context *); - int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, - unsigned, unsigned, unsigned, unsigned, unsigned); -}; - -extern GLboolean nouveau_context_create(const __GLcontextModes *, - __DRIcontextPrivate *, void *); -extern void nouveau_context_destroy(__DRIcontextPrivate *); -extern GLboolean nouveau_context_bind(__DRIcontextPrivate *, - __DRIdrawablePrivate *draw, - __DRIdrawablePrivate *read); -extern GLboolean nouveau_context_unbind(__DRIcontextPrivate *); - -#ifdef DEBUG -extern int __nouveau_debug; - -#define DEBUG_BO (1 << 0) - -#define DBG(flag, ...) do { \ - if (__nouveau_debug & (DEBUG_##flag)) \ - NOUVEAU_ERR(__VA_ARGS__); \ -} while(0) -#else -#define DBG(flag, ...) -#endif - -extern void LOCK_HARDWARE(struct nouveau_context *); -extern void UNLOCK_HARDWARE(struct nouveau_context *); - -extern int nouveau_surface_init_nv04(struct nouveau_context *); -extern int nouveau_surface_init_nv50(struct nouveau_context *); - -extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int); -extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *); - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c deleted file mode 100644 index 409e4415f76..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.c +++ /dev/null @@ -1,146 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" - -int -nouveau_device_open_existing(struct nouveau_device **dev, int close, - int fd, drm_context_t ctx) -{ - struct nouveau_device_priv *nvdev; - int ret; - - if (!dev || *dev) - return -EINVAL; - - nvdev = calloc(1, sizeof(*nvdev)); - if (!nvdev) - return -ENOMEM; - nvdev->fd = fd; - nvdev->ctx = ctx; - nvdev->needs_close = close; - - drmCommandNone(nvdev->fd, DRM_NOUVEAU_CARD_INIT); - - if ((ret = nouveau_bo_init(&nvdev->base))) { - nouveau_device_close((void *)&nvdev); - return ret; - } - - *dev = &nvdev->base; - return 0; -} - -int -nouveau_device_open(struct nouveau_device **dev, const char *busid) -{ - drm_context_t ctx; - int fd, ret; - - if (!dev || *dev) - return -EINVAL; - - fd = drmOpen("nouveau", busid); - if (fd < 0) - return -EINVAL; - - ret = drmCreateContext(fd, &ctx); - if (ret) { - drmClose(fd); - return ret; - } - - ret = nouveau_device_open_existing(dev, 1, fd, ctx); - if (ret) { - drmDestroyContext(fd, ctx); - drmClose(fd); - return ret; - } - - return 0; -} - -void -nouveau_device_close(struct nouveau_device **dev) -{ - struct nouveau_device_priv *nvdev; - - if (dev || !*dev) - return; - nvdev = nouveau_device(*dev); - *dev = NULL; - - nouveau_bo_takedown(&nvdev->base); - - if (nvdev->needs_close) { - drmDestroyContext(nvdev->fd, nvdev->ctx); - drmClose(nvdev->fd); - } - free(nvdev); -} - -int -nouveau_device_get_param(struct nouveau_device *dev, - uint64_t param, uint64_t *value) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct drm_nouveau_getparam g; - int ret; - - if (!nvdev || !value) - return -EINVAL; - - g.param = param; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_GETPARAM, - &g, sizeof(g)); - if (ret) - return ret; - - *value = g.value; - return 0; -} - -int -nouveau_device_set_param(struct nouveau_device *dev, - uint64_t param, uint64_t value) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct drm_nouveau_setparam s; - int ret; - - if (!nvdev) - return -EINVAL; - - s.param = param; - s.value = value; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_SETPARAM, - &s, sizeof(s)); - if (ret) - return ret; - - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h deleted file mode 100644 index 744a89f74bf..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_device.h +++ /dev/null @@ -1,29 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_DEVICE_H__ -#define __NOUVEAU_DEVICE_H__ - -struct nouveau_device { -}; - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c deleted file mode 100644 index f8a8ba04f6d..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.c +++ /dev/null @@ -1,219 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" -#include "nouveau_local.h" - -static inline uint32_t -READ_GET(struct nouveau_channel_priv *nvchan) -{ - return *nvchan->get; -} - -static inline void -WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) -{ - uint32_t put = ((val << 2) + nvchan->dma->base); - volatile int dum; - - NOUVEAU_DMA_BARRIER; - dum = READ_GET(nvchan); - - *nvchan->put = put; - nvchan->dma->put = val; -#ifdef NOUVEAU_DMA_TRACE - NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put); -#endif - - NOUVEAU_DMA_BARRIER; -} - -static inline int -LOCAL_GET(struct nouveau_dma_priv *dma, uint32_t *val) -{ - uint32_t get = *val; - - if (get >= dma->base && get <= (dma->base + (dma->max << 2))) { - *val = (get - dma->base) >> 2; - return 1; - } - - return 0; -} - -void -nouveau_dma_channel_init(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - int i; - - nvchan->dma = &nvchan->dma_master; - nvchan->dma->base = nvchan->drm.put_base; - nvchan->dma->cur = nvchan->dma->put = 0; - nvchan->dma->max = (nvchan->drm.cmdbuf_size >> 2) - 2; - nvchan->dma->free = nvchan->dma->max - nvchan->dma->cur; - - RING_SPACE_CH(chan, RING_SKIPS); - for (i = 0; i < RING_SKIPS; i++) - OUT_RING_CH(chan, 0); -} - -#define CHECK_TIMEOUT() do { \ - if ((NOUVEAU_TIME_MSEC() - t_start) > NOUVEAU_DMA_TIMEOUT) \ - return - EBUSY; \ -} while(0) - -int -nouveau_dma_wait(struct nouveau_channel *chan, int size) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - uint32_t get, t_start; - - FIRE_RING_CH(chan); - - t_start = NOUVEAU_TIME_MSEC(); - while (dma->free < size) { - CHECK_TIMEOUT(); - - get = READ_GET(nvchan); - if (!LOCAL_GET(dma, &get)) - continue; - - if (dma->put >= get) { - dma->free = dma->max - dma->cur; - - if (dma->free < size) { -#ifdef NOUVEAU_DMA_DEBUG - dma->push_free = 1; -#endif - OUT_RING_CH(chan, 0x20000000 | dma->base); - if (get <= RING_SKIPS) { - /*corner case - will be idle*/ - if (dma->put <= RING_SKIPS) - WRITE_PUT(nvchan, - RING_SKIPS + 1); - - do { - CHECK_TIMEOUT(); - get = READ_GET(nvchan); - if (!LOCAL_GET(dma, &get)) - get = 0; - } while (get <= RING_SKIPS); - } - - WRITE_PUT(nvchan, RING_SKIPS); - dma->cur = dma->put = RING_SKIPS; - dma->free = get - (RING_SKIPS + 1); - } - } else { - dma->free = get - dma->cur - 1; - } - } - - return 0; -} - -#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF -static void -nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - unsigned mthd_count = 0; - - while (get != put) { - uint32_t gpuget = (get << 2) + nvchan->drm.put_base; - uint32_t data; - - if (get < 0 || get >= nvchan->drm.cmdbuf_size) { - NOUVEAU_ERR("DMA_PT 0x%08x\n", gpuget); - assert(0); - } - data = nvchan->pushbuf[get++]; - - if (mthd_count) { - NOUVEAU_MSG("0x%08x 0x%08x\n", gpuget, data); - mthd_count--; - continue; - } - - switch (data & 0x60000000) { - case 0x00000000: - mthd_count = (data >> 18) & 0x7ff; - NOUVEAU_MSG("0x%08x 0x%08x MTHD " - "Sc %d Mthd 0x%04x Size %d\n", - gpuget, data, (data>>13) & 7, data & 0x1ffc, - mthd_count); - break; - case 0x20000000: - get = (data & 0x1ffffffc) >> 2; - NOUVEAU_MSG("0x%08x 0x%08x JUMP 0x%08x\n", - gpuget, data, data & 0x1ffffffc); - continue; - case 0x40000000: - mthd_count = (data >> 18) & 0x7ff; - NOUVEAU_MSG("0x%08x 0x%08x NINC " - "Sc %d Mthd 0x%04x Size %d\n", - gpuget, data, (data>>13) & 7, data & 0x1ffc, - mthd_count); - break; - case 0x60000000: - /* DMA_OPCODE_CALL apparently, doesn't seem to work on - * my NV40 at least.. - */ - /* fall-through */ - default: - NOUVEAU_MSG("DMA_PUSHER 0x%08x 0x%08x\n", - gpuget, data); - assert(0); - } - } -} -#endif - -void -nouveau_dma_kickoff(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - - if (dma->cur == dma->put) - return; - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free) { - NOUVEAU_ERR("Packet incomplete: %d left\n", dma->push_free); - return; - } -#endif - -#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF - nouveau_dma_parse_pushbuf(chan, dma->put, dma->cur); -#endif - - WRITE_PUT(nvchan, dma->cur); -} diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h deleted file mode 100644 index cfa6d26e828..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dma.h +++ /dev/null @@ -1,143 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_DMA_H__ -#define __NOUVEAU_DMA_H__ - -#include -#include "nouveau_drmif.h" -#include "nouveau_local.h" - -#define RING_SKIPS 8 - -extern int nouveau_dma_wait(struct nouveau_channel *chan, int size); -extern void nouveau_dma_subc_bind(struct nouveau_grobj *); -extern void nouveau_dma_channel_init(struct nouveau_channel *); -extern void nouveau_dma_kickoff(struct nouveau_channel *); - -#ifdef NOUVEAU_DMA_DEBUG -static char faulty[1024]; -#endif - -static inline void -nouveau_dma_out(struct nouveau_channel *chan, uint32_t data) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free == 0) { - NOUVEAU_ERR("No space left in packet at %s\n", faulty); - return; - } - dma->push_free--; -#endif -#ifdef NOUVEAU_DMA_TRACE - { - uint32_t offset = (dma->cur << 2) + dma->base; - NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", - nvchan->drm.channel, offset, data); - } -#endif - nvchan->pushbuf[dma->cur + (dma->base - nvchan->drm.put_base)/4] = data; - dma->cur++; -} - -static inline void -nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - (void)dma; - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free < size) { - NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", - dma->push_free, size); - return; - } -#endif -#ifdef NOUVEAU_DMA_TRACE - while (size--) { - nouveau_dma_out(chan, *ptr); - ptr++; - } -#else - memcpy(&nvchan->pushbuf[dma->cur], ptr, size << 2); -#ifdef NOUVEAU_DMA_DEBUG - dma->push_free -= size; -#endif - dma->cur += size; -#endif -} - -static inline void -nouveau_dma_space(struct nouveau_channel *chan, int size) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - - if (dma->free < size) { - if (nouveau_dma_wait(chan, size) && chan->hang_notify) - chan->hang_notify(chan); - } - dma->free -= size; -#ifdef NOUVEAU_DMA_DEBUG - dma->push_free = size; -#endif -} - -static inline void -nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, - int method, int size, const char* file, int line) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - (void)dma; - -#ifdef NOUVEAU_DMA_TRACE - NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, - grobj->handle, grobj->subc, method, size); -#endif - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free) { - NOUVEAU_ERR("Previous packet incomplete: %d left at %s\n", - dma->push_free, faulty); - return; - } - sprintf(faulty,"%s:%d",file,line); -#endif - - nouveau_dma_space(chan, (size + 1)); - nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); -} - -#define RING_SPACE_CH(ch,sz) nouveau_dma_space((ch), (sz)) -#define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) -#define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) -#define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ - (dwords)) -#define FIRE_RING_CH(ch) nouveau_dma_kickoff((ch)) -#define WAIT_RING_CH(ch,sz) nouveau_dma_wait((ch), (sz)) - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h deleted file mode 100644 index 1207c2d609c..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_dri.h +++ /dev/null @@ -1,28 +0,0 @@ -#ifndef _NOUVEAU_DRI_ -#define _NOUVEAU_DRI_ - -#include "xf86drm.h" -#include "drm.h" -#include "nouveau_drm.h" - -struct nouveau_dri { - uint32_t device_id; /**< \brief PCI device ID */ - uint32_t width; /**< \brief width in pixels of display */ - uint32_t height; /**< \brief height in scanlines of display */ - uint32_t depth; /**< \brief depth of display (8, 15, 16, 24) */ - uint32_t bpp; /**< \brief bit depth of display (8, 16, 24, 32) */ - - uint32_t bus_type; /**< \brief ths bus type */ - uint32_t bus_mode; /**< \brief bus mode (used for AGP, maybe also for PCI-E ?) */ - - uint32_t front_offset; /**< \brief front buffer offset */ - uint32_t front_pitch; /**< \brief front buffer pitch */ - uint32_t back_offset; /**< \brief private back buffer offset */ - uint32_t back_pitch; /**< \brief private back buffer pitch */ - uint32_t depth_offset; /**< \brief private depth buffer offset */ - uint32_t depth_pitch; /**< \brief private depth buffer pitch */ - -}; - -#endif - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h deleted file mode 100644 index 67e19f1cfe8..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_drmif.h +++ /dev/null @@ -1,304 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_DRMIF_H__ -#define __NOUVEAU_DRMIF_H__ - -#include -#include -#include - -#include "nouveau_device.h" -#include "pipe/nouveau/nouveau_channel.h" -#include "pipe/nouveau/nouveau_grobj.h" -#include "pipe/nouveau/nouveau_notifier.h" -#include "pipe/nouveau/nouveau_bo.h" -#include "pipe/nouveau/nouveau_resource.h" -#include "pipe/nouveau/nouveau_pushbuf.h" - -struct nouveau_device_priv { - struct nouveau_device base; - - int fd; - drm_context_t ctx; - drmLock *lock; - int needs_close; - - struct drm_nouveau_mem_alloc sa; - void *sa_map; - struct nouveau_resource *sa_heap; -}; -#define nouveau_device(n) ((struct nouveau_device_priv *)(n)) - -extern int -nouveau_device_open_existing(struct nouveau_device **, int close, - int fd, drm_context_t ctx); - -extern int -nouveau_device_open(struct nouveau_device **, const char *busid); - -extern void -nouveau_device_close(struct nouveau_device **); - -extern int -nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v); - -extern int -nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val); - -struct nouveau_fence { - struct nouveau_channel *channel; -}; - -struct nouveau_fence_cb { - struct nouveau_fence_cb *next; - void (*func)(void *); - void *priv; -}; - -struct nouveau_fence_priv { - struct nouveau_fence base; - int refcount; - - struct nouveau_fence *next; - struct nouveau_fence_cb *signal_cb; - - uint32_t sequence; - int emitted; - int signalled; -}; -#define nouveau_fence(n) ((struct nouveau_fence_priv *)(n)) - -extern int -nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **); - -extern int -nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); - -extern int -nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *); - -extern void -nouveau_fence_emit(struct nouveau_fence *); - -extern int -nouveau_fence_wait(struct nouveau_fence **); - -extern void -nouveau_fence_flush(struct nouveau_channel *); - -struct nouveau_pushbuf_reloc { - uint64_t next; - uint64_t handle; - uint32_t *ptr; - uint32_t flags; - uint32_t data; - uint32_t vor; - uint32_t tor; -}; - -struct nouveau_pushbuf_bo { - uint64_t next; - uint64_t handle; - uint64_t flags; - uint64_t relocs; - int nr_relocs; -}; - -struct nouveau_pushbuf_priv { - struct nouveau_pushbuf base; - - unsigned nop_jump; - unsigned start; - unsigned size; - - uint64_t buffers; - int nr_buffers; -}; -#define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) - -#define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o)) -#define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h)) -#define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o)) -#define ptr_to_pbrel(h) ((struct nouveau_pushbuf_reloc *)(unsigned long)(h)) -#define bo_to_ptr(o) ((uint64_t)(unsigned long)(o)) -#define ptr_to_bo(h) ((struct nouveau_bo_priv *)(unsigned long)(h)) - -extern int -nouveau_pushbuf_init(struct nouveau_channel *); - -extern int -nouveau_pushbuf_flush(struct nouveau_channel *, unsigned min); - -extern int -nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr, - struct nouveau_bo *, uint32_t data, uint32_t flags, - uint32_t vor, uint32_t tor); - -struct nouveau_dma_priv { - uint32_t base; - uint32_t max; - uint32_t cur; - uint32_t put; - uint32_t free; - - int push_free; -} dma; - -struct nouveau_channel_priv { - struct nouveau_channel base; - - struct drm_nouveau_channel_alloc drm; - - uint32_t *pushbuf; - void *notifier_block; - - volatile uint32_t *user; - volatile uint32_t *put; - volatile uint32_t *get; - volatile uint32_t *ref_cnt; - - struct nouveau_dma_priv dma_master; - struct nouveau_dma_priv dma_bufmgr; - struct nouveau_dma_priv *dma; - - struct nouveau_fence *fence_head; - struct nouveau_fence *fence_tail; - uint32_t fence_sequence; - - struct nouveau_pushbuf_priv pb; - - unsigned user_charge; -}; -#define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) - -extern int -nouveau_channel_alloc(struct nouveau_device *, uint32_t fb, uint32_t tt, - struct nouveau_channel **); - -extern void -nouveau_channel_free(struct nouveau_channel **); - -struct nouveau_grobj_priv { - struct nouveau_grobj base; -}; -#define nouveau_grobj(n) ((struct nouveau_grobj_priv *)(n)) - -extern int nouveau_grobj_alloc(struct nouveau_channel *, uint32_t handle, - int class, struct nouveau_grobj **); -extern int nouveau_grobj_ref(struct nouveau_channel *, uint32_t handle, - struct nouveau_grobj **); -extern void nouveau_grobj_free(struct nouveau_grobj **); - - -struct nouveau_notifier_priv { - struct nouveau_notifier base; - - struct drm_nouveau_notifierobj_alloc drm; - volatile void *map; -}; -#define nouveau_notifier(n) ((struct nouveau_notifier_priv *)(n)) - -extern int -nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, int count, - struct nouveau_notifier **); - -extern void -nouveau_notifier_free(struct nouveau_notifier **); - -extern void -nouveau_notifier_reset(struct nouveau_notifier *, int id); - -extern uint32_t -nouveau_notifier_status(struct nouveau_notifier *, int id); - -extern uint32_t -nouveau_notifier_return_val(struct nouveau_notifier *, int id); - -extern int -nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, - int timeout); - -struct nouveau_bo_priv { - struct nouveau_bo base; - - struct nouveau_fence *fence; - struct nouveau_fence *wr_fence; - - struct drm_nouveau_mem_alloc drm; - void *map; - - void *sysmem; - int user; - - int refcount; - - uint64_t offset; - uint64_t flags; -}; -#define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) - -extern int -nouveau_bo_init(struct nouveau_device *); - -extern void -nouveau_bo_takedown(struct nouveau_device *); - -extern int -nouveau_bo_new(struct nouveau_device *, uint32_t flags, int align, int size, - struct nouveau_bo **); - -extern int -nouveau_bo_user(struct nouveau_device *, void *ptr, int size, - struct nouveau_bo **); - -extern int -nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); - -extern int -nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags); - -extern void -nouveau_bo_del(struct nouveau_bo **); - -extern int -nouveau_bo_map(struct nouveau_bo *, uint32_t flags); - -extern void -nouveau_bo_unmap(struct nouveau_bo *); - -extern int -nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, - struct nouveau_fence *fence, uint32_t flags); - -extern int -nouveau_resource_init(struct nouveau_resource **heap, unsigned start, - unsigned size); - -extern int -nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, - struct nouveau_resource **); - -extern void -nouveau_resource_free(struct nouveau_resource **); - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c deleted file mode 100644 index 7714e6f2485..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_fence.c +++ /dev/null @@ -1,215 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" -#include "nouveau_local.h" - -static void -nouveau_fence_del_unsignalled(struct nouveau_fence *fence) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); - struct nouveau_fence *le; - - if (nvchan->fence_head == fence) { - nvchan->fence_head = nouveau_fence(fence)->next; - if (nvchan->fence_head == NULL) - nvchan->fence_tail = NULL; - return; - } - - le = nvchan->fence_head; - while (le && nouveau_fence(le)->next != fence) - le = nouveau_fence(le)->next; - assert(le && nouveau_fence(le)->next == fence); - nouveau_fence(le)->next = nouveau_fence(fence)->next; - if (nvchan->fence_tail == fence) - nvchan->fence_tail = le; -} - -static void -nouveau_fence_del(struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!fence || !*fence) - return; - nvfence = nouveau_fence(*fence); - *fence = NULL; - - if (--nvfence->refcount) - return; - - if (nvfence->emitted && !nvfence->signalled) { - if (nvfence->signal_cb) { - nvfence->refcount++; - nouveau_fence_wait((void *)&nvfence); - return; - } - - nouveau_fence_del_unsignalled(&nvfence->base); - } - free(nvfence); -} - -int -nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!chan || !fence || *fence) - return -EINVAL; - - nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); - if (!nvfence) - return -ENOMEM; - nvfence->base.channel = chan; - nvfence->refcount = 1; - - *fence = &nvfence->base; - return 0; -} - -int -nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!fence) - return -EINVAL; - - if (*fence) { - nouveau_fence_del(fence); - *fence = NULL; - } - - if (ref) { - nvfence = nouveau_fence(ref); - nvfence->refcount++; - *fence = &nvfence->base; - } - - return 0; -} - -int -nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *), - void *priv) -{ - struct nouveau_fence_priv *nvfence = nouveau_fence(fence); - struct nouveau_fence_cb *cb; - - if (!nvfence || !func) - return -EINVAL; - - cb = malloc(sizeof(struct nouveau_fence_cb)); - if (!cb) - return -ENOMEM; - - cb->func = func; - cb->priv = priv; - cb->next = nvfence->signal_cb; - nvfence->signal_cb = cb; - return 0; -} - -void -nouveau_fence_emit(struct nouveau_fence *fence) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); - struct nouveau_fence_priv *nvfence = nouveau_fence(fence); - - nvfence->emitted = 1; - nvfence->sequence = ++nvchan->fence_sequence; - if (nvfence->sequence == 0xffffffff) - NOUVEAU_ERR("AII wrap unhandled\n"); - - /*XXX: assumes subc 0 is populated */ - RING_SPACE_CH(fence->channel, 2); - OUT_RING_CH (fence->channel, 0x00040050); - OUT_RING_CH (fence->channel, nvfence->sequence); - - if (nvchan->fence_tail) { - nouveau_fence(nvchan->fence_tail)->next = fence; - } else { - nvchan->fence_head = fence; - } - nvchan->fence_tail = fence; -} - -void -nouveau_fence_flush(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - uint32_t sequence = *nvchan->ref_cnt; - - while (nvchan->fence_head) { - struct nouveau_fence_priv *nvfence; - - nvfence = nouveau_fence(nvchan->fence_head); - if (nvfence->sequence > sequence) - break; - - nouveau_fence_del_unsignalled(&nvfence->base); - nvfence->signalled = 1; - - if (nvfence->signal_cb) { - struct nouveau_fence *fence = NULL; - - nouveau_fence_ref(nvchan->fence_head, &fence); - - while (nvfence->signal_cb) { - struct nouveau_fence_cb *cb; - - cb = nvfence->signal_cb; - nvfence->signal_cb = cb->next; - cb->func(cb->priv); - free(cb); - } - - nouveau_fence_ref(NULL, &fence); - } - } -} - -int -nouveau_fence_wait(struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!fence || !*fence) - return -EINVAL; - nvfence = nouveau_fence(*fence); - - if (nvfence->emitted) { - while (!nvfence->signalled) - nouveau_fence_flush(nvfence->base.channel); - } - nouveau_fence_ref(NULL, fence); - - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c deleted file mode 100644 index 55dfeb99aa7..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_grobj.c +++ /dev/null @@ -1,107 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include - -#include "nouveau_drmif.h" - -int -nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle, - int class, struct nouveau_grobj **grobj) -{ - struct nouveau_device_priv *nvdev = nouveau_device(chan->device); - struct nouveau_grobj_priv *nvgrobj; - struct drm_nouveau_grobj_alloc g; - int ret; - - if (!nvdev || !grobj || *grobj) - return -EINVAL; - - nvgrobj = calloc(1, sizeof(*nvgrobj)); - if (!nvgrobj) - return -ENOMEM; - nvgrobj->base.channel = chan; - nvgrobj->base.handle = handle; - nvgrobj->base.grclass = class; - - g.channel = chan->id; - g.handle = handle; - g.class = class; - ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC, - &g, sizeof(g)); - if (ret) { - nouveau_grobj_free((void *)&grobj); - return ret; - } - - *grobj = &nvgrobj->base; - return 0; -} - -int -nouveau_grobj_ref(struct nouveau_channel *chan, uint32_t handle, - struct nouveau_grobj **grobj) -{ - struct nouveau_grobj_priv *nvgrobj; - - if (!chan || !grobj || *grobj) - return -EINVAL; - - nvgrobj = calloc(1, sizeof(struct nouveau_grobj_priv)); - if (!nvgrobj) - return -ENOMEM; - nvgrobj->base.channel = chan; - nvgrobj->base.handle = handle; - nvgrobj->base.grclass = 0; - - *grobj = &nvgrobj->base; - return 0; -} - -void -nouveau_grobj_free(struct nouveau_grobj **grobj) -{ - struct nouveau_device_priv *nvdev; - struct nouveau_channel_priv *chan; - struct nouveau_grobj_priv *nvgrobj; - - if (!grobj || !*grobj) - return; - nvgrobj = nouveau_grobj(*grobj); - *grobj = NULL; - - - chan = nouveau_channel(nvgrobj->base.channel); - nvdev = nouveau_device(chan->base.device); - - if (nvgrobj->base.grclass) { - struct drm_nouveau_gpuobj_free f; - - f.channel = chan->drm.channel; - f.handle = nvgrobj->base.handle; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, - &f, sizeof(f)); - } - free(nvgrobj); -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h deleted file mode 100644 index 59febca2929..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_local.h +++ /dev/null @@ -1,89 +0,0 @@ -#ifndef __NOUVEAU_LOCAL_H__ -#define __NOUVEAU_LOCAL_H__ - -#include - -/* Debug output */ -#define NOUVEAU_MSG(fmt, args...) do { \ - fprintf(stdout, "nouveau: "fmt, ##args); \ - fflush(stdout); \ -} while(0) - -#define NOUVEAU_ERR(fmt, args...) do { \ - fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); \ - fflush(stderr); \ -} while(0) - -#define NOUVEAU_TIME_MSEC() 0 - -/* User FIFO control */ -//#define NOUVEAU_DMA_TRACE -//#define NOUVEAU_DMA_DEBUG -//#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF -#define NOUVEAU_DMA_BARRIER -#define NOUVEAU_DMA_TIMEOUT 2000 - -/* Push buffer access macros */ -#define OUT_RING(data) do { \ - (*nv->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - memcpy(nv->channel->pushbuf->cur, (src), (size)<<2); \ - nv->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define FIRE_RING() do { \ - nouveau_pushbuf_flush(nv->channel, 0); \ -} while(0) - -#define BEGIN_RING_GR(obj,mthd,size) do { \ - if (nv->channel->pushbuf->remaining < ((size) + 1)) \ - nouveau_pushbuf_flush(nv->channel, ((size) + 1)); \ - OUT_RING(((obj)->subc << 13) | ((size) << 18) | (mthd)); \ - nv->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - BEGIN_RING_GR(nv->obj, (mthd), (size)); \ -} while(0) - -#define BIND_RING(o,s) do { \ - nv->o->subc = (s); \ - BEGIN_RING(o, 0x0000, 1); \ - OUT_RING (nv->o->handle); \ -} while(0) - -#define OUT_RELOC(buf,data,flags,vor,tor) do { \ - nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur++, \ - buf, (data), (flags), (vor), (tor)); \ -} while(0) - -/* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) - -/* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - nv->channel->vram->handle, nv->channel->gart->handle); \ -} while(0) - -/* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) - -/* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c deleted file mode 100644 index 9adb9ac8547..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_lock.c +++ /dev/null @@ -1,94 +0,0 @@ -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "main/glheader.h" -#include "glapi/glthread.h" -#include - -#include "nouveau_context.h" -#include "nouveau_screen.h" - -_glthread_DECLARE_STATIC_MUTEX( lockMutex ); - -static void -nouveau_contended_lock(struct nouveau_context *nv, GLuint flags) -{ - __DRIdrawablePrivate *dPriv = nv->dri_drawable; - __DRIscreenPrivate *sPriv = nv->dri_screen; - struct nouveau_screen *nv_screen = nv->nv_screen; - struct nouveau_device *dev = nv_screen->device; - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - drmGetLock(nvdev->fd, nvdev->ctx, flags); - - /* If the window moved, may need to set a new cliprect now. - * - * NOTE: This releases and regains the hw lock, so all state - * checking must be done *after* this call: - */ - if (dPriv) - DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv); -} - -/* Lock the hardware and validate our state. - */ -void -LOCK_HARDWARE(struct nouveau_context *nv) -{ - struct nouveau_screen *nv_screen = nv->nv_screen; - struct nouveau_device *dev = nv_screen->device; - struct nouveau_device_priv *nvdev = nouveau_device(dev); - char __ret=0; - - _glthread_LOCK_MUTEX(lockMutex); - assert(!nv->locked); - - DRM_CAS(nvdev->lock, nvdev->ctx, - (DRM_LOCK_HELD | nvdev->ctx), __ret); - - if (__ret) - nouveau_contended_lock(nv, 0); - nv->locked = GL_TRUE; -} - - - /* Unlock the hardware using the global current context - */ -void -UNLOCK_HARDWARE(struct nouveau_context *nv) -{ - struct nouveau_screen *nv_screen = nv->nv_screen; - struct nouveau_device *dev = nv_screen->device; - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - assert(nv->locked); - nv->locked = GL_FALSE; - - DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx); - - _glthread_UNLOCK_MUTEX(lockMutex); -} diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c deleted file mode 100644 index 01e8f38440e..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_notifier.c +++ /dev/null @@ -1,137 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_local.h" - -#define NOTIFIER(__v) \ - struct nouveau_notifier_priv *nvnotify = nouveau_notifier(notifier); \ - volatile uint32_t *__v = (void*)nvnotify->map + (id * 32) - -int -nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, - int count, struct nouveau_notifier **notifier) -{ - struct nouveau_notifier_priv *nvnotify; - int ret; - - if (!chan || !notifier || *notifier) - return -EINVAL; - - nvnotify = calloc(1, sizeof(struct nouveau_notifier_priv)); - if (!nvnotify) - return -ENOMEM; - nvnotify->base.channel = chan; - nvnotify->base.handle = handle; - - nvnotify->drm.channel = chan->id; - nvnotify->drm.handle = handle; - nvnotify->drm.count = count; - if ((ret = drmCommandWriteRead(nouveau_device(chan->device)->fd, - DRM_NOUVEAU_NOTIFIEROBJ_ALLOC, - &nvnotify->drm, - sizeof(nvnotify->drm)))) { - nouveau_notifier_free((void *)&nvnotify); - return ret; - } - - nvnotify->map = (void *)nouveau_channel(chan)->notifier_block + - nvnotify->drm.offset; - *notifier = &nvnotify->base; - return 0; -} - -void -nouveau_notifier_free(struct nouveau_notifier **notifier) -{ - - struct nouveau_notifier_priv *nvnotify; - struct nouveau_channel_priv *nvchan; - struct nouveau_device_priv *nvdev; - struct drm_nouveau_gpuobj_free f; - - if (!notifier || !*notifier) - return; - nvnotify = nouveau_notifier(*notifier); - *notifier = NULL; - - nvchan = nouveau_channel(nvnotify->base.channel); - nvdev = nouveau_device(nvchan->base.device); - - f.channel = nvchan->drm.channel; - f.handle = nvnotify->base.handle; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); - free(nvnotify); -} - -void -nouveau_notifier_reset(struct nouveau_notifier *notifier, int id) -{ - NOTIFIER(n); - - n[NV_NOTIFY_TIME_0 /4] = 0x00000000; - n[NV_NOTIFY_TIME_1 /4] = 0x00000000; - n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000; - n[NV_NOTIFY_STATE /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS << - NV_NOTIFY_STATE_STATUS_SHIFT); -} - -uint32_t -nouveau_notifier_status(struct nouveau_notifier *notifier, int id) -{ - NOTIFIER(n); - - return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; -} - -uint32_t -nouveau_notifier_return_val(struct nouveau_notifier *notifier, int id) -{ - NOTIFIER(n); - - return n[NV_NOTIFY_RETURN_VALUE/4]; -} - -int -nouveau_notifier_wait_status(struct nouveau_notifier *notifier, int id, - int status, int timeout) -{ - NOTIFIER(n); - uint32_t time = 0, t_start = NOUVEAU_TIME_MSEC(); - - while (time <= timeout) { - uint32_t v; - - v = n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; - if (v == status) - return 0; - - if (timeout) - time = NOUVEAU_TIME_MSEC() - t_start; - } - - return -EBUSY; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c deleted file mode 100644 index 7d5eddb92ff..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_pushbuf.c +++ /dev/null @@ -1,261 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" - -#define PB_BUFMGR_DWORDS (4096 / 2) -#define PB_MIN_USER_DWORDS 2048 - -static int -nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; - - assert((min + 1) <= nvchan->dma->max); - - /* Wait for enough space in push buffer */ - min = min < PB_MIN_USER_DWORDS ? PB_MIN_USER_DWORDS : min; - min += 1; /* a bit extra for the NOP */ - if (nvchan->dma->free < min) - WAIT_RING_CH(chan, min); - - /* Insert NOP, may turn into a jump later */ - RING_SPACE_CH(chan, 1); - nvpb->nop_jump = nvchan->dma->cur; - OUT_RING_CH(chan, 0); - - /* Any remaining space is available to the user */ - nvpb->start = nvchan->dma->cur; - nvpb->size = nvchan->dma->free; - nvpb->base.channel = chan; - nvpb->base.remaining = nvpb->size; - nvpb->base.cur = &nvchan->pushbuf[nvpb->start]; - - return 0; -} - -int -nouveau_pushbuf_init(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *m = &nvchan->dma_master; - struct nouveau_dma_priv *b = &nvchan->dma_bufmgr; - int i; - - if (!nvchan) - return -EINVAL; - - /* Reassign last bit of push buffer for a "separate" bufmgr - * ring buffer - */ - m->max -= PB_BUFMGR_DWORDS; - m->free -= PB_BUFMGR_DWORDS; - - b->base = m->base + ((m->max + 2) << 2); - b->max = PB_BUFMGR_DWORDS - 2; - b->cur = b->put = 0; - b->free = b->max - b->cur; - - /* Some NOPs just to be safe - *XXX: RING_SKIPS - */ - nvchan->dma = b; - RING_SPACE_CH(chan, 8); - for (i = 0; i < 8; i++) - OUT_RING_CH(chan, 0); - nvchan->dma = m; - - nouveau_pushbuf_space(chan, 0); - chan->pushbuf = &nvchan->pb.base; - - return 0; -} - -static uint32_t -nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo, - struct nouveau_pushbuf_reloc *r) -{ - uint32_t push; - - if (r->flags & NOUVEAU_BO_LOW) { - push = bo->offset + r->data; - } else - if (r->flags & NOUVEAU_BO_HIGH) { - push = (bo->offset + r->data) >> 32; - } else { - push = r->data; - } - - if (r->flags & NOUVEAU_BO_OR) { - if (bo->flags & NOUVEAU_BO_VRAM) - push |= r->vor; - else - push |= r->tor; - } - - return push; -} - -/* This would be our TTM "superioctl" */ -int -nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; - struct nouveau_pushbuf_bo *pbbo; - struct nouveau_fence *fence = NULL; - int ret; - - if (nvpb->base.remaining == nvpb->size) - return 0; - - nvpb->size -= nvpb->base.remaining; - nvchan->dma->cur += nvpb->size; - nvchan->dma->free -= nvpb->size; - assert(nvchan->dma->cur <= nvchan->dma->max); - - ret = nouveau_fence_new(chan, &fence); - if (ret) - return ret; - - nvchan->dma = &nvchan->dma_bufmgr; - nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 | - (nvchan->dma->base + (nvchan->dma->cur << 2)); - - /* Validate buffers + apply relocations */ - nvchan->user_charge = 0; - while ((pbbo = ptr_to_pbbo(nvpb->buffers))) { - struct nouveau_pushbuf_reloc *r; - struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; - - ret = nouveau_bo_validate(chan, bo, fence, pbbo->flags); - assert (ret == 0); - - if (bo->offset == nouveau_bo(bo)->offset && - bo->flags == nouveau_bo(bo)->flags) { - while ((r = ptr_to_pbrel(pbbo->relocs))) { - pbbo->relocs = r->next; - free(r); - } - - nvpb->buffers = pbbo->next; - free(pbbo); - continue; - } - bo->offset = nouveau_bo(bo)->offset; - bo->flags = nouveau_bo(bo)->flags; - - while ((r = ptr_to_pbrel(pbbo->relocs))) { - *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); - pbbo->relocs = r->next; - free(r); - } - - nvpb->buffers = pbbo->next; - free(pbbo); - } - nvpb->nr_buffers = 0; - - /* Switch back to user's ring */ - RING_SPACE_CH(chan, 1); - OUT_RING_CH(chan, 0x20000000 | ((nvpb->start << 2) + - nvchan->dma_master.base)); - nvchan->dma = &nvchan->dma_master; - - /* Fence + kickoff */ - nouveau_fence_emit(fence); - FIRE_RING_CH(chan); - nouveau_fence_ref(NULL, &fence); - - /* Allocate space for next push buffer */ - assert(!nouveau_pushbuf_space(chan, min)); - - return 0; -} - -static struct nouveau_pushbuf_bo * -nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) -{ - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); - struct nouveau_pushbuf_bo *pbbo = ptr_to_pbbo(nvpb->buffers); - - while (pbbo) { - if (pbbo->handle == bo->handle) - return pbbo; - pbbo = ptr_to_pbbo(pbbo->next); - } - - pbbo = malloc(sizeof(struct nouveau_pushbuf_bo)); - pbbo->next = nvpb->buffers; - nvpb->buffers = pbbo_to_ptr(pbbo); - nvpb->nr_buffers++; - - pbbo->handle = bo_to_ptr(bo); - pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; - pbbo->relocs = 0; - pbbo->nr_relocs = 0; - return pbbo; -} - -int -nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, - struct nouveau_bo *bo, uint32_t data, uint32_t flags, - uint32_t vor, uint32_t tor) -{ - struct nouveau_pushbuf_bo *pbbo; - struct nouveau_pushbuf_reloc *r; - - if (!chan) - return -EINVAL; - - pbbo = nouveau_pushbuf_emit_buffer(chan, bo); - if (!pbbo) - return -EFAULT; - - r = malloc(sizeof(struct nouveau_pushbuf_reloc)); - r->next = pbbo->relocs; - pbbo->relocs = pbrel_to_ptr(r); - pbbo->nr_relocs++; - - pbbo->flags |= (flags & NOUVEAU_BO_RDWR); - pbbo->flags &= (flags | NOUVEAU_BO_RDWR); - - r->handle = bo_to_ptr(r); - r->ptr = ptr; - r->flags = flags; - r->data = data; - r->vor = vor; - r->tor = tor; - - if (flags & NOUVEAU_BO_DUMMY) - *(uint32_t *)ptr = 0; - else - *(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r); - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c deleted file mode 100644 index 5d9d578b4fe..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_resource.c +++ /dev/null @@ -1,111 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_local.h" - -int -nouveau_resource_init(struct nouveau_resource **heap, - unsigned start, unsigned size) -{ - struct nouveau_resource *r; - - r = calloc(1, sizeof(struct nouveau_resource)); - if (!r) - return 1; - - r->start = start; - r->size = size; - *heap = r; - return 0; -} - -int -nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, - struct nouveau_resource **res) -{ - struct nouveau_resource *r; - - if (!heap || !size || !res || *res) - return 1; - - while (heap) { - if (!heap->in_use && heap->size >= size) { - r = calloc(1, sizeof(struct nouveau_resource)); - if (!r) - return 1; - - r->start = (heap->start + heap->size) - size; - r->size = size; - r->in_use = 1; - r->priv = priv; - - heap->size -= size; - - r->next = heap->next; - if (heap->next) - heap->next->prev = r; - r->prev = heap; - heap->next = r; - - *res = r; - return 0; - } - - heap = heap->next; - } - - return 1; -} - -void -nouveau_resource_free(struct nouveau_resource **res) -{ - struct nouveau_resource *r; - - if (!res || !*res) - return; - r = *res; - - if (r->prev && !r->prev->in_use) { - r->prev->next = r->next; - if (r->next) - r->next->prev = r->prev; - r->prev->size += r->size; - free(r); - } else - if (r->next && !r->next->in_use) { - r->next->prev = r->prev; - if (r->prev) - r->prev->next = r->next; - r->next->size += r->size; - r->next->start = r->start; - free(r); - } else { - r->in_use = 0; - } - - *res = NULL; -} diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c deleted file mode 100644 index f06e1784831..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.c +++ /dev/null @@ -1,308 +0,0 @@ -#include "utils.h" -#include "vblank.h" -#include "xmlpool.h" - -#include "pipe/p_context.h" -#include "state_tracker/st_public.h" -#include "state_tracker/st_cb_fbo.h" - -#include "nouveau_context.h" -#include "nouveau_device.h" -#include "nouveau_drm.h" -#include "nouveau_dri.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_swapbuffers.h" - -#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10 -#error nouveau_drm.h version does not match expected version -#endif - -/* Extension stuff, enabling of extensions handled by Gallium's GL state - * tracker. But, we still need to define the entry points we want. - */ -#define need_GL_ARB_fragment_program -#define need_GL_ARB_multisample -#define need_GL_ARB_occlusion_query -#define need_GL_ARB_point_parameters -#define need_GL_ARB_shader_objects -#define need_GL_ARB_texture_compression -#define need_GL_ARB_vertex_program -#define need_GL_ARB_vertex_shader -#define need_GL_ARB_vertex_buffer_object -#define need_GL_EXT_compiled_vertex_array -#define need_GL_EXT_fog_coord -#define need_GL_EXT_secondary_color -#define need_GL_EXT_framebuffer_object -#define need_GL_VERSION_2_0 -#define need_GL_VERSION_2_1 -#include "extension_helper.h" - -const struct dri_extension card_extensions[] = -{ - { "GL_ARB_multisample", GL_ARB_multisample_functions }, - { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, - { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, - { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, - { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, - { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, - { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, - { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, - { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, - { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, - { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, - { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, - { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, - { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, - { NULL, 0 } -}; - -PUBLIC const char __driConfigOptions[] = -DRI_CONF_BEGIN -DRI_CONF_END; -static const GLuint __driNConfigOptions = 0; - -extern const struct dri_extension common_extensions[]; -extern const struct dri_extension nv40_extensions[]; - -static GLboolean -nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) -{ - struct nouveau_dri *nv_dri = driScrnPriv->pDevPriv; - struct nouveau_screen *nv_screen; - int ret; - - if (driScrnPriv->devPrivSize != sizeof(struct nouveau_dri)) { - NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); - return GL_FALSE; - } - - nv_screen = CALLOC_STRUCT(nouveau_screen); - if (!nv_screen) - return GL_FALSE; - nv_screen->driScrnPriv = driScrnPriv; - driScrnPriv->private = (void *)nv_screen; - - driParseOptionInfo(&nv_screen->option_cache, - __driConfigOptions, __driNConfigOptions); - - if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, - driScrnPriv->fd, 0))) { - NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); - return GL_FALSE; - } - - nv_screen->front_offset = nv_dri->front_offset; - nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); - nv_screen->front_cpp = nv_dri->bpp / 8; - nv_screen->front_height = nv_dri->height; - - return GL_TRUE; -} - -static void -nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) -{ - struct nouveau_screen *nv_screen = driScrnPriv->private; - - driScrnPriv->private = NULL; - FREE(nv_screen); -} - -static GLboolean -nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, - __DRIdrawablePrivate * driDrawPriv, - const __GLcontextModes *glVis, GLboolean pixmapBuffer) -{ - struct nouveau_framebuffer *nvfb; - enum pipe_format colour, depth, stencil; - - if (pixmapBuffer) - return GL_FALSE; - - nvfb = CALLOC_STRUCT(nouveau_framebuffer); - if (!nvfb) - return GL_FALSE; - - if (glVis->redBits == 5) - colour = PIPE_FORMAT_R5G6B5_UNORM; - else - colour = PIPE_FORMAT_A8R8G8B8_UNORM; - - if (glVis->depthBits == 16) - depth = PIPE_FORMAT_Z16_UNORM; - else if (glVis->depthBits == 24) - depth = PIPE_FORMAT_Z24S8_UNORM; - else - depth = PIPE_FORMAT_NONE; - - if (glVis->stencilBits == 8) - stencil = PIPE_FORMAT_Z24S8_UNORM; - else - stencil = PIPE_FORMAT_NONE; - - nvfb->stfb = st_create_framebuffer(glVis, colour, depth, stencil, - driDrawPriv->w, driDrawPriv->h, - (void*)nvfb); - if (!nvfb->stfb) { - free(nvfb); - return GL_FALSE; - } - - driDrawPriv->driverPrivate = (void *)nvfb; - return GL_TRUE; -} - -static void -nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) -{ - struct nouveau_framebuffer *nvfb; - - nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate; - st_unreference_framebuffer(&nvfb->stfb); - free(nvfb); -} - -static struct __DriverAPIRec -nouveau_api = { - .InitDriver = nouveau_screen_create, - .DestroyScreen = nouveau_screen_destroy, - .CreateContext = nouveau_context_create, - .DestroyContext = nouveau_context_destroy, - .CreateBuffer = nouveau_create_buffer, - .DestroyBuffer = nouveau_destroy_buffer, - .SwapBuffers = nouveau_swap_buffers, - .MakeCurrent = nouveau_context_bind, - .UnbindContext = nouveau_context_unbind, - .GetSwapInfo = NULL, - .GetMSC = NULL, - .WaitForMSC = NULL, - .WaitForSBC = NULL, - .SwapBuffersMSC = NULL, - .CopySubBuffer = nouveau_copy_sub_buffer, - .setTexOffset = NULL -}; - -static __GLcontextModes * -nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, - unsigned stencil_bits, GLboolean have_back_buffer) -{ - __GLcontextModes * modes; - __GLcontextModes * m; - unsigned num_modes; - unsigned depth_buffer_factor; - unsigned back_buffer_factor; - int i; - - static const struct { - GLenum format; - GLenum type; - } fb_format_array[] = { - { GL_RGB , GL_UNSIGNED_SHORT_5_6_5 }, - { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV }, - { GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV }, - }; - - /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't - * support pageflipping at all. - */ - static const GLenum back_buffer_modes[] = { - GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML - }; - - u_int8_t depth_bits_array[4] = { 0, 16, 24, 24 }; - u_int8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; - - depth_buffer_factor = 4; - back_buffer_factor = (have_back_buffer) ? 3 : 1; - - num_modes = ((pixel_bits==16) ? 1 : 2) * - depth_buffer_factor * back_buffer_factor * 4; - modes = (*dri_interface->createContextModes)(num_modes, - sizeof(__GLcontextModes)); - m = modes; - - for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) { - if (!driFillInModes(&m, fb_format_array[i].format, - fb_format_array[i].type, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - GLX_TRUE_COLOR)) { - fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", - __func__, __LINE__ ); - return NULL; - } - - if (!driFillInModes(&m, fb_format_array[i].format, - fb_format_array[i].type, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - GLX_DIRECT_COLOR)) { - fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", - __func__, __LINE__ ); - return NULL; - } - } - - return modes; -} -PUBLIC void * -__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, - __DRIscreen *psc, const __GLcontextModes * modes, - const __DRIversion * ddx_version, - const __DRIversion * dri_version, - const __DRIversion * drm_version, - const __DRIframebuffer * frame_buffer, - void * pSAREA, int fd, int internal_api_version, - const __DRIinterfaceMethods * interface, - __GLcontextModes ** driver_modes) -{ - __DRIscreenPrivate *psp; - static const __DRIversion ddx_expected = - { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; - static const __DRIversion dri_expected = { 4, 0, 0 }; - static const __DRIversion drm_expected = - { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; - struct nouveau_dri *nv_dri = NULL; - - dri_interface = interface; - - if (!driCheckDriDdxDrmVersions2("nouveau", - dri_version, &dri_expected, - ddx_version, &ddx_expected, - drm_version, &drm_expected)) { - return NULL; - } - - if (drm_expected.patch != drm_version->patch) { - fprintf(stderr, "Incompatible DRM patch level.\n" - "Expected: %d\n" "Current : %d\n", - drm_expected.patch, drm_version->patch); - return NULL; - } - - psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, - ddx_version, dri_version, drm_version, - frame_buffer, pSAREA, fd, - internal_api_version, - &nouveau_api); - if (psp == NULL) - return NULL; - nv_dri = psp->pDevPriv; - - *driver_modes = nouveau_fill_in_modes(nv_dri->bpp, - (nv_dri->bpp == 16) ? 16 : 24, - (nv_dri->bpp == 16) ? 0 : 8, - 1); - - driInitExtensions(NULL, card_extensions, GL_FALSE); - - return (void *)psp; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h deleted file mode 100644 index 019823bd44d..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_screen.h +++ /dev/null @@ -1,19 +0,0 @@ -#ifndef __NOUVEAU_SCREEN_H__ -#define __NOUVEAU_SCREEN_H__ - -#include "xmlconfig.h" -#include "nouveau_device.h" - -struct nouveau_screen { - __DRIscreenPrivate *driScrnPriv; - driOptionCache option_cache; - - struct nouveau_device *device; - - uint32_t front_offset; - uint32_t front_pitch; - uint32_t front_cpp; - uint32_t front_height; -}; - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c deleted file mode 100644 index 91bf243f424..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.c +++ /dev/null @@ -1,86 +0,0 @@ -#include "main/glheader.h" -#include "glapi/glthread.h" -#include - -#include "pipe/p_context.h" -#include "state_tracker/st_public.h" -#include "state_tracker/st_context.h" -#include "state_tracker/st_cb_fbo.h" - -#include "nouveau_context.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_swapbuffers.h" - -void -nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, - const drm_clip_rect_t *rect) -{ - struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; - drm_clip_rect_t *pbox; - int nbox, i; - - LOCK_HARDWARE(nv); - if (!dPriv->numClipRects) { - UNLOCK_HARDWARE(nv); - return; - } - pbox = dPriv->pClipRects; - nbox = dPriv->numClipRects; - - nv->surface_copy_prep(nv, nv->frontbuffer, surf); - for (i = 0; i < nbox; i++, pbox++) { - int sx, sy, dx, dy, w, h; - - sx = pbox->x1 - dPriv->x; - sy = pbox->y1 - dPriv->y; - dx = pbox->x1; - dy = pbox->y1; - w = pbox->x2 - pbox->x1; - h = pbox->y2 - pbox->y1; - - nv->surface_copy(nv, dx, dy, sx, sy, w, h); - } - - FIRE_RING(); - UNLOCK_HARDWARE(nv); - - if (nv->last_stamp != dPriv->lastStamp) { - struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; - st_resize_framebuffer(nvfb->stfb, dPriv->w, dPriv->h); - nv->last_stamp = dPriv->lastStamp; - } -} - -void -nouveau_copy_sub_buffer(__DRIdrawablePrivate *dPriv, int x, int y, int w, int h) -{ - struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; - struct pipe_surface *surf; - - surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); - if (surf) { - drm_clip_rect_t rect; - rect.x1 = x; - rect.y1 = y; - rect.x2 = x + w; - rect.y2 = y + h; - - st_notify_swapbuffers(nvfb->stfb); - nouveau_copy_buffer(dPriv, surf, &rect); - } -} - -void -nouveau_swap_buffers(__DRIdrawablePrivate *dPriv) -{ - struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; - struct pipe_surface *surf; - - surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); - if (surf) { - st_notify_swapbuffers(nvfb->stfb); - nouveau_copy_buffer(dPriv, surf, NULL); - } -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h deleted file mode 100644 index 825d3da6da5..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_swapbuffers.h +++ /dev/null @@ -1,10 +0,0 @@ -#ifndef __NOUVEAU_SWAPBUFFERS_H__ -#define __NOUVEAU_SWAPBUFFERS_H__ - -extern void nouveau_copy_buffer(__DRIdrawablePrivate *, struct pipe_surface *, - const drm_clip_rect_t *); -extern void nouveau_copy_sub_buffer(__DRIdrawablePrivate *, - int x, int y, int w, int h); -extern void nouveau_swap_buffers(__DRIdrawablePrivate *); - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c deleted file mode 100644 index 1494bd48dd4..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys.c +++ /dev/null @@ -1,124 +0,0 @@ -#include "pipe/p_util.h" - -#include "nouveau_context.h" -#include "nouveau_winsys_pipe.h" - -#include "pipe/nouveau/nouveau_winsys.h" - -static int -nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, - struct nouveau_notifier **notify) -{ - struct nouveau_context *nv = nvws->nv; - - return nouveau_notifier_alloc(nv->channel, nv->next_handle++, - count, notify); -} - -static int -nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, - struct nouveau_grobj **grobj) -{ - struct nouveau_context *nv = nvws->nv; - int ret; - - ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, - grclass, grobj); - if (ret) - return ret; - - (*grobj)->subc = nv->next_subchannel++; - assert((*grobj)->subc <= 7); - BEGIN_RING_GR(*grobj, 0x0000, 1); - OUT_RING ((*grobj)->handle); - return 0; -} - -static int -nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst, - unsigned dx, unsigned dy, struct pipe_surface *src, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct nouveau_context *nv = nvws->nv; - - if (nv->surface_copy_prep(nv, dst, src)) - return 1; - nv->surface_copy(nv, dx, dy, sx, sy, w, h); - nv->surface_copy_done(nv); - - return 0; -} - -static int -nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - if (nvws->nv->surface_fill(nvws->nv, dst, dx, dy, w, h, value)) - return 1; - return 0; -} - -int -nouveau_pipe_emit_reloc(struct nouveau_channel *chan, void *ptr, - struct pipe_buffer *buf, uint32_t data, - uint32_t flags, uint32_t vor, uint32_t tor) -{ - return nouveau_pushbuf_emit_reloc(chan, ptr, nouveau_buffer(buf)->bo, - data, flags, vor, tor); -} - -struct pipe_context * -nouveau_pipe_create(struct nouveau_context *nv) -{ - struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); - struct pipe_context *(*hw_create)(struct pipe_winsys *, - struct nouveau_winsys *, - unsigned); - - if (!nvws) - return NULL; - - switch (nv->chipset & 0xf0) { - case 0x30: - hw_create = nv30_create; - break; - case 0x40: - case 0x60: - hw_create = nv40_create; - break; - case 0x50: - case 0x80: - hw_create = nv50_create; - break; - default: - NOUVEAU_ERR("Unknown chipset NV%02x\n", (int)nv->chipset); - return NULL; - } - - nvws->nv = nv; - nvws->channel = nv->channel; - - nvws->res_init = nouveau_resource_init; - nvws->res_alloc = nouveau_resource_alloc; - nvws->res_free = nouveau_resource_free; - - nvws->push_reloc = nouveau_pipe_emit_reloc; - nvws->push_flush = nouveau_pushbuf_flush; - - nvws->grobj_alloc = nouveau_pipe_grobj_alloc; - nvws->grobj_free = nouveau_grobj_free; - - nvws->notifier_alloc = nouveau_pipe_notifier_alloc; - nvws->notifier_free = nouveau_notifier_free; - nvws->notifier_reset = nouveau_notifier_reset; - nvws->notifier_status = nouveau_notifier_status; - nvws->notifier_retval = nouveau_notifier_return_val; - nvws->notifier_wait = nouveau_notifier_wait_status; - - nvws->surface_copy = nouveau_pipe_surface_copy; - nvws->surface_fill = nouveau_pipe_surface_fill; - - return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c deleted file mode 100644 index e1a9271395b..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.c +++ /dev/null @@ -1,196 +0,0 @@ -#include "pipe/p_winsys.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_inlines.h" - -#include "nouveau_context.h" -#include "nouveau_device.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_swapbuffers.h" -#include "nouveau_winsys_pipe.h" - -static void -nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, - void *context_private) -{ - struct nouveau_context *nv = context_private; - __DRIdrawablePrivate *dPriv = nv->dri_drawable; - - nouveau_copy_buffer(dPriv, surf, NULL); -} - -static void -nouveau_printf(struct pipe_winsys *pws, const char *fmt, ...) -{ - va_list args; - va_start(args, fmt); - vfprintf(stderr, fmt, args); - va_end(args); -} - -static const char * -nouveau_get_name(struct pipe_winsys *pws) -{ - return "Nouveau/DRI"; -} - -static struct pipe_surface * -nouveau_surface_alloc(struct pipe_winsys *ws) -{ - struct pipe_surface *surf; - - surf = CALLOC_STRUCT(pipe_surface); - if (!surf) - return NULL; - - surf->refcount = 1; - surf->winsys = ws; - return surf; -} - -static int -nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, - unsigned width, unsigned height, - enum pipe_format format, unsigned flags) -{ - unsigned pitch = ((width * pf_get_size(format)) + 63) & ~63; - - surf->format = format; - surf->width = width; - surf->height = height; - surf->cpp = pf_get_size(format); - surf->pitch = pitch / surf->cpp; - - surf->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - pitch * height); - if (!surf->buffer) - return 1; - - return 0; -} - -static void -nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) -{ - struct pipe_surface *surf = *s; - - *s = NULL; - if (--surf->refcount <= 0) { - if (surf->buffer) - pipe_buffer_reference(ws, &surf->buffer, NULL); - free(surf); - } -} - -static struct pipe_buffer * -nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, - unsigned usage, unsigned size) -{ - struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; - struct nouveau_device *dev = nvpws->nv->nv_screen->device; - struct nouveau_pipe_buffer *nvbuf; - uint32_t flags = 0; - - nvbuf = calloc(1, sizeof(*nvbuf)); - if (!nvbuf) - return NULL; - nvbuf->base.refcount = 1; - nvbuf->base.alignment = alignment; - nvbuf->base.usage = usage; - nvbuf->base.size = size; - - flags = NOUVEAU_BO_LOCAL; - if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { - free(nvbuf); - return NULL; - } - - return &nvbuf->base; -} - -static struct pipe_buffer * -nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) -{ - struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; - struct nouveau_device *dev = nvpws->nv->nv_screen->device; - struct nouveau_pipe_buffer *nvbuf; - - nvbuf = calloc(1, sizeof(*nvbuf)); - if (!nvbuf) - return NULL; - nvbuf->base.refcount = 1; - nvbuf->base.size = bytes; - - if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) { - free(nvbuf); - return NULL; - } - - return &nvbuf->base; -} - -static void -nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) -{ - struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - - nouveau_bo_del(&nvbuf->bo); - free(nvbuf); -} - -static void * -nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf, - unsigned flags) -{ - struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - uint32_t map_flags = 0; - - if (flags & PIPE_BUFFER_USAGE_CPU_READ) - map_flags |= NOUVEAU_BO_RD; - if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) - map_flags |= NOUVEAU_BO_WR; - - if (nouveau_bo_map(nvbuf->bo, map_flags)) - return NULL; - return nvbuf->bo->map; -} - -static void -nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) -{ - struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - - nouveau_bo_unmap(nvbuf->bo); -} - -struct pipe_winsys * -nouveau_create_pipe_winsys(struct nouveau_context *nv) -{ - struct nouveau_pipe_winsys *nvpws; - struct pipe_winsys *pws; - - nvpws = CALLOC_STRUCT(nouveau_pipe_winsys); - if (!nvpws) - return NULL; - nvpws->nv = nv; - pws = &nvpws->pws; - - pws->flush_frontbuffer = nouveau_flush_frontbuffer; - pws->printf = nouveau_printf; - - pws->surface_alloc = nouveau_surface_alloc; - pws->surface_alloc_storage = nouveau_surface_alloc_storage; - pws->surface_release = nouveau_surface_release; - - pws->buffer_create = nouveau_pipe_bo_create; - pws->buffer_destroy = nouveau_pipe_bo_del; - pws->user_buffer_create = nouveau_pipe_bo_user_create; - pws->buffer_map = nouveau_pipe_bo_map; - pws->buffer_unmap = nouveau_pipe_bo_unmap; - - pws->get_name = nouveau_get_name; - - return &nvpws->pws; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h deleted file mode 100644 index 6a03ac0d773..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_pipe.h +++ /dev/null @@ -1,34 +0,0 @@ -#ifndef NOUVEAU_PIPE_WINSYS_H -#define NOUVEAU_PIPE_WINSYS_H - -#include "pipe/p_context.h" -#include "pipe/p_winsys.h" -#include "nouveau_context.h" - -struct nouveau_pipe_buffer { - struct pipe_buffer base; - struct nouveau_bo *bo; -}; - -static inline struct nouveau_pipe_buffer * -nouveau_buffer(struct pipe_buffer *buf) -{ - return (struct nouveau_pipe_buffer *)buf; -} - -struct nouveau_pipe_winsys { - struct pipe_winsys pws; - - struct nouveau_context *nv; -}; - -extern struct pipe_winsys * -nouveau_create_pipe_winsys(struct nouveau_context *nv); - -struct pipe_context * -nouveau_create_softpipe(struct nouveau_context *nv); - -struct pipe_context * -nouveau_pipe_create(struct nouveau_context *nv); - -#endif diff --git a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c b/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c deleted file mode 100644 index 3908c175088..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nouveau_winsys_softpipe.c +++ /dev/null @@ -1,83 +0,0 @@ -/************************************************************************** - * - * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * - **************************************************************************/ -/* - * Authors: Keith Whitwell - */ - -#include "imports.h" - -#include "pipe/p_defines.h" -#include "pipe/p_format.h" -#include "pipe/softpipe/sp_winsys.h" - -#include "nouveau_context.h" -#include "nouveau_winsys_pipe.h" - -struct nouveau_softpipe_winsys { - struct softpipe_winsys sws; - struct nouveau_context *nv; -}; - -/** - * Return list of surface formats supported by this driver. - */ -static boolean -nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) -{ - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - }; - - return FALSE; -} - - - -struct pipe_context * -nouveau_create_softpipe(struct nouveau_context *nv) -{ - struct nouveau_softpipe_winsys *nvsws; - - nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); - - /* Fill in this struct with callbacks that softpipe will need to - * communicate with the window system, buffer manager, etc. - */ - nvsws->sws.is_format_supported = nouveau_is_format_supported; - nvsws->nv = nv; - - /* Create the softpipe context: - */ - return softpipe_create(nouveau_create_pipe_winsys(nv), &nvsws->sws); -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c deleted file mode 100644 index fe1ea4ed70f..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nv04_surface.c +++ /dev/null @@ -1,226 +0,0 @@ -#include "pipe/p_context.h" - -#include "nouveau_context.h" - -static INLINE int -nv04_surface_format(int cpp) -{ - switch (cpp) { - case 1: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; - case 2: return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; - case 4: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; - default: - return -1; - } -} - -static INLINE int -nv04_rect_format(int cpp) -{ - switch (cpp) { - case 1: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - case 2: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; - case 4: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - default: - return -1; - } -} - -static void -nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct pipe_surface *dst = nv->surf_dst; - struct pipe_surface *src = nv->surf_src; - unsigned dst_offset, src_offset; - - dst_offset = dst->offset + (dy * dst->pitch + dx) * dst->cpp; - src_offset = src->offset + (sy * src->pitch + sx) * src->cpp; - - while (h) { - int count = (h > 2047) ? 2047 : h; - - BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); - OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_WR); - OUT_RING (src->pitch * src->cpp); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (w * src->cpp); - OUT_RING (count); - OUT_RING (0x0101); - OUT_RING (0); - - h -= count; - src_offset += src->pitch * src->cpp * count; - dst_offset += dst->pitch * dst->cpp * count; - } -} - -static void -nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - BEGIN_RING(NvImageBlit, 0x0300, 3); - OUT_RING ((sy << 16) | sx); - OUT_RING ((dy << 16) | dx); - OUT_RING (( h << 16) | w); -} - -static int -nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, - struct pipe_surface *src) -{ - int format; - - if (src->cpp != dst->cpp) - return 1; - - /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback - * to NV_MEMORY_TO_MEMORY_FORMAT in this case. - */ - if ((src->offset & 63) || (dst->offset & 63)) { - BEGIN_RING(NvM2MF, - NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_WR); - - nv->surface_copy = nv04_surface_copy_m2mf; - nv->surf_dst = dst; - nv->surf_src = src; - return 0; - - } - - if ((format = nv04_surface_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); - return 1; - } - nv->surface_copy = nv04_surface_copy_blit; - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (format); - OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); - OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - return 0; -} - -static void -nv04_surface_copy_done(struct nouveau_context *nv) -{ - FIRE_RING(); -} - -static int -nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - int cs2d_format, gdirect_format; - - if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); - return 1; - } - - if ((gdirect_format = nv04_rect_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); - return 1; - } - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (cs2d_format); - OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); - OUT_RING (gdirect_format); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); - OUT_RING (value); - BEGIN_RING(NvGdiRect, - NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); - OUT_RING ((dx << 16) | dy); - OUT_RING (( w << 16) | h); - - FIRE_RING(); - return 0; -} - -int -nouveau_surface_init_nv04(struct nouveau_context *nv) -{ - unsigned class; - int ret; - - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, - &nv->NvM2MF))) { - NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); - return 1; - } - BIND_RING (NvM2MF, nv->next_subchannel++); - BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - - class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : - NV10_CONTEXT_SURFACES_2D; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvCtxSurf2D))) { - NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); - return 1; - } - BIND_RING (NvCtxSurf2D, nv->next_subchannel++); - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RING (nv->channel->vram->handle); - OUT_RING (nv->channel->vram->handle); - - class = nv->chipset < 0x10 ? NV04_IMAGE_BLIT : - NV12_IMAGE_BLIT; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvImageBlit))) { - NOUVEAU_ERR("Error creating blit object: %d\n", ret); - return 1; - } - BIND_RING (NvImageBlit, nv->next_subchannel++); - BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); - OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); - OUT_RING (NV04_IMAGE_BLIT_OPERATION_SRCCOPY); - - class = NV04_GDI_RECTANGLE_TEXT; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvGdiRect))) { - NOUVEAU_ERR("Error creating rect object: %d\n", ret); - return 1; - } - BIND_RING (NvGdiRect, nv->next_subchannel++); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); - OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); - OUT_RING (NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); - OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); - - nv->surface_copy_prep = nv04_surface_copy_prep; - nv->surface_copy = nv04_surface_copy_blit; - nv->surface_copy_done = nv04_surface_copy_done; - nv->surface_fill = nv04_surface_fill; - return 0; -} - diff --git a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c b/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c deleted file mode 100644 index 15a10028613..00000000000 --- a/src/mesa/drivers/dri/nouveau_winsys/nv50_surface.c +++ /dev/null @@ -1,160 +0,0 @@ -#include "pipe/p_context.h" - -#include "nouveau_context.h" - -static INLINE int -nv50_format(int cpp) -{ - switch (cpp) { - case 4: return NV50_2D_DST_FORMAT_32BPP; - case 3: return NV50_2D_DST_FORMAT_24BPP; - case 2: return NV50_2D_DST_FORMAT_16BPP; - case 1: return NV50_2D_DST_FORMAT_8BPP; - default: - return -1; - } -} - -static int -nv50_surface_copy_prep(struct nouveau_context *nv, - struct pipe_surface *dst, struct pipe_surface *src) -{ - int surf_format; - - assert(src->cpp == dst->cpp); - - surf_format = nv50_format(dst->cpp); - assert(surf_format >= 0); - - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); - OUT_RING (0); - OUT_RING (0); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - - BEGIN_RING(Nv2D, NV50_2D_SRC_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_SRC_PITCH, 5); - OUT_RING (src->pitch * src->cpp); - OUT_RING (src->pitch); - OUT_RING (src->height); - OUT_RELOCh(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - - return 0; -} - -static void -nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - BEGIN_RING(Nv2D, 0x0110, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, NV50_2D_BLIT_DST_X, 12); - OUT_RING (dx); - OUT_RING (dy); - OUT_RING (w); - OUT_RING (h); - OUT_RING (0); - OUT_RING (1); - OUT_RING (0); - OUT_RING (1); - OUT_RING (0); - OUT_RING (sx); - OUT_RING (0); - OUT_RING (sy); -} - -static void -nv50_surface_copy_done(struct nouveau_context *nv) -{ - FIRE_RING(); -} - -static int -nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - int surf_format, rect_format; - - surf_format = nv50_format(dst->cpp); - if (surf_format < 0) - return 1; - - rect_format = nv50_format(dst->cpp); - if (rect_format < 0) - return 1; - - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY1, 1); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); - OUT_RING (0); - OUT_RING (0); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - - BEGIN_RING(Nv2D, 0x0580, 3); - OUT_RING (4); - OUT_RING (rect_format); - OUT_RING (value); - - BEGIN_RING(Nv2D, NV50_2D_RECT_X1, 4); - OUT_RING (dx); - OUT_RING (dy); - OUT_RING (dx + w); - OUT_RING (dy + h); - - FIRE_RING(); - - return 0; -} - -int -nouveau_surface_init_nv50(struct nouveau_context *nv) -{ - int ret; - - ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, NV50_2D, - &nv->Nv2D); - if (ret) - return ret; - BIND_RING (Nv2D, 0); - BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RING (nv->channel->vram->handle); - OUT_RING (nv->channel->vram->handle); - BEGIN_RING(Nv2D, NV50_2D_OPERATION, 1); - OUT_RING (NV50_2D_OPERATION_SRCCOPY); - - nv->surface_copy_prep = nv50_surface_copy_prep; - nv->surface_copy = nv50_surface_copy; - nv->surface_copy_done = nv50_surface_copy_done; - nv->surface_fill = nv50_surface_fill; - return 0; -} - diff --git a/src/mesa/pipe/nouveau/nouveau_bo.h b/src/mesa/pipe/nouveau/nouveau_bo.h deleted file mode 100644 index 18020e9c652..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_bo.h +++ /dev/null @@ -1,51 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_BO_H__ -#define __NOUVEAU_BO_H__ - -/* Relocation/Buffer type flags */ -#define NOUVEAU_BO_VRAM (1 << 0) -#define NOUVEAU_BO_GART (1 << 1) -#define NOUVEAU_BO_RD (1 << 2) -#define NOUVEAU_BO_WR (1 << 3) -#define NOUVEAU_BO_RDWR (NOUVEAU_BO_RD | NOUVEAU_BO_WR) -#define NOUVEAU_BO_MAP (1 << 4) -#define NOUVEAU_BO_PIN (1 << 5) -#define NOUVEAU_BO_LOW (1 << 6) -#define NOUVEAU_BO_HIGH (1 << 7) -#define NOUVEAU_BO_OR (1 << 8) -#define NOUVEAU_BO_LOCAL (1 << 9) -#define NOUVEAU_BO_DUMMY (1 << 31) - -struct nouveau_bo { - struct nouveau_device *device; - uint64_t handle; - - uint64_t size; - void *map; - - uint32_t flags; - uint64_t offset; -}; - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_channel.h b/src/mesa/pipe/nouveau/nouveau_channel.h deleted file mode 100644 index b99de9add86..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_channel.h +++ /dev/null @@ -1,39 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_CHANNEL_H__ -#define __NOUVEAU_CHANNEL_H__ - -struct nouveau_channel { - struct nouveau_device *device; - int id; - - struct nouveau_pushbuf *pushbuf; - - struct nouveau_grobj *vram; - struct nouveau_grobj *gart; - - void *user_private; - void (*hang_notify)(struct nouveau_channel *); -}; - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_class.h b/src/mesa/pipe/nouveau/nouveau_class.h deleted file mode 100644 index 5998945677b..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_class.h +++ /dev/null @@ -1,6134 +0,0 @@ -/************************************************************************* - - Autogenerated file, do not edit ! - -************************************************************************** - - Copyright (C) 2006-2007 : - Dmitry Baryshkov, - Laurent Carlier, - Matthieu Castet, - Dawid Gajownik, - Jeremy Kolb, - Stephane Loeuillet, - Patrice Mandin, - Stephane Marchesin, - Serge Martin, - Sylvain Munaut, - Simon Raffeiner, - Ben Skeggs, - Erik Waling, - koala_br, - -All Rights Reserved. - -Permission is hereby granted, free of charge, to any person obtaining -a copy of this software and associated documentation files (the -"Software"), to deal in the Software without restriction, including -without limitation the rights to use, copy, modify, merge, publish, -distribute, sublicense, and/or sell copies of the Software, and to -permit persons to whom the Software is furnished to do so, subject to -the following conditions: - -The above copyright notice and this permission notice (including the -next paragraph) shall be included in all copies or substantial -portions of the Software. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, -EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF -MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. -IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE -LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION -OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION -WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - -*************************************************************************/ - - -#ifndef NOUVEAU_REG_H -#define NOUVEAU_REG_H 1 - - -#define NV01_ROOT 0x00000001 - - - -#define NV01_CONTEXT_DMA 0x00000002 - - - -#define NV01_DEVICE 0x00000003 - - - -#define NV01_TIMER 0x00000004 - -#define NV01_TIMER_SYNCHRONIZE 0x00000100 -#define NV01_TIMER_STOP_ALARM 0x00000104 -#define NV01_TIMER_DMA_NOTIFY 0x00000180 -#define NV01_TIMER_TIME(x) (0x00000300+((x)*4)) -#define NV01_TIMER_TIME__SIZE 0x00000002 -#define NV01_TIMER_ALARM_NOTIFY 0x00000308 - - -#define NV_IMAGE_STENCIL 0x00000010 - -#define NV_IMAGE_STENCIL_NOTIFY 0x00000104 -#define NV_IMAGE_STENCIL_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_STENCIL_IMAGE_OUTPUT 0x00000200 -#define NV_IMAGE_STENCIL_IMAGE_INPUT(x) (0x00000204+((x)*4)) -#define NV_IMAGE_STENCIL_IMAGE_INPUT__SIZE 0x00000002 - - -#define NV_IMAGE_BLEND_AND 0x00000011 - -#define NV_IMAGE_BLEND_AND_NOP 0x00000100 -#define NV_IMAGE_BLEND_AND_NOTIFY 0x00000104 -#define NV_IMAGE_BLEND_AND_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_BLEND_AND_IMAGE_OUTPUT 0x00000200 -#define NV_IMAGE_BLEND_AND_BETA_INPUT 0x00000204 -#define NV_IMAGE_BLEND_AND_IMAGE_INPUT 0x00000208 - - -#define NV01_CONTEXT_BETA1 0x00000012 - -#define NV01_CONTEXT_BETA1_NOP 0x00000100 -#define NV01_CONTEXT_BETA1_NOTIFY 0x00000104 -#define NV01_CONTEXT_BETA1_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_BETA1_BETA_1D31 0x00000300 - - -#define NV_IMAGE_ROP_AND 0x00000013 - -#define NV_IMAGE_ROP_AND_NOTIFY 0x00000104 -#define NV_IMAGE_ROP_AND_DMA_NOTIFY 0x00000180 -#define NV_IMAGE_ROP_AND_IMAGE_OUTPUT 0x00000200 -#define NV_IMAGE_ROP_AND_ROP_INPUT 0x00000204 -#define NV_IMAGE_ROP_AND_IMAGE_INPUT(x) (0x00000208+((x)*4)) -#define NV_IMAGE_ROP_AND_IMAGE_INPUT__SIZE 0x00000002 - - -#define NV_IMAGE_COLOR_KEY 0x00000015 - - - -#define NV01_CONTEXT_COLOR_KEY 0x00000017 - -#define NV01_CONTEXT_COLOR_KEY_NOP 0x00000100 -#define NV01_CONTEXT_COLOR_KEY_NOTIFY 0x00000104 -#define NV01_CONTEXT_COLOR_KEY_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT 0x00000300 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A8Y8 0x00000001 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X24Y8 0x00000002 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A1R5G5B5 0x00000003 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X17R5G5B5 0x00000004 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A8R8G8B8 0x00000005 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X8R8G8B8 0x00000006 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A16Y16 0x00000007 -#define NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16Y16 0x00000008 -#define NV01_CONTEXT_COLOR_KEY_COLOR 0x00000304 - - -#define NV01_CONTEXT_PATTERN 0x00000018 - -#define NV01_CONTEXT_PATTERN_NOP 0x00000100 -#define NV01_CONTEXT_PATTERN_NOTIFY 0x00000104 -#define NV01_CONTEXT_PATTERN_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_PATTERN_COLOR_FORMAT 0x00000300 -#define NV01_CONTEXT_PATTERN_MONOCHROME_FORMAT 0x00000304 -#define NV01_CONTEXT_PATTERN_SHAPE 0x00000308 -#define NV01_CONTEXT_PATTERN_COLOR(x) (0x00000310+((x)*4)) -#define NV01_CONTEXT_PATTERN_COLOR__SIZE 0x00000002 -#define NV01_CONTEXT_PATTERN_PATTERN(x) (0x00000318+((x)*4)) -#define NV01_CONTEXT_PATTERN_PATTERN__SIZE 0x00000002 - - -#define NV01_CONTEXT_CLIP_RECTANGLE 0x00000019 - -#define NV01_CONTEXT_CLIP_RECTANGLE_NOP 0x00000100 -#define NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY 0x00000104 -#define NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY 0x00000180 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT 0x00000300 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_SHIFT 0 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_MASK 0x0000ffff -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_SHIFT 16 -#define NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_MASK 0xffff0000 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE 0x00000304 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_SHIFT 0 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_MASK 0x0000ffff -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_SHIFT 16 -#define NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_MASK 0xffff0000 - - -#define NV01_RENDER_SOLID_LINE 0x0000001c - -#define NV01_RENDER_SOLID_LINE_NOP 0x00000100 -#define NV01_RENDER_SOLID_LINE_NOTIFY 0x00000104 -#define NV01_RENDER_SOLID_LINE_PATCH 0x0000010c -#define NV01_RENDER_SOLID_LINE_DMA_NOTIFY 0x00000180 -#define NV01_RENDER_SOLID_LINE_CLIP_RECTANGLE 0x00000184 -#define NV01_RENDER_SOLID_LINE_PATTERN 0x00000188 -#define NV01_RENDER_SOLID_LINE_ROP 0x0000018c -#define NV01_RENDER_SOLID_LINE_BETA1 0x00000190 -#define NV01_RENDER_SOLID_LINE_SURFACE 0x00000194 -#define NV01_RENDER_SOLID_LINE_OPERATION 0x000002fc -#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_RENDER_SOLID_LINE_OPERATION_ROP_AND 0x00000001 -#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_AND 0x00000002 -#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY 0x00000003 -#define NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV01_RENDER_SOLID_LINE_OPERATION_BLEND_PREMULT 0x00000005 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT 0x00000300 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A8Y8 0x00000001 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X24Y8 0x00000002 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A1R5G5B5 0x00000003 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X17R5G5B5 0x00000004 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A8R8G8B8 0x00000005 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X8R8G8B8 0x00000006 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A16Y16 0x00000007 -#define NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16Y16 0x00000008 -#define NV01_RENDER_SOLID_LINE_COLOR 0x00000304 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0(x) (0x00000400+((x)*8)) -#define NV01_RENDER_SOLID_LINE_LINE_POINT0__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1(x) (0x00000404+((x)*8)) -#define NV01_RENDER_SOLID_LINE_LINE_POINT1__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X(x) (0x00000480+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y(x) (0x00000484+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X(x) (0x00000488+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y(x) (0x0000048c+((x)*16)) -#define NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_POLYLINE(x) (0x00000500+((x)*4)) -#define NV01_RENDER_SOLID_LINE_POLYLINE__SIZE 0x00000020 -#define NV01_RENDER_SOLID_LINE_POLYLINE_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_POLYLINE_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_POLYLINE_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X(x) (0x00000580+((x)*8)) -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y(x) (0x00000584+((x)*8)) -#define NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR(x) (0x00000600+((x)*8)) -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT(x) (0x00000604+((x)*8)) -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT__SIZE 0x00000010 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_SHIFT 0 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_SHIFT 16 -#define NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_MASK 0xffff0000 - - -#define NV01_RENDER_SOLID_TRIANGLE 0x0000001d - -#define NV01_RENDER_SOLID_TRIANGLE_NOP 0x00000100 -#define NV01_RENDER_SOLID_TRIANGLE_NOTIFY 0x00000104 -#define NV01_RENDER_SOLID_TRIANGLE_PATCH 0x0000010c -#define NV01_RENDER_SOLID_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV01_RENDER_SOLID_TRIANGLE_CLIP_RECTANGLE 0x00000184 -#define NV01_RENDER_SOLID_TRIANGLE_PATTERN 0x00000188 -#define NV01_RENDER_SOLID_TRIANGLE_ROP 0x0000018c -#define NV01_RENDER_SOLID_TRIANGLE_BETA1 0x00000190 -#define NV01_RENDER_SOLID_TRIANGLE_SURFACE 0x00000194 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION 0x000002fc -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_ROP_AND 0x00000001 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_AND 0x00000002 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY 0x00000003 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_PREMULT 0x00000005 -#define NV01_RENDER_SOLID_TRIANGLE_COLOR_FORMAT 0x00000300 -#define NV01_RENDER_SOLID_TRIANGLE_COLOR 0x00000304 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0 0x00000310 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1 0x00000314 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2 0x00000318 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_X 0x00000320 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_Y 0x00000324 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_X 0x00000328 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_Y 0x0000032c -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_X 0x00000330 -#define NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_Y 0x00000334 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH(x) (0x00000400+((x)*4)) -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH__SIZE 0x00000020 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X(x) (0x00000480+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y(x) (0x00000484+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR(x) (0x00000500+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0(x) (0x00000504+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1(x) (0x00000508+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2(x) (0x0000050c+((x)*16)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2__SIZE 0x00000008 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR(x) (0x00000580+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT(x) (0x00000584+((x)*8)) -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT__SIZE 0x00000010 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_SHIFT 0 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_SHIFT 16 -#define NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_MASK 0xffff0000 - - -#define NV01_RENDER_SOLID_RECTANGLE 0x0000001e - -#define NV01_RENDER_SOLID_RECTANGLE_NOP 0x00000100 -#define NV01_RENDER_SOLID_RECTANGLE_NOTIFY 0x00000104 -#define NV01_RENDER_SOLID_RECTANGLE_PATCH 0x0000010c -#define NV01_RENDER_SOLID_RECTANGLE_DMA_NOTIFY 0x00000180 -#define NV01_RENDER_SOLID_RECTANGLE_CLIP_RECTANGLE 0x00000184 -#define NV01_RENDER_SOLID_RECTANGLE_PATTERN 0x00000188 -#define NV01_RENDER_SOLID_RECTANGLE_ROP 0x0000018c -#define NV01_RENDER_SOLID_RECTANGLE_BETA1 0x00000190 -#define NV01_RENDER_SOLID_RECTANGLE_SURFACE 0x00000194 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION 0x000002fc -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_ROP_AND 0x00000001 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_AND 0x00000002 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY 0x00000003 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_PREMULT 0x00000005 -#define NV01_RENDER_SOLID_RECTANGLE_COLOR_FORMAT 0x00000300 -#define NV01_RENDER_SOLID_RECTANGLE_COLOR 0x00000304 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT(x) (0x00000400+((x)*8)) -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT__SIZE 0x00000010 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_SHIFT 0 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_MASK 0x0000ffff -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_SHIFT 16 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_MASK 0xffff0000 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE__SIZE 0x00000010 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_SHIFT 0 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_MASK 0x0000ffff -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_SHIFT 16 -#define NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_MASK 0xffff0000 - - -#define NV01_IMAGE_BLIT 0x0000001f - -#define NV01_IMAGE_BLIT_NOP 0x00000100 -#define NV01_IMAGE_BLIT_NOTIFY 0x00000104 -#define NV01_IMAGE_BLIT_PATCH 0x0000010c -#define NV01_IMAGE_BLIT_DMA_NOTIFY 0x00000180 -#define NV01_IMAGE_BLIT_COLOR_KEY 0x00000184 -#define NV01_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 -#define NV01_IMAGE_BLIT_PATTERN 0x0000018c -#define NV01_IMAGE_BLIT_ROP 0x00000190 -#define NV01_IMAGE_BLIT_BETA1 0x00000194 -#define NV01_IMAGE_BLIT_SURFACE 0x0000019c -#define NV01_IMAGE_BLIT_OPERATION 0x000002fc -#define NV01_IMAGE_BLIT_IMAGE_INPUT 0x00000204 -#define NV01_IMAGE_BLIT_POINT_IN 0x00000300 -#define NV01_IMAGE_BLIT_POINT_IN_X_SHIFT 0 -#define NV01_IMAGE_BLIT_POINT_IN_X_MASK 0x0000ffff -#define NV01_IMAGE_BLIT_POINT_IN_Y_SHIFT 16 -#define NV01_IMAGE_BLIT_POINT_IN_Y_MASK 0xffff0000 -#define NV01_IMAGE_BLIT_POINT_OUT 0x00000304 -#define NV01_IMAGE_BLIT_POINT_OUT_X_SHIFT 0 -#define NV01_IMAGE_BLIT_POINT_OUT_X_MASK 0x0000ffff -#define NV01_IMAGE_BLIT_POINT_OUT_Y_SHIFT 16 -#define NV01_IMAGE_BLIT_POINT_OUT_Y_MASK 0xffff0000 -#define NV01_IMAGE_BLIT_SIZE 0x00000308 -#define NV01_IMAGE_BLIT_SIZE_W_SHIFT 0 -#define NV01_IMAGE_BLIT_SIZE_W_MASK 0x0000ffff -#define NV01_IMAGE_BLIT_SIZE_H_SHIFT 16 -#define NV01_IMAGE_BLIT_SIZE_H_MASK 0xffff0000 - - -#define NV01_IMAGE_FROM_CPU 0x00000021 - -#define NV01_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV01_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV01_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV01_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV01_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 -#define NV01_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x00000188 -#define NV01_IMAGE_FROM_CPU_PATTERN 0x0000018c -#define NV01_IMAGE_FROM_CPU_ROP 0x00000190 -#define NV01_IMAGE_FROM_CPU_BETA1 0x00000194 -#define NV01_IMAGE_FROM_CPU_SURFACE 0x00000198 -#define NV01_IMAGE_FROM_CPU_OPERATION 0x000002fc -#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_AND 0x00000000 -#define NV01_IMAGE_FROM_CPU_OPERATION_ROP_AND 0x00000001 -#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_AND 0x00000002 -#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY 0x00000003 -#define NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV01_IMAGE_FROM_CPU_OPERATION_BLEND_PREMULT 0x00000005 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_Y8 0x00000001 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A1R5G5B5 0x00000002 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X1R5G5B5 0x00000003 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A8R8G8B8 0x00000004 -#define NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X8R8G8B8 0x00000005 -#define NV01_IMAGE_FROM_CPU_POINT 0x00000304 -#define NV01_IMAGE_FROM_CPU_POINT_X_SHIFT 0 -#define NV01_IMAGE_FROM_CPU_POINT_X_MASK 0x0000ffff -#define NV01_IMAGE_FROM_CPU_POINT_Y_SHIFT 16 -#define NV01_IMAGE_FROM_CPU_POINT_Y_MASK 0xffff0000 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT 0x00000308 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_SHIFT 0 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_W_MASK 0x0000ffff -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_SHIFT 16 -#define NV01_IMAGE_FROM_CPU_SIZE_OUT_H_MASK 0xffff0000 -#define NV01_IMAGE_FROM_CPU_SIZE_IN 0x0000030c -#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 -#define NV01_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff -#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 -#define NV01_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 -#define NV01_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) -#define NV01_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 - - -#define NV01_NULL 0x00000030 - - - -#define NV03_STRETCHED_IMAGE_FROM_CPU 0x00000036 - -#define NV03_STRETCHED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV03_STRETCHED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV03_STRETCHED_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV03_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY 0x00000184 -#define NV03_STRETCHED_IMAGE_FROM_CPU_PATTERN 0x00000188 -#define NV03_STRETCHED_IMAGE_FROM_CPU_ROP 0x0000018c -#define NV03_STRETCHED_IMAGE_FROM_CPU_BETA1 0x00000190 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000194 -#define NV03_STRETCHED_IMAGE_FROM_CPU_OPERATION 0x000002fc -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN 0x00000304 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_DX_DU 0x00000308 -#define NV03_STRETCHED_IMAGE_FROM_CPU_DY_DV 0x0000030c -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT 0x00000310 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE 0x00000314 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4 0x00000318 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_SHIFT 0 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_MASK 0x0000ffff -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_SHIFT 16 -#define NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_MASK 0xffff0000 -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) -#define NV03_STRETCHED_IMAGE_FROM_CPU_COLOR__SIZE 0x00000020 - - -#define NV03_SCALED_IMAGE_FROM_MEMORY 0x00000037 - -#define NV03_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 -#define NV03_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 -#define NV03_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 -#define NV03_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c -#define NV03_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 -#define NV03_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000194 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 -#define NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY 0x00000003 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000005 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT 0x00000310 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE 0x00000314 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DU_DX 0x00000318 -#define NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DV_DY 0x0000031c -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE 0x00000400 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_MASK 0xffff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT 0x00000404 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_MASK 0x00ff0000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CENTER 0x00010000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CORNER 0x00020000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_SHIFT 24 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_MASK 0xff000000 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_OFFSET 0x00000408 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT 0x0000040c -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_SHIFT 0 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_MASK 0x0000ffff -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_SHIFT 16 -#define NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_MASK 0xffff0000 - - -#define NV04_DVD_SUBPICTURE 0x00000038 - -#define NV04_DVD_SUBPICTURE_NOP 0x00000100 -#define NV04_DVD_SUBPICTURE_NOTIFY 0x00000104 -#define NV04_DVD_SUBPICTURE_WAIT_FOR_IDLE 0x00000108 -#define NV04_DVD_SUBPICTURE_DMA_NOTIFY 0x00000180 -#define NV04_DVD_SUBPICTURE_DMA_OVERLAY 0x00000184 -#define NV04_DVD_SUBPICTURE_DMA_IMAGEIN 0x00000188 -#define NV04_DVD_SUBPICTURE_DMA_IMAGEOUT 0x0000018c -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT 0x00000300 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE 0x00000304 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT 0x00000308 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEOUT_OFFSET 0x0000030c -#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DU_DX 0x00000310 -#define NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DV_DY 0x00000314 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE 0x00000318 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT 0x0000031c -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_IMAGEIN_OFFSET 0x00000320 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT 0x00000324 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_SHIFT 0 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_SHIFT 16 -#define NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DU_DX 0x00000328 -#define NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DV_DY 0x0000032c -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE 0x00000330 -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_SHIFT 0 -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_SHIFT 16 -#define NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT 0x00000334 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_SHIFT 0 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_SHIFT 16 -#define NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_MASK 0xffff0000 -#define NV04_DVD_SUBPICTURE_OVERLAY_OFFSET 0x00000338 -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT 0x0000033c -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_SHIFT 0 -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_MASK 0x0000ffff -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_SHIFT 16 -#define NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK 0xffff0000 - - -#define NV04_MEMORY_TO_MEMORY_FORMAT 0x00000039 - -#define NV04_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 -#define NV04_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 -#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 -#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN 0x00000184 -#define NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT 0x00000188 -#define NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c -#define NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT 0x00000310 -#define NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_IN 0x00000314 -#define NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT 0x00000318 -#define NV04_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN 0x0000031c -#define NV04_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT 0x00000320 -#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT 0x00000324 -#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT 0 -#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK 0x0000000f -#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT 8 -#define NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK 0x00000f00 -#define NV04_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY 0x00000328 - - -#define NV01_MEMORY_LOCAL_BANKED 0x0000003d - - - -#define NV01_MAPPING_SYSTEM 0x0000003e - - - -#define NV03_MEMORY_LOCAL_CURSOR 0x0000003f - - - -#define NV01_MEMORY_LOCAL_LINEAR 0x00000040 - - - -#define NV01_MAPPING_LOCAL 0x00000041 - - - -#define NV04_CONTEXT_SURFACES_2D 0x00000042 - -#define NV04_CONTEXT_SURFACES_2D_NOP 0x00000100 -#define NV04_CONTEXT_SURFACES_2D_NOTIFY 0x00000104 -#define NV04_CONTEXT_SURFACES_2D_PM_TRIGGER 0x00000140 -#define NV04_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 -#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE 0x00000184 -#define NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_DESTIN 0x00000188 -#define NV04_CONTEXT_SURFACES_2D_FORMAT 0x00000300 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y8 0x00000001 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_Z1R5G5B5 0x00000002 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_X1R5G5B5 0x00000003 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5 0x00000004 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y16 0x00000005 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_Z8R8G8B8 0x00000006 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_X8R8G8B8 0x00000007 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_X1A7R8G8B8 0x00000009 -#define NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8 0x0000000a -#define NV04_CONTEXT_SURFACES_2D_FORMAT_Y32 0x0000000b -#define NV04_CONTEXT_SURFACES_2D_PITCH 0x00000304 -#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 -#define NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 -#define NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 -#define NV04_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c - - -#define NV03_CONTEXT_ROP 0x00000043 - -#define NV03_CONTEXT_ROP_NOP 0x00000100 -#define NV03_CONTEXT_ROP_NOTIFY 0x00000104 -#define NV03_CONTEXT_ROP_DMA_NOTIFY 0x00000180 -#define NV03_CONTEXT_ROP_ROP 0x00000300 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SHIFT 0 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_MASK 0x0000000f -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_CLEAR 0x00000000 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOR 0x00000001 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_INVERTED 0x00000002 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY_INVERTED 0x00000003 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_REVERSE 0x00000004 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_INVERT 0x00000005 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_XOR 0x00000006 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NAND 0x00000007 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND 0x00000008 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_EQUI 0x00000009 -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOOP 0x0000000a -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_INVERTED 0x0000000b -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY 0x0000000c -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_REVERSE 0x0000000d -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR 0x0000000e -#define NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SET 0x0000000f -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SHIFT 4 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_MASK 0x000000f0 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_CLEAR 0x00000000 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOR 0x00000010 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_INVERTED 0x00000020 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY_INVERTED 0x00000030 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_REVERSE 0x00000040 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_INVERT 0x00000050 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_XOR 0x00000060 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NAND 0x00000070 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND 0x00000080 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_EQUI 0x00000090 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOOP 0x000000a0 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_INVERTED 0x000000b0 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY 0x000000c0 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_REVERSE 0x000000d0 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR 0x000000e0 -#define NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SET 0x000000f0 - - -#define NV04_IMAGE_PATTERN 0x00000044 - -#define NV04_IMAGE_PATTERN_NOP 0x00000100 -#define NV04_IMAGE_PATTERN_NOTIFY 0x00000104 -#define NV04_IMAGE_PATTERN_DMA_NOTIFY 0x00000180 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT 0x00000300 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A16R5G6B5 0x00000001 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT_X16A1R5G5B5 0x00000002 -#define NV04_IMAGE_PATTERN_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT 0x00000304 -#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_CGA6 0x00000001 -#define NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_LE 0x00000002 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE 0x00000308 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_8X8 0x00000000 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_64X1 0x00000001 -#define NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_1X64 0x00000002 -#define NV04_IMAGE_PATTERN_PATTERN_SELECT 0x0000030c -#define NV04_IMAGE_PATTERN_PATTERN_SELECT_MONO 0x00000001 -#define NV04_IMAGE_PATTERN_PATTERN_SELECT_COLOR 0x00000002 -#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR0 0x00000310 -#define NV04_IMAGE_PATTERN_MONOCHROME_COLOR1 0x00000314 -#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0 0x00000318 -#define NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1 0x0000031c -#define NV04_IMAGE_PATTERN_PATTERN_Y8(x) (0x00000400+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_Y8__SIZE 0x00000010 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_MASK 0x000000ff -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_SHIFT 8 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_MASK 0x0000ff00 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_MASK 0x00ff0000 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_SHIFT 24 -#define NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_MASK 0xff000000 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5(x) (0x00000500+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5__SIZE 0x00000020 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_MASK 0x0000001f -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_SHIFT 5 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_MASK 0x000007e0 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_SHIFT 11 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_MASK 0x0000f800 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_MASK 0x001f0000 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_SHIFT 21 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_MASK 0x07e00000 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_SHIFT 27 -#define NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_MASK 0xf8000000 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5(x) (0x00000600+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5__SIZE 0x00000020 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_MASK 0x0000001f -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_SHIFT 5 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_MASK 0x000003e0 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_SHIFT 10 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_MASK 0x00007c00 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_MASK 0x001f0000 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_SHIFT 21 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_MASK 0x03e00000 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_SHIFT 26 -#define NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_MASK 0x7c000000 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8(x) (0x00000700+((x)*4)) -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8__SIZE 0x00000040 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_SHIFT 0 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_MASK 0x000000ff -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_SHIFT 8 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_MASK 0x0000ff00 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_SHIFT 16 -#define NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_MASK 0x00ff0000 - - -#define NV03_VIDEO_LUT_CURSOR_DAC 0x00000046 - -#define NV03_VIDEO_LUT_CURSOR_DAC_SYNCHRONIZE 0x00000100 -#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_IMAGE 0x00000104 -#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_CURSOR 0x00000108 -#define NV03_VIDEO_LUT_CURSOR_DAC_STOP_DAC 0x0000010c -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_NOTIFY 0x00000180 -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE(x) (0x00000184+((x)*4)) -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT(x) (0x0000018c+((x)*4)) -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR(x) (0x00000194+((x)*4)) -#define NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_GET 0x000002fc -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET(x) (0x00000300+((x)*8)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT(x) (0x00000304+((x)*8)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET(x) (0x00000340+((x)*12)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT(x) (0x00000344+((x)*12)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_MASK 0xffff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT(x) (0x00000348+((x)*12)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A 0x00000358 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_MASK 0xffff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE(x) (0x00000380+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_MASK 0xffff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC(x) (0x00000384+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC(x) (0x00000388+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE(x) (0x0000038c+((x)*16)) -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE__SIZE 0x00000002 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_SHIFT 0 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_MASK 0x0000ffff -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_SHIFT 16 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_MASK 0x0fff0000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_SHIFT 28 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_MASK 0xf0000000 -#define NV03_VIDEO_LUT_CURSOR_DAC_SET_PIXEL_CLOCK 0x000003a0 - - -#define NV03_DX3_TEXTURED_TRIANGLE 0x00000048 - -#define NV03_DX3_TEXTURED_TRIANGLE_NOP 0x00000100 -#define NV03_DX3_TEXTURED_TRIANGLE_NOTIFY 0x00000104 -#define NV03_DX3_TEXTURED_TRIANGLE_PATCH 0x0000010c -#define NV03_DX3_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV03_DX3_TEXTURED_TRIANGLE_DMA_TEXTURE 0x00000184 -#define NV03_DX3_TEXTURED_TRIANGLE_CLIP_RECTANGLE 0x00000188 -#define NV03_DX3_TEXTURED_TRIANGLE_SURFACE 0x0000018c -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_OFFSET 0x00000304 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT 0x00000308 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_MASK 0x0000ffff -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_MASK 0x000f0000 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT 20 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK 0x00f00000 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_SHIFT 24 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_MASK 0x0f000000 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_SHIFT 28 -#define NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_MASK 0xf0000000 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER 0x0000030c -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_MASK 0x0000001f -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_MASK 0x00001f00 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_MASK 0x00ff0000 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR 0x00000310 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_MASK 0x000000ff -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_MASK 0x0000ff00 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_MASK 0x00ff0000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT 0x00000314 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_MASK 0x0000000f -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_SHIFT 4 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_MASK 0x00000030 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_SHIFT 6 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_MASK 0x000000c0 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_MASK 0x00000f00 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_SHIFT 12 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_MASK 0x00007000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_PERSPECTIVE_ENABLE (1 << 15) -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_MASK 0x000f0000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_SHIFT 20 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_MASK 0x00f00000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_SHIFT 24 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_MASK 0x07000000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_SHIFT 27 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_MASK 0x18000000 -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_BETA (1 << 29) -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_DST_BLEND (1 << 30) -#define NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SRC_BLEND (1 << 31) -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL 0x00000318 -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_MASK 0x000000ff -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_MASK 0xffffff00 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR(x) (0x00001000+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_SHIFT 0 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_MASK 0x0000000f -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_SHIFT 4 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_MASK 0x000000f0 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_SHIFT 8 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_MASK 0x00000f00 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_SHIFT 12 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_MASK 0x0000f000 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_SHIFT 16 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_MASK 0x000f0000 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_SHIFT 20 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_MASK 0x00f00000 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_SHIFT 24 -#define NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_MASK 0xff000000 -#define NV03_DX3_TEXTURED_TRIANGLE_COLOR(x) (0x00001004+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_COLOR__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_X(x) (0x00001008+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_X__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_Y(x) (0x0000100c+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_Y__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_Z(x) (0x00001010+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_Z__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_M(x) (0x00001014+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_M__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_U(x) (0x00001018+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_U__SIZE 0x00000040 -#define NV03_DX3_TEXTURED_TRIANGLE_V(x) (0x0000101c+((x)*32)) -#define NV03_DX3_TEXTURED_TRIANGLE_V__SIZE 0x00000040 - - -#define NV04_GDI_RECTANGLE_TEXT 0x0000004a - -#define NV04_GDI_RECTANGLE_TEXT_NOP 0x00000100 -#define NV04_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 -#define NV04_GDI_RECTANGLE_TEXT_PATCH 0x0000010c -#define NV04_GDI_RECTANGLE_TEXT_PM_TRIGGER 0x00000140 -#define NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 -#define NV04_GDI_RECTANGLE_TEXT_DMA_FONTS 0x00000184 -#define NV04_GDI_RECTANGLE_TEXT_PATTERN 0x00000188 -#define NV04_GDI_RECTANGLE_TEXT_ROP 0x0000018c -#define NV04_GDI_RECTANGLE_TEXT_BETA1 0x00000190 -#define NV04_GDI_RECTANGLE_TEXT_BETA4 0x00000194 -#define NV04_GDI_RECTANGLE_TEXT_SURFACE 0x00000198 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc -#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_AND 0x00000000 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION_ROP_AND 0x00000001 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_AND 0x00000002 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY 0x00000003 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_PREMULT 0x00000005 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5 0x00000001 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_X16A1R5G5B5 0x00000002 -#define NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_CGA6 0x00000001 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE 0x00000002 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(x) (0x00000400+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE(x) (0x00000404+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0 0x000005f4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1 0x000005f8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_B 0x000005fc -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0(x) (0x00000600+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1(x) (0x00000604+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1__SIZE 0x00000020 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x000007ec -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x000007f0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_C 0x000007f4 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C 0x000007f8 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_POINT_C 0x000007fc -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000800+((x)*4)) -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000080 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x00000be4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x00000be8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR0_E 0x00000bec -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_E 0x00000bf0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x00000bf4 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x00000bf8 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_POINT_E 0x00000bfc -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00000c00+((x)*4)) -#define NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000080 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F 0x00000ff0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_MASK 0x0fffffff -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_SHIFT 28 -#define NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_MASK 0xf0000000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0 0x00000ff4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1 0x00000ff8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_F 0x00000ffc -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F(x) (0x00001000+((x)*4)) -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F__SIZE 0x00000100 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_MASK 0x000000ff -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_SHIFT 8 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_MASK 0x000fff00 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_SHIFT 20 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_MASK 0xfff00000 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G 0x000017f0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_MASK 0x0fffffff -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_SHIFT 28 -#define NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_MASK 0xf0000000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0 0x000017f4 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1 0x000017f8 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_COLOR1_G 0x000017fc -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT(x) (0x00001800+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT__SIZE 0x00000100 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_SHIFT 0 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_MASK 0x0000ffff -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_SHIFT 16 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_MASK 0xffff0000 -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX(x) (0x00001804+((x)*8)) -#define NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX__SIZE 0x00000100 - - -#define NV03_GDI_RECTANGLE_TEXT 0x0000004b - -#define NV03_GDI_RECTANGLE_TEXT_NOP 0x00000100 -#define NV03_GDI_RECTANGLE_TEXT_NOTIFY 0x00000104 -#define NV03_GDI_RECTANGLE_TEXT_DMA_NOTIFY 0x00000180 -#define NV03_GDI_RECTANGLE_TEXT_PATTERN 0x00000184 -#define NV03_GDI_RECTANGLE_TEXT_ROP 0x00000188 -#define NV03_GDI_RECTANGLE_TEXT_BETA1 0x0000018c -#define NV03_GDI_RECTANGLE_TEXT_SURFACE 0x00000190 -#define NV03_GDI_RECTANGLE_TEXT_OPERATION 0x000002fc -#define NV03_GDI_RECTANGLE_TEXT_COLOR_FORMAT 0x00000300 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT 0x00000304 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_A 0x000003fc -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT 0x00000400 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE 0x00000404 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B 0x000007f4 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B 0x000007f8 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_B 0x000007fc -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0 0x00000800 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1 0x00000804 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0 0x00000bec -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1 0x00000bf0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_C 0x00000bf4 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C 0x00000bf8 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_POINT_C 0x00000bfc -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_X_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x) (0x00000c00+((x)*4)) -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE 0x00000020 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0 0x00000fe8 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1 0x00000fec -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_D 0x00000ff0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D 0x00000ff4 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D 0x00000ff8 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_POINT_D 0x00000ffc -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_X_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D(x) (0x00001000+((x)*4)) -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D__SIZE 0x00000020 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0 0x000013e4 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1 0x000013e8 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_COLOR0_E 0x000013ec -#define NV03_GDI_RECTANGLE_TEXT_COLOR1_E 0x000013f0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E 0x000013f4 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E 0x000013f8 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_POINT_E 0x000013fc -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT 0 -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_X_MASK 0x0000ffff -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT 16 -#define NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK 0xffff0000 -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x) (0x00001400+((x)*4)) -#define NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE 0x00000020 - - -#define NV04_SWIZZLED_SURFACE 0x00000052 - -#define NV04_SWIZZLED_SURFACE_NOP 0x00000100 -#define NV04_SWIZZLED_SURFACE_NOTIFY 0x00000104 -#define NV04_SWIZZLED_SURFACE_DMA_NOTIFY 0x00000180 -#define NV04_SWIZZLED_SURFACE_DMA_IMAGE 0x00000184 -#define NV04_SWIZZLED_SURFACE_FORMAT 0x00000300 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_SHIFT 0 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK 0x000000ff -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y8 0x00000001 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000002 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000003 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_R5G6B5 0x00000004 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y16 0x00000005 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000006 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000007 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000008 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000009 -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8 0x0000000a -#define NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y32 0x0000000b -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_MASK 0xff000000 -#define NV04_SWIZZLED_SURFACE_OFFSET 0x00000304 - - -#define NV04_CONTEXT_SURFACES_3D 0x00000053 - -#define NV04_CONTEXT_SURFACES_3D_NOP 0x00000100 -#define NV04_CONTEXT_SURFACES_3D_NOTIFY 0x00000104 -#define NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 -#define NV04_CONTEXT_SURFACES_3D_DMA_COLOR 0x00000184 -#define NV04_CONTEXT_SURFACES_3D_DMA_ZETA 0x00000188 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL 0x000002f8 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL 0x000002fc -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_FORMAT 0x00000300 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK 0x000000ff -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5 0x00000001 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_X1R5G5B5 0x00000002 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_R5G6B5 0x00000003 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8 0x00000004 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_X8R8G8B8 0x00000005 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8 0x00000006 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8 0x00000007 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_A8R8G8B8 0x00000008 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT 8 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK 0x0000ff00 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_PITCH 0x00000100 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SWIZZLE 0x00000200 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_MASK 0x00ff0000 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_MASK 0xff000000 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE 0x00000304 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_PITCH 0x00000308 -#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_SHIFT 0 -#define NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK 0x0000ffff -#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT 16 -#define NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK 0xffff0000 -#define NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x0000030c -#define NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000310 - - -#define NV04_DX5_TEXTURED_TRIANGLE 0x00000054 - -#define NV04_DX5_TEXTURED_TRIANGLE_NOP 0x00000100 -#define NV04_DX5_TEXTURED_TRIANGLE_NOTIFY 0x00000104 -#define NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV04_DX5_TEXTURED_TRIANGLE_DMA_A 0x00000184 -#define NV04_DX5_TEXTURED_TRIANGLE_DMA_B 0x00000188 -#define NV04_DX5_TEXTURED_TRIANGLE_SURFACE 0x0000018c -#define NV04_DX5_TEXTURED_TRIANGLE_COLORKEY 0x00000300 -#define NV04_DX5_TEXTURED_TRIANGLE_OFFSET 0x00000304 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT 0x00000308 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_MASK 0x00000003 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_SHIFT 2 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_MASK 0x0000000c -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CENTER 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER 0x00000020 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CENTER 0x00000040 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER 0x00000080 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_Y8 0x00000100 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A1R5G5B5 0x00000200 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X1R5G5B5 0x00000300 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A4R4G4B4 0x00000400 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_R5G6B5 0x00000500 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A8R8G8B8 0x00000600 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X8R8G8B8 0x00000700 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT 0x01000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT 0x02000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE 0x03000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER 0x04000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP 0x05000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPU (1 << 27) -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_REPEAT 0x10000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MIRRORED_REPEAT 0x20000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE 0x30000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_BORDER 0x40000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP 0x50000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPV (1 << 31) -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER 0x0000030c -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST 0x01000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR 0x02000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST 0x10000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR 0x20000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND 0x00000310 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_MASK 0x0000000f -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT 0x00000040 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD 0x00000080 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_PHONG 0x000000c0 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK 0x0f000000 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT 28 -#define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK 0xf0000000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL 0x00000314 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE (1 << 12) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN (1 << 13) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT 14 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_MASK 0x0000c000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_MASK 0x00300000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE (1 << 22) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE (1 << 23) -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_MASK 0x3f000000 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT 30 -#define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_MASK 0xc0000000 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR 0x00000318 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(x) (0x00000400+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY(x) (0x00000404+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ(x) (0x00000408+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW(x) (0x0000040c+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR(x) (0x00000410+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_MASK 0x0000ff00 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_MASK 0xff000000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR(x) (0x00000414+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_MASK 0x000000ff -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_MASK 0x0000ff00 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_MASK 0x00ff0000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_SHIFT 24 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_MASK 0xff000000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU(x) (0x00000418+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV(x) (0x0000041c+((x)*32)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV__SIZE 0x00000010 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(x) (0x00000600+((x)*4)) -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE__SIZE 0x00000040 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 -#define NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 - - -#define NV04_DX6_MULTITEX_TRIANGLE 0x00000055 - -#define NV04_DX6_MULTITEX_TRIANGLE_NOP 0x00000100 -#define NV04_DX6_MULTITEX_TRIANGLE_NOTIFY 0x00000104 -#define NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY 0x00000180 -#define NV04_DX6_MULTITEX_TRIANGLE_DMA_A 0x00000184 -#define NV04_DX6_MULTITEX_TRIANGLE_DMA_B 0x00000188 -#define NV04_DX6_MULTITEX_TRIANGLE_SURFACE 0x0000018c -#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET(x) (0x00000308+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_OFFSET__SIZE 0x00000002 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT(x) (0x00000310+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT__SIZE 0x00000002 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK 0x00000030 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT 6 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_MASK 0x000000c0 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_MASK 0x07000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPU (1 << 27) -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_SHIFT 28 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_MASK 0x70000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPV (1 << 31) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER(x) (0x00000318+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER__SIZE 0x00000002 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK 0x00007f00 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE (1 << 15) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_MASK 0x07000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE (1 << 27) -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_SHIFT 28 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_MASK 0x70000000 -#define NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE (1 << 31) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA 0x00000320 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR 0x00000324 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA 0x0000032c -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR 0x00000330 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE0 (1 << 0) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA0 (1 << 1) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_SHIFT 2 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_MASK 0x000000fc -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE1 (1 << 8) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA1 (1 << 9) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_SHIFT 10 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_MASK 0x0000fc00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE2 (1 << 16) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA2 (1 << 17) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_SHIFT 18 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_MASK 0x00fc0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE3 (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA3 (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_SHIFT 26 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_MASK 0x1c000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_SHIFT 29 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_MASK 0xe0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR 0x00000334 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND 0x00000338 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_MASK 0x00000030 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_SHIFT 6 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_MASK 0x000000c0 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT 12 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK 0x0000f000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_MASK 0x00f00000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK 0x0f000000 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT 28 -#define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK 0xf0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0 0x0000033c -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_TEST_ENABLE (1 << 12) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ORIGIN (1 << 13) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_SHIFT 14 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_MASK 0x0000c000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_MASK 0x00300000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_DITHER_ENABLE (1 << 22) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_PERSPECTIVE_ENABLE (1 << 23) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_WRITE_ENABLE (1 << 24) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_ENABLE (1 << 25) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_ENABLE (1 << 26) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_ENABLE (1 << 27) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_ENABLE (1 << 28) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_ENABLE (1 << 29) -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_SHIFT 30 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_MASK 0xc0000000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1 0x00000340 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_MASK 0x000000f0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2 0x00000344 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_MASK 0x000000f0 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR 0x00000348 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX(x) (0x00000400+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY(x) (0x00000404+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ(x) (0x00000408+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW(x) (0x0000040c+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR(x) (0x00000410+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR(x) (0x00000414+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_MASK 0x000000ff -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_MASK 0x0000ff00 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_MASK 0x00ff0000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_SHIFT 24 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_MASK 0xff000000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0(x) (0x00000418+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0(x) (0x0000041c+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1(x) (0x00000420+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1(x) (0x00000424+((x)*40)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1__SIZE 0x00000008 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE(x) (0x00000540+((x)*4)) -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE__SIZE 0x00000030 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_SHIFT 0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_MASK 0x0000000f -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_SHIFT 4 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_MASK 0x000000f0 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_SHIFT 8 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_MASK 0x00000f00 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_SHIFT 12 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_MASK 0x0000f000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_SHIFT 16 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_MASK 0x000f0000 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_SHIFT 20 -#define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 - - -#define NV10TCL 0x00000056 - -#define NV10TCL_NOP 0x00000100 -#define NV10TCL_NOTIFY 0x00000104 -#define NV10TCL_DMA_NOTIFY 0x00000180 -#define NV10TCL_DMA_IN_MEMORY0 0x00000184 -#define NV10TCL_DMA_IN_MEMORY1 0x00000188 -#define NV10TCL_DISPLAY_LIST 0x0000018c -#define NV10TCL_DMA_IN_MEMORY2 0x00000194 -#define NV10TCL_DMA_IN_MEMORY3 0x00000198 -#define NV10TCL_VIEWPORT_HORIZ 0x00000200 -#define NV10TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV10TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV10TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV10TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV10TCL_VIEWPORT_VERT 0x00000204 -#define NV10TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV10TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV10TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV10TCL_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV10TCL_BUFFER_FORMAT 0x00000208 -#define NV10TCL_BUFFER_PITCH 0x0000020c -#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 -#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff -#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 -#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 -#define NV10TCL_COLOR_OFFSET 0x00000210 -#define NV10TCL_ZETA_OFFSET 0x00000214 -#define NV10TCL_TX_OFFSET(x) (0x00000218+((x)*4)) -#define NV10TCL_TX_OFFSET__SIZE 0x00000002 -#define NV10TCL_TX_FORMAT(x) (0x00000220+((x)*4)) -#define NV10TCL_TX_FORMAT__SIZE 0x00000002 -#define NV10TCL_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV10TCL_TX_FORMAT_FORMAT_SHIFT 7 -#define NV10TCL_TX_FORMAT_FORMAT_MASK 0x00000780 -#define NV10TCL_TX_FORMAT_FORMAT_L8 0x00000000 -#define NV10TCL_TX_FORMAT_FORMAT_A8 0x00000080 -#define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000100 -#define NV10TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000180 -#define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000200 -#define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000300 -#define NV10TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000380 -#define NV10TCL_TX_FORMAT_FORMAT_INDEX8 0x00000580 -#define NV10TCL_TX_FORMAT_FORMAT_DXT1 0x00000600 -#define NV10TCL_TX_FORMAT_FORMAT_DXT3 0x00000700 -#define NV10TCL_TX_FORMAT_FORMAT_DXT5 0x00000780 -#define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00000800 -#define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00000900 -#define NV10TCL_TX_FORMAT_FORMAT_L8_RECT 0x00000980 -#define NV10TCL_TX_FORMAT_FORMAT_A8L8 0x00000d00 -#define NV10TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00000d80 -#define NV10TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 -#define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00000e80 -#define NV10TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00001000 -#define NV10TCL_TX_FORMAT_FORMAT_A16 0x00001900 -#define NV10TCL_TX_FORMAT_FORMAT_A16_RECT 0x00001a80 -#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00002500 -#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00002580 -#define NV10TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00002600 -#define NV10TCL_TX_FORMAT_NPOT (1 << 11) -#define NV10TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 12 -#define NV10TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x0000f000 -#define NV10TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 16 -#define NV10TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x000f0000 -#define NV10TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 20 -#define NV10TCL_TX_FORMAT_BASE_SIZE_V_MASK 0x00f00000 -#define NV10TCL_TX_FORMAT_WRAP_S_SHIFT 24 -#define NV10TCL_TX_FORMAT_WRAP_S_MASK 0x0f000000 -#define NV10TCL_TX_FORMAT_WRAP_S_REPEAT 0x01000000 -#define NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT 0x02000000 -#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE 0x03000000 -#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER 0x04000000 -#define NV10TCL_TX_FORMAT_WRAP_S_CLAMP 0x05000000 -#define NV10TCL_TX_FORMAT_WRAP_T_SHIFT 28 -#define NV10TCL_TX_FORMAT_WRAP_T_MASK 0xf0000000 -#define NV10TCL_TX_FORMAT_WRAP_T_REPEAT 0x10000000 -#define NV10TCL_TX_FORMAT_WRAP_T_MIRRORED_REPEAT 0x20000000 -#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE 0x30000000 -#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER 0x40000000 -#define NV10TCL_TX_FORMAT_WRAP_T_CLAMP 0x50000000 -#define NV10TCL_TX_ENABLE(x) (0x00000228+((x)*4)) -#define NV10TCL_TX_ENABLE__SIZE 0x00000002 -#define NV10TCL_TX_ENABLE_ANISOTROPY_SHIFT 4 -#define NV10TCL_TX_ENABLE_ANISOTROPY_MASK 0x00000030 -#define NV10TCL_TX_ENABLE_ENABLE (1 << 30) -#define NV10TCL_TX_NPOT_PITCH(x) (0x00000230+((x)*4)) -#define NV10TCL_TX_NPOT_PITCH__SIZE 0x00000002 -#define NV10TCL_TX_NPOT_PITCH_PITCH_SHIFT 16 -#define NV10TCL_TX_NPOT_PITCH_PITCH_MASK 0xffff0000 -#define NV10TCL_TX_NPOT_SIZE(x) (0x00000240+((x)*4)) -#define NV10TCL_TX_NPOT_SIZE__SIZE 0x00000002 -#define NV10TCL_TX_NPOT_SIZE_H_SHIFT 0 -#define NV10TCL_TX_NPOT_SIZE_H_MASK 0x0000ffff -#define NV10TCL_TX_NPOT_SIZE_W_SHIFT 16 -#define NV10TCL_TX_NPOT_SIZE_W_MASK 0xffff0000 -#define NV10TCL_TX_FILTER(x) (0x00000248+((x)*4)) -#define NV10TCL_TX_FILTER__SIZE 0x00000002 -#define NV10TCL_TX_FILTER_MINIFY_SHIFT 24 -#define NV10TCL_TX_FILTER_MINIFY_MASK 0x0f000000 -#define NV10TCL_TX_FILTER_MINIFY_NEAREST 0x01000000 -#define NV10TCL_TX_FILTER_MINIFY_LINEAR 0x02000000 -#define NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x03000000 -#define NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x04000000 -#define NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x05000000 -#define NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x06000000 -#define NV10TCL_TX_FILTER_MAGNIFY_SHIFT 28 -#define NV10TCL_TX_FILTER_MAGNIFY_MASK 0xf0000000 -#define NV10TCL_TX_FILTER_MAGNIFY_NEAREST 0x10000000 -#define NV10TCL_TX_FILTER_MAGNIFY_LINEAR 0x20000000 -#define NV10TCL_TX_PALETTE_OFFSET(x) (0x00000250+((x)*4)) -#define NV10TCL_TX_PALETTE_OFFSET__SIZE 0x00000002 -#define NV10TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) -#define NV10TCL_RC_IN_ALPHA__SIZE 0x00000002 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_SHIFT 0 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_MASK 0x0000000f -#define NV10TCL_RC_IN_ALPHA_D_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_FOG 0x00000003 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB 0x00000008 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB 0x00000009 -#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_NV 0x0000000c -#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE1_NV 0x0000000d -#define NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10TCL_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE (1 << 4) -#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA 0x00000010 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SHIFT 5 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_MASK 0x000000e0 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 -#define NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_SHIFT 8 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_MASK 0x00000f00 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_FOG 0x00000300 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_NV 0x00000c00 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE1_NV 0x00000d00 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10TCL_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE (1 << 12) -#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SHIFT 13 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_MASK 0x0000e000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_SHIFT 16 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_MASK 0x000f0000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_FOG 0x00030000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_NV 0x000c0000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE1_NV 0x000d0000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10TCL_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE (1 << 20) -#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SHIFT 21 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_MASK 0x00e00000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_SHIFT 24 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_MASK 0x0f000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_FOG 0x03000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_NV 0x0c000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE1_NV 0x0d000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10TCL_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE (1 << 28) -#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE 0x00000000 -#define NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SHIFT 29 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_MASK 0xe0000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10TCL_RC_IN_RGB(x) (0x00000268+((x)*4)) -#define NV10TCL_RC_IN_RGB__SIZE 0x00000002 -#define NV10TCL_RC_IN_RGB_D_INPUT_SHIFT 0 -#define NV10TCL_RC_IN_RGB_D_INPUT_MASK 0x0000000f -#define NV10TCL_RC_IN_RGB_D_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10TCL_RC_IN_RGB_D_INPUT_FOG 0x00000003 -#define NV10TCL_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10TCL_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE0_ARB 0x00000008 -#define NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE1_ARB 0x00000009 -#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_NV 0x0000000c -#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE1_NV 0x0000000d -#define NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10TCL_RC_IN_RGB_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE (1 << 4) -#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA 0x00000010 -#define NV10TCL_RC_IN_RGB_D_MAPPING_SHIFT 5 -#define NV10TCL_RC_IN_RGB_D_MAPPING_MASK 0x000000e0 -#define NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 -#define NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 -#define NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 -#define NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 -#define NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 -#define NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 -#define NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 -#define NV10TCL_RC_IN_RGB_C_INPUT_SHIFT 8 -#define NV10TCL_RC_IN_RGB_C_INPUT_MASK 0x00000f00 -#define NV10TCL_RC_IN_RGB_C_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10TCL_RC_IN_RGB_C_INPUT_FOG 0x00000300 -#define NV10TCL_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10TCL_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_NV 0x00000c00 -#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE1_NV 0x00000d00 -#define NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10TCL_RC_IN_RGB_C_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE (1 << 12) -#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_SHIFT 13 -#define NV10TCL_RC_IN_RGB_C_MAPPING_MASK 0x0000e000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10TCL_RC_IN_RGB_B_INPUT_SHIFT 16 -#define NV10TCL_RC_IN_RGB_B_INPUT_MASK 0x000f0000 -#define NV10TCL_RC_IN_RGB_B_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10TCL_RC_IN_RGB_B_INPUT_FOG 0x00030000 -#define NV10TCL_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10TCL_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_NV 0x000c0000 -#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE1_NV 0x000d0000 -#define NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10TCL_RC_IN_RGB_B_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE (1 << 20) -#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_SHIFT 21 -#define NV10TCL_RC_IN_RGB_B_MAPPING_MASK 0x00e00000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10TCL_RC_IN_RGB_A_INPUT_SHIFT 24 -#define NV10TCL_RC_IN_RGB_A_INPUT_MASK 0x0f000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_FOG 0x03000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_NV 0x0c000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE1_NV 0x0d000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10TCL_RC_IN_RGB_A_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE (1 << 28) -#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_SHIFT 29 -#define NV10TCL_RC_IN_RGB_A_MAPPING_MASK 0xe0000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10TCL_RC_COLOR(x) (0x00000270+((x)*4)) -#define NV10TCL_RC_COLOR__SIZE 0x00000002 -#define NV10TCL_RC_COLOR_B_SHIFT 0 -#define NV10TCL_RC_COLOR_B_MASK 0x000000ff -#define NV10TCL_RC_COLOR_G_SHIFT 8 -#define NV10TCL_RC_COLOR_G_MASK 0x0000ff00 -#define NV10TCL_RC_COLOR_R_SHIFT 16 -#define NV10TCL_RC_COLOR_R_MASK 0x00ff0000 -#define NV10TCL_RC_COLOR_A_SHIFT 24 -#define NV10TCL_RC_COLOR_A_MASK 0xff000000 -#define NV10TCL_RC_OUT_ALPHA(x) (0x00000278+((x)*4)) -#define NV10TCL_RC_OUT_ALPHA__SIZE 0x00000002 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SHIFT 0 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_MASK 0x0000000f -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_ZERO 0x00000000 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_FOG 0x00000003 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB 0x00000008 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB 0x00000009 -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV 0x0000000c -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV 0x0000000d -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SHIFT 4 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_MASK 0x000000f0 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_ZERO 0x00000000 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_FOG 0x00000030 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB 0x00000080 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB 0x00000090 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV 0x000000c0 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV 0x000000d0 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 -#define NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT 8 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_MASK 0x00000f00 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_ZERO 0x00000000 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_FOG 0x00000300 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV 0x00000c00 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV 0x00000d00 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 -#define NV10TCL_RC_OUT_ALPHA_CD_DOT_PRODUCT (1 << 12) -#define NV10TCL_RC_OUT_ALPHA_AB_DOT_PRODUCT (1 << 13) -#define NV10TCL_RC_OUT_ALPHA_MUX_SUM (1 << 14) -#define NV10TCL_RC_OUT_ALPHA_BIAS (1 << 15) -#define NV10TCL_RC_OUT_ALPHA_BIAS_NONE 0x00000000 -#define NV10TCL_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 -#define NV10TCL_RC_OUT_ALPHA_SCALE_SHIFT 17 -#define NV10TCL_RC_OUT_ALPHA_SCALE_MASK 0x00000000 -#define NV10TCL_RC_OUT_ALPHA_SCALE_NONE 0x00000000 -#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV 0x00020000 -#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV 0x00040000 -#define NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 -#define NV10TCL_RC_OUT_RGB(x) (0x00000280+((x)*4)) -#define NV10TCL_RC_OUT_RGB__SIZE 0x00000002 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SHIFT 0 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_MASK 0x0000000f -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_ZERO 0x00000000 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_FOG 0x00000003 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB 0x00000008 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB 0x00000009 -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV 0x0000000c -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV 0x0000000d -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10TCL_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV 0x0000000f -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SHIFT 4 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_MASK 0x000000f0 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_ZERO 0x00000000 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV 0x00000010 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV 0x00000020 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_FOG 0x00000030 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV 0x00000040 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV 0x00000050 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB 0x00000080 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB 0x00000090 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV 0x000000c0 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV 0x000000d0 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000000e0 -#define NV10TCL_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV 0x000000f0 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SHIFT 8 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_MASK 0x00000f00 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_ZERO 0x00000000 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_FOG 0x00000300 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB 0x00000800 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB 0x00000900 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV 0x00000c00 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV 0x00000d00 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10TCL_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV 0x00000f00 -#define NV10TCL_RC_OUT_RGB_CD_DOT_PRODUCT (1 << 12) -#define NV10TCL_RC_OUT_RGB_AB_DOT_PRODUCT (1 << 13) -#define NV10TCL_RC_OUT_RGB_MUX_SUM (1 << 14) -#define NV10TCL_RC_OUT_RGB_BIAS (1 << 15) -#define NV10TCL_RC_OUT_RGB_BIAS_NONE 0x00000000 -#define NV10TCL_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x00008000 -#define NV10TCL_RC_OUT_RGB_SCALE_SHIFT 17 -#define NV10TCL_RC_OUT_RGB_SCALE_MASK 0x00000000 -#define NV10TCL_RC_OUT_RGB_SCALE_NONE 0x00000000 -#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV 0x00020000 -#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV 0x00040000 -#define NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV 0x00060000 -#define NV10TCL_RC_OUT_RGB_OPERATION_SHIFT 27 -#define NV10TCL_RC_OUT_RGB_OPERATION_MASK 0x38000000 -#define NV10TCL_RC_FINAL0 0x00000288 -#define NV10TCL_RC_FINAL0_D_INPUT_SHIFT 0 -#define NV10TCL_RC_FINAL0_D_INPUT_MASK 0x0000000f -#define NV10TCL_RC_FINAL0_D_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV 0x00000001 -#define NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV 0x00000002 -#define NV10TCL_RC_FINAL0_D_INPUT_FOG 0x00000003 -#define NV10TCL_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV 0x00000004 -#define NV10TCL_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV 0x00000005 -#define NV10TCL_RC_FINAL0_D_INPUT_TEXTURE0_ARB 0x00000008 -#define NV10TCL_RC_FINAL0_D_INPUT_TEXTURE1_ARB 0x00000009 -#define NV10TCL_RC_FINAL0_D_INPUT_SPARE0_NV 0x0000000c -#define NV10TCL_RC_FINAL0_D_INPUT_SPARE1_NV 0x0000000d -#define NV10TCL_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0000000e -#define NV10TCL_RC_FINAL0_D_INPUT_E_TIMES_F_NV 0x0000000f -#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE (1 << 4) -#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_ALPHA 0x00000010 -#define NV10TCL_RC_FINAL0_D_MAPPING_SHIFT 5 -#define NV10TCL_RC_FINAL0_D_MAPPING_MASK 0x000000e0 -#define NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV 0x00000020 -#define NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV 0x00000040 -#define NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV 0x00000060 -#define NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV 0x00000080 -#define NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV 0x000000a0 -#define NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV 0x000000c0 -#define NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV 0x000000e0 -#define NV10TCL_RC_FINAL0_C_INPUT_SHIFT 8 -#define NV10TCL_RC_FINAL0_C_INPUT_MASK 0x00000f00 -#define NV10TCL_RC_FINAL0_C_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10TCL_RC_FINAL0_C_INPUT_FOG 0x00000300 -#define NV10TCL_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10TCL_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10TCL_RC_FINAL0_C_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10TCL_RC_FINAL0_C_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10TCL_RC_FINAL0_C_INPUT_SPARE0_NV 0x00000c00 -#define NV10TCL_RC_FINAL0_C_INPUT_SPARE1_NV 0x00000d00 -#define NV10TCL_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10TCL_RC_FINAL0_C_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE (1 << 12) -#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10TCL_RC_FINAL0_C_MAPPING_SHIFT 13 -#define NV10TCL_RC_FINAL0_C_MAPPING_MASK 0x0000e000 -#define NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10TCL_RC_FINAL0_B_INPUT_SHIFT 16 -#define NV10TCL_RC_FINAL0_B_INPUT_MASK 0x000f0000 -#define NV10TCL_RC_FINAL0_B_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10TCL_RC_FINAL0_B_INPUT_FOG 0x00030000 -#define NV10TCL_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10TCL_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10TCL_RC_FINAL0_B_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10TCL_RC_FINAL0_B_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10TCL_RC_FINAL0_B_INPUT_SPARE0_NV 0x000c0000 -#define NV10TCL_RC_FINAL0_B_INPUT_SPARE1_NV 0x000d0000 -#define NV10TCL_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10TCL_RC_FINAL0_B_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE (1 << 20) -#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10TCL_RC_FINAL0_B_MAPPING_SHIFT 21 -#define NV10TCL_RC_FINAL0_B_MAPPING_MASK 0x00e00000 -#define NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10TCL_RC_FINAL0_A_INPUT_SHIFT 24 -#define NV10TCL_RC_FINAL0_A_INPUT_MASK 0x0f000000 -#define NV10TCL_RC_FINAL0_A_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10TCL_RC_FINAL0_A_INPUT_FOG 0x03000000 -#define NV10TCL_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10TCL_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10TCL_RC_FINAL0_A_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10TCL_RC_FINAL0_A_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10TCL_RC_FINAL0_A_INPUT_SPARE0_NV 0x0c000000 -#define NV10TCL_RC_FINAL0_A_INPUT_SPARE1_NV 0x0d000000 -#define NV10TCL_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10TCL_RC_FINAL0_A_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE (1 << 28) -#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_SHIFT 29 -#define NV10TCL_RC_FINAL0_A_MAPPING_MASK 0xe0000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10TCL_RC_FINAL1 0x0000028c -#define NV10TCL_RC_FINAL1_COLOR_SUM_CLAMP (1 << 7) -#define NV10TCL_RC_FINAL1_G_INPUT_SHIFT 8 -#define NV10TCL_RC_FINAL1_G_INPUT_MASK 0x00000f00 -#define NV10TCL_RC_FINAL1_G_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV 0x00000100 -#define NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV 0x00000200 -#define NV10TCL_RC_FINAL1_G_INPUT_FOG 0x00000300 -#define NV10TCL_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV 0x00000400 -#define NV10TCL_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV 0x00000500 -#define NV10TCL_RC_FINAL1_G_INPUT_TEXTURE0_ARB 0x00000800 -#define NV10TCL_RC_FINAL1_G_INPUT_TEXTURE1_ARB 0x00000900 -#define NV10TCL_RC_FINAL1_G_INPUT_SPARE0_NV 0x00000c00 -#define NV10TCL_RC_FINAL1_G_INPUT_SPARE1_NV 0x00000d00 -#define NV10TCL_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x00000e00 -#define NV10TCL_RC_FINAL1_G_INPUT_E_TIMES_F_NV 0x00000f00 -#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE (1 << 12) -#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_ALPHA 0x00001000 -#define NV10TCL_RC_FINAL1_G_MAPPING_SHIFT 13 -#define NV10TCL_RC_FINAL1_G_MAPPING_MASK 0x0000e000 -#define NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV 0x00002000 -#define NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV 0x00004000 -#define NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV 0x00006000 -#define NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV 0x00008000 -#define NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV 0x0000a000 -#define NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV 0x0000c000 -#define NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV 0x0000e000 -#define NV10TCL_RC_FINAL1_F_INPUT_SHIFT 16 -#define NV10TCL_RC_FINAL1_F_INPUT_MASK 0x000f0000 -#define NV10TCL_RC_FINAL1_F_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV 0x00010000 -#define NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV 0x00020000 -#define NV10TCL_RC_FINAL1_F_INPUT_FOG 0x00030000 -#define NV10TCL_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV 0x00040000 -#define NV10TCL_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV 0x00050000 -#define NV10TCL_RC_FINAL1_F_INPUT_TEXTURE0_ARB 0x00080000 -#define NV10TCL_RC_FINAL1_F_INPUT_TEXTURE1_ARB 0x00090000 -#define NV10TCL_RC_FINAL1_F_INPUT_SPARE0_NV 0x000c0000 -#define NV10TCL_RC_FINAL1_F_INPUT_SPARE1_NV 0x000d0000 -#define NV10TCL_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x000e0000 -#define NV10TCL_RC_FINAL1_F_INPUT_E_TIMES_F_NV 0x000f0000 -#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE (1 << 20) -#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_ALPHA 0x00100000 -#define NV10TCL_RC_FINAL1_F_MAPPING_SHIFT 21 -#define NV10TCL_RC_FINAL1_F_MAPPING_MASK 0x00e00000 -#define NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV 0x00200000 -#define NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV 0x00400000 -#define NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV 0x00600000 -#define NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV 0x00800000 -#define NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV 0x00a00000 -#define NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV 0x00c00000 -#define NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV 0x00e00000 -#define NV10TCL_RC_FINAL1_E_INPUT_SHIFT 24 -#define NV10TCL_RC_FINAL1_E_INPUT_MASK 0x0f000000 -#define NV10TCL_RC_FINAL1_E_INPUT_ZERO 0x00000000 -#define NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV 0x01000000 -#define NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV 0x02000000 -#define NV10TCL_RC_FINAL1_E_INPUT_FOG 0x03000000 -#define NV10TCL_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV 0x04000000 -#define NV10TCL_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV 0x05000000 -#define NV10TCL_RC_FINAL1_E_INPUT_TEXTURE0_ARB 0x08000000 -#define NV10TCL_RC_FINAL1_E_INPUT_TEXTURE1_ARB 0x09000000 -#define NV10TCL_RC_FINAL1_E_INPUT_SPARE0_NV 0x0c000000 -#define NV10TCL_RC_FINAL1_E_INPUT_SPARE1_NV 0x0d000000 -#define NV10TCL_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV 0x0e000000 -#define NV10TCL_RC_FINAL1_E_INPUT_E_TIMES_F_NV 0x0f000000 -#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE (1 << 28) -#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_RGB 0x00000000 -#define NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_ALPHA 0x10000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_SHIFT 29 -#define NV10TCL_RC_FINAL1_E_MAPPING_MASK 0xe0000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV 0x00000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV 0x20000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV 0x40000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV 0x60000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV 0x80000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV 0xa0000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV 0xc0000000 -#define NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV 0xe0000000 -#define NV10TCL_LIGHT_MODEL 0x00000294 -#define NV10TCL_LIGHT_MODEL_COLOR_CONTROL (1 << 1) -#define NV10TCL_LIGHT_MODEL_LOCAL_VIEWER (1 << 16) -#define NV10TCL_COLOR_MATERIAL_ENABLE 0x00000298 -#define NV10TCL_COLOR_MATERIAL_ENABLE_SPECULAR (1 << 0) -#define NV10TCL_COLOR_MATERIAL_ENABLE_DIFFUSE (1 << 1) -#define NV10TCL_COLOR_MATERIAL_ENABLE_AMBIENT (1 << 2) -#define NV10TCL_COLOR_MATERIAL_ENABLE_EMISSION (1 << 3) -#define NV10TCL_FOG_MODE 0x0000029c -#define NV10TCL_FOG_MODE_EXP 0x00000800 -#define NV10TCL_FOG_MODE_EXP_2 0x00000802 -#define NV10TCL_FOG_MODE_EXP2 0x00000803 -#define NV10TCL_FOG_MODE_LINEAR 0x00000804 -#define NV10TCL_FOG_MODE_LINEAR_2 0x00002601 -#define NV10TCL_FOG_COORD_DIST 0x000002a0 -#define NV10TCL_FOG_ENABLE 0x000002a4 -#define NV10TCL_FOG_COLOR 0x000002a8 -#define NV10TCL_FOG_COLOR_R_SHIFT 0 -#define NV10TCL_FOG_COLOR_R_MASK 0x000000ff -#define NV10TCL_FOG_COLOR_G_SHIFT 8 -#define NV10TCL_FOG_COLOR_G_MASK 0x0000ff00 -#define NV10TCL_FOG_COLOR_B_SHIFT 16 -#define NV10TCL_FOG_COLOR_B_MASK 0x00ff0000 -#define NV10TCL_FOG_COLOR_A_SHIFT 24 -#define NV10TCL_FOG_COLOR_A_MASK 0xff000000 -#define NV10TCL_VIEWPORT_CLIP_MODE 0x000002b4 -#define NV10TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) -#define NV10TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT 0 -#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK 0x000007ff -#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE (1 << 11) -#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT 16 -#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK 0x07ff0000 -#define NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE (1 << 27) -#define NV10TCL_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) -#define NV10TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT 0 -#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_MASK 0x000007ff -#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE (1 << 11) -#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT 16 -#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_MASK 0x07ff0000 -#define NV10TCL_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE (1 << 27) -#define NV10TCL_ALPHA_FUNC_ENABLE 0x00000300 -#define NV10TCL_BLEND_FUNC_ENABLE 0x00000304 -#define NV10TCL_CULL_FACE_ENABLE 0x00000308 -#define NV10TCL_DEPTH_TEST_ENABLE 0x0000030c -#define NV10TCL_DITHER_ENABLE 0x00000310 -#define NV10TCL_LIGHTING_ENABLE 0x00000314 -#define NV10TCL_POINT_PARAMETERS_ENABLE 0x00000318 -#define NV10TCL_POINT_SMOOTH_ENABLE 0x0000031c -#define NV10TCL_LINE_SMOOTH_ENABLE 0x00000320 -#define NV10TCL_POLYGON_SMOOTH_ENABLE 0x00000324 -#define NV10TCL_VERTEX_WEIGHT_ENABLE 0x00000328 -#define NV10TCL_STENCIL_ENABLE 0x0000032c -#define NV10TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000330 -#define NV10TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000334 -#define NV10TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000338 -#define NV10TCL_ALPHA_FUNC_FUNC 0x0000033c -#define NV10TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV10TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV10TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV10TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV10TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV10TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV10TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV10TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV10TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV10TCL_ALPHA_FUNC_REF 0x00000340 -#define NV10TCL_BLEND_FUNC_SRC 0x00000344 -#define NV10TCL_BLEND_FUNC_SRC_ZERO 0x00000000 -#define NV10TCL_BLEND_FUNC_SRC_ONE 0x00000001 -#define NV10TCL_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 -#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 -#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV10TCL_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 -#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV10TCL_BLEND_FUNC_SRC_DST_COLOR 0x00000306 -#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 -#define NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 -#define NV10TCL_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 -#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV10TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 -#define NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV10TCL_BLEND_FUNC_DST 0x00000348 -#define NV10TCL_BLEND_FUNC_DST_ZERO 0x00000000 -#define NV10TCL_BLEND_FUNC_DST_ONE 0x00000001 -#define NV10TCL_BLEND_FUNC_DST_SRC_COLOR 0x00000300 -#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV10TCL_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 -#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV10TCL_BLEND_FUNC_DST_DST_ALPHA 0x00000304 -#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV10TCL_BLEND_FUNC_DST_DST_COLOR 0x00000306 -#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 -#define NV10TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 -#define NV10TCL_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 -#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV10TCL_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 -#define NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV10TCL_BLEND_COLOR 0x0000034c -#define NV10TCL_BLEND_COLOR_B_SHIFT 0 -#define NV10TCL_BLEND_COLOR_B_MASK 0x000000ff -#define NV10TCL_BLEND_COLOR_G_SHIFT 8 -#define NV10TCL_BLEND_COLOR_G_MASK 0x0000ff00 -#define NV10TCL_BLEND_COLOR_R_SHIFT 16 -#define NV10TCL_BLEND_COLOR_R_MASK 0x00ff0000 -#define NV10TCL_BLEND_COLOR_A_SHIFT 24 -#define NV10TCL_BLEND_COLOR_A_MASK 0xff000000 -#define NV10TCL_BLEND_EQUATION 0x00000350 -#define NV10TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 -#define NV10TCL_BLEND_EQUATION_MIN 0x00008007 -#define NV10TCL_BLEND_EQUATION_MAX 0x00008008 -#define NV10TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV10TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV10TCL_DEPTH_FUNC 0x00000354 -#define NV10TCL_DEPTH_FUNC_NEVER 0x00000200 -#define NV10TCL_DEPTH_FUNC_LESS 0x00000201 -#define NV10TCL_DEPTH_FUNC_EQUAL 0x00000202 -#define NV10TCL_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV10TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV10TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV10TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV10TCL_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV10TCL_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV10TCL_COLOR_MASK 0x00000358 -#define NV10TCL_COLOR_MASK_B (1 << 0) -#define NV10TCL_COLOR_MASK_G (1 << 8) -#define NV10TCL_COLOR_MASK_R (1 << 16) -#define NV10TCL_COLOR_MASK_A (1 << 24) -#define NV10TCL_DEPTH_WRITE_ENABLE 0x0000035c -#define NV10TCL_STENCIL_MASK 0x00000360 -#define NV10TCL_STENCIL_FUNC_FUNC 0x00000364 -#define NV10TCL_STENCIL_FUNC_FUNC_NEVER 0x00000200 -#define NV10TCL_STENCIL_FUNC_FUNC_LESS 0x00000201 -#define NV10TCL_STENCIL_FUNC_FUNC_EQUAL 0x00000202 -#define NV10TCL_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 -#define NV10TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV10TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV10TCL_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV10TCL_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 -#define NV10TCL_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 -#define NV10TCL_STENCIL_FUNC_REF 0x00000368 -#define NV10TCL_STENCIL_FUNC_MASK 0x0000036c -#define NV10TCL_STENCIL_OP_FAIL 0x00000370 -#define NV10TCL_STENCIL_OP_FAIL_ZERO 0x00000000 -#define NV10TCL_STENCIL_OP_FAIL_INVERT 0x0000150a -#define NV10TCL_STENCIL_OP_FAIL_KEEP 0x00001e00 -#define NV10TCL_STENCIL_OP_FAIL_REPLACE 0x00001e01 -#define NV10TCL_STENCIL_OP_FAIL_INCR 0x00001e02 -#define NV10TCL_STENCIL_OP_FAIL_DECR 0x00001e03 -#define NV10TCL_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 -#define NV10TCL_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 -#define NV10TCL_STENCIL_OP_ZFAIL 0x00000374 -#define NV10TCL_STENCIL_OP_ZFAIL_ZERO 0x00000000 -#define NV10TCL_STENCIL_OP_ZFAIL_INVERT 0x0000150a -#define NV10TCL_STENCIL_OP_ZFAIL_KEEP 0x00001e00 -#define NV10TCL_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 -#define NV10TCL_STENCIL_OP_ZFAIL_INCR 0x00001e02 -#define NV10TCL_STENCIL_OP_ZFAIL_DECR 0x00001e03 -#define NV10TCL_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV10TCL_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV10TCL_STENCIL_OP_ZPASS 0x00000378 -#define NV10TCL_STENCIL_OP_ZPASS_ZERO 0x00000000 -#define NV10TCL_STENCIL_OP_ZPASS_INVERT 0x0000150a -#define NV10TCL_STENCIL_OP_ZPASS_KEEP 0x00001e00 -#define NV10TCL_STENCIL_OP_ZPASS_REPLACE 0x00001e01 -#define NV10TCL_STENCIL_OP_ZPASS_INCR 0x00001e02 -#define NV10TCL_STENCIL_OP_ZPASS_DECR 0x00001e03 -#define NV10TCL_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV10TCL_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV10TCL_SHADE_MODEL 0x0000037c -#define NV10TCL_SHADE_MODEL_FLAT 0x00001d00 -#define NV10TCL_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV10TCL_LINE_WIDTH 0x00000380 -#define NV10TCL_POLYGON_OFFSET_FACTOR 0x00000384 -#define NV10TCL_POLYGON_OFFSET_UNITS 0x00000388 -#define NV10TCL_POLYGON_MODE_FRONT 0x0000038c -#define NV10TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV10TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV10TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV10TCL_POLYGON_MODE_BACK 0x00000390 -#define NV10TCL_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV10TCL_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV10TCL_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV10TCL_DEPTH_RANGE_NEAR 0x00000394 -#define NV10TCL_DEPTH_RANGE_FAR 0x00000398 -#define NV10TCL_CULL_FACE 0x0000039c -#define NV10TCL_CULL_FACE_FRONT 0x00000404 -#define NV10TCL_CULL_FACE_BACK 0x00000405 -#define NV10TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV10TCL_FRONT_FACE 0x000003a0 -#define NV10TCL_FRONT_FACE_CW 0x00000900 -#define NV10TCL_FRONT_FACE_CCW 0x00000901 -#define NV10TCL_NORMALIZE_ENABLE 0x000003a4 -#define NV10TCL_COLOR_MATERIAL_R 0x000003a8 -#define NV10TCL_COLOR_MATERIAL_G 0x000003ac -#define NV10TCL_COLOR_MATERIAL_B 0x000003b0 -#define NV10TCL_COLOR_MATERIAL_A 0x000003b4 -#define NV10TCL_COLOR_CONTROL 0x000003b8 -#define NV10TCL_ENABLED_LIGHTS 0x000003bc -#define NV10TCL_ENABLED_LIGHTS_LIGHT0 (1 << 0) -#define NV10TCL_ENABLED_LIGHTS_LIGHT1 (1 << 2) -#define NV10TCL_ENABLED_LIGHTS_LIGHT2 (1 << 4) -#define NV10TCL_ENABLED_LIGHTS_LIGHT3 (1 << 6) -#define NV10TCL_ENABLED_LIGHTS_LIGHT4 (1 << 8) -#define NV10TCL_ENABLED_LIGHTS_LIGHT5 (1 << 10) -#define NV10TCL_ENABLED_LIGHTS_LIGHT6 (1 << 12) -#define NV10TCL_ENABLED_LIGHTS_LIGHT7 (1 << 14) -#define NV10TCL_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) -#define NV10TCL_CLIP_PLANE_ENABLE__SIZE 0x00000008 -#define NV10TCL_CLIP_PLANE_ENABLE_FALSE 0x00000000 -#define NV10TCL_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 -#define NV10TCL_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 -#define NV10TCL_TX_MATRIX_ENABLE(x) (0x000003e0+((x)*4)) -#define NV10TCL_TX_MATRIX_ENABLE__SIZE 0x00000002 -#define NV10TCL_VIEW_MATRIX_ENABLE 0x000003e8 -#define NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW1 (1 << 0) -#define NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW0 (1 << 1) -#define NV10TCL_VIEW_MATRIX_ENABLE_PROJECTION (1 << 2) -#define NV10TCL_POINT_SIZE 0x000003ec -#define NV10TCL_MODELVIEW0_MATRIX(x) (0x00000400+((x)*4)) -#define NV10TCL_MODELVIEW0_MATRIX__SIZE 0x00000010 -#define NV10TCL_MODELVIEW1_MATRIX(x) (0x00000440+((x)*4)) -#define NV10TCL_MODELVIEW1_MATRIX__SIZE 0x00000010 -#define NV10TCL_INVERSE_MODELVIEW0_MATRIX(x) (0x00000480+((x)*4)) -#define NV10TCL_INVERSE_MODELVIEW0_MATRIX__SIZE 0x00000010 -#define NV10TCL_INVERSE_MODELVIEW1_MATRIX(x) (0x000004c0+((x)*4)) -#define NV10TCL_INVERSE_MODELVIEW1_MATRIX__SIZE 0x00000010 -#define NV10TCL_PROJECTION_MATRIX(x) (0x00000500+((x)*4)) -#define NV10TCL_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV10TCL_TX0_MATRIX(x) (0x00000540+((x)*4)) -#define NV10TCL_TX0_MATRIX__SIZE 0x00000010 -#define NV10TCL_TX1_MATRIX(x) (0x00000580+((x)*4)) -#define NV10TCL_TX1_MATRIX__SIZE 0x00000010 -#define NV10TCL_CLIP_PLANE_A(x) (0x00000600+((x)*16)) -#define NV10TCL_CLIP_PLANE_A__SIZE 0x00000008 -#define NV10TCL_CLIP_PLANE_B(x) (0x00000604+((x)*16)) -#define NV10TCL_CLIP_PLANE_B__SIZE 0x00000008 -#define NV10TCL_CLIP_PLANE_C(x) (0x00000608+((x)*16)) -#define NV10TCL_CLIP_PLANE_C__SIZE 0x00000008 -#define NV10TCL_CLIP_PLANE_D(x) (0x0000060c+((x)*16)) -#define NV10TCL_CLIP_PLANE_D__SIZE 0x00000008 -#define NV10TCL_FOG_EQUATION_CONSTANT 0x00000680 -#define NV10TCL_FOG_EQUATION_LINEAR 0x00000684 -#define NV10TCL_FOG_EQUATION_QUADRATIC 0x00000688 -#define NV10TCL_FRONT_MATERIAL_SHININESS(x) (0x000006a0+((x)*4)) -#define NV10TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000006c4 -#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000006c8 -#define NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000006cc -#define NV10TCL_VIEWPORT_SCALE_X 0x000006e8 -#define NV10TCL_VIEWPORT_SCALE_Y 0x000006ec -#define NV10TCL_VIEWPORT_SCALE_Z 0x000006f0 -#define NV10TCL_VIEWPORT_SCALE_W 0x000006f4 -#define NV10TCL_POINT_PARAMETER(x) (0x000006f8+((x)*4)) -#define NV10TCL_POINT_PARAMETER__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00000800+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00000804+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00000808+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000080c+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00000810+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00000814+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00000818+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000081c+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00000820+((x)*128)) -#define NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 -#define NV10TCL_LIGHT_HALF_VECTOR_X(x) (0x00000828+((x)*128)) -#define NV10TCL_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 -#define NV10TCL_LIGHT_HALF_VECTOR_Y(x) (0x0000082c+((x)*128)) -#define NV10TCL_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 -#define NV10TCL_LIGHT_HALF_VECTOR_Z(x) (0x00000830+((x)*128)) -#define NV10TCL_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 -#define NV10TCL_LIGHT_DIRECTION_X(x) (0x00000834+((x)*128)) -#define NV10TCL_LIGHT_DIRECTION_X__SIZE 0x00000008 -#define NV10TCL_LIGHT_DIRECTION_Y(x) (0x00000838+((x)*128)) -#define NV10TCL_LIGHT_DIRECTION_Y__SIZE 0x00000008 -#define NV10TCL_LIGHT_DIRECTION_Z(x) (0x0000083c+((x)*128)) -#define NV10TCL_LIGHT_DIRECTION_Z__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_CUTOFF_A(x) (0x00000840+((x)*128)) -#define NV10TCL_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_CUTOFF_B(x) (0x00000844+((x)*128)) -#define NV10TCL_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_CUTOFF_C(x) (0x00000848+((x)*128)) -#define NV10TCL_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_DIR_X(x) (0x0000084c+((x)*128)) -#define NV10TCL_LIGHT_SPOT_DIR_X__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_DIR_Y(x) (0x00000850+((x)*128)) -#define NV10TCL_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_DIR_Z(x) (0x00000854+((x)*128)) -#define NV10TCL_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 -#define NV10TCL_LIGHT_SPOT_CUTOFF_D(x) (0x00000858+((x)*128)) -#define NV10TCL_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 -#define NV10TCL_LIGHT_POSITION_X(x) (0x0000085c+((x)*128)) -#define NV10TCL_LIGHT_POSITION_X__SIZE 0x00000008 -#define NV10TCL_LIGHT_POSITION_Y(x) (0x00000860+((x)*128)) -#define NV10TCL_LIGHT_POSITION_Y__SIZE 0x00000008 -#define NV10TCL_LIGHT_POSITION_Z(x) (0x00000864+((x)*128)) -#define NV10TCL_LIGHT_POSITION_Z__SIZE 0x00000008 -#define NV10TCL_LIGHT_ATTENUATION_CONSTANT(x) (0x00000868+((x)*128)) -#define NV10TCL_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 -#define NV10TCL_LIGHT_ATTENUATION_LINEAR(x) (0x0000086c+((x)*128)) -#define NV10TCL_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 -#define NV10TCL_LIGHT_ATTENUATION_QUADRATIC(x) (0x00000870+((x)*128)) -#define NV10TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 -#define NV10TCL_VERTEX_POS_3F_X 0x00000c00 -#define NV10TCL_VERTEX_POS_3F_Y 0x00000c04 -#define NV10TCL_VERTEX_POS_3F_Z 0x00000c08 -#define NV10TCL_VERTEX_POS_4F_X 0x00000c18 -#define NV10TCL_VERTEX_POS_4F_Y 0x00000c1c -#define NV10TCL_VERTEX_POS_4F_Z 0x00000c20 -#define NV10TCL_VERTEX_POS_4F_W 0x00000c24 -#define NV10TCL_VERTEX_NOR_3F_X 0x00000c30 -#define NV10TCL_VERTEX_NOR_3F_Y 0x00000c34 -#define NV10TCL_VERTEX_NOR_3F_Z 0x00000c38 -#define NV10TCL_VERTEX_NOR_3I_XY 0x00000c40 -#define NV10TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV10TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV10TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV10TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV10TCL_VERTEX_NOR_3I_Z 0x00000c44 -#define NV10TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV10TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV10TCL_VERTEX_COL_4F_R 0x00000c50 -#define NV10TCL_VERTEX_COL_4F_G 0x00000c54 -#define NV10TCL_VERTEX_COL_4F_B 0x00000c58 -#define NV10TCL_VERTEX_COL_4F_A 0x00000c5c -#define NV10TCL_VERTEX_COL_3F_R 0x00000c60 -#define NV10TCL_VERTEX_COL_3F_G 0x00000c64 -#define NV10TCL_VERTEX_COL_3F_B 0x00000c68 -#define NV10TCL_VERTEX_COL_4I 0x00000c6c -#define NV10TCL_VERTEX_COL_4I_R_SHIFT 0 -#define NV10TCL_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV10TCL_VERTEX_COL_4I_G_SHIFT 8 -#define NV10TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV10TCL_VERTEX_COL_4I_B_SHIFT 16 -#define NV10TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV10TCL_VERTEX_COL_4I_A_SHIFT 24 -#define NV10TCL_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV10TCL_VERTEX_COL2_3F_R 0x00000c80 -#define NV10TCL_VERTEX_COL2_3F_G 0x00000c84 -#define NV10TCL_VERTEX_COL2_3F_B 0x00000c88 -#define NV10TCL_VERTEX_COL2_3I 0x00000c8c -#define NV10TCL_VERTEX_COL2_3I_R_SHIFT 0 -#define NV10TCL_VERTEX_COL2_3I_R_MASK 0x000000ff -#define NV10TCL_VERTEX_COL2_3I_G_SHIFT 8 -#define NV10TCL_VERTEX_COL2_3I_G_MASK 0x0000ff00 -#define NV10TCL_VERTEX_COL2_3I_B_SHIFT 16 -#define NV10TCL_VERTEX_COL2_3I_B_MASK 0x00ff0000 -#define NV10TCL_VERTEX_TX0_2F_S 0x00000c90 -#define NV10TCL_VERTEX_TX0_2F_T 0x00000c94 -#define NV10TCL_VERTEX_TX0_2I 0x00000c98 -#define NV10TCL_VERTEX_TX0_2I_S_SHIFT 0 -#define NV10TCL_VERTEX_TX0_2I_S_MASK 0x0000ffff -#define NV10TCL_VERTEX_TX0_2I_T_SHIFT 16 -#define NV10TCL_VERTEX_TX0_2I_T_MASK 0xffff0000 -#define NV10TCL_VERTEX_TX0_4F_S 0x00000ca0 -#define NV10TCL_VERTEX_TX0_4F_T 0x00000ca4 -#define NV10TCL_VERTEX_TX0_4F_R 0x00000ca8 -#define NV10TCL_VERTEX_TX0_4F_Q 0x00000cac -#define NV10TCL_VERTEX_TX0_4I_ST 0x00000cb0 -#define NV10TCL_VERTEX_TX0_4I_ST_S_SHIFT 0 -#define NV10TCL_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff -#define NV10TCL_VERTEX_TX0_4I_ST_T_SHIFT 16 -#define NV10TCL_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 -#define NV10TCL_VERTEX_TX0_4I_RQ 0x00000cb4 -#define NV10TCL_VERTEX_TX0_4I_RQ_R_SHIFT 0 -#define NV10TCL_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff -#define NV10TCL_VERTEX_TX0_4I_RQ_Q_SHIFT 16 -#define NV10TCL_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 -#define NV10TCL_VERTEX_TX1_2F_S 0x00000cb8 -#define NV10TCL_VERTEX_TX1_2F_T 0x00000cbc -#define NV10TCL_VERTEX_TX1_2I 0x00000cc0 -#define NV10TCL_VERTEX_TX1_2I_S_SHIFT 0 -#define NV10TCL_VERTEX_TX1_2I_S_MASK 0x0000ffff -#define NV10TCL_VERTEX_TX1_2I_T_SHIFT 16 -#define NV10TCL_VERTEX_TX1_2I_T_MASK 0xffff0000 -#define NV10TCL_VERTEX_TX1_4F_S 0x00000cc8 -#define NV10TCL_VERTEX_TX1_4F_T 0x00000ccc -#define NV10TCL_VERTEX_TX1_4F_R 0x00000cd0 -#define NV10TCL_VERTEX_TX1_4F_Q 0x00000cd4 -#define NV10TCL_VERTEX_TX1_4I_ST 0x00000cd8 -#define NV10TCL_VERTEX_TX1_4I_ST_S_SHIFT 0 -#define NV10TCL_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff -#define NV10TCL_VERTEX_TX1_4I_ST_T_SHIFT 16 -#define NV10TCL_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 -#define NV10TCL_VERTEX_TX1_4I_RQ 0x00000cdc -#define NV10TCL_VERTEX_TX1_4I_RQ_R_SHIFT 0 -#define NV10TCL_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff -#define NV10TCL_VERTEX_TX1_4I_RQ_Q_SHIFT 16 -#define NV10TCL_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 -#define NV10TCL_VERTEX_FOG_1F 0x00000ce0 -#define NV10TCL_VERTEX_WGH_1F 0x00000ce4 -#define NV10TCL_EDGEFLAG_ENABLE 0x00000cec -#define NV10TCL_VERTEX_ARRAY_VALIDATE 0x00000cf0 -#define NV10TCL_VERTEX_ARRAY_OFFSET_POS 0x00000d00 -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS 0x00000d04 -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_COL 0x00000d08 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL 0x00000d0c -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_COL2 0x00000d10 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2 0x00000d14 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_TX0 0x00000d18 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0 0x00000d1c -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_TX1 0x00000d20 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1 0x00000d24 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_NOR 0x00000d28 -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR 0x00000d2c -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_WGH 0x00000d30 -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH 0x00000d34 -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_ARRAY_OFFSET_FOG 0x00000d38 -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG 0x00000d3c -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT 0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK 0x0000000f -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT 4 -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK 0x000000f0 -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT 8 -#define NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK 0x0000ff00 -#define NV10TCL_VERTEX_BEGIN_END 0x00000dfc -#define NV10TCL_VERTEX_BEGIN_END_STOP 0x00000000 -#define NV10TCL_VERTEX_BEGIN_END_POINTS 0x00000001 -#define NV10TCL_VERTEX_BEGIN_END_LINES 0x00000002 -#define NV10TCL_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 -#define NV10TCL_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 -#define NV10TCL_VERTEX_BEGIN_END_TRIANGLES 0x00000005 -#define NV10TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV10TCL_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV10TCL_VERTEX_BEGIN_END_QUADS 0x00000008 -#define NV10TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV10TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV10TCL_DRAW_INDEX 0x00000e00 -#define NV10TCL_DRAW_INDEX_I0_SHIFT 0 -#define NV10TCL_DRAW_INDEX_I0_MASK 0x0000ffff -#define NV10TCL_DRAW_INDEX_I1_SHIFT 24 -#define NV10TCL_DRAW_INDEX_I1_MASK 0xff000000 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END 0x000013fc -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINES 0x00000002 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_LOOP 0x00000003 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_STRIP 0x00000004 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLES 0x00000005 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_QUADS 0x00000008 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV10TCL_VERTEX_BUFFER_BEGIN_END_POLYGON 0x0000000a -#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS 0x00001400 -#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT 0 -#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK 0x0000ffff -#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT 24 -#define NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK 0xff000000 -#define NV10TCL_VERTEX_ARRAY_DATA 0x00001800 - - -#define NV04_CONTEXT_COLOR_KEY 0x00000057 - - - -#define NV03_CONTEXT_SURFACES_2D 0x00000058 - -#define NV03_CONTEXT_SURFACES_2D_SYNCHRONIZE 0x00000100 -#define NV03_CONTEXT_SURFACES_2D_DMA_NOTIFY 0x00000180 -#define NV03_CONTEXT_SURFACES_2D_DMA_SOURCE 0x00000184 -#define NV03_CONTEXT_SURFACES_2D_DMA_DESTIN 0x00000188 -#define NV03_CONTEXT_SURFACES_2D_COLOR_FORMAT 0x00000300 -#define NV03_CONTEXT_SURFACES_2D_PITCH 0x00000304 -#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT 0 -#define NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK 0x0000ffff -#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT 16 -#define NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK 0xffff0000 -#define NV03_CONTEXT_SURFACES_2D_OFFSET_SOURCE 0x00000308 -#define NV03_CONTEXT_SURFACES_2D_OFFSET_DESTIN 0x0000030c - - -#define NV03_CONTEXT_SURFACES_3D 0x0000005a - -#define NV03_CONTEXT_SURFACES_3D_SYNCHRONIZE 0x00000100 -#define NV03_CONTEXT_SURFACES_3D_DMA_NOTIFY 0x00000180 -#define NV03_CONTEXT_SURFACES_3D_DMA_SURFACE 0x00000184 -#define NV03_CONTEXT_SURFACES_3D_PITCH 0x00000300 -#define NV03_CONTEXT_SURFACES_3D_OFFSET_COLOR 0x00000304 -#define NV03_CONTEXT_SURFACES_3D_OFFSET_ZETA 0x00000308 - - -#define NV04_RENDER_SOLID_LINE 0x0000005c - -#define NV04_RENDER_SOLID_LINE_SURFACE 0x00000198 - - -#define NV04_RENDER_SOLID_TRIANGLE 0x0000005d - - - -#define NV04_RENDER_SOLID_RECTANGLE 0x0000005e - -#define NV04_RENDER_SOLID_RECTANGLE_SURFACE 0x00000198 - - -#define NV04_IMAGE_BLIT 0x0000005f - -#define NV04_IMAGE_BLIT_NOP 0x00000100 -#define NV04_IMAGE_BLIT_NOTIFY 0x00000104 -#define NV04_IMAGE_BLIT_DMA_NOTIFY 0x00000180 -#define NV04_IMAGE_BLIT_COLOR_KEY 0x00000184 -#define NV04_IMAGE_BLIT_CLIP_RECTANGLE 0x00000188 -#define NV04_IMAGE_BLIT_PATTERN 0x0000018c -#define NV04_IMAGE_BLIT_ROP 0x00000190 -#define NV04_IMAGE_BLIT_BETA4 0x00000198 -#define NV04_IMAGE_BLIT_SURFACE 0x0000019c -#define NV04_IMAGE_BLIT_OPERATION 0x000002fc -#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY_AND 0x00000000 -#define NV04_IMAGE_BLIT_OPERATION_ROP_AND 0x00000001 -#define NV04_IMAGE_BLIT_OPERATION_BLEND_AND 0x00000002 -#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY 0x00000003 -#define NV04_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV04_IMAGE_BLIT_OPERATION_BLEND_PREMULT 0x00000005 - - -#define NV04_INDEXED_IMAGE_FROM_CPU 0x00000060 - -#define NV04_INDEXED_IMAGE_FROM_CPU_NOP 0x00000100 -#define NV04_INDEXED_IMAGE_FROM_CPU_NOTIFY 0x00000104 -#define NV04_INDEXED_IMAGE_FROM_CPU_PATCH 0x0000010c -#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV04_INDEXED_IMAGE_FROM_CPU_DMA_LUT 0x00000184 -#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT 0x000003e8 -#define NV04_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT 0x000003ec -#define NV04_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET 0x000003f0 -#define NV04_INDEXED_IMAGE_FROM_CPU_POINT 0x000003f4 -#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_OUT 0x000003f8 -#define NV04_INDEXED_IMAGE_FROM_CPU_SIZE_IN 0x000003fc -#define NV04_INDEXED_IMAGE_FROM_CPU_COLOR 0x00000400 - - -#define NV04_IMAGE_FROM_CPU 0x00000061 - -#define NV04_IMAGE_FROM_CPU_BETA4 0x00000198 -#define NV04_IMAGE_FROM_CPU_SURFACE 0x0000019c - - -#define NV10_CONTEXT_SURFACES_2D 0x00000062 - - - -#define NV05_SCALED_IMAGE_FROM_MEMORY 0x00000063 - -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 -#define NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 - - -#define NV01_IMAGE_SRCCOPY_AND 0x00000064 - -#define NV01_IMAGE_SRCCOPY_AND_NOTIFY 0x00000104 -#define NV01_IMAGE_SRCCOPY_AND_DMA_NOTIFY 0x00000180 -#define NV01_IMAGE_SRCCOPY_AND_IMAGE_OUTPUT 0x00000200 -#define NV01_IMAGE_SRCCOPY_AND_IMAGE_INPUT 0x00000204 - - -#define NV05_INDEXED_IMAGE_FROM_CPU 0x00000064 - -#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_KEY 0x00000188 -#define NV05_INDEXED_IMAGE_FROM_CPU_CLIP_RECTANGLE 0x0000018c -#define NV05_INDEXED_IMAGE_FROM_CPU_PATTERN 0x00000190 -#define NV05_INDEXED_IMAGE_FROM_CPU_ROP 0x00000194 -#define NV05_INDEXED_IMAGE_FROM_CPU_BETA1 0x00000198 -#define NV05_INDEXED_IMAGE_FROM_CPU_BETA4 0x0000019c -#define NV05_INDEXED_IMAGE_FROM_CPU_SURFACE 0x000001a0 -#define NV05_INDEXED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000003e0 -#define NV05_INDEXED_IMAGE_FROM_CPU_OPERATION 0x000003e4 -#define NV05_INDEXED_IMAGE_FROM_CPU_INDICES 0x00000400 - - -#define NV05_IMAGE_FROM_CPU 0x00000065 - -#define NV05_IMAGE_FROM_CPU_BETA4 0x00000198 -#define NV05_IMAGE_FROM_CPU_SURFACE 0x0000019c - - -#define NV05_STRETCHED_IMAGE_FROM_CPU 0x00000066 - -#define NV05_STRETCHED_IMAGE_FROM_CPU_BETA4 0x00000194 -#define NV05_STRETCHED_IMAGE_FROM_CPU_SURFACE 0x00000198 -#define NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 - - -#define NV04_IMAGE_BLEND_PREMULT 0x00000067 - -#define NV04_IMAGE_BLEND_PREMULT_NOP 0x00000100 -#define NV04_IMAGE_BLEND_PREMULT_NOTIFY 0x00000104 -#define NV04_IMAGE_BLEND_PREMULT_DMA_NOTIFY 0x00000180 -#define NV04_IMAGE_BLEND_PREMULT_IMAGE_OUTPUT 0x00000200 -#define NV04_IMAGE_BLEND_PREMULT_BETA_INPUT 0x00000204 -#define NV04_IMAGE_BLEND_PREMULT_IMAGE_INPUT 0x00000208 - - -#define NV03_CHANNEL_PIO 0x0000006a - - - -#define NV03_CHANNEL_DMA 0x0000006b - - - -#define NV04_BETA_SOLID 0x00000072 - -#define NV04_BETA_SOLID_NOP 0x00000100 -#define NV04_BETA_SOLID_NOTIFY 0x00000104 -#define NV04_BETA_SOLID_DMA_NOTIFY 0x00000180 -#define NV04_BETA_SOLID_BETA_OUTPUT 0x00000200 -#define NV04_BETA_SOLID_BETA_FACTOR 0x00000300 - - -#define NV04_STRETCHED_IMAGE_FROM_CPU 0x00000076 - - - -#define NV04_SCALED_IMAGE_FROM_MEMORY 0x00000077 - -#define NV04_SCALED_IMAGE_FROM_MEMORY_NOP 0x00000100 -#define NV04_SCALED_IMAGE_FROM_MEMORY_NOTIFY 0x00000104 -#define NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY 0x00000180 -#define NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE 0x00000184 -#define NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN 0x00000188 -#define NV04_SCALED_IMAGE_FROM_MEMORY_ROP 0x0000018c -#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA1 0x00000190 -#define NV04_SCALED_IMAGE_FROM_MEMORY_BETA4 0x00000194 -#define NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE 0x00000198 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION 0x000002fc -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER 0x00000000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE 0x00000001 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE 0x00000002 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT 0x00000300 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5 0x00000001 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5 0x00000002 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8 0x00000003 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8 0x00000004 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8 0x00000005 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8 0x00000006 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5 0x00000007 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8 0x00000008 -#define NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8 0x00000009 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION 0x00000304 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND 0x00000000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND 0x00000001 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND 0x00000002 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY 0x00000003 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT 0x00000005 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT 0x00000308 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK 0xffff0000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE 0x0000030c -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK 0xffff0000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT 0x00000310 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_MASK 0xffff0000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE 0x00000314 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_MASK 0xffff0000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX 0x00000318 -#define NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY 0x0000031c -#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE 0x00000400 -#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_MASK 0xffff0000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT 0x00000404 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_MASK 0x00ff0000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER 0x00010000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CORNER 0x00020000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_SHIFT 24 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_MASK 0xff000000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE 0x00000000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_BILINEAR 0x01000000 -#define NV04_SCALED_IMAGE_FROM_MEMORY_ADDRESS 0x00000408 -#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT 0x0000040c -#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_SHIFT 0 -#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_MASK 0x0000ffff -#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_SHIFT 16 -#define NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_MASK 0xffff0000 - - -#define NV10_TEXTURE_FROM_CPU 0x0000007b - -#define NV10_TEXTURE_FROM_CPU_NOP 0x00000100 -#define NV10_TEXTURE_FROM_CPU_NOTIFY 0x00000104 -#define NV10_TEXTURE_FROM_CPU_WAIT_FOR_IDLE 0x00000108 -#define NV10_TEXTURE_FROM_CPU_PM_TRIGGER 0x00000140 -#define NV10_TEXTURE_FROM_CPU_DMA_NOTIFY 0x00000180 -#define NV10_TEXTURE_FROM_CPU_SURFACE 0x00000184 -#define NV10_TEXTURE_FROM_CPU_COLOR_FORMAT 0x00000300 -#define NV10_TEXTURE_FROM_CPU_POINT 0x00000304 -#define NV10_TEXTURE_FROM_CPU_POINT_X_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_POINT_X_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_POINT_Y_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_POINT_Y_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_SIZE 0x00000308 -#define NV10_TEXTURE_FROM_CPU_SIZE_W_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_SIZE_W_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_SIZE_H_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_SIZE_H_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL 0x0000030c -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL 0x00000310 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_SHIFT 0 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_MASK 0x0000ffff -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_SHIFT 16 -#define NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_MASK 0xffff0000 -#define NV10_TEXTURE_FROM_CPU_COLOR(x) (0x00000400+((x)*4)) -#define NV10_TEXTURE_FROM_CPU_COLOR__SIZE 0x00000700 - - -#define NV10_VIDEO_DISPLAY 0x0000007c - - - -#define NV10_DVD_SUBPICTURE 0x00000088 - - - -#define NV10_SCALED_IMAGE_FROM_MEMORY 0x00000089 - -#define NV10_SCALED_IMAGE_FROM_MEMORY_WAIT_FOR_IDLE 0x00000108 - - -#define NV10_IMAGE_FROM_CPU 0x0000008a - -#define NV10_IMAGE_FROM_CPU_COLOR_CONVERSION 0x000002f8 - - -#define NV10_CONTEXT_SURFACES_3D 0x00000093 - - - -#define NV10_DX5_TEXTURE_TRIANGLE 0x00000094 - - - -#define NV10_DX6_MULTI_TEXTURE_TRIANGLE 0x00000095 - - - -#define NV11TCL 0x00000096 - -#define NV11TCL_COLOR_LOGIC_OP_ENABLE 0x00000d40 -#define NV11TCL_COLOR_LOGIC_OP_OP 0x00000d44 -#define NV11TCL_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV11TCL_COLOR_LOGIC_OP_OP_AND 0x00001501 -#define NV11TCL_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV11TCL_COLOR_LOGIC_OP_OP_COPY 0x00001503 -#define NV11TCL_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV11TCL_COLOR_LOGIC_OP_OP_NOOP 0x00001505 -#define NV11TCL_COLOR_LOGIC_OP_OP_XOR 0x00001506 -#define NV11TCL_COLOR_LOGIC_OP_OP_OR 0x00001507 -#define NV11TCL_COLOR_LOGIC_OP_OP_NOR 0x00001508 -#define NV11TCL_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV11TCL_COLOR_LOGIC_OP_OP_INVERT 0x0000150a -#define NV11TCL_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV11TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV11TCL_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV11TCL_COLOR_LOGIC_OP_OP_NAND 0x0000150e -#define NV11TCL_COLOR_LOGIC_OP_OP_SET 0x0000150f - - -#define NV20TCL 0x00000097 - -#define NV20TCL_NOP 0x00000100 -#define NV20TCL_NOTIFY 0x00000104 -#define NV20TCL_DMA_NOTIFY 0x00000180 -#define NV20TCL_DMA_IN_MEMORY0 0x00000184 -#define NV20TCL_DMA_IN_MEMORY1 0x00000188 -#define NV20TCL_DMA_IN_MEMORY2 0x00000194 -#define NV20TCL_DMA_IN_MEMORY3 0x00000198 -#define NV20TCL_DMA_IN_MEMORY6 0x000001a4 -#define NV20TCL_DMA_IN_MEMORY7 0x000001a8 -#define NV20TCL_VIEWPORT_HORIZ 0x00000200 -#define NV20TCL_VIEWPORT_VERT 0x00000204 -#define NV20TCL_BUFFER_FORMAT 0x00000208 -#define NV20TCL_BUFFER_PITCH 0x0000020c -#define NV20TCL_COLOR_OFFSET 0x00000210 -#define NV20TCL_ZETA_OFFSET 0x00000214 -#define NV20TCL_RC_IN_ALPHA(x) (0x00000260+((x)*4)) -#define NV20TCL_RC_IN_ALPHA__SIZE 0x00000008 -#define NV20TCL_RC_FINAL0 0x00000288 -#define NV20TCL_RC_FINAL1 0x0000028c -#define NV20TCL_LIGHT_CONTROL 0x00000294 -#define NV20TCL_FOG_MODE 0x0000029c -#define NV20TCL_FOG_COORD_DIST 0x000002a0 -#define NV20TCL_FOG_ENABLE 0x000002a4 -#define NV20TCL_FOG_COLOR 0x000002a8 -#define NV20TCL_VIEWPORT_CLIP_MODE 0x000002b4 -#define NV20TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*4)) -#define NV20TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV20TCL_VIEWPORT_CLIP_VERT(x) (0x000002e0+((x)*4)) -#define NV20TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV20TCL_ALPHA_FUNC_ENABLE 0x00000300 -#define NV20TCL_BLEND_FUNC_ENABLE 0x00000304 -#define NV20TCL_CULL_FACE_ENABLE 0x00000308 -#define NV20TCL_DEPTH_TEST_ENABLE 0x0000030c -#define NV20TCL_DITHER_ENABLE 0x00000310 -#define NV20TCL_LIGHTING_ENABLE 0x00000314 -#define NV20TCL_POINT_PARAMETERS_ENABLE 0x00000318 -#define NV20TCL_LINE_SMOOTH_ENABLE 0x00000320 -#define NV20TCL_POLYGON_SMOOTH_ENABLE 0x00000324 -#define NV20TCL_STENCIL_ENABLE 0x0000032c -#define NV20TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000330 -#define NV20TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000334 -#define NV20TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000338 -#define NV20TCL_ALPHA_FUNC_FUNC 0x0000033c -#define NV20TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV20TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV20TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV20TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV20TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV20TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV20TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV20TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV20TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV20TCL_ALPHA_FUNC_REF 0x00000340 -#define NV20TCL_BLEND_FUNC_SRC 0x00000344 -#define NV20TCL_BLEND_FUNC_SRC_ZERO 0x00000000 -#define NV20TCL_BLEND_FUNC_SRC_ONE 0x00000001 -#define NV20TCL_BLEND_FUNC_SRC_SRC_COLOR 0x00000300 -#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA 0x00000302 -#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV20TCL_BLEND_FUNC_SRC_DST_ALPHA 0x00000304 -#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV20TCL_BLEND_FUNC_SRC_DST_COLOR 0x00000306 -#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR 0x00000307 -#define NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE 0x00000308 -#define NV20TCL_BLEND_FUNC_SRC_CONSTANT_COLOR 0x00008001 -#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV20TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA 0x00008003 -#define NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV20TCL_BLEND_FUNC_DST 0x00000348 -#define NV20TCL_BLEND_FUNC_DST_ZERO 0x00000000 -#define NV20TCL_BLEND_FUNC_DST_ONE 0x00000001 -#define NV20TCL_BLEND_FUNC_DST_SRC_COLOR 0x00000300 -#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV20TCL_BLEND_FUNC_DST_SRC_ALPHA 0x00000302 -#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV20TCL_BLEND_FUNC_DST_DST_ALPHA 0x00000304 -#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV20TCL_BLEND_FUNC_DST_DST_COLOR 0x00000306 -#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR 0x00000307 -#define NV20TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE 0x00000308 -#define NV20TCL_BLEND_FUNC_DST_CONSTANT_COLOR 0x00008001 -#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV20TCL_BLEND_FUNC_DST_CONSTANT_ALPHA 0x00008003 -#define NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV20TCL_BLEND_COLOR 0x0000034c -#define NV20TCL_BLEND_EQUATION 0x00000350 -#define NV20TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 -#define NV20TCL_BLEND_EQUATION_MIN 0x00008007 -#define NV20TCL_BLEND_EQUATION_MAX 0x00008008 -#define NV20TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV20TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV20TCL_DEPTH_FUNC 0x00000354 -#define NV20TCL_DEPTH_FUNC_NEVER 0x00000200 -#define NV20TCL_DEPTH_FUNC_LESS 0x00000201 -#define NV20TCL_DEPTH_FUNC_EQUAL 0x00000202 -#define NV20TCL_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV20TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV20TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV20TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV20TCL_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV20TCL_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV20TCL_COLOR_MASK 0x00000358 -#define NV20TCL_DEPTH_WRITE_ENABLE 0x0000035c -#define NV20TCL_STENCIL_MASK 0x00000360 -#define NV20TCL_STENCIL_FUNC_FUNC 0x00000364 -#define NV20TCL_STENCIL_FUNC_FUNC_NEVER 0x00000200 -#define NV20TCL_STENCIL_FUNC_FUNC_LESS 0x00000201 -#define NV20TCL_STENCIL_FUNC_FUNC_EQUAL 0x00000202 -#define NV20TCL_STENCIL_FUNC_FUNC_LEQUAL 0x00000203 -#define NV20TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV20TCL_STENCIL_FUNC_FUNC_GREATER 0x00000204 -#define NV20TCL_STENCIL_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV20TCL_STENCIL_FUNC_FUNC_GEQUAL 0x00000206 -#define NV20TCL_STENCIL_FUNC_FUNC_ALWAYS 0x00000207 -#define NV20TCL_STENCIL_FUNC_REF 0x00000368 -#define NV20TCL_STENCIL_FUNC_MASK 0x0000036c -#define NV20TCL_STENCIL_OP_FAIL 0x00000370 -#define NV20TCL_STENCIL_OP_FAIL_ZERO 0x00000000 -#define NV20TCL_STENCIL_OP_FAIL_INVERT 0x0000150a -#define NV20TCL_STENCIL_OP_FAIL_KEEP 0x00001e00 -#define NV20TCL_STENCIL_OP_FAIL_REPLACE 0x00001e01 -#define NV20TCL_STENCIL_OP_FAIL_INCR 0x00001e02 -#define NV20TCL_STENCIL_OP_FAIL_DECR 0x00001e03 -#define NV20TCL_STENCIL_OP_FAIL_INCR_WRAP 0x00008507 -#define NV20TCL_STENCIL_OP_FAIL_DECR_WRAP 0x00008508 -#define NV20TCL_STENCIL_OP_ZFAIL 0x00000374 -#define NV20TCL_STENCIL_OP_ZFAIL_ZERO 0x00000000 -#define NV20TCL_STENCIL_OP_ZFAIL_INVERT 0x0000150a -#define NV20TCL_STENCIL_OP_ZFAIL_KEEP 0x00001e00 -#define NV20TCL_STENCIL_OP_ZFAIL_REPLACE 0x00001e01 -#define NV20TCL_STENCIL_OP_ZFAIL_INCR 0x00001e02 -#define NV20TCL_STENCIL_OP_ZFAIL_DECR 0x00001e03 -#define NV20TCL_STENCIL_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV20TCL_STENCIL_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV20TCL_STENCIL_OP_ZPASS 0x00000378 -#define NV20TCL_STENCIL_OP_ZPASS_ZERO 0x00000000 -#define NV20TCL_STENCIL_OP_ZPASS_INVERT 0x0000150a -#define NV20TCL_STENCIL_OP_ZPASS_KEEP 0x00001e00 -#define NV20TCL_STENCIL_OP_ZPASS_REPLACE 0x00001e01 -#define NV20TCL_STENCIL_OP_ZPASS_INCR 0x00001e02 -#define NV20TCL_STENCIL_OP_ZPASS_DECR 0x00001e03 -#define NV20TCL_STENCIL_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV20TCL_STENCIL_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV20TCL_SHADE_MODEL 0x0000037c -#define NV20TCL_SHADE_MODEL_FLAT 0x00001d00 -#define NV20TCL_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV20TCL_LINE_WIDTH 0x00000380 -#define NV20TCL_POLYGON_OFFSET_FACTOR 0x00000384 -#define NV20TCL_POLYGON_OFFSET_UNITS 0x00000388 -#define NV20TCL_POLYGON_MODE_FRONT 0x0000038c -#define NV20TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV20TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV20TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV20TCL_POLYGON_MODE_BACK 0x00000390 -#define NV20TCL_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV20TCL_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV20TCL_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV20TCL_DEPTH_RANGE_NEAR 0x00000394 -#define NV20TCL_DEPTH_RANGE_FAR 0x00000398 -#define NV20TCL_CULL_FACE 0x0000039c -#define NV20TCL_CULL_FACE_FRONT 0x00000404 -#define NV20TCL_CULL_FACE_BACK 0x00000405 -#define NV20TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV20TCL_FRONT_FACE 0x000003a0 -#define NV20TCL_FRONT_FACE_CW 0x00000900 -#define NV20TCL_FRONT_FACE_CCW 0x00000901 -#define NV20TCL_NORMALIZE_ENABLE 0x000003a4 -#define NV20TCL_SEPARATE_SPECULAR_ENABLE 0x000003b8 -#define NV20TCL_ENABLED_LIGHTS 0x000003bc -#define NV20TCL_CLIP_PLANE_ENABLE(x) (0x000003c0+((x)*4)) -#define NV20TCL_CLIP_PLANE_ENABLE__SIZE 0x00000010 -#define NV20TCL_TX_MATRIX_ENABLE(x) (0x00000420+((x)*4)) -#define NV20TCL_TX_MATRIX_ENABLE__SIZE 0x00000004 -#define NV20TCL_POINT_SIZE 0x0000043c -#define NV20TCL_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) -#define NV20TCL_MODELVIEW_MATRIX__SIZE 0x00000010 -#define NV20TCL_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) -#define NV20TCL_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV20TCL_TX0_MATRIX(x) (0x000006c0+((x)*4)) -#define NV20TCL_TX0_MATRIX__SIZE 0x00000010 -#define NV20TCL_TX1_MATRIX(x) (0x00000700+((x)*4)) -#define NV20TCL_TX1_MATRIX__SIZE 0x00000010 -#define NV20TCL_TX2_MATRIX(x) (0x00000740+((x)*4)) -#define NV20TCL_TX2_MATRIX__SIZE 0x00000010 -#define NV20TCL_TX3_MATRIX(x) (0x00000780+((x)*4)) -#define NV20TCL_TX3_MATRIX__SIZE 0x00000010 -#define NV20TCL_FOG_EQUATION_CONSTANT 0x000009c0 -#define NV20TCL_FOG_EQUATION_LINEAR 0x000009c4 -#define NV20TCL_FOG_EQUATION_QUADRATIC 0x000009c8 -#define NV20TCL_VIEWPORT_SCALE0_X 0x00000a20 -#define NV20TCL_VIEWPORT_SCALE0_Y 0x00000a24 -#define NV20TCL_VIEWPORT_SCALE0_Z 0x00000a28 -#define NV20TCL_VIEWPORT_SCALE0_W 0x00000a2c -#define NV20TCL_POINT_PARAMETER(x) (0x00000a30+((x)*4)) -#define NV20TCL_POINT_PARAMETER__SIZE 0x00000007 -#define NV20TCL_RC_CONSTANT_COLOR0(x) (0x00000a60+((x)*4)) -#define NV20TCL_RC_CONSTANT_COLOR0__SIZE 0x00000008 -#define NV20TCL_RC_CONSTANT_COLOR1(x) (0x00000a80+((x)*4)) -#define NV20TCL_RC_CONSTANT_COLOR1__SIZE 0x00000008 -#define NV20TCL_RC_OUT_ALPHA(x) (0x00000aa0+((x)*4)) -#define NV20TCL_RC_OUT_ALPHA__SIZE 0x00000008 -#define NV20TCL_RC_IN_RGB(x) (0x00000ac0+((x)*4)) -#define NV20TCL_RC_IN_RGB__SIZE 0x00000008 -#define NV20TCL_VIEWPORT_SCALE1_X 0x00000af0 -#define NV20TCL_VIEWPORT_SCALE1_Y 0x00000af4 -#define NV20TCL_VIEWPORT_SCALE1_Z 0x00000af8 -#define NV20TCL_VIEWPORT_SCALE1_W 0x00000afc -#define NV20TCL_VP_UPLOAD_INST(x) (0x00000b00+((x)*4)) -#define NV20TCL_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV20TCL_VP_UPLOAD_CONST(x) (0x00000b80+((x)*4)) -#define NV20TCL_VP_UPLOAD_CONST__SIZE 0x00000004 -#define NV20TCL_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV20TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV20TCL_VERTEX_POS_3F_X 0x00001500 -#define NV20TCL_VERTEX_POS_3F_Y 0x00001504 -#define NV20TCL_VERTEX_POS_3F_Z 0x00001508 -#define NV20TCL_VERTEX_POS_4F_X 0x00001518 -#define NV20TCL_VERTEX_POS_4F_Y 0x0000151c -#define NV20TCL_VERTEX_POS_4F_Z 0x00001520 -#define NV20TCL_VERTEX_POS_3I_XY 0x00001528 -#define NV20TCL_VERTEX_POS_3I_XY_X_SHIFT 0 -#define NV20TCL_VERTEX_POS_3I_XY_X_MASK 0x0000ffff -#define NV20TCL_VERTEX_POS_3I_XY_Y_SHIFT 16 -#define NV20TCL_VERTEX_POS_3I_XY_Y_MASK 0xffff0000 -#define NV20TCL_VERTEX_POS_3I_Z 0x0000152c -#define NV20TCL_VERTEX_POS_3I_Z_Z_SHIFT 0 -#define NV20TCL_VERTEX_POS_3I_Z_Z_MASK 0x0000ffff -#define NV20TCL_VERTEX_NOR_3F_X 0x00001530 -#define NV20TCL_VERTEX_NOR_3F_Y 0x00001534 -#define NV20TCL_VERTEX_NOR_3F_Z 0x00001538 -#define NV20TCL_VERTEX_NOR_3I_XY 0x00001540 -#define NV20TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV20TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV20TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV20TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV20TCL_VERTEX_NOR_3I_Z 0x00001544 -#define NV20TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV20TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV20TCL_VERTEX_COL_4F_X 0x00001550 -#define NV20TCL_VERTEX_COL_4F_Y 0x00001554 -#define NV20TCL_VERTEX_COL_4F_Z 0x00001558 -#define NV20TCL_VERTEX_COL_4F_W 0x0000155c -#define NV20TCL_VERTEX_COL_3F_X 0x00001560 -#define NV20TCL_VERTEX_COL_3F_Y 0x00001564 -#define NV20TCL_VERTEX_COL_3F_Z 0x00001568 -#define NV20TCL_VERTEX_COL_4I 0x0000156c -#define NV20TCL_VERTEX_COL_4I_R_SHIFT 0 -#define NV20TCL_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV20TCL_VERTEX_COL_4I_G_SHIFT 8 -#define NV20TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV20TCL_VERTEX_COL_4I_B_SHIFT 16 -#define NV20TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV20TCL_VERTEX_COL_4I_A_SHIFT 24 -#define NV20TCL_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV20TCL_VERTEX_COL2_3F_X 0x00001580 -#define NV20TCL_VERTEX_COL2_3F_Y 0x00001584 -#define NV20TCL_VERTEX_COL2_3F_Z 0x00001588 -#define NV20TCL_VERTEX_COL2_4I 0x0000158c -#define NV20TCL_VERTEX_COL2_4I_R_SHIFT 0 -#define NV20TCL_VERTEX_COL2_4I_R_MASK 0x000000ff -#define NV20TCL_VERTEX_COL2_4I_G_SHIFT 8 -#define NV20TCL_VERTEX_COL2_4I_G_MASK 0x0000ff00 -#define NV20TCL_VERTEX_COL2_4I_B_SHIFT 16 -#define NV20TCL_VERTEX_COL2_4I_B_MASK 0x00ff0000 -#define NV20TCL_VERTEX_COL2_4I_A_SHIFT 24 -#define NV20TCL_VERTEX_COL2_4I_A_MASK 0xff000000 -#define NV20TCL_VERTEX_TX0_2F_S 0x00001590 -#define NV20TCL_VERTEX_TX0_2F_T 0x00001594 -#define NV20TCL_VERTEX_TX0_2I 0x00001598 -#define NV20TCL_VERTEX_TX0_2I_S_SHIFT 0 -#define NV20TCL_VERTEX_TX0_2I_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX0_2I_T_SHIFT 16 -#define NV20TCL_VERTEX_TX0_2I_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX0_4F_S 0x000015a0 -#define NV20TCL_VERTEX_TX0_4F_T 0x000015a4 -#define NV20TCL_VERTEX_TX0_4F_R 0x000015a8 -#define NV20TCL_VERTEX_TX0_4F_Q 0x000015ac -#define NV20TCL_VERTEX_TX0_4I_ST 0x000015b0 -#define NV20TCL_VERTEX_TX0_4I_ST_S_SHIFT 0 -#define NV20TCL_VERTEX_TX0_4I_ST_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX0_4I_ST_T_SHIFT 16 -#define NV20TCL_VERTEX_TX0_4I_ST_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX0_4I_RQ 0x000015b4 -#define NV20TCL_VERTEX_TX0_4I_RQ_R_SHIFT 0 -#define NV20TCL_VERTEX_TX0_4I_RQ_R_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX0_4I_RQ_Q_SHIFT 16 -#define NV20TCL_VERTEX_TX0_4I_RQ_Q_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX1_2F_S 0x000015b8 -#define NV20TCL_VERTEX_TX1_2F_T 0x000015bc -#define NV20TCL_VERTEX_TX1_2I 0x000015c0 -#define NV20TCL_VERTEX_TX1_2I_S_SHIFT 0 -#define NV20TCL_VERTEX_TX1_2I_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX1_2I_T_SHIFT 16 -#define NV20TCL_VERTEX_TX1_2I_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX1_4F_S 0x000015c8 -#define NV20TCL_VERTEX_TX1_4F_T 0x000015cc -#define NV20TCL_VERTEX_TX1_4F_R 0x000015d0 -#define NV20TCL_VERTEX_TX1_4F_Q 0x000015d4 -#define NV20TCL_VERTEX_TX1_4I_ST 0x000015d8 -#define NV20TCL_VERTEX_TX1_4I_ST_S_SHIFT 0 -#define NV20TCL_VERTEX_TX1_4I_ST_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX1_4I_ST_T_SHIFT 16 -#define NV20TCL_VERTEX_TX1_4I_ST_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX1_4I_RQ 0x000015dc -#define NV20TCL_VERTEX_TX1_4I_RQ_R_SHIFT 0 -#define NV20TCL_VERTEX_TX1_4I_RQ_R_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX1_4I_RQ_Q_SHIFT 16 -#define NV20TCL_VERTEX_TX1_4I_RQ_Q_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX2_2F_S 0x000015e0 -#define NV20TCL_VERTEX_TX2_2F_T 0x000015e4 -#define NV20TCL_VERTEX_TX2_2I 0x000015e8 -#define NV20TCL_VERTEX_TX2_2I_S_SHIFT 0 -#define NV20TCL_VERTEX_TX2_2I_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX2_2I_T_SHIFT 16 -#define NV20TCL_VERTEX_TX2_2I_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX2_4F_S 0x000015f0 -#define NV20TCL_VERTEX_TX2_4F_T 0x000015f4 -#define NV20TCL_VERTEX_TX2_4F_R 0x000015f8 -#define NV20TCL_VERTEX_TX2_4F_Q 0x000015fc -#define NV20TCL_VERTEX_TX2_4I_ST 0x00001600 -#define NV20TCL_VERTEX_TX2_4I_ST_S_SHIFT 0 -#define NV20TCL_VERTEX_TX2_4I_ST_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX2_4I_ST_T_SHIFT 16 -#define NV20TCL_VERTEX_TX2_4I_ST_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX2_4I_RQ 0x00001604 -#define NV20TCL_VERTEX_TX2_4I_RQ_R_SHIFT 0 -#define NV20TCL_VERTEX_TX2_4I_RQ_R_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX2_4I_RQ_Q_SHIFT 16 -#define NV20TCL_VERTEX_TX2_4I_RQ_Q_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX3_2F_S 0x00001608 -#define NV20TCL_VERTEX_TX3_2F_T 0x0000160c -#define NV20TCL_VERTEX_TX3_2I 0x00001610 -#define NV20TCL_VERTEX_TX3_2I_S_SHIFT 0 -#define NV20TCL_VERTEX_TX3_2I_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX3_2I_T_SHIFT 16 -#define NV20TCL_VERTEX_TX3_2I_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX3_4F_S 0x00001620 -#define NV20TCL_VERTEX_TX3_4F_T 0x00001624 -#define NV20TCL_VERTEX_TX3_4F_R 0x00001628 -#define NV20TCL_VERTEX_TX3_4F_Q 0x0000162c -#define NV20TCL_VERTEX_TX3_4I_ST 0x00001630 -#define NV20TCL_VERTEX_TX3_4I_ST_S_SHIFT 0 -#define NV20TCL_VERTEX_TX3_4I_ST_S_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX3_4I_ST_T_SHIFT 16 -#define NV20TCL_VERTEX_TX3_4I_ST_T_MASK 0xffff0000 -#define NV20TCL_VERTEX_TX3_4I_RQ 0x00001634 -#define NV20TCL_VERTEX_TX3_4I_RQ_R_SHIFT 0 -#define NV20TCL_VERTEX_TX3_4I_RQ_R_MASK 0x0000ffff -#define NV20TCL_VERTEX_TX3_4I_RQ_Q_SHIFT 16 -#define NV20TCL_VERTEX_TX3_4I_RQ_Q_MASK 0xffff0000 -#define NV20TCL_EDGEFLAG_ENABLE 0x000016bc -#define NV20TCL_VERTEX_ATTR_OFFSET(x) (0x00001720+((x)*4)) -#define NV20TCL_VERTEX_ATTR_OFFSET__SIZE 0x00000010 -#define NV20TCL_VERTEX_ARRAY_FORMAT(x) (0x00001760+((x)*4)) -#define NV20TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV20TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV20TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV20TCL_VERTEX_ARRAY_FORMAT_FIELDS_SHIFT 4 -#define NV20TCL_VERTEX_ARRAY_FORMAT_FIELDS_MASK 0x000000f0 -#define NV20TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV20TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 -#define NV20TCL_COLOR_LOGIC_OP_ENABLE 0x000017bc -#define NV20TCL_COLOR_LOGIC_OP_OP 0x000017c0 -#define NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE 0x000017c4 -#define NV20TCL_TX_SHADER_CULL_MODE 0x000017f8 -#define NV20TCL_VERTEX_BEGIN_END 0x000017fc -#define NV20TCL_VERTEX_BUFFER_DRAW_ARRAYS 0x00001810 -#define NV20TCL_VERTEX_ARRAY_DATA 0x00001818 -#define NV20TCL_TX_OFFSET(x) (0x00001b00+((x)*64)) -#define NV20TCL_TX_OFFSET__SIZE 0x00000004 -#define NV20TCL_TX_FORMAT(x) (0x00001b04+((x)*64)) -#define NV20TCL_TX_FORMAT__SIZE 0x00000004 -#define NV20TCL_TX_WRAP(x) (0x00001b08+((x)*64)) -#define NV20TCL_TX_WRAP__SIZE 0x00000004 -#define NV20TCL_TX_ENABLE(x) (0x00001b0c+((x)*64)) -#define NV20TCL_TX_ENABLE__SIZE 0x00000004 -#define NV20TCL_TX_FILTER(x) (0x00001b14+((x)*64)) -#define NV20TCL_TX_FILTER__SIZE 0x00000004 -#define NV20TCL_TX_BORDER_COLOR(x) (0x00001b24+((x)*64)) -#define NV20TCL_TX_BORDER_COLOR__SIZE 0x00000004 -#define NV20TCL_SCISSOR_HORIZ 0x00001c30 -#define NV20TCL_SCISSOR_VERT 0x00001c50 -#define NV20TCL_MULTISAMPLE_CONTROL 0x00001d7c -#define NV20TCL_CLEAR_VALUE 0x00001d90 -#define NV20TCL_CLEAR_BUFFERS 0x00001d94 -#define NV20TCL_RC_COLOR0 0x00001e20 -#define NV20TCL_RC_COLOR1 0x00001e24 -#define NV20TCL_RC_OUT_RGB(x) (0x00001e40+((x)*4)) -#define NV20TCL_RC_OUT_RGB__SIZE 0x00000008 -#define NV20TCL_RC_ENABLE 0x00001e60 -#define NV20TCL_TX_SHADER_OP 0x00001e70 -#define NV20TCL_VP_UPLOAD_CONST_ID 0x00001ea4 - - -#define NV17TCL 0x00000099 - -#define NV17TCL_DMA_IN_MEMORY4 0x000001ac -#define NV17TCL_DMA_IN_MEMORY5 0x000001b0 -#define NV17TCL_COLOR_MASK_ENABLE 0x000002bc -#define NV17TCL_LMA_DEPTH_BUFFER_PITCH 0x00000d5c -#define NV17TCL_LMA_DEPTH_BUFFER_OFFSET 0x00000d60 -#define NV17TCL_LMA_DEPTH_FILL_VALUE 0x00000d68 -#define NV17TCL_LMA_DEPTH_BUFFER_CLEAR 0x00000d6c -#define NV17TCL_LMA_DEPTH_ENABLE 0x00001658 - - -#define NV20_SWIZZLED_SURFACE 0x0000009e - - - -#define NV12_IMAGE_BLIT 0x0000009f - - - -#define NV30_CONTEXT_SURFACES_2D 0x00000362 - - - -#define NV30_STRETCHED_IMAGE_FROM_CPU 0x00000366 - - - -#define NV30_TEXTURE_FROM_CPU 0x0000037b - - - -#define NV30_SCALED_IMAGE_FROM_MEMORY 0x00000389 - - - -#define NV30_IMAGE_FROM_CPU 0x0000038a - - - -#define NV30TCL 0x00000397 - - - -#define NV30_SWIZZLED_SURFACE 0x0000039e - - - -#define NV35TCL 0x00000497 - - - -#define NV25TCL 0x00000597 - -#define NV25TCL_DMA_IN_MEMORY4 0x0000019c -#define NV25TCL_DMA_IN_MEMORY5 0x000001a0 -#define NV25TCL_DMA_IN_MEMORY8 0x000001ac -#define NV25TCL_DMA_IN_MEMORY9 0x000001b0 - - -#define NV34TCL 0x00000697 - -#define NV34TCL_NOP 0x00000100 -#define NV34TCL_NOTIFY 0x00000104 -#define NV34TCL_DMA_NOTIFY 0x00000180 -#define NV34TCL_DMA_TEXTURE0 0x00000184 -#define NV34TCL_DMA_TEXTURE1 0x00000188 -#define NV34TCL_DMA_COLOR1 0x0000018c -#define NV34TCL_DMA_COLOR0 0x00000194 -#define NV34TCL_DMA_ZETA 0x00000198 -#define NV34TCL_DMA_VTXBUF0 0x0000019c -#define NV34TCL_DMA_VTXBUF1 0x000001a0 -#define NV34TCL_DMA_FENCE 0x000001a4 -#define NV34TCL_DMA_QUERY 0x000001a8 -#define NV34TCL_DMA_IN_MEMORY7 0x000001ac -#define NV34TCL_DMA_IN_MEMORY8 0x000001b0 -#define NV34TCL_RT_HORIZ 0x00000200 -#define NV34TCL_RT_HORIZ_X_SHIFT 0 -#define NV34TCL_RT_HORIZ_X_MASK 0x0000ffff -#define NV34TCL_RT_HORIZ_W_SHIFT 16 -#define NV34TCL_RT_HORIZ_W_MASK 0xffff0000 -#define NV34TCL_RT_VERT 0x00000204 -#define NV34TCL_RT_VERT_Y_SHIFT 0 -#define NV34TCL_RT_VERT_Y_MASK 0x0000ffff -#define NV34TCL_RT_VERT_H_SHIFT 16 -#define NV34TCL_RT_VERT_H_MASK 0xffff0000 -#define NV34TCL_RT_FORMAT 0x00000208 -#define NV34TCL_RT_FORMAT_TYPE_SHIFT 8 -#define NV34TCL_RT_FORMAT_TYPE_MASK 0x00000f00 -#define NV34TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 -#define NV34TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 -#define NV34TCL_RT_FORMAT_ZETA_SHIFT 5 -#define NV34TCL_RT_FORMAT_ZETA_MASK 0x000000e0 -#define NV34TCL_RT_FORMAT_ZETA_Z16 0x00000020 -#define NV34TCL_RT_FORMAT_ZETA_Z24S8 0x00000040 -#define NV34TCL_RT_FORMAT_COLOR_SHIFT 0 -#define NV34TCL_RT_FORMAT_COLOR_MASK 0x0000001f -#define NV34TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 -#define NV34TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 -#define NV34TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 -#define NV34TCL_RT_FORMAT_COLOR_B8 0x00000009 -#define NV34TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f -#define NV34TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 -#define NV34TCL_COLOR0_PITCH 0x0000020c -#define NV34TCL_COLOR0_PITCH_COLOR0_SHIFT 0 -#define NV34TCL_COLOR0_PITCH_COLOR0_MASK 0x0000ffff -#define NV34TCL_COLOR0_PITCH_ZETA_SHIFT 16 -#define NV34TCL_COLOR0_PITCH_ZETA_MASK 0xffff0000 -#define NV34TCL_COLOR0_OFFSET 0x00000210 -#define NV34TCL_ZETA_OFFSET 0x00000214 -#define NV34TCL_COLOR1_OFFSET 0x00000218 -#define NV34TCL_COLOR1_PITCH 0x0000021c -#define NV34TCL_RT_ENABLE 0x00000220 -#define NV34TCL_RT_ENABLE_MRT (1 << 4) -#define NV34TCL_RT_ENABLE_COLOR3 (1 << 3) -#define NV34TCL_RT_ENABLE_COLOR2 (1 << 2) -#define NV34TCL_RT_ENABLE_COLOR1 (1 << 1) -#define NV34TCL_RT_ENABLE_COLOR0 (1 << 0) -#define NV34TCL_ZETA_PITCH 0x0000022c -#define NV34TCL_LMA_DEPTH_OFFSET 0x00000230 -#define NV34TCL_TX_UNITS_ENABLE 0x0000023c -#define NV34TCL_TX_UNITS_ENABLE_TX0 (1 << 0) -#define NV34TCL_TX_UNITS_ENABLE_TX1 (1 << 1) -#define NV34TCL_TX_UNITS_ENABLE_TX2 (1 << 2) -#define NV34TCL_TX_UNITS_ENABLE_TX3 (1 << 3) -#define NV34TCL_TX_MATRIX_ENABLE(x) (0x00000240+((x)*4)) -#define NV34TCL_TX_MATRIX_ENABLE__SIZE 0x00000004 -#define NV34TCL_UNK0250(x) (0x00000250+((x)*4)) -#define NV34TCL_UNK0250__SIZE 0x00000004 -#define NV34TCL_VIEWPORT_TX_ORIGIN 0x000002b8 -#define NV34TCL_VIEWPORT_TX_ORIGIN_X_SHIFT 0 -#define NV34TCL_VIEWPORT_TX_ORIGIN_X_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_TX_ORIGIN_Y_SHIFT 16 -#define NV34TCL_VIEWPORT_TX_ORIGIN_Y_MASK 0xffff0000 -#define NV34TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) -#define NV34TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV34TCL_VIEWPORT_CLIP_HORIZ_L_SHIFT 0 -#define NV34TCL_VIEWPORT_CLIP_HORIZ_L_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_CLIP_HORIZ_R_SHIFT 16 -#define NV34TCL_VIEWPORT_CLIP_HORIZ_R_MASK 0xffff0000 -#define NV34TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) -#define NV34TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV34TCL_VIEWPORT_CLIP_VERT_T_SHIFT 0 -#define NV34TCL_VIEWPORT_CLIP_VERT_T_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_CLIP_VERT_D_SHIFT 16 -#define NV34TCL_VIEWPORT_CLIP_VERT_D_MASK 0xffff0000 -#define NV34TCL_DITHER_ENABLE 0x00000300 -#define NV34TCL_ALPHA_FUNC_ENABLE 0x00000304 -#define NV34TCL_ALPHA_FUNC_FUNC 0x00000308 -#define NV34TCL_ALPHA_FUNC_FUNC_NEVER 0x00000200 -#define NV34TCL_ALPHA_FUNC_FUNC_LESS 0x00000201 -#define NV34TCL_ALPHA_FUNC_FUNC_EQUAL 0x00000202 -#define NV34TCL_ALPHA_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV34TCL_ALPHA_FUNC_FUNC_GREATER 0x00000204 -#define NV34TCL_ALPHA_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34TCL_ALPHA_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34TCL_ALPHA_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34TCL_ALPHA_FUNC_REF 0x0000030c -#define NV34TCL_BLEND_FUNC_ENABLE 0x00000310 -#define NV34TCL_BLEND_FUNC_SRC 0x00000314 -#define NV34TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 -#define NV34TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff -#define NV34TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 -#define NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV34TCL_BLEND_FUNC_DST 0x00000318 -#define NV34TCL_BLEND_FUNC_DST_RGB_SHIFT 0 -#define NV34TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff -#define NV34TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV34TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV34TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 -#define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV34TCL_BLEND_FUNC_COLOR 0x0000031c -#define NV34TCL_BLEND_FUNC_EQUATION 0x00000320 -#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 -#define NV34TCL_BLEND_FUNC_EQUATION_MIN 0x00008007 -#define NV34TCL_BLEND_FUNC_EQUATION_MAX 0x00008008 -#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV34TCL_COLOR_MASK 0x00000324 -#define NV34TCL_COLOR_MASK_B_SHIFT 0 -#define NV34TCL_COLOR_MASK_B_MASK 0x000000ff -#define NV34TCL_COLOR_MASK_G_SHIFT 8 -#define NV34TCL_COLOR_MASK_G_MASK 0x0000ff00 -#define NV34TCL_COLOR_MASK_R_SHIFT 16 -#define NV34TCL_COLOR_MASK_R_MASK 0x00ff0000 -#define NV34TCL_COLOR_MASK_A_SHIFT 24 -#define NV34TCL_COLOR_MASK_A_MASK 0xff000000 -#define NV34TCL_STENCIL_BACK_ENABLE 0x00000328 -#define NV34TCL_STENCIL_BACK_MASK 0x0000032c -#define NV34TCL_STENCIL_BACK_FUNC_FUNC 0x00000330 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34TCL_STENCIL_BACK_FUNC_REF 0x00000334 -#define NV34TCL_STENCIL_BACK_FUNC_MASK 0x00000338 -#define NV34TCL_STENCIL_BACK_OP_FAIL 0x0000033c -#define NV34TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV34TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV34TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV34TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV34TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV34TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV34TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV34TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL 0x00000340 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV34TCL_STENCIL_BACK_OP_ZPASS 0x00000344 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV34TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV34TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV34TCL_STENCIL_FRONT_ENABLE 0x00000348 -#define NV34TCL_STENCIL_FRONT_MASK 0x0000034c -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC 0x00000350 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV34TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV34TCL_STENCIL_FRONT_FUNC_REF 0x00000354 -#define NV34TCL_STENCIL_FRONT_FUNC_MASK 0x00000358 -#define NV34TCL_STENCIL_FRONT_OP_FAIL 0x0000035c -#define NV34TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV34TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV34TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV34TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV34TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV34TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV34TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV34TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL 0x00000360 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS 0x00000364 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV34TCL_SHADE_MODEL 0x00000368 -#define NV34TCL_SHADE_MODEL_FLAT 0x00001d00 -#define NV34TCL_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV34TCL_FOG_ENABLE 0x0000036c -#define NV34TCL_FOG_COLOR 0x00000370 -#define NV34TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 -#define NV34TCL_COLOR_LOGIC_OP_OP 0x00000378 -#define NV34TCL_COLOR_LOGIC_OP_OP_CLEAR 0x00001500 -#define NV34TCL_COLOR_LOGIC_OP_OP_AND 0x00001501 -#define NV34TCL_COLOR_LOGIC_OP_OP_AND_REVERSE 0x00001502 -#define NV34TCL_COLOR_LOGIC_OP_OP_COPY 0x00001503 -#define NV34TCL_COLOR_LOGIC_OP_OP_AND_INVERTED 0x00001504 -#define NV34TCL_COLOR_LOGIC_OP_OP_NOOP 0x00001505 -#define NV34TCL_COLOR_LOGIC_OP_OP_XOR 0x00001506 -#define NV34TCL_COLOR_LOGIC_OP_OP_OR 0x00001507 -#define NV34TCL_COLOR_LOGIC_OP_OP_NOR 0x00001508 -#define NV34TCL_COLOR_LOGIC_OP_OP_EQUIV 0x00001509 -#define NV34TCL_COLOR_LOGIC_OP_OP_INVERT 0x0000150a -#define NV34TCL_COLOR_LOGIC_OP_OP_OR_REVERSE 0x0000150b -#define NV34TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED 0x0000150c -#define NV34TCL_COLOR_LOGIC_OP_OP_OR_INVERTED 0x0000150d -#define NV34TCL_COLOR_LOGIC_OP_OP_NAND 0x0000150e -#define NV34TCL_COLOR_LOGIC_OP_OP_SET 0x0000150f -#define NV34TCL_NORMALIZE_ENABLE 0x0000037c -#define NV34TCL_COLOR_MATERIAL 0x00000390 -#define NV34TCL_COLOR_MATERIAL_FRONT_EMISSION_ENABLE (1 << 0) -#define NV34TCL_COLOR_MATERIAL_FRONT_AMBIENT_ENABLE (1 << 2) -#define NV34TCL_COLOR_MATERIAL_FRONT_DIFFUSE_ENABLE (1 << 4) -#define NV34TCL_COLOR_MATERIAL_FRONT_SPECULAR_ENABLE (1 << 6) -#define NV34TCL_COLOR_MATERIAL_BACK_EMISSION_ENABLE (1 << 8) -#define NV34TCL_COLOR_MATERIAL_BACK_AMBIENT_ENABLE (1 << 10) -#define NV34TCL_COLOR_MATERIAL_BACK_DIFFUSE_ENABLE (1 << 12) -#define NV34TCL_COLOR_MATERIAL_BACK_SPECULAR_ENABLE (1 << 14) -#define NV34TCL_DEPTH_RANGE_NEAR 0x00000394 -#define NV34TCL_DEPTH_RANGE_FAR 0x00000398 -#define NV34TCL_COLOR_MATERIAL_FRONT_R 0x000003a0 -#define NV34TCL_COLOR_MATERIAL_FRONT_G 0x000003a4 -#define NV34TCL_COLOR_MATERIAL_FRONT_B 0x000003a8 -#define NV34TCL_COLOR_MATERIAL_FRONT_A 0x000003b4 -#define NV34TCL_LINE_WIDTH 0x000003b8 -#define NV34TCL_LINE_SMOOTH_ENABLE 0x000003bc -#define NV34TCL_CLIP_PLANE_ENABLE(x) (0x00000400+((x)*4)) -#define NV34TCL_CLIP_PLANE_ENABLE__SIZE 0x00000020 -#define NV34TCL_CLIP_PLANE_ENABLE_FALSE 0x00000000 -#define NV34TCL_CLIP_PLANE_ENABLE_EYE_LINEAR 0x00002400 -#define NV34TCL_CLIP_PLANE_ENABLE_OBJECT_LINEAR 0x00002401 -#define NV34TCL_MODELVIEW_MATRIX(x) (0x00000480+((x)*4)) -#define NV34TCL_MODELVIEW_MATRIX__SIZE 0x00000010 -#define NV34TCL_INVERSE_MODELVIEW_MATRIX(x) (0x00000580+((x)*4)) -#define NV34TCL_INVERSE_MODELVIEW_MATRIX__SIZE 0x0000000c -#define NV34TCL_PROJECTION_MATRIX(x) (0x00000680+((x)*4)) -#define NV34TCL_PROJECTION_MATRIX__SIZE 0x00000010 -#define NV34TCL_TX0_MATRIX(x) (0x000006c0+((x)*4)) -#define NV34TCL_TX0_MATRIX__SIZE 0x00000010 -#define NV34TCL_TX1_MATRIX(x) (0x00000700+((x)*4)) -#define NV34TCL_TX1_MATRIX__SIZE 0x00000010 -#define NV34TCL_TX2_MATRIX(x) (0x00000740+((x)*4)) -#define NV34TCL_TX2_MATRIX__SIZE 0x00000010 -#define NV34TCL_TX3_MATRIX(x) (0x00000780+((x)*4)) -#define NV34TCL_TX3_MATRIX__SIZE 0x00000010 -#define NV34TCL_SCISSOR_HORIZ 0x000008c0 -#define NV34TCL_SCISSOR_HORIZ_X_SHIFT 0 -#define NV34TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff -#define NV34TCL_SCISSOR_HORIZ_W_SHIFT 16 -#define NV34TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 -#define NV34TCL_SCISSOR_VERT 0x000008c4 -#define NV34TCL_SCISSOR_VERT_Y_SHIFT 0 -#define NV34TCL_SCISSOR_VERT_Y_MASK 0x0000ffff -#define NV34TCL_SCISSOR_VERT_H_SHIFT 16 -#define NV34TCL_SCISSOR_VERT_H_MASK 0xffff0000 -#define NV34TCL_FOG_COORD_DIST 0x000008c8 -#define NV34TCL_FOG_MODE 0x000008cc -#define NV34TCL_FOG_MODE_EXP 0x00000800 -#define NV34TCL_FOG_MODE_EXP_2 0x00000802 -#define NV34TCL_FOG_MODE_EXP2 0x00000803 -#define NV34TCL_FOG_MODE_LINEAR 0x00000804 -#define NV34TCL_FOG_MODE_LINEAR_2 0x00002601 -#define NV34TCL_FOG_EQUATION_CONSTANT 0x000008d0 -#define NV34TCL_FOG_EQUATION_LINEAR 0x000008d4 -#define NV34TCL_FOG_EQUATION_QUADRATIC 0x000008d8 -#define NV34TCL_FP_ACTIVE_PROGRAM 0x000008e4 -#define NV34TCL_FP_ACTIVE_PROGRAM_DMA0 (1 << 0) -#define NV34TCL_FP_ACTIVE_PROGRAM_DMA1 (1 << 1) -#define NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_SHIFT 2 -#define NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_MASK 0xfffffffc -#define NV34TCL_RC_COLOR0 0x000008ec -#define NV34TCL_RC_COLOR0_B_SHIFT 0 -#define NV34TCL_RC_COLOR0_B_MASK 0x000000ff -#define NV34TCL_RC_COLOR0_G_SHIFT 8 -#define NV34TCL_RC_COLOR0_G_MASK 0x0000ff00 -#define NV34TCL_RC_COLOR0_R_SHIFT 16 -#define NV34TCL_RC_COLOR0_R_MASK 0x00ff0000 -#define NV34TCL_RC_COLOR0_A_SHIFT 24 -#define NV34TCL_RC_COLOR0_A_MASK 0xff000000 -#define NV34TCL_RC_COLOR1 0x000008f0 -#define NV34TCL_RC_COLOR1_B_SHIFT 0 -#define NV34TCL_RC_COLOR1_B_MASK 0x000000ff -#define NV34TCL_RC_COLOR1_G_SHIFT 8 -#define NV34TCL_RC_COLOR1_G_MASK 0x0000ff00 -#define NV34TCL_RC_COLOR1_R_SHIFT 16 -#define NV34TCL_RC_COLOR1_R_MASK 0x00ff0000 -#define NV34TCL_RC_COLOR1_A_SHIFT 24 -#define NV34TCL_RC_COLOR1_A_MASK 0xff000000 -#define NV34TCL_RC_FINAL0 0x000008f4 -#define NV34TCL_RC_FINAL1 0x000008f8 -#define NV34TCL_RC_ENABLE 0x000008fc -#define NV34TCL_RC_IN_ALPHA(x) (0x00000900+((x)*32)) -#define NV34TCL_RC_IN_ALPHA__SIZE 0x00000008 -#define NV34TCL_RC_IN_RGB(x) (0x00000904+((x)*32)) -#define NV34TCL_RC_IN_RGB__SIZE 0x00000008 -#define NV34TCL_RC_CONSTANT_COLOR0(x) (0x00000908+((x)*32)) -#define NV34TCL_RC_CONSTANT_COLOR0__SIZE 0x00000008 -#define NV34TCL_RC_CONSTANT_COLOR0_B_SHIFT 0 -#define NV34TCL_RC_CONSTANT_COLOR0_B_MASK 0x000000ff -#define NV34TCL_RC_CONSTANT_COLOR0_G_SHIFT 8 -#define NV34TCL_RC_CONSTANT_COLOR0_G_MASK 0x0000ff00 -#define NV34TCL_RC_CONSTANT_COLOR0_R_SHIFT 16 -#define NV34TCL_RC_CONSTANT_COLOR0_R_MASK 0x00ff0000 -#define NV34TCL_RC_CONSTANT_COLOR0_A_SHIFT 24 -#define NV34TCL_RC_CONSTANT_COLOR0_A_MASK 0xff000000 -#define NV34TCL_RC_CONSTANT_COLOR1(x) (0x0000090c+((x)*32)) -#define NV34TCL_RC_CONSTANT_COLOR1__SIZE 0x00000008 -#define NV34TCL_RC_CONSTANT_COLOR1_B_SHIFT 0 -#define NV34TCL_RC_CONSTANT_COLOR1_B_MASK 0x000000ff -#define NV34TCL_RC_CONSTANT_COLOR1_G_SHIFT 8 -#define NV34TCL_RC_CONSTANT_COLOR1_G_MASK 0x0000ff00 -#define NV34TCL_RC_CONSTANT_COLOR1_R_SHIFT 16 -#define NV34TCL_RC_CONSTANT_COLOR1_R_MASK 0x00ff0000 -#define NV34TCL_RC_CONSTANT_COLOR1_A_SHIFT 24 -#define NV34TCL_RC_CONSTANT_COLOR1_A_MASK 0xff000000 -#define NV34TCL_RC_OUT_ALPHA(x) (0x00000910+((x)*32)) -#define NV34TCL_RC_OUT_ALPHA__SIZE 0x00000008 -#define NV34TCL_RC_OUT_RGB(x) (0x00000914+((x)*32)) -#define NV34TCL_RC_OUT_RGB__SIZE 0x00000008 -#define NV34TCL_VIEWPORT_HORIZ 0x00000a00 -#define NV34TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV34TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV34TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV34TCL_VIEWPORT_VERT 0x00000a04 -#define NV34TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV34TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV34TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV34TCL_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x00000a10 -#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x00000a14 -#define NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x00000a18 -#define NV34TCL_VIEWPORT_TRANSLATE_X 0x00000a20 -#define NV34TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 -#define NV34TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 -#define NV34TCL_VIEWPORT_TRANSLATE_W 0x00000a2c -#define NV34TCL_VIEWPORT_SCALE_X 0x00000a30 -#define NV34TCL_VIEWPORT_SCALE_Y 0x00000a34 -#define NV34TCL_VIEWPORT_SCALE_Z 0x00000a38 -#define NV34TCL_VIEWPORT_SCALE_W 0x00000a3c -#define NV34TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 -#define NV34TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV34TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 -#define NV34TCL_DEPTH_FUNC 0x00000a6c -#define NV34TCL_DEPTH_FUNC_NEVER 0x00000200 -#define NV34TCL_DEPTH_FUNC_LESS 0x00000201 -#define NV34TCL_DEPTH_FUNC_EQUAL 0x00000202 -#define NV34TCL_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV34TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV34TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV34TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV34TCL_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV34TCL_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV34TCL_DEPTH_WRITE_ENABLE 0x00000a70 -#define NV34TCL_DEPTH_TEST_ENABLE 0x00000a74 -#define NV34TCL_POLYGON_OFFSET_FACTOR 0x00000a78 -#define NV34TCL_POLYGON_OFFSET_UNITS 0x00000a7c -#define NV34TCL_VERTEX_NOR_3I_XY 0x00000a90 -#define NV34TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV34TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV34TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV34TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV34TCL_VERTEX_NOR_3I_Z 0x00000a94 -#define NV34TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV34TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff -#define NV34TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) -#define NV34TCL_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV34TCL_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) -#define NV34TCL_CLIP_PLANE_A__SIZE 0x00000020 -#define NV34TCL_CLIP_PLANE_B(x) (0x00000e04+((x)*16)) -#define NV34TCL_CLIP_PLANE_B__SIZE 0x00000020 -#define NV34TCL_CLIP_PLANE_C(x) (0x00000e08+((x)*16)) -#define NV34TCL_CLIP_PLANE_C__SIZE 0x00000020 -#define NV34TCL_CLIP_PLANE_D(x) (0x00000e0c+((x)*16)) -#define NV34TCL_CLIP_PLANE_D__SIZE 0x00000020 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x) (0x00001000+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x) (0x00001004+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x) (0x00001008+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x) (0x0000100c+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x) (0x00001010+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x) (0x00001014+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x) (0x00001018+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x) (0x0000101c+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE 0x00000008 -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x) (0x00001020+((x)*64)) -#define NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE 0x00000008 -#define NV34TCL_LIGHT_HALF_VECTOR_X(x) (0x00001028+((x)*64)) -#define NV34TCL_LIGHT_HALF_VECTOR_X__SIZE 0x00000008 -#define NV34TCL_LIGHT_HALF_VECTOR_Y(x) (0x0000102c+((x)*64)) -#define NV34TCL_LIGHT_HALF_VECTOR_Y__SIZE 0x00000008 -#define NV34TCL_LIGHT_HALF_VECTOR_Z(x) (0x00001030+((x)*64)) -#define NV34TCL_LIGHT_HALF_VECTOR_Z__SIZE 0x00000008 -#define NV34TCL_LIGHT_DIRECTION_X(x) (0x00001034+((x)*64)) -#define NV34TCL_LIGHT_DIRECTION_X__SIZE 0x00000008 -#define NV34TCL_LIGHT_DIRECTION_Y(x) (0x00001038+((x)*64)) -#define NV34TCL_LIGHT_DIRECTION_Y__SIZE 0x00000008 -#define NV34TCL_LIGHT_DIRECTION_Z(x) (0x0000103c+((x)*64)) -#define NV34TCL_LIGHT_DIRECTION_Z__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_CUTOFF_A(x) (0x00001200+((x)*64)) -#define NV34TCL_LIGHT_SPOT_CUTOFF_A__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_CUTOFF_B(x) (0x00001204+((x)*64)) -#define NV34TCL_LIGHT_SPOT_CUTOFF_B__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_CUTOFF_C(x) (0x00001208+((x)*64)) -#define NV34TCL_LIGHT_SPOT_CUTOFF_C__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_DIR_X(x) (0x0000120c+((x)*64)) -#define NV34TCL_LIGHT_SPOT_DIR_X__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_DIR_Y(x) (0x00001210+((x)*64)) -#define NV34TCL_LIGHT_SPOT_DIR_Y__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_DIR_Z(x) (0x00001214+((x)*64)) -#define NV34TCL_LIGHT_SPOT_DIR_Z__SIZE 0x00000008 -#define NV34TCL_LIGHT_SPOT_CUTOFF_D(x) (0x00001218+((x)*64)) -#define NV34TCL_LIGHT_SPOT_CUTOFF_D__SIZE 0x00000008 -#define NV34TCL_LIGHT_POSITION_X(x) (0x0000121c+((x)*64)) -#define NV34TCL_LIGHT_POSITION_X__SIZE 0x00000008 -#define NV34TCL_LIGHT_POSITION_Y(x) (0x00001220+((x)*64)) -#define NV34TCL_LIGHT_POSITION_Y__SIZE 0x00000008 -#define NV34TCL_LIGHT_POSITION_Z(x) (0x00001224+((x)*64)) -#define NV34TCL_LIGHT_POSITION_Z__SIZE 0x00000008 -#define NV34TCL_LIGHT_ATTENUATION_CONSTANT(x) (0x00001228+((x)*64)) -#define NV34TCL_LIGHT_ATTENUATION_CONSTANT__SIZE 0x00000008 -#define NV34TCL_LIGHT_ATTENUATION_LINEAR(x) (0x0000122c+((x)*64)) -#define NV34TCL_LIGHT_ATTENUATION_LINEAR__SIZE 0x00000008 -#define NV34TCL_LIGHT_ATTENUATION_QUADRATIC(x) (0x00001230+((x)*64)) -#define NV34TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 -#define NV34TCL_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) -#define NV34TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34TCL_FP_REG_CONTROL 0x00001450 -#define NV34TCL_FP_REG_CONTROL_UNK1_SHIFT 16 -#define NV34TCL_FP_REG_CONTROL_UNK1_MASK 0xffff0000 -#define NV34TCL_FP_REG_CONTROL_UNK0_SHIFT 0 -#define NV34TCL_FP_REG_CONTROL_UNK0_MASK 0x0000ffff -#define NV34TCL_VP_CLIP_PLANES_ENABLE 0x00001478 -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE0 (1 << 1) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE1 (1 << 5) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE2 (1 << 9) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE3 (1 << 13) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE4 (1 << 17) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE5 (1 << 21) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE6 (1 << 25) -#define NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE7 (1 << 29) -#define NV34TCL_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV34TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV34TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV34TCL_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) -#define NV34TCL_VERTEX_ATTR_3F_X__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) -#define NV34TCL_VERTEX_ATTR_3F_Y__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) -#define NV34TCL_VERTEX_ATTR_3F_Z__SIZE 0x00000010 -#define NV34TCL_VP_CLIP_PLANE_A(x) (0x00001600+((x)*16)) -#define NV34TCL_VP_CLIP_PLANE_A__SIZE 0x00000006 -#define NV34TCL_VP_CLIP_PLANE_B(x) (0x00001604+((x)*16)) -#define NV34TCL_VP_CLIP_PLANE_B__SIZE 0x00000006 -#define NV34TCL_VP_CLIP_PLANE_C(x) (0x00001608+((x)*16)) -#define NV34TCL_VP_CLIP_PLANE_C__SIZE 0x00000006 -#define NV34TCL_VP_CLIP_PLANE_D(x) (0x0000160c+((x)*16)) -#define NV34TCL_VP_CLIP_PLANE_D__SIZE 0x00000006 -#define NV34TCL_VERTEX_BUFFER_ADDRESS(x) (0x00001680+((x)*4)) -#define NV34TCL_VERTEX_BUFFER_ADDRESS__SIZE 0x00000010 -#define NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1 (1 << 31) -#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_SHIFT 0 -#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_MASK 0x0fffffff -#define NV34TCL_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) -#define NV34TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT 0x00000002 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE 0x00000004 -#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT 4 -#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_MASK 0x000000f0 -#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 -#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000017a0 -#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000017a4 -#define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000017a8 -#define NV34TCL_COLOR_MATERIAL_BACK_R 0x000017b0 -#define NV34TCL_COLOR_MATERIAL_BACK_G 0x000017b4 -#define NV34TCL_COLOR_MATERIAL_BACK_B 0x000017b8 -#define NV34TCL_COLOR_MATERIAL_BACK_A 0x000017c0 -#define NV34TCL_QUERY_RESET 0x000017c8 -#define NV34TCL_QUERY_UNK17CC 0x000017cc -#define NV34TCL_QUERY_GET 0x00001800 -#define NV34TCL_QUERY_GET_UNK24_SHIFT 24 -#define NV34TCL_QUERY_GET_UNK24_MASK 0xff000000 -#define NV34TCL_QUERY_GET_OFFSET_SHIFT 0 -#define NV34TCL_QUERY_GET_OFFSET_MASK 0x00ffffff -#define NV34TCL_VERTEX_BEGIN_END 0x00001808 -#define NV34TCL_VERTEX_BEGIN_END_STOP 0x00000000 -#define NV34TCL_VERTEX_BEGIN_END_POINTS 0x00000001 -#define NV34TCL_VERTEX_BEGIN_END_LINES 0x00000002 -#define NV34TCL_VERTEX_BEGIN_END_LINE_LOOP 0x00000003 -#define NV34TCL_VERTEX_BEGIN_END_LINE_STRIP 0x00000004 -#define NV34TCL_VERTEX_BEGIN_END_TRIANGLES 0x00000005 -#define NV34TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV34TCL_VERTEX_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV34TCL_VERTEX_BEGIN_END_QUADS 0x00000008 -#define NV34TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV34TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV34TCL_VB_ELEMENT_U16 0x0000180c -#define NV34TCL_VB_ELEMENT_U16_I0_SHIFT 0 -#define NV34TCL_VB_ELEMENT_U16_I0_MASK 0x0000ffff -#define NV34TCL_VB_ELEMENT_U16_I1_SHIFT 16 -#define NV34TCL_VB_ELEMENT_U16_I1_MASK 0xffff0000 -#define NV34TCL_VB_ELEMENT_U32 0x00001810 -#define NV34TCL_VB_VERTEX_BATCH 0x00001814 -#define NV34TCL_VB_VERTEX_BATCH_OFFSET_SHIFT 0 -#define NV34TCL_VB_VERTEX_BATCH_OFFSET_MASK 0x00ffffff -#define NV34TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 -#define NV34TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 -#define NV34TCL_VERTEX_DATA 0x00001818 -#define NV34TCL_POLYGON_MODE_FRONT 0x00001828 -#define NV34TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV34TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV34TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV34TCL_POLYGON_MODE_BACK 0x0000182c -#define NV34TCL_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV34TCL_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV34TCL_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV34TCL_CULL_FACE 0x00001830 -#define NV34TCL_CULL_FACE_FRONT 0x00000404 -#define NV34TCL_CULL_FACE_BACK 0x00000405 -#define NV34TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV34TCL_FRONT_FACE 0x00001834 -#define NV34TCL_FRONT_FACE_CW 0x00000900 -#define NV34TCL_FRONT_FACE_CCW 0x00000901 -#define NV34TCL_POLYGON_SMOOTH_ENABLE 0x00001838 -#define NV34TCL_CULL_FACE_ENABLE 0x0000183c -#define NV34TCL_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) -#define NV34TCL_VERTEX_ATTR_2F_X__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) -#define NV34TCL_VERTEX_ATTR_2F_Y__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_2I(x) (0x00001900+((x)*4)) -#define NV34TCL_VERTEX_ATTR_2I__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_2I_Y_SHIFT 16 -#define NV34TCL_VERTEX_ATTR_2I_Y_MASK 0xffff0000 -#define NV34TCL_VERTEX_ATTR_2I_X_SHIFT 0 -#define NV34TCL_VERTEX_ATTR_2I_X_MASK 0x0000ffff -#define NV34TCL_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) -#define NV34TCL_VERTEX_COL_4I__SIZE 0x00000002 -#define NV34TCL_VERTEX_COL_4I_R_SHIFT 0 -#define NV34TCL_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV34TCL_VERTEX_COL_4I_G_SHIFT 8 -#define NV34TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV34TCL_VERTEX_COL_4I_B_SHIFT 16 -#define NV34TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV34TCL_VERTEX_COL_4I_A_SHIFT 24 -#define NV34TCL_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV34TCL_VERTEX_POS_4I_XY 0x00001980 -#define NV34TCL_VERTEX_POS_4I_XY_X_SHIFT 0 -#define NV34TCL_VERTEX_POS_4I_XY_X_MASK 0x0000ffff -#define NV34TCL_VERTEX_POS_4I_XY_Y_SHIFT 16 -#define NV34TCL_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 -#define NV34TCL_VERTEX_POS_4I_ZW 0x00001984 -#define NV34TCL_VERTEX_POS_4I_ZW_Z_SHIFT 0 -#define NV34TCL_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff -#define NV34TCL_VERTEX_POS_4I_ZW_W_SHIFT 16 -#define NV34TCL_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 -#define NV34TCL_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) -#define NV34TCL_VERTEX_TX_4I_ST__SIZE 0x00000004 -#define NV34TCL_VERTEX_TX_4I_ST_S_SHIFT 0 -#define NV34TCL_VERTEX_TX_4I_ST_S_MASK 0x0000ffff -#define NV34TCL_VERTEX_TX_4I_ST_T_SHIFT 16 -#define NV34TCL_VERTEX_TX_4I_ST_T_MASK 0xffff0000 -#define NV34TCL_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) -#define NV34TCL_VERTEX_TX_4I_RQ__SIZE 0x00000004 -#define NV34TCL_VERTEX_TX_4I_RQ_R_SHIFT 0 -#define NV34TCL_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff -#define NV34TCL_VERTEX_TX_4I_RQ_Q_SHIFT 16 -#define NV34TCL_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 -#define NV34TCL_TX_OFFSET(x) (0x00001a00+((x)*32)) -#define NV34TCL_TX_OFFSET__SIZE 0x00000004 -#define NV34TCL_TX_FORMAT(x) (0x00001a04+((x)*32)) -#define NV34TCL_TX_FORMAT__SIZE 0x00000004 -#define NV34TCL_TX_FORMAT_DMA0 (1 << 0) -#define NV34TCL_TX_FORMAT_DMA1 (1 << 1) -#define NV34TCL_TX_FORMAT_CUBE_MAP (1 << 2) -#define NV34TCL_TX_FORMAT_DIMS_SHIFT 4 -#define NV34TCL_TX_FORMAT_DIMS_MASK 0x000000f0 -#define NV34TCL_TX_FORMAT_DIMS_1D 0x00000010 -#define NV34TCL_TX_FORMAT_DIMS_2D 0x00000020 -#define NV34TCL_TX_FORMAT_DIMS_3D 0x00000030 -#define NV34TCL_TX_FORMAT_FORMAT_SHIFT 8 -#define NV34TCL_TX_FORMAT_FORMAT_MASK 0x0000ff00 -#define NV34TCL_TX_FORMAT_FORMAT_L8 0x00000000 -#define NV34TCL_TX_FORMAT_FORMAT_A8 0x00000100 -#define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000200 -#define NV34TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000300 -#define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000400 -#define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000600 -#define NV34TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000700 -#define NV34TCL_TX_FORMAT_FORMAT_INDEX8 0x00000b00 -#define NV34TCL_TX_FORMAT_FORMAT_DXT1 0x00000c00 -#define NV34TCL_TX_FORMAT_FORMAT_DXT3 0x00000e00 -#define NV34TCL_TX_FORMAT_FORMAT_DXT5 0x00000f00 -#define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00001000 -#define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00001200 -#define NV34TCL_TX_FORMAT_FORMAT_L8_RECT 0x00001300 -#define NV34TCL_TX_FORMAT_FORMAT_A8L8 0x00001a00 -#define NV34TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00001b00 -#define NV34TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 -#define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00001d00 -#define NV34TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00002000 -#define NV34TCL_TX_FORMAT_FORMAT_A16 0x00003200 -#define NV34TCL_TX_FORMAT_FORMAT_A16_RECT 0x00003500 -#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00004a00 -#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00004b00 -#define NV34TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00004c00 -#define NV34TCL_TX_FORMAT_NPOT (1 << 12) -#define NV34TCL_TX_FORMAT_RECT (1 << 14) -#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 -#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 -#define NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 20 -#define NV34TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 -#define NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 24 -#define NV34TCL_TX_FORMAT_BASE_SIZE_V_MASK 0x0f000000 -#define NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT 28 -#define NV34TCL_TX_FORMAT_BASE_SIZE_W_MASK 0xf0000000 -#define NV34TCL_TX_WRAP(x) (0x00001a08+((x)*32)) -#define NV34TCL_TX_WRAP__SIZE 0x00000004 -#define NV34TCL_TX_WRAP_S_SHIFT 0 -#define NV34TCL_TX_WRAP_S_MASK 0x000000ff -#define NV34TCL_TX_WRAP_S_REPEAT 0x00000001 -#define NV34TCL_TX_WRAP_S_MIRRORED_REPEAT 0x00000002 -#define NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE 0x00000003 -#define NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 -#define NV34TCL_TX_WRAP_S_CLAMP 0x00000005 -#define NV34TCL_TX_WRAP_T_SHIFT 8 -#define NV34TCL_TX_WRAP_T_MASK 0x0000ff00 -#define NV34TCL_TX_WRAP_T_REPEAT 0x00000100 -#define NV34TCL_TX_WRAP_T_MIRRORED_REPEAT 0x00000200 -#define NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE 0x00000300 -#define NV34TCL_TX_WRAP_T_CLAMP_TO_BORDER 0x00000400 -#define NV34TCL_TX_WRAP_T_CLAMP 0x00000500 -#define NV34TCL_TX_WRAP_R_SHIFT 16 -#define NV34TCL_TX_WRAP_R_MASK 0x00ff0000 -#define NV34TCL_TX_WRAP_R_REPEAT 0x00010000 -#define NV34TCL_TX_WRAP_R_MIRRORED_REPEAT 0x00020000 -#define NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE 0x00030000 -#define NV34TCL_TX_WRAP_R_CLAMP_TO_BORDER 0x00040000 -#define NV34TCL_TX_WRAP_R_CLAMP 0x00050000 -#define NV34TCL_TX_ENABLE(x) (0x00001a0c+((x)*32)) -#define NV34TCL_TX_ENABLE__SIZE 0x00000004 -#define NV34TCL_TX_ENABLE_ENABLE (1 << 30) -#define NV34TCL_TX_SWIZZLE(x) (0x00001a10+((x)*32)) -#define NV34TCL_TX_SWIZZLE__SIZE 0x00000004 -#define NV34TCL_TX_SWIZZLE_S0_X_SHIFT 14 -#define NV34TCL_TX_SWIZZLE_S0_X_MASK 0x0000c000 -#define NV34TCL_TX_SWIZZLE_S0_X_ZERO 0x00000000 -#define NV34TCL_TX_SWIZZLE_S0_X_ONE 0x00004000 -#define NV34TCL_TX_SWIZZLE_S0_X_S1 0x00008000 -#define NV34TCL_TX_SWIZZLE_S0_Y_SHIFT 12 -#define NV34TCL_TX_SWIZZLE_S0_Y_MASK 0x00003000 -#define NV34TCL_TX_SWIZZLE_S0_Y_ZERO 0x00000000 -#define NV34TCL_TX_SWIZZLE_S0_Y_ONE 0x00001000 -#define NV34TCL_TX_SWIZZLE_S0_Y_S1 0x00002000 -#define NV34TCL_TX_SWIZZLE_S0_Z_SHIFT 10 -#define NV34TCL_TX_SWIZZLE_S0_Z_MASK 0x00000c00 -#define NV34TCL_TX_SWIZZLE_S0_Z_ZERO 0x00000000 -#define NV34TCL_TX_SWIZZLE_S0_Z_ONE 0x00000400 -#define NV34TCL_TX_SWIZZLE_S0_Z_S1 0x00000800 -#define NV34TCL_TX_SWIZZLE_S0_W_SHIFT 8 -#define NV34TCL_TX_SWIZZLE_S0_W_MASK 0x00000300 -#define NV34TCL_TX_SWIZZLE_S0_W_ZERO 0x00000000 -#define NV34TCL_TX_SWIZZLE_S0_W_ONE 0x00000100 -#define NV34TCL_TX_SWIZZLE_S0_W_S1 0x00000200 -#define NV34TCL_TX_SWIZZLE_S1_X_SHIFT 6 -#define NV34TCL_TX_SWIZZLE_S1_X_MASK 0x000000c0 -#define NV34TCL_TX_SWIZZLE_S1_X_W 0x00000000 -#define NV34TCL_TX_SWIZZLE_S1_X_Z 0x00000040 -#define NV34TCL_TX_SWIZZLE_S1_X_Y 0x00000080 -#define NV34TCL_TX_SWIZZLE_S1_X_X 0x000000c0 -#define NV34TCL_TX_SWIZZLE_S1_Y_SHIFT 4 -#define NV34TCL_TX_SWIZZLE_S1_Y_MASK 0x00000030 -#define NV34TCL_TX_SWIZZLE_S1_Y_W 0x00000000 -#define NV34TCL_TX_SWIZZLE_S1_Y_Z 0x00000010 -#define NV34TCL_TX_SWIZZLE_S1_Y_Y 0x00000020 -#define NV34TCL_TX_SWIZZLE_S1_Y_X 0x00000030 -#define NV34TCL_TX_SWIZZLE_S1_Z_SHIFT 2 -#define NV34TCL_TX_SWIZZLE_S1_Z_MASK 0x0000000c -#define NV34TCL_TX_SWIZZLE_S1_Z_W 0x00000000 -#define NV34TCL_TX_SWIZZLE_S1_Z_Z 0x00000004 -#define NV34TCL_TX_SWIZZLE_S1_Z_Y 0x00000008 -#define NV34TCL_TX_SWIZZLE_S1_Z_X 0x0000000c -#define NV34TCL_TX_SWIZZLE_S1_W_SHIFT 0 -#define NV34TCL_TX_SWIZZLE_S1_W_MASK 0x00000003 -#define NV34TCL_TX_SWIZZLE_S1_W_W 0x00000000 -#define NV34TCL_TX_SWIZZLE_S1_W_Z 0x00000001 -#define NV34TCL_TX_SWIZZLE_S1_W_Y 0x00000002 -#define NV34TCL_TX_SWIZZLE_S1_W_X 0x00000003 -#define NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT 16 -#define NV34TCL_TX_SWIZZLE_RECT_PITCH_MASK 0xffff0000 -#define NV34TCL_TX_FILTER(x) (0x00001a14+((x)*32)) -#define NV34TCL_TX_FILTER__SIZE 0x00000004 -#define NV34TCL_TX_FILTER_MINIFY_SHIFT 16 -#define NV34TCL_TX_FILTER_MINIFY_MASK 0x000f0000 -#define NV34TCL_TX_FILTER_MINIFY_NEAREST 0x00010000 -#define NV34TCL_TX_FILTER_MINIFY_LINEAR 0x00020000 -#define NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST 0x00030000 -#define NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST 0x00040000 -#define NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR 0x00050000 -#define NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR 0x00060000 -#define NV34TCL_TX_FILTER_MAGNIFY_SHIFT 24 -#define NV34TCL_TX_FILTER_MAGNIFY_MASK 0x0f000000 -#define NV34TCL_TX_FILTER_MAGNIFY_NEAREST 0x01000000 -#define NV34TCL_TX_FILTER_MAGNIFY_LINEAR 0x02000000 -#define NV34TCL_TX_FILTER_SIGNED_BLUE (1 << 28) -#define NV34TCL_TX_FILTER_SIGNED_GREEN (1 << 29) -#define NV34TCL_TX_FILTER_SIGNED_RED (1 << 30) -#define NV34TCL_TX_FILTER_SIGNED_ALPHA (1 << 31) -#define NV34TCL_TX_NPOT_SIZE(x) (0x00001a18+((x)*32)) -#define NV34TCL_TX_NPOT_SIZE__SIZE 0x00000004 -#define NV34TCL_TX_NPOT_SIZE_H_SHIFT 0 -#define NV34TCL_TX_NPOT_SIZE_H_MASK 0x0000ffff -#define NV34TCL_TX_NPOT_SIZE_W_SHIFT 16 -#define NV34TCL_TX_NPOT_SIZE_W_MASK 0xffff0000 -#define NV34TCL_TX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) -#define NV34TCL_TX_BORDER_COLOR__SIZE 0x00000004 -#define NV34TCL_TX_BORDER_COLOR_B_SHIFT 0 -#define NV34TCL_TX_BORDER_COLOR_B_MASK 0x000000ff -#define NV34TCL_TX_BORDER_COLOR_G_SHIFT 8 -#define NV34TCL_TX_BORDER_COLOR_G_MASK 0x0000ff00 -#define NV34TCL_TX_BORDER_COLOR_R_SHIFT 16 -#define NV34TCL_TX_BORDER_COLOR_R_MASK 0x00ff0000 -#define NV34TCL_TX_BORDER_COLOR_A_SHIFT 24 -#define NV34TCL_TX_BORDER_COLOR_A_MASK 0xff000000 -#define NV34TCL_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_X__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_Y__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_Z__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_W__SIZE 0x00000010 -#define NV34TCL_FP_CONTROL 0x00001d60 -#define NV34TCL_FP_CONTROL_USES_KIL (1 << 7) -#define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_SHIFT 0 -#define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_MASK 0x0000000f -#define NV34TCL_MULTISAMPLE_CONTROL 0x00001d7c -#define NV34TCL_CLEAR_DEPTH_VALUE 0x00001d8c -#define NV34TCL_CLEAR_COLOR_VALUE 0x00001d90 -#define NV34TCL_CLEAR_COLOR_VALUE_B_SHIFT 0 -#define NV34TCL_CLEAR_COLOR_VALUE_B_MASK 0x000000ff -#define NV34TCL_CLEAR_COLOR_VALUE_G_SHIFT 8 -#define NV34TCL_CLEAR_COLOR_VALUE_G_MASK 0x0000ff00 -#define NV34TCL_CLEAR_COLOR_VALUE_R_SHIFT 16 -#define NV34TCL_CLEAR_COLOR_VALUE_R_MASK 0x00ff0000 -#define NV34TCL_CLEAR_COLOR_VALUE_A_SHIFT 24 -#define NV34TCL_CLEAR_COLOR_VALUE_A_MASK 0xff000000 -#define NV34TCL_CLEAR_BUFFERS 0x00001d94 -#define NV34TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) -#define NV34TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) -#define NV34TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) -#define NV34TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) -#define NV34TCL_CLEAR_BUFFERS_STENCIL (1 << 1) -#define NV34TCL_CLEAR_BUFFERS_DEPTH (1 << 0) -#define NV34TCL_DO_VERTICES 0x00001dac -#define NV34TCL_LINE_STIPPLE_ENABLE 0x00001db4 -#define NV34TCL_LINE_STIPPLE_PATTERN 0x00001db8 -#define NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 -#define NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff -#define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 -#define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 -#define NV34TCL_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) -#define NV34TCL_BACK_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34TCL_VERTEX_FOG_1F 0x00001e54 -#define NV34TCL_VP_UPLOAD_FROM_ID 0x00001e9c -#define NV34TCL_VP_START_FROM_ID 0x00001ea0 -#define NV34TCL_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) -#define NV34TCL_POINT_PARAMETERS__SIZE 0x00000008 -#define NV34TCL_POINT_SIZE 0x00001ee0 -#define NV34TCL_POINT_PARAMETERS_ENABLE 0x00001ee4 -#define NV34TCL_POINT_SPRITE 0x00001ee8 -#define NV34TCL_POINT_SPRITE_ENABLE (1 << 0) -#define NV34TCL_POINT_SPRITE_R_MODE_SHIFT 1 -#define NV34TCL_POINT_SPRITE_R_MODE_MASK 0x00000006 -#define NV34TCL_POINT_SPRITE_R_MODE_GL_ZERO 0x00000000 -#define NV34TCL_POINT_SPRITE_R_MODE_GL_R 0x00000002 -#define NV34TCL_POINT_SPRITE_R_MODE_GL_S 0x00000004 -#define NV34TCL_POINT_SPRITE_COORD_REPLACE (1 << 11) -#define NV34TCL_VP_UPLOAD_CONST_ID 0x00001efc -#define NV34TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) -#define NV34TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 -#define NV34TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) -#define NV34TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 -#define NV34TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) -#define NV34TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 -#define NV34TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) -#define NV34TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 -#define NV34TCL_UNK1f80(x) (0x00001f80+((x)*4)) -#define NV34TCL_UNK1f80__SIZE 0x00000010 -#define NV34TCL_VP_ATTRIB_EN 0x00001ff0 -#define NV34TCL_VP_RESULT_EN 0x00001ff4 - - -#define NV40_CONTEXT_SURFACES_2D 0x00003062 - - - -#define NV40_STRETCHED_IMAGE_FROM_CPU 0x00003066 - - - -#define NV40_TEXTURE_FROM_CPU 0x0000307b - - - -#define NV40_SCALED_IMAGE_FROM_MEMORY 0x00003089 - - - -#define NV40_IMAGE_FROM_CPU 0x0000308a - - - -#define NV40_SWIZZLED_SURFACE 0x0000309e - - - -#define NV40TCL 0x00004097 - -#define NV40TCL_REF_CNT 0x00000050 -#define NV40TCL_NOP 0x00000100 -#define NV40TCL_NOTIFY 0x00000104 -#define NV40TCL_DMA_NOTIFY 0x00000180 -#define NV40TCL_DMA_TEXTURE0 0x00000184 -#define NV40TCL_DMA_TEXTURE1 0x00000188 -#define NV40TCL_DMA_COLOR1 0x0000018c -#define NV40TCL_DMA_COLOR0 0x00000194 -#define NV40TCL_DMA_ZETA 0x00000198 -#define NV40TCL_DMA_VTXBUF0 0x0000019c -#define NV40TCL_DMA_VTXBUF1 0x000001a0 -#define NV40TCL_DMA_FENCE 0x000001a4 -#define NV40TCL_DMA_QUERY 0x000001a8 -#define NV40TCL_DMA_UNK01AC 0x000001ac -#define NV40TCL_DMA_UNK01B0 0x000001b0 -#define NV40TCL_DMA_COLOR2 0x000001b4 -#define NV40TCL_DMA_COLOR3 0x000001b8 -#define NV40TCL_RT_HORIZ 0x00000200 -#define NV40TCL_RT_HORIZ_W_SHIFT 16 -#define NV40TCL_RT_HORIZ_W_MASK 0xffff0000 -#define NV40TCL_RT_HORIZ_X_SHIFT 0 -#define NV40TCL_RT_HORIZ_X_MASK 0x0000ffff -#define NV40TCL_RT_VERT 0x00000204 -#define NV40TCL_RT_VERT_H_SHIFT 16 -#define NV40TCL_RT_VERT_H_MASK 0xffff0000 -#define NV40TCL_RT_VERT_Y_SHIFT 0 -#define NV40TCL_RT_VERT_Y_MASK 0x0000ffff -#define NV40TCL_RT_FORMAT 0x00000208 -#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT 24 -#define NV40TCL_RT_FORMAT_LOG2_HEIGHT_MASK 0xff000000 -#define NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT 16 -#define NV40TCL_RT_FORMAT_LOG2_WIDTH_MASK 0x00ff0000 -#define NV40TCL_RT_FORMAT_TYPE_SHIFT 8 -#define NV40TCL_RT_FORMAT_TYPE_MASK 0x00000f00 -#define NV40TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 -#define NV40TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 -#define NV40TCL_RT_FORMAT_ZETA_SHIFT 5 -#define NV40TCL_RT_FORMAT_ZETA_MASK 0x000000e0 -#define NV40TCL_RT_FORMAT_ZETA_Z16 0x00000020 -#define NV40TCL_RT_FORMAT_ZETA_Z24S8 0x00000040 -#define NV40TCL_RT_FORMAT_COLOR_SHIFT 0 -#define NV40TCL_RT_FORMAT_COLOR_MASK 0x0000001f -#define NV40TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 -#define NV40TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 -#define NV40TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 -#define NV40TCL_RT_FORMAT_COLOR_B8 0x00000009 -#define NV40TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f -#define NV40TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 -#define NV40TCL_COLOR0_PITCH 0x0000020c -#define NV40TCL_COLOR0_OFFSET 0x00000210 -#define NV40TCL_ZETA_OFFSET 0x00000214 -#define NV40TCL_COLOR1_OFFSET 0x00000218 -#define NV40TCL_COLOR1_PITCH 0x0000021c -#define NV40TCL_RT_ENABLE 0x00000220 -#define NV40TCL_RT_ENABLE_MRT (1 << 4) -#define NV40TCL_RT_ENABLE_COLOR3 (1 << 3) -#define NV40TCL_RT_ENABLE_COLOR2 (1 << 2) -#define NV40TCL_RT_ENABLE_COLOR1 (1 << 1) -#define NV40TCL_RT_ENABLE_COLOR0 (1 << 0) -#define NV40TCL_ZETA_PITCH 0x0000022c -#define NV40TCL_COLOR2_PITCH 0x00000280 -#define NV40TCL_COLOR3_PITCH 0x00000284 -#define NV40TCL_COLOR2_OFFSET 0x00000288 -#define NV40TCL_COLOR3_OFFSET 0x0000028c -#define NV40TCL_VIEWPORT_CLIP_HORIZ(x) (0x000002c0+((x)*8)) -#define NV40TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV40TCL_VIEWPORT_CLIP_VERT(x) (0x000002c4+((x)*8)) -#define NV40TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV40TCL_DITHER_ENABLE 0x00000300 -#define NV40TCL_ALPHA_TEST_ENABLE 0x00000304 -#define NV40TCL_ALPHA_TEST_FUNC 0x00000308 -#define NV40TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 -#define NV40TCL_ALPHA_TEST_FUNC_LESS 0x00000201 -#define NV40TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 -#define NV40TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 -#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 -#define NV40TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 -#define NV40TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 -#define NV40TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 -#define NV40TCL_ALPHA_TEST_REF 0x0000030c -#define NV40TCL_BLEND_ENABLE 0x00000310 -#define NV40TCL_BLEND_FUNC_SRC 0x00000314 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SHIFT 0 -#define NV40TCL_BLEND_FUNC_SRC_RGB_MASK 0x0000ffff -#define NV40TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV40TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SHIFT 16 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_MASK 0xffff0000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00010000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x03000000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x03020000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x03040000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x03060000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x03080000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x80010000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x80030000 -#define NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV40TCL_BLEND_FUNC_DST 0x00000318 -#define NV40TCL_BLEND_FUNC_DST_RGB_SHIFT 0 -#define NV40TCL_BLEND_FUNC_DST_RGB_MASK 0x0000ffff -#define NV40TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV40TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV40TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SHIFT 16 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_MASK 0xffff0000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00010000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x03000000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x03010000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x03020000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x03030000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x03040000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x03050000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x03060000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x03070000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x03080000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x80010000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 -#define NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV40TCL_BLEND_COLOR 0x0000031c -#define NV40TCL_BLEND_EQUATION 0x00000320 -#define NV40TCL_BLEND_EQUATION_RGB_SHIFT 0 -#define NV40TCL_BLEND_EQUATION_RGB_MASK 0x0000ffff -#define NV40TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 -#define NV40TCL_BLEND_EQUATION_RGB_MIN 0x00008007 -#define NV40TCL_BLEND_EQUATION_RGB_MAX 0x00008008 -#define NV40TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a -#define NV40TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV40TCL_BLEND_EQUATION_ALPHA_SHIFT 16 -#define NV40TCL_BLEND_EQUATION_ALPHA_MASK 0xffff0000 -#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x80060000 -#define NV40TCL_BLEND_EQUATION_ALPHA_MIN 0x80070000 -#define NV40TCL_BLEND_EQUATION_ALPHA_MAX 0x80080000 -#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x800a0000 -#define NV40TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x800b0000 -#define NV40TCL_COLOR_MASK 0x00000324 -#define NV40TCL_COLOR_MASK_BUFFER0_B_SHIFT 0 -#define NV40TCL_COLOR_MASK_BUFFER0_B_MASK 0x000000ff -#define NV40TCL_COLOR_MASK_BUFFER0_G_SHIFT 8 -#define NV40TCL_COLOR_MASK_BUFFER0_G_MASK 0x0000ff00 -#define NV40TCL_COLOR_MASK_BUFFER0_R_SHIFT 16 -#define NV40TCL_COLOR_MASK_BUFFER0_R_MASK 0x00ff0000 -#define NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT 24 -#define NV40TCL_COLOR_MASK_BUFFER0_A_MASK 0xff000000 -#define NV40TCL_STENCIL_FRONT_ENABLE 0x00000328 -#define NV40TCL_STENCIL_FRONT_MASK 0x0000032c -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC 0x00000330 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV40TCL_STENCIL_FRONT_FUNC_REF 0x00000334 -#define NV40TCL_STENCIL_FRONT_FUNC_MASK 0x00000338 -#define NV40TCL_STENCIL_FRONT_OP_FAIL 0x0000033c -#define NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL 0x00000340 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS 0x00000344 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_ENABLE 0x00000348 -#define NV40TCL_STENCIL_BACK_MASK 0x0000034c -#define NV40TCL_STENCIL_BACK_FUNC_FUNC 0x00000350 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV40TCL_STENCIL_BACK_FUNC_REF 0x00000354 -#define NV40TCL_STENCIL_BACK_FUNC_MASK 0x00000358 -#define NV40TCL_STENCIL_BACK_OP_FAIL 0x0000035c -#define NV40TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL 0x00000360 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV40TCL_STENCIL_BACK_OP_ZPASS 0x00000364 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV40TCL_SHADE_MODEL 0x00000368 -#define NV40TCL_SHADE_MODEL_FLAT 0x00001d00 -#define NV40TCL_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV40TCL_MRT_COLOR_MASK 0x00000370 -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_A (1 << 4) -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_R (1 << 5) -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_G (1 << 6) -#define NV40TCL_MRT_COLOR_MASK_BUFFER1_B (1 << 7) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_A (1 << 8) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_R (1 << 9) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_G (1 << 10) -#define NV40TCL_MRT_COLOR_MASK_BUFFER2_B (1 << 11) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_A (1 << 12) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_R (1 << 13) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_G (1 << 14) -#define NV40TCL_MRT_COLOR_MASK_BUFFER3_B (1 << 15) -#define NV40TCL_COLOR_LOGIC_OP_ENABLE 0x00000374 -#define NV40TCL_COLOR_LOGIC_OP 0x00000378 -#define NV40TCL_COLOR_LOGIC_OP_CLEAR 0x00001500 -#define NV40TCL_COLOR_LOGIC_OP_AND 0x00001501 -#define NV40TCL_COLOR_LOGIC_OP_AND_REVERSE 0x00001502 -#define NV40TCL_COLOR_LOGIC_OP_COPY 0x00001503 -#define NV40TCL_COLOR_LOGIC_OP_AND_INVERTED 0x00001504 -#define NV40TCL_COLOR_LOGIC_OP_NOOP 0x00001505 -#define NV40TCL_COLOR_LOGIC_OP_XOR 0x00001506 -#define NV40TCL_COLOR_LOGIC_OP_OR 0x00001507 -#define NV40TCL_COLOR_LOGIC_OP_NOR 0x00001508 -#define NV40TCL_COLOR_LOGIC_OP_EQUIV 0x00001509 -#define NV40TCL_COLOR_LOGIC_OP_INVERT 0x0000150a -#define NV40TCL_COLOR_LOGIC_OP_OR_REVERSE 0x0000150b -#define NV40TCL_COLOR_LOGIC_OP_COPY_INVERTED 0x0000150c -#define NV40TCL_COLOR_LOGIC_OP_OR_INVERTED 0x0000150d -#define NV40TCL_COLOR_LOGIC_OP_NAND 0x0000150e -#define NV40TCL_COLOR_LOGIC_OP_SET 0x0000150f -#define NV40TCL_DEPTH_RANGE_NEAR 0x00000394 -#define NV40TCL_DEPTH_RANGE_FAR 0x00000398 -#define NV40TCL_LINE_WIDTH 0x000003b8 -#define NV40TCL_LINE_SMOOTH_ENABLE 0x000003bc -#define NV40TCL_UNK03C0(x) (0x000003c0+((x)*4)) -#define NV40TCL_UNK03C0__SIZE 0x00000010 -#define NV40TCL_UNK0400(x) (0x00000400+((x)*4)) -#define NV40TCL_UNK0400__SIZE 0x00000010 -#define NV40TCL_UNK0440(x) (0x00000440+((x)*4)) -#define NV40TCL_UNK0440__SIZE 0x00000020 -#define NV40TCL_SCISSOR_HORIZ 0x000008c0 -#define NV40TCL_SCISSOR_HORIZ_X_SHIFT 0 -#define NV40TCL_SCISSOR_HORIZ_X_MASK 0x0000ffff -#define NV40TCL_SCISSOR_HORIZ_W_SHIFT 16 -#define NV40TCL_SCISSOR_HORIZ_W_MASK 0xffff0000 -#define NV40TCL_SCISSOR_VERT 0x000008c4 -#define NV40TCL_SCISSOR_VERT_Y_SHIFT 0 -#define NV40TCL_SCISSOR_VERT_Y_MASK 0x0000ffff -#define NV40TCL_SCISSOR_VERT_H_SHIFT 16 -#define NV40TCL_SCISSOR_VERT_H_MASK 0xffff0000 -#define NV40TCL_FOG_MODE 0x000008cc -#define NV40TCL_FOG_EQUATION_CONSTANT 0x000008d0 -#define NV40TCL_FOG_EQUATION_LINEAR 0x000008d4 -#define NV40TCL_FOG_EQUATION_QUADRATIC 0x000008d8 -#define NV40TCL_FP_ADDRESS 0x000008e4 -#define NV40TCL_FP_ADDRESS_OFFSET_SHIFT 8 -#define NV40TCL_FP_ADDRESS_OFFSET_MASK 0xffffff00 -#define NV40TCL_FP_ADDRESS_DMA1 (1 << 1) -#define NV40TCL_FP_ADDRESS_DMA0 (1 << 0) -#define NV40TCL_VIEWPORT_HORIZ 0x00000a00 -#define NV40TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV40TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV40TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV40TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV40TCL_VIEWPORT_VERT 0x00000a04 -#define NV40TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV40TCL_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV40TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV40TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV40TCL_VIEWPORT_TRANSLATE_X 0x00000a20 -#define NV40TCL_VIEWPORT_TRANSLATE_Y 0x00000a24 -#define NV40TCL_VIEWPORT_TRANSLATE_Z 0x00000a28 -#define NV40TCL_VIEWPORT_TRANSLATE_W 0x00000a2c -#define NV40TCL_VIEWPORT_SCALE_X 0x00000a30 -#define NV40TCL_VIEWPORT_SCALE_Y 0x00000a34 -#define NV40TCL_VIEWPORT_SCALE_Z 0x00000a38 -#define NV40TCL_VIEWPORT_SCALE_W 0x00000a3c -#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 -#define NV40TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 -#define NV40TCL_DEPTH_FUNC 0x00000a6c -#define NV40TCL_DEPTH_FUNC_NEVER 0x00000200 -#define NV40TCL_DEPTH_FUNC_LESS 0x00000201 -#define NV40TCL_DEPTH_FUNC_EQUAL 0x00000202 -#define NV40TCL_DEPTH_FUNC_LEQUAL 0x00000203 -#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV40TCL_DEPTH_FUNC_GREATER 0x00000204 -#define NV40TCL_DEPTH_FUNC_NOTEQUAL 0x00000205 -#define NV40TCL_DEPTH_FUNC_GEQUAL 0x00000206 -#define NV40TCL_DEPTH_FUNC_ALWAYS 0x00000207 -#define NV40TCL_DEPTH_WRITE_ENABLE 0x00000a70 -#define NV40TCL_DEPTH_TEST_ENABLE 0x00000a74 -#define NV40TCL_POLYGON_OFFSET_FACTOR 0x00000a78 -#define NV40TCL_POLYGON_OFFSET_UNITS 0x00000a7c -#define NV40TCL_UNK0B40(x) (0x00000b40+((x)*4)) -#define NV40TCL_UNK0B40__SIZE 0x00000008 -#define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) -#define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 -#define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) -#define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c -#define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) -#define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV40TCL_VTX_ATTR_3F_X(x) (0x00001500+((x)*16)) -#define NV40TCL_VTX_ATTR_3F_X__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_3F_Y(x) (0x00001504+((x)*16)) -#define NV40TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_3F_Z(x) (0x00001508+((x)*16)) -#define NV40TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 -#define NV40TCL_VTXBUF_ADDRESS(x) (0x00001680+((x)*4)) -#define NV40TCL_VTXBUF_ADDRESS__SIZE 0x00000010 -#define NV40TCL_VTXBUF_ADDRESS_DMA1 (1 << 31) -#define NV40TCL_VTXBUF_ADDRESS_OFFSET_SHIFT 0 -#define NV40TCL_VTXBUF_ADDRESS_OFFSET_MASK 0x0fffffff -#define NV40TCL_VTX_CACHE_INVALIDATE 0x00001714 -#define NV40TCL_VTXFMT(x) (0x00001740+((x)*4)) -#define NV40TCL_VTXFMT__SIZE 0x00000010 -#define NV40TCL_VTXFMT_TYPE_SHIFT 0 -#define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f -#define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 -#define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 -#define NV40TCL_VTXFMT_SIZE_SHIFT 4 -#define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 -#define NV40TCL_VTXFMT_STRIDE_SHIFT 8 -#define NV40TCL_VTXFMT_STRIDE_MASK 0x0000ff00 -#define NV40TCL_QUERY_RESET 0x000017c8 -#define NV40TCL_QUERY_UNK17CC 0x000017cc -#define NV40TCL_QUERY_GET 0x00001800 -#define NV40TCL_QUERY_GET_UNK24_SHIFT 24 -#define NV40TCL_QUERY_GET_UNK24_MASK 0xff000000 -#define NV40TCL_QUERY_GET_OFFSET_SHIFT 0 -#define NV40TCL_QUERY_GET_OFFSET_MASK 0x00ffffff -#define NV40TCL_BEGIN_END 0x00001808 -#define NV40TCL_BEGIN_END_STOP 0x00000000 -#define NV40TCL_BEGIN_END_POINTS 0x00000001 -#define NV40TCL_BEGIN_END_LINES 0x00000002 -#define NV40TCL_BEGIN_END_LINE_LOOP 0x00000003 -#define NV40TCL_BEGIN_END_LINE_STRIP 0x00000004 -#define NV40TCL_BEGIN_END_TRIANGLES 0x00000005 -#define NV40TCL_BEGIN_END_TRIANGLE_STRIP 0x00000006 -#define NV40TCL_BEGIN_END_TRIANGLE_FAN 0x00000007 -#define NV40TCL_BEGIN_END_QUADS 0x00000008 -#define NV40TCL_BEGIN_END_QUAD_STRIP 0x00000009 -#define NV40TCL_BEGIN_END_POLYGON 0x0000000a -#define NV40TCL_VB_ELEMENT_U16 0x0000180c -#define NV40TCL_VB_ELEMENT_U16_1_SHIFT 16 -#define NV40TCL_VB_ELEMENT_U16_1_MASK 0xffff0000 -#define NV40TCL_VB_ELEMENT_U16_0_SHIFT 0 -#define NV40TCL_VB_ELEMENT_U16_0_MASK 0x0000ffff -#define NV40TCL_VB_ELEMENT_U32 0x00001810 -#define NV40TCL_VB_VERTEX_BATCH 0x00001814 -#define NV40TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 -#define NV40TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 -#define NV40TCL_VB_VERTEX_BATCH_START_SHIFT 0 -#define NV40TCL_VB_VERTEX_BATCH_START_MASK 0x00ffffff -#define NV40TCL_VERTEX_DATA 0x00001818 -#define NV40TCL_IDXBUF_ADDRESS 0x0000181c -#define NV40TCL_IDXBUF_FORMAT 0x00001820 -#define NV40TCL_IDXBUF_FORMAT_TYPE_SHIFT 4 -#define NV40TCL_IDXBUF_FORMAT_TYPE_MASK 0x000000f0 -#define NV40TCL_IDXBUF_FORMAT_TYPE_U32 0x00000000 -#define NV40TCL_IDXBUF_FORMAT_TYPE_U16 0x00000010 -#define NV40TCL_IDXBUF_FORMAT_DMA1 (1 << 0) -#define NV40TCL_VB_INDEX_BATCH 0x00001824 -#define NV40TCL_VB_INDEX_BATCH_COUNT_SHIFT 24 -#define NV40TCL_VB_INDEX_BATCH_COUNT_MASK 0xff000000 -#define NV40TCL_VB_INDEX_BATCH_START_SHIFT 0 -#define NV40TCL_VB_INDEX_BATCH_START_MASK 0x00ffffff -#define NV40TCL_POLYGON_MODE_FRONT 0x00001828 -#define NV40TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV40TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV40TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV40TCL_POLYGON_MODE_BACK 0x0000182c -#define NV40TCL_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV40TCL_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV40TCL_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV40TCL_CULL_FACE 0x00001830 -#define NV40TCL_CULL_FACE_FRONT 0x00000404 -#define NV40TCL_CULL_FACE_BACK 0x00000405 -#define NV40TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV40TCL_FRONT_FACE 0x00001834 -#define NV40TCL_FRONT_FACE_CW 0x00000900 -#define NV40TCL_FRONT_FACE_CCW 0x00000901 -#define NV40TCL_POLYGON_SMOOTH_ENABLE 0x00001838 -#define NV40TCL_CULL_FACE_ENABLE 0x0000183c -#define NV40TCL_TEX_SIZE1(x) (0x00001840+((x)*4)) -#define NV40TCL_TEX_SIZE1__SIZE 0x00000008 -#define NV40TCL_TEX_SIZE1_DEPTH_SHIFT 20 -#define NV40TCL_TEX_SIZE1_DEPTH_MASK 0xfff00000 -#define NV40TCL_TEX_SIZE1_PITCH_SHIFT 0 -#define NV40TCL_TEX_SIZE1_PITCH_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) -#define NV40TCL_VTX_ATTR_2F_X__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) -#define NV40TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) -#define NV40TCL_VTX_ATTR_2I__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 -#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_2I_X_SHIFT 0 -#define NV40TCL_VTX_ATTR_2I_X_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_4I_0(x) (0x00001900+((x)*8)) -#define NV40TCL_VTX_ATTR_4I_0__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4I_0_Y_SHIFT 16 -#define NV40TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_4I_0_X_SHIFT 0 -#define NV40TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_4I_1(x) (0x00001904+((x)*8)) -#define NV40TCL_VTX_ATTR_4I_1__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4I_1_W_SHIFT 16 -#define NV40TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_4I_1_Z_SHIFT 0 -#define NV40TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff -#define NV40TCL_TEX_OFFSET(x) (0x00001a00+((x)*32)) -#define NV40TCL_TEX_OFFSET__SIZE 0x00000010 -#define NV40TCL_TEX_FORMAT(x) (0x00001a04+((x)*32)) -#define NV40TCL_TEX_FORMAT__SIZE 0x00000010 -#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT 16 -#define NV40TCL_TEX_FORMAT_MIPMAP_COUNT_MASK 0x000f0000 -#define NV40TCL_TEX_FORMAT_RECT (1 << 14) -#define NV40TCL_TEX_FORMAT_LINEAR (1 << 13) -#define NV40TCL_TEX_FORMAT_FORMAT_SHIFT 8 -#define NV40TCL_TEX_FORMAT_FORMAT_MASK 0x00001f00 -#define NV40TCL_TEX_FORMAT_FORMAT_L8 0x00000100 -#define NV40TCL_TEX_FORMAT_FORMAT_A1R5G5B5 0x00000200 -#define NV40TCL_TEX_FORMAT_FORMAT_A4R4G4B4 0x00000300 -#define NV40TCL_TEX_FORMAT_FORMAT_R5G6B5 0x00000400 -#define NV40TCL_TEX_FORMAT_FORMAT_A8R8G8B8 0x00000500 -#define NV40TCL_TEX_FORMAT_FORMAT_DXT1 0x00000600 -#define NV40TCL_TEX_FORMAT_FORMAT_DXT3 0x00000700 -#define NV40TCL_TEX_FORMAT_FORMAT_DXT5 0x00000800 -#define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00000b00 -#define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x00001000 -#define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x00001200 -#define NV40TCL_TEX_FORMAT_FORMAT_HILO8 0x00001800 -#define NV40TCL_TEX_FORMAT_FORMAT_RGBA16F 0x00001a00 -#define NV40TCL_TEX_FORMAT_FORMAT_RGBA32F 0x00001b00 -#define NV40TCL_TEX_FORMAT_DIMS_SHIFT 4 -#define NV40TCL_TEX_FORMAT_DIMS_MASK 0x000000f0 -#define NV40TCL_TEX_FORMAT_DIMS_1D 0x00000010 -#define NV40TCL_TEX_FORMAT_DIMS_2D 0x00000020 -#define NV40TCL_TEX_FORMAT_DIMS_3D 0x00000030 -#define NV40TCL_TEX_FORMAT_NO_BORDER (1 << 3) -#define NV40TCL_TEX_FORMAT_CUBIC (1 << 2) -#define NV40TCL_TEX_FORMAT_DMA1 (1 << 1) -#define NV40TCL_TEX_FORMAT_DMA0 (1 << 0) -#define NV40TCL_TEX_WRAP(x) (0x00001a08+((x)*32)) -#define NV40TCL_TEX_WRAP__SIZE 0x00000010 -#define NV40TCL_TEX_WRAP_S_SHIFT 0 -#define NV40TCL_TEX_WRAP_S_MASK 0x000000ff -#define NV40TCL_TEX_WRAP_S_REPEAT 0x00000001 -#define NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT 0x00000002 -#define NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE 0x00000003 -#define NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER 0x00000004 -#define NV40TCL_TEX_WRAP_S_CLAMP 0x00000005 -#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE 0x00000006 -#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER 0x00000007 -#define NV40TCL_TEX_WRAP_S_MIRROR_CLAMP 0x00000008 -#define NV40TCL_TEX_WRAP_T_SHIFT 8 -#define NV40TCL_TEX_WRAP_T_MASK 0x0000ff00 -#define NV40TCL_TEX_WRAP_T_REPEAT 0x00000100 -#define NV40TCL_TEX_WRAP_T_MIRRORED_REPEAT 0x00000200 -#define NV40TCL_TEX_WRAP_T_CLAMP_TO_EDGE 0x00000300 -#define NV40TCL_TEX_WRAP_T_CLAMP_TO_BORDER 0x00000400 -#define NV40TCL_TEX_WRAP_T_CLAMP 0x00000500 -#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_EDGE 0x00000600 -#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_BORDER 0x00000700 -#define NV40TCL_TEX_WRAP_T_MIRROR_CLAMP 0x00000800 -#define NV40TCL_TEX_WRAP_R_SHIFT 16 -#define NV40TCL_TEX_WRAP_R_MASK 0x00ff0000 -#define NV40TCL_TEX_WRAP_R_REPEAT 0x00010000 -#define NV40TCL_TEX_WRAP_R_MIRRORED_REPEAT 0x00020000 -#define NV40TCL_TEX_WRAP_R_CLAMP_TO_EDGE 0x00030000 -#define NV40TCL_TEX_WRAP_R_CLAMP_TO_BORDER 0x00040000 -#define NV40TCL_TEX_WRAP_R_CLAMP 0x00050000 -#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE 0x00060000 -#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER 0x00070000 -#define NV40TCL_TEX_WRAP_R_MIRROR_CLAMP 0x00080000 -#define NV40TCL_TEX_WRAP_RCOMP_SHIFT 28 -#define NV40TCL_TEX_WRAP_RCOMP_MASK 0xf0000000 -#define NV40TCL_TEX_WRAP_RCOMP_NEVER 0x00000000 -#define NV40TCL_TEX_WRAP_RCOMP_GREATER 0x10000000 -#define NV40TCL_TEX_WRAP_RCOMP_EQUAL 0x20000000 -#define NV40TCL_TEX_WRAP_RCOMP_GEQUAL 0x30000000 -#define NV40TCL_TEX_WRAP_RCOMP_LESS 0x40000000 -#define NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL 0x50000000 -#define NV40TCL_TEX_WRAP_RCOMP_LEQUAL 0x60000000 -#define NV40TCL_TEX_WRAP_RCOMP_ALWAYS 0x70000000 -#define NV40TCL_TEX_ENABLE(x) (0x00001a0c+((x)*32)) -#define NV40TCL_TEX_ENABLE__SIZE 0x00000010 -#define NV40TCL_TEX_ENABLE_ENABLE (1 << 31) -#define NV40TCL_TEX_ENABLE_ANISO_SHIFT 4 -#define NV40TCL_TEX_ENABLE_ANISO_MASK 0x000000f0 -#define NV40TCL_TEX_ENABLE_ANISO_NONE 0x00000000 -#define NV40TCL_TEX_ENABLE_ANISO_2X 0x00000010 -#define NV40TCL_TEX_ENABLE_ANISO_4X 0x00000020 -#define NV40TCL_TEX_ENABLE_ANISO_6X 0x00000030 -#define NV40TCL_TEX_ENABLE_ANISO_8X 0x00000040 -#define NV40TCL_TEX_ENABLE_ANISO_10X 0x00000050 -#define NV40TCL_TEX_ENABLE_ANISO_12X 0x00000060 -#define NV40TCL_TEX_ENABLE_ANISO_16X 0x00000070 -#define NV40TCL_TEX_SWIZZLE(x) (0x00001a10+((x)*32)) -#define NV40TCL_TEX_SWIZZLE__SIZE 0x00000010 -#define NV40TCL_TEX_SWIZZLE_S0_X_SHIFT 14 -#define NV40TCL_TEX_SWIZZLE_S0_X_MASK 0x0000c000 -#define NV40TCL_TEX_SWIZZLE_S0_X_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_X_ONE 0x00004000 -#define NV40TCL_TEX_SWIZZLE_S0_X_S1 0x00008000 -#define NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT 12 -#define NV40TCL_TEX_SWIZZLE_S0_Y_MASK 0x00003000 -#define NV40TCL_TEX_SWIZZLE_S0_Y_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_Y_ONE 0x00001000 -#define NV40TCL_TEX_SWIZZLE_S0_Y_S1 0x00002000 -#define NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT 10 -#define NV40TCL_TEX_SWIZZLE_S0_Z_MASK 0x00000c00 -#define NV40TCL_TEX_SWIZZLE_S0_Z_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_Z_ONE 0x00000400 -#define NV40TCL_TEX_SWIZZLE_S0_Z_S1 0x00000800 -#define NV40TCL_TEX_SWIZZLE_S0_W_SHIFT 8 -#define NV40TCL_TEX_SWIZZLE_S0_W_MASK 0x00000300 -#define NV40TCL_TEX_SWIZZLE_S0_W_ZERO 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S0_W_ONE 0x00000100 -#define NV40TCL_TEX_SWIZZLE_S0_W_S1 0x00000200 -#define NV40TCL_TEX_SWIZZLE_S1_X_SHIFT 6 -#define NV40TCL_TEX_SWIZZLE_S1_X_MASK 0x000000c0 -#define NV40TCL_TEX_SWIZZLE_S1_X_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_X_Z 0x00000040 -#define NV40TCL_TEX_SWIZZLE_S1_X_Y 0x00000080 -#define NV40TCL_TEX_SWIZZLE_S1_X_X 0x000000c0 -#define NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT 4 -#define NV40TCL_TEX_SWIZZLE_S1_Y_MASK 0x00000030 -#define NV40TCL_TEX_SWIZZLE_S1_Y_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_Y_Z 0x00000010 -#define NV40TCL_TEX_SWIZZLE_S1_Y_Y 0x00000020 -#define NV40TCL_TEX_SWIZZLE_S1_Y_X 0x00000030 -#define NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT 2 -#define NV40TCL_TEX_SWIZZLE_S1_Z_MASK 0x0000000c -#define NV40TCL_TEX_SWIZZLE_S1_Z_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_Z_Z 0x00000004 -#define NV40TCL_TEX_SWIZZLE_S1_Z_Y 0x00000008 -#define NV40TCL_TEX_SWIZZLE_S1_Z_X 0x0000000c -#define NV40TCL_TEX_SWIZZLE_S1_W_SHIFT 0 -#define NV40TCL_TEX_SWIZZLE_S1_W_MASK 0x00000003 -#define NV40TCL_TEX_SWIZZLE_S1_W_W 0x00000000 -#define NV40TCL_TEX_SWIZZLE_S1_W_Z 0x00000001 -#define NV40TCL_TEX_SWIZZLE_S1_W_Y 0x00000002 -#define NV40TCL_TEX_SWIZZLE_S1_W_X 0x00000003 -#define NV40TCL_TEX_FILTER(x) (0x00001a14+((x)*32)) -#define NV40TCL_TEX_FILTER__SIZE 0x00000010 -#define NV40TCL_TEX_FILTER_SIGNED_ALPHA (1 << 31) -#define NV40TCL_TEX_FILTER_SIGNED_RED (1 << 30) -#define NV40TCL_TEX_FILTER_SIGNED_GREEN (1 << 29) -#define NV40TCL_TEX_FILTER_SIGNED_BLUE (1 << 28) -#define NV40TCL_TEX_FILTER_MIN_SHIFT 16 -#define NV40TCL_TEX_FILTER_MIN_MASK 0x000f0000 -#define NV40TCL_TEX_FILTER_MIN_NEAREST 0x00010000 -#define NV40TCL_TEX_FILTER_MIN_LINEAR 0x00020000 -#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST 0x00030000 -#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST 0x00040000 -#define NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR 0x00050000 -#define NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR 0x00060000 -#define NV40TCL_TEX_FILTER_MAG_SHIFT 24 -#define NV40TCL_TEX_FILTER_MAG_MASK 0x0f000000 -#define NV40TCL_TEX_FILTER_MAG_NEAREST 0x01000000 -#define NV40TCL_TEX_FILTER_MAG_LINEAR 0x02000000 -#define NV40TCL_TEX_SIZE0(x) (0x00001a18+((x)*32)) -#define NV40TCL_TEX_SIZE0__SIZE 0x00000010 -#define NV40TCL_TEX_SIZE0_H_SHIFT 0 -#define NV40TCL_TEX_SIZE0_H_MASK 0x0000ffff -#define NV40TCL_TEX_SIZE0_W_SHIFT 16 -#define NV40TCL_TEX_SIZE0_W_MASK 0xffff0000 -#define NV40TCL_TEX_BORDER_COLOR(x) (0x00001a1c+((x)*32)) -#define NV40TCL_TEX_BORDER_COLOR__SIZE 0x00000010 -#define NV40TCL_TEX_BORDER_COLOR_B_SHIFT 0 -#define NV40TCL_TEX_BORDER_COLOR_B_MASK 0x000000ff -#define NV40TCL_TEX_BORDER_COLOR_G_SHIFT 8 -#define NV40TCL_TEX_BORDER_COLOR_G_MASK 0x0000ff00 -#define NV40TCL_TEX_BORDER_COLOR_R_SHIFT 16 -#define NV40TCL_TEX_BORDER_COLOR_R_MASK 0x00ff0000 -#define NV40TCL_TEX_BORDER_COLOR_A_SHIFT 24 -#define NV40TCL_TEX_BORDER_COLOR_A_MASK 0xff000000 -#define NV40TCL_VTX_ATTR_4F_X(x) (0x00001c00+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_X__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) -#define NV40TCL_VTX_ATTR_4F_W__SIZE 0x00000010 -#define NV40TCL_FP_CONTROL 0x00001d60 -#define NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT 24 -#define NV40TCL_FP_CONTROL_TEMP_COUNT_MASK 0xff000000 -#define NV40TCL_FP_CONTROL_KIL (1 << 7) -#define NV40TCL_MULTISAMPLE_CONTROL 0x00001d7c -#define NV40TCL_CLEAR_VALUE_DEPTH 0x00001d8c -#define NV40TCL_CLEAR_VALUE_COLOR 0x00001d90 -#define NV40TCL_CLEAR_BUFFERS 0x00001d94 -#define NV40TCL_CLEAR_BUFFERS_COLOR_A (1 << 7) -#define NV40TCL_CLEAR_BUFFERS_COLOR_B (1 << 6) -#define NV40TCL_CLEAR_BUFFERS_COLOR_G (1 << 5) -#define NV40TCL_CLEAR_BUFFERS_COLOR_R (1 << 4) -#define NV40TCL_CLEAR_BUFFERS_STENCIL (1 << 1) -#define NV40TCL_CLEAR_BUFFERS_DEPTH (1 << 0) -#define NV40TCL_LINE_STIPPLE_ENABLE 0x00001db4 -#define NV40TCL_LINE_STIPPLE_PATTERN 0x00001db8 -#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT 0 -#define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff -#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 -#define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 -#define NV40TCL_VP_UPLOAD_FROM_ID 0x00001e9c -#define NV40TCL_VP_START_FROM_ID 0x00001ea0 -#define NV40TCL_POINT_SIZE 0x00001ee0 -#define NV40TCL_POINT_SPRITE 0x00001ee8 -#define NV40TCL_VP_UPLOAD_CONST_ID 0x00001efc -#define NV40TCL_VP_UPLOAD_CONST_X(x) (0x00001f00+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_X__SIZE 0x00000004 -#define NV40TCL_VP_UPLOAD_CONST_Y(x) (0x00001f04+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_Y__SIZE 0x00000004 -#define NV40TCL_VP_UPLOAD_CONST_Z(x) (0x00001f08+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_Z__SIZE 0x00000004 -#define NV40TCL_VP_UPLOAD_CONST_W(x) (0x00001f0c+((x)*16)) -#define NV40TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 -#define NV40TCL_TEX_CACHE_CTL 0x00001fd8 -#define NV40TCL_VP_ATTRIB_EN 0x00001ff0 -#define NV40TCL_VP_RESULT_EN 0x00001ff4 - - -#define NV44TCL 0x00004497 - - - -#define NV50_2D 0x0000502d - -#define NV50_2D_NOP 0x00000100 -#define NV50_2D_NOTIFY 0x00000104 -#define NV50_2D_DMA_NOTIFY 0x00000180 -#define NV50_2D_DMA_IN_MEMORY0 0x00000184 -#define NV50_2D_DMA_IN_MEMORY1 0x00000188 -#define NV50_2D_DMA_IN_MEMORY2 0x0000018c -#define NV50_2D_DST_FORMAT 0x00000200 -#define NV50_2D_DST_FORMAT_32BPP 0x000000cf -#define NV50_2D_DST_FORMAT_24BPP 0x000000e6 -#define NV50_2D_DST_FORMAT_16BPP 0x000000e8 -#define NV50_2D_DST_FORMAT_8BPP 0x000000f3 -#define NV50_2D_DST_FORMAT_15BPP 0x000000f8 -#define NV50_2D_DST_PITCH 0x00000214 -#define NV50_2D_DST_WIDTH 0x00000218 -#define NV50_2D_DST_HEIGHT 0x0000021c -#define NV50_2D_DST_ADDRESS_HIGH 0x00000220 -#define NV50_2D_DST_ADDRESS_LOW 0x00000224 -#define NV50_2D_SRC_FORMAT 0x00000230 -#define NV50_2D_SRC_FORMAT_32BPP 0x000000cf -#define NV50_2D_SRC_FORMAT_24BPP 0x000000e6 -#define NV50_2D_SRC_FORMAT_16BPP 0x000000e8 -#define NV50_2D_SRC_FORMAT_8BPP 0x000000f3 -#define NV50_2D_SRC_FORMAT_15BPP 0x000000f8 -#define NV50_2D_SRC_PITCH 0x00000244 -#define NV50_2D_SRC_WIDTH 0x00000248 -#define NV50_2D_SRC_HEIGHT 0x0000024c -#define NV50_2D_SRC_ADDRESS_HIGH 0x00000250 -#define NV50_2D_SRC_ADDRESS_LOW 0x00000254 -#define NV50_2D_CLIP_X 0x00000280 -#define NV50_2D_CLIP_Y 0x00000284 -#define NV50_2D_CLIP_Z 0x00000288 -#define NV50_2D_CLIP_W 0x0000028c -#define NV50_2D_ROP 0x000002a0 -#define NV50_2D_OPERATION 0x000002ac -#define NV50_2D_OPERATION_SRCCOPY_AND 0x00000000 -#define NV50_2D_OPERATION_ROP_AND 0x00000001 -#define NV50_2D_OPERATION_BLEND_AND 0x00000002 -#define NV50_2D_OPERATION_SRCCOPY 0x00000003 -#define NV50_2D_OPERATION_SRCCOPY_PREMULT 0x00000004 -#define NV50_2D_OPERATION_BLEND_PREMULT 0x00000005 -#define NV50_2D_PATTERN_FORMAT 0x000002e8 -#define NV50_2D_PATTERN_FORMAT_16BPP 0x00000000 -#define NV50_2D_PATTERN_FORMAT_15BPP 0x00000001 -#define NV50_2D_PATTERN_FORMAT_32BPP 0x00000002 -#define NV50_2D_PATTERN_FORMAT_8BPP 0x00000003 -#define NV50_2D_PATTERN_COLOR(x) (0x000002f0+((x)*4)) -#define NV50_2D_PATTERN_COLOR__SIZE 0x00000002 -#define NV50_2D_PATTERN_BITMAP(x) (0x000002f8+((x)*4)) -#define NV50_2D_PATTERN_BITMAP__SIZE 0x00000002 -#define NV50_2D_RECT_FORMAT 0x00000584 -#define NV50_2D_RECT_FORMAT_32BPP 0x000000cf -#define NV50_2D_RECT_FORMAT_24BPP 0x000000e6 -#define NV50_2D_RECT_FORMAT_16BPP 0x000000e8 -#define NV50_2D_RECT_FORMAT_8BPP 0x000000f3 -#define NV50_2D_RECT_FORMAT_15BPP 0x000000f8 -#define NV50_2D_RECT_COLOR 0x00000588 -#define NV50_2D_RECT_X1 0x00000600 -#define NV50_2D_RECT_Y1 0x00000604 -#define NV50_2D_RECT_X2 0x00000608 -#define NV50_2D_RECT_Y2 0x0000060c -#define NV50_2D_BLIT_DST_X 0x000008b0 -#define NV50_2D_BLIT_DST_Y 0x000008b4 -#define NV50_2D_BLIT_DST_W 0x000008b8 -#define NV50_2D_BLIT_DST_H 0x000008bc -#define NV50_2D_BLIT_SRC_X 0x000008d4 -#define NV50_2D_BLIT_SRC_Y 0x000008dc -#define NV50_2D_SIFC_UNK0800 0x00000800 -#define NV50_2D_SIFC_FORMAT 0x00000804 -#define NV50_2D_SIFC_FORMAT_32BPP 0x000000cf -#define NV50_2D_SIFC_FORMAT_24BPP 0x000000e6 -#define NV50_2D_SIFC_FORMAT_16BPP 0x000000e8 -#define NV50_2D_SIFC_FORMAT_8BPP 0x000000f3 -#define NV50_2D_SIFC_FORMAT_15BPP 0x000000f8 -#define NV50_2D_SIFC_WIDTH 0x00000838 -#define NV50_2D_SIFC_HEIGHT 0x0000083c -#define NV50_2D_SIFC_SCALE_UNK0840 0x00000840 -#define NV50_2D_SIFC_SCALE_UNK0844 0x00000844 -#define NV50_2D_SIFC_SCALE_UNK0848 0x00000848 -#define NV50_2D_SIFC_SCALE_UNK084C 0x0000084c -#define NV50_2D_SIFC_UNK0850 0x00000850 -#define NV50_2D_SIFC_DST_X 0x00000854 -#define NV50_2D_SIFC_UNK0858 0x00000858 -#define NV50_2D_SIFC_DST_Y 0x0000085c -#define NV50_2D_SIFC_DATA 0x00000860 - - -#define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 - -#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH 0x00000238 -#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT_HIGH 0x0000023c - - -#define NV50TCL 0x00005097 - -#define NV50TCL_NOP 0x00000100 -#define NV50TCL_NOTIFY 0x00000104 -#define NV50TCL_DMA_NOTIFY 0x00000180 -#define NV50TCL_DMA_IN_MEMORY0(x) (0x00000184+((x)*4)) -#define NV50TCL_DMA_IN_MEMORY0__SIZE 0x0000000b -#define NV50TCL_DMA_IN_MEMORY1(x) (0x000001c0+((x)*4)) -#define NV50TCL_DMA_IN_MEMORY1__SIZE 0x00000008 -#define NV50TCL_RT_ADDRESS_HIGH(x) (0x00000200+((x)*32)) -#define NV50TCL_RT_ADDRESS_HIGH__SIZE 0x00000008 -#define NV50TCL_RT_ADDRESS_LOW(x) (0x00000204+((x)*32)) -#define NV50TCL_RT_ADDRESS_LOW__SIZE 0x00000008 -#define NV50TCL_RT_FORMAT(x) (0x00000208+((x)*32)) -#define NV50TCL_RT_FORMAT__SIZE 0x00000008 -#define NV50TCL_RT_UNK3(x) (0x0000020c+((x)*32)) -#define NV50TCL_RT_UNK3__SIZE 0x00000008 -#define NV50TCL_RT_UNK4(x) (0x00000210+((x)*32)) -#define NV50TCL_RT_UNK4__SIZE 0x00000008 -#define NV50TCL_VTX_ATTR_1F(x) (0x00000300+((x)*4)) -#define NV50TCL_VTX_ATTR_1F__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_2F_X(x) (0x00000380+((x)*8)) -#define NV50TCL_VTX_ATTR_2F_X__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_2F_Y(x) (0x00000384+((x)*8)) -#define NV50TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_3F_X(x) (0x00000400+((x)*16)) -#define NV50TCL_VTX_ATTR_3F_X__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_3F_Y(x) (0x00000404+((x)*16)) -#define NV50TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_3F_Z(x) (0x00000408+((x)*16)) -#define NV50TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_3F_W(x) (0x0000040c+((x)*16)) -#define NV50TCL_VTX_ATTR_3F_W__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4F_X(x) (0x00000500+((x)*16)) -#define NV50TCL_VTX_ATTR_4F_X__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4F_Y(x) (0x00000504+((x)*16)) -#define NV50TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4F_Z(x) (0x00000508+((x)*16)) -#define NV50TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4F_W(x) (0x0000050c+((x)*16)) -#define NV50TCL_VTX_ATTR_4F_W__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_2I(x) (0x00000680+((x)*4)) -#define NV50TCL_VTX_ATTR_2I__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_2I_X_SHIFT 0 -#define NV50TCL_VTX_ATTR_2I_X_MASK 0x0000ffff -#define NV50TCL_VTX_ATTR_2I_Y_SHIFT 16 -#define NV50TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 -#define NV50TCL_VTX_ATTR_4I_0(x) (0x00000700+((x)*8)) -#define NV50TCL_VTX_ATTR_4I_0__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4I_0_X_SHIFT 0 -#define NV50TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff -#define NV50TCL_VTX_ATTR_4I_0_Y_SHIFT 16 -#define NV50TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 -#define NV50TCL_VTX_ATTR_4I_1(x) (0x00000704+((x)*8)) -#define NV50TCL_VTX_ATTR_4I_1__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4I_1_Z_SHIFT 0 -#define NV50TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff -#define NV50TCL_VTX_ATTR_4I_1_W_SHIFT 16 -#define NV50TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 -#define NV50TCL_VTX_ATTR_4NI_0(x) (0x00000780+((x)*8)) -#define NV50TCL_VTX_ATTR_4NI_0__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4NI_0_X_SHIFT 0 -#define NV50TCL_VTX_ATTR_4NI_0_X_MASK 0x0000ffff -#define NV50TCL_VTX_ATTR_4NI_0_Y_SHIFT 16 -#define NV50TCL_VTX_ATTR_4NI_0_Y_MASK 0xffff0000 -#define NV50TCL_VTX_ATTR_4NI_1(x) (0x00000784+((x)*8)) -#define NV50TCL_VTX_ATTR_4NI_1__SIZE 0x00000010 -#define NV50TCL_VTX_ATTR_4NI_1_Z_SHIFT 0 -#define NV50TCL_VTX_ATTR_4NI_1_Z_MASK 0x0000ffff -#define NV50TCL_VTX_ATTR_4NI_1_W_SHIFT 16 -#define NV50TCL_VTX_ATTR_4NI_1_W_MASK 0xffff0000 -#define NV50TCL_VERTEX_ARRAY_FORMAT(x) (0x00000900+((x)*16)) -#define NV50TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV50TCL_VIEWPORT_UNK0(x) (0x00000a00+((x)*4)) -#define NV50TCL_VIEWPORT_UNK0__SIZE 0x00000003 -#define NV50TCL_VIEWPORT_UNK1(x) (0x00000a0c+((x)*4)) -#define NV50TCL_VIEWPORT_UNK1__SIZE 0x00000003 -#define NV50TCL_VIEWPORT_HORIZ 0x00000c00 -#define NV50TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV50TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV50TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV50TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV50TCL_VIEWPORT_VERT 0x00000c04 -#define NV50TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV50TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV50TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV50TCL_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV50TCL_DEPTH_RANGE_NEAR 0x00000c08 -#define NV50TCL_DEPTH_RANGE_FAR 0x00000c0c -#define NV50TCL_VIEWPORT_CLIP_HORIZ(x) (0x00000d00+((x)*8)) -#define NV50TCL_VIEWPORT_CLIP_HORIZ__SIZE 0x00000008 -#define NV50TCL_VIEWPORT_CLIP_VERT(x) (0x00000d04+((x)*8)) -#define NV50TCL_VIEWPORT_CLIP_VERT__SIZE 0x00000008 -#define NV50TCL_VERTEX_BUFFER_FIRST 0x00000d74 -#define NV50TCL_VERTEX_BUFFER_COUNT 0x00000d78 -#define NV50TCL_CLEAR_COLOR(x) (0x00000d80+((x)*4)) -#define NV50TCL_CLEAR_COLOR__SIZE 0x00000004 -#define NV50TCL_CLEAR_DEPTH 0x00000d90 -#define NV50TCL_CLEAR_STENCIL 0x00000da0 -#define NV50TCL_POLYGON_MODE_FRONT 0x00000dac -#define NV50TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 -#define NV50TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 -#define NV50TCL_POLYGON_MODE_FRONT_FILL 0x00001b02 -#define NV50TCL_POLYGON_MODE_BACK 0x00000db0 -#define NV50TCL_POLYGON_MODE_BACK_POINT 0x00001b00 -#define NV50TCL_POLYGON_MODE_BACK_LINE 0x00001b01 -#define NV50TCL_POLYGON_MODE_BACK_FILL 0x00001b02 -#define NV50TCL_POLYGON_SMOOTH_ENABLE 0x00000db4 -#define NV50TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000dc0 -#define NV50TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000dc4 -#define NV50TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000dc8 -#define NV50TCL_SCISSOR_HORIZ 0x00000e04 -#define NV50TCL_SCISSOR_HORIZ_L_SHIFT 0 -#define NV50TCL_SCISSOR_HORIZ_L_MASK 0x0000ffff -#define NV50TCL_SCISSOR_HORIZ_R_SHIFT 16 -#define NV50TCL_SCISSOR_HORIZ_R_MASK 0xffff0000 -#define NV50TCL_SCISSOR_VERT 0x00000e08 -#define NV50TCL_SCISSOR_VERT_T_SHIFT 0 -#define NV50TCL_SCISSOR_VERT_T_MASK 0x0000ffff -#define NV50TCL_SCISSOR_VERT_B_SHIFT 16 -#define NV50TCL_SCISSOR_VERT_B_MASK 0xffff0000 -#define NV50TCL_VP_UPLOAD_CONST_ID 0x00000f00 -#define NV50TCL_VP_UPLOAD_CONST(x) (0x00000f04+((x)*4)) -#define NV50TCL_VP_UPLOAD_CONST__SIZE 0x00000010 -#define NV50TCL_STENCIL_FRONT_FUNC_REF 0x00000f54 -#define NV50TCL_STENCIL_FRONT_MASK 0x00000f58 -#define NV50TCL_STENCIL_FRONT_FUNC_MASK 0x00000f5c -#define NV50TCL_GP_ADDRESS_HIGH 0x00000f70 -#define NV50TCL_GP_ADDRESS_LOW 0x00000f74 -#define NV50TCL_VP_ADDRESS_HIGH 0x00000f7c -#define NV50TCL_VP_ADDRESS_LOW 0x00000f80 -#define NV50TCL_FP_ADDRESS_HIGH 0x00000fa4 -#define NV50TCL_FP_ADDRESS_LOW 0x00000fa8 -#define NV50TCL_ZETA_ADDRESS_HIGH 0x00000fe0 -#define NV50TCL_ZETA_ADDRESS_LOW 0x00000fe4 -#define NV50TCL_UNKFF4 0x00000ff4 -#define NV50TCL_UNKFF4_W_SHIFT 16 -#define NV50TCL_UNKFF4_W_MASK 0xffff0000 -#define NV50TCL_UNKFF8 0x00000ff8 -#define NV50TCL_UNKFF8_H_SHIFT 16 -#define NV50TCL_UNKFF8_H_MASK 0xffff0000 -#define NV50TCL_RT_HORIZ(x) (0x00001240+((x)*8)) -#define NV50TCL_RT_HORIZ__SIZE 0x00000008 -#define NV50TCL_RT_VERT(x) (0x00001244+((x)*8)) -#define NV50TCL_RT_VERT__SIZE 0x00000008 -#define NV50TCL_DEPTH_TEST_ENABLE 0x000012cc -#define NV50TCL_SHADE_MODEL 0x000012d4 -#define NV50TCL_SHADE_MODEL_FLAT 0x00001d00 -#define NV50TCL_SHADE_MODEL_SMOOTH 0x00001d01 -#define NV50TCL_DEPTH_WRITE_ENABLE 0x000012e8 -#define NV50TCL_ALPHA_TEST_ENABLE 0x000012ec -#define NV50TCL_DEPTH_TEST_FUNC 0x0000130c -#define NV50TCL_DEPTH_TEST_FUNC_NEVER 0x00000200 -#define NV50TCL_DEPTH_TEST_FUNC_LESS 0x00000201 -#define NV50TCL_DEPTH_TEST_FUNC_EQUAL 0x00000202 -#define NV50TCL_DEPTH_TEST_FUNC_LEQUAL 0x00000203 -#define NV50TCL_DEPTH_TEST_FUNC_GREATER 0x00000204 -#define NV50TCL_DEPTH_TEST_FUNC_GREATER 0x00000204 -#define NV50TCL_DEPTH_TEST_FUNC_NOTEQUAL 0x00000205 -#define NV50TCL_DEPTH_TEST_FUNC_GEQUAL 0x00000206 -#define NV50TCL_DEPTH_TEST_FUNC_ALWAYS 0x00000207 -#define NV50TCL_ALPHA_TEST_REF 0x00001310 -#define NV50TCL_ALPHA_TEST_FUNC 0x00001314 -#define NV50TCL_ALPHA_TEST_FUNC_NEVER 0x00000200 -#define NV50TCL_ALPHA_TEST_FUNC_LESS 0x00000201 -#define NV50TCL_ALPHA_TEST_FUNC_EQUAL 0x00000202 -#define NV50TCL_ALPHA_TEST_FUNC_LEQUAL 0x00000203 -#define NV50TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 -#define NV50TCL_ALPHA_TEST_FUNC_GREATER 0x00000204 -#define NV50TCL_ALPHA_TEST_FUNC_NOTEQUAL 0x00000205 -#define NV50TCL_ALPHA_TEST_FUNC_GEQUAL 0x00000206 -#define NV50TCL_ALPHA_TEST_FUNC_ALWAYS 0x00000207 -#define NV50TCL_BLEND_COLOR(x) (0x0000131c+((x)*4)) -#define NV50TCL_BLEND_COLOR__SIZE 0x00000004 -#define NV50TCL_BLEND_EQUATION_RGB 0x00001340 -#define NV50TCL_BLEND_EQUATION_RGB_FUNC_ADD 0x00008006 -#define NV50TCL_BLEND_EQUATION_RGB_MIN 0x00008007 -#define NV50TCL_BLEND_EQUATION_RGB_MAX 0x00008008 -#define NV50TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT 0x0000800a -#define NV50TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV50TCL_BLEND_FUNC_SRC_RGB 0x00001344 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ZERO 0x00000000 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE 0x00000001 -#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR 0x00000300 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA 0x00000302 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA 0x00000304 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50TCL_BLEND_FUNC_SRC_RGB_DST_COLOR 0x00000306 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR 0x00008001 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA 0x00008003 -#define NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50TCL_BLEND_FUNC_DST_RGB 0x00001348 -#define NV50TCL_BLEND_FUNC_DST_RGB_ZERO 0x00000000 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE 0x00000001 -#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_COLOR 0x00000300 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA 0x00000302 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50TCL_BLEND_FUNC_DST_RGB_DST_ALPHA 0x00000304 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50TCL_BLEND_FUNC_DST_RGB_DST_COLOR 0x00000306 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE 0x00000308 -#define NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR 0x00008001 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA 0x00008003 -#define NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50TCL_BLEND_EQUATION_ALPHA 0x0000134c -#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_ADD 0x00008006 -#define NV50TCL_BLEND_EQUATION_ALPHA_MIN 0x00008007 -#define NV50TCL_BLEND_EQUATION_ALPHA_MAX 0x00008008 -#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT 0x0000800a -#define NV50TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT 0x0000800b -#define NV50TCL_BLEND_FUNC_SRC_ALPHA 0x00001350 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ZERO 0x00000000 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE 0x00000001 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR 0x00000300 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA 0x00000302 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA 0x00000304 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR 0x00000306 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50TCL_BLEND_FUNC_DST_ALPHA 0x00001358 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ZERO 0x00000000 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE 0x00000001 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR 0x00000300 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR 0x00000301 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA 0x00000302 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA 0x00000303 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA 0x00000304 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA 0x00000305 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR 0x00000306 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR 0x00000307 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE 0x00000308 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR 0x00008001 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x00008002 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x00008003 -#define NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x00008004 -#define NV50TCL_BLEND_ENABLE(x) (0x00001360+((x)*4)) -#define NV50TCL_BLEND_ENABLE__SIZE 0x00000008 -#define NV50TCL_STENCIL_BACK_ENABLE 0x00001380 -#define NV50TCL_STENCIL_BACK_OP_FAIL 0x00001384 -#define NV50TCL_STENCIL_BACK_OP_FAIL_ZERO 0x00000000 -#define NV50TCL_STENCIL_BACK_OP_FAIL_INVERT 0x0000150a -#define NV50TCL_STENCIL_BACK_OP_FAIL_KEEP 0x00001e00 -#define NV50TCL_STENCIL_BACK_OP_FAIL_REPLACE 0x00001e01 -#define NV50TCL_STENCIL_BACK_OP_FAIL_INCR 0x00001e02 -#define NV50TCL_STENCIL_BACK_OP_FAIL_DECR 0x00001e03 -#define NV50TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP 0x00008507 -#define NV50TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP 0x00008508 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL 0x00001388 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_ZERO 0x00000000 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INVERT 0x0000150a -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_KEEP 0x00001e00 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_REPLACE 0x00001e01 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR 0x00001e02 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR 0x00001e03 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV50TCL_STENCIL_BACK_OP_ZPASS 0x0000138c -#define NV50TCL_STENCIL_BACK_OP_ZPASS_ZERO 0x00000000 -#define NV50TCL_STENCIL_BACK_OP_ZPASS_INVERT 0x0000150a -#define NV50TCL_STENCIL_BACK_OP_ZPASS_KEEP 0x00001e00 -#define NV50TCL_STENCIL_BACK_OP_ZPASS_REPLACE 0x00001e01 -#define NV50TCL_STENCIL_BACK_OP_ZPASS_INCR 0x00001e02 -#define NV50TCL_STENCIL_BACK_OP_ZPASS_DECR 0x00001e03 -#define NV50TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV50TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC 0x00001390 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_NEVER 0x00000200 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_LESS 0x00000201 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_EQUAL 0x00000202 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL 0x00000203 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER 0x00000204 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL 0x00000206 -#define NV50TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS 0x00000207 -#define NV50TCL_STENCIL_BACK_FUNC_REF 0x00001394 -#define NV50TCL_STENCIL_BACK_MASK 0x00001398 -#define NV50TCL_STENCIL_BACK_FUNC_MASK 0x0000139c -#define NV50TCL_LINE_WIDTH 0x000013b0 -#define NV50TCL_VP_START_ID 0x0000140c -#define NV50TCL_GP_START_ID 0x00001410 -#define NV50TCL_FP_START_ID 0x00001414 -#define NV50TCL_POINT_SIZE 0x00001518 -#define NV50TCL_TEX_CB0_ADDRESS_HIGH 0x0000155c -#define NV50TCL_TEX_CB0_ADDRESS_LOW 0x00001560 -#define NV50TCL_POLYGON_OFFSET_FACTOR 0x0000156c -#define NV50TCL_LINE_SMOOTH_ENABLE 0x00001570 -#define NV50TCL_TEX_CB1_ADDRESS_HIGH 0x00001574 -#define NV50TCL_TEX_CB1_ADDRESS_LOW 0x00001578 -#define NV50TCL_STENCIL_FRONT_ENABLE 0x00001594 -#define NV50TCL_STENCIL_FRONT_OP_FAIL 0x00001598 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_INVERT 0x0000150a -#define NV50TCL_STENCIL_FRONT_OP_FAIL_KEEP 0x00001e00 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_REPLACE 0x00001e01 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_INCR 0x00001e02 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_DECR 0x00001e03 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP 0x00008507 -#define NV50TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP 0x00008508 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL 0x0000159c -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_ZERO 0x00000000 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INVERT 0x0000150a -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_KEEP 0x00001e00 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE 0x00001e01 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR 0x00001e02 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR 0x00001e03 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP 0x00008507 -#define NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP 0x00008508 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS 0x000015a0 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_ZERO 0x00000000 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INVERT 0x0000150a -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_KEEP 0x00001e00 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_REPLACE 0x00001e01 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR 0x00001e02 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR 0x00001e03 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP 0x00008507 -#define NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP 0x00008508 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC 0x000015a4 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_NEVER 0x00000200 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_LESS 0x00000201 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL 0x00000202 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL 0x00000203 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER 0x00000204 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL 0x00000205 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL 0x00000206 -#define NV50TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS 0x00000207 -#define NV50TCL_POLYGON_OFFSET_UNITS 0x000015bc -#define NV50TCL_VERTEX_BEGIN 0x000015dc -#define NV50TCL_VERTEX_BEGIN_POINTS 0x00000000 -#define NV50TCL_VERTEX_BEGIN_LINES 0x00000001 -#define NV50TCL_VERTEX_BEGIN_LINE_LOOP 0x00000002 -#define NV50TCL_VERTEX_BEGIN_LINE_STRIP 0x00000003 -#define NV50TCL_VERTEX_BEGIN_TRIANGLES 0x00000004 -#define NV50TCL_VERTEX_BEGIN_TRIANGLE_STRIP 0x00000005 -#define NV50TCL_VERTEX_BEGIN_TRIANGLE_FAN 0x00000006 -#define NV50TCL_VERTEX_BEGIN_QUADS 0x00000007 -#define NV50TCL_VERTEX_BEGIN_QUAD_STRIP 0x00000008 -#define NV50TCL_VERTEX_BEGIN_POLYGON 0x00000009 -#define NV50TCL_VERTEX_END 0x000015e0 -#define NV50TCL_VERTEX_DATA 0x00001640 -#define NV50TCL_VP_ATTR_EN_0 0x00001650 -#define NV50TCL_VP_ATTR_EN_0_7_SHIFT 28 -#define NV50TCL_VP_ATTR_EN_0_7_MASK 0xf0000000 -#define NV50TCL_VP_ATTR_EN_0_7_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_7_XNNN 0x10000000 -#define NV50TCL_VP_ATTR_EN_0_7_NYNN 0x20000000 -#define NV50TCL_VP_ATTR_EN_0_7_XYNN 0x30000000 -#define NV50TCL_VP_ATTR_EN_0_7_NNZN 0x40000000 -#define NV50TCL_VP_ATTR_EN_0_7_XNZN 0x50000000 -#define NV50TCL_VP_ATTR_EN_0_7_NYZN 0x60000000 -#define NV50TCL_VP_ATTR_EN_0_7_XYZN 0x70000000 -#define NV50TCL_VP_ATTR_EN_0_7_NNNW 0x80000000 -#define NV50TCL_VP_ATTR_EN_0_7_XNNW 0x90000000 -#define NV50TCL_VP_ATTR_EN_0_7_NYNW 0xa0000000 -#define NV50TCL_VP_ATTR_EN_0_7_XYNW 0xb0000000 -#define NV50TCL_VP_ATTR_EN_0_7_NNZW 0xc0000000 -#define NV50TCL_VP_ATTR_EN_0_7_XNZW 0xd0000000 -#define NV50TCL_VP_ATTR_EN_0_7_NYZW 0xe0000000 -#define NV50TCL_VP_ATTR_EN_0_7_XYZW 0xf0000000 -#define NV50TCL_VP_ATTR_EN_0_6_SHIFT 24 -#define NV50TCL_VP_ATTR_EN_0_6_MASK 0x0f000000 -#define NV50TCL_VP_ATTR_EN_0_6_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_6_XNNN 0x01000000 -#define NV50TCL_VP_ATTR_EN_0_6_NYNN 0x02000000 -#define NV50TCL_VP_ATTR_EN_0_6_XYNN 0x03000000 -#define NV50TCL_VP_ATTR_EN_0_6_NNZN 0x04000000 -#define NV50TCL_VP_ATTR_EN_0_6_XNZN 0x05000000 -#define NV50TCL_VP_ATTR_EN_0_6_NYZN 0x06000000 -#define NV50TCL_VP_ATTR_EN_0_6_XYZN 0x07000000 -#define NV50TCL_VP_ATTR_EN_0_6_NNNW 0x08000000 -#define NV50TCL_VP_ATTR_EN_0_6_XNNW 0x09000000 -#define NV50TCL_VP_ATTR_EN_0_6_NYNW 0x0a000000 -#define NV50TCL_VP_ATTR_EN_0_6_XYNW 0x0b000000 -#define NV50TCL_VP_ATTR_EN_0_6_NNZW 0x0c000000 -#define NV50TCL_VP_ATTR_EN_0_6_XNZW 0x0d000000 -#define NV50TCL_VP_ATTR_EN_0_6_NYZW 0x0e000000 -#define NV50TCL_VP_ATTR_EN_0_6_XYZW 0x0f000000 -#define NV50TCL_VP_ATTR_EN_0_5_SHIFT 20 -#define NV50TCL_VP_ATTR_EN_0_5_MASK 0x00f00000 -#define NV50TCL_VP_ATTR_EN_0_5_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_5_XNNN 0x00100000 -#define NV50TCL_VP_ATTR_EN_0_5_NYNN 0x00200000 -#define NV50TCL_VP_ATTR_EN_0_5_XYNN 0x00300000 -#define NV50TCL_VP_ATTR_EN_0_5_NNZN 0x00400000 -#define NV50TCL_VP_ATTR_EN_0_5_XNZN 0x00500000 -#define NV50TCL_VP_ATTR_EN_0_5_NYZN 0x00600000 -#define NV50TCL_VP_ATTR_EN_0_5_XYZN 0x00700000 -#define NV50TCL_VP_ATTR_EN_0_5_NNNW 0x00800000 -#define NV50TCL_VP_ATTR_EN_0_5_XNNW 0x00900000 -#define NV50TCL_VP_ATTR_EN_0_5_NYNW 0x00a00000 -#define NV50TCL_VP_ATTR_EN_0_5_XYNW 0x00b00000 -#define NV50TCL_VP_ATTR_EN_0_5_NNZW 0x00c00000 -#define NV50TCL_VP_ATTR_EN_0_5_XNZW 0x00d00000 -#define NV50TCL_VP_ATTR_EN_0_5_NYZW 0x00e00000 -#define NV50TCL_VP_ATTR_EN_0_5_XYZW 0x00f00000 -#define NV50TCL_VP_ATTR_EN_0_4_SHIFT 16 -#define NV50TCL_VP_ATTR_EN_0_4_MASK 0x000f0000 -#define NV50TCL_VP_ATTR_EN_0_4_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_4_XNNN 0x00010000 -#define NV50TCL_VP_ATTR_EN_0_4_NYNN 0x00020000 -#define NV50TCL_VP_ATTR_EN_0_4_XYNN 0x00030000 -#define NV50TCL_VP_ATTR_EN_0_4_NNZN 0x00040000 -#define NV50TCL_VP_ATTR_EN_0_4_XNZN 0x00050000 -#define NV50TCL_VP_ATTR_EN_0_4_NYZN 0x00060000 -#define NV50TCL_VP_ATTR_EN_0_4_XYZN 0x00070000 -#define NV50TCL_VP_ATTR_EN_0_4_NNNW 0x00080000 -#define NV50TCL_VP_ATTR_EN_0_4_XNNW 0x00090000 -#define NV50TCL_VP_ATTR_EN_0_4_NYNW 0x000a0000 -#define NV50TCL_VP_ATTR_EN_0_4_XYNW 0x000b0000 -#define NV50TCL_VP_ATTR_EN_0_4_NNZW 0x000c0000 -#define NV50TCL_VP_ATTR_EN_0_4_XNZW 0x000d0000 -#define NV50TCL_VP_ATTR_EN_0_4_NYZW 0x000e0000 -#define NV50TCL_VP_ATTR_EN_0_4_XYZW 0x000f0000 -#define NV50TCL_VP_ATTR_EN_0_3_SHIFT 12 -#define NV50TCL_VP_ATTR_EN_0_3_MASK 0x0000f000 -#define NV50TCL_VP_ATTR_EN_0_3_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_3_XNNN 0x00001000 -#define NV50TCL_VP_ATTR_EN_0_3_NYNN 0x00002000 -#define NV50TCL_VP_ATTR_EN_0_3_XYNN 0x00003000 -#define NV50TCL_VP_ATTR_EN_0_3_NNZN 0x00004000 -#define NV50TCL_VP_ATTR_EN_0_3_XNZN 0x00005000 -#define NV50TCL_VP_ATTR_EN_0_3_NYZN 0x00006000 -#define NV50TCL_VP_ATTR_EN_0_3_XYZN 0x00007000 -#define NV50TCL_VP_ATTR_EN_0_3_NNNW 0x00008000 -#define NV50TCL_VP_ATTR_EN_0_3_XNNW 0x00009000 -#define NV50TCL_VP_ATTR_EN_0_3_NYNW 0x0000a000 -#define NV50TCL_VP_ATTR_EN_0_3_XYNW 0x0000b000 -#define NV50TCL_VP_ATTR_EN_0_3_NNZW 0x0000c000 -#define NV50TCL_VP_ATTR_EN_0_3_XNZW 0x0000d000 -#define NV50TCL_VP_ATTR_EN_0_3_NYZW 0x0000e000 -#define NV50TCL_VP_ATTR_EN_0_3_XYZW 0x0000f000 -#define NV50TCL_VP_ATTR_EN_0_2_SHIFT 8 -#define NV50TCL_VP_ATTR_EN_0_2_MASK 0x00000f00 -#define NV50TCL_VP_ATTR_EN_0_2_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_2_XNNN 0x00000100 -#define NV50TCL_VP_ATTR_EN_0_2_NYNN 0x00000200 -#define NV50TCL_VP_ATTR_EN_0_2_XYNN 0x00000300 -#define NV50TCL_VP_ATTR_EN_0_2_NNZN 0x00000400 -#define NV50TCL_VP_ATTR_EN_0_2_XNZN 0x00000500 -#define NV50TCL_VP_ATTR_EN_0_2_NYZN 0x00000600 -#define NV50TCL_VP_ATTR_EN_0_2_XYZN 0x00000700 -#define NV50TCL_VP_ATTR_EN_0_2_NNNW 0x00000800 -#define NV50TCL_VP_ATTR_EN_0_2_XNNW 0x00000900 -#define NV50TCL_VP_ATTR_EN_0_2_NYNW 0x00000a00 -#define NV50TCL_VP_ATTR_EN_0_2_XYNW 0x00000b00 -#define NV50TCL_VP_ATTR_EN_0_2_NNZW 0x00000c00 -#define NV50TCL_VP_ATTR_EN_0_2_XNZW 0x00000d00 -#define NV50TCL_VP_ATTR_EN_0_2_NYZW 0x00000e00 -#define NV50TCL_VP_ATTR_EN_0_2_XYZW 0x00000f00 -#define NV50TCL_VP_ATTR_EN_0_1_SHIFT 4 -#define NV50TCL_VP_ATTR_EN_0_1_MASK 0x000000f0 -#define NV50TCL_VP_ATTR_EN_0_1_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_1_XNNN 0x00000010 -#define NV50TCL_VP_ATTR_EN_0_1_NYNN 0x00000020 -#define NV50TCL_VP_ATTR_EN_0_1_XYNN 0x00000030 -#define NV50TCL_VP_ATTR_EN_0_1_NNZN 0x00000040 -#define NV50TCL_VP_ATTR_EN_0_1_XNZN 0x00000050 -#define NV50TCL_VP_ATTR_EN_0_1_NYZN 0x00000060 -#define NV50TCL_VP_ATTR_EN_0_1_XYZN 0x00000070 -#define NV50TCL_VP_ATTR_EN_0_1_NNNW 0x00000080 -#define NV50TCL_VP_ATTR_EN_0_1_XNNW 0x00000090 -#define NV50TCL_VP_ATTR_EN_0_1_NYNW 0x000000a0 -#define NV50TCL_VP_ATTR_EN_0_1_XYNW 0x000000b0 -#define NV50TCL_VP_ATTR_EN_0_1_NNZW 0x000000c0 -#define NV50TCL_VP_ATTR_EN_0_1_XNZW 0x000000d0 -#define NV50TCL_VP_ATTR_EN_0_1_NYZW 0x000000e0 -#define NV50TCL_VP_ATTR_EN_0_1_XYZW 0x000000f0 -#define NV50TCL_VP_ATTR_EN_0_0_SHIFT 0 -#define NV50TCL_VP_ATTR_EN_0_0_MASK 0x0000000f -#define NV50TCL_VP_ATTR_EN_0_0_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_0_0_XNNN 0x00000001 -#define NV50TCL_VP_ATTR_EN_0_0_NYNN 0x00000002 -#define NV50TCL_VP_ATTR_EN_0_0_XYNN 0x00000003 -#define NV50TCL_VP_ATTR_EN_0_0_NNZN 0x00000004 -#define NV50TCL_VP_ATTR_EN_0_0_XNZN 0x00000005 -#define NV50TCL_VP_ATTR_EN_0_0_NYZN 0x00000006 -#define NV50TCL_VP_ATTR_EN_0_0_XYZN 0x00000007 -#define NV50TCL_VP_ATTR_EN_0_0_NNNW 0x00000008 -#define NV50TCL_VP_ATTR_EN_0_0_XNNW 0x00000009 -#define NV50TCL_VP_ATTR_EN_0_0_NYNW 0x0000000a -#define NV50TCL_VP_ATTR_EN_0_0_XYNW 0x0000000b -#define NV50TCL_VP_ATTR_EN_0_0_NNZW 0x0000000c -#define NV50TCL_VP_ATTR_EN_0_0_XNZW 0x0000000d -#define NV50TCL_VP_ATTR_EN_0_0_NYZW 0x0000000e -#define NV50TCL_VP_ATTR_EN_0_0_XYZW 0x0000000f -#define NV50TCL_VP_ATTR_EN_1 0x00001654 -#define NV50TCL_VP_ATTR_EN_1_15_SHIFT 28 -#define NV50TCL_VP_ATTR_EN_1_15_MASK 0xf0000000 -#define NV50TCL_VP_ATTR_EN_1_15_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_15_XNNN 0x10000000 -#define NV50TCL_VP_ATTR_EN_1_15_NYNN 0x20000000 -#define NV50TCL_VP_ATTR_EN_1_15_XYNN 0x30000000 -#define NV50TCL_VP_ATTR_EN_1_15_NNZN 0x40000000 -#define NV50TCL_VP_ATTR_EN_1_15_XNZN 0x50000000 -#define NV50TCL_VP_ATTR_EN_1_15_NYZN 0x60000000 -#define NV50TCL_VP_ATTR_EN_1_15_XYZN 0x70000000 -#define NV50TCL_VP_ATTR_EN_1_15_NNNW 0x80000000 -#define NV50TCL_VP_ATTR_EN_1_15_XNNW 0x90000000 -#define NV50TCL_VP_ATTR_EN_1_15_NYNW 0xa0000000 -#define NV50TCL_VP_ATTR_EN_1_15_XYNW 0xb0000000 -#define NV50TCL_VP_ATTR_EN_1_15_NNZW 0xc0000000 -#define NV50TCL_VP_ATTR_EN_1_15_XNZW 0xd0000000 -#define NV50TCL_VP_ATTR_EN_1_15_NYZW 0xe0000000 -#define NV50TCL_VP_ATTR_EN_1_15_XYZW 0xf0000000 -#define NV50TCL_VP_ATTR_EN_1_14_SHIFT 24 -#define NV50TCL_VP_ATTR_EN_1_14_MASK 0x0f000000 -#define NV50TCL_VP_ATTR_EN_1_14_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_14_XNNN 0x01000000 -#define NV50TCL_VP_ATTR_EN_1_14_NYNN 0x02000000 -#define NV50TCL_VP_ATTR_EN_1_14_XYNN 0x03000000 -#define NV50TCL_VP_ATTR_EN_1_14_NNZN 0x04000000 -#define NV50TCL_VP_ATTR_EN_1_14_XNZN 0x05000000 -#define NV50TCL_VP_ATTR_EN_1_14_NYZN 0x06000000 -#define NV50TCL_VP_ATTR_EN_1_14_XYZN 0x07000000 -#define NV50TCL_VP_ATTR_EN_1_14_NNNW 0x08000000 -#define NV50TCL_VP_ATTR_EN_1_14_XNNW 0x09000000 -#define NV50TCL_VP_ATTR_EN_1_14_NYNW 0x0a000000 -#define NV50TCL_VP_ATTR_EN_1_14_XYNW 0x0b000000 -#define NV50TCL_VP_ATTR_EN_1_14_NNZW 0x0c000000 -#define NV50TCL_VP_ATTR_EN_1_14_XNZW 0x0d000000 -#define NV50TCL_VP_ATTR_EN_1_14_NYZW 0x0e000000 -#define NV50TCL_VP_ATTR_EN_1_14_XYZW 0x0f000000 -#define NV50TCL_VP_ATTR_EN_1_13_SHIFT 20 -#define NV50TCL_VP_ATTR_EN_1_13_MASK 0x00f00000 -#define NV50TCL_VP_ATTR_EN_1_13_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_13_XNNN 0x00100000 -#define NV50TCL_VP_ATTR_EN_1_13_NYNN 0x00200000 -#define NV50TCL_VP_ATTR_EN_1_13_XYNN 0x00300000 -#define NV50TCL_VP_ATTR_EN_1_13_NNZN 0x00400000 -#define NV50TCL_VP_ATTR_EN_1_13_XNZN 0x00500000 -#define NV50TCL_VP_ATTR_EN_1_13_NYZN 0x00600000 -#define NV50TCL_VP_ATTR_EN_1_13_XYZN 0x00700000 -#define NV50TCL_VP_ATTR_EN_1_13_NNNW 0x00800000 -#define NV50TCL_VP_ATTR_EN_1_13_XNNW 0x00900000 -#define NV50TCL_VP_ATTR_EN_1_13_NYNW 0x00a00000 -#define NV50TCL_VP_ATTR_EN_1_13_XYNW 0x00b00000 -#define NV50TCL_VP_ATTR_EN_1_13_NNZW 0x00c00000 -#define NV50TCL_VP_ATTR_EN_1_13_XNZW 0x00d00000 -#define NV50TCL_VP_ATTR_EN_1_13_NYZW 0x00e00000 -#define NV50TCL_VP_ATTR_EN_1_13_XYZW 0x00f00000 -#define NV50TCL_VP_ATTR_EN_1_12_SHIFT 16 -#define NV50TCL_VP_ATTR_EN_1_12_MASK 0x000f0000 -#define NV50TCL_VP_ATTR_EN_1_12_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_12_XNNN 0x00010000 -#define NV50TCL_VP_ATTR_EN_1_12_NYNN 0x00020000 -#define NV50TCL_VP_ATTR_EN_1_12_XYNN 0x00030000 -#define NV50TCL_VP_ATTR_EN_1_12_NNZN 0x00040000 -#define NV50TCL_VP_ATTR_EN_1_12_XNZN 0x00050000 -#define NV50TCL_VP_ATTR_EN_1_12_NYZN 0x00060000 -#define NV50TCL_VP_ATTR_EN_1_12_XYZN 0x00070000 -#define NV50TCL_VP_ATTR_EN_1_12_NNNW 0x00080000 -#define NV50TCL_VP_ATTR_EN_1_12_XNNW 0x00090000 -#define NV50TCL_VP_ATTR_EN_1_12_NYNW 0x000a0000 -#define NV50TCL_VP_ATTR_EN_1_12_XYNW 0x000b0000 -#define NV50TCL_VP_ATTR_EN_1_12_NNZW 0x000c0000 -#define NV50TCL_VP_ATTR_EN_1_12_XNZW 0x000d0000 -#define NV50TCL_VP_ATTR_EN_1_12_NYZW 0x000e0000 -#define NV50TCL_VP_ATTR_EN_1_12_XYZW 0x000f0000 -#define NV50TCL_VP_ATTR_EN_1_11_SHIFT 12 -#define NV50TCL_VP_ATTR_EN_1_11_MASK 0x0000f000 -#define NV50TCL_VP_ATTR_EN_1_11_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_11_XNNN 0x00001000 -#define NV50TCL_VP_ATTR_EN_1_11_NYNN 0x00002000 -#define NV50TCL_VP_ATTR_EN_1_11_XYNN 0x00003000 -#define NV50TCL_VP_ATTR_EN_1_11_NNZN 0x00004000 -#define NV50TCL_VP_ATTR_EN_1_11_XNZN 0x00005000 -#define NV50TCL_VP_ATTR_EN_1_11_NYZN 0x00006000 -#define NV50TCL_VP_ATTR_EN_1_11_XYZN 0x00007000 -#define NV50TCL_VP_ATTR_EN_1_11_NNNW 0x00008000 -#define NV50TCL_VP_ATTR_EN_1_11_XNNW 0x00009000 -#define NV50TCL_VP_ATTR_EN_1_11_NYNW 0x0000a000 -#define NV50TCL_VP_ATTR_EN_1_11_XYNW 0x0000b000 -#define NV50TCL_VP_ATTR_EN_1_11_NNZW 0x0000c000 -#define NV50TCL_VP_ATTR_EN_1_11_XNZW 0x0000d000 -#define NV50TCL_VP_ATTR_EN_1_11_NYZW 0x0000e000 -#define NV50TCL_VP_ATTR_EN_1_11_XYZW 0x0000f000 -#define NV50TCL_VP_ATTR_EN_1_10_SHIFT 8 -#define NV50TCL_VP_ATTR_EN_1_10_MASK 0x00000f00 -#define NV50TCL_VP_ATTR_EN_1_10_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_10_XNNN 0x00000100 -#define NV50TCL_VP_ATTR_EN_1_10_NYNN 0x00000200 -#define NV50TCL_VP_ATTR_EN_1_10_XYNN 0x00000300 -#define NV50TCL_VP_ATTR_EN_1_10_NNZN 0x00000400 -#define NV50TCL_VP_ATTR_EN_1_10_XNZN 0x00000500 -#define NV50TCL_VP_ATTR_EN_1_10_NYZN 0x00000600 -#define NV50TCL_VP_ATTR_EN_1_10_XYZN 0x00000700 -#define NV50TCL_VP_ATTR_EN_1_10_NNNW 0x00000800 -#define NV50TCL_VP_ATTR_EN_1_10_XNNW 0x00000900 -#define NV50TCL_VP_ATTR_EN_1_10_NYNW 0x00000a00 -#define NV50TCL_VP_ATTR_EN_1_10_XYNW 0x00000b00 -#define NV50TCL_VP_ATTR_EN_1_10_NNZW 0x00000c00 -#define NV50TCL_VP_ATTR_EN_1_10_XNZW 0x00000d00 -#define NV50TCL_VP_ATTR_EN_1_10_NYZW 0x00000e00 -#define NV50TCL_VP_ATTR_EN_1_10_XYZW 0x00000f00 -#define NV50TCL_VP_ATTR_EN_1_9_SHIFT 4 -#define NV50TCL_VP_ATTR_EN_1_9_MASK 0x000000f0 -#define NV50TCL_VP_ATTR_EN_1_9_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_9_XNNN 0x00000010 -#define NV50TCL_VP_ATTR_EN_1_9_NYNN 0x00000020 -#define NV50TCL_VP_ATTR_EN_1_9_XYNN 0x00000030 -#define NV50TCL_VP_ATTR_EN_1_9_NNZN 0x00000040 -#define NV50TCL_VP_ATTR_EN_1_9_XNZN 0x00000050 -#define NV50TCL_VP_ATTR_EN_1_9_NYZN 0x00000060 -#define NV50TCL_VP_ATTR_EN_1_9_XYZN 0x00000070 -#define NV50TCL_VP_ATTR_EN_1_9_NNNW 0x00000080 -#define NV50TCL_VP_ATTR_EN_1_9_XNNW 0x00000090 -#define NV50TCL_VP_ATTR_EN_1_9_NYNW 0x000000a0 -#define NV50TCL_VP_ATTR_EN_1_9_XYNW 0x000000b0 -#define NV50TCL_VP_ATTR_EN_1_9_NNZW 0x000000c0 -#define NV50TCL_VP_ATTR_EN_1_9_XNZW 0x000000d0 -#define NV50TCL_VP_ATTR_EN_1_9_NYZW 0x000000e0 -#define NV50TCL_VP_ATTR_EN_1_9_XYZW 0x000000f0 -#define NV50TCL_VP_ATTR_EN_1_8_SHIFT 0 -#define NV50TCL_VP_ATTR_EN_1_8_MASK 0x0000000f -#define NV50TCL_VP_ATTR_EN_1_8_NONE 0x00000000 -#define NV50TCL_VP_ATTR_EN_1_8_XNNN 0x00000001 -#define NV50TCL_VP_ATTR_EN_1_8_NYNN 0x00000002 -#define NV50TCL_VP_ATTR_EN_1_8_XYNN 0x00000003 -#define NV50TCL_VP_ATTR_EN_1_8_NNZN 0x00000004 -#define NV50TCL_VP_ATTR_EN_1_8_XNZN 0x00000005 -#define NV50TCL_VP_ATTR_EN_1_8_NYZN 0x00000006 -#define NV50TCL_VP_ATTR_EN_1_8_XYZN 0x00000007 -#define NV50TCL_VP_ATTR_EN_1_8_NNNW 0x00000008 -#define NV50TCL_VP_ATTR_EN_1_8_XNNW 0x00000009 -#define NV50TCL_VP_ATTR_EN_1_8_NYNW 0x0000000a -#define NV50TCL_VP_ATTR_EN_1_8_XYNW 0x0000000b -#define NV50TCL_VP_ATTR_EN_1_8_NNZW 0x0000000c -#define NV50TCL_VP_ATTR_EN_1_8_XNZW 0x0000000d -#define NV50TCL_VP_ATTR_EN_1_8_NYZW 0x0000000e -#define NV50TCL_VP_ATTR_EN_1_8_XYZW 0x0000000f -#define NV50TCL_LINE_STIPPLE_ENABLE 0x0000166c -#define NV50TCL_LINE_STIPPLE_PATTERN 0x00001680 -#define NV50TCL_POLYGON_STIPPLE_ENABLE 0x0000168c -#define NV50TCL_VP_REG_HPOS 0x000016bc -#define NV50TCL_VP_REG_HPOS_X_SHIFT 0 -#define NV50TCL_VP_REG_HPOS_X_MASK 0x000000ff -#define NV50TCL_VP_REG_HPOS_Y_SHIFT 8 -#define NV50TCL_VP_REG_HPOS_Y_MASK 0x0000ff00 -#define NV50TCL_VP_REG_HPOS_Z_SHIFT 16 -#define NV50TCL_VP_REG_HPOS_Z_MASK 0x00ff0000 -#define NV50TCL_VP_REG_HPOS_W_SHIFT 24 -#define NV50TCL_VP_REG_HPOS_W_MASK 0xff000000 -#define NV50TCL_VP_REG_COL0 0x000016c0 -#define NV50TCL_VP_REG_COL0_X_SHIFT 0 -#define NV50TCL_VP_REG_COL0_X_MASK 0x000000ff -#define NV50TCL_VP_REG_COL0_Y_SHIFT 8 -#define NV50TCL_VP_REG_COL0_Y_MASK 0x0000ff00 -#define NV50TCL_VP_REG_COL0_Z_SHIFT 16 -#define NV50TCL_VP_REG_COL0_Z_MASK 0x00ff0000 -#define NV50TCL_VP_REG_COL0_W_SHIFT 24 -#define NV50TCL_VP_REG_COL0_W_MASK 0xff000000 -#define NV50TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001700+((x)*4)) -#define NV50TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV50TCL_CULL_FACE_ENABLE 0x00001918 -#define NV50TCL_FRONT_FACE 0x0000191c -#define NV50TCL_FRONT_FACE_CW 0x00000900 -#define NV50TCL_FRONT_FACE_CCW 0x00000901 -#define NV50TCL_CULL_FACE 0x00001920 -#define NV50TCL_CULL_FACE_FRONT 0x00000404 -#define NV50TCL_CULL_FACE_BACK 0x00000405 -#define NV50TCL_CULL_FACE_FRONT_AND_BACK 0x00000408 -#define NV50TCL_LOGIC_OP_ENABLE 0x000019c4 -#define NV50TCL_LOGIC_OP 0x000019c8 -#define NV50TCL_LOGIC_OP_CLEAR 0x00001500 -#define NV50TCL_LOGIC_OP_AND 0x00001501 -#define NV50TCL_LOGIC_OP_AND_REVERSE 0x00001502 -#define NV50TCL_LOGIC_OP_COPY 0x00001503 -#define NV50TCL_LOGIC_OP_AND_INVERTED 0x00001504 -#define NV50TCL_LOGIC_OP_NOOP 0x00001505 -#define NV50TCL_LOGIC_OP_XOR 0x00001506 -#define NV50TCL_LOGIC_OP_OR 0x00001507 -#define NV50TCL_LOGIC_OP_NOR 0x00001508 -#define NV50TCL_LOGIC_OP_EQUIV 0x00001509 -#define NV50TCL_LOGIC_OP_INVERT 0x0000150a -#define NV50TCL_LOGIC_OP_OR_REVERSE 0x0000150b -#define NV50TCL_LOGIC_OP_COPY_INVERTED 0x0000150c -#define NV50TCL_LOGIC_OP_OR_INVERTED 0x0000150d -#define NV50TCL_LOGIC_OP_NAND 0x0000150e -#define NV50TCL_LOGIC_OP_SET 0x0000150f -#define NV50TCL_CLEAR_BUFFERS 0x000019d0 -#define NV50TCL_COLOR_MASK(x) (0x00001a00+((x)*4)) -#define NV50TCL_COLOR_MASK__SIZE 0x00000008 -#define NV50TCL_COLOR_MASK_R_SHIFT 0 -#define NV50TCL_COLOR_MASK_R_MASK 0x0000000f -#define NV50TCL_COLOR_MASK_G_SHIFT 4 -#define NV50TCL_COLOR_MASK_G_MASK 0x000000f0 -#define NV50TCL_COLOR_MASK_B_SHIFT 8 -#define NV50TCL_COLOR_MASK_B_MASK 0x00000f00 -#define NV50TCL_COLOR_MASK_A_SHIFT 12 -#define NV50TCL_COLOR_MASK_A_MASK 0x0000f000 - - -#define NV50_COMPUTE 0x000050c0 - -#define NV50_COMPUTE_DMA_UNK0 0x000001a0 -#define NV50_COMPUTE_DMA_STATUS 0x000001a4 -#define NV50_COMPUTE_DMA_UNK1 0x000001b8 -#define NV50_COMPUTE_DMA_UNK2 0x000001bc -#define NV50_COMPUTE_DMA_UNK3 0x000001c0 -#define NV50_COMPUTE_UNK4_HIGH 0x00000210 -#define NV50_COMPUTE_UNK4_LOW 0x00000214 -#define NV50_COMPUTE_UNK5_HIGH 0x00000218 -#define NV50_COMPUTE_UNK5_LOW 0x0000021c -#define NV50_COMPUTE_UNK6_HIGH 0x00000294 -#define NV50_COMPUTE_UNK6_LOW 0x00000298 -#define NV50_COMPUTE_CONST_BASE_HIGH 0x000002a4 -#define NV50_COMPUTE_CONST_BASE_LO 0x000002a8 -#define NV50_COMPUTE_CONST_SIZE_SEG 0x000002ac -#define NV50_COMPUTE_REG_COUNT 0x000002c0 -#define NV50_COMPUTE_STATUS_HIGH 0x00000310 -#define NV50_COMPUTE_STATUS_LOW 0x00000314 -#define NV50_COMPUTE_EXECUTE 0x0000031c -#define NV50_COMPUTE_USER_PARAM_COUNT 0x00000374 -#define NV50_COMPUTE_GRIDDIM_YX 0x000003a4 -#define NV50_COMPUTE_SHARED_SIZE 0x000003a8 -#define NV50_COMPUTE_BLOCKDIM_YX 0x000003ac -#define NV50_COMPUTE_BLOCKDIM_Z 0x000003b0 -#define NV50_COMPUTE_CALL_ADDRESS 0x000003b4 -#define NV50_COMPUTE_GLOBAL_BASE_HIGH(x) (0x00000400+((x)*32)) -#define NV50_COMPUTE_GLOBAL_BASE_HIGH__SIZE 0x00000010 -#define NV50_COMPUTE_GLOBAL_BASE_LOW(x) (0x00000404+((x)*32)) -#define NV50_COMPUTE_GLOBAL_BASE_LOW__SIZE 0x00000010 -#define NV50_COMPUTE_GLOBAL_LIMIT_HIGH(x) (0x00000408+((x)*32)) -#define NV50_COMPUTE_GLOBAL_LIMIT_HIGH__SIZE 0x00000010 -#define NV50_COMPUTE_GLOBAL_LIMIT_LOW(x) (0x0000040c+((x)*32)) -#define NV50_COMPUTE_GLOBAL_LIMIT_LOW__SIZE 0x00000010 -#define NV50_COMPUTE_GLOBAL_UNK(x) (0x00000410+((x)*32)) -#define NV50_COMPUTE_GLOBAL_UNK__SIZE 0x00000010 -#define NV50_COMPUTE_USER_PARAM(x) (0x00000600+((x)*4)) -#define NV50_COMPUTE_USER_PARAM__SIZE 0x00000040 - - -#define NV54TCL 0x00008297 - - - -#endif /* NOUVEAU_REG_H */ diff --git a/src/mesa/pipe/nouveau/nouveau_gldefs.h b/src/mesa/pipe/nouveau/nouveau_gldefs.h deleted file mode 100644 index e1015c93a27..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_gldefs.h +++ /dev/null @@ -1,196 +0,0 @@ -#ifndef __NOUVEAU_GLDEFS_H__ -#define __NOUVEAU_GLDEFS_H__ - -static INLINE unsigned -nvgl_blend_func(unsigned factor) -{ - switch (factor) { - case PIPE_BLENDFACTOR_ZERO: - return 0x0000; - case PIPE_BLENDFACTOR_ONE: - return 0x0001; - case PIPE_BLENDFACTOR_SRC_COLOR: - return 0x0300; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - return 0x0301; - case PIPE_BLENDFACTOR_SRC_ALPHA: - return 0x0302; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - return 0x0303; - case PIPE_BLENDFACTOR_DST_ALPHA: - return 0x0304; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - return 0x0305; - case PIPE_BLENDFACTOR_DST_COLOR: - return 0x0306; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - return 0x0307; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - return 0x0308; - case PIPE_BLENDFACTOR_CONST_COLOR: - return 0x8001; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - return 0x8002; - case PIPE_BLENDFACTOR_CONST_ALPHA: - return 0x8003; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - return 0x8004; - default: - return 0x0000; - } -} - -static INLINE unsigned -nvgl_blend_eqn(unsigned func) -{ - switch (func) { - case PIPE_BLEND_ADD: - return 0x8006; - case PIPE_BLEND_MIN: - return 0x8007; - case PIPE_BLEND_MAX: - return 0x8008; - case PIPE_BLEND_SUBTRACT: - return 0x800a; - case PIPE_BLEND_REVERSE_SUBTRACT: - return 0x800b; - default: - return 0x8006; - } -} - -static INLINE unsigned -nvgl_logicop_func(unsigned func) -{ - switch (func) { - case PIPE_LOGICOP_CLEAR: - return 0x1500; - case PIPE_LOGICOP_NOR: - return 0x1508; - case PIPE_LOGICOP_AND_INVERTED: - return 0x1504; - case PIPE_LOGICOP_COPY_INVERTED: - return 0x150c; - case PIPE_LOGICOP_AND_REVERSE: - return 0x1502; - case PIPE_LOGICOP_INVERT: - return 0x150a; - case PIPE_LOGICOP_XOR: - return 0x1506; - case PIPE_LOGICOP_NAND: - return 0x150e; - case PIPE_LOGICOP_AND: - return 0x1501; - case PIPE_LOGICOP_EQUIV: - return 0x1509; - case PIPE_LOGICOP_NOOP: - return 0x1505; - case PIPE_LOGICOP_OR_INVERTED: - return 0x150d; - case PIPE_LOGICOP_COPY: - return 0x1503; - case PIPE_LOGICOP_OR_REVERSE: - return 0x150b; - case PIPE_LOGICOP_OR: - return 0x1507; - case PIPE_LOGICOP_SET: - return 0x150f; - default: - return 0x1505; - } -} - -static INLINE unsigned -nvgl_comparison_op(unsigned op) -{ - switch (op) { - case PIPE_FUNC_NEVER: - return 0x0200; - case PIPE_FUNC_LESS: - return 0x0201; - case PIPE_FUNC_EQUAL: - return 0x0202; - case PIPE_FUNC_LEQUAL: - return 0x0203; - case PIPE_FUNC_GREATER: - return 0x0204; - case PIPE_FUNC_NOTEQUAL: - return 0x0205; - case PIPE_FUNC_GEQUAL: - return 0x0206; - case PIPE_FUNC_ALWAYS: - return 0x0207; - default: - return 0x0207; - } -} - -static INLINE unsigned -nvgl_polygon_mode(unsigned mode) -{ - switch (mode) { - case PIPE_POLYGON_MODE_POINT: - return 0x1b00; - case PIPE_POLYGON_MODE_LINE: - return 0x1b01; - case PIPE_POLYGON_MODE_FILL: - return 0x1b02; - default: - return 0x1b02; - } -} - -static INLINE unsigned -nvgl_stencil_op(unsigned op) -{ - switch (op) { - case PIPE_STENCIL_OP_ZERO: - return 0x0000; - case PIPE_STENCIL_OP_INVERT: - return 0x150a; - case PIPE_STENCIL_OP_KEEP: - return 0x1e00; - case PIPE_STENCIL_OP_REPLACE: - return 0x1e01; - case PIPE_STENCIL_OP_INCR: - return 0x1e02; - case PIPE_STENCIL_OP_DECR: - return 0x1e03; - case PIPE_STENCIL_OP_INCR_WRAP: - return 0x8507; - case PIPE_STENCIL_OP_DECR_WRAP: - return 0x8508; - default: - return 0x1e00; - } -} - -static INLINE unsigned -nvgl_primitive(unsigned prim) { - switch (prim) { - case PIPE_PRIM_POINTS: - return 0x0001; - case PIPE_PRIM_LINES: - return 0x0002; - case PIPE_PRIM_LINE_LOOP: - return 0x0003; - case PIPE_PRIM_LINE_STRIP: - return 0x0004; - case PIPE_PRIM_TRIANGLES: - return 0x0005; - case PIPE_PRIM_TRIANGLE_STRIP: - return 0x0006; - case PIPE_PRIM_TRIANGLE_FAN: - return 0x0007; - case PIPE_PRIM_QUADS: - return 0x0008; - case PIPE_PRIM_QUAD_STRIP: - return 0x0009; - case PIPE_PRIM_POLYGON: - return 0x000a; - default: - return 0x0001; - } -} - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_grobj.h b/src/mesa/pipe/nouveau/nouveau_grobj.h deleted file mode 100644 index 8f5abf90514..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_grobj.h +++ /dev/null @@ -1,35 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_GROBJ_H__ -#define __NOUVEAU_GROBJ_H__ - -#include "nouveau_channel.h" - -struct nouveau_grobj { - struct nouveau_channel *channel; - int grclass; - uint32_t handle; - int subc; -}; - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_notifier.h b/src/mesa/pipe/nouveau/nouveau_notifier.h deleted file mode 100644 index 35adde1e324..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_notifier.h +++ /dev/null @@ -1,43 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_NOTIFIER_H__ -#define __NOUVEAU_NOTIFIER_H__ - -#define NV_NOTIFIER_SIZE 32 -#define NV_NOTIFY_TIME_0 0x00000000 -#define NV_NOTIFY_TIME_1 0x00000004 -#define NV_NOTIFY_RETURN_VALUE 0x00000008 -#define NV_NOTIFY_STATE 0x0000000C -#define NV_NOTIFY_STATE_STATUS_MASK 0xFF000000 -#define NV_NOTIFY_STATE_STATUS_SHIFT 24 -#define NV_NOTIFY_STATE_STATUS_COMPLETED 0x00 -#define NV_NOTIFY_STATE_STATUS_IN_PROCESS 0x01 -#define NV_NOTIFY_STATE_ERROR_CODE_MASK 0x0000FFFF -#define NV_NOTIFY_STATE_ERROR_CODE_SHIFT 0 - -struct nouveau_notifier { - struct nouveau_channel *channel; - uint32_t handle; -}; - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_push.h b/src/mesa/pipe/nouveau/nouveau_push.h deleted file mode 100644 index 679472669b9..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_push.h +++ /dev/null @@ -1,83 +0,0 @@ -#ifndef __NOUVEAU_PUSH_H__ -#define __NOUVEAU_PUSH_H__ - -#include "pipe/nouveau/nouveau_winsys.h" - -#ifndef NOUVEAU_PUSH_CONTEXT -#error undefined push context -#endif - -#define OUT_RING(data) do { \ - NOUVEAU_PUSH_CONTEXT(pc); \ - (*pc->nvws->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - NOUVEAU_PUSH_CONTEXT(pc); \ - memcpy(pc->nvws->channel->pushbuf->cur, (src), (size) * 4); \ - pc->nvws->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - NOUVEAU_PUSH_CONTEXT(pc); \ - if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ - OUT_RING((pc->obj->subc << 13) | ((size) << 18) | (mthd)); \ - pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - -#define FIRE_RING() do { \ - NOUVEAU_PUSH_CONTEXT(pc); \ - pc->nvws->push_flush(pc->nvws->channel, 0); \ -} while(0) - -#define OUT_RELOC(bo,data,flags,vor,tor) do { \ - NOUVEAU_PUSH_CONTEXT(pc); \ - pc->nvws->push_reloc(pc->nvws->channel, \ - pc->nvws->channel->pushbuf->cur++, \ - (bo), (data), (flags), (vor), (tor)); \ -} while(0) - -/* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) - -/* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - pc->nvws->channel->vram->handle, \ - pc->nvws->channel->gart->handle); \ -} while(0) - -/* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) - -/* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) - -/* A reloc which'll recombine into a NV_DMA_METHOD packet header */ -#define OUT_RELOCm(bo, flags, obj, mthd, size) do { \ - NOUVEAU_PUSH_CONTEXT(pc); \ - if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ - OUT_RELOCd((bo), (pc->obj->subc << 13) | ((size) << 18) | (mthd), \ - (flags), 0, 0); \ - pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_pushbuf.h b/src/mesa/pipe/nouveau/nouveau_pushbuf.h deleted file mode 100644 index 19097650982..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_pushbuf.h +++ /dev/null @@ -1,32 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_PUSHBUF_H__ -#define __NOUVEAU_PUSHBUF_H__ - -struct nouveau_pushbuf { - struct nouveau_channel *channel; - unsigned remaining; - uint32_t *cur; -}; - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_resource.h b/src/mesa/pipe/nouveau/nouveau_resource.h deleted file mode 100644 index 1af7961d301..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_resource.h +++ /dev/null @@ -1,37 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_RESOURCE_H__ -#define __NOUVEAU_RESOURCE_H__ - -struct nouveau_resource { - struct nouveau_resource *prev; - struct nouveau_resource *next; - - int in_use; - void *priv; - - unsigned int start; - unsigned int size; -}; - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_stateobj.h b/src/mesa/pipe/nouveau/nouveau_stateobj.h deleted file mode 100644 index 07c31b014a5..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_stateobj.h +++ /dev/null @@ -1,141 +0,0 @@ -#ifndef __NOUVEAU_STATEOBJ_H__ -#define __NOUVEAU_STATEOBJ_H__ - -#include "pipe/p_util.h" - -struct nouveau_stateobj_reloc { - struct pipe_buffer *bo; - - unsigned offset; - unsigned packet; - - unsigned data; - unsigned flags; - unsigned vor; - unsigned tor; -}; - -struct nouveau_stateobj { - int refcount; - - unsigned *push; - struct nouveau_stateobj_reloc *reloc; - - unsigned *cur; - unsigned cur_packet; - unsigned cur_reloc; -}; - -static INLINE struct nouveau_stateobj * -so_new(unsigned push, unsigned reloc) -{ - struct nouveau_stateobj *so; - - so = MALLOC(sizeof(struct nouveau_stateobj)); - so->refcount = 1; - so->push = MALLOC(sizeof(unsigned) * push); - so->reloc = MALLOC(sizeof(struct nouveau_stateobj_reloc) * reloc); - - so->cur = so->push; - so->cur_reloc = so->cur_packet = 0; - - return so; -} - -static INLINE void -so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) -{ - struct nouveau_stateobj *so; - - so = *pso; - if (so) { - if (--so->refcount <= 0) { - free(so->push); - free(so->reloc); - free(so); - } - *pso = NULL; - } - - if (ref) { - ref->refcount++; - *pso = ref; - } -} - -static INLINE void -so_data(struct nouveau_stateobj *so, unsigned data) -{ - (*so->cur++) = (data); - so->cur_packet += 4; -} - -static INLINE void -so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, - unsigned mthd, unsigned size) -{ - so->cur_packet = (gr->subc << 13) | (1 << 18) | (mthd - 4); - so_data(so, (gr->subc << 13) | (size << 18) | mthd); -} - -static INLINE void -so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, - unsigned data, unsigned flags, unsigned vor, unsigned tor) -{ - struct nouveau_stateobj_reloc *r = &so->reloc[so->cur_reloc++]; - - r->bo = bo; - r->offset = so->cur - so->push; - r->packet = so->cur_packet; - r->data = data; - r->flags = flags; - r->vor = vor; - r->tor = tor; - so_data(so, data); -} - -static INLINE void -so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) -{ - struct nouveau_pushbuf *pb = nvws->channel->pushbuf; - unsigned nr, i; - - nr = so->cur - so->push; - if (pb->remaining < nr) - nvws->push_flush(nvws->channel, nr); - pb->remaining -= nr; - - memcpy(pb->cur, so->push, nr * 4); - for (i = 0; i < so->cur_reloc; i++) { - struct nouveau_stateobj_reloc *r = &so->reloc[i]; - - nvws->push_reloc(nvws->channel, pb->cur + r->offset, r->bo, - r->data, r->flags, r->vor, r->tor); - } - pb->cur += nr; -} - -static INLINE void -so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) -{ - struct nouveau_pushbuf *pb = nvws->channel->pushbuf; - unsigned i; - - i = so->cur_reloc << 1; - if (nvws->channel->pushbuf->remaining < i) - nvws->push_flush(nvws->channel, i); - nvws->channel->pushbuf->remaining -= i; - - for (i = 0; i < so->cur_reloc; i++) { - struct nouveau_stateobj_reloc *r = &so->reloc[i]; - - nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->packet, - (r->flags & - (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART)) | - NOUVEAU_BO_DUMMY, 0, 0); - nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->data, - r->flags | NOUVEAU_BO_DUMMY, r->vor, r->tor); - } -} - -#endif diff --git a/src/mesa/pipe/nouveau/nouveau_winsys.h b/src/mesa/pipe/nouveau/nouveau_winsys.h deleted file mode 100644 index 818ae9afae2..00000000000 --- a/src/mesa/pipe/nouveau/nouveau_winsys.h +++ /dev/null @@ -1,61 +0,0 @@ -#ifndef NOUVEAU_WINSYS_H -#define NOUVEAU_WINSYS_H - -#include -#include "pipe/p_winsys.h" -#include "pipe/p_defines.h" - -#include "pipe/nouveau/nouveau_bo.h" -#include "pipe/nouveau/nouveau_channel.h" -#include "pipe/nouveau/nouveau_class.h" -#include "pipe/nouveau/nouveau_grobj.h" -#include "pipe/nouveau/nouveau_notifier.h" -#include "pipe/nouveau/nouveau_resource.h" -#include "pipe/nouveau/nouveau_pushbuf.h" - -struct nouveau_winsys { - struct nouveau_context *nv; - - struct nouveau_channel *channel; - - int (*res_init)(struct nouveau_resource **heap, unsigned start, - unsigned size); - int (*res_alloc)(struct nouveau_resource *heap, int size, void *priv, - struct nouveau_resource **); - void (*res_free)(struct nouveau_resource **); - - int (*push_reloc)(struct nouveau_channel *, void *ptr, - struct pipe_buffer *, uint32_t data, - uint32_t flags, uint32_t vor, uint32_t tor); - int (*push_flush)(struct nouveau_channel *, unsigned size); - - int (*grobj_alloc)(struct nouveau_winsys *, int grclass, - struct nouveau_grobj **); - void (*grobj_free)(struct nouveau_grobj **); - - int (*notifier_alloc)(struct nouveau_winsys *, int count, - struct nouveau_notifier **); - void (*notifier_free)(struct nouveau_notifier **); - void (*notifier_reset)(struct nouveau_notifier *, int id); - uint32_t (*notifier_status)(struct nouveau_notifier *, int id); - uint32_t (*notifier_retval)(struct nouveau_notifier *, int id); - int (*notifier_wait)(struct nouveau_notifier *, int id, - int status, int timeout); - - int (*surface_copy)(struct nouveau_winsys *, struct pipe_surface *, - unsigned, unsigned, struct pipe_surface *, - unsigned, unsigned, unsigned, unsigned); - int (*surface_fill)(struct nouveau_winsys *, struct pipe_surface *, - unsigned, unsigned, unsigned, unsigned, unsigned); -}; - -extern struct pipe_context * -nv30_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); - -extern struct pipe_context * -nv40_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); - -extern struct pipe_context * -nv50_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); - -#endif diff --git a/src/mesa/pipe/nv30/Makefile b/src/mesa/pipe/nv30/Makefile deleted file mode 100644 index dd4b7e73cda..00000000000 --- a/src/mesa/pipe/nv30/Makefile +++ /dev/null @@ -1,29 +0,0 @@ -TOP = ../../../.. -include $(TOP)/configs/current - -LIBNAME = nv30 - -DRIVER_SOURCES = \ - nv30_clear.c \ - nv30_context.c \ - nv30_draw.c \ - nv30_fragprog.c \ - nv30_fragtex.c \ - nv30_miptree.c \ - nv30_query.c \ - nv30_state.c \ - nv30_state_emit.c \ - nv30_surface.c \ - nv30_vbo.c \ - nv30_vertprog.c - -C_SOURCES = \ - $(COMMON_SOURCES) \ - $(DRIVER_SOURCES) - -ASM_SOURCES = - -include ../Makefile.template - -symlinks: - diff --git a/src/mesa/pipe/nv30/nv30_clear.c b/src/mesa/pipe/nv30/nv30_clear.c deleted file mode 100644 index 71f413588ee..00000000000 --- a/src/mesa/pipe/nv30/nv30_clear.c +++ /dev/null @@ -1,12 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "nv30_context.h" - -void -nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue) -{ - pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); -} diff --git a/src/mesa/pipe/nv30/nv30_context.c b/src/mesa/pipe/nv30/nv30_context.c deleted file mode 100644 index d12aab85d8c..00000000000 --- a/src/mesa/pipe/nv30/nv30_context.c +++ /dev/null @@ -1,431 +0,0 @@ -#include "pipe/draw/draw_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "pipe/p_util.h" - -#include "nv30_context.h" - -static const char * -nv30_get_name(struct pipe_context *pipe) -{ - struct nv30_context *nv30 = nv30_context(pipe); - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", nv30->chipset); - return buffer; -} - -static const char * -nv30_get_vendor(struct pipe_context *pipe) -{ - return "nouveau"; -} - -static int -nv30_get_param(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 16; - case PIPE_CAP_NPOT_TEXTURES: - return 0; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 1; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 1; - case PIPE_CAP_POINT_SPRITE: - return 1; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 2; - case PIPE_CAP_OCCLUSION_QUERY: - return 1; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 1; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 13; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 10; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 13; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv30_get_paramf(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} - -static void -nv30_flush(struct pipe_context *pipe, unsigned flags) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_winsys *nvws = nv30->nvws; - - if (flags & PIPE_FLUSH_TEXTURE_CACHE) { - BEGIN_RING(rankine, 0x1fd8, 1); - OUT_RING (2); - BEGIN_RING(rankine, 0x1fd8, 1); - OUT_RING (1); - } - - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv30->sync, 0); - BEGIN_RING(rankine, 0x104, 1); - OUT_RING (0); - BEGIN_RING(rankine, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv30->sync, 0, 0, 2000); -} - -static void -nv30_destroy(struct pipe_context *pipe) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_winsys *nvws = nv30->nvws; - - if (nv30->draw) - draw_destroy(nv30->draw); - - nvws->res_free(&nv30->vertprog.exec_heap); - nvws->res_free(&nv30->vertprog.data_heap); - - nvws->res_free(&nv30->query_heap); - nvws->notifier_free(&nv30->query); - - nvws->notifier_free(&nv30->sync); - - nvws->grobj_free(&nv30->rankine); - - free(nv30); -} - -static boolean -nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) -{ - struct nouveau_winsys *nvws = nv30->nvws; - int ret; - int i; - - ret = nvws->grobj_alloc(nvws, rankine_class, &nv30->rankine); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - BEGIN_RING(rankine, NV34TCL_DMA_NOTIFY, 1); - OUT_RING (nv30->sync->handle); - BEGIN_RING(rankine, NV34TCL_DMA_TEXTURE0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(rankine, NV34TCL_DMA_VTXBUF0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); -/* BEGIN_RING(rankine, NV34TCL_DMA_FENCE, 2); - OUT_RING (0); - OUT_RING (nv30->query->handle);*/ - BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY7, 1); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY8, 1); - OUT_RING (nvws->channel->vram->handle); - - for (i=1; i<8; i++) { - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(i), 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_VERT(i), 1); - OUT_RING (0); - } - - BEGIN_RING(rankine, 0x220, 1); - OUT_RING (1); - - BEGIN_RING(rankine, 0x03b0, 1); - OUT_RING (0x00100000); - BEGIN_RING(rankine, 0x1454, 1); - OUT_RING (0); - BEGIN_RING(rankine, 0x1d80, 1); - OUT_RING (3); - BEGIN_RING(rankine, 0x1450, 1); - OUT_RING (0x00030004); - - /* NEW */ - BEGIN_RING(rankine, 0x1e98, 1); - OUT_RING (0); - BEGIN_RING(rankine, 0x17e0, 3); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0x3f800000); - BEGIN_RING(rankine, 0x1f80, 16); - OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); - OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); - OUT_RING (0x0000ffff); - OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); - OUT_RING (0); OUT_RING (0); OUT_RING (0); - - BEGIN_RING(rankine, 0x120, 3); - OUT_RING (0); - OUT_RING (1); - OUT_RING (2); - - BEGIN_RING(rankine, 0x1d88, 1); - OUT_RING (0x00001200); - - BEGIN_RING(rankine, NV34TCL_RC_ENABLE, 1); - OUT_RING (0); - - /* Attempt to setup a known state.. Probably missing a heap of - * stuff here.. - */ - BEGIN_RING(rankine, NV34TCL_STENCIL_FRONT_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_DEPTH_WRITE_ENABLE, 2); - OUT_RING (0); /* wr disable */ - OUT_RING (0); /* test disable */ - BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); - OUT_RING (0x01010101); /* TR,TR,TR,TR */ - BEGIN_RING(rankine, NV34TCL_CULL_FACE_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 5); - OUT_RING (0); /* Blend enable */ - OUT_RING (0); /* Blend src */ - OUT_RING (0); /* Blend dst */ - OUT_RING (0x00000000); /* Blend colour */ - OUT_RING (0x8006); /* FUNC_ADD */ - BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (0); - OUT_RING (0x1503 /*GL_COPY*/); - BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); - OUT_RING (1); - BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); - OUT_RING (0x1d01 /*GL_SMOOTH*/); - BEGIN_RING(rankine, NV34TCL_POLYGON_OFFSET_FACTOR,2); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 2); - OUT_RING (0x1b02 /*GL_FILL*/); - OUT_RING (0x1b02 /*GL_FILL*/); - /* - Disable texture units - * - Set fragprog to MOVR result.color, fragment.color */ - for (i=0;i<16;i++) { - BEGIN_RING(rankine, - NV34TCL_TX_ENABLE(i), 1); - OUT_RING (0); - } - /* Polygon stipple */ - BEGIN_RING(rankine, - NV34TCL_POLYGON_STIPPLE_PATTERN(0), 0x20); - for (i=0;i<0x20;i++) - OUT_RING (0xFFFFFFFF); - - int w=4096; - int h=4096; - int pitch=4096*4; - BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 5); - OUT_RING (w<<16); - OUT_RING (h<<16); - OUT_RING (0x148); /* format */ - OUT_RING (pitch << 16 | pitch); - OUT_RING (0x0); - BEGIN_RING(rankine, 0x0a00, 2); - OUT_RING ((w<<16) | 0); - OUT_RING ((h<<16) | 0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING ((w-1)<<16); - OUT_RING ((h-1)<<16); - BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); - OUT_RING (w<<16); - OUT_RING (h<<16); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); - OUT_RING (w<<16); - OUT_RING (h<<16); - - BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - - BEGIN_RING(rankine, NV34TCL_MODELVIEW_MATRIX(0), 16); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - - BEGIN_RING(rankine, NV34TCL_PROJECTION_MATRIX(0), 16); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - - BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); - OUT_RING (4096<<16); - OUT_RING (4096<<16); - - BEGIN_RING(rankine, NV34TCL_MULTISAMPLE_CONTROL, 1); - OUT_RING (0xffff0000); - - FIRE_RING (); - return TRUE; -} - -#define NV30TCL_CHIPSET_3X_MASK 0x00000003 -#define NV34TCL_CHIPSET_3X_MASK 0x00000010 -#define NV35TCL_CHIPSET_3X_MASK 0x000001e0 - -struct pipe_context * -nv30_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv30_context *nv30; - int rankine_class = 0, ret; - - if ((chipset & 0xf0) != 0x30) { - NOUVEAU_ERR("Not a NV3X chipset\n"); - return NULL; - } - - if (NV30TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { - rankine_class = 0x0397; - } else if (NV34TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { - rankine_class = 0x0697; - } else if (NV35TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { - rankine_class = 0x0497; - } else { - NOUVEAU_ERR("Unknown NV3X chipset: NV%02x\n", chipset); - return NULL; - } - - nv30 = CALLOC_STRUCT(nv30_context); - if (!nv30) - return NULL; - nv30->chipset = chipset; - nv30->nvws = nvws; - - /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &nv30->sync); - if (ret) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Query objects */ - ret = nvws->notifier_alloc(nvws, 32, &nv30->query); - if (ret) { - NOUVEAU_ERR("Error initialising query objects: %d\n", ret); - nv30_destroy(&nv30->pipe); - return NULL; - } - - ret = nvws->res_init(&nv30->query_heap, 0, 32); - if (ret) { - NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Vtxprog resources */ - if (nvws->res_init(&nv30->vertprog.exec_heap, 0, 512) || - nvws->res_init(&nv30->vertprog.data_heap, 0, 256)) { - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Static rankine initialisation */ - if (!nv30_init_hwctx(nv30, rankine_class)) { - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Pipe context setup */ - nv30->pipe.winsys = pipe_winsys; - - nv30->pipe.destroy = nv30_destroy; - nv30->pipe.get_name = nv30_get_name; - nv30->pipe.get_vendor = nv30_get_vendor; - nv30->pipe.get_param = nv30_get_param; - nv30->pipe.get_paramf = nv30_get_paramf; - - nv30->pipe.draw_arrays = nv30_draw_arrays; - nv30->pipe.draw_elements = nv30_draw_elements; - nv30->pipe.clear = nv30_clear; - - nv30->pipe.flush = nv30_flush; - - nv30_init_query_functions(nv30); - nv30_init_surface_functions(nv30); - nv30_init_state_functions(nv30); - nv30_init_miptree_functions(nv30); - - nv30->draw = draw_create(); - assert(nv30->draw); - draw_set_rasterize_stage(nv30->draw, nv30_draw_render_stage(nv30)); - - return &nv30->pipe; -} - diff --git a/src/mesa/pipe/nv30/nv30_context.h b/src/mesa/pipe/nv30/nv30_context.h deleted file mode 100644 index f6c69545991..00000000000 --- a/src/mesa/pipe/nv30/nv30_context.h +++ /dev/null @@ -1,136 +0,0 @@ -#ifndef __NV30_CONTEXT_H__ -#define __NV30_CONTEXT_H__ - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/draw/draw_vertex.h" - -#include "pipe/nouveau/nouveau_winsys.h" -#include "pipe/nouveau/nouveau_gldefs.h" - -#define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv30_context *ctx = nv30 -#include "pipe/nouveau/nouveau_push.h" - -#include "nv30_state.h" - -#define NOUVEAU_ERR(fmt, args...) \ - fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); -#define NOUVEAU_MSG(fmt, args...) \ - fprintf(stderr, "nouveau: "fmt, ##args); - -#define NV30_NEW_VERTPROG (1 << 1) -#define NV30_NEW_FRAGPROG (1 << 2) -#define NV30_NEW_ARRAYS (1 << 3) - -struct nv30_context { - struct pipe_context pipe; - struct nouveau_winsys *nvws; - - struct draw_context *draw; - - int chipset; - struct nouveau_grobj *rankine; - struct nouveau_notifier *sync; - - /* query objects */ - struct nouveau_notifier *query; - struct nouveau_resource *query_heap; - - uint32_t dirty; - - struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; - unsigned dirty_samplers; - unsigned fp_samplers; - unsigned vp_samplers; - - uint32_t rt_enable; - struct pipe_buffer *rt[4]; - struct pipe_buffer *zeta; - - struct { - struct pipe_buffer *buffer; - uint32_t format; - } tex[16]; - - unsigned vb_enable; - struct { - struct pipe_buffer *buffer; - unsigned delta; - } vb[16]; - - struct { - struct nouveau_resource *exec_heap; - struct nouveau_resource *data_heap; - - struct nv30_vertex_program *active; - - struct nv30_vertex_program *current; - struct pipe_buffer *constant_buf; - } vertprog; - - struct { - struct nv30_fragment_program *active; - - struct nv30_fragment_program *current; - struct pipe_buffer *constant_buf; - } fragprog; - - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; - struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; -}; - -static INLINE struct nv30_context * -nv30_context(struct pipe_context *pipe) -{ - return (struct nv30_context *)pipe; -} - -extern void nv30_init_state_functions(struct nv30_context *nv30); -extern void nv30_init_surface_functions(struct nv30_context *nv30); -extern void nv30_init_miptree_functions(struct nv30_context *nv30); -extern void nv30_init_query_functions(struct nv30_context *nv30); - -/* nv30_draw.c */ -extern struct draw_stage *nv30_draw_render_stage(struct nv30_context *nv30); - -/* nv30_vertprog.c */ -extern void nv30_vertprog_translate(struct nv30_context *, - struct nv30_vertex_program *); -extern void nv30_vertprog_bind(struct nv30_context *, - struct nv30_vertex_program *); -extern void nv30_vertprog_destroy(struct nv30_context *, - struct nv30_vertex_program *); - -/* nv30_fragprog.c */ -extern void nv30_fragprog_translate(struct nv30_context *, - struct nv30_fragment_program *); -extern void nv30_fragprog_bind(struct nv30_context *, - struct nv30_fragment_program *); -extern void nv30_fragprog_destroy(struct nv30_context *, - struct nv30_fragment_program *); - -/* nv30_fragtex.c */ -extern void nv30_fragtex_bind(struct nv30_context *); - -/* nv30_state.c and friends */ -extern void nv30_emit_hw_state(struct nv30_context *nv30); -extern void nv30_state_tex_update(struct nv30_context *nv30); - -/* nv30_vbo.c */ -extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, - unsigned start, unsigned count); -extern boolean nv30_draw_elements(struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, - unsigned indexSize, - unsigned mode, unsigned start, - unsigned count); - -/* nv30_clear.c */ -extern void nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue); - -#endif diff --git a/src/mesa/pipe/nv30/nv30_draw.c b/src/mesa/pipe/nv30/nv30_draw.c deleted file mode 100644 index bdeb975ca1e..00000000000 --- a/src/mesa/pipe/nv30/nv30_draw.c +++ /dev/null @@ -1,62 +0,0 @@ -#include "pipe/draw/draw_private.h" -#include "pipe/p_util.h" - -#include "nv30_context.h" - -struct nv30_draw_stage { - struct draw_stage draw; - struct nv30_context *nv30; -}; - -static void -nv30_draw_point(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv30_draw_line(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv30_draw_tri(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv30_draw_flush(struct draw_stage *draw, unsigned flags) -{ -} - -static void -nv30_draw_reset_stipple_counter(struct draw_stage *draw) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv30_draw_destroy(struct draw_stage *draw) -{ - free(draw); -} - -struct draw_stage * -nv30_draw_render_stage(struct nv30_context *nv30) -{ - struct nv30_draw_stage *nv30draw = CALLOC_STRUCT(nv30_draw_stage); - - nv30draw->nv30 = nv30; - nv30draw->draw.draw = nv30->draw; - nv30draw->draw.point = nv30_draw_point; - nv30draw->draw.line = nv30_draw_line; - nv30draw->draw.tri = nv30_draw_tri; - nv30draw->draw.flush = nv30_draw_flush; - nv30draw->draw.reset_stipple_counter = nv30_draw_reset_stipple_counter; - nv30draw->draw.destroy = nv30_draw_destroy; - - return &nv30draw->draw; -} - diff --git a/src/mesa/pipe/nv30/nv30_fragprog.c b/src/mesa/pipe/nv30/nv30_fragprog.c deleted file mode 100644 index 0db1ac868c7..00000000000 --- a/src/mesa/pipe/nv30/nv30_fragprog.c +++ /dev/null @@ -1,835 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" -#include "pipe/p_util.h" - -#include "pipe/p_shader_tokens.h" -#include "pipe/tgsi/util/tgsi_parse.h" -#include "pipe/tgsi/util/tgsi_util.h" - -#include "nv30_context.h" - -#define SWZ_X 0 -#define SWZ_Y 1 -#define SWZ_Z 2 -#define SWZ_W 3 -#define MASK_X 1 -#define MASK_Y 2 -#define MASK_Z 4 -#define MASK_W 8 -#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) -#define DEF_SCALE NV30_FP_OP_DST_SCALE_1X -#define DEF_CTEST NV30_FP_OP_COND_TR -#include "nv30_shader.h" - -#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) -#define neg(s) nv30_sr_neg((s)) -#define abs(s) nv30_sr_abs((s)) -#define scale(s,v) nv30_sr_scale((s), NV30_FP_OP_DST_SCALE_##v) - -#define MAX_CONSTS 128 -#define MAX_IMM 32 -struct nv30_fpc { - struct nv30_fragment_program *fp; - - uint attrib_map[PIPE_MAX_SHADER_INPUTS]; - - int high_temp; - int temp_temp_count; - int num_regs; - - uint depth_id; - uint colour_id; - - unsigned inst_offset; - - struct { - int pipe; - float vals[4]; - } consts[MAX_CONSTS]; - int nr_consts; - - struct nv30_sreg imm[MAX_IMM]; - unsigned nr_imm; -}; - -static INLINE struct nv30_sreg -temp(struct nv30_fpc *fpc) -{ - int idx; - - idx = fpc->temp_temp_count++; - idx += fpc->high_temp + 1; - return nv30_sr(NV30SR_TEMP, idx); -} - -static INLINE struct nv30_sreg -constant(struct nv30_fpc *fpc, int pipe, float vals[4]) -{ - int idx; - - if (fpc->nr_consts == MAX_CONSTS) - assert(0); - idx = fpc->nr_consts++; - - fpc->consts[idx].pipe = pipe; - if (pipe == -1) - memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float)); - return nv30_sr(NV30SR_CONST, idx); -} - -#define arith(cc,s,o,d,m,s0,s1,s2) \ - nv30_fp_arith((cc), (s), NV30_FP_OP_OPCODE_##o, \ - (d), (m), (s0), (s1), (s2)) -#define tex(cc,s,o,u,d,m,s0,s1,s2) \ - nv30_fp_tex((cc), (s), NV30_FP_OP_OPCODE_##o, (u), \ - (d), (m), (s0), none, none) - -static void -grow_insns(struct nv30_fpc *fpc, int size) -{ - struct nv30_fragment_program *fp = fpc->fp; - - fp->insn_len += size; - fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len); -} - -static void -emit_src(struct nv30_fpc *fpc, int pos, struct nv30_sreg src) -{ - struct nv30_fragment_program *fp = fpc->fp; - uint32_t *hw = &fp->insn[fpc->inst_offset]; - uint32_t sr = 0; - - switch (src.type) { - case NV30SR_INPUT: - sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); - hw[0] |= (src.index << NV30_FP_OP_INPUT_SRC_SHIFT); - break; - case NV30SR_OUTPUT: - sr |= NV30_FP_REG_SRC_HALF; - /* fall-through */ - case NV30SR_TEMP: - sr |= (NV30_FP_REG_TYPE_TEMP << NV30_FP_REG_TYPE_SHIFT); - sr |= (src.index << NV30_FP_REG_SRC_SHIFT); - break; - case NV30SR_CONST: - grow_insns(fpc, 4); - hw = &fp->insn[fpc->inst_offset]; - if (fpc->consts[src.index].pipe >= 0) { - struct nv30_fragment_program_data *fpd; - - fp->consts = realloc(fp->consts, ++fp->nr_consts * - sizeof(*fpd)); - fpd = &fp->consts[fp->nr_consts - 1]; - fpd->offset = fpc->inst_offset + 4; - fpd->index = fpc->consts[src.index].pipe; - memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4); - } else { - memcpy(&fp->insn[fpc->inst_offset + 4], - fpc->consts[src.index].vals, - sizeof(uint32_t) * 4); - } - - sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT); - break; - case NV30SR_NONE: - sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT); - break; - default: - assert(0); - } - - if (src.negate) - sr |= NV30_FP_REG_NEGATE; - - if (src.abs) - hw[1] |= (1 << (29 + pos)); - - sr |= ((src.swz[0] << NV30_FP_REG_SWZ_X_SHIFT) | - (src.swz[1] << NV30_FP_REG_SWZ_Y_SHIFT) | - (src.swz[2] << NV30_FP_REG_SWZ_Z_SHIFT) | - (src.swz[3] << NV30_FP_REG_SWZ_W_SHIFT)); - - hw[pos + 1] |= sr; -} - -static void -emit_dst(struct nv30_fpc *fpc, struct nv30_sreg dst) -{ - struct nv30_fragment_program *fp = fpc->fp; - uint32_t *hw = &fp->insn[fpc->inst_offset]; - - switch (dst.type) { - case NV30SR_TEMP: - if (fpc->num_regs < (dst.index + 1)) - fpc->num_regs = dst.index + 1; - break; - case NV30SR_OUTPUT: - if (dst.index == 1) { - fp->fp_control |= 0xe; - } else { - hw[0] |= NV30_FP_OP_OUT_REG_HALF; - } - break; - case NV30SR_NONE: - hw[0] |= (1 << 30); - break; - default: - assert(0); - } - - hw[0] |= (dst.index << NV30_FP_OP_OUT_REG_SHIFT); -} - -static void -nv30_fp_arith(struct nv30_fpc *fpc, int sat, int op, - struct nv30_sreg dst, int mask, - struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2) -{ - struct nv30_fragment_program *fp = fpc->fp; - uint32_t *hw; - - fpc->inst_offset = fp->insn_len; - grow_insns(fpc, 4); - hw = &fp->insn[fpc->inst_offset]; - memset(hw, 0, sizeof(uint32_t) * 4); - - if (op == NV30_FP_OP_OPCODE_KIL) - fp->fp_control |= NV34TCL_FP_CONTROL_USES_KIL; - hw[0] |= (op << NV30_FP_OP_OPCODE_SHIFT); - hw[0] |= (mask << NV30_FP_OP_OUTMASK_SHIFT); - hw[2] |= (dst.dst_scale << NV30_FP_OP_DST_SCALE_SHIFT); - - if (sat) - hw[0] |= NV30_FP_OP_OUT_SAT; - - if (dst.cc_update) - hw[0] |= NV30_FP_OP_COND_WRITE_ENABLE; - hw[1] |= (dst.cc_test << NV30_FP_OP_COND_SHIFT); - hw[1] |= ((dst.cc_swz[0] << NV30_FP_OP_COND_SWZ_X_SHIFT) | - (dst.cc_swz[1] << NV30_FP_OP_COND_SWZ_Y_SHIFT) | - (dst.cc_swz[2] << NV30_FP_OP_COND_SWZ_Z_SHIFT) | - (dst.cc_swz[3] << NV30_FP_OP_COND_SWZ_W_SHIFT)); - - emit_dst(fpc, dst); - emit_src(fpc, 0, s0); - emit_src(fpc, 1, s1); - emit_src(fpc, 2, s2); -} - -static void -nv30_fp_tex(struct nv30_fpc *fpc, int sat, int op, int unit, - struct nv30_sreg dst, int mask, - struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2) -{ - struct nv30_fragment_program *fp = fpc->fp; - - nv30_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); - - fp->insn[fpc->inst_offset] |= (unit << NV30_FP_OP_TEX_UNIT_SHIFT); - fp->samplers |= (1 << unit); -} - -static INLINE struct nv30_sreg -tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc) -{ - struct nv30_sreg src; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - src = nv30_sr(NV30SR_INPUT, - fpc->attrib_map[fsrc->SrcRegister.Index]); - break; - case TGSI_FILE_CONSTANT: - src = constant(fpc, fsrc->SrcRegister.Index, NULL); - break; - case TGSI_FILE_IMMEDIATE: - assert(fsrc->SrcRegister.Index < fpc->nr_imm); - src = fpc->imm[fsrc->SrcRegister.Index]; - break; - case TGSI_FILE_TEMPORARY: - src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index + 1); - if (fpc->high_temp < src.index) - fpc->high_temp = src.index; - break; - /* This is clearly insane, but gallium hands us shaders like this. - * Luckily fragprog results are just temp regs.. - */ - case TGSI_FILE_OUTPUT: - if (fsrc->SrcRegister.Index == fpc->colour_id) - return nv30_sr(NV30SR_OUTPUT, 0); - else - return nv30_sr(NV30SR_OUTPUT, 1); - break; - default: - NOUVEAU_ERR("bad src file\n"); - break; - } - - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; - return src; -} - -static INLINE struct nv30_sreg -tgsi_dst(struct nv30_fpc *fpc, const struct tgsi_full_dst_register *fdst) { - int idx; - - switch (fdst->DstRegister.File) { - case TGSI_FILE_OUTPUT: - if (fdst->DstRegister.Index == fpc->colour_id) - return nv30_sr(NV30SR_OUTPUT, 0); - else - return nv30_sr(NV30SR_OUTPUT, 1); - break; - case TGSI_FILE_TEMPORARY: - idx = fdst->DstRegister.Index + 1; - if (fpc->high_temp < idx) - fpc->high_temp = idx; - return nv30_sr(NV30SR_TEMP, idx); - case TGSI_FILE_NULL: - return nv30_sr(NV30SR_NONE, 0); - default: - NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); - return nv30_sr(NV30SR_NONE, 0); - } -} - -static INLINE int -tgsi_mask(uint tgsi) -{ - int mask = 0; - - if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; - if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; - if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; - if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; - return mask; -} - -static boolean -src_native_swz(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc, - struct nv30_sreg *src) -{ - const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); - struct nv30_sreg tgsi = tgsi_src(fpc, fsrc); - uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; - uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, - fsrc->SrcRegisterExtSwz.NegateY, - fsrc->SrcRegisterExtSwz.NegateZ, - fsrc->SrcRegisterExtSwz.NegateW }; - uint c; - - for (c = 0; c < 4; c++) { - switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: - mask |= (1 << c); - break; - case TGSI_EXTSWIZZLE_ZERO: - zero_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; - case TGSI_EXTSWIZZLE_ONE: - one_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; - default: - assert(0); - } - - if (!tgsi.negate && neg[c]) - neg_mask |= (1 << c); - } - - if (mask == MASK_ALL && !neg_mask) - return TRUE; - - *src = temp(fpc); - - if (mask) - arith(fpc, 0, MOV, *src, mask, tgsi, none, none); - - if (zero_mask) - arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); - - if (one_mask) - arith(fpc, 0, STR, *src, one_mask, *src, none, none); - - if (neg_mask) { - struct nv30_sreg one = temp(fpc); - arith(fpc, 0, STR, one, neg_mask, one, none, none); - arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none); - } - - return FALSE; -} - -static boolean -nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, - const struct tgsi_full_instruction *finst) -{ - const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); - struct nv30_sreg src[3], dst, tmp; - int mask, sat, unit; - int ai = -1, ci = -1; - int i; - - if (finst->Instruction.Opcode == TGSI_OPCODE_END) - return TRUE; - - fpc->temp_temp_count = 0; - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { - src[i] = tgsi_src(fpc, fsrc); - } - } - - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - case TGSI_FILE_CONSTANT: - case TGSI_FILE_TEMPORARY: - if (!src_native_swz(fpc, fsrc, &src[i])) - continue; - break; - default: - break; - } - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; - src[i] = tgsi_src(fpc, fsrc); - } else { - NOUVEAU_MSG("extra src attr %d\n", - fsrc->SrcRegister.Index); - src[i] = temp(fpc); - arith(fpc, 0, MOV, src[i], MASK_ALL, - tgsi_src(fpc, fsrc), none, none); - } - break; - case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; - src[i] = tgsi_src(fpc, fsrc); - } else { - src[i] = temp(fpc); - arith(fpc, 0, MOV, src[i], MASK_ALL, - tgsi_src(fpc, fsrc), none, none); - } - break; - case TGSI_FILE_TEMPORARY: - /* handled above */ - break; - case TGSI_FILE_SAMPLER: - unit = fsrc->SrcRegister.Index; - break; - case TGSI_FILE_OUTPUT: - break; - default: - NOUVEAU_ERR("bad src file\n"); - return FALSE; - } - } - - dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); - sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); - - switch (finst->Instruction.Opcode) { - case TGSI_OPCODE_ABS: - arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none); - break; - case TGSI_OPCODE_ADD: - arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_CMP: - tmp = temp(fpc); - arith(fpc, sat, MOV, dst, mask, src[2], none, none); - tmp.cc_update = 1; - arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); - dst.cc_test = NV30_VP_INST_COND_LT; - arith(fpc, sat, MOV, dst, mask, src[1], none, none); - break; - case TGSI_OPCODE_COS: - arith(fpc, sat, COS, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_DP3: - arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DP4: - arith(fpc, sat, DP4, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DPH: - tmp = temp(fpc); - arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none); - arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X), - swz(src[1], W, W, W, W), none); - break; - case TGSI_OPCODE_DST: - arith(fpc, sat, DST, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_EX2: - arith(fpc, sat, EX2, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FLR: - arith(fpc, sat, FLR, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FRC: - arith(fpc, sat, FRC, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_KIL: - arith(fpc, 0, KIL, none, 0, none, none, none); - break; - case TGSI_OPCODE_KILP: - dst = nv30_sr(NV30SR_NONE, 0); - dst.cc_update = 1; - arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); - dst.cc_update = 0; dst.cc_test = NV30_FP_OP_COND_LT; - arith(fpc, 0, KIL, dst, 0, none, none, none); - break; - case TGSI_OPCODE_LG2: - arith(fpc, sat, LG2, dst, mask, src[0], none, none); - break; -// case TGSI_OPCODE_LIT: - case TGSI_OPCODE_LRP: - tmp = temp(fpc); - arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); - arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); - break; - case TGSI_OPCODE_MAD: - arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); - break; - case TGSI_OPCODE_MAX: - arith(fpc, sat, MAX, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MIN: - arith(fpc, sat, MIN, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MOV: - arith(fpc, sat, MOV, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_MUL: - arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_POW: - tmp = temp(fpc); - arith(fpc, 0, LG2, tmp, MASK_X, - swz(src[0], X, X, X, X), none, none); - arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), - swz(src[1], X, X, X, X), none); - arith(fpc, sat, EX2, dst, mask, - swz(tmp, X, X, X, X), none, none); - break; - case TGSI_OPCODE_RCP: - arith(fpc, sat, RCP, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_RET: - assert(0); - break; - case TGSI_OPCODE_RFL: - tmp = temp(fpc); - arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); - arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); - arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, - swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); - arith(fpc, sat, MAD, dst, mask, - swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); - break; - case TGSI_OPCODE_RSQ: - tmp = temp(fpc); - arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, - abs(swz(src[0], X, X, X, X)), none, none); - arith(fpc, sat, EX2, dst, mask, - neg(swz(tmp, X, X, X, X)), none, none); - break; - case TGSI_OPCODE_SCS: - if (mask & MASK_X) { - arith(fpc, sat, COS, dst, MASK_X, - swz(src[0], X, X, X, X), none, none); - } - if (mask & MASK_Y) { - arith(fpc, sat, SIN, dst, MASK_Y, - swz(src[0], X, X, X, X), none, none); - } - break; - case TGSI_OPCODE_SIN: - arith(fpc, sat, SIN, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_SGE: - arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SLT: - arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SUB: - arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); - break; - case TGSI_OPCODE_TEX: - if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == - TGSI_EXTSWIZZLE_W) { - tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); - } else - tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_TXB: - tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_XPD: - tmp = temp(fpc); - arith(fpc, 0, MUL, tmp, mask, - swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); - arith(fpc, sat, MAD, dst, (mask & ~MASK_W), - swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), - neg(tmp)); - break; - default: - NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); - return FALSE; - } - - return TRUE; -} - -static boolean -nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, - const struct tgsi_full_declaration *fdec) -{ - int hw; - - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - hw = NV30_FP_OP_INPUT_SRC_POSITION; - break; - case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV30_FP_OP_INPUT_SRC_COL0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV30_FP_OP_INPUT_SRC_COL1; - } else { - NOUVEAU_ERR("bad colour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_FOG: - hw = NV30_FP_OP_INPUT_SRC_FOGC; - break; - case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { - hw = NV30_FP_OP_INPUT_SRC_TC(fdec->Semantic. - SemanticIndex); - } else { - NOUVEAU_ERR("bad generic semantic index\n"); - return FALSE; - } - break; - default: - NOUVEAU_ERR("bad input semantic\n"); - return FALSE; - } - - fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; - return TRUE; -} - -static boolean -nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc, - const struct tgsi_full_declaration *fdec) -{ - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - fpc->depth_id = fdec->u.DeclarationRange.First; - break; - case TGSI_SEMANTIC_COLOR: - fpc->colour_id = fdec->u.DeclarationRange.First; - break; - default: - NOUVEAU_ERR("bad output semantic\n"); - return FALSE; - } - - return TRUE; -} - -void -nv30_fragprog_translate(struct nv30_context *nv30, - struct nv30_fragment_program *fp) -{ - struct tgsi_parse_context parse; - struct nv30_fpc *fpc = NULL; - - fpc = CALLOC(1, sizeof(struct nv30_fpc)); - if (!fpc) - return; - fpc->fp = fp; - fpc->high_temp = -1; - fpc->num_regs = 2; - - tgsi_parse_init(&parse, fp->pipe->tokens); - - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - { - const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; - switch (fdec->Declaration.File) { - case TGSI_FILE_INPUT: - if (!nv30_fragprog_parse_decl_attrib(fpc, fdec)) - goto out_err; - break; - case TGSI_FILE_OUTPUT: - if (!nv30_fragprog_parse_decl_output(fpc, fdec)) - goto out_err; - break; - default: - break; - } - } - break; - case TGSI_TOKEN_TYPE_IMMEDIATE: - { - struct tgsi_full_immediate *imm; - float vals[4]; - int i; - - imm = &parse.FullToken.FullImmediate; - assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); - assert(fpc->nr_imm < MAX_IMM); - - for (i = 0; i < 4; i++) - vals[i] = imm->u.ImmediateFloat32[i].Float; - fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); - } - break; - case TGSI_TOKEN_TYPE_INSTRUCTION: - { - const struct tgsi_full_instruction *finst; - - finst = &parse.FullToken.FullInstruction; - if (!nv30_fragprog_parse_instruction(fpc, finst)) - goto out_err; - } - break; - default: - break; - } - } - - fp->fp_control |= (fpc->num_regs-1)/2; - fp->fp_reg_control = (1<<16)|0x4; - - /* Terminate final instruction */ - fp->insn[fpc->inst_offset] |= 0x00000001; - - /* Append NOP + END instruction, may or may not be necessary. */ - fpc->inst_offset = fp->insn_len; - grow_insns(fpc, 4); - fp->insn[fpc->inst_offset + 0] = 0x00000001; - fp->insn[fpc->inst_offset + 1] = 0x00000000; - fp->insn[fpc->inst_offset + 2] = 0x00000000; - fp->insn[fpc->inst_offset + 3] = 0x00000000; - - fp->translated = TRUE; - fp->on_hw = FALSE; -out_err: - tgsi_parse_free(&parse); - free(fpc); -} - -void -nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) -{ - struct pipe_winsys *ws = nv30->pipe.winsys; - int i; - - if (!fp->translated) { - nv30_fragprog_translate(nv30, fp); - if (!fp->translated) - assert(0); - } - - if (fp->nr_consts) { - float *map = ws->buffer_map(ws, nv30->fragprog.constant_buf, - PIPE_BUFFER_USAGE_CPU_READ); - for (i = 0; i < fp->nr_consts; i++) { - struct nv30_fragment_program_data *fpd = &fp->consts[i]; - uint32_t *p = &fp->insn[fpd->offset]; - uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; - - if (!memcmp(p, cb, 4 * sizeof(float))) - continue; - memcpy(p, cb, 4 * sizeof(float)); - fp->on_hw = 0; - } - ws->buffer_unmap(ws, nv30->fragprog.constant_buf); - } - - if (!fp->on_hw) { - const uint32_t le = 1; - uint32_t *map; - - if (!fp->buffer) - fp->buffer = ws->buffer_create(ws, 0x100, 0, - fp->insn_len * 4); - map = ws->buffer_map(ws, fp->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); - -#if 0 - for (i = 0; i < fp->insn_len; i++) { - NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); - } -#endif - - if ((*(const uint8_t *)&le)) { - for (i = 0; i < fp->insn_len; i++) { - map[i] = fp->insn[i]; - } - } else { - /* Weird swapping for big-endian chips */ - for (i = 0; i < fp->insn_len; i++) { - map[i] = ((fp->insn[i] & 0xffff) << 16) | - ((fp->insn[i] >> 16) & 0xffff); - } - } - - ws->buffer_unmap(ws, fp->buffer); - fp->on_hw = TRUE; - } - - BEGIN_RING(rankine, NV34TCL_FP_CONTROL, 1); - OUT_RING (fp->fp_control); - BEGIN_RING(rankine, NV34TCL_FP_REG_CONTROL, 1); - OUT_RING (fp->fp_reg_control); - - nv30->fragprog.active = fp; -} - -void -nv30_fragprog_destroy(struct nv30_context *nv30, - struct nv30_fragment_program *fp) -{ - if (fp->insn_len) - free(fp->insn); -} - diff --git a/src/mesa/pipe/nv30/nv30_fragtex.c b/src/mesa/pipe/nv30/nv30_fragtex.c deleted file mode 100644 index 45ee6db8d6a..00000000000 --- a/src/mesa/pipe/nv30/nv30_fragtex.c +++ /dev/null @@ -1,160 +0,0 @@ -#include "nv30_context.h" - -static INLINE int log2i(int i) -{ - int r = 0; - - if (i & 0xffff0000) { - i >>= 16; - r += 16; - } - if (i & 0x0000ff00) { - i >>= 8; - r += 8; - } - if (i & 0x000000f0) { - i >>= 4; - r += 4; - } - if (i & 0x0000000c) { - i >>= 2; - r += 2; - } - if (i & 0x00000002) { - r += 1; - } - return r; -} - -#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ -{ \ - TRUE, \ - PIPE_FORMAT_##m, \ - NV34TCL_TX_FORMAT_FORMAT_##tf, \ - (NV34TCL_TX_SWIZZLE_S0_X_##ts0x | NV34TCL_TX_SWIZZLE_S0_Y_##ts0y | \ - NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w | \ - NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y | \ - NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w), \ -} - -struct nv30_texture_format { - boolean defined; - uint pipe; - int format; - int swizzle; -}; - -static struct nv30_texture_format -nv30_texture_formats[] = { - _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), - _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), - _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), -// _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), - _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), -// _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), -// _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), -// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), - {}, -}; - -static struct nv30_texture_format * -nv30_fragtex_format(uint pipe_format) -{ - struct nv30_texture_format *tf = nv30_texture_formats; - - while (tf->defined) { - if (tf->pipe == pipe_format) - return tf; - tf++; - } - - return NULL; -} - - -static void -nv30_fragtex_build(struct nv30_context *nv30, int unit) -{ - struct nv30_sampler_state *ps = nv30->tex_sampler[unit]; - struct nv30_miptree *nv30mt = nv30->tex_miptree[unit]; - struct pipe_texture *pt = &nv30mt->base; - struct nv30_texture_format *tf; - uint32_t txf, txs, txp; - int swizzled = 0; /*XXX: implement in region code? */ - - tf = nv30_fragtex_format(pt->format); - if (!tf || !tf->defined) { - NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); - return; - } - - txf = tf->format << 8; - txf |= (pt->last_level + 1) << 16; - txf |= log2i(pt->width[0]) << 20; - txf |= log2i(pt->height[0]) << 24; - txf |= log2i(pt->depth[0]) << 28; - txf |= 8; - - switch (pt->target) { -/* case PIPE_TEXTURE_CUBE: - txf |= NV34TCL_TEX_FORMAT_CUBIC;*/ - /* fall-through */ - case PIPE_TEXTURE_2D: - txf |= (2<<4); - break; - case PIPE_TEXTURE_3D: - txf |= (3<<4); - break; - case PIPE_TEXTURE_1D: - txf |= (1<<4); - break; - default: - NOUVEAU_ERR("Unknown target %d\n", pt->target); - return; - } - - txs = tf->swizzle; - - BEGIN_RING(rankine, NV34TCL_TX_OFFSET(unit), 8); - OUT_RELOCl(nv30mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv30mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); - OUT_RING (ps->wrap); - OUT_RING (0x40000000); /* enable */ - OUT_RING (txs); - OUT_RING (ps->filt | 0x2000 /* magic */); - OUT_RING ((pt->width[0] << 16) | pt->height[0]); - OUT_RING (ps->bcol); -} - -void -nv30_fragtex_bind(struct nv30_context *nv30) -{ - struct nv30_fragment_program *fp = nv30->fragprog.active; - unsigned samplers, unit; - - samplers = nv30->fp_samplers & ~fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - BEGIN_RING(rankine, NV34TCL_TX_ENABLE(unit), 1); - OUT_RING (0); - } - - samplers = nv30->dirty_samplers & fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - nv30_fragtex_build(nv30, unit); - } - - nv30->fp_samplers = fp->samplers; -} - diff --git a/src/mesa/pipe/nv30/nv30_miptree.c b/src/mesa/pipe/nv30/nv30_miptree.c deleted file mode 100644 index 5fb89f4cfdc..00000000000 --- a/src/mesa/pipe/nv30/nv30_miptree.c +++ /dev/null @@ -1,105 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_inlines.h" - -#include "nv30_context.h" - -static void -nv30_miptree_layout(struct nv30_miptree *nv30mt) -{ - struct pipe_texture *pt = &nv30mt->base; - boolean swizzled = FALSE; - uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; - uint offset = 0; - int nr_faces, l, f; - - if (pt->target == PIPE_TEXTURE_CUBE) { - nr_faces = 6; - } else - if (pt->target == PIPE_TEXTURE_3D) { - nr_faces = pt->depth[0]; - } else { - nr_faces = 1; - } - - for (l = 0; l <= pt->last_level; l++) { - pt->width[l] = width; - pt->height[l] = height; - pt->depth[l] = depth; - - if (swizzled) - nv30mt->level[l].pitch = pt->width[l] * pt->cpp; - else - nv30mt->level[l].pitch = pt->width[0] * pt->cpp; - nv30mt->level[l].pitch = (nv30mt->level[l].pitch + 63) & ~63; - - nv30mt->level[l].image_offset = - CALLOC(nr_faces, sizeof(unsigned)); - - width = MAX2(1, width >> 1); - height = MAX2(1, height >> 1); - depth = MAX2(1, depth >> 1); - - } - - for (f = 0; f < nr_faces; f++) { - for (l = 0; l <= pt->last_level; l++) { - nv30mt->level[l].image_offset[f] = offset; - offset += nv30mt->level[l].pitch * pt->height[l]; - } - } - - nv30mt->total_size = offset; -} - -static void -nv30_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) -{ - struct pipe_winsys *ws = pipe->winsys; - struct nv30_miptree *nv30mt; - - nv30mt = realloc(*pt, sizeof(struct nv30_miptree)); - if (!nv30mt) - return; - *pt = NULL; - - nv30_miptree_layout(nv30mt); - - nv30mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - nv30mt->total_size); - if (!nv30mt->buffer) { - free(nv30mt); - return; - } - - *pt = &nv30mt->base; -} - -static void -nv30_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) -{ - struct pipe_winsys *ws = pipe->winsys; - struct pipe_texture *mt = *pt; - - *pt = NULL; - if (--mt->refcount <= 0) { - struct nv30_miptree *nv30mt = (struct nv30_miptree *)mt; - int l; - - pipe_buffer_reference(ws, &nv30mt->buffer, NULL); - for (l = 0; l <= mt->last_level; l++) { - if (nv30mt->level[l].image_offset) - free(nv30mt->level[l].image_offset); - } - free(nv30mt); - } -} - -void -nv30_init_miptree_functions(struct nv30_context *nv30) -{ - nv30->pipe.texture_create = nv30_miptree_create; - nv30->pipe.texture_release = nv30_miptree_release; -} - diff --git a/src/mesa/pipe/nv30/nv30_query.c b/src/mesa/pipe/nv30/nv30_query.c deleted file mode 100644 index 71fdcfa24df..00000000000 --- a/src/mesa/pipe/nv30/nv30_query.c +++ /dev/null @@ -1,113 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_util.h" - -#include "nv30_context.h" - -struct nv30_query { - struct nouveau_resource *object; - unsigned type; - boolean ready; - uint64_t result; -}; - -static inline struct nv30_query * -nv30_query(struct pipe_query *pipe) -{ - return (struct nv30_query *)pipe; -} - -static struct pipe_query * -nv30_query_create(struct pipe_context *pipe, unsigned query_type) -{ - struct nv30_query *q; - - q = CALLOC(1, sizeof(struct nv30_query)); - q->type = query_type; - - return (struct pipe_query *)q; -} - -static void -nv30_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_query *q = nv30_query(pq); - - if (q->object) - nv30->nvws->res_free(&q->object); - free(q); -} - -static void -nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_query *q = nv30_query(pq); - - assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - - if (nv30->nvws->res_alloc(nv30->query_heap, 1, NULL, &q->object)) - assert(0); - nv30->nvws->notifier_reset(nv30->query, q->object->start); - - BEGIN_RING(rankine, NV34TCL_QUERY_RESET, 1); - OUT_RING (1); - BEGIN_RING(rankine, NV34TCL_QUERY_UNK17CC, 1); - OUT_RING (1); - - q->ready = FALSE; -} - -static void -nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_query *q = nv30_query(pq); - - BEGIN_RING(rankine, NV34TCL_QUERY_GET, 1); - OUT_RING ((0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) | - ((q->object->start * 32) << NV34TCL_QUERY_GET_OFFSET_SHIFT)); - FIRE_RING(); -} - -static boolean -nv30_query_result(struct pipe_context *pipe, struct pipe_query *pq, - boolean wait, uint64 *result) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_query *q = nv30_query(pq); - struct nouveau_winsys *nvws = nv30->nvws; - - assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER); - - if (!q->ready) { - unsigned status; - - status = nvws->notifier_status(nv30->query, q->object->start); - if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { - if (wait == FALSE) - return FALSE; - nvws->notifier_wait(nv30->query, q->object->start, - NV_NOTIFY_STATE_STATUS_COMPLETED, - 0); - } - - q->result = nvws->notifier_retval(nv30->query, - q->object->start); - q->ready = TRUE; - nvws->res_free(&q->object); - } - - *result = q->result; - return TRUE; -} - -void -nv30_init_query_functions(struct nv30_context *nv30) -{ - nv30->pipe.create_query = nv30_query_create; - nv30->pipe.destroy_query = nv30_query_destroy; - nv30->pipe.begin_query = nv30_query_begin; - nv30->pipe.end_query = nv30_query_end; - nv30->pipe.get_query_result = nv30_query_result; -} diff --git a/src/mesa/pipe/nv30/nv30_shader.h b/src/mesa/pipe/nv30/nv30_shader.h deleted file mode 100644 index dd3a36f78f3..00000000000 --- a/src/mesa/pipe/nv30/nv30_shader.h +++ /dev/null @@ -1,490 +0,0 @@ -#ifndef __NV30_SHADER_H__ -#define __NV30_SHADER_H__ - -/* Vertex programs instruction set - * - * 128bit opcodes, split into 4 32-bit ones for ease of use. - * - * Non-native instructions - * ABS - MOV + NV40_VP_INST0_DEST_ABS - * POW - EX2 + MUL + LG2 - * SUB - ADD, second source negated - * SWZ - MOV - * XPD - - * - * Register access - * - Only one INPUT can be accessed per-instruction (move extras into TEMPs) - * - Only one CONST can be accessed per-instruction (move extras into TEMPs) - * - * Relative Addressing - * According to the value returned for - * MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB - * - * there are only two address registers available. The destination in the - * ARL instruction is set to TEMP (The temp isn't actually written). - * - * When using vanilla ARB_v_p, the proprietary driver will squish both the - * available ADDRESS regs into the first hardware reg in the X and Y - * components. - * - * To use an address reg as an index into consts, the CONST_SRC is set to - * (const_base + offset) and INDEX_CONST is set. - * - * To access the second address reg use ADDR_REG_SELECT_1. A particular - * component of the address regs is selected with ADDR_SWZ. - * - * Only one address register can be accessed per instruction. - * - * Conditional execution (see NV_vertex_program{2,3} for details) Conditional - * execution of an instruction is enabled by setting COND_TEST_ENABLE, and - * selecting the condition which will allow the test to pass with - * COND_{FL,LT,...}. It is possible to swizzle the values in the condition - * register, which allows for testing against an individual component. - * - * Branching: - * - * The BRA/CAL instructions seem to follow a slightly different opcode - * layout. The destination instruction ID (IADDR) overlaps a source field. - * Instruction ID's seem to be numbered based on the UPLOAD_FROM_ID FIFO - * command, and is incremented automatically on each UPLOAD_INST FIFO - * command. - * - * Conditional branching is achieved by using the condition tests described - * above. There doesn't appear to be dedicated looping instructions, but - * this can be done using a temp reg + conditional branching. - * - * Subroutines may be uploaded before the main program itself, but the first - * executed instruction is determined by the PROGRAM_START_ID FIFO command. - * - */ - -/* DWORD 0 */ - -#define NV30_VP_INST_ADDR_REG_SELECT_1 (1 << 24) -#define NV30_VP_INST_SRC2_ABS (1 << 23) /* guess */ -#define NV30_VP_INST_SRC1_ABS (1 << 22) /* guess */ -#define NV30_VP_INST_SRC0_ABS (1 << 21) /* guess */ -#define NV30_VP_INST_VEC_RESULT (1 << 20) -#define NV30_VP_INST_DEST_TEMP_ID_SHIFT 16 -#define NV30_VP_INST_DEST_TEMP_ID_MASK (0x0F << 16) -#define NV30_VP_INST_COND_UPDATE_ENABLE (1<<15) -#define NV30_VP_INST_VEC_DEST_TEMP_MASK (0xF << 16) -#define NV30_VP_INST_COND_TEST_ENABLE (1<<14) -#define NV30_VP_INST_COND_SHIFT 11 -#define NV30_VP_INST_COND_MASK (0x07 << 11) -# define NV30_VP_INST_COND_FL 0 /* guess */ -# define NV30_VP_INST_COND_LT 1 -# define NV30_VP_INST_COND_EQ 2 -# define NV30_VP_INST_COND_LE 3 -# define NV30_VP_INST_COND_GT 4 -# define NV30_VP_INST_COND_NE 5 -# define NV30_VP_INST_COND_GE 6 -# define NV30_VP_INST_COND_TR 7 /* guess */ -#define NV30_VP_INST_COND_SWZ_X_SHIFT 9 -#define NV30_VP_INST_COND_SWZ_X_MASK (0x03 << 9) -#define NV30_VP_INST_COND_SWZ_Y_SHIFT 7 -#define NV30_VP_INST_COND_SWZ_Y_MASK (0x03 << 7) -#define NV30_VP_INST_COND_SWZ_Z_SHIFT 5 -#define NV30_VP_INST_COND_SWZ_Z_MASK (0x03 << 5) -#define NV30_VP_INST_COND_SWZ_W_SHIFT 3 -#define NV30_VP_INST_COND_SWZ_W_MASK (0x03 << 3) -#define NV30_VP_INST_COND_SWZ_ALL_SHIFT 3 -#define NV30_VP_INST_COND_SWZ_ALL_MASK (0xFF << 3) -#define NV30_VP_INST_ADDR_SWZ_SHIFT 1 -#define NV30_VP_INST_ADDR_SWZ_MASK (0x03 << 1) -#define NV30_VP_INST_SCA_OPCODEH_SHIFT 0 -#define NV30_VP_INST_SCA_OPCODEH_MASK (0x01 << 0) - -/* DWORD 1 */ -#define NV30_VP_INST_SCA_OPCODEL_SHIFT 28 -#define NV30_VP_INST_SCA_OPCODEL_MASK (0x0F << 28) -# define NV30_VP_INST_OP_NOP 0x00 -# define NV30_VP_INST_OP_RCP 0x02 -# define NV30_VP_INST_OP_RCC 0x03 -# define NV30_VP_INST_OP_RSQ 0x04 -# define NV30_VP_INST_OP_EXP 0x05 -# define NV30_VP_INST_OP_LOG 0x06 -# define NV30_VP_INST_OP_LIT 0x07 -# define NV30_VP_INST_OP_BRA 0x09 -# define NV30_VP_INST_OP_CAL 0x0B -# define NV30_VP_INST_OP_RET 0x0C -# define NV30_VP_INST_OP_LG2 0x0D -# define NV30_VP_INST_OP_EX2 0x0E -# define NV30_VP_INST_OP_SIN 0x0F -# define NV30_VP_INST_OP_COS 0x10 -#define NV30_VP_INST_VEC_OPCODE_SHIFT 23 -#define NV30_VP_INST_VEC_OPCODE_MASK (0x1F << 23) -# define NV30_VP_INST_OP_NOPV 0x00 -# define NV30_VP_INST_OP_MOV 0x01 -# define NV30_VP_INST_OP_MUL 0x02 -# define NV30_VP_INST_OP_ADD 0x03 -# define NV30_VP_INST_OP_MAD 0x04 -# define NV30_VP_INST_OP_DP3 0x05 -# define NV30_VP_INST_OP_DP4 0x07 -# define NV30_VP_INST_OP_DPH 0x06 -# define NV30_VP_INST_OP_DST 0x08 -# define NV30_VP_INST_OP_MIN 0x09 -# define NV30_VP_INST_OP_MAX 0x0A -# define NV30_VP_INST_OP_SLT 0x0B -# define NV30_VP_INST_OP_SGE 0x0C -# define NV30_VP_INST_OP_ARL 0x0D -# define NV30_VP_INST_OP_FRC 0x0E -# define NV30_VP_INST_OP_FLR 0x0F -# define NV30_VP_INST_OP_SEQ 0x10 -# define NV30_VP_INST_OP_SFL 0x11 -# define NV30_VP_INST_OP_SGT 0x12 -# define NV30_VP_INST_OP_SLE 0x13 -# define NV30_VP_INST_OP_SNE 0x14 -# define NV30_VP_INST_OP_STR 0x15 -# define NV30_VP_INST_OP_SSG 0x16 -# define NV30_VP_INST_OP_ARR 0x17 -# define NV30_VP_INST_OP_ARA 0x18 -#define NV30_VP_INST_CONST_SRC_SHIFT 14 -#define NV30_VP_INST_CONST_SRC_MASK (0xFF << 14) -#define NV30_VP_INST_INPUT_SRC_SHIFT 9 /*NV20*/ -#define NV30_VP_INST_INPUT_SRC_MASK (0x0F << 9) /*NV20*/ -# define NV30_VP_INST_IN_POS 0 /* These seem to match the bindings specified in */ -# define NV30_VP_INST_IN_WEIGHT 1 /* the ARB_v_p spec (2.14.3.1) */ -# define NV30_VP_INST_IN_NORMAL 2 -# define NV30_VP_INST_IN_COL0 3 /* Should probably confirm them all though */ -# define NV30_VP_INST_IN_COL1 4 -# define NV30_VP_INST_IN_FOGC 5 -# define NV30_VP_INST_IN_TC0 8 -# define NV30_VP_INST_IN_TC(n) (8+n) -#define NV30_VP_INST_SRC0H_SHIFT 0 /*NV20*/ -#define NV30_VP_INST_SRC0H_MASK (0x1FF << 0) /*NV20*/ - -/* Please note: the IADDR fields overlap other fields because they are used - * only for branch instructions. See Branching: label above - * - * DWORD 2 - */ -#define NV30_VP_INST_SRC0L_SHIFT 26 /*NV20*/ -#define NV30_VP_INST_SRC0L_MASK (0x3F <<26) /* NV30_VP_SRC0_LOW_MASK << 26 */ -#define NV30_VP_INST_SRC1_SHIFT 11 /*NV20*/ -#define NV30_VP_INST_SRC1_MASK (0x7FFF<<11) /*NV20*/ -#define NV30_VP_INST_SRC2H_SHIFT 0 /*NV20*/ -#define NV30_VP_INST_SRC2H_MASK (0x7FF << 0) /* NV30_VP_SRC2_HIGH_MASK >> 4*/ -#define NV30_VP_INST_IADDR_SHIFT 2 -#define NV30_VP_INST_IADDR_MASK (0xF << 28) /* NV30_VP_SRC2_LOW_MASK << 28 */ - -/* DWORD 3 */ -#define NV30_VP_INST_SRC2L_SHIFT 28 /*NV20*/ -#define NV30_VP_INST_SRC2L_MASK (0x0F <<28) /*NV20*/ -#define NV30_VP_INST_STEMP_WRITEMASK_SHIFT 24 -#define NV30_VP_INST_STEMP_WRITEMASK_MASK (0x0F << 24) -#define NV30_VP_INST_VTEMP_WRITEMASK_SHIFT 20 -#define NV30_VP_INST_VTEMP_WRITEMASK_MASK (0x0F << 20) -#define NV30_VP_INST_SDEST_WRITEMASK_SHIFT 16 -#define NV30_VP_INST_SDEST_WRITEMASK_MASK (0x0F << 16) -#define NV30_VP_INST_VDEST_WRITEMASK_SHIFT 12 /*NV20*/ -#define NV30_VP_INST_VDEST_WRITEMASK_MASK (0x0F << 12) /*NV20*/ -#define NV30_VP_INST_DEST_SHIFT 2 -#define NV30_VP_INST_DEST_MASK (0x0F << 2) -# define NV30_VP_INST_DEST_POS 0 -# define NV30_VP_INST_DEST_BFC0 1 -# define NV30_VP_INST_DEST_BFC1 2 -# define NV30_VP_INST_DEST_COL0 3 -# define NV30_VP_INST_DEST_COL1 4 -# define NV30_VP_INST_DEST_FOGC 5 -# define NV30_VP_INST_DEST_PSZ 6 -# define NV30_VP_INST_DEST_TC(n) (8+n) - -#define NV30_VP_INST_LAST (1 << 0) - -/* Useful to split the source selection regs into their pieces */ -#define NV30_VP_SRC0_HIGH_SHIFT 6 -#define NV30_VP_SRC0_HIGH_MASK 0x00007FC0 -#define NV30_VP_SRC0_LOW_MASK 0x0000003F -#define NV30_VP_SRC2_HIGH_SHIFT 4 -#define NV30_VP_SRC2_HIGH_MASK 0x00007FF0 -#define NV30_VP_SRC2_LOW_MASK 0x0000000F - - -/* Source-register definition - matches NV20 exactly */ -#define NV30_VP_SRC_NEGATE (1<<14) -#define NV30_VP_SRC_SWZ_X_SHIFT 12 -#define NV30_VP_SRC_REG_SWZ_X_MASK (0x03 <<12) -#define NV30_VP_SRC_SWZ_Y_SHIFT 10 -#define NV30_VP_SRC_REG_SWZ_Y_MASK (0x03 <<10) -#define NV30_VP_SRC_SWZ_Z_SHIFT 8 -#define NV30_VP_SRC_REG_SWZ_Z_MASK (0x03 << 8) -#define NV30_VP_SRC_SWZ_W_SHIFT 6 -#define NV30_VP_SRC_REG_SWZ_W_MASK (0x03 << 6) -#define NV30_VP_SRC_REG_SWZ_ALL_SHIFT 6 -#define NV30_VP_SRC_REG_SWZ_ALL_MASK (0xFF << 6) -#define NV30_VP_SRC_TEMP_SRC_SHIFT 2 -#define NV30_VP_SRC_REG_TEMP_ID_MASK (0x0F << 0) -#define NV30_VP_SRC_REG_TYPE_SHIFT 0 -#define NV30_VP_SRC_REG_TYPE_MASK (0x03 << 0) -#define NV30_VP_SRC_REG_TYPE_TEMP 1 -#define NV30_VP_SRC_REG_TYPE_INPUT 2 -#define NV30_VP_SRC_REG_TYPE_CONST 3 /* guess */ - -/* - * Each fragment program opcode appears to be comprised of 4 32-bit values. - * - * 0 - Opcode, output reg/mask, ATTRIB source - * 1 - Source 0 - * 2 - Source 1 - * 3 - Source 2 - * - * There appears to be no special difference between result regs and temp regs. - * result.color == R0.xyzw - * result.depth == R1.z - * When the fragprog contains instructions to write depth, NV30_TCL_PRIMITIVE_3D_UNK1D78=0 - * otherwise it is set to 1. - * - * Constants are inserted directly after the instruction that uses them. - * - * It appears that it's not possible to use two input registers in one - * instruction as the input sourcing is done in the instruction dword - * and not the source selection dwords. As such instructions such as: - * - * ADD result.color, fragment.color, fragment.texcoord[0]; - * - * must be split into two MOV's and then an ADD (nvidia does this) but - * I'm not sure why it's not just one MOV and then source the second input - * in the ADD instruction.. - * - * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary - * negation requires multiplication with a const. - * - * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO/SWIZZLE_ONE - * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as SWIZZLE_ZERO - * is implemented simply by not writing to the relevant components of the destination. - * - * Conditional execution - * TODO - * - * Non-native instructions: - * LIT - * LRP - MAD+MAD - * SUB - ADD, negate second source - * RSQ - LG2 + EX2 - * POW - LG2 + MUL + EX2 - * SCS - COS + SIN - * XPD - */ - -//== Opcode / Destination selection == -#define NV30_FP_OP_PROGRAM_END (1 << 0) -#define NV30_FP_OP_OUT_REG_SHIFT 1 -#define NV30_FP_OP_OUT_REG_MASK (31 << 1) /* uncertain */ -/* Needs to be set when writing outputs to get expected result.. */ -#define NV30_FP_OP_OUT_REG_HALF (1 << 7) -#define NV30_FP_OP_COND_WRITE_ENABLE (1 << 8) -#define NV30_FP_OP_OUTMASK_SHIFT 9 -#define NV30_FP_OP_OUTMASK_MASK (0xF << 9) -# define NV30_FP_OP_OUT_X (1<<9) -# define NV30_FP_OP_OUT_Y (1<<10) -# define NV30_FP_OP_OUT_Z (1<<11) -# define NV30_FP_OP_OUT_W (1<<12) -/* Uncertain about these, especially the input_src values.. it's possible that - * they can be dynamically changed. - */ -#define NV30_FP_OP_INPUT_SRC_SHIFT 13 -#define NV30_FP_OP_INPUT_SRC_MASK (15 << 13) -# define NV30_FP_OP_INPUT_SRC_POSITION 0x0 -# define NV30_FP_OP_INPUT_SRC_COL0 0x1 -# define NV30_FP_OP_INPUT_SRC_COL1 0x2 -# define NV30_FP_OP_INPUT_SRC_FOGC 0x3 -# define NV30_FP_OP_INPUT_SRC_TC0 0x4 -# define NV30_FP_OP_INPUT_SRC_TC(n) (0x4 + n) -#define NV30_FP_OP_TEX_UNIT_SHIFT 17 -#define NV30_FP_OP_TEX_UNIT_MASK (0xF << 17) /* guess */ -#define NV30_FP_OP_PRECISION_SHIFT 22 -#define NV30_FP_OP_PRECISION_MASK (3 << 22) -# define NV30_FP_PRECISION_FP32 0 -# define NV30_FP_PRECISION_FP16 1 -# define NV30_FP_PRECISION_FX12 2 -#define NV30_FP_OP_OPCODE_SHIFT 24 -#define NV30_FP_OP_OPCODE_MASK (0x3F << 24) -# define NV30_FP_OP_OPCODE_NOP 0x00 -# define NV30_FP_OP_OPCODE_MOV 0x01 -# define NV30_FP_OP_OPCODE_MUL 0x02 -# define NV30_FP_OP_OPCODE_ADD 0x03 -# define NV30_FP_OP_OPCODE_MAD 0x04 -# define NV30_FP_OP_OPCODE_DP3 0x05 -# define NV30_FP_OP_OPCODE_DP4 0x06 -# define NV30_FP_OP_OPCODE_DST 0x07 -# define NV30_FP_OP_OPCODE_MIN 0x08 -# define NV30_FP_OP_OPCODE_MAX 0x09 -# define NV30_FP_OP_OPCODE_SLT 0x0A -# define NV30_FP_OP_OPCODE_SGE 0x0B -# define NV30_FP_OP_OPCODE_SLE 0x0C -# define NV30_FP_OP_OPCODE_SGT 0x0D -# define NV30_FP_OP_OPCODE_SNE 0x0E -# define NV30_FP_OP_OPCODE_SEQ 0x0F -# define NV30_FP_OP_OPCODE_FRC 0x10 -# define NV30_FP_OP_OPCODE_FLR 0x11 -# define NV30_FP_OP_OPCODE_KIL 0x12 -# define NV30_FP_OP_OPCODE_PK4B 0x13 -# define NV30_FP_OP_OPCODE_UP4B 0x14 -# define NV30_FP_OP_OPCODE_DDX 0x15 /* can only write XY */ -# define NV30_FP_OP_OPCODE_DDY 0x16 /* can only write XY */ -# define NV30_FP_OP_OPCODE_TEX 0x17 -# define NV30_FP_OP_OPCODE_TXP 0x18 -# define NV30_FP_OP_OPCODE_TXD 0x19 -# define NV30_FP_OP_OPCODE_RCP 0x1A -# define NV30_FP_OP_OPCODE_RSQ 0x1B -# define NV30_FP_OP_OPCODE_EX2 0x1C -# define NV30_FP_OP_OPCODE_LG2 0x1D -# define NV30_FP_OP_OPCODE_LIT 0x1E -# define NV30_FP_OP_OPCODE_LRP 0x1F -# define NV30_FP_OP_OPCODE_STR 0x20 -# define NV30_FP_OP_OPCODE_SFL 0x21 -# define NV30_FP_OP_OPCODE_COS 0x22 -# define NV30_FP_OP_OPCODE_SIN 0x23 -# define NV30_FP_OP_OPCODE_PK2H 0x24 -# define NV30_FP_OP_OPCODE_UP2H 0x25 -# define NV30_FP_OP_OPCODE_POW 0x26 -# define NV30_FP_OP_OPCODE_PK4UB 0x27 -# define NV30_FP_OP_OPCODE_UP4UB 0x28 -# define NV30_FP_OP_OPCODE_PK2US 0x29 -# define NV30_FP_OP_OPCODE_UP2US 0x2A -# define NV30_FP_OP_OPCODE_DP2A 0x2E -# define NV30_FP_OP_OPCODE_TXB 0x31 -# define NV30_FP_OP_OPCODE_RFL 0x36 -# define NV30_FP_OP_OPCODE_DIV 0x3A -#define NV30_FP_OP_OUT_SAT (1 << 31) - -/* high order bits of SRC0 */ -#define NV30_FP_OP_OUT_ABS (1 << 29) -#define NV30_FP_OP_COND_SWZ_W_SHIFT 27 -#define NV30_FP_OP_COND_SWZ_W_MASK (3 << 27) -#define NV30_FP_OP_COND_SWZ_Z_SHIFT 25 -#define NV30_FP_OP_COND_SWZ_Z_MASK (3 << 25) -#define NV30_FP_OP_COND_SWZ_Y_SHIFT 23 -#define NV30_FP_OP_COND_SWZ_Y_MASK (3 << 23) -#define NV30_FP_OP_COND_SWZ_X_SHIFT 21 -#define NV30_FP_OP_COND_SWZ_X_MASK (3 << 21) -#define NV30_FP_OP_COND_SWZ_ALL_SHIFT 21 -#define NV30_FP_OP_COND_SWZ_ALL_MASK (0xFF << 21) -#define NV30_FP_OP_COND_SHIFT 18 -#define NV30_FP_OP_COND_MASK (0x07 << 18) -# define NV30_FP_OP_COND_FL 0 -# define NV30_FP_OP_COND_LT 1 -# define NV30_FP_OP_COND_EQ 2 -# define NV30_FP_OP_COND_LE 3 -# define NV30_FP_OP_COND_GT 4 -# define NV30_FP_OP_COND_NE 5 -# define NV30_FP_OP_COND_GE 6 -# define NV30_FP_OP_COND_TR 7 - -/* high order bits of SRC1 */ -#define NV30_FP_OP_DST_SCALE_SHIFT 28 -#define NV30_FP_OP_DST_SCALE_MASK (3 << 28) -#define NV30_FP_OP_DST_SCALE_1X 0 -#define NV30_FP_OP_DST_SCALE_2X 1 -#define NV30_FP_OP_DST_SCALE_4X 2 -#define NV30_FP_OP_DST_SCALE_8X 3 -#define NV30_FP_OP_DST_SCALE_INV_2X 5 -#define NV30_FP_OP_DST_SCALE_INV_4X 6 -#define NV30_FP_OP_DST_SCALE_INV_8X 7 - - -/* high order bits of SRC2 */ -#define NV30_FP_OP_INDEX_INPUT (1 << 30) - -//== Register selection == -#define NV30_FP_REG_TYPE_SHIFT 0 -#define NV30_FP_REG_TYPE_MASK (3 << 0) -# define NV30_FP_REG_TYPE_TEMP 0 -# define NV30_FP_REG_TYPE_INPUT 1 -# define NV30_FP_REG_TYPE_CONST 2 -#define NV30_FP_REG_SRC_SHIFT 2 /* uncertain */ -#define NV30_FP_REG_SRC_MASK (31 << 2) -#define NV30_FP_REG_SRC_HALF (1 << 8) -#define NV30_FP_REG_SWZ_ALL_SHIFT 9 -#define NV30_FP_REG_SWZ_ALL_MASK (255 << 9) -#define NV30_FP_REG_SWZ_X_SHIFT 9 -#define NV30_FP_REG_SWZ_X_MASK (3 << 9) -#define NV30_FP_REG_SWZ_Y_SHIFT 11 -#define NV30_FP_REG_SWZ_Y_MASK (3 << 11) -#define NV30_FP_REG_SWZ_Z_SHIFT 13 -#define NV30_FP_REG_SWZ_Z_MASK (3 << 13) -#define NV30_FP_REG_SWZ_W_SHIFT 15 -#define NV30_FP_REG_SWZ_W_MASK (3 << 15) -# define NV30_FP_SWIZZLE_X 0 -# define NV30_FP_SWIZZLE_Y 1 -# define NV30_FP_SWIZZLE_Z 2 -# define NV30_FP_SWIZZLE_W 3 -#define NV30_FP_REG_NEGATE (1 << 17) - -#define NV30SR_NONE 0 -#define NV30SR_OUTPUT 1 -#define NV30SR_INPUT 2 -#define NV30SR_TEMP 3 -#define NV30SR_CONST 4 - -struct nv30_sreg { - int type; - int index; - - int dst_scale; - - int negate; - int abs; - int swz[4]; - - int cc_update; - int cc_update_reg; - int cc_test; - int cc_test_reg; - int cc_swz[4]; -}; - -static INLINE struct nv30_sreg -nv30_sr(int type, int index) -{ - struct nv30_sreg temp = { - .type = type, - .index = index, - .dst_scale = DEF_SCALE, - .abs = 0, - .negate = 0, - .swz = { 0, 1, 2, 3 }, - .cc_update = 0, - .cc_update_reg = 0, - .cc_test = DEF_CTEST, - .cc_test_reg = 0, - .cc_swz = { 0, 1, 2, 3 }, - }; - return temp; -} - -static INLINE struct nv30_sreg -nv30_sr_swz(struct nv30_sreg src, int x, int y, int z, int w) -{ - struct nv30_sreg dst = src; - - dst.swz[SWZ_X] = src.swz[x]; - dst.swz[SWZ_Y] = src.swz[y]; - dst.swz[SWZ_Z] = src.swz[z]; - dst.swz[SWZ_W] = src.swz[w]; - return dst; -} - -static INLINE struct nv30_sreg -nv30_sr_neg(struct nv30_sreg src) -{ - src.negate = !src.negate; - return src; -} - -static INLINE struct nv30_sreg -nv30_sr_abs(struct nv30_sreg src) -{ - src.abs = 1; - return src; -} - -static INLINE struct nv30_sreg -nv30_sr_scale(struct nv30_sreg src, int scale) -{ - src.dst_scale = scale; - return src; -} - -#endif diff --git a/src/mesa/pipe/nv30/nv30_state.c b/src/mesa/pipe/nv30/nv30_state.c deleted file mode 100644 index 53368561e07..00000000000 --- a/src/mesa/pipe/nv30/nv30_state.c +++ /dev/null @@ -1,739 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" - -#include "nv30_context.h" -#include "nv30_state.h" - -static void * -nv30_blend_state_create(struct pipe_context *pipe, - const struct pipe_blend_state *cso) -{ - struct nv30_blend_state *cb; - - cb = malloc(sizeof(struct nv30_blend_state)); - - cb->b_enable = cso->blend_enable ? 1 : 0; - cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | - (nvgl_blend_func(cso->rgb_src_factor))); - cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | - (nvgl_blend_func(cso->rgb_dst_factor))); - cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | - (nvgl_blend_eqn(cso->rgb_func))); - - cb->l_enable = cso->logicop_enable ? 1 : 0; - cb->l_op = nvgl_logicop_func(cso->logicop_func); - - cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | - ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | - ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | - ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); - - cb->d_enable = cso->dither ? 1 : 0; - - return (void *)cb; -} - -static void -nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_blend_state *cb = hwcso; - - BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); - OUT_RING (cb->d_enable); - - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 3); - OUT_RING (cb->b_enable); - OUT_RING (cb->b_srcfunc); - OUT_RING (cb->b_dstfunc); - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_EQUATION, 1); - OUT_RING (cb->b_eqn); - - BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); - OUT_RING (cb->c_mask); - - BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (cb->l_enable); - OUT_RING (cb->l_op); -} - -static void -nv30_blend_state_delete(struct pipe_context *pipe, void *hwcso) -{ - free(hwcso); -} - - -static INLINE unsigned -wrap_mode(unsigned wrap) { - unsigned ret; - - switch (wrap) { - case PIPE_TEX_WRAP_REPEAT: - ret = NV34TCL_TX_WRAP_S_REPEAT; - break; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - ret = NV34TCL_TX_WRAP_S_MIRRORED_REPEAT; - break; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - ret = NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE; - break; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - ret = NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER; - break; - case PIPE_TEX_WRAP_CLAMP: - ret = NV34TCL_TX_WRAP_S_CLAMP; - break; -/* case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_EDGE; - break; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_BORDER; - break; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP; - break;*/ - default: - NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); - ret = NV34TCL_TX_WRAP_S_REPEAT; - break; - } - - return ret >> NV34TCL_TX_WRAP_S_SHIFT; -} - -static void * -nv30_sampler_state_create(struct pipe_context *pipe, - const struct pipe_sampler_state *cso) -{ - struct nv30_sampler_state *ps; - uint32_t filter = 0; - - ps = malloc(sizeof(struct nv30_sampler_state)); - - ps->fmt = 0; - if (!cso->normalized_coords) - ps->fmt |= NV34TCL_TX_FORMAT_RECT; - - ps->wrap = ((wrap_mode(cso->wrap_s) << NV34TCL_TX_WRAP_S_SHIFT) | - (wrap_mode(cso->wrap_t) << NV34TCL_TX_WRAP_T_SHIFT) | - (wrap_mode(cso->wrap_r) << NV34TCL_TX_WRAP_R_SHIFT)); - - ps->en = 0; - if (cso->max_anisotropy >= 2.0) { - /* no idea, binary driver sets it, works without it.. meh.. */ - ps->wrap |= (1 << 5); - -/* if (cso->max_anisotropy >= 16.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_16X; - } else - if (cso->max_anisotropy >= 12.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_12X; - } else - if (cso->max_anisotropy >= 10.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_10X; - } else - if (cso->max_anisotropy >= 8.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_8X; - } else - if (cso->max_anisotropy >= 6.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_6X; - } else - if (cso->max_anisotropy >= 4.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_4X; - } else { - ps->en |= NV34TCL_TX_ENABLE_ANISO_2X; - }*/ - } - - switch (cso->mag_img_filter) { - case PIPE_TEX_FILTER_LINEAR: - filter |= NV34TCL_TX_FILTER_MAGNIFY_LINEAR; - break; - case PIPE_TEX_FILTER_NEAREST: - default: - filter |= NV34TCL_TX_FILTER_MAGNIFY_NEAREST; - break; - } - - switch (cso->min_img_filter) { - case PIPE_TEX_FILTER_LINEAR: - switch (cso->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NEAREST: - filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; - break; - case PIPE_TEX_MIPFILTER_LINEAR: - filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; - break; - case PIPE_TEX_MIPFILTER_NONE: - default: - filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR; - break; - } - break; - case PIPE_TEX_FILTER_NEAREST: - default: - switch (cso->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NEAREST: - filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; - break; - case PIPE_TEX_MIPFILTER_LINEAR: - filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; - break; - case PIPE_TEX_MIPFILTER_NONE: - default: - filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST; - break; - } - break; - } - - ps->filt = filter; - -/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - switch (cso->compare_func) { - case PIPE_FUNC_NEVER: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NEVER; - break; - case PIPE_FUNC_GREATER: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GREATER; - break; - case PIPE_FUNC_EQUAL: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_EQUAL; - break; - case PIPE_FUNC_GEQUAL: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GEQUAL; - break; - case PIPE_FUNC_LESS: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LESS; - break; - case PIPE_FUNC_NOTEQUAL: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NOTEQUAL; - break; - case PIPE_FUNC_LEQUAL: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LEQUAL; - break; - case PIPE_FUNC_ALWAYS: - ps->wrap |= NV34TCL_TX_WRAP_RCOMP_ALWAYS; - break; - default: - break; - } - }*/ - - ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | - (float_to_ubyte(cso->border_color[0]) << 16) | - (float_to_ubyte(cso->border_color[1]) << 8) | - (float_to_ubyte(cso->border_color[2]) << 0)); - - return (void *)ps; -} - -static void -nv30_sampler_state_bind(struct pipe_context *pipe, unsigned unit, - void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_sampler_state *ps = hwcso; - - nv30->tex_sampler[unit] = ps; - nv30->dirty_samplers |= (1 << unit); -} - -static void -nv30_sampler_state_delete(struct pipe_context *pipe, void *hwcso) -{ - free(hwcso); -} - -static void -nv30_set_sampler_texture(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *miptree) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - nv30->tex_miptree[unit] = (struct nv30_miptree *)miptree; - nv30->dirty_samplers |= (1 << unit); -} - -static void * -nv30_rasterizer_state_create(struct pipe_context *pipe, - const struct pipe_rasterizer_state *cso) -{ - struct nv30_rasterizer_state *rs; - int i; - - /*XXX: ignored: - * light_twoside - * offset_cw/ccw -nohw - * scissor - * point_smooth -nohw - * multisample - * offset_units / offset_scale - */ - rs = malloc(sizeof(struct nv30_rasterizer_state)); - - rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; - - rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; - rs->line_smooth_en = cso->line_smooth ? 1 : 0; - rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; - rs->line_stipple = (cso->line_stipple_pattern << 16) | - cso->line_stipple_factor; - - rs->point_size = *(uint32_t*)&cso->point_size; - - rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; - rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; - - if (cso->front_winding == PIPE_WINDING_CCW) { - rs->front_face = NV34TCL_FRONT_FACE_CCW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); - } else { - rs->front_face = NV34TCL_FRONT_FACE_CW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); - } - - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CCW) - rs->cull_face = NV34TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV34TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_CW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CW) - rs->cull_face = NV34TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV34TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_BOTH: - rs->cull_face_en = 1; - rs->cull_face = NV34TCL_CULL_FACE_FRONT_AND_BACK; - break; - case PIPE_WINDING_NONE: - default: - rs->cull_face_en = 0; - rs->cull_face = 0; - break; - } - - if (cso->point_sprite) { - rs->point_sprite = (1 << 0); - for (i = 0; i < 8; i++) { - if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) - rs->point_sprite |= (1 << (8 + i)); - } - } else { - rs->point_sprite = 0; - } - - return (void *)rs; -} - -static void -nv30_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_rasterizer_state *rs = hwcso; - - BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); - OUT_RING (rs->shade_model); - - BEGIN_RING(rankine, NV34TCL_LINE_WIDTH, 2); - OUT_RING (rs->line_width); - OUT_RING (rs->line_smooth_en); - BEGIN_RING(rankine, NV34TCL_LINE_STIPPLE_ENABLE, 2); - OUT_RING (rs->line_stipple_en); - OUT_RING (rs->line_stipple); - - BEGIN_RING(rankine, NV34TCL_POINT_SIZE, 1); - OUT_RING (rs->point_size); - - BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 6); - OUT_RING (rs->poly_mode_front); - OUT_RING (rs->poly_mode_back); - OUT_RING (rs->cull_face); - OUT_RING (rs->front_face); - OUT_RING (rs->poly_smooth_en); - OUT_RING (rs->cull_face_en); - - BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); - OUT_RING (rs->poly_stipple_en); - - BEGIN_RING(rankine, NV34TCL_POINT_SPRITE, 1); - OUT_RING (rs->point_sprite); -} - -static void -nv30_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) -{ - free(hwcso); -} - -static void -nv30_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, - unsigned idx, struct nv30_stencil_push *hw) -{ - hw->enable = cso->stencil[idx].enabled ? 1 : 0; - hw->wmask = cso->stencil[idx].write_mask; - hw->func = nvgl_comparison_op(cso->stencil[idx].func); - hw->ref = cso->stencil[idx].ref_value; - hw->vmask = cso->stencil[idx].value_mask; - hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); - hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); - hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); -} - -static void * -nv30_depth_stencil_alpha_state_create(struct pipe_context *pipe, - const struct pipe_depth_stencil_alpha_state *cso) -{ - struct nv30_depth_stencil_alpha_state *hw; - - hw = malloc(sizeof(struct nv30_depth_stencil_alpha_state)); - - hw->depth.func = nvgl_comparison_op(cso->depth.func); - hw->depth.write_enable = cso->depth.writemask ? 1 : 0; - hw->depth.test_enable = cso->depth.enabled ? 1 : 0; - - nv30_translate_stencil(cso, 0, &hw->stencil.front); - nv30_translate_stencil(cso, 1, &hw->stencil.back); - - hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; - hw->alpha.func = nvgl_comparison_op(cso->alpha.func); - hw->alpha.ref = float_to_ubyte(cso->alpha.ref); - - return (void *)hw; -} - -static void -nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_depth_stencil_alpha_state *hw = hwcso; - - BEGIN_RING(rankine, NV34TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&hw->depth, 3); - BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 16); - OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); - OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); - BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3); - OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); -} - -static void -nv30_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) -{ - free(hwcso); -} - -static void * -nv30_vp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - struct nv30_vertex_program *vp; - - vp = CALLOC(1, sizeof(struct nv30_vertex_program)); - vp->pipe = cso; - - return (void *)vp; -} - -static void -nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_vertex_program *vp = hwcso; - - nv30->vertprog.current = vp; - nv30->dirty |= NV30_NEW_VERTPROG; -} - -static void -nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_vertex_program *vp = hwcso; - - nv30_vertprog_destroy(nv30, vp); - free(vp); -} - -static void * -nv30_fp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - struct nv30_fragment_program *fp; - - fp = CALLOC(1, sizeof(struct nv30_fragment_program)); - fp->pipe = cso; - - return (void *)fp; -} - -static void -nv30_fp_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_fragment_program *fp = hwcso; - - nv30->fragprog.current = fp; - nv30->dirty |= NV30_NEW_FRAGPROG; -} - -static void -nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_fragment_program *fp = hwcso; - - nv30_fragprog_destroy(nv30, fp); - free(fp); -} - -static void -nv30_set_blend_color(struct pipe_context *pipe, - const struct pipe_blend_color *bcol) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_COLOR, 1); - OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0)); -} - -static void -nv30_set_clip_state(struct pipe_context *pipe, - const struct pipe_clip_state *clip) -{ -} - -static void -nv30_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, - const struct pipe_constant_buffer *buf ) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - if (shader == PIPE_SHADER_VERTEX) { - nv30->vertprog.constant_buf = buf->buffer; - nv30->dirty |= NV30_NEW_VERTPROG; - } else - if (shader == PIPE_SHADER_FRAGMENT) { - nv30->fragprog.constant_buf = buf->buffer; - nv30->dirty |= NV30_NEW_FRAGPROG; - } -} - -static void -nv30_set_framebuffer_state(struct pipe_context *pipe, - const struct pipe_framebuffer_state *fb) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct pipe_surface *rt[4], *zeta; - uint32_t rt_enable, rt_format, w, h; - int i, colour_format = 0, zeta_format = 0; - - rt_enable = 0; - for (i = 0; i < 4; i++) { - if (!fb->cbufs[i]) - continue; - - if (colour_format) { - assert(w == fb->cbufs[i]->width); - assert(h == fb->cbufs[i]->height); - assert(colour_format == fb->cbufs[i]->format); - } else { - w = fb->cbufs[i]->width; - h = fb->cbufs[i]->height; - colour_format = fb->cbufs[i]->format; - rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i); - rt[i] = fb->cbufs[i]; - } - } - - if (rt_enable & (NV34TCL_RT_ENABLE_COLOR1 | NV34TCL_RT_ENABLE_COLOR2 | - NV34TCL_RT_ENABLE_COLOR3)) - rt_enable |= NV34TCL_RT_ENABLE_MRT; - - if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - - zeta_format = fb->zsbuf->format; - zeta = fb->zsbuf; - } - - rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR; - - switch (colour_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case 0: - rt_format |= NV34TCL_RT_FORMAT_COLOR_A8R8G8B8; - break; - case PIPE_FORMAT_R5G6B5_UNORM: - rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5; - break; - default: - assert(0); - } - - switch (zeta_format) { - case PIPE_FORMAT_Z16_UNORM: - rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16; - break; - case PIPE_FORMAT_Z24S8_UNORM: - case 0: - rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8; - break; - default: - assert(0); - } - - if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); - OUT_RING ( (rt[0]->pitch * rt[0]->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); - nv30->rt[0] = rt[0]->buffer; - } - - if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 2); - OUT_RING (rt[1]->pitch * rt[1]->cpp); - nv30->rt[1] = rt[1]->buffer; - } - - if (zeta_format) - { - nv30->zeta = zeta->buffer; - } - - nv30->rt_enable = rt_enable; - BEGIN_RING(rankine, NV34TCL_RT_ENABLE, 1); - OUT_RING (rt_enable); - BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 3); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - OUT_RING (rt_format); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0); - OUT_RING (((h - 1) << 16) | 0); -} - -static void -nv30_set_polygon_stipple(struct pipe_context *pipe, - const struct pipe_poly_stipple *stipple) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32); - OUT_RINGp ((uint32_t *)stipple->stipple, 32); -} - -static void -nv30_set_scissor_state(struct pipe_context *pipe, - const struct pipe_scissor_state *s) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); - OUT_RING (((s->maxx - s->minx) << 16) | s->minx); - OUT_RING (((s->maxy - s->miny) << 16) | s->miny); -} - -static void -nv30_set_viewport_state(struct pipe_context *pipe, - const struct pipe_viewport_state *vpt) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); - OUT_RINGf (vpt->translate[0]); - OUT_RINGf (vpt->translate[1]); - OUT_RINGf (vpt->translate[2]); - OUT_RINGf (vpt->translate[3]); - OUT_RINGf (vpt->scale[0]); - OUT_RINGf (vpt->scale[1]); - OUT_RINGf (vpt->scale[2]); - OUT_RINGf (vpt->scale[3]); -} - -static void -nv30_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - nv30->vtxbuf[index] = *vb; - - nv30->dirty |= NV30_NEW_ARRAYS; -} - -static void -nv30_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) -{ - struct nv30_context *nv30 = nv30_context(pipe); - - nv30->vtxelt[index] = *ve; - - nv30->dirty |= NV30_NEW_ARRAYS; -} - -void -nv30_init_state_functions(struct nv30_context *nv30) -{ - nv30->pipe.create_blend_state = nv30_blend_state_create; - nv30->pipe.bind_blend_state = nv30_blend_state_bind; - nv30->pipe.delete_blend_state = nv30_blend_state_delete; - - nv30->pipe.create_sampler_state = nv30_sampler_state_create; - nv30->pipe.bind_sampler_state = nv30_sampler_state_bind; - nv30->pipe.delete_sampler_state = nv30_sampler_state_delete; - nv30->pipe.set_sampler_texture = nv30_set_sampler_texture; - - nv30->pipe.create_rasterizer_state = nv30_rasterizer_state_create; - nv30->pipe.bind_rasterizer_state = nv30_rasterizer_state_bind; - nv30->pipe.delete_rasterizer_state = nv30_rasterizer_state_delete; - - nv30->pipe.create_depth_stencil_alpha_state = - nv30_depth_stencil_alpha_state_create; - nv30->pipe.bind_depth_stencil_alpha_state = - nv30_depth_stencil_alpha_state_bind; - nv30->pipe.delete_depth_stencil_alpha_state = - nv30_depth_stencil_alpha_state_delete; - - nv30->pipe.create_vs_state = nv30_vp_state_create; - nv30->pipe.bind_vs_state = nv30_vp_state_bind; - nv30->pipe.delete_vs_state = nv30_vp_state_delete; - - nv30->pipe.create_fs_state = nv30_fp_state_create; - nv30->pipe.bind_fs_state = nv30_fp_state_bind; - nv30->pipe.delete_fs_state = nv30_fp_state_delete; - - nv30->pipe.set_blend_color = nv30_set_blend_color; - nv30->pipe.set_clip_state = nv30_set_clip_state; - nv30->pipe.set_constant_buffer = nv30_set_constant_buffer; - nv30->pipe.set_framebuffer_state = nv30_set_framebuffer_state; - nv30->pipe.set_polygon_stipple = nv30_set_polygon_stipple; - nv30->pipe.set_scissor_state = nv30_set_scissor_state; - nv30->pipe.set_viewport_state = nv30_set_viewport_state; - - nv30->pipe.set_vertex_buffer = nv30_set_vertex_buffer; - nv30->pipe.set_vertex_element = nv30_set_vertex_element; -} - diff --git a/src/mesa/pipe/nv30/nv30_state.h b/src/mesa/pipe/nv30/nv30_state.h deleted file mode 100644 index 233600f69ab..00000000000 --- a/src/mesa/pipe/nv30/nv30_state.h +++ /dev/null @@ -1,147 +0,0 @@ -#ifndef __NV30_STATE_H__ -#define __NV30_STATE_H__ - -#include "pipe/p_state.h" - -struct nv30_blend_state { - uint32_t b_enable; - uint32_t b_srcfunc; - uint32_t b_dstfunc; - uint32_t b_eqn; - - uint32_t l_enable; - uint32_t l_op; - - uint32_t c_mask; - - uint32_t d_enable; -}; - -struct nv30_sampler_state { - uint32_t fmt; - uint32_t wrap; - uint32_t en; - uint32_t filt; - uint32_t bcol; -}; - -struct nv30_rasterizer_state { - uint32_t shade_model; - - uint32_t line_width; - uint32_t line_smooth_en; - uint32_t line_stipple_en; - uint32_t line_stipple; - - uint32_t point_size; - - uint32_t poly_smooth_en; - uint32_t poly_stipple_en; - - uint32_t poly_mode_front; - uint32_t poly_mode_back; - - uint32_t front_face; - uint32_t cull_face; - uint32_t cull_face_en; - - uint32_t point_sprite; -}; - -struct nv30_vertex_program_exec { - uint32_t data[4]; - boolean has_branch_offset; - int const_index; -}; - -struct nv30_vertex_program_data { - int index; /* immediates == -1 */ - float value[4]; -}; - -struct nv30_vertex_program { - const struct pipe_shader_state *pipe; - - boolean translated; - struct nv30_vertex_program_exec *insns; - unsigned nr_insns; - struct nv30_vertex_program_data *consts; - unsigned nr_consts; - - struct nouveau_resource *exec; - unsigned exec_start; - struct nouveau_resource *data; - unsigned data_start; - unsigned data_start_min; - - uint32_t ir; - uint32_t or; -}; - -struct nv30_fragment_program_data { - unsigned offset; - unsigned index; -}; - -struct nv30_fragment_program { - const struct pipe_shader_state *pipe; - - boolean translated; - boolean on_hw; - unsigned samplers; - - uint32_t *insn; - int insn_len; - - struct nv30_fragment_program_data *consts; - unsigned nr_consts; - - struct pipe_buffer *buffer; - - uint32_t fp_control; - uint32_t fp_reg_control; -}; - -struct nv30_stencil_push { - uint32_t enable; - uint32_t wmask; - uint32_t func; - uint32_t ref; - uint32_t vmask; - uint32_t fail; - uint32_t zfail; - uint32_t zpass; -}; - -struct nv30_depth_stencil_alpha_state { - struct { - uint32_t func; - uint32_t write_enable; - uint32_t test_enable; - } depth; - - struct { - struct nv30_stencil_push back; - struct nv30_stencil_push front; - } stencil; - - struct { - uint32_t enabled; - uint32_t func; - uint32_t ref; - } alpha; -}; - -struct nv30_miptree { - struct pipe_texture base; - - struct pipe_buffer *buffer; - uint total_size; - - struct { - uint pitch; - uint *image_offset; - } level[PIPE_MAX_TEXTURE_LEVELS]; -}; - -#endif diff --git a/src/mesa/pipe/nv30/nv30_state_emit.c b/src/mesa/pipe/nv30/nv30_state_emit.c deleted file mode 100644 index 70b98836f0f..00000000000 --- a/src/mesa/pipe/nv30/nv30_state_emit.c +++ /dev/null @@ -1,83 +0,0 @@ -#include "nv30_context.h" -#include "nv30_state.h" - -void -nv30_emit_hw_state(struct nv30_context *nv30) -{ - int i; - - if (nv30->dirty & NV30_NEW_FRAGPROG) { - nv30_fragprog_bind(nv30, nv30->fragprog.current); - /*XXX: clear NV30_NEW_FRAGPROG if no new program uploaded */ - } - - if (nv30->dirty_samplers || (nv30->dirty & NV30_NEW_FRAGPROG)) { - nv30_fragtex_bind(nv30); -/* - BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); - OUT_RING (1);*/ - nv30->dirty &= ~NV30_NEW_FRAGPROG; - } - - if (nv30->dirty & NV30_NEW_VERTPROG) { - nv30_vertprog_bind(nv30, nv30->vertprog.current); - nv30->dirty &= ~NV30_NEW_VERTPROG; - } - - nv30->dirty_samplers = 0; - - /* Emit relocs for every referenced buffer. - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. This isn't very efficient, but we don't - * seem to take a significant performance hit. Will be improved - * at some point. Vertex arrays are emitted by nv30_vbo.c - */ - - /* Render targets */ - if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv30->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(rankine, NV34TCL_COLOR0_OFFSET, 1); - OUT_RELOCl(nv30->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - } - - if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv30->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(rankine, NV34TCL_COLOR1_OFFSET, 1); - OUT_RELOCl(nv30->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - } - - if (nv30->zeta) { - BEGIN_RING(rankine, NV34TCL_DMA_ZETA, 1); - OUT_RELOCo(nv30->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(rankine, NV34TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv30->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - /* XXX allocate LMA */ -/* BEGIN_RING(rankine, NV34TCL_LMA_DEPTH_OFFSET, 1); - OUT_RING(0);*/ - } - - /* Texture images */ - for (i = 0; i < 16; i++) { - if (!(nv30->fp_samplers & (1 << i))) - continue; - BEGIN_RING(rankine, NV34TCL_TX_OFFSET(i), 2); - OUT_RELOCl(nv30->tex[i].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv30->tex[i].buffer, nv30->tex[i].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV34TCL_TX_FORMAT_DMA0, - NV34TCL_TX_FORMAT_DMA1); - } - - /* Fragment program */ - BEGIN_RING(rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); - OUT_RELOC (nv30->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR, NV34TCL_FP_ACTIVE_PROGRAM_DMA0, - NV34TCL_FP_ACTIVE_PROGRAM_DMA1); -} - diff --git a/src/mesa/pipe/nv30/nv30_surface.c b/src/mesa/pipe/nv30/nv30_surface.c deleted file mode 100644 index 31745e3d6e8..00000000000 --- a/src/mesa/pipe/nv30/nv30_surface.c +++ /dev/null @@ -1,136 +0,0 @@ - -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "nv30_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_winsys.h" -#include "pipe/p_inlines.h" -#include "pipe/util/p_tile.h" - -static boolean -nv30_surface_format_supported(struct pipe_context *pipe, - enum pipe_format format, uint type) -{ - switch (type) { - case PIPE_SURFACE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return TRUE; - default: - break; - } - break; - case PIPE_TEXTURE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A1R5G5B5_UNORM: - case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: - case PIPE_FORMAT_Z16_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - } - break; - default: - assert(0); - }; - - return FALSE; -} - -static struct pipe_surface * -nv30_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) -{ - struct pipe_winsys *ws = pipe->winsys; - struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; - struct pipe_surface *ps; - - ps = ws->surface_alloc(ws); - if (!ps) - return NULL; - pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); - ps->format = pt->format; - ps->cpp = pt->cpp; - ps->width = pt->width[level]; - ps->height = pt->height[level]; - ps->pitch = nv30mt->level[level].pitch / ps->cpp; - - if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv30mt->level[level].image_offset[face]; - } else - if (pt->target == PIPE_TEXTURE_3D) { - ps->offset = nv30mt->level[level].image_offset[zslice]; - } else { - ps->offset = nv30mt->level[level].image_offset[0]; - } - - return ps; -} - -static void -nv30_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, struct pipe_surface *src, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_winsys *nvws = nv30->nvws; - - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); -} - -static void -nv30_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, unsigned width, - unsigned height, unsigned value) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_winsys *nvws = nv30->nvws; - - nvws->surface_fill(nvws, dest, destx, desty, width, height, value); -} - -void -nv30_init_surface_functions(struct nv30_context *nv30) -{ - nv30->pipe.is_format_supported = nv30_surface_format_supported; - nv30->pipe.get_tex_surface = nv30_get_tex_surface; - nv30->pipe.surface_copy = nv30_surface_copy; - nv30->pipe.surface_fill = nv30_surface_fill; -} diff --git a/src/mesa/pipe/nv30/nv30_vbo.c b/src/mesa/pipe/nv30/nv30_vbo.c deleted file mode 100644 index 57fb9bc8a57..00000000000 --- a/src/mesa/pipe/nv30/nv30_vbo.c +++ /dev/null @@ -1,425 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_state.h" -#include "pipe/p_util.h" - -#include "nv30_context.h" -#include "nv30_state.h" - -#include "pipe/nouveau/nouveau_channel.h" -#include "pipe/nouveau/nouveau_pushbuf.h" - -static INLINE int -nv30_vbo_ncomp(uint format) -{ - int ncomp = 0; - - if (pf_size_x(format)) ncomp++; - if (pf_size_y(format)) ncomp++; - if (pf_size_z(format)) ncomp++; - if (pf_size_w(format)) ncomp++; - - return ncomp; -} - -static INLINE int -nv30_vbo_type(uint format) -{ - switch (pf_type(format)) { - case PIPE_FORMAT_TYPE_FLOAT: - return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; - case PIPE_FORMAT_TYPE_UNORM: - return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE; - default: - NOUVEAU_ERR("Unknown format 0x%08x\n", format); - return NV40TCL_VTXFMT_TYPE_FLOAT; - } -} - -static boolean -nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, - struct pipe_vertex_element *ve, - struct pipe_vertex_buffer *vb) -{ - struct pipe_winsys *ws = nv30->pipe.winsys; - int type, ncomp; - void *map; - - type = nv30_vbo_type(ve->src_format); - ncomp = nv30_vbo_ncomp(ve->src_format); - - map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); - map += vb->buffer_offset + ve->src_offset; - - switch (type) { - case NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT: - { - float *v = map; - - BEGIN_RING(rankine, NV34TCL_VERTEX_ATTR_4F_X(attrib), 4); - switch (ncomp) { - case 4: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(v[3]); - break; - case 3: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(1.0); - break; - case 2: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(0.0); - OUT_RINGf(1.0); - break; - case 1: - OUT_RINGf(v[0]); - OUT_RINGf(0.0); - OUT_RINGf(0.0); - OUT_RINGf(1.0); - break; - default: - ws->buffer_unmap(ws, vb->buffer); - return FALSE; - } - } - break; - default: - ws->buffer_unmap(ws, vb->buffer); - return FALSE; - } - - ws->buffer_unmap(ws, vb->buffer); - - return TRUE; -} - -static void -nv30_vbo_arrays_update(struct nv30_context *nv30) -{ - struct nv30_vertex_program *vp = nv30->vertprog.active; - uint32_t inputs, vtxfmt[16]; - int hw, num_hw; - - nv30->vb_enable = 0; - - inputs = vp->ir; - for (hw = 0; hw < 16 && inputs; hw++) { - if (inputs & (1 << hw)) { - num_hw = hw; - inputs &= ~(1 << hw); - } - } - num_hw++; - - inputs = vp->ir; - for (hw = 0; hw < num_hw; hw++) { - struct pipe_vertex_element *ve; - struct pipe_vertex_buffer *vb; - - if (!(inputs & (1 << hw))) { - vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; - continue; - } - - ve = &nv30->vtxelt[hw]; - vb = &nv30->vtxbuf[ve->vertex_buffer_index]; - - if (vb->pitch == 0) { - vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; - if (nv30_vbo_static_attrib(nv30, hw, ve, vb) == TRUE) - continue; - } - - nv30->vb_enable |= (1 << hw); - nv30->vb[hw].delta = vb->buffer_offset + ve->src_offset; - nv30->vb[hw].buffer = vb->buffer; - - vtxfmt[hw] = ((vb->pitch << NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT) | - (nv30_vbo_ncomp(ve->src_format) << - NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT) | - nv30_vbo_type(ve->src_format)); - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_ARRAY_FORMAT(0), num_hw); - OUT_RINGp (vtxfmt, num_hw); -} - -static boolean -nv30_vbo_validate_state(struct nv30_context *nv30, - struct pipe_buffer *ib, unsigned ib_format) -{ - unsigned inputs; - - nv30_emit_hw_state(nv30); - - if (nv30->dirty & NV30_NEW_ARRAYS) { - nv30_vbo_arrays_update(nv30); - nv30->dirty &= ~NV30_NEW_ARRAYS; - } - - inputs = nv30->vb_enable; - while (inputs) { - unsigned a = ffs(inputs) - 1; - - inputs &= ~(1 << a); - - BEGIN_RING(rankine, NV34TCL_VERTEX_BUFFER_ADDRESS(a), 1); - OUT_RELOC (nv30->vb[a].buffer, nv30->vb[a].delta, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1); - } - - if (ib) { - BEGIN_RING(rankine, NV40TCL_IDXBUF_ADDRESS, 2); - OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD); - OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_OR, - 0, NV40TCL_IDXBUF_FORMAT_DMA1); - } - - BEGIN_RING(rankine, 0x1710, 1); - OUT_RING (0); /* vtx cache flush */ - - return TRUE; -} - -boolean -nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, - unsigned count) -{ - struct nv30_context *nv30 = nv30_context(pipe); - unsigned nr; - boolean ret; - - ret = nv30_vbo_validate_state(nv30, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - nr = (count & 0xff); - if (nr) { - BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } - - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; - - nr -= push; - - BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; - } - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); - - pipe->flush(pipe, 0); - return TRUE; -} - -static INLINE void -nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, - unsigned start, unsigned count) -{ - uint8_t *elts = (uint8_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } - - while (count) { - push = MIN2(count, 2047 * 2); - - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); - - count -= push; - elts += push; - } -} - -static INLINE void -nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, - unsigned start, unsigned count) -{ - uint16_t *elts = (uint16_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } - - while (count) { - push = MIN2(count, 2047 * 2); - - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); - - count -= push; - elts += push; - } -} - -static INLINE void -nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, - unsigned start, unsigned count) -{ - uint32_t *elts = (uint32_t *)ib + start; - int push; - - while (count) { - push = MIN2(count, 2047); - - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push); - OUT_RINGp (elts, push); - - count -= push; - elts += push; - } -} - -static boolean -nv30_draw_elements_inline(struct pipe_context *pipe, - struct pipe_buffer *ib, unsigned ib_size, - unsigned mode, unsigned start, unsigned count) -{ - struct nv30_context *nv30 = nv30_context(pipe); - struct pipe_winsys *ws = pipe->winsys; - boolean ret; - void *map; - - ret = nv30_vbo_validate_state(nv30, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } - - map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); - if (!ib) { - NOUVEAU_ERR("failed mapping ib\n"); - return FALSE; - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - switch (ib_size) { - case 1: - nv30_draw_elements_u08(nv30, map, start, count); - break; - case 2: - nv30_draw_elements_u16(nv30, map, start, count); - break; - case 4: - nv30_draw_elements_u32(nv30, map, start, count); - break; - default: - NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); - break; - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); - - ws->buffer_unmap(ws, ib); - - return TRUE; -} - -static boolean -nv30_draw_elements_vbo(struct pipe_context *pipe, - struct pipe_buffer *ib, unsigned ib_size, - unsigned mode, unsigned start, unsigned count) -{ - struct nv30_context *nv30 = nv30_context(pipe); - unsigned nr, type; - boolean ret; - - switch (ib_size) { - case 2: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; - break; - case 4: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; - break; - default: - NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); - return FALSE; - } - - ret = nv30_vbo_validate_state(nv30, ib, type); - if (!ret) { - NOUVEAU_ERR("failed state validation\n"); - return FALSE; - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - nr = (count & 0xff); - if (nr) { - BEGIN_RING(rankine, NV40TCL_VB_INDEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } - - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; - - nr -= push; - - BEGIN_RING_NI(rankine, NV40TCL_VB_INDEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; - } - } - - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); - - return TRUE; -} - -boolean -nv30_draw_elements(struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, unsigned indexSize, - unsigned mode, unsigned start, unsigned count) -{ -/* if (indexSize != 1) { - nv30_draw_elements_vbo(pipe, indexBuffer, indexSize, - mode, start, count); - } else */{ - nv30_draw_elements_inline(pipe, indexBuffer, indexSize, - mode, start, count); - } - - pipe->flush(pipe, 0); - return TRUE; -} - - diff --git a/src/mesa/pipe/nv30/nv30_vertprog.c b/src/mesa/pipe/nv30/nv30_vertprog.c deleted file mode 100644 index c96210d3fa7..00000000000 --- a/src/mesa/pipe/nv30/nv30_vertprog.c +++ /dev/null @@ -1,777 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" -#include "pipe/p_util.h" - -#include "pipe/p_shader_tokens.h" -#include "pipe/tgsi/util/tgsi_parse.h" - -#include "nv30_context.h" -#include "nv30_state.h" - -/* TODO (at least...): - * 1. Indexed consts + ARL - * 2. Arb. swz/negation - * 3. NV_vp11, NV_vp2, NV_vp3 features - * - extra arith opcodes - * - branching - * - texture sampling - * - indexed attribs - * - indexed results - * 4. bugs - */ - -#define SWZ_X 0 -#define SWZ_Y 1 -#define SWZ_Z 2 -#define SWZ_W 3 -#define MASK_X 8 -#define MASK_Y 4 -#define MASK_Z 2 -#define MASK_W 1 -#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) -#define DEF_SCALE 0 -#define DEF_CTEST 0 -#include "nv30_shader.h" - -#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) -#define neg(s) nv30_sr_neg((s)) -#define abs(s) nv30_sr_abs((s)) - -struct nv30_vpc { - struct nv30_vertex_program *vp; - - struct nv30_vertex_program_exec *vpi; - - unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; - - int high_temp; - int temp_temp_count; - - struct nv30_sreg *imm; - unsigned nr_imm; -}; - -static struct nv30_sreg -temp(struct nv30_vpc *vpc) -{ - int idx; - - idx = vpc->temp_temp_count++; - idx += vpc->high_temp + 1; - return nv30_sr(NV30SR_TEMP, idx); -} - -static struct nv30_sreg -constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) -{ - struct nv30_vertex_program *vp = vpc->vp; - struct nv30_vertex_program_data *vpd; - int idx; - - if (pipe >= 0) { - for (idx = 0; idx < vp->nr_consts; idx++) { - if (vp->consts[idx].index == pipe) - return nv30_sr(NV30SR_CONST, idx); - } - } - - idx = vp->nr_consts++; - vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); - vpd = &vp->consts[idx]; - - vpd->index = pipe; - vpd->value[0] = x; - vpd->value[1] = y; - vpd->value[2] = z; - vpd->value[3] = w; - return nv30_sr(NV30SR_CONST, idx); -} - -#define arith(cc,s,o,d,m,s0,s1,s2) \ - nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) - -static void -emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) -{ - struct nv30_vertex_program *vp = vpc->vp; - uint32_t sr = 0; - - switch (src.type) { - case NV30SR_TEMP: - sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT); - sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT); - break; - case NV30SR_INPUT: - sr |= (NV30_VP_SRC_REG_TYPE_INPUT << - NV30_VP_SRC_REG_TYPE_SHIFT); - vp->ir |= (1 << src.index); - hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT); - break; - case NV30SR_CONST: - sr |= (NV30_VP_SRC_REG_TYPE_CONST << - NV30_VP_SRC_REG_TYPE_SHIFT); - assert(vpc->vpi->const_index == -1 || - vpc->vpi->const_index == src.index); - vpc->vpi->const_index = src.index; - break; - case NV30SR_NONE: - sr |= (NV30_VP_SRC_REG_TYPE_INPUT << - NV30_VP_SRC_REG_TYPE_SHIFT); - break; - default: - assert(0); - } - - if (src.negate) - sr |= NV30_VP_SRC_NEGATE; - - if (src.abs) - hw[0] |= (1 << (21 + pos)); - - sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) | - (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) | - (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) | - (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT)); - -/* - * |VVV| - * d°.°b - * \u/ - * - */ - - switch (pos) { - case 0: - hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >> - NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT; - hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) << - NV30_VP_INST_SRC0L_SHIFT; - break; - case 1: - hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT; - break; - case 2: - hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >> - NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT; - hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) << - NV30_VP_INST_SRC2L_SHIFT; - break; - default: - assert(0); - } -} - -static void -emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) -{ - struct nv30_vertex_program *vp = vpc->vp; - - switch (dst.type) { - case NV30SR_TEMP: - hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT); - break; - case NV30SR_OUTPUT: - switch (dst.index) { - case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; - case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; - case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; - case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; - case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; - case NV30_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; - case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; - case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; - case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; - case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; - case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; - case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; - case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; - case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; - default: - break; - } - - hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); - hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); - break; - default: - assert(0); - } -} - -static void -nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, - struct nv30_sreg dst, int mask, - struct nv30_sreg s0, struct nv30_sreg s1, - struct nv30_sreg s2) -{ - struct nv30_vertex_program *vp = vpc->vp; - uint32_t *hw; - - vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); - vpc->vpi = &vp->insns[vp->nr_insns - 1]; - memset(vpc->vpi, 0, sizeof(*vpc->vpi)); - vpc->vpi->const_index = -1; - - hw = vpc->vpi->data; - - hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT); - hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) | - (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) | - (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) | - (3 << NV30_VP_INST_COND_SWZ_W_SHIFT)); - - hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT); -// hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; -// hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); - - emit_dst(vpc, hw, slot, dst); - emit_src(vpc, hw, 0, s0); - emit_src(vpc, hw, 1, s1); - emit_src(vpc, hw, 2, s2); -} - -static INLINE struct nv30_sreg -tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { - struct nv30_sreg src; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); - break; - case TGSI_FILE_CONSTANT: - src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); - break; - case TGSI_FILE_IMMEDIATE: - src = vpc->imm[fsrc->SrcRegister.Index]; - break; - case TGSI_FILE_TEMPORARY: - if (vpc->high_temp < fsrc->SrcRegister.Index) - vpc->high_temp = fsrc->SrcRegister.Index; - src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); - break; - default: - NOUVEAU_ERR("bad src file\n"); - break; - } - - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; - return src; -} - -static INLINE struct nv30_sreg -tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { - struct nv30_sreg dst; - - switch (fdst->DstRegister.File) { - case TGSI_FILE_OUTPUT: - dst = nv30_sr(NV30SR_OUTPUT, - vpc->output_map[fdst->DstRegister.Index]); - - break; - case TGSI_FILE_TEMPORARY: - dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); - if (vpc->high_temp < dst.index) - vpc->high_temp = dst.index; - break; - default: - NOUVEAU_ERR("bad dst file\n"); - break; - } - - return dst; -} - -static INLINE int -tgsi_mask(uint tgsi) -{ - int mask = 0; - - if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; - if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; - if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; - if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; - return mask; -} - -static boolean -nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, - const struct tgsi_full_instruction *finst) -{ - struct nv30_sreg src[3], dst, tmp; - struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); - int mask; - int ai = -1, ci = -1; - int i; - - if (finst->Instruction.Opcode == TGSI_OPCODE_END) - return TRUE; - - vpc->temp_temp_count = 0; - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { - src[i] = tgsi_src(vpc, fsrc); - } - } - - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; - src[i] = tgsi_src(vpc, fsrc); - } else { - src[i] = temp(vpc); - arith(vpc, 0, OP_MOV, src[i], MASK_ALL, - tgsi_src(vpc, fsrc), none, none); - } - break; - /*XXX: index comparison is broken now that consts come from - * two different register files. - */ - case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; - src[i] = tgsi_src(vpc, fsrc); - } else { - src[i] = temp(vpc); - arith(vpc, 0, OP_MOV, src[i], MASK_ALL, - tgsi_src(vpc, fsrc), none, none); - } - break; - case TGSI_FILE_TEMPORARY: - /* handled above */ - break; - default: - NOUVEAU_ERR("bad src file\n"); - return FALSE; - } - } - - dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); - - switch (finst->Instruction.Opcode) { - case TGSI_OPCODE_ABS: - arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); - break; - case TGSI_OPCODE_ADD: - arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); - break; - case TGSI_OPCODE_ARL: - arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_DP3: - arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DP4: - arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DPH: - arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DST: - arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_EX2: - arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_EXP: - arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_FLR: - arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FRC: - arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_LG2: - arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_LIT: - arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_LOG: - arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_MAD: - arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); - break; - case TGSI_OPCODE_MAX: - arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MIN: - arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MOV: - arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_MUL: - arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_POW: - tmp = temp(vpc); - arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, - swz(src[0], X, X, X, X)); - arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), - swz(src[1], X, X, X, X), none); - arith(vpc, 1, OP_EX2, dst, mask, none, none, - swz(tmp, X, X, X, X)); - break; - case TGSI_OPCODE_RCP: - arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_RET: - break; - case TGSI_OPCODE_RSQ: - arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_SGE: - arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SLT: - arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SUB: - arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); - break; - case TGSI_OPCODE_XPD: - tmp = temp(vpc); - arith(vpc, 0, OP_MUL, tmp, mask, - swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); - arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), - swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), - neg(tmp)); - break; - default: - NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); - return FALSE; - } - - return TRUE; -} - -static boolean -nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, - const struct tgsi_full_declaration *fdec) -{ - int hw; - - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - hw = NV30_VP_INST_DEST_POS; - break; - case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV30_VP_INST_DEST_COL0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV30_VP_INST_DEST_COL1; - } else { - NOUVEAU_ERR("bad colour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_BCOLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV30_VP_INST_DEST_BFC0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV30_VP_INST_DEST_BFC1; - } else { - NOUVEAU_ERR("bad bcolour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_FOG: - hw = NV30_VP_INST_DEST_FOGC; - break; - case TGSI_SEMANTIC_PSIZE: - hw = NV30_VP_INST_DEST_PSZ; - break; - case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { - hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); - } else { - NOUVEAU_ERR("bad generic semantic index\n"); - return FALSE; - } - break; - default: - NOUVEAU_ERR("bad output semantic\n"); - return FALSE; - } - - vpc->output_map[fdec->u.DeclarationRange.First] = hw; - return TRUE; -} - -static boolean -nv30_vertprog_prepare(struct nv30_vpc *vpc) -{ - struct tgsi_parse_context p; - int nr_imm = 0; - - tgsi_parse_init(&p, vpc->vp->pipe->tokens); - while (!tgsi_parse_end_of_tokens(&p)) { - const union tgsi_full_token *tok = &p.FullToken; - - tgsi_parse_token(&p); - switch(tok->Token.Type) { - case TGSI_TOKEN_TYPE_IMMEDIATE: - nr_imm++; - break; - default: - break; - } - } - tgsi_parse_free(&p); - - if (nr_imm) { - vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); - assert(vpc->imm); - } - - return TRUE; -} - -void -nv30_vertprog_translate(struct nv30_context *nv30, - struct nv30_vertex_program *vp) -{ - struct tgsi_parse_context parse; - struct nv30_vpc *vpc = NULL; - - vpc = CALLOC(1, sizeof(struct nv30_vpc)); - if (!vpc) - return; - vpc->vp = vp; - vpc->high_temp = -1; - - if (!nv30_vertprog_prepare(vpc)) { - free(vpc); - return; - } - - tgsi_parse_init(&parse, vp->pipe->tokens); - - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - { - const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; - switch (fdec->Declaration.File) { - case TGSI_FILE_OUTPUT: - if (!nv30_vertprog_parse_decl_output(vpc, fdec)) - goto out_err; - break; - default: - break; - } - } - break; - case TGSI_TOKEN_TYPE_IMMEDIATE: - { - const struct tgsi_full_immediate *imm; - - imm = &parse.FullToken.FullImmediate; - assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); -// assert(imm->Immediate.Size == 4); - vpc->imm[vpc->nr_imm++] = - constant(vpc, -1, - imm->u.ImmediateFloat32[0].Float, - imm->u.ImmediateFloat32[1].Float, - imm->u.ImmediateFloat32[2].Float, - imm->u.ImmediateFloat32[3].Float); - } - break; - case TGSI_TOKEN_TYPE_INSTRUCTION: - { - const struct tgsi_full_instruction *finst; - finst = &parse.FullToken.FullInstruction; - if (!nv30_vertprog_parse_instruction(vpc, finst)) - goto out_err; - } - break; - default: - break; - } - } - - vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST; - vp->translated = TRUE; -out_err: - tgsi_parse_free(&parse); - free(vpc); -} - -void -nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) -{ - struct nouveau_winsys *nvws = nv30->nvws; - struct pipe_winsys *ws = nv30->pipe.winsys; - boolean upload_code = FALSE, upload_data = FALSE; - int i; - - /* Translate TGSI shader into hw bytecode */ - if (!vp->translated) { - nv30_vertprog_translate(nv30, vp); - if (!vp->translated) - assert(0); - } - - /* Allocate hw vtxprog exec slots */ - if (!vp->exec) { - struct nouveau_resource *heap = nv30->vertprog.exec_heap; - uint vplen = vp->nr_insns; - - if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { - while (heap->next && heap->size < vplen) { - struct nv30_vertex_program *evict; - - evict = heap->next->priv; - nvws->res_free(&evict->exec); - } - - if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) - assert(0); - } - - upload_code = TRUE; - } - - /* Allocate hw vtxprog const slots */ - if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv30->vertprog.data_heap; - - if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { - while (heap->next && heap->size < vp->nr_consts) { - struct nv30_vertex_program *evict; - - evict = heap->next->priv; - nvws->res_free(&evict->data); - } - - if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) - assert(0); - } - - /*XXX: handle this some day */ - assert(vp->data->start >= vp->data_start_min); - - upload_data = TRUE; - if (vp->data_start != vp->data->start) - upload_code = TRUE; - } - - /* If exec or data segments moved we need to patch the program to - * fixup offsets and register IDs. - */ - if (vp->exec_start != vp->exec->start) { - for (i = 0; i < vp->nr_insns; i++) { - struct nv30_vertex_program_exec *vpi = &vp->insns[i]; - - if (vpi->has_branch_offset) { - assert(0); - } - } - - vp->exec_start = vp->exec->start; - } - - if (vp->nr_consts && vp->data_start != vp->data->start) { - for (i = 0; i < vp->nr_insns; i++) { - struct nv30_vertex_program_exec *vpi = &vp->insns[i]; - - if (vpi->const_index >= 0) { - vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; - vpi->data[1] |= - (vpi->const_index + vp->data->start) << - NV30_VP_INST_CONST_SRC_SHIFT; - - } - } - - vp->data_start = vp->data->start; - } - - /* Update + Upload constant values */ - if (vp->nr_consts) { - float *map = NULL; - - if (nv30->vertprog.constant_buf) { - map = ws->buffer_map(ws, nv30->vertprog.constant_buf, - PIPE_BUFFER_USAGE_CPU_READ); - } - - for (i = 0; i < vp->nr_consts; i++) { - struct nv30_vertex_program_data *vpd = &vp->consts[i]; - - if (vpd->index >= 0) { - if (!upload_data && - !memcmp(vpd->value, &map[vpd->index * 4], - 4 * sizeof(float))) - continue; - memcpy(vpd->value, &map[vpd->index * 4], - 4 * sizeof(float)); - } - - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); - OUT_RING (i + vp->data->start); - OUT_RINGp ((uint32_t *)vpd->value, 4); - } - - if (map) { - ws->buffer_unmap(ws, nv30->vertprog.constant_buf); - } - } - - /* Upload vtxprog */ - if (upload_code) { -#if 0 - for (i = 0; i < vp->nr_insns; i++) { - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); - } -#endif - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (/*vp->exec->start*/0); - for (i = 0; i < vp->nr_insns; i++) { - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); - OUT_RINGp (vp->insns[i].data, 4); - } - } - - BEGIN_RING(rankine, NV34TCL_VP_START_FROM_ID, 1); -// OUT_RING (vp->exec->start); - OUT_RING (0); - - nv30->vertprog.active = vp; -} - -void -nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) -{ - if (vp->nr_consts) - free(vp->consts); - if (vp->nr_insns) - free(vp->insns); -} - diff --git a/src/mesa/pipe/nv40/Makefile b/src/mesa/pipe/nv40/Makefile deleted file mode 100644 index c9c3a8032e5..00000000000 --- a/src/mesa/pipe/nv40/Makefile +++ /dev/null @@ -1,29 +0,0 @@ -TOP = ../../../.. -include $(TOP)/configs/current - -LIBNAME = nv40 - -DRIVER_SOURCES = \ - nv40_clear.c \ - nv40_context.c \ - nv40_draw.c \ - nv40_fragprog.c \ - nv40_fragtex.c \ - nv40_miptree.c \ - nv40_query.c \ - nv40_state.c \ - nv40_state_emit.c \ - nv40_surface.c \ - nv40_vbo.c \ - nv40_vertprog.c - -C_SOURCES = \ - $(COMMON_SOURCES) \ - $(DRIVER_SOURCES) - -ASM_SOURCES = - -include ../Makefile.template - -symlinks: - diff --git a/src/mesa/pipe/nv40/nv40_clear.c b/src/mesa/pipe/nv40/nv40_clear.c deleted file mode 100644 index 2c4e8f01fda..00000000000 --- a/src/mesa/pipe/nv40/nv40_clear.c +++ /dev/null @@ -1,12 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "nv40_context.h" - -void -nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue) -{ - pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); -} diff --git a/src/mesa/pipe/nv40/nv40_context.c b/src/mesa/pipe/nv40/nv40_context.c deleted file mode 100644 index 6e86ca00816..00000000000 --- a/src/mesa/pipe/nv40/nv40_context.c +++ /dev/null @@ -1,312 +0,0 @@ -#include "pipe/draw/draw_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "pipe/p_util.h" - -#include "nv40_context.h" - -#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf -#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 -#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 - -static const char * -nv40_get_name(struct pipe_context *pipe) -{ - struct nv40_context *nv40 = nv40_context(pipe); - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", nv40->chipset); - return buffer; -} - -static const char * -nv40_get_vendor(struct pipe_context *pipe) -{ - return "nouveau"; -} - -static int -nv40_get_param(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 16; - case PIPE_CAP_NPOT_TEXTURES: - return 1; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 1; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 1; - case PIPE_CAP_POINT_SPRITE: - return 1; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 4; - case PIPE_CAP_OCCLUSION_QUERY: - return 1; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 1; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 13; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 10; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 13; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv40_get_paramf(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 16.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} - -static void -nv40_flush(struct pipe_context *pipe, unsigned flags) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; - - if (flags & PIPE_FLUSH_TEXTURE_CACHE) { - BEGIN_RING(curie, 0x1fd8, 1); - OUT_RING (2); - BEGIN_RING(curie, 0x1fd8, 1); - OUT_RING (1); - } - - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv40->hw->sync, 0); - BEGIN_RING(curie, 0x104, 1); - OUT_RING (0); - BEGIN_RING(curie, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv40->hw->sync, 0, 0, 2000); -} - -static void -nv40_channel_takedown(struct nv40_channel_context *cnv40) -{ - struct nouveau_winsys *nvws = cnv40->nvws; - - nvws->res_free(&cnv40->vp_exec_heap); - nvws->res_free(&cnv40->vp_data_heap); - nvws->res_free(&cnv40->query_heap); - nvws->notifier_free(&cnv40->query); - nvws->notifier_free(&cnv40->sync); - nvws->grobj_free(&cnv40->curie); - free(cnv40); -} - -static struct nv40_channel_context * -nv40_channel_init(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv40_channel_context *cnv40 = NULL; - struct nouveau_stateobj *so; - unsigned curie_class = 0; - int ret; - - switch (chipset & 0xf0) { - case 0x40: - if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) - curie_class = NV40TCL; - else - if (NV4X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) - curie_class = NV44TCL; - break; - case 0x60: - if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) - curie_class = NV44TCL; - break; - default: - break; - } - - if (!curie_class) { - NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset); - return NULL; - } - - cnv40 = CALLOC(1, sizeof(struct nv40_channel_context)); - if (!cnv40) - return NULL; - cnv40->chipset = chipset; - cnv40->nvws = nvws; - - /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &cnv40->sync); - if (ret) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv40_channel_takedown(cnv40); - return NULL; - } - - /* Query objects */ - ret = nvws->notifier_alloc(nvws, 32, &cnv40->query); - if (ret) { - NOUVEAU_ERR("Error initialising query objects: %d\n", ret); - nv40_channel_takedown(cnv40); - return NULL; - } - - ret = nvws->res_init(&cnv40->query_heap, 0, 32); - if (ret) { - NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); - nv40_channel_takedown(cnv40); - return NULL; - } - - /* Vtxprog resources */ - if (nvws->res_init(&cnv40->vp_exec_heap, 0, 512) || - nvws->res_init(&cnv40->vp_data_heap, 0, 256)) { - nv40_channel_takedown(cnv40); - return NULL; - } - - /* 3D object */ - ret = nvws->grobj_alloc(nvws, curie_class, &cnv40->curie); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - /* Static curie initialisation */ - so = so_new(128, 0); - so_method(so, cnv40->curie, NV40TCL_DMA_NOTIFY, 1); - so_data (so, cnv40->sync->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_TEXTURE0, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->gart->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_COLOR1, 1); - so_data (so, nvws->channel->vram->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_COLOR0, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->vram->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_VTXBUF0, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->gart->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_FENCE, 2); - so_data (so, 0); - so_data (so, cnv40->query->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_UNK01AC, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->vram->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_COLOR2, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->vram->handle); - - so_method(so, cnv40->curie, 0x1ea4, 3); - so_data (so, 0x00000010); - so_data (so, 0x01000100); - so_data (so, 0xff800006); - - /* vtxprog output routing */ - so_method(so, cnv40->curie, 0x1fc4, 1); - so_data (so, 0x06144321); - so_method(so, cnv40->curie, 0x1fc8, 2); - so_data (so, 0xedcba987); - so_data (so, 0x00000021); - so_method(so, cnv40->curie, 0x1fd0, 1); - so_data (so, 0x00171615); - so_method(so, cnv40->curie, 0x1fd4, 1); - so_data (so, 0x001b1a19); - - so_method(so, cnv40->curie, 0x1ef8, 1); - so_data (so, 0x0020ffff); - so_method(so, cnv40->curie, 0x1d64, 1); - so_data (so, 0x00d30000); - so_method(so, cnv40->curie, 0x1e94, 1); - so_data (so, 0x00000001); - - so_emit(nvws, so); - so_ref(NULL, &so); - nvws->push_flush(nvws->channel, 0); - - return cnv40; -} - -static void -nv40_destroy(struct pipe_context *pipe) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - if (nv40->draw) - draw_destroy(nv40->draw); - - if (nv40->hw) { - if (--nv40->hw->refcount == 0) - nv40_channel_takedown(nv40->hw); - } - - free(nv40); -} - -struct pipe_context * -nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv40_context *nv40; - - nv40 = CALLOC(1, sizeof(struct nv40_context)); - if (!nv40) - return NULL; - - nv40->hw = nv40_channel_init(ws, nvws, chipset); - if (!nv40->hw) { - nv40_destroy(&nv40->pipe); - return NULL; - } - - nv40->chipset = chipset; - nv40->nvws = nvws; - - nv40->pipe.winsys = ws; - nv40->pipe.destroy = nv40_destroy; - nv40->pipe.get_name = nv40_get_name; - nv40->pipe.get_vendor = nv40_get_vendor; - nv40->pipe.get_param = nv40_get_param; - nv40->pipe.get_paramf = nv40_get_paramf; - nv40->pipe.draw_arrays = nv40_draw_arrays; - nv40->pipe.draw_elements = nv40_draw_elements; - nv40->pipe.clear = nv40_clear; - nv40->pipe.flush = nv40_flush; - - nv40_init_query_functions(nv40); - nv40_init_surface_functions(nv40); - nv40_init_state_functions(nv40); - nv40_init_miptree_functions(nv40); - - nv40->draw = draw_create(); - assert(nv40->draw); - draw_set_rasterize_stage(nv40->draw, nv40_draw_render_stage(nv40)); - - return &nv40->pipe; -} - diff --git a/src/mesa/pipe/nv40/nv40_context.h b/src/mesa/pipe/nv40/nv40_context.h deleted file mode 100644 index cf2a14405a4..00000000000 --- a/src/mesa/pipe/nv40/nv40_context.h +++ /dev/null @@ -1,153 +0,0 @@ -#ifndef __NV40_CONTEXT_H__ -#define __NV40_CONTEXT_H__ - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/draw/draw_vertex.h" - -#include "pipe/nouveau/nouveau_winsys.h" -#include "pipe/nouveau/nouveau_gldefs.h" - -#define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv40_channel_context *ctx = nv40->hw -#include "pipe/nouveau/nouveau_push.h" -#include "pipe/nouveau/nouveau_stateobj.h" - -#include "nv40_state.h" - -#define NOUVEAU_ERR(fmt, args...) \ - fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); -#define NOUVEAU_MSG(fmt, args...) \ - fprintf(stderr, "nouveau: "fmt, ##args); - -#define NV40_NEW_BLEND (1 << 0) -#define NV40_NEW_RAST (1 << 1) -#define NV40_NEW_ZSA (1 << 2) -#define NV40_NEW_SAMPLER (1 << 3) -#define NV40_NEW_FB (1 << 4) -#define NV40_NEW_STIPPLE (1 << 5) -#define NV40_NEW_SCISSOR (1 << 6) -#define NV40_NEW_VIEWPORT (1 << 7) -#define NV40_NEW_BCOL (1 << 8) -#define NV40_NEW_VERTPROG (1 << 9) -#define NV40_NEW_FRAGPROG (1 << 10) -#define NV40_NEW_ARRAYS (1 << 11) - -struct nv40_channel_context { - struct nouveau_winsys *nvws; - unsigned refcount; - - unsigned chipset; - - /* HW graphics objects */ - struct nouveau_grobj *curie; - struct nouveau_notifier *sync; - - /* Query object resources */ - struct nouveau_notifier *query; - struct nouveau_resource *query_heap; - - /* Vtxprog resources */ - struct nouveau_resource *vp_exec_heap; - struct nouveau_resource *vp_data_heap; -}; - -struct nv40_context { - struct pipe_context pipe; - struct nouveau_winsys *nvws; - - struct nv40_channel_context *hw; - struct draw_context *draw; - - int chipset; - - uint32_t dirty; - - struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; - unsigned dirty_samplers; - unsigned fp_samplers; - unsigned vp_samplers; - - struct nouveau_stateobj *so_framebuffer; - struct nouveau_stateobj *so_fragtex[16]; - struct nouveau_stateobj *so_vtxbuf; - struct nouveau_stateobj *so_blend; - struct nouveau_stateobj *so_rast; - struct nouveau_stateobj *so_zsa; - struct nouveau_stateobj *so_bcol; - struct nouveau_stateobj *so_scissor; - struct nouveau_stateobj *so_viewport; - struct nouveau_stateobj *so_stipple; - - struct { - struct nv40_vertex_program *active; - - struct nv40_vertex_program *current; - struct pipe_buffer *constant_buf; - } vertprog; - - struct { - struct nv40_fragment_program *active; - - struct nv40_fragment_program *current; - struct pipe_buffer *constant_buf; - } fragprog; - - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; - struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; -}; - -static INLINE struct nv40_context * -nv40_context(struct pipe_context *pipe) -{ - return (struct nv40_context *)pipe; -} - -extern void nv40_init_state_functions(struct nv40_context *nv40); -extern void nv40_init_surface_functions(struct nv40_context *nv40); -extern void nv40_init_miptree_functions(struct nv40_context *nv40); -extern void nv40_init_query_functions(struct nv40_context *nv40); - -/* nv40_draw.c */ -extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); - -/* nv40_vertprog.c */ -extern void nv40_vertprog_translate(struct nv40_context *, - struct nv40_vertex_program *); -extern void nv40_vertprog_bind(struct nv40_context *, - struct nv40_vertex_program *); -extern void nv40_vertprog_destroy(struct nv40_context *, - struct nv40_vertex_program *); - -/* nv40_fragprog.c */ -extern void nv40_fragprog_translate(struct nv40_context *, - struct nv40_fragment_program *); -extern void nv40_fragprog_bind(struct nv40_context *, - struct nv40_fragment_program *); -extern void nv40_fragprog_destroy(struct nv40_context *, - struct nv40_fragment_program *); - -/* nv40_fragtex.c */ -extern void nv40_fragtex_bind(struct nv40_context *); - -/* nv40_state.c and friends */ -extern void nv40_emit_hw_state(struct nv40_context *nv40); -extern void nv40_state_tex_update(struct nv40_context *nv40); - -/* nv40_vbo.c */ -extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, - unsigned start, unsigned count); -extern boolean nv40_draw_elements(struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, - unsigned indexSize, - unsigned mode, unsigned start, - unsigned count); - -/* nv40_clear.c */ -extern void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue); - -#endif diff --git a/src/mesa/pipe/nv40/nv40_dma.h b/src/mesa/pipe/nv40/nv40_dma.h deleted file mode 100644 index 1fb82677689..00000000000 --- a/src/mesa/pipe/nv40/nv40_dma.h +++ /dev/null @@ -1,66 +0,0 @@ -#ifndef __NV40_DMA_H__ -#define __NV40_DMA_H__ - -#include "pipe/nouveau/nouveau_winsys.h" - -#define OUT_RING(data) do { \ - (*nv40->nvws->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - memcpy(nv40->nvws->channel->pushbuf->cur, (src), (size) * 4); \ - nv40->nvws->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - if (nv40->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - nv40->nvws->push_flush(nv40->nvws->channel, ((size) + 1)); \ - OUT_RING((nv40->obj->subc << 13) | ((size) << 18) | (mthd)); \ - nv40->nvws->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - -#define FIRE_RING() do { \ - nv40->nvws->push_flush(nv40->nvws->channel, 0); \ -} while(0) - -#define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nv40->nvws->push_reloc(nv40->nvws->channel, \ - nv40->nvws->channel->pushbuf->cur, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ -} while(0) - -/* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) - -/* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - nv40->nvws->channel->vram->handle, \ - nv40->nvws->channel->gart->handle); \ -} while(0) - -/* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) - -/* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) - -#endif diff --git a/src/mesa/pipe/nv40/nv40_draw.c b/src/mesa/pipe/nv40/nv40_draw.c deleted file mode 100644 index d361d5f07d9..00000000000 --- a/src/mesa/pipe/nv40/nv40_draw.c +++ /dev/null @@ -1,62 +0,0 @@ -#include "pipe/draw/draw_private.h" -#include "pipe/p_util.h" - -#include "nv40_context.h" - -struct nv40_draw_stage { - struct draw_stage draw; - struct nv40_context *nv40; -}; - -static void -nv40_draw_point(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv40_draw_line(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv40_draw_tri(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv40_draw_flush(struct draw_stage *draw, unsigned flags) -{ -} - -static void -nv40_draw_reset_stipple_counter(struct draw_stage *draw) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv40_draw_destroy(struct draw_stage *draw) -{ - free(draw); -} - -struct draw_stage * -nv40_draw_render_stage(struct nv40_context *nv40) -{ - struct nv40_draw_stage *nv40draw = CALLOC_STRUCT(nv40_draw_stage); - - nv40draw->nv40 = nv40; - nv40draw->draw.draw = nv40->draw; - nv40draw->draw.point = nv40_draw_point; - nv40draw->draw.line = nv40_draw_line; - nv40draw->draw.tri = nv40_draw_tri; - nv40draw->draw.flush = nv40_draw_flush; - nv40draw->draw.reset_stipple_counter = nv40_draw_reset_stipple_counter; - nv40draw->draw.destroy = nv40_draw_destroy; - - return &nv40draw->draw; -} - diff --git a/src/mesa/pipe/nv40/nv40_fragprog.c b/src/mesa/pipe/nv40/nv40_fragprog.c deleted file mode 100644 index 7487fb896f3..00000000000 --- a/src/mesa/pipe/nv40/nv40_fragprog.c +++ /dev/null @@ -1,842 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/p_shader_tokens.h" -#include "pipe/tgsi/util/tgsi_parse.h" -#include "pipe/tgsi/util/tgsi_util.h" - -#include "nv40_context.h" - -#define SWZ_X 0 -#define SWZ_Y 1 -#define SWZ_Z 2 -#define SWZ_W 3 -#define MASK_X 1 -#define MASK_Y 2 -#define MASK_Z 4 -#define MASK_W 8 -#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) -#define DEF_SCALE NV40_FP_OP_DST_SCALE_1X -#define DEF_CTEST NV40_FP_OP_COND_TR -#include "nv40_shader.h" - -#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) -#define neg(s) nv40_sr_neg((s)) -#define abs(s) nv40_sr_abs((s)) -#define scale(s,v) nv40_sr_scale((s), NV40_FP_OP_DST_SCALE_##v) - -#define MAX_CONSTS 128 -#define MAX_IMM 32 -struct nv40_fpc { - struct nv40_fragment_program *fp; - - uint attrib_map[PIPE_MAX_SHADER_INPUTS]; - - int high_temp; - int temp_temp_count; - int num_regs; - - uint depth_id; - uint colour_id; - - unsigned inst_offset; - - struct { - int pipe; - float vals[4]; - } consts[MAX_CONSTS]; - int nr_consts; - - struct nv40_sreg imm[MAX_IMM]; - unsigned nr_imm; -}; - -static INLINE struct nv40_sreg -temp(struct nv40_fpc *fpc) -{ - int idx; - - idx = fpc->temp_temp_count++; - idx += fpc->high_temp + 1; - return nv40_sr(NV40SR_TEMP, idx); -} - -static INLINE struct nv40_sreg -constant(struct nv40_fpc *fpc, int pipe, float vals[4]) -{ - int idx; - - if (fpc->nr_consts == MAX_CONSTS) - assert(0); - idx = fpc->nr_consts++; - - fpc->consts[idx].pipe = pipe; - if (pipe == -1) - memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float)); - return nv40_sr(NV40SR_CONST, idx); -} - -#define arith(cc,s,o,d,m,s0,s1,s2) \ - nv40_fp_arith((cc), (s), NV40_FP_OP_OPCODE_##o, \ - (d), (m), (s0), (s1), (s2)) -#define tex(cc,s,o,u,d,m,s0,s1,s2) \ - nv40_fp_tex((cc), (s), NV40_FP_OP_OPCODE_##o, (u), \ - (d), (m), (s0), none, none) - -static void -grow_insns(struct nv40_fpc *fpc, int size) -{ - struct nv40_fragment_program *fp = fpc->fp; - - fp->insn_len += size; - fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len); -} - -static void -emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) -{ - struct nv40_fragment_program *fp = fpc->fp; - uint32_t *hw = &fp->insn[fpc->inst_offset]; - uint32_t sr = 0; - - switch (src.type) { - case NV40SR_INPUT: - sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); - hw[0] |= (src.index << NV40_FP_OP_INPUT_SRC_SHIFT); - break; - case NV40SR_OUTPUT: - sr |= NV40_FP_REG_SRC_HALF; - /* fall-through */ - case NV40SR_TEMP: - sr |= (NV40_FP_REG_TYPE_TEMP << NV40_FP_REG_TYPE_SHIFT); - sr |= (src.index << NV40_FP_REG_SRC_SHIFT); - break; - case NV40SR_CONST: - grow_insns(fpc, 4); - hw = &fp->insn[fpc->inst_offset]; - if (fpc->consts[src.index].pipe >= 0) { - struct nv40_fragment_program_data *fpd; - - fp->consts = realloc(fp->consts, ++fp->nr_consts * - sizeof(*fpd)); - fpd = &fp->consts[fp->nr_consts - 1]; - fpd->offset = fpc->inst_offset + 4; - fpd->index = fpc->consts[src.index].pipe; - memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4); - } else { - memcpy(&fp->insn[fpc->inst_offset + 4], - fpc->consts[src.index].vals, - sizeof(uint32_t) * 4); - } - - sr |= (NV40_FP_REG_TYPE_CONST << NV40_FP_REG_TYPE_SHIFT); - break; - case NV40SR_NONE: - sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT); - break; - default: - assert(0); - } - - if (src.negate) - sr |= NV40_FP_REG_NEGATE; - - if (src.abs) - hw[1] |= (1 << (29 + pos)); - - sr |= ((src.swz[0] << NV40_FP_REG_SWZ_X_SHIFT) | - (src.swz[1] << NV40_FP_REG_SWZ_Y_SHIFT) | - (src.swz[2] << NV40_FP_REG_SWZ_Z_SHIFT) | - (src.swz[3] << NV40_FP_REG_SWZ_W_SHIFT)); - - hw[pos + 1] |= sr; -} - -static void -emit_dst(struct nv40_fpc *fpc, struct nv40_sreg dst) -{ - struct nv40_fragment_program *fp = fpc->fp; - uint32_t *hw = &fp->insn[fpc->inst_offset]; - - switch (dst.type) { - case NV40SR_TEMP: - if (fpc->num_regs < (dst.index + 1)) - fpc->num_regs = dst.index + 1; - break; - case NV40SR_OUTPUT: - if (dst.index == 1) { - fp->fp_control |= 0xe; - } else { - hw[0] |= NV40_FP_OP_OUT_REG_HALF; - } - break; - case NV40SR_NONE: - hw[0] |= (1 << 30); - break; - default: - assert(0); - } - - hw[0] |= (dst.index << NV40_FP_OP_OUT_REG_SHIFT); -} - -static void -nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, - struct nv40_sreg dst, int mask, - struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) -{ - struct nv40_fragment_program *fp = fpc->fp; - uint32_t *hw; - - fpc->inst_offset = fp->insn_len; - grow_insns(fpc, 4); - hw = &fp->insn[fpc->inst_offset]; - memset(hw, 0, sizeof(uint32_t) * 4); - - if (op == NV40_FP_OP_OPCODE_KIL) - fp->fp_control |= NV40TCL_FP_CONTROL_KIL; - hw[0] |= (op << NV40_FP_OP_OPCODE_SHIFT); - hw[0] |= (mask << NV40_FP_OP_OUTMASK_SHIFT); - hw[2] |= (dst.dst_scale << NV40_FP_OP_DST_SCALE_SHIFT); - - if (sat) - hw[0] |= NV40_FP_OP_OUT_SAT; - - if (dst.cc_update) - hw[0] |= NV40_FP_OP_COND_WRITE_ENABLE; - hw[1] |= (dst.cc_test << NV40_FP_OP_COND_SHIFT); - hw[1] |= ((dst.cc_swz[0] << NV40_FP_OP_COND_SWZ_X_SHIFT) | - (dst.cc_swz[1] << NV40_FP_OP_COND_SWZ_Y_SHIFT) | - (dst.cc_swz[2] << NV40_FP_OP_COND_SWZ_Z_SHIFT) | - (dst.cc_swz[3] << NV40_FP_OP_COND_SWZ_W_SHIFT)); - - emit_dst(fpc, dst); - emit_src(fpc, 0, s0); - emit_src(fpc, 1, s1); - emit_src(fpc, 2, s2); -} - -static void -nv40_fp_tex(struct nv40_fpc *fpc, int sat, int op, int unit, - struct nv40_sreg dst, int mask, - struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2) -{ - struct nv40_fragment_program *fp = fpc->fp; - - nv40_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2); - - fp->insn[fpc->inst_offset] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT); - fp->samplers |= (1 << unit); -} - -static INLINE struct nv40_sreg -tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) -{ - struct nv40_sreg src; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - src = nv40_sr(NV40SR_INPUT, - fpc->attrib_map[fsrc->SrcRegister.Index]); - break; - case TGSI_FILE_CONSTANT: - src = constant(fpc, fsrc->SrcRegister.Index, NULL); - break; - case TGSI_FILE_IMMEDIATE: - assert(fsrc->SrcRegister.Index < fpc->nr_imm); - src = fpc->imm[fsrc->SrcRegister.Index]; - break; - case TGSI_FILE_TEMPORARY: - src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index + 1); - if (fpc->high_temp < src.index) - fpc->high_temp = src.index; - break; - /* This is clearly insane, but gallium hands us shaders like this. - * Luckily fragprog results are just temp regs.. - */ - case TGSI_FILE_OUTPUT: - if (fsrc->SrcRegister.Index == fpc->colour_id) - return nv40_sr(NV40SR_OUTPUT, 0); - else - return nv40_sr(NV40SR_OUTPUT, 1); - break; - default: - NOUVEAU_ERR("bad src file\n"); - break; - } - - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; - return src; -} - -static INLINE struct nv40_sreg -tgsi_dst(struct nv40_fpc *fpc, const struct tgsi_full_dst_register *fdst) { - int idx; - - switch (fdst->DstRegister.File) { - case TGSI_FILE_OUTPUT: - if (fdst->DstRegister.Index == fpc->colour_id) - return nv40_sr(NV40SR_OUTPUT, 0); - else - return nv40_sr(NV40SR_OUTPUT, 1); - break; - case TGSI_FILE_TEMPORARY: - idx = fdst->DstRegister.Index + 1; - if (fpc->high_temp < idx) - fpc->high_temp = idx; - return nv40_sr(NV40SR_TEMP, idx); - case TGSI_FILE_NULL: - return nv40_sr(NV40SR_NONE, 0); - default: - NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File); - return nv40_sr(NV40SR_NONE, 0); - } -} - -static INLINE int -tgsi_mask(uint tgsi) -{ - int mask = 0; - - if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; - if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; - if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; - if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; - return mask; -} - -static boolean -src_native_swz(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc, - struct nv40_sreg *src) -{ - const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); - struct nv40_sreg tgsi = tgsi_src(fpc, fsrc); - uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; - uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, - fsrc->SrcRegisterExtSwz.NegateY, - fsrc->SrcRegisterExtSwz.NegateZ, - fsrc->SrcRegisterExtSwz.NegateW }; - uint c; - - for (c = 0; c < 4; c++) { - switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: - mask |= (1 << c); - break; - case TGSI_EXTSWIZZLE_ZERO: - zero_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; - case TGSI_EXTSWIZZLE_ONE: - one_mask |= (1 << c); - tgsi.swz[c] = SWZ_X; - break; - default: - assert(0); - } - - if (!tgsi.negate && neg[c]) - neg_mask |= (1 << c); - } - - if (mask == MASK_ALL && !neg_mask) - return TRUE; - - *src = temp(fpc); - - if (mask) - arith(fpc, 0, MOV, *src, mask, tgsi, none, none); - - if (zero_mask) - arith(fpc, 0, SFL, *src, zero_mask, *src, none, none); - - if (one_mask) - arith(fpc, 0, STR, *src, one_mask, *src, none, none); - - if (neg_mask) { - struct nv40_sreg one = temp(fpc); - arith(fpc, 0, STR, one, neg_mask, one, none, none); - arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none); - } - - return FALSE; -} - -static boolean -nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, - const struct tgsi_full_instruction *finst) -{ - const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); - struct nv40_sreg src[3], dst, tmp; - int mask, sat, unit; - int ai = -1, ci = -1; - int i; - - if (finst->Instruction.Opcode == TGSI_OPCODE_END) - return TRUE; - - fpc->temp_temp_count = 0; - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { - src[i] = tgsi_src(fpc, fsrc); - } - } - - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - case TGSI_FILE_CONSTANT: - case TGSI_FILE_TEMPORARY: - if (!src_native_swz(fpc, fsrc, &src[i])) - continue; - break; - default: - break; - } - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; - src[i] = tgsi_src(fpc, fsrc); - } else { - NOUVEAU_MSG("extra src attr %d\n", - fsrc->SrcRegister.Index); - src[i] = temp(fpc); - arith(fpc, 0, MOV, src[i], MASK_ALL, - tgsi_src(fpc, fsrc), none, none); - } - break; - case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; - src[i] = tgsi_src(fpc, fsrc); - } else { - src[i] = temp(fpc); - arith(fpc, 0, MOV, src[i], MASK_ALL, - tgsi_src(fpc, fsrc), none, none); - } - break; - case TGSI_FILE_TEMPORARY: - /* handled above */ - break; - case TGSI_FILE_SAMPLER: - unit = fsrc->SrcRegister.Index; - break; - case TGSI_FILE_OUTPUT: - break; - default: - NOUVEAU_ERR("bad src file\n"); - return FALSE; - } - } - - dst = tgsi_dst(fpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); - sat = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE); - - switch (finst->Instruction.Opcode) { - case TGSI_OPCODE_ABS: - arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none); - break; - case TGSI_OPCODE_ADD: - arith(fpc, sat, ADD, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_CMP: - tmp = temp(fpc); - arith(fpc, sat, MOV, dst, mask, src[2], none, none); - tmp.cc_update = 1; - arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none); - dst.cc_test = NV40_VP_INST_COND_LT; - arith(fpc, sat, MOV, dst, mask, src[1], none, none); - break; - case TGSI_OPCODE_COS: - arith(fpc, sat, COS, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_DP3: - arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DP4: - arith(fpc, sat, DP4, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DPH: - tmp = temp(fpc); - arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none); - arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X), - swz(src[1], W, W, W, W), none); - break; - case TGSI_OPCODE_DST: - arith(fpc, sat, DST, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_EX2: - arith(fpc, sat, EX2, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FLR: - arith(fpc, sat, FLR, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FRC: - arith(fpc, sat, FRC, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_KIL: - arith(fpc, 0, KIL, none, 0, none, none, none); - break; - case TGSI_OPCODE_KILP: - dst = nv40_sr(NV40SR_NONE, 0); - dst.cc_update = 1; - arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); - dst.cc_update = 0; dst.cc_test = NV40_FP_OP_COND_LT; - arith(fpc, 0, KIL, dst, 0, none, none, none); - break; - case TGSI_OPCODE_LG2: - arith(fpc, sat, LG2, dst, mask, src[0], none, none); - break; -// case TGSI_OPCODE_LIT: - case TGSI_OPCODE_LRP: - tmp = temp(fpc); - arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); - arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); - break; - case TGSI_OPCODE_MAD: - arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); - break; - case TGSI_OPCODE_MAX: - arith(fpc, sat, MAX, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MIN: - arith(fpc, sat, MIN, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MOV: - arith(fpc, sat, MOV, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_MUL: - arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_POW: - tmp = temp(fpc); - arith(fpc, 0, LG2, tmp, MASK_X, - swz(src[0], X, X, X, X), none, none); - arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), - swz(src[1], X, X, X, X), none); - arith(fpc, sat, EX2, dst, mask, - swz(tmp, X, X, X, X), none, none); - break; - case TGSI_OPCODE_RCP: - arith(fpc, sat, RCP, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_RET: - assert(0); - break; - case TGSI_OPCODE_RFL: - tmp = temp(fpc); - arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); - arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); - arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, - swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); - arith(fpc, sat, MAD, dst, mask, - swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); - break; - case TGSI_OPCODE_RSQ: - tmp = temp(fpc); - arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, - abs(swz(src[0], X, X, X, X)), none, none); - arith(fpc, sat, EX2, dst, mask, - neg(swz(tmp, X, X, X, X)), none, none); - break; - case TGSI_OPCODE_SCS: - if (mask & MASK_X) { - arith(fpc, sat, COS, dst, MASK_X, - swz(src[0], X, X, X, X), none, none); - } - if (mask & MASK_Y) { - arith(fpc, sat, SIN, dst, MASK_Y, - swz(src[0], X, X, X, X), none, none); - } - break; - case TGSI_OPCODE_SIN: - arith(fpc, sat, SIN, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_SGE: - arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SLT: - arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SUB: - arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); - break; - case TGSI_OPCODE_TEX: - if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == - TGSI_EXTSWIZZLE_W) { - tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); - } else - tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_TXB: - tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_XPD: - tmp = temp(fpc); - arith(fpc, 0, MUL, tmp, mask, - swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); - arith(fpc, sat, MAD, dst, (mask & ~MASK_W), - swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), - neg(tmp)); - break; - default: - NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); - return FALSE; - } - - return TRUE; -} - -static boolean -nv40_fragprog_parse_decl_attrib(struct nv40_fpc *fpc, - const struct tgsi_full_declaration *fdec) -{ - int hw; - - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - hw = NV40_FP_OP_INPUT_SRC_POSITION; - break; - case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV40_FP_OP_INPUT_SRC_COL0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV40_FP_OP_INPUT_SRC_COL1; - } else { - NOUVEAU_ERR("bad colour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_FOG: - hw = NV40_FP_OP_INPUT_SRC_FOGC; - break; - case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { - hw = NV40_FP_OP_INPUT_SRC_TC(fdec->Semantic. - SemanticIndex); - } else { - NOUVEAU_ERR("bad generic semantic index\n"); - return FALSE; - } - break; - default: - NOUVEAU_ERR("bad input semantic\n"); - return FALSE; - } - - fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; - return TRUE; -} - -static boolean -nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, - const struct tgsi_full_declaration *fdec) -{ - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - fpc->depth_id = fdec->u.DeclarationRange.First; - break; - case TGSI_SEMANTIC_COLOR: - fpc->colour_id = fdec->u.DeclarationRange.First; - break; - default: - NOUVEAU_ERR("bad output semantic\n"); - return FALSE; - } - - return TRUE; -} - -void -nv40_fragprog_translate(struct nv40_context *nv40, - struct nv40_fragment_program *fp) -{ - struct tgsi_parse_context parse; - struct nv40_fpc *fpc = NULL; - - fpc = CALLOC(1, sizeof(struct nv40_fpc)); - if (!fpc) - return; - fpc->fp = fp; - fpc->high_temp = -1; - fpc->num_regs = 2; - - tgsi_parse_init(&parse, fp->pipe->tokens); - - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - { - const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; - switch (fdec->Declaration.File) { - case TGSI_FILE_INPUT: - if (!nv40_fragprog_parse_decl_attrib(fpc, fdec)) - goto out_err; - break; - case TGSI_FILE_OUTPUT: - if (!nv40_fragprog_parse_decl_output(fpc, fdec)) - goto out_err; - break; - default: - break; - } - } - break; - case TGSI_TOKEN_TYPE_IMMEDIATE: - { - struct tgsi_full_immediate *imm; - float vals[4]; - - imm = &parse.FullToken.FullImmediate; - assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); - assert(fpc->nr_imm < MAX_IMM); - - vals[0] = imm->u.ImmediateFloat32[0].Float; - vals[1] = imm->u.ImmediateFloat32[1].Float; - vals[2] = imm->u.ImmediateFloat32[2].Float; - vals[3] = imm->u.ImmediateFloat32[3].Float; - fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); - } - break; - case TGSI_TOKEN_TYPE_INSTRUCTION: - { - const struct tgsi_full_instruction *finst; - - finst = &parse.FullToken.FullInstruction; - if (!nv40_fragprog_parse_instruction(fpc, finst)) - goto out_err; - } - break; - default: - break; - } - } - - fp->fp_control |= fpc->num_regs << NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT; - - /* Terminate final instruction */ - fp->insn[fpc->inst_offset] |= 0x00000001; - - /* Append NOP + END instruction, may or may not be necessary. */ - fpc->inst_offset = fp->insn_len; - grow_insns(fpc, 4); - fp->insn[fpc->inst_offset + 0] = 0x00000001; - fp->insn[fpc->inst_offset + 1] = 0x00000000; - fp->insn[fpc->inst_offset + 2] = 0x00000000; - fp->insn[fpc->inst_offset + 3] = 0x00000000; - - fp->translated = TRUE; - fp->on_hw = FALSE; -out_err: - tgsi_parse_free(&parse); - free(fpc); -} - -void -nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) -{ - struct pipe_winsys *ws = nv40->pipe.winsys; - struct nouveau_stateobj *so; - int i; - - if (!fp->translated) { - nv40_fragprog_translate(nv40, fp); - if (!fp->translated) - assert(0); - } - - if (fp->nr_consts) { - float *map = ws->buffer_map(ws, nv40->fragprog.constant_buf, - PIPE_BUFFER_USAGE_CPU_READ); - for (i = 0; i < fp->nr_consts; i++) { - struct nv40_fragment_program_data *fpd = &fp->consts[i]; - uint32_t *p = &fp->insn[fpd->offset]; - uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; - - if (!memcmp(p, cb, 4 * sizeof(float))) - continue; - memcpy(p, cb, 4 * sizeof(float)); - fp->on_hw = 0; - } - ws->buffer_unmap(ws, nv40->fragprog.constant_buf); - } - - if (!fp->on_hw) { - const uint32_t le = 1; - uint32_t *map; - - if (!fp->buffer) - fp->buffer = ws->buffer_create(ws, 0x100, 0, - fp->insn_len * 4); - map = ws->buffer_map(ws, fp->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); - -#if 0 - for (i = 0; i < fp->insn_len; i++) { - NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); - } -#endif - - if ((*(const uint8_t *)&le)) { - for (i = 0; i < fp->insn_len; i++) { - map[i] = fp->insn[i]; - } - } else { - /* Weird swapping for big-endian chips */ - for (i = 0; i < fp->insn_len; i++) { - map[i] = ((fp->insn[i] & 0xffff) << 16) | - ((fp->insn[i] >> 16) & 0xffff); - } - } - - ws->buffer_unmap(ws, fp->buffer); - fp->on_hw = TRUE; - } - - so = so_new(4, 1); - so_method(so, nv40->hw->curie, NV40TCL_FP_ADDRESS, 1); - so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, - NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); - so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); - so_data (so, fp->fp_control); - - so_emit(nv40->nvws, so); - so_ref(so, &fp->so); - so_ref(NULL, &so); - - nv40->fragprog.active = fp; -} - -void -nv40_fragprog_destroy(struct nv40_context *nv40, - struct nv40_fragment_program *fp) -{ - if (fp->insn_len) - free(fp->insn); -} - diff --git a/src/mesa/pipe/nv40/nv40_fragtex.c b/src/mesa/pipe/nv40/nv40_fragtex.c deleted file mode 100644 index 5af5fbe7465..00000000000 --- a/src/mesa/pipe/nv40/nv40_fragtex.c +++ /dev/null @@ -1,151 +0,0 @@ -#include "nv40_context.h" - -#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ -{ \ - TRUE, \ - PIPE_FORMAT_##m, \ - NV40TCL_TEX_FORMAT_FORMAT_##tf, \ - (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x | NV40TCL_TEX_SWIZZLE_S0_Y_##ts0y | \ - NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w | \ - NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y | \ - NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ -} - -struct nv40_texture_format { - boolean defined; - uint pipe; - int format; - int swizzle; -}; - -static struct nv40_texture_format -nv40_texture_formats[] = { - _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), - _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), - _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), - _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), - _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), - _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), - _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), -// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), - {}, -}; - -static struct nv40_texture_format * -nv40_fragtex_format(uint pipe_format) -{ - struct nv40_texture_format *tf = nv40_texture_formats; - - while (tf->defined) { - if (tf->pipe == pipe_format) - return tf; - tf++; - } - - return NULL; -} - - -static void -nv40_fragtex_build(struct nv40_context *nv40, int unit) -{ - struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; - struct nv40_miptree *nv40mt = nv40->tex_miptree[unit]; - struct pipe_texture *pt = &nv40mt->base; - struct nv40_texture_format *tf; - struct nouveau_stateobj *so; - uint32_t txf, txs, txp; - int swizzled = 0; /*XXX: implement in region code? */ - unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; - - tf = nv40_fragtex_format(pt->format); - if (!tf) - assert(0); - - txf = ps->fmt; - txf |= tf->format | 0x8000; - txf |= ((pt->last_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT); - - if (1) /* XXX */ - txf |= NV40TCL_TEX_FORMAT_NO_BORDER; - - switch (pt->target) { - case PIPE_TEXTURE_CUBE: - txf |= NV40TCL_TEX_FORMAT_CUBIC; - /* fall-through */ - case PIPE_TEXTURE_2D: - txf |= NV40TCL_TEX_FORMAT_DIMS_2D; - break; - case PIPE_TEXTURE_3D: - txf |= NV40TCL_TEX_FORMAT_DIMS_3D; - break; - case PIPE_TEXTURE_1D: - txf |= NV40TCL_TEX_FORMAT_DIMS_1D; - break; - default: - NOUVEAU_ERR("Unknown target %d\n", pt->target); - return; - } - - if (swizzled) { - txp = 0; - } else { - txp = nv40mt->level[0].pitch; - txf |= NV40TCL_TEX_FORMAT_LINEAR; - } - - txs = tf->swizzle; - - so = so_new(16, 2); - so_method(so, nv40->hw->curie, NV40TCL_TEX_OFFSET(unit), 8); - so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); - so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, - NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); - so_data (so, ps->wrap); - so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en); - so_data (so, txs); - so_data (so, ps->filt | 0x2000 /*voodoo*/); - so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | - pt->height[0]); - so_data (so, ps->bcol); - so_method(so, nv40->hw->curie, NV40TCL_TEX_SIZE1(unit), 1); - so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); - - so_emit(nv40->nvws, so); - so_ref (so, &nv40->so_fragtex[unit]); - so_ref (NULL, &so); -} - -void -nv40_fragtex_bind(struct nv40_context *nv40) -{ - struct nv40_fragment_program *fp = nv40->fragprog.active; - unsigned samplers, unit; - - samplers = nv40->fp_samplers & ~fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - so_ref(NULL, &nv40->so_fragtex[unit]); - BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); - OUT_RING (0); - } - - samplers = nv40->dirty_samplers & fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - nv40_fragtex_build(nv40, unit); - } - - nv40->fp_samplers = fp->samplers; -} - diff --git a/src/mesa/pipe/nv40/nv40_miptree.c b/src/mesa/pipe/nv40/nv40_miptree.c deleted file mode 100644 index 92e6b3a43df..00000000000 --- a/src/mesa/pipe/nv40/nv40_miptree.c +++ /dev/null @@ -1,104 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_inlines.h" - -#include "nv40_context.h" - -static void -nv40_miptree_layout(struct nv40_miptree *nv40mt) -{ - struct pipe_texture *pt = &nv40mt->base; - boolean swizzled = FALSE; - uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; - uint offset = 0; - int nr_faces, l, f; - - if (pt->target == PIPE_TEXTURE_CUBE) { - nr_faces = 6; - } else - if (pt->target == PIPE_TEXTURE_3D) { - nr_faces = pt->depth[0]; - } else { - nr_faces = 1; - } - - for (l = 0; l <= pt->last_level; l++) { - pt->width[l] = width; - pt->height[l] = height; - pt->depth[l] = depth; - - if (swizzled) - nv40mt->level[l].pitch = pt->width[l] * pt->cpp; - else - nv40mt->level[l].pitch = pt->width[0] * pt->cpp; - nv40mt->level[l].pitch = (nv40mt->level[l].pitch + 63) & ~63; - - nv40mt->level[l].image_offset = - CALLOC(nr_faces, sizeof(unsigned)); - - width = MAX2(1, width >> 1); - height = MAX2(1, height >> 1); - depth = MAX2(1, depth >> 1); - - } - - for (f = 0; f < nr_faces; f++) { - for (l = 0; l <= pt->last_level; l++) { - nv40mt->level[l].image_offset[f] = offset; - offset += nv40mt->level[l].pitch * pt->height[l]; - } - } - - nv40mt->total_size = offset; -} - -static struct pipe_texture * -nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) -{ - struct pipe_winsys *ws = pipe->winsys; - struct nv40_miptree *mt; - - mt = MALLOC(sizeof(struct nv40_miptree)); - if (!mt) - return NULL; - mt->base = *pt; - nv40_miptree_layout(mt); - - mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - mt->total_size); - if (!mt->buffer) { - free(mt); - return NULL; - } - - return &mt->base; -} - -static void -nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) -{ - struct pipe_winsys *ws = pipe->winsys; - struct pipe_texture *mt = *pt; - - *pt = NULL; - if (--mt->refcount <= 0) { - struct nv40_miptree *nv40mt = (struct nv40_miptree *)mt; - int l; - - pipe_buffer_reference(ws, &nv40mt->buffer, NULL); - for (l = 0; l <= mt->last_level; l++) { - if (nv40mt->level[l].image_offset) - free(nv40mt->level[l].image_offset); - } - free(nv40mt); - } -} - -void -nv40_init_miptree_functions(struct nv40_context *nv40) -{ - nv40->pipe.texture_create = nv40_miptree_create; - nv40->pipe.texture_release = nv40_miptree_release; -} - diff --git a/src/mesa/pipe/nv40/nv40_query.c b/src/mesa/pipe/nv40/nv40_query.c deleted file mode 100644 index 8bca2788b93..00000000000 --- a/src/mesa/pipe/nv40/nv40_query.c +++ /dev/null @@ -1,113 +0,0 @@ -#include "pipe/p_context.h" - -#include "nv40_context.h" - -struct nv40_query { - struct nouveau_resource *object; - unsigned type; - boolean ready; - uint64_t result; -}; - -static INLINE struct nv40_query * -nv40_query(struct pipe_query *pipe) -{ - return (struct nv40_query *)pipe; -} - -static struct pipe_query * -nv40_query_create(struct pipe_context *pipe, unsigned query_type) -{ - struct nv40_query *q; - - q = CALLOC(1, sizeof(struct nv40_query)); - q->type = query_type; - - return (struct pipe_query *)q; -} - -static void -nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *q = nv40_query(pq); - - if (q->object) - nv40->nvws->res_free(&q->object); - free(q); -} - -static void -nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *q = nv40_query(pq); - - assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - - if (nv40->nvws->res_alloc(nv40->hw->query_heap, 1, NULL, &q->object)) - assert(0); - nv40->nvws->notifier_reset(nv40->hw->query, q->object->start); - - BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); - OUT_RING (1); - BEGIN_RING(curie, NV40TCL_QUERY_UNK17CC, 1); - OUT_RING (1); - - q->ready = FALSE; -} - -static void -nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *q = nv40_query(pq); - - BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); - OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | - ((q->object->start * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); - FIRE_RING(); -} - -static boolean -nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, - boolean wait, uint64 *result) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_query *q = nv40_query(pq); - struct nouveau_winsys *nvws = nv40->nvws; - - assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER); - - if (!q->ready) { - unsigned status; - - status = nvws->notifier_status(nv40->hw->query, - q->object->start); - if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { - if (wait == FALSE) - return FALSE; - nvws->notifier_wait(nv40->hw->query, q->object->start, - NV_NOTIFY_STATE_STATUS_COMPLETED, - 0); - } - - q->result = nvws->notifier_retval(nv40->hw->query, - q->object->start); - q->ready = TRUE; - nvws->res_free(&q->object); - } - - *result = q->result; - return TRUE; -} - -void -nv40_init_query_functions(struct nv40_context *nv40) -{ - nv40->pipe.create_query = nv40_query_create; - nv40->pipe.destroy_query = nv40_query_destroy; - nv40->pipe.begin_query = nv40_query_begin; - nv40->pipe.end_query = nv40_query_end; - nv40->pipe.get_query_result = nv40_query_result; -} diff --git a/src/mesa/pipe/nv40/nv40_shader.h b/src/mesa/pipe/nv40/nv40_shader.h deleted file mode 100644 index 5909c70713c..00000000000 --- a/src/mesa/pipe/nv40/nv40_shader.h +++ /dev/null @@ -1,554 +0,0 @@ -#ifndef __NV40_SHADER_H__ -#define __NV40_SHADER_H__ - -/* Vertex programs instruction set - * - * The NV40 instruction set is very similar to NV30. Most fields are in - * a slightly different position in the instruction however. - * - * Merged instructions - * In some cases it is possible to put two instructions into one opcode - * slot. The rules for when this is OK is not entirely clear to me yet. - * - * There are separate writemasks and dest temp register fields for each - * grouping of instructions. There is however only one field with the - * ID of a result register. Writing to temp/result regs is selected by - * setting VEC_RESULT/SCA_RESULT. - * - * Temporary registers - * The source/dest temp register fields have been extended by 1 bit, to - * give a total of 32 temporary registers. - * - * Relative Addressing - * NV40 can use an address register to index into vertex attribute regs. - * This is done by putting the offset value into INPUT_SRC and setting - * the INDEX_INPUT flag. - * - * Conditional execution (see NV_vertex_program{2,3} for details) - * There is a second condition code register on NV40, it's use is enabled - * by setting the COND_REG_SELECT_1 flag. - * - * Texture lookup - * TODO - */ - -/* ---- OPCODE BITS 127:96 / data DWORD 0 --- */ -#define NV40_VP_INST_VEC_RESULT (1 << 30) -/* uncertain.. */ -#define NV40_VP_INST_COND_UPDATE_ENABLE ((1 << 14)|1<<29) -/* use address reg as index into attribs */ -#define NV40_VP_INST_INDEX_INPUT (1 << 27) -#define NV40_VP_INST_COND_REG_SELECT_1 (1 << 25) -#define NV40_VP_INST_ADDR_REG_SELECT_1 (1 << 24) -#define NV40_VP_INST_SRC2_ABS (1 << 23) -#define NV40_VP_INST_SRC1_ABS (1 << 22) -#define NV40_VP_INST_SRC0_ABS (1 << 21) -#define NV40_VP_INST_VEC_DEST_TEMP_SHIFT 15 -#define NV40_VP_INST_VEC_DEST_TEMP_MASK (0x1F << 15) -#define NV40_VP_INST_COND_TEST_ENABLE (1 << 13) -#define NV40_VP_INST_COND_SHIFT 10 -#define NV40_VP_INST_COND_MASK (0x7 << 10) -# define NV40_VP_INST_COND_FL 0 -# define NV40_VP_INST_COND_LT 1 -# define NV40_VP_INST_COND_EQ 2 -# define NV40_VP_INST_COND_LE 3 -# define NV40_VP_INST_COND_GT 4 -# define NV40_VP_INST_COND_NE 5 -# define NV40_VP_INST_COND_GE 6 -# define NV40_VP_INST_COND_TR 7 -#define NV40_VP_INST_COND_SWZ_X_SHIFT 8 -#define NV40_VP_INST_COND_SWZ_X_MASK (3 << 8) -#define NV40_VP_INST_COND_SWZ_Y_SHIFT 6 -#define NV40_VP_INST_COND_SWZ_Y_MASK (3 << 6) -#define NV40_VP_INST_COND_SWZ_Z_SHIFT 4 -#define NV40_VP_INST_COND_SWZ_Z_MASK (3 << 4) -#define NV40_VP_INST_COND_SWZ_W_SHIFT 2 -#define NV40_VP_INST_COND_SWZ_W_MASK (3 << 2) -#define NV40_VP_INST_COND_SWZ_ALL_SHIFT 2 -#define NV40_VP_INST_COND_SWZ_ALL_MASK (0xFF << 2) -#define NV40_VP_INST_ADDR_SWZ_SHIFT 0 -#define NV40_VP_INST_ADDR_SWZ_MASK (0x03 << 0) -#define NV40_VP_INST0_KNOWN ( \ - NV40_VP_INST_INDEX_INPUT | \ - NV40_VP_INST_COND_REG_SELECT_1 | \ - NV40_VP_INST_ADDR_REG_SELECT_1 | \ - NV40_VP_INST_SRC2_ABS | \ - NV40_VP_INST_SRC1_ABS | \ - NV40_VP_INST_SRC0_ABS | \ - NV40_VP_INST_VEC_DEST_TEMP_MASK | \ - NV40_VP_INST_COND_TEST_ENABLE | \ - NV40_VP_INST_COND_MASK | \ - NV40_VP_INST_COND_SWZ_ALL_MASK | \ - NV40_VP_INST_ADDR_SWZ_MASK) - -/* ---- OPCODE BITS 95:64 / data DWORD 1 --- */ -#define NV40_VP_INST_VEC_OPCODE_SHIFT 22 -#define NV40_VP_INST_VEC_OPCODE_MASK (0x1F << 22) -# define NV40_VP_INST_OP_NOP 0x00 -# define NV40_VP_INST_OP_MOV 0x01 -# define NV40_VP_INST_OP_MUL 0x02 -# define NV40_VP_INST_OP_ADD 0x03 -# define NV40_VP_INST_OP_MAD 0x04 -# define NV40_VP_INST_OP_DP3 0x05 -# define NV40_VP_INST_OP_DPH 0x06 -# define NV40_VP_INST_OP_DP4 0x07 -# define NV40_VP_INST_OP_DST 0x08 -# define NV40_VP_INST_OP_MIN 0x09 -# define NV40_VP_INST_OP_MAX 0x0A -# define NV40_VP_INST_OP_SLT 0x0B -# define NV40_VP_INST_OP_SGE 0x0C -# define NV40_VP_INST_OP_ARL 0x0D -# define NV40_VP_INST_OP_FRC 0x0E -# define NV40_VP_INST_OP_FLR 0x0F -# define NV40_VP_INST_OP_SEQ 0x10 -# define NV40_VP_INST_OP_SFL 0x11 -# define NV40_VP_INST_OP_SGT 0x12 -# define NV40_VP_INST_OP_SLE 0x13 -# define NV40_VP_INST_OP_SNE 0x14 -# define NV40_VP_INST_OP_STR 0x15 -# define NV40_VP_INST_OP_SSG 0x16 -# define NV40_VP_INST_OP_ARR 0x17 -# define NV40_VP_INST_OP_ARA 0x18 -# define NV40_VP_INST_OP_TXL 0x19 -#define NV40_VP_INST_SCA_OPCODE_SHIFT 27 -#define NV40_VP_INST_SCA_OPCODE_MASK (0x1F << 27) -# define NV40_VP_INST_OP_NOP 0x00 -# define NV40_VP_INST_OP_MOV 0x01 -# define NV40_VP_INST_OP_RCP 0x02 -# define NV40_VP_INST_OP_RCC 0x03 -# define NV40_VP_INST_OP_RSQ 0x04 -# define NV40_VP_INST_OP_EXP 0x05 -# define NV40_VP_INST_OP_LOG 0x06 -# define NV40_VP_INST_OP_LIT 0x07 -# define NV40_VP_INST_OP_BRA 0x09 -# define NV40_VP_INST_OP_CAL 0x0B -# define NV40_VP_INST_OP_RET 0x0C -# define NV40_VP_INST_OP_LG2 0x0D -# define NV40_VP_INST_OP_EX2 0x0E -# define NV40_VP_INST_OP_SIN 0x0F -# define NV40_VP_INST_OP_COS 0x10 -# define NV40_VP_INST_OP_PUSHA 0x13 -# define NV40_VP_INST_OP_POPA 0x14 -#define NV40_VP_INST_CONST_SRC_SHIFT 12 -#define NV40_VP_INST_CONST_SRC_MASK (0xFF << 12) -#define NV40_VP_INST_INPUT_SRC_SHIFT 8 -#define NV40_VP_INST_INPUT_SRC_MASK (0x0F << 8) -# define NV40_VP_INST_IN_POS 0 -# define NV40_VP_INST_IN_WEIGHT 1 -# define NV40_VP_INST_IN_NORMAL 2 -# define NV40_VP_INST_IN_COL0 3 -# define NV40_VP_INST_IN_COL1 4 -# define NV40_VP_INST_IN_FOGC 5 -# define NV40_VP_INST_IN_TC0 8 -# define NV40_VP_INST_IN_TC(n) (8+n) -#define NV40_VP_INST_SRC0H_SHIFT 0 -#define NV40_VP_INST_SRC0H_MASK (0xFF << 0) -#define NV40_VP_INST1_KNOWN ( \ - NV40_VP_INST_VEC_OPCODE_MASK | \ - NV40_VP_INST_SCA_OPCODE_MASK | \ - NV40_VP_INST_CONST_SRC_MASK | \ - NV40_VP_INST_INPUT_SRC_MASK | \ - NV40_VP_INST_SRC0H_MASK \ - ) - -/* ---- OPCODE BITS 63:32 / data DWORD 2 --- */ -#define NV40_VP_INST_SRC0L_SHIFT 23 -#define NV40_VP_INST_SRC0L_MASK (0x1FF << 23) -#define NV40_VP_INST_SRC1_SHIFT 6 -#define NV40_VP_INST_SRC1_MASK (0x1FFFF << 6) -#define NV40_VP_INST_SRC2H_SHIFT 0 -#define NV40_VP_INST_SRC2H_MASK (0x3F << 0) -#define NV40_VP_INST_IADDRH_SHIFT 0 -#define NV40_VP_INST_IADDRH_MASK (0x1F << 0) - -/* ---- OPCODE BITS 31:0 / data DWORD 3 --- */ -#define NV40_VP_INST_IADDRL_SHIFT 29 -#define NV40_VP_INST_IADDRL_MASK (7 << 29) -#define NV40_VP_INST_SRC2L_SHIFT 21 -#define NV40_VP_INST_SRC2L_MASK (0x7FF << 21) -#define NV40_VP_INST_SCA_WRITEMASK_SHIFT 17 -#define NV40_VP_INST_SCA_WRITEMASK_MASK (0xF << 17) -# define NV40_VP_INST_SCA_WRITEMASK_X (1 << 20) -# define NV40_VP_INST_SCA_WRITEMASK_Y (1 << 19) -# define NV40_VP_INST_SCA_WRITEMASK_Z (1 << 18) -# define NV40_VP_INST_SCA_WRITEMASK_W (1 << 17) -#define NV40_VP_INST_VEC_WRITEMASK_SHIFT 13 -#define NV40_VP_INST_VEC_WRITEMASK_MASK (0xF << 13) -# define NV40_VP_INST_VEC_WRITEMASK_X (1 << 16) -# define NV40_VP_INST_VEC_WRITEMASK_Y (1 << 15) -# define NV40_VP_INST_VEC_WRITEMASK_Z (1 << 14) -# define NV40_VP_INST_VEC_WRITEMASK_W (1 << 13) -#define NV40_VP_INST_SCA_RESULT (1 << 12) -#define NV40_VP_INST_SCA_DEST_TEMP_SHIFT 7 -#define NV40_VP_INST_SCA_DEST_TEMP_MASK (0x1F << 7) -#define NV40_VP_INST_DEST_SHIFT 2 -#define NV40_VP_INST_DEST_MASK (31 << 2) -# define NV40_VP_INST_DEST_POS 0 -# define NV40_VP_INST_DEST_COL0 1 -# define NV40_VP_INST_DEST_COL1 2 -# define NV40_VP_INST_DEST_BFC0 3 -# define NV40_VP_INST_DEST_BFC1 4 -# define NV40_VP_INST_DEST_FOGC 5 -# define NV40_VP_INST_DEST_PSZ 6 -# define NV40_VP_INST_DEST_TC0 7 -# define NV40_VP_INST_DEST_TC(n) (7+n) -# define NV40_VP_INST_DEST_TEMP 0x1F -#define NV40_VP_INST_INDEX_CONST (1 << 1) -#define NV40_VP_INST_LAST (1 << 0) -#define NV40_VP_INST3_KNOWN ( \ - NV40_VP_INST_SRC2L_MASK |\ - NV40_VP_INST_SCA_WRITEMASK_MASK |\ - NV40_VP_INST_VEC_WRITEMASK_MASK |\ - NV40_VP_INST_SCA_DEST_TEMP_MASK |\ - NV40_VP_INST_DEST_MASK |\ - NV40_VP_INST_INDEX_CONST) - -/* Useful to split the source selection regs into their pieces */ -#define NV40_VP_SRC0_HIGH_SHIFT 9 -#define NV40_VP_SRC0_HIGH_MASK 0x0001FE00 -#define NV40_VP_SRC0_LOW_MASK 0x000001FF -#define NV40_VP_SRC2_HIGH_SHIFT 11 -#define NV40_VP_SRC2_HIGH_MASK 0x0001F800 -#define NV40_VP_SRC2_LOW_MASK 0x000007FF - -/* Source selection - these are the bits you fill NV40_VP_INST_SRCn with */ -#define NV40_VP_SRC_NEGATE (1 << 16) -#define NV40_VP_SRC_SWZ_X_SHIFT 14 -#define NV40_VP_SRC_SWZ_X_MASK (3 << 14) -#define NV40_VP_SRC_SWZ_Y_SHIFT 12 -#define NV40_VP_SRC_SWZ_Y_MASK (3 << 12) -#define NV40_VP_SRC_SWZ_Z_SHIFT 10 -#define NV40_VP_SRC_SWZ_Z_MASK (3 << 10) -#define NV40_VP_SRC_SWZ_W_SHIFT 8 -#define NV40_VP_SRC_SWZ_W_MASK (3 << 8) -#define NV40_VP_SRC_SWZ_ALL_SHIFT 8 -#define NV40_VP_SRC_SWZ_ALL_MASK (0xFF << 8) -#define NV40_VP_SRC_TEMP_SRC_SHIFT 2 -#define NV40_VP_SRC_TEMP_SRC_MASK (0x1F << 2) -#define NV40_VP_SRC_REG_TYPE_SHIFT 0 -#define NV40_VP_SRC_REG_TYPE_MASK (3 << 0) -# define NV40_VP_SRC_REG_TYPE_UNK0 0 -# define NV40_VP_SRC_REG_TYPE_TEMP 1 -# define NV40_VP_SRC_REG_TYPE_INPUT 2 -# define NV40_VP_SRC_REG_TYPE_CONST 3 - - -/* - * Each fragment program opcode appears to be comprised of 4 32-bit values. - * - * 0 - Opcode, output reg/mask, ATTRIB source - * 1 - Source 0 - * 2 - Source 1 - * 3 - Source 2 - * - * There appears to be no special difference between result regs and temp regs. - * result.color == R0.xyzw - * result.depth == R1.z - * When the fragprog contains instructions to write depth, - * NV30_TCL_PRIMITIVE_3D_UNK1D78=0 otherwise it is set to 1. - * - * Constants are inserted directly after the instruction that uses them. - * - * It appears that it's not possible to use two input registers in one - * instruction as the input sourcing is done in the instruction dword - * and not the source selection dwords. As such instructions such as: - * - * ADD result.color, fragment.color, fragment.texcoord[0]; - * - * must be split into two MOV's and then an ADD (nvidia does this) but - * I'm not sure why it's not just one MOV and then source the second input - * in the ADD instruction.. - * - * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary - * negation requires multiplication with a const. - * - * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO and - * SWIZZLE_ONE. - * - * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as - * SWIZZLE_ZERO is implemented simply by not writing to the relevant components - * of the destination. - * - * Looping - * Loops appear to be fairly expensive on NV40 at least, the proprietary - * driver goes to a lot of effort to avoid using the native looping - * instructions. If the total number of *executed* instructions between - * REP/ENDREP or LOOP/ENDLOOP is <=500, the driver will unroll the loop. - * The maximum loop count is 255. - * - * Conditional execution - * TODO - * - * Non-native instructions: - * LIT - * LRP - MAD+MAD - * SUB - ADD, negate second source - * RSQ - LG2 + EX2 - * POW - LG2 + MUL + EX2 - * SCS - COS + SIN - * XPD - * DP2 - MUL + ADD - * NRM - */ - -//== Opcode / Destination selection == -#define NV40_FP_OP_PROGRAM_END (1 << 0) -#define NV40_FP_OP_OUT_REG_SHIFT 1 -#define NV40_FP_OP_OUT_REG_MASK (63 << 1) -/* Needs to be set when writing outputs to get expected result.. */ -#define NV40_FP_OP_OUT_REG_HALF (1 << 7) -#define NV40_FP_OP_COND_WRITE_ENABLE (1 << 8) -#define NV40_FP_OP_OUTMASK_SHIFT 9 -#define NV40_FP_OP_OUTMASK_MASK (0xF << 9) -# define NV40_FP_OP_OUT_X (1 << 9) -# define NV40_FP_OP_OUT_Y (1 <<10) -# define NV40_FP_OP_OUT_Z (1 <<11) -# define NV40_FP_OP_OUT_W (1 <<12) -/* Uncertain about these, especially the input_src values.. it's possible that - * they can be dynamically changed. - */ -#define NV40_FP_OP_INPUT_SRC_SHIFT 13 -#define NV40_FP_OP_INPUT_SRC_MASK (15 << 13) -# define NV40_FP_OP_INPUT_SRC_POSITION 0x0 -# define NV40_FP_OP_INPUT_SRC_COL0 0x1 -# define NV40_FP_OP_INPUT_SRC_COL1 0x2 -# define NV40_FP_OP_INPUT_SRC_FOGC 0x3 -# define NV40_FP_OP_INPUT_SRC_TC0 0x4 -# define NV40_FP_OP_INPUT_SRC_TC(n) (0x4 + n) -# define NV40_FP_OP_INPUT_SRC_FACING 0xE -#define NV40_FP_OP_TEX_UNIT_SHIFT 17 -#define NV40_FP_OP_TEX_UNIT_MASK (0xF << 17) -#define NV40_FP_OP_PRECISION_SHIFT 22 -#define NV40_FP_OP_PRECISION_MASK (3 << 22) -# define NV40_FP_PRECISION_FP32 0 -# define NV40_FP_PRECISION_FP16 1 -# define NV40_FP_PRECISION_FX12 2 -#define NV40_FP_OP_OPCODE_SHIFT 24 -#define NV40_FP_OP_OPCODE_MASK (0x3F << 24) -# define NV40_FP_OP_OPCODE_NOP 0x00 -# define NV40_FP_OP_OPCODE_MOV 0x01 -# define NV40_FP_OP_OPCODE_MUL 0x02 -# define NV40_FP_OP_OPCODE_ADD 0x03 -# define NV40_FP_OP_OPCODE_MAD 0x04 -# define NV40_FP_OP_OPCODE_DP3 0x05 -# define NV40_FP_OP_OPCODE_DP4 0x06 -# define NV40_FP_OP_OPCODE_DST 0x07 -# define NV40_FP_OP_OPCODE_MIN 0x08 -# define NV40_FP_OP_OPCODE_MAX 0x09 -# define NV40_FP_OP_OPCODE_SLT 0x0A -# define NV40_FP_OP_OPCODE_SGE 0x0B -# define NV40_FP_OP_OPCODE_SLE 0x0C -# define NV40_FP_OP_OPCODE_SGT 0x0D -# define NV40_FP_OP_OPCODE_SNE 0x0E -# define NV40_FP_OP_OPCODE_SEQ 0x0F -# define NV40_FP_OP_OPCODE_FRC 0x10 -# define NV40_FP_OP_OPCODE_FLR 0x11 -# define NV40_FP_OP_OPCODE_KIL 0x12 -# define NV40_FP_OP_OPCODE_PK4B 0x13 -# define NV40_FP_OP_OPCODE_UP4B 0x14 -/* DDX/DDY can only write to XY */ -# define NV40_FP_OP_OPCODE_DDX 0x15 -# define NV40_FP_OP_OPCODE_DDY 0x16 -# define NV40_FP_OP_OPCODE_TEX 0x17 -# define NV40_FP_OP_OPCODE_TXP 0x18 -# define NV40_FP_OP_OPCODE_TXD 0x19 -# define NV40_FP_OP_OPCODE_RCP 0x1A -# define NV40_FP_OP_OPCODE_EX2 0x1C -# define NV40_FP_OP_OPCODE_LG2 0x1D -# define NV40_FP_OP_OPCODE_STR 0x20 -# define NV40_FP_OP_OPCODE_SFL 0x21 -# define NV40_FP_OP_OPCODE_COS 0x22 -# define NV40_FP_OP_OPCODE_SIN 0x23 -# define NV40_FP_OP_OPCODE_PK2H 0x24 -# define NV40_FP_OP_OPCODE_UP2H 0x25 -# define NV40_FP_OP_OPCODE_PK4UB 0x27 -# define NV40_FP_OP_OPCODE_UP4UB 0x28 -# define NV40_FP_OP_OPCODE_PK2US 0x29 -# define NV40_FP_OP_OPCODE_UP2US 0x2A -# define NV40_FP_OP_OPCODE_DP2A 0x2E -# define NV40_FP_OP_OPCODE_TXL 0x2F -# define NV40_FP_OP_OPCODE_TXB 0x31 -# define NV40_FP_OP_OPCODE_DIV 0x3A -# define NV40_FP_OP_OPCODE_UNK_LIT 0x3C -/* The use of these instructions appears to be indicated by bit 31 of DWORD 2.*/ -# define NV40_FP_OP_BRA_OPCODE_BRK 0x0 -# define NV40_FP_OP_BRA_OPCODE_CAL 0x1 -# define NV40_FP_OP_BRA_OPCODE_IF 0x2 -# define NV40_FP_OP_BRA_OPCODE_LOOP 0x3 -# define NV40_FP_OP_BRA_OPCODE_REP 0x4 -# define NV40_FP_OP_BRA_OPCODE_RET 0x5 -#define NV40_FP_OP_OUT_SAT (1 << 31) - -/* high order bits of SRC0 */ -#define NV40_FP_OP_OUT_ABS (1 << 29) -#define NV40_FP_OP_COND_SWZ_W_SHIFT 27 -#define NV40_FP_OP_COND_SWZ_W_MASK (3 << 27) -#define NV40_FP_OP_COND_SWZ_Z_SHIFT 25 -#define NV40_FP_OP_COND_SWZ_Z_MASK (3 << 25) -#define NV40_FP_OP_COND_SWZ_Y_SHIFT 23 -#define NV40_FP_OP_COND_SWZ_Y_MASK (3 << 23) -#define NV40_FP_OP_COND_SWZ_X_SHIFT 21 -#define NV40_FP_OP_COND_SWZ_X_MASK (3 << 21) -#define NV40_FP_OP_COND_SWZ_ALL_SHIFT 21 -#define NV40_FP_OP_COND_SWZ_ALL_MASK (0xFF << 21) -#define NV40_FP_OP_COND_SHIFT 18 -#define NV40_FP_OP_COND_MASK (0x07 << 18) -# define NV40_FP_OP_COND_FL 0 -# define NV40_FP_OP_COND_LT 1 -# define NV40_FP_OP_COND_EQ 2 -# define NV40_FP_OP_COND_LE 3 -# define NV40_FP_OP_COND_GT 4 -# define NV40_FP_OP_COND_NE 5 -# define NV40_FP_OP_COND_GE 6 -# define NV40_FP_OP_COND_TR 7 - -/* high order bits of SRC1 */ -#define NV40_FP_OP_OPCODE_IS_BRANCH (1<<31) -#define NV40_FP_OP_DST_SCALE_SHIFT 28 -#define NV40_FP_OP_DST_SCALE_MASK (3 << 28) -#define NV40_FP_OP_DST_SCALE_1X 0 -#define NV40_FP_OP_DST_SCALE_2X 1 -#define NV40_FP_OP_DST_SCALE_4X 2 -#define NV40_FP_OP_DST_SCALE_8X 3 -#define NV40_FP_OP_DST_SCALE_INV_2X 5 -#define NV40_FP_OP_DST_SCALE_INV_4X 6 -#define NV40_FP_OP_DST_SCALE_INV_8X 7 - -/* SRC1 LOOP */ -#define NV40_FP_OP_LOOP_INCR_SHIFT 19 -#define NV40_FP_OP_LOOP_INCR_MASK (0xFF << 19) -#define NV40_FP_OP_LOOP_INDEX_SHIFT 10 -#define NV40_FP_OP_LOOP_INDEX_MASK (0xFF << 10) -#define NV40_FP_OP_LOOP_COUNT_SHIFT 2 -#define NV40_FP_OP_LOOP_COUNT_MASK (0xFF << 2) - -/* SRC1 IF */ -#define NV40_FP_OP_ELSE_ID_SHIFT 2 -#define NV40_FP_OP_ELSE_ID_MASK (0xFF << 2) - -/* SRC1 CAL */ -#define NV40_FP_OP_IADDR_SHIFT 2 -#define NV40_FP_OP_IADDR_MASK (0xFF << 2) - -/* SRC1 REP - * I have no idea why there are 3 count values here.. but they - * have always been filled with the same value in my tests so - * far.. - */ -#define NV40_FP_OP_REP_COUNT1_SHIFT 2 -#define NV40_FP_OP_REP_COUNT1_MASK (0xFF << 2) -#define NV40_FP_OP_REP_COUNT2_SHIFT 10 -#define NV40_FP_OP_REP_COUNT2_MASK (0xFF << 10) -#define NV40_FP_OP_REP_COUNT3_SHIFT 19 -#define NV40_FP_OP_REP_COUNT3_MASK (0xFF << 19) - -/* SRC2 REP/IF */ -#define NV40_FP_OP_END_ID_SHIFT 2 -#define NV40_FP_OP_END_ID_MASK (0xFF << 2) - -// SRC2 high-order -#define NV40_FP_OP_INDEX_INPUT (1 << 30) -#define NV40_FP_OP_ADDR_INDEX_SHIFT 19 -#define NV40_FP_OP_ADDR_INDEX_MASK (0xF << 19) - -//== Register selection == -#define NV40_FP_REG_TYPE_SHIFT 0 -#define NV40_FP_REG_TYPE_MASK (3 << 0) -# define NV40_FP_REG_TYPE_TEMP 0 -# define NV40_FP_REG_TYPE_INPUT 1 -# define NV40_FP_REG_TYPE_CONST 2 -#define NV40_FP_REG_SRC_SHIFT 2 -#define NV40_FP_REG_SRC_MASK (63 << 2) -#define NV40_FP_REG_SRC_HALF (1 << 8) -#define NV40_FP_REG_SWZ_ALL_SHIFT 9 -#define NV40_FP_REG_SWZ_ALL_MASK (255 << 9) -#define NV40_FP_REG_SWZ_X_SHIFT 9 -#define NV40_FP_REG_SWZ_X_MASK (3 << 9) -#define NV40_FP_REG_SWZ_Y_SHIFT 11 -#define NV40_FP_REG_SWZ_Y_MASK (3 << 11) -#define NV40_FP_REG_SWZ_Z_SHIFT 13 -#define NV40_FP_REG_SWZ_Z_MASK (3 << 13) -#define NV40_FP_REG_SWZ_W_SHIFT 15 -#define NV40_FP_REG_SWZ_W_MASK (3 << 15) -# define NV40_FP_SWIZZLE_X 0 -# define NV40_FP_SWIZZLE_Y 1 -# define NV40_FP_SWIZZLE_Z 2 -# define NV40_FP_SWIZZLE_W 3 -#define NV40_FP_REG_NEGATE (1 << 17) - -#define NV40SR_NONE 0 -#define NV40SR_OUTPUT 1 -#define NV40SR_INPUT 2 -#define NV40SR_TEMP 3 -#define NV40SR_CONST 4 - -struct nv40_sreg { - int type; - int index; - - int dst_scale; - - int negate; - int abs; - int swz[4]; - - int cc_update; - int cc_update_reg; - int cc_test; - int cc_test_reg; - int cc_swz[4]; -}; - -static INLINE struct nv40_sreg -nv40_sr(int type, int index) -{ - struct nv40_sreg temp = { - .type = type, - .index = index, - .dst_scale = DEF_SCALE, - .abs = 0, - .negate = 0, - .swz = { 0, 1, 2, 3 }, - .cc_update = 0, - .cc_update_reg = 0, - .cc_test = DEF_CTEST, - .cc_test_reg = 0, - .cc_swz = { 0, 1, 2, 3 }, - }; - return temp; -} - -static INLINE struct nv40_sreg -nv40_sr_swz(struct nv40_sreg src, int x, int y, int z, int w) -{ - struct nv40_sreg dst = src; - - dst.swz[SWZ_X] = src.swz[x]; - dst.swz[SWZ_Y] = src.swz[y]; - dst.swz[SWZ_Z] = src.swz[z]; - dst.swz[SWZ_W] = src.swz[w]; - return dst; -} - -static INLINE struct nv40_sreg -nv40_sr_neg(struct nv40_sreg src) -{ - src.negate = !src.negate; - return src; -} - -static INLINE struct nv40_sreg -nv40_sr_abs(struct nv40_sreg src) -{ - src.abs = 1; - return src; -} - -static INLINE struct nv40_sreg -nv40_sr_scale(struct nv40_sreg src, int scale) -{ - src.dst_scale = scale; - return src; -} - -#endif diff --git a/src/mesa/pipe/nv40/nv40_state.c b/src/mesa/pipe/nv40/nv40_state.c deleted file mode 100644 index 713f31dbb12..00000000000 --- a/src/mesa/pipe/nv40/nv40_state.c +++ /dev/null @@ -1,823 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" - -#include "nv40_context.h" -#include "nv40_state.h" - -static void * -nv40_blend_state_create(struct pipe_context *pipe, - const struct pipe_blend_state *cso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_grobj *curie = nv40->hw->curie; - struct nouveau_stateobj *so = so_new(16, 0); - - if (cso->blend_enable) { - so_method(so, curie, NV40TCL_BLEND_ENABLE, 3); - so_data (so, 1); - so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | - nvgl_blend_func(cso->rgb_src_factor)); - so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | - nvgl_blend_func(cso->rgb_dst_factor)); - so_method(so, curie, NV40TCL_BLEND_EQUATION, 1); - so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | - nvgl_blend_eqn(cso->rgb_func)); - } else { - so_method(so, curie, NV40TCL_BLEND_ENABLE, 1); - so_data (so, 0); - } - - so_method(so, curie, NV40TCL_COLOR_MASK, 1); - so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | - ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | - ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | - ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); - - if (cso->logicop_enable) { - so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2); - so_data (so, 1); - so_data (so, nvgl_logicop_func(cso->logicop_func)); - } else { - so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1); - so_data (so, 0); - } - - so_method(so, curie, NV40TCL_DITHER_ENABLE, 1); - so_data (so, cso->dither ? 1 : 0); - - return (void *)so; -} - -static void -nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - so_ref(hwcso, &nv40->so_blend); - nv40->dirty |= NV40_NEW_BLEND; -} - -static void -nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nouveau_stateobj *so = hwcso; - - so_ref(NULL, &so); -} - - -static INLINE unsigned -wrap_mode(unsigned wrap) { - unsigned ret; - - switch (wrap) { - case PIPE_TEX_WRAP_REPEAT: - ret = NV40TCL_TEX_WRAP_S_REPEAT; - break; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - ret = NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT; - break; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE; - break; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER; - break; - case PIPE_TEX_WRAP_CLAMP: - ret = NV40TCL_TEX_WRAP_S_CLAMP; - break; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE; - break; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER; - break; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP; - break; - default: - NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); - ret = NV40TCL_TEX_WRAP_S_REPEAT; - break; - } - - return ret >> NV40TCL_TEX_WRAP_S_SHIFT; -} - -static void * -nv40_sampler_state_create(struct pipe_context *pipe, - const struct pipe_sampler_state *cso) -{ - struct nv40_sampler_state *ps; - uint32_t filter = 0; - - ps = MALLOC(sizeof(struct nv40_sampler_state)); - - ps->fmt = 0; - if (!cso->normalized_coords) - ps->fmt |= NV40TCL_TEX_FORMAT_RECT; - - ps->wrap = ((wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_S_SHIFT) | - (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) | - (wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_R_SHIFT)); - - ps->en = 0; - if (cso->max_anisotropy >= 2.0) { - /* no idea, binary driver sets it, works without it.. meh.. */ - ps->wrap |= (1 << 5); - - if (cso->max_anisotropy >= 16.0) { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_16X; - } else - if (cso->max_anisotropy >= 12.0) { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_12X; - } else - if (cso->max_anisotropy >= 10.0) { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_10X; - } else - if (cso->max_anisotropy >= 8.0) { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_8X; - } else - if (cso->max_anisotropy >= 6.0) { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_6X; - } else - if (cso->max_anisotropy >= 4.0) { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_4X; - } else { - ps->en |= NV40TCL_TEX_ENABLE_ANISO_2X; - } - } - - switch (cso->mag_img_filter) { - case PIPE_TEX_FILTER_LINEAR: - filter |= NV40TCL_TEX_FILTER_MAG_LINEAR; - break; - case PIPE_TEX_FILTER_NEAREST: - default: - filter |= NV40TCL_TEX_FILTER_MAG_NEAREST; - break; - } - - switch (cso->min_img_filter) { - case PIPE_TEX_FILTER_LINEAR: - switch (cso->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NEAREST: - filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST; - break; - case PIPE_TEX_MIPFILTER_LINEAR: - filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR; - break; - case PIPE_TEX_MIPFILTER_NONE: - default: - filter |= NV40TCL_TEX_FILTER_MIN_LINEAR; - break; - } - break; - case PIPE_TEX_FILTER_NEAREST: - default: - switch (cso->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NEAREST: - filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST; - break; - case PIPE_TEX_MIPFILTER_LINEAR: - filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR; - break; - case PIPE_TEX_MIPFILTER_NONE: - default: - filter |= NV40TCL_TEX_FILTER_MIN_NEAREST; - break; - } - break; - } - - ps->filt = filter; - - { - float limit; - - limit = CLAMP(cso->lod_bias, -16.0, 15.0); - ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff; - - limit = CLAMP(cso->max_lod, 0.0, 15.0); - ps->en |= (int)(limit * 256.0) << 7; - - limit = CLAMP(cso->min_lod, 0.0, 15.0); - ps->en |= (int)(limit * 256.0) << 19; - } - - - if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - switch (cso->compare_func) { - case PIPE_FUNC_NEVER: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NEVER; - break; - case PIPE_FUNC_GREATER: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GREATER; - break; - case PIPE_FUNC_EQUAL: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_EQUAL; - break; - case PIPE_FUNC_GEQUAL: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GEQUAL; - break; - case PIPE_FUNC_LESS: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LESS; - break; - case PIPE_FUNC_NOTEQUAL: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL; - break; - case PIPE_FUNC_LEQUAL: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LEQUAL; - break; - case PIPE_FUNC_ALWAYS: - ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_ALWAYS; - break; - default: - break; - } - } - - ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | - (float_to_ubyte(cso->border_color[0]) << 16) | - (float_to_ubyte(cso->border_color[1]) << 8) | - (float_to_ubyte(cso->border_color[2]) << 0)); - - return (void *)ps; -} - -static void -nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, - void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_sampler_state *ps = hwcso; - - nv40->tex_sampler[unit] = ps; - nv40->dirty_samplers |= (1 << unit); -} - -static void -nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso) -{ - free(hwcso); -} - -static void -nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *miptree) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - nv40->tex_miptree[unit] = (struct nv40_miptree *)miptree; - nv40->dirty_samplers |= (1 << unit); -} - -static void * -nv40_rasterizer_state_create(struct pipe_context *pipe, - const struct pipe_rasterizer_state *cso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(32, 0); - - /*XXX: ignored: - * light_twoside - * offset_cw/ccw -nohw - * scissor - * point_smooth -nohw - * multisample - * offset_units / offset_scale - */ - - so_method(so, nv40->hw->curie, NV40TCL_SHADE_MODEL, 1); - so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : - NV40TCL_SHADE_MODEL_SMOOTH); - - so_method(so, nv40->hw->curie, NV40TCL_LINE_WIDTH, 2); - so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); - so_data (so, cso->line_smooth ? 1 : 0); - so_method(so, nv40->hw->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); - so_data (so, cso->line_stipple_enable ? 1 : 0); - so_data (so, (cso->line_stipple_pattern << 16) | - cso->line_stipple_factor); - - so_method(so, nv40->hw->curie, NV40TCL_POINT_SIZE, 1); - so_data (so, fui(cso->point_size)); - - so_method(so, nv40->hw->curie, NV40TCL_POLYGON_MODE_FRONT, 6); - if (cso->front_winding == PIPE_WINDING_CCW) { - so_data(so, nvgl_polygon_mode(cso->fill_ccw)); - so_data(so, nvgl_polygon_mode(cso->fill_cw)); - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - so_data(so, NV40TCL_CULL_FACE_FRONT); - break; - case PIPE_WINDING_CW: - so_data(so, NV40TCL_CULL_FACE_BACK); - break; - case PIPE_WINDING_BOTH: - so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); - break; - default: - so_data(so, 0); - break; - } - so_data(so, NV40TCL_FRONT_FACE_CCW); - } else { - so_data(so, nvgl_polygon_mode(cso->fill_cw)); - so_data(so, nvgl_polygon_mode(cso->fill_ccw)); - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - so_data(so, NV40TCL_CULL_FACE_BACK); - break; - case PIPE_WINDING_CW: - so_data(so, NV40TCL_CULL_FACE_FRONT); - break; - case PIPE_WINDING_BOTH: - so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); - break; - default: - so_data(so, 0); - break; - } - so_data(so, NV40TCL_FRONT_FACE_CW); - } - so_data(so, cso->poly_smooth ? 1 : 0); - so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); - - so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); - so_data (so, cso->poly_stipple_enable ? 1 : 0); - - so_method(so, nv40->hw->curie, NV40TCL_POINT_SPRITE, 1); - if (cso->point_sprite) { - unsigned psctl = (1 << 0), i; - - for (i = 0; i < 8; i++) { - if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) - psctl |= (1 << (8 + i)); - } - - so_data(so, psctl); - } else { - so_data(so, 0); - } - - return (void *)so; -} - -static void -nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - so_ref(hwcso, &nv40->so_rast); - nv40->dirty |= NV40_NEW_RAST; -} - -static void -nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nouveau_stateobj *so = hwcso; - - so_ref(NULL, &so); -} - -static void * -nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, - const struct pipe_depth_stencil_alpha_state *cso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(32, 0); - - so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); - so_data (so, nvgl_comparison_op(cso->depth.func)); - so_data (so, cso->depth.writemask ? 1 : 0); - so_data (so, cso->depth.enabled ? 1 : 0); - - so_method(so, nv40->hw->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); - so_data (so, cso->alpha.enabled ? 1 : 0); - so_data (so, nvgl_comparison_op(cso->alpha.func)); - so_data (so, float_to_ubyte(cso->alpha.ref)); - - if (cso->stencil[0].enabled) { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); - so_data (so, cso->stencil[0].enabled ? 1 : 0); - so_data (so, cso->stencil[0].write_mask); - so_data (so, nvgl_comparison_op(cso->stencil[0].func)); - so_data (so, cso->stencil[0].ref_value); - so_data (so, cso->stencil[0].value_mask); - so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); - so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); - so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); - } else { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); - so_data (so, 0); - } - - if (cso->stencil[1].enabled) { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); - so_data (so, cso->stencil[1].enabled ? 1 : 0); - so_data (so, cso->stencil[1].write_mask); - so_data (so, nvgl_comparison_op(cso->stencil[1].func)); - so_data (so, cso->stencil[1].ref_value); - so_data (so, cso->stencil[1].value_mask); - so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); - so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); - so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); - } else { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); - so_data (so, 0); - } - - return (void *)so; -} - -static void -nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - so_ref(hwcso, &nv40->so_zsa); - nv40->dirty |= NV40_NEW_ZSA; -} - -static void -nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nouveau_stateobj *so = hwcso; - - so_ref(NULL, &so); -} - -static void * -nv40_vp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - struct nv40_vertex_program *vp; - - vp = CALLOC(1, sizeof(struct nv40_vertex_program)); - vp->pipe = cso; - - return (void *)vp; -} - -static void -nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_vertex_program *vp = hwcso; - - nv40->vertprog.current = vp; - nv40->dirty |= NV40_NEW_VERTPROG; -} - -static void -nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_vertex_program *vp = hwcso; - - nv40_vertprog_destroy(nv40, vp); - free(vp); -} - -static void * -nv40_fp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - struct nv40_fragment_program *fp; - - fp = CALLOC(1, sizeof(struct nv40_fragment_program)); - fp->pipe = cso; - - return (void *)fp; -} - -static void -nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_fragment_program *fp = hwcso; - - nv40->fragprog.current = fp; - nv40->dirty |= NV40_NEW_FRAGPROG; -} - -static void -nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_fragment_program *fp = hwcso; - - nv40_fragprog_destroy(nv40, fp); - free(fp); -} - -static void -nv40_set_blend_color(struct pipe_context *pipe, - const struct pipe_blend_color *bcol) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(2, 0); - - so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); - so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0))); - - so_ref(so, &nv40->so_bcol); - so_ref(NULL, &so); - nv40->dirty |= NV40_NEW_BCOL; -} - -static void -nv40_set_clip_state(struct pipe_context *pipe, - const struct pipe_clip_state *clip) -{ -} - -static void -nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, - const struct pipe_constant_buffer *buf ) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - if (shader == PIPE_SHADER_VERTEX) { - nv40->vertprog.constant_buf = buf->buffer; - nv40->dirty |= NV40_NEW_VERTPROG; - } else - if (shader == PIPE_SHADER_FRAGMENT) { - nv40->fragprog.constant_buf = buf->buffer; - nv40->dirty |= NV40_NEW_FRAGPROG; - } -} - -static void -nv40_set_framebuffer_state(struct pipe_context *pipe, - const struct pipe_framebuffer_state *fb) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct pipe_surface *rt[4], *zeta; - uint32_t rt_enable, rt_format, w, h; - int i, colour_format = 0, zeta_format = 0; - struct nouveau_stateobj *so = so_new(64, 10); - unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; - - rt_enable = 0; - for (i = 0; i < 4; i++) { - if (!fb->cbufs[i]) - continue; - - if (colour_format) { - assert(w == fb->cbufs[i]->width); - assert(h == fb->cbufs[i]->height); - assert(colour_format == fb->cbufs[i]->format); - } else { - w = fb->cbufs[i]->width; - h = fb->cbufs[i]->height; - colour_format = fb->cbufs[i]->format; - rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); - rt[i] = fb->cbufs[i]; - } - } - - if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | - NV40TCL_RT_ENABLE_COLOR3)) - rt_enable |= NV40TCL_RT_ENABLE_MRT; - - if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - - zeta_format = fb->zsbuf->format; - zeta = fb->zsbuf; - } - - rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; - - switch (colour_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case 0: - rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; - break; - case PIPE_FORMAT_R5G6B5_UNORM: - rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; - break; - default: - assert(0); - } - - switch (zeta_format) { - case PIPE_FORMAT_Z16_UNORM: - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; - break; - case PIPE_FORMAT_Z24S8_UNORM: - case 0: - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; - break; - default: - assert(0); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); - so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); - so_data (so, rt[0]->pitch * rt[0]->cpp); - so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); - so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); - so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_data (so, rt[1]->pitch * rt[1]->cpp); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); - so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); - so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); - so_data (so, rt[2]->pitch * rt[2]->cpp); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); - so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); - so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); - so_data (so, rt[3]->pitch * rt[3]->cpp); - } - - if (zeta_format) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); - so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); - so_reloc (so, zeta->buffer, zeta->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); - so_data (so, zeta->pitch * zeta->cpp); - } - - so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); - so_data (so, rt_enable); - so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); - so_data (so, (w << 16) | 0); - so_data (so, (h << 16) | 0); - so_data (so, rt_format); - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); - so_data (so, (w << 16) | 0); - so_data (so, (h << 16) | 0); - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); - so_data (so, ((w - 1) << 16) | 0); - so_data (so, ((h - 1) << 16) | 0); - - so_ref(so, &nv40->so_framebuffer); - so_ref(NULL, &so); - nv40->dirty |= NV40_NEW_FB; -} - -static void -nv40_set_polygon_stipple(struct pipe_context *pipe, - const struct pipe_poly_stipple *stipple) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(33, 0); - unsigned i; - - so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); - for (i = 0; i < 32; i++) - so_data(so, stipple->stipple[i]); - - so_ref(so, &nv40->so_stipple); - so_ref(NULL, &so); - nv40->dirty |= NV40_NEW_STIPPLE; -} - -static void -nv40_set_scissor_state(struct pipe_context *pipe, - const struct pipe_scissor_state *s) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(3, 0); - - so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); - so_data (so, ((s->maxx - s->minx) << 16) | s->minx); - so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - - so_ref(so, &nv40->so_scissor); - so_ref(NULL, &so); - nv40->dirty |= NV40_NEW_SCISSOR; -} - -static void -nv40_set_viewport_state(struct pipe_context *pipe, - const struct pipe_viewport_state *vpt) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(9, 0); - - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); - so_data (so, fui(vpt->translate[0])); - so_data (so, fui(vpt->translate[1])); - so_data (so, fui(vpt->translate[2])); - so_data (so, fui(vpt->translate[3])); - so_data (so, fui(vpt->scale[0])); - so_data (so, fui(vpt->scale[1])); - so_data (so, fui(vpt->scale[2])); - so_data (so, fui(vpt->scale[3])); - - so_ref(so, &nv40->so_viewport); - so_ref(NULL, &so); - nv40->dirty |= NV40_NEW_VIEWPORT; -} - -static void -nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - nv40->vtxbuf[index] = *vb; - - nv40->dirty |= NV40_NEW_ARRAYS; -} - -static void -nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - nv40->vtxelt[index] = *ve; - - nv40->dirty |= NV40_NEW_ARRAYS; -} - -void -nv40_init_state_functions(struct nv40_context *nv40) -{ - nv40->pipe.create_blend_state = nv40_blend_state_create; - nv40->pipe.bind_blend_state = nv40_blend_state_bind; - nv40->pipe.delete_blend_state = nv40_blend_state_delete; - - nv40->pipe.create_sampler_state = nv40_sampler_state_create; - nv40->pipe.bind_sampler_state = nv40_sampler_state_bind; - nv40->pipe.delete_sampler_state = nv40_sampler_state_delete; - nv40->pipe.set_sampler_texture = nv40_set_sampler_texture; - - nv40->pipe.create_rasterizer_state = nv40_rasterizer_state_create; - nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind; - nv40->pipe.delete_rasterizer_state = nv40_rasterizer_state_delete; - - nv40->pipe.create_depth_stencil_alpha_state = - nv40_depth_stencil_alpha_state_create; - nv40->pipe.bind_depth_stencil_alpha_state = - nv40_depth_stencil_alpha_state_bind; - nv40->pipe.delete_depth_stencil_alpha_state = - nv40_depth_stencil_alpha_state_delete; - - nv40->pipe.create_vs_state = nv40_vp_state_create; - nv40->pipe.bind_vs_state = nv40_vp_state_bind; - nv40->pipe.delete_vs_state = nv40_vp_state_delete; - - nv40->pipe.create_fs_state = nv40_fp_state_create; - nv40->pipe.bind_fs_state = nv40_fp_state_bind; - nv40->pipe.delete_fs_state = nv40_fp_state_delete; - - nv40->pipe.set_blend_color = nv40_set_blend_color; - nv40->pipe.set_clip_state = nv40_set_clip_state; - nv40->pipe.set_constant_buffer = nv40_set_constant_buffer; - nv40->pipe.set_framebuffer_state = nv40_set_framebuffer_state; - nv40->pipe.set_polygon_stipple = nv40_set_polygon_stipple; - nv40->pipe.set_scissor_state = nv40_set_scissor_state; - nv40->pipe.set_viewport_state = nv40_set_viewport_state; - - nv40->pipe.set_vertex_buffer = nv40_set_vertex_buffer; - nv40->pipe.set_vertex_element = nv40_set_vertex_element; -} - diff --git a/src/mesa/pipe/nv40/nv40_state.h b/src/mesa/pipe/nv40/nv40_state.h deleted file mode 100644 index e82ab9de98a..00000000000 --- a/src/mesa/pipe/nv40/nv40_state.h +++ /dev/null @@ -1,80 +0,0 @@ -#ifndef __NV40_STATE_H__ -#define __NV40_STATE_H__ - -#include "pipe/p_state.h" - -struct nv40_sampler_state { - uint32_t fmt; - uint32_t wrap; - uint32_t en; - uint32_t filt; - uint32_t bcol; -}; - -struct nv40_vertex_program_exec { - uint32_t data[4]; - boolean has_branch_offset; - int const_index; -}; - -struct nv40_vertex_program_data { - int index; /* immediates == -1 */ - float value[4]; -}; - -struct nv40_vertex_program { - const struct pipe_shader_state *pipe; - - boolean translated; - struct nv40_vertex_program_exec *insns; - unsigned nr_insns; - struct nv40_vertex_program_data *consts; - unsigned nr_consts; - - struct nouveau_resource *exec; - unsigned exec_start; - struct nouveau_resource *data; - unsigned data_start; - unsigned data_start_min; - - uint32_t ir; - uint32_t or; -}; - -struct nv40_fragment_program_data { - unsigned offset; - unsigned index; -}; - -struct nv40_fragment_program { - const struct pipe_shader_state *pipe; - - boolean translated; - boolean on_hw; - unsigned samplers; - - uint32_t *insn; - int insn_len; - - struct nv40_fragment_program_data *consts; - unsigned nr_consts; - - struct pipe_buffer *buffer; - - uint32_t fp_control; - struct nouveau_stateobj *so; -}; - -struct nv40_miptree { - struct pipe_texture base; - - struct pipe_buffer *buffer; - uint total_size; - - struct { - uint pitch; - uint *image_offset; - } level[PIPE_MAX_TEXTURE_LEVELS]; -}; - -#endif diff --git a/src/mesa/pipe/nv40/nv40_state_emit.c b/src/mesa/pipe/nv40/nv40_state_emit.c deleted file mode 100644 index a10c9955480..00000000000 --- a/src/mesa/pipe/nv40/nv40_state_emit.c +++ /dev/null @@ -1,77 +0,0 @@ -#include "nv40_context.h" -#include "nv40_state.h" - -/* Emit relocs for every referenced buffer. - * - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. These relocs appear in the push buffer as - * NOPs, and will only be turned into state changes if a buffer - * actually moves. - */ -static void -nv40_state_emit_dummy_relocs(struct nv40_context *nv40) -{ - unsigned i; - - so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); - for (i = 0; i < 16; i++) { - if (!(nv40->fp_samplers & (1 << i))) - continue; - so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); - } - so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); -} - -void -nv40_emit_hw_state(struct nv40_context *nv40) -{ - if (nv40->dirty & NV40_NEW_FB) - so_emit(nv40->nvws, nv40->so_framebuffer); - - if (nv40->dirty & NV40_NEW_BLEND) - so_emit(nv40->nvws, nv40->so_blend); - - if (nv40->dirty & NV40_NEW_RAST) - so_emit(nv40->nvws, nv40->so_rast); - - if (nv40->dirty & NV40_NEW_ZSA) - so_emit(nv40->nvws, nv40->so_zsa); - - if (nv40->dirty & NV40_NEW_BCOL) - so_emit(nv40->nvws, nv40->so_bcol); - - if (nv40->dirty & NV40_NEW_SCISSOR) - so_emit(nv40->nvws, nv40->so_scissor); - - if (nv40->dirty & NV40_NEW_VIEWPORT) - so_emit(nv40->nvws, nv40->so_viewport); - - if (nv40->dirty & NV40_NEW_STIPPLE) - so_emit(nv40->nvws, nv40->so_stipple); - - if (nv40->dirty & NV40_NEW_FRAGPROG) { - nv40_fragprog_bind(nv40, nv40->fragprog.current); - /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ - } - - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { - nv40_fragtex_bind(nv40); - - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (1); - nv40->dirty &= ~NV40_NEW_FRAGPROG; - } - - if (nv40->dirty & NV40_NEW_VERTPROG) { - nv40_vertprog_bind(nv40, nv40->vertprog.current); - nv40->dirty &= ~NV40_NEW_VERTPROG; - } - - nv40->dirty_samplers = 0; - nv40->dirty = 0; - - nv40_state_emit_dummy_relocs(nv40); -} - diff --git a/src/mesa/pipe/nv40/nv40_surface.c b/src/mesa/pipe/nv40/nv40_surface.c deleted file mode 100644 index d8f87d9adc2..00000000000 --- a/src/mesa/pipe/nv40/nv40_surface.c +++ /dev/null @@ -1,136 +0,0 @@ - -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "nv40_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_winsys.h" -#include "pipe/p_inlines.h" -#include "pipe/util/p_tile.h" - -static boolean -nv40_surface_format_supported(struct pipe_context *pipe, - enum pipe_format format, uint type) -{ - switch (type) { - case PIPE_SURFACE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return TRUE; - default: - break; - } - break; - case PIPE_TEXTURE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A1R5G5B5_UNORM: - case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: - case PIPE_FORMAT_Z16_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - } - break; - default: - assert(0); - }; - - return FALSE; -} - -static struct pipe_surface * -nv40_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) -{ - struct pipe_winsys *ws = pipe->winsys; - struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; - struct pipe_surface *ps; - - ps = ws->surface_alloc(ws); - if (!ps) - return NULL; - pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); - ps->format = pt->format; - ps->cpp = pt->cpp; - ps->width = pt->width[level]; - ps->height = pt->height[level]; - ps->pitch = nv40mt->level[level].pitch / ps->cpp; - - if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv40mt->level[level].image_offset[face]; - } else - if (pt->target == PIPE_TEXTURE_3D) { - ps->offset = nv40mt->level[level].image_offset[zslice]; - } else { - ps->offset = nv40mt->level[level].image_offset[0]; - } - - return ps; -} - -static void -nv40_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, struct pipe_surface *src, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; - - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); -} - -static void -nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, unsigned width, - unsigned height, unsigned value) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; - - nvws->surface_fill(nvws, dest, destx, desty, width, height, value); -} - -void -nv40_init_surface_functions(struct nv40_context *nv40) -{ - nv40->pipe.is_format_supported = nv40_surface_format_supported; - nv40->pipe.get_tex_surface = nv40_get_tex_surface; - nv40->pipe.surface_copy = nv40_surface_copy; - nv40->pipe.surface_fill = nv40_surface_fill; -} diff --git a/src/mesa/pipe/nv40/nv40_vbo.c b/src/mesa/pipe/nv40/nv40_vbo.c deleted file mode 100644 index 6b1ac65b49e..00000000000 --- a/src/mesa/pipe/nv40/nv40_vbo.c +++ /dev/null @@ -1,424 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_state.h" -#include "pipe/p_util.h" - -#include "nv40_context.h" -#include "nv40_state.h" - -#include "pipe/nouveau/nouveau_channel.h" -#include "pipe/nouveau/nouveau_pushbuf.h" - -static INLINE int -nv40_vbo_ncomp(uint format) -{ - int ncomp = 0; - - if (pf_size_x(format)) ncomp++; - if (pf_size_y(format)) ncomp++; - if (pf_size_z(format)) ncomp++; - if (pf_size_w(format)) ncomp++; - - return ncomp; -} - -static INLINE int -nv40_vbo_type(uint format) -{ - switch (pf_type(format)) { - case PIPE_FORMAT_TYPE_FLOAT: - return NV40TCL_VTXFMT_TYPE_FLOAT; - case PIPE_FORMAT_TYPE_UNORM: - return NV40TCL_VTXFMT_TYPE_UBYTE; - default: - NOUVEAU_ERR("Unknown format 0x%08x\n", format); - return NV40TCL_VTXFMT_TYPE_FLOAT; - } -} - -static boolean -nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, - struct pipe_vertex_element *ve, - struct pipe_vertex_buffer *vb) -{ - struct pipe_winsys *ws = nv40->pipe.winsys; - int type, ncomp; - void *map; - - type = nv40_vbo_type(ve->src_format); - ncomp = nv40_vbo_ncomp(ve->src_format); - - map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); - map += vb->buffer_offset + ve->src_offset; - - switch (type) { - case NV40TCL_VTXFMT_TYPE_FLOAT: - { - float *v = map; - - BEGIN_RING(curie, NV40TCL_VTX_ATTR_4F_X(attrib), 4); - switch (ncomp) { - case 4: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(v[3]); - break; - case 3: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(1.0); - break; - case 2: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(0.0); - OUT_RINGf(1.0); - break; - case 1: - OUT_RINGf(v[0]); - OUT_RINGf(0.0); - OUT_RINGf(0.0); - OUT_RINGf(1.0); - break; - default: - ws->buffer_unmap(ws, vb->buffer); - return FALSE; - } - } - break; - default: - ws->buffer_unmap(ws, vb->buffer); - return FALSE; - } - - ws->buffer_unmap(ws, vb->buffer); - - return TRUE; -} - -static void -nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, - unsigned ib_format) -{ - struct nv40_vertex_program *vp = nv40->vertprog.active; - struct nouveau_stateobj *vtxbuf, *vtxfmt; - unsigned inputs, hw, num_hw; - unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; - - inputs = vp->ir; - for (hw = 0; hw < 16 && inputs; hw++) { - if (inputs & (1 << hw)) { - num_hw = hw; - inputs &= ~(1 << hw); - } - } - num_hw++; - - vtxbuf = so_new(20, 18); - so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); - vtxfmt = so_new(17, 0); - so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); - - inputs = vp->ir; - for (hw = 0; hw < num_hw; hw++) { - struct pipe_vertex_element *ve; - struct pipe_vertex_buffer *vb; - - if (!(inputs & (1 << hw))) { - so_data(vtxbuf, 0); - so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); - continue; - } - - ve = &nv40->vtxelt[hw]; - vb = &nv40->vtxbuf[ve->vertex_buffer_index]; - - if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { - so_data(vtxbuf, 0); - so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); - continue; - } - - so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, - vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, - 0, NV40TCL_VTXBUF_ADDRESS_DMA1); - so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_ncomp(ve->src_format) << - NV40TCL_VTXFMT_SIZE_SHIFT) | - nv40_vbo_type(ve->src_format))); - } - - if (ib) { - so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); - so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); - so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, - 0, NV40TCL_IDXBUF_FORMAT_DMA1); - } - - so_emit(nv40->nvws, vtxfmt); - so_emit(nv40->nvws, vtxbuf); - so_ref (vtxbuf, &nv40->so_vtxbuf); - so_ref (NULL, &vtxbuf); - so_ref (NULL, &vtxfmt); -} - -static boolean -nv40_vbo_validate_state(struct nv40_context *nv40, - struct pipe_buffer *ib, unsigned ib_format) -{ - unsigned vdn = nv40->dirty & NV40_NEW_ARRAYS; - - nv40_emit_hw_state(nv40); - if (vdn || ib) { - nv40_vbo_arrays_update(nv40, ib, ib_format); - nv40->dirty &= ~NV40_NEW_ARRAYS; - } - - so_emit_reloc_markers(nv40->nvws, nv40->so_vtxbuf); - - BEGIN_RING(curie, 0x1710, 1); - OUT_RING (0); /* vtx cache flush */ - - return TRUE; -} - -boolean -nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, - unsigned count) -{ - struct nv40_context *nv40 = nv40_context(pipe); - unsigned nr; - boolean ret; - - ret = nv40_vbo_validate_state(nv40, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } - - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - nr = (count & 0xff); - if (nr) { - BEGIN_RING(curie, NV40TCL_VB_VERTEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } - - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; - - nr -= push; - - BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; - } - } - - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (0); - - pipe->flush(pipe, 0); - return TRUE; -} - -static INLINE void -nv40_draw_elements_u08(struct nv40_context *nv40, void *ib, - unsigned start, unsigned count) -{ - uint8_t *elts = (uint8_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } - - while (count) { - push = MIN2(count, 2047 * 2); - - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); - - count -= push; - elts += push; - } -} - -static INLINE void -nv40_draw_elements_u16(struct nv40_context *nv40, void *ib, - unsigned start, unsigned count) -{ - uint16_t *elts = (uint16_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } - - while (count) { - push = MIN2(count, 2047 * 2); - - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); - - count -= push; - elts += push; - } -} - -static INLINE void -nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, - unsigned start, unsigned count) -{ - uint32_t *elts = (uint32_t *)ib + start; - int push; - - while (count) { - push = MIN2(count, 2047); - - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U32, push); - OUT_RINGp (elts, push); - - count -= push; - elts += push; - } -} - -static boolean -nv40_draw_elements_inline(struct pipe_context *pipe, - struct pipe_buffer *ib, unsigned ib_size, - unsigned mode, unsigned start, unsigned count) -{ - struct nv40_context *nv40 = nv40_context(pipe); - struct pipe_winsys *ws = pipe->winsys; - boolean ret; - void *map; - - ret = nv40_vbo_validate_state(nv40, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } - - map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); - if (!ib) { - NOUVEAU_ERR("failed mapping ib\n"); - return FALSE; - } - - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - switch (ib_size) { - case 1: - nv40_draw_elements_u08(nv40, map, start, count); - break; - case 2: - nv40_draw_elements_u16(nv40, map, start, count); - break; - case 4: - nv40_draw_elements_u32(nv40, map, start, count); - break; - default: - NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); - break; - } - - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (0); - - ws->buffer_unmap(ws, ib); - - return TRUE; -} - -static boolean -nv40_draw_elements_vbo(struct pipe_context *pipe, - struct pipe_buffer *ib, unsigned ib_size, - unsigned mode, unsigned start, unsigned count) -{ - struct nv40_context *nv40 = nv40_context(pipe); - unsigned nr, type; - boolean ret; - - switch (ib_size) { - case 2: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; - break; - case 4: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; - break; - default: - NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); - return FALSE; - } - - ret = nv40_vbo_validate_state(nv40, ib, type); - if (!ret) { - NOUVEAU_ERR("failed state validation\n"); - return FALSE; - } - - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - nr = (count & 0xff); - if (nr) { - BEGIN_RING(curie, NV40TCL_VB_INDEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } - - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; - - nr -= push; - - BEGIN_RING_NI(curie, NV40TCL_VB_INDEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; - } - } - - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (0); - - return TRUE; -} - -boolean -nv40_draw_elements(struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, unsigned indexSize, - unsigned mode, unsigned start, unsigned count) -{ - struct nv40_context *nv40 = nv40_context(pipe); - - /* 0x4497 doesn't support real index buffers, and there doesn't appear - * to be support on any chipset for 8-bit indices. - */ - if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { - nv40_draw_elements_inline(pipe, indexBuffer, indexSize, - mode, start, count); - } else { - nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, - mode, start, count); - } - - pipe->flush(pipe, 0); - return TRUE; -} - - diff --git a/src/mesa/pipe/nv40/nv40_vertprog.c b/src/mesa/pipe/nv40/nv40_vertprog.c deleted file mode 100644 index d57e3ca350b..00000000000 --- a/src/mesa/pipe/nv40/nv40_vertprog.c +++ /dev/null @@ -1,790 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/p_shader_tokens.h" -#include "pipe/tgsi/util/tgsi_parse.h" - -#include "nv40_context.h" -#include "nv40_state.h" - -/* TODO (at least...): - * 1. Indexed consts + ARL - * 2. Arb. swz/negation - * 3. NV_vp11, NV_vp2, NV_vp3 features - * - extra arith opcodes - * - branching - * - texture sampling - * - indexed attribs - * - indexed results - * 4. bugs - */ - -#define SWZ_X 0 -#define SWZ_Y 1 -#define SWZ_Z 2 -#define SWZ_W 3 -#define MASK_X 8 -#define MASK_Y 4 -#define MASK_Z 2 -#define MASK_W 1 -#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) -#define DEF_SCALE 0 -#define DEF_CTEST 0 -#include "nv40_shader.h" - -#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) -#define neg(s) nv40_sr_neg((s)) -#define abs(s) nv40_sr_abs((s)) - -struct nv40_vpc { - struct nv40_vertex_program *vp; - - struct nv40_vertex_program_exec *vpi; - - unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; - - int high_temp; - int temp_temp_count; - - struct nv40_sreg *imm; - unsigned nr_imm; -}; - -static struct nv40_sreg -temp(struct nv40_vpc *vpc) -{ - int idx; - - idx = vpc->temp_temp_count++; - idx += vpc->high_temp + 1; - return nv40_sr(NV40SR_TEMP, idx); -} - -static struct nv40_sreg -constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w) -{ - struct nv40_vertex_program *vp = vpc->vp; - struct nv40_vertex_program_data *vpd; - int idx; - - if (pipe >= 0) { - for (idx = 0; idx < vp->nr_consts; idx++) { - if (vp->consts[idx].index == pipe) - return nv40_sr(NV40SR_CONST, idx); - } - } - - idx = vp->nr_consts++; - vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); - vpd = &vp->consts[idx]; - - vpd->index = pipe; - vpd->value[0] = x; - vpd->value[1] = y; - vpd->value[2] = z; - vpd->value[3] = w; - return nv40_sr(NV40SR_CONST, idx); -} - -#define arith(cc,s,o,d,m,s0,s1,s2) \ - nv40_vp_arith((cc), (s), NV40_VP_INST_##o, (d), (m), (s0), (s1), (s2)) - -static void -emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src) -{ - struct nv40_vertex_program *vp = vpc->vp; - uint32_t sr = 0; - - switch (src.type) { - case NV40SR_TEMP: - sr |= (NV40_VP_SRC_REG_TYPE_TEMP << NV40_VP_SRC_REG_TYPE_SHIFT); - sr |= (src.index << NV40_VP_SRC_TEMP_SRC_SHIFT); - break; - case NV40SR_INPUT: - sr |= (NV40_VP_SRC_REG_TYPE_INPUT << - NV40_VP_SRC_REG_TYPE_SHIFT); - vp->ir |= (1 << src.index); - hw[1] |= (src.index << NV40_VP_INST_INPUT_SRC_SHIFT); - break; - case NV40SR_CONST: - sr |= (NV40_VP_SRC_REG_TYPE_CONST << - NV40_VP_SRC_REG_TYPE_SHIFT); - assert(vpc->vpi->const_index == -1 || - vpc->vpi->const_index == src.index); - vpc->vpi->const_index = src.index; - break; - case NV40SR_NONE: - sr |= (NV40_VP_SRC_REG_TYPE_INPUT << - NV40_VP_SRC_REG_TYPE_SHIFT); - break; - default: - assert(0); - } - - if (src.negate) - sr |= NV40_VP_SRC_NEGATE; - - if (src.abs) - hw[0] |= (1 << (21 + pos)); - - sr |= ((src.swz[0] << NV40_VP_SRC_SWZ_X_SHIFT) | - (src.swz[1] << NV40_VP_SRC_SWZ_Y_SHIFT) | - (src.swz[2] << NV40_VP_SRC_SWZ_Z_SHIFT) | - (src.swz[3] << NV40_VP_SRC_SWZ_W_SHIFT)); - - switch (pos) { - case 0: - hw[1] |= ((sr & NV40_VP_SRC0_HIGH_MASK) >> - NV40_VP_SRC0_HIGH_SHIFT) << NV40_VP_INST_SRC0H_SHIFT; - hw[2] |= (sr & NV40_VP_SRC0_LOW_MASK) << - NV40_VP_INST_SRC0L_SHIFT; - break; - case 1: - hw[2] |= sr << NV40_VP_INST_SRC1_SHIFT; - break; - case 2: - hw[2] |= ((sr & NV40_VP_SRC2_HIGH_MASK) >> - NV40_VP_SRC2_HIGH_SHIFT) << NV40_VP_INST_SRC2H_SHIFT; - hw[3] |= (sr & NV40_VP_SRC2_LOW_MASK) << - NV40_VP_INST_SRC2L_SHIFT; - break; - default: - assert(0); - } -} - -static void -emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) -{ - struct nv40_vertex_program *vp = vpc->vp; - - switch (dst.type) { - case NV40SR_TEMP: - hw[3] |= NV40_VP_INST_DEST_MASK; - if (slot == 0) { - hw[0] |= (dst.index << - NV40_VP_INST_VEC_DEST_TEMP_SHIFT); - } else { - hw[3] |= (dst.index << - NV40_VP_INST_SCA_DEST_TEMP_SHIFT); - } - break; - case NV40SR_OUTPUT: - switch (dst.index) { - case NV40_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; - case NV40_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; - case NV40_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; - case NV40_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; - case NV40_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; - case NV40_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; - case NV40_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; - case NV40_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; - case NV40_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; - case NV40_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; - case NV40_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; - case NV40_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; - case NV40_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; - case NV40_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; - default: - break; - } - - hw[3] |= (dst.index << NV40_VP_INST_DEST_SHIFT); - if (slot == 0) { - hw[0] |= NV40_VP_INST_VEC_RESULT; - hw[0] |= NV40_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); - } else { - hw[3] |= NV40_VP_INST_SCA_RESULT; - hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; - } - break; - default: - assert(0); - } -} - -static void -nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op, - struct nv40_sreg dst, int mask, - struct nv40_sreg s0, struct nv40_sreg s1, - struct nv40_sreg s2) -{ - struct nv40_vertex_program *vp = vpc->vp; - uint32_t *hw; - - vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); - vpc->vpi = &vp->insns[vp->nr_insns - 1]; - memset(vpc->vpi, 0, sizeof(*vpc->vpi)); - vpc->vpi->const_index = -1; - - hw = vpc->vpi->data; - - hw[0] |= (NV40_VP_INST_COND_TR << NV40_VP_INST_COND_SHIFT); - hw[0] |= ((0 << NV40_VP_INST_COND_SWZ_X_SHIFT) | - (1 << NV40_VP_INST_COND_SWZ_Y_SHIFT) | - (2 << NV40_VP_INST_COND_SWZ_Z_SHIFT) | - (3 << NV40_VP_INST_COND_SWZ_W_SHIFT)); - - if (slot == 0) { - hw[1] |= (op << NV40_VP_INST_VEC_OPCODE_SHIFT); - hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK; - hw[3] |= (mask << NV40_VP_INST_VEC_WRITEMASK_SHIFT); - } else { - hw[1] |= (op << NV40_VP_INST_SCA_OPCODE_SHIFT); - hw[0] |= (NV40_VP_INST_VEC_DEST_TEMP_MASK | (1 << 20)); - hw[3] |= (mask << NV40_VP_INST_SCA_WRITEMASK_SHIFT); - } - - emit_dst(vpc, hw, slot, dst); - emit_src(vpc, hw, 0, s0); - emit_src(vpc, hw, 1, s1); - emit_src(vpc, hw, 2, s2); -} - -static INLINE struct nv40_sreg -tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { - struct nv40_sreg src; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - src = nv40_sr(NV40SR_INPUT, fsrc->SrcRegister.Index); - break; - case TGSI_FILE_CONSTANT: - src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); - break; - case TGSI_FILE_IMMEDIATE: - src = vpc->imm[fsrc->SrcRegister.Index]; - break; - case TGSI_FILE_TEMPORARY: - if (vpc->high_temp < fsrc->SrcRegister.Index) - vpc->high_temp = fsrc->SrcRegister.Index; - src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index); - break; - default: - NOUVEAU_ERR("bad src file\n"); - break; - } - - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; - return src; -} - -static INLINE struct nv40_sreg -tgsi_dst(struct nv40_vpc *vpc, const struct tgsi_full_dst_register *fdst) { - struct nv40_sreg dst; - - switch (fdst->DstRegister.File) { - case TGSI_FILE_OUTPUT: - dst = nv40_sr(NV40SR_OUTPUT, - vpc->output_map[fdst->DstRegister.Index]); - - break; - case TGSI_FILE_TEMPORARY: - dst = nv40_sr(NV40SR_TEMP, fdst->DstRegister.Index); - if (vpc->high_temp < dst.index) - vpc->high_temp = dst.index; - break; - default: - NOUVEAU_ERR("bad dst file\n"); - break; - } - - return dst; -} - -static INLINE int -tgsi_mask(uint tgsi) -{ - int mask = 0; - - if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; - if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; - if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; - if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; - return mask; -} - -static boolean -nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, - const struct tgsi_full_instruction *finst) -{ - struct nv40_sreg src[3], dst, tmp; - struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); - int mask; - int ai = -1, ci = -1; - int i; - - if (finst->Instruction.Opcode == TGSI_OPCODE_END) - return TRUE; - - vpc->temp_temp_count = 0; - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { - src[i] = tgsi_src(vpc, fsrc); - } - } - - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; - src[i] = tgsi_src(vpc, fsrc); - } else { - src[i] = temp(vpc); - arith(vpc, 0, OP_MOV, src[i], MASK_ALL, - tgsi_src(vpc, fsrc), none, none); - } - break; - /*XXX: index comparison is broken now that consts come from - * two different register files. - */ - case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; - src[i] = tgsi_src(vpc, fsrc); - } else { - src[i] = temp(vpc); - arith(vpc, 0, OP_MOV, src[i], MASK_ALL, - tgsi_src(vpc, fsrc), none, none); - } - break; - case TGSI_FILE_TEMPORARY: - /* handled above */ - break; - default: - NOUVEAU_ERR("bad src file\n"); - return FALSE; - } - } - - dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); - - switch (finst->Instruction.Opcode) { - case TGSI_OPCODE_ABS: - arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); - break; - case TGSI_OPCODE_ADD: - arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); - break; - case TGSI_OPCODE_ARL: - arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_DP3: - arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DP4: - arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DPH: - arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DST: - arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_EX2: - arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_EXP: - arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_FLR: - arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FRC: - arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_LG2: - arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_LIT: - arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_LOG: - arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_MAD: - arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); - break; - case TGSI_OPCODE_MAX: - arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MIN: - arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MOV: - arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_MUL: - arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_POW: - tmp = temp(vpc); - arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, - swz(src[0], X, X, X, X)); - arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), - swz(src[1], X, X, X, X), none); - arith(vpc, 1, OP_EX2, dst, mask, none, none, - swz(tmp, X, X, X, X)); - break; - case TGSI_OPCODE_RCP: - arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_RET: - break; - case TGSI_OPCODE_RSQ: - arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_SGE: - arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SLT: - arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SUB: - arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); - break; - case TGSI_OPCODE_XPD: - tmp = temp(vpc); - arith(vpc, 0, OP_MUL, tmp, mask, - swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); - arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), - swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), - neg(tmp)); - break; - default: - NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); - return FALSE; - } - - return TRUE; -} - -static boolean -nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, - const struct tgsi_full_declaration *fdec) -{ - int hw; - - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - hw = NV40_VP_INST_DEST_POS; - break; - case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV40_VP_INST_DEST_COL0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV40_VP_INST_DEST_COL1; - } else { - NOUVEAU_ERR("bad colour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_BCOLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV40_VP_INST_DEST_BFC0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV40_VP_INST_DEST_BFC1; - } else { - NOUVEAU_ERR("bad bcolour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_FOG: - hw = NV40_VP_INST_DEST_FOGC; - break; - case TGSI_SEMANTIC_PSIZE: - hw = NV40_VP_INST_DEST_PSZ; - break; - case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { - hw = NV40_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); - } else { - NOUVEAU_ERR("bad generic semantic index\n"); - return FALSE; - } - break; - default: - NOUVEAU_ERR("bad output semantic\n"); - return FALSE; - } - - vpc->output_map[fdec->u.DeclarationRange.First] = hw; - return TRUE; -} - -static boolean -nv40_vertprog_prepare(struct nv40_vpc *vpc) -{ - struct tgsi_parse_context p; - int nr_imm = 0; - - tgsi_parse_init(&p, vpc->vp->pipe->tokens); - while (!tgsi_parse_end_of_tokens(&p)) { - const union tgsi_full_token *tok = &p.FullToken; - - tgsi_parse_token(&p); - switch(tok->Token.Type) { - case TGSI_TOKEN_TYPE_IMMEDIATE: - nr_imm++; - break; - default: - break; - } - } - tgsi_parse_free(&p); - - if (nr_imm) { - vpc->imm = CALLOC(nr_imm, sizeof(struct nv40_sreg)); - assert(vpc->imm); - } - - return TRUE; -} - -void -nv40_vertprog_translate(struct nv40_context *nv40, - struct nv40_vertex_program *vp) -{ - struct tgsi_parse_context parse; - struct nv40_vpc *vpc = NULL; - - vpc = CALLOC(1, sizeof(struct nv40_vpc)); - if (!vpc) - return; - vpc->vp = vp; - vpc->high_temp = -1; - - if (!nv40_vertprog_prepare(vpc)) { - free(vpc); - return; - } - - tgsi_parse_init(&parse, vp->pipe->tokens); - - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - { - const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; - switch (fdec->Declaration.File) { - case TGSI_FILE_OUTPUT: - if (!nv40_vertprog_parse_decl_output(vpc, fdec)) - goto out_err; - break; - default: - break; - } - } - break; - case TGSI_TOKEN_TYPE_IMMEDIATE: - { - const struct tgsi_full_immediate *imm; - - imm = &parse.FullToken.FullImmediate; - assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); -// assert(imm->Immediate.Size == 4); - vpc->imm[vpc->nr_imm++] = - constant(vpc, -1, - imm->u.ImmediateFloat32[0].Float, - imm->u.ImmediateFloat32[1].Float, - imm->u.ImmediateFloat32[2].Float, - imm->u.ImmediateFloat32[3].Float); - } - break; - case TGSI_TOKEN_TYPE_INSTRUCTION: - { - const struct tgsi_full_instruction *finst; - finst = &parse.FullToken.FullInstruction; - if (!nv40_vertprog_parse_instruction(vpc, finst)) - goto out_err; - } - break; - default: - break; - } - } - - vp->insns[vp->nr_insns - 1].data[3] |= NV40_VP_INST_LAST; - vp->translated = TRUE; -out_err: - tgsi_parse_free(&parse); - free(vpc); -} - -void -nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) -{ - struct nouveau_winsys *nvws = nv40->nvws; - struct pipe_winsys *ws = nv40->pipe.winsys; - boolean upload_code = FALSE, upload_data = FALSE; - int i; - - /* Translate TGSI shader into hw bytecode */ - if (!vp->translated) { - nv40_vertprog_translate(nv40, vp); - if (!vp->translated) - assert(0); - } - - /* Allocate hw vtxprog exec slots */ - if (!vp->exec) { - struct nouveau_resource *heap = nv40->hw->vp_exec_heap; - uint vplen = vp->nr_insns; - - if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { - while (heap->next && heap->size < vplen) { - struct nv40_vertex_program *evict; - - evict = heap->next->priv; - nvws->res_free(&evict->exec); - } - - if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) - assert(0); - } - - upload_code = TRUE; - } - - /* Allocate hw vtxprog const slots */ - if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv40->hw->vp_data_heap; - - if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { - while (heap->next && heap->size < vp->nr_consts) { - struct nv40_vertex_program *evict; - - evict = heap->next->priv; - nvws->res_free(&evict->data); - } - - if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) - assert(0); - } - - /*XXX: handle this some day */ - assert(vp->data->start >= vp->data_start_min); - - upload_data = TRUE; - if (vp->data_start != vp->data->start) - upload_code = TRUE; - } - - /* If exec or data segments moved we need to patch the program to - * fixup offsets and register IDs. - */ - if (vp->exec_start != vp->exec->start) { - for (i = 0; i < vp->nr_insns; i++) { - struct nv40_vertex_program_exec *vpi = &vp->insns[i]; - - if (vpi->has_branch_offset) { - assert(0); - } - } - - vp->exec_start = vp->exec->start; - } - - if (vp->nr_consts && vp->data_start != vp->data->start) { - for (i = 0; i < vp->nr_insns; i++) { - struct nv40_vertex_program_exec *vpi = &vp->insns[i]; - - if (vpi->const_index >= 0) { - vpi->data[1] &= ~NV40_VP_INST_CONST_SRC_MASK; - vpi->data[1] |= - (vpi->const_index + vp->data->start) << - NV40_VP_INST_CONST_SRC_SHIFT; - - } - } - - vp->data_start = vp->data->start; - } - - /* Update + Upload constant values */ - if (vp->nr_consts) { - float *map = NULL; - - if (nv40->vertprog.constant_buf) { - map = ws->buffer_map(ws, nv40->vertprog.constant_buf, - PIPE_BUFFER_USAGE_CPU_READ); - } - - for (i = 0; i < vp->nr_consts; i++) { - struct nv40_vertex_program_data *vpd = &vp->consts[i]; - - if (vpd->index >= 0) { - if (!upload_data && - !memcmp(vpd->value, &map[vpd->index * 4], - 4 * sizeof(float))) - continue; - memcpy(vpd->value, &map[vpd->index * 4], - 4 * sizeof(float)); - } - - BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5); - OUT_RING (i + vp->data->start); - OUT_RINGp ((uint32_t *)vpd->value, 4); - } - - if (map) { - ws->buffer_unmap(ws, nv40->vertprog.constant_buf); - } - } - - /* Upload vtxprog */ - if (upload_code) { -#if 0 - for (i = 0; i < vp->nr_insns; i++) { - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); - } -#endif - BEGIN_RING(curie, NV40TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (vp->exec->start); - for (i = 0; i < vp->nr_insns; i++) { - BEGIN_RING(curie, NV40TCL_VP_UPLOAD_INST(0), 4); - OUT_RINGp (vp->insns[i].data, 4); - } - } - - BEGIN_RING(curie, NV40TCL_VP_START_FROM_ID, 1); - OUT_RING (vp->exec->start); - BEGIN_RING(curie, NV40TCL_VP_ATTRIB_EN, 2); - OUT_RING (vp->ir); - OUT_RING (vp->or); - - nv40->vertprog.active = vp; -} - -void -nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp) -{ - if (vp->nr_consts) - free(vp->consts); - if (vp->nr_insns) - free(vp->insns); -} - diff --git a/src/mesa/pipe/nv50/Makefile b/src/mesa/pipe/nv50/Makefile deleted file mode 100644 index d3d011b14b7..00000000000 --- a/src/mesa/pipe/nv50/Makefile +++ /dev/null @@ -1,25 +0,0 @@ -TOP = ../../../.. -include $(TOP)/configs/current - -LIBNAME = nv50 - -DRIVER_SOURCES = \ - nv50_clear.c \ - nv50_context.c \ - nv50_draw.c \ - nv50_miptree.c \ - nv50_query.c \ - nv50_state.c \ - nv50_surface.c \ - nv50_vbo.c - -C_SOURCES = \ - $(COMMON_SOURCES) \ - $(DRIVER_SOURCES) - -ASM_SOURCES = - -include ../Makefile.template - -symlinks: - diff --git a/src/mesa/pipe/nv50/nv50_clear.c b/src/mesa/pipe/nv50/nv50_clear.c deleted file mode 100644 index 552b92f72e2..00000000000 --- a/src/mesa/pipe/nv50/nv50_clear.c +++ /dev/null @@ -1,12 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "nv50_context.h" - -void -nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue) -{ - pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); -} diff --git a/src/mesa/pipe/nv50/nv50_context.c b/src/mesa/pipe/nv50/nv50_context.c deleted file mode 100644 index 33c8eebb0bc..00000000000 --- a/src/mesa/pipe/nv50/nv50_context.c +++ /dev/null @@ -1,202 +0,0 @@ -#include "pipe/draw/draw_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "pipe/p_util.h" - -#include "nv50_context.h" - -static boolean -nv50_is_format_supported(struct pipe_context *pipe, enum pipe_format format, - uint type) -{ - return FALSE; -} - -static const char * -nv50_get_name(struct pipe_context *pipe) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", nv50->chipset); - return buffer; -} - -static const char * -nv50_get_vendor(struct pipe_context *pipe) -{ - return "nouveau"; -} - -static int -nv50_get_param(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 32; - case PIPE_CAP_NPOT_TEXTURES: - return 0; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 1; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 0; - case PIPE_CAP_POINT_SPRITE: - return 0; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 8; - case PIPE_CAP_OCCLUSION_QUERY: - return 0; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 0; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 13; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 10; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 13; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv50_get_paramf(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} - -static void -nv50_flush(struct pipe_context *pipe, unsigned flags) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv50->sync, 0); - BEGIN_RING(tesla, 0x104, 1); - OUT_RING (0); - BEGIN_RING(tesla, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv50->sync, 0, 0, 2000); -} - -static void -nv50_destroy(struct pipe_context *pipe) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - - draw_destroy(nv50->draw); - free(nv50); -} - -static boolean -nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) -{ - struct nouveau_winsys *nvws = nv50->nvws; - int ret; - - if ((ret = nvws->grobj_alloc(nvws, tesla_class, &nv50->tesla))) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - BEGIN_RING(tesla, NV50TCL_DMA_NOTIFY, 1); - OUT_RING (nv50->sync->handle); - - FIRE_RING (); - return TRUE; -} - -#define GRCLASS5097_CHIPSETS 0x00000000 -#define GRCLASS8297_CHIPSETS 0x00000010 -struct pipe_context * -nv50_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv50_context *nv50; - int tesla_class, ret; - - if ((chipset & 0xf0) != 0x50 && (chipset & 0xf0) != 0x80) { - NOUVEAU_ERR("Not a G8x chipset\n"); - return NULL; - } - - if (GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) { - tesla_class = 0x5097; - } else - if (GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) { - tesla_class = 0x8297; - } else { - NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset); - return NULL; - } - - nv50 = CALLOC_STRUCT(nv50_context); - if (!nv50) - return NULL; - nv50->chipset = chipset; - nv50->nvws = nvws; - - if ((ret = nvws->notifier_alloc(nvws, 1, &nv50->sync))) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - free(nv50); - return NULL; - } - - if (!nv50_init_hwctx(nv50, tesla_class)) { - free(nv50); - return NULL; - } - - nv50->pipe.winsys = pipe_winsys; - - nv50->pipe.destroy = nv50_destroy; - nv50->pipe.is_format_supported = nv50_is_format_supported; - nv50->pipe.get_name = nv50_get_name; - nv50->pipe.get_vendor = nv50_get_vendor; - nv50->pipe.get_param = nv50_get_param; - nv50->pipe.get_paramf = nv50_get_paramf; - - nv50->pipe.draw_arrays = nv50_draw_arrays; - nv50->pipe.draw_elements = nv50_draw_elements; - nv50->pipe.clear = nv50_clear; - - nv50->pipe.flush = nv50_flush; - - nv50_init_miptree_functions(nv50); - nv50_init_surface_functions(nv50); - nv50_init_state_functions(nv50); - nv50_init_query_functions(nv50); - - nv50->draw = draw_create(); - assert(nv50->draw); - draw_set_rasterize_stage(nv50->draw, nv50_draw_render_stage(nv50)); - - return &nv50->pipe; -} - - diff --git a/src/mesa/pipe/nv50/nv50_context.h b/src/mesa/pipe/nv50/nv50_context.h deleted file mode 100644 index 5491c0cbb5d..00000000000 --- a/src/mesa/pipe/nv50/nv50_context.h +++ /dev/null @@ -1,57 +0,0 @@ -#ifndef __NV50_CONTEXT_H__ -#define __NV50_CONTEXT_H__ - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/draw/draw_vertex.h" - -#include "pipe/nouveau/nouveau_winsys.h" -#include "pipe/nouveau/nouveau_gldefs.h" - -#define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv50_context *ctx = nv50 -#include "pipe/nouveau/nouveau_push.h" - -#include "nv50_state.h" - -#define NOUVEAU_ERR(fmt, args...) \ - fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); -#define NOUVEAU_MSG(fmt, args...) \ - fprintf(stderr, "nouveau: "fmt, ##args); - -struct nv50_context { - struct pipe_context pipe; - struct nouveau_winsys *nvws; - - struct draw_context *draw; - - int chipset; - struct nouveau_grobj *tesla; - struct nouveau_notifier *sync; -}; - - -extern void nv50_init_miptree_functions(struct nv50_context *nv50); -extern void nv50_init_surface_functions(struct nv50_context *nv50); -extern void nv50_init_state_functions(struct nv50_context *nv50); -extern void nv50_init_query_functions(struct nv50_context *nv50); - -/* nv50_draw.c */ -extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50); - -/* nv50_vbo.c */ -extern boolean nv50_draw_arrays(struct pipe_context *, unsigned mode, - unsigned start, unsigned count); -extern boolean nv50_draw_elements(struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, - unsigned indexSize, - unsigned mode, unsigned start, - unsigned count); - -/* nv50_clear.c */ -extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue); - -#endif diff --git a/src/mesa/pipe/nv50/nv50_draw.c b/src/mesa/pipe/nv50/nv50_draw.c deleted file mode 100644 index 85d347f5e00..00000000000 --- a/src/mesa/pipe/nv50/nv50_draw.c +++ /dev/null @@ -1,55 +0,0 @@ -#include "pipe/draw/draw_private.h" -#include "pipe/p_util.h" - -#include "nv50_context.h" - -struct nv50_draw_stage { - struct draw_stage draw; - struct nv50_context *nv50; -}; - -static void -nv50_draw_point(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv50_draw_line(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv50_draw_tri(struct draw_stage *draw, struct prim_header *prim) -{ - NOUVEAU_ERR("\n"); -} - -static void -nv50_draw_flush(struct draw_stage *draw, unsigned flags) -{ -} - -static void -nv50_draw_reset_stipple_counter(struct draw_stage *draw) -{ - NOUVEAU_ERR("\n"); -} - -struct draw_stage * -nv50_draw_render_stage(struct nv50_context *nv50) -{ - struct nv50_draw_stage *nv50draw = CALLOC_STRUCT(nv50_draw_stage); - - nv50draw->nv50 = nv50; - nv50draw->draw.draw = nv50->draw; - nv50draw->draw.point = nv50_draw_point; - nv50draw->draw.line = nv50_draw_line; - nv50draw->draw.tri = nv50_draw_tri; - nv50draw->draw.flush = nv50_draw_flush; - nv50draw->draw.reset_stipple_counter = nv50_draw_reset_stipple_counter; - - return &nv50draw->draw; -} - diff --git a/src/mesa/pipe/nv50/nv50_miptree.c b/src/mesa/pipe/nv50/nv50_miptree.c deleted file mode 100644 index 0c034ed4387..00000000000 --- a/src/mesa/pipe/nv50/nv50_miptree.c +++ /dev/null @@ -1,25 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" - -#include "nv50_context.h" - -static struct pipe_texture * -nv50_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) -{ - NOUVEAU_ERR("unimplemented\n"); - return NULL; -} - -static void -nv50_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) -{ - NOUVEAU_ERR("unimplemented\n"); -} - -void -nv50_init_miptree_functions(struct nv50_context *nv50) -{ - nv50->pipe.texture_create = nv50_miptree_create; - nv50->pipe.texture_release = nv50_miptree_release; -} diff --git a/src/mesa/pipe/nv50/nv50_query.c b/src/mesa/pipe/nv50/nv50_query.c deleted file mode 100644 index d8c3491c2c5..00000000000 --- a/src/mesa/pipe/nv50/nv50_query.c +++ /dev/null @@ -1,47 +0,0 @@ -#include "pipe/p_context.h" - -#include "nv50_context.h" - -static struct pipe_query * -nv50_query_create(struct pipe_context *pipe, unsigned type) -{ - NOUVEAU_ERR("unimplemented\n"); - return NULL; -} - -static void -nv50_query_destroy(struct pipe_context *pipe, struct pipe_query *q) -{ - NOUVEAU_ERR("unimplemented\n"); -} - -static void -nv50_query_begin(struct pipe_context *pipe, struct pipe_query *q) -{ - NOUVEAU_ERR("unimplemented\n"); -} - -static void -nv50_query_end(struct pipe_context *pipe, struct pipe_query *q) -{ - NOUVEAU_ERR("unimplemented\n"); -} - -static boolean -nv50_query_result(struct pipe_context *pipe, struct pipe_query *q, - boolean wait, uint64 *result) -{ - NOUVEAU_ERR("unimplemented\n"); - *result = 0xdeadcafe; - return TRUE; -} - -void -nv50_init_query_functions(struct nv50_context *nv50) -{ - nv50->pipe.create_query = nv50_query_create; - nv50->pipe.destroy_query = nv50_query_destroy; - nv50->pipe.begin_query = nv50_query_begin; - nv50->pipe.end_query = nv50_query_end; - nv50->pipe.get_query_result = nv50_query_result; -} diff --git a/src/mesa/pipe/nv50/nv50_state.c b/src/mesa/pipe/nv50/nv50_state.c deleted file mode 100644 index 99dcab51b26..00000000000 --- a/src/mesa/pipe/nv50/nv50_state.c +++ /dev/null @@ -1,213 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" - -#include "nv50_context.h" -#include "nv50_state.h" - -static void * -nv50_blend_state_create(struct pipe_context *pipe, - const struct pipe_blend_state *cso) -{ - return NULL; -} - -static void -nv50_blend_state_bind(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_blend_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - -static void * -nv50_sampler_state_create(struct pipe_context *pipe, - const struct pipe_sampler_state *cso) -{ - return NULL; -} - -static void -nv50_sampler_state_bind(struct pipe_context *pipe, unsigned unit, - void *hwcso) -{ -} - -static void -nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_set_sampler_texture(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *pt) -{ -} - -static void * -nv50_rasterizer_state_create(struct pipe_context *pipe, - const struct pipe_rasterizer_state *cso) -{ - return NULL; -} - -static void -nv50_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - -static void * -nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, - const struct pipe_depth_stencil_alpha_state *cso) -{ - return NULL; -} - -static void -nv50_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - -static void * -nv50_vp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - return NULL; -} - -static void -nv50_vp_state_bind(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_vp_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - -static void * -nv50_fp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - return NULL; -} - -static void -nv50_fp_state_bind(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_fp_state_delete(struct pipe_context *pipe, void *hwcso) -{ -} - -static void -nv50_set_blend_color(struct pipe_context *pipe, - const struct pipe_blend_color *bcol) -{ -} - -static void -nv50_set_clip_state(struct pipe_context *pipe, - const struct pipe_clip_state *clip) -{ -} - -static void -nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, - const struct pipe_constant_buffer *buf ) -{ -} - -static void -nv50_set_framebuffer_state(struct pipe_context *pipe, - const struct pipe_framebuffer_state *fb) -{ -} - -static void -nv50_set_polygon_stipple(struct pipe_context *pipe, - const struct pipe_poly_stipple *stipple) -{ -} - -static void -nv50_set_scissor_state(struct pipe_context *pipe, - const struct pipe_scissor_state *s) -{ -} - -static void -nv50_set_viewport_state(struct pipe_context *pipe, - const struct pipe_viewport_state *vpt) -{ -} - -static void -nv50_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) -{ -} - -static void -nv50_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) -{ -} - -void -nv50_init_state_functions(struct nv50_context *nv50) -{ - nv50->pipe.create_blend_state = nv50_blend_state_create; - nv50->pipe.bind_blend_state = nv50_blend_state_bind; - nv50->pipe.delete_blend_state = nv50_blend_state_delete; - - nv50->pipe.create_sampler_state = nv50_sampler_state_create; - nv50->pipe.bind_sampler_state = nv50_sampler_state_bind; - nv50->pipe.delete_sampler_state = nv50_sampler_state_delete; - nv50->pipe.set_sampler_texture = nv50_set_sampler_texture; - - nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create; - nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; - nv50->pipe.delete_rasterizer_state = nv50_rasterizer_state_delete; - - nv50->pipe.create_depth_stencil_alpha_state = - nv50_depth_stencil_alpha_state_create; - nv50->pipe.bind_depth_stencil_alpha_state = - nv50_depth_stencil_alpha_state_bind; - nv50->pipe.delete_depth_stencil_alpha_state = - nv50_depth_stencil_alpha_state_delete; - - nv50->pipe.create_vs_state = nv50_vp_state_create; - nv50->pipe.bind_vs_state = nv50_vp_state_bind; - nv50->pipe.delete_vs_state = nv50_vp_state_delete; - - nv50->pipe.create_fs_state = nv50_fp_state_create; - nv50->pipe.bind_fs_state = nv50_fp_state_bind; - nv50->pipe.delete_fs_state = nv50_fp_state_delete; - - nv50->pipe.set_blend_color = nv50_set_blend_color; - nv50->pipe.set_clip_state = nv50_set_clip_state; - nv50->pipe.set_constant_buffer = nv50_set_constant_buffer; - nv50->pipe.set_framebuffer_state = nv50_set_framebuffer_state; - nv50->pipe.set_polygon_stipple = nv50_set_polygon_stipple; - nv50->pipe.set_scissor_state = nv50_set_scissor_state; - nv50->pipe.set_viewport_state = nv50_set_viewport_state; - - nv50->pipe.set_vertex_buffer = nv50_set_vertex_buffer; - nv50->pipe.set_vertex_element = nv50_set_vertex_element; -} - diff --git a/src/mesa/pipe/nv50/nv50_state.h b/src/mesa/pipe/nv50/nv50_state.h deleted file mode 100644 index a3b781d4c61..00000000000 --- a/src/mesa/pipe/nv50/nv50_state.h +++ /dev/null @@ -1,7 +0,0 @@ -#ifndef __NV50_STATE_H__ -#define __NV50_STATE_H__ - -#include "pipe/p_state.h" - - -#endif diff --git a/src/mesa/pipe/nv50/nv50_surface.c b/src/mesa/pipe/nv50/nv50_surface.c deleted file mode 100644 index cfb370da77e..00000000000 --- a/src/mesa/pipe/nv50/nv50_surface.c +++ /dev/null @@ -1,74 +0,0 @@ - -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "nv50_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_winsys.h" -#include "pipe/p_inlines.h" -#include "pipe/util/p_tile.h" - -static struct pipe_surface * -nv50_get_tex_surface(struct pipe_context *pipe, - struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) -{ - NOUVEAU_ERR("unimplemented\n"); - return NULL; -} - -static void -nv50_surface_copy(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, struct pipe_surface *src, - unsigned srcx, unsigned srcy, unsigned width, unsigned height) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); -} - -static void -nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, unsigned width, - unsigned height, unsigned value) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; - - nvws->surface_fill(nvws, dest, destx, desty, width, height, value); -} - -void -nv50_init_surface_functions(struct nv50_context *nv50) -{ - nv50->pipe.get_tex_surface = nv50_get_tex_surface; - nv50->pipe.surface_copy = nv50_surface_copy; - nv50->pipe.surface_fill = nv50_surface_fill; -} diff --git a/src/mesa/pipe/nv50/nv50_vbo.c b/src/mesa/pipe/nv50/nv50_vbo.c deleted file mode 100644 index 6c0dc23a439..00000000000 --- a/src/mesa/pipe/nv50/nv50_vbo.c +++ /dev/null @@ -1,24 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_state.h" -#include "pipe/p_util.h" - -#include "nv50_context.h" -#include "nv50_state.h" - -boolean -nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, - unsigned count) -{ - NOUVEAU_ERR("unimplemented\n"); - return TRUE; -} - -boolean -nv50_draw_elements(struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, unsigned indexSize, - unsigned mode, unsigned start, unsigned count) -{ - NOUVEAU_ERR("unimplemented\n"); - return TRUE; -} - -- cgit v1.2.3 From c303cf15dcf2744028f920cf71d7e6fda709bd15 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 16 Feb 2008 17:51:44 +1100 Subject: nouveau: fix potential userbuf crash. --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index 6887ffa6886..ad587bafdff 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -295,7 +295,8 @@ nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) &new, &new_map); if (ret) return ret; - } else { + } else + if (!nvbo->user) { new_sysmem = malloc(bo->size); } @@ -311,12 +312,13 @@ nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) if (nvbo->fence) nouveau_fence_wait(&nvbo->fence); nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); - if (nvbo->sysmem) + if (nvbo->sysmem && !nvbo->user) free(nvbo->sysmem); nvbo->drm = new; nvbo->map = new_map; - nvbo->sysmem = new_sysmem; + if (!nvbo->user) + nvbo->sysmem = new_sysmem; bo->flags = flags; bo->offset = nvbo->drm.offset; return 0; @@ -333,7 +335,6 @@ nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, if (nvchan->user_charge + bo->size > nvdev->sa.size) return 1; - nvchan->user_charge += bo->size; if (!(flags & NOUVEAU_BO_GART)) return 1; @@ -341,6 +342,7 @@ nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, r = nouveau_bo_tmp(chan, bo->size, fence); if (!r) return 1; + nvchan->user_charge += bo->size; memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); -- cgit v1.2.3 From 5e091b573aa0a0c45f8ff34429f2a9d4198bb80a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 16 Feb 2008 21:27:53 +1100 Subject: nv40: ensure scissor gets disabled where necessary Fixes progs/demos/lodbias. Makes a complete mess of things, but now there's a motivation to finish this off :) --- src/gallium/drivers/nv40/nv40_context.h | 19 ++++++++- src/gallium/drivers/nv40/nv40_state.c | 23 +++++------ src/gallium/drivers/nv40/nv40_state_emit.c | 65 +++++++++++++++++++++++++++++- 3 files changed, 91 insertions(+), 16 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index f511759e3be..c4523112dbc 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -54,6 +54,11 @@ struct nv40_channel_context { struct nouveau_resource *vp_data_heap; }; +struct nv40_rasterizer_state { + struct pipe_rasterizer_state pipe; + struct nouveau_stateobj *so; +}; + struct nv40_context { struct pipe_context pipe; struct nouveau_winsys *nvws; @@ -63,7 +68,8 @@ struct nv40_context { int chipset; - uint32_t dirty; + unsigned dirty; + unsigned hw_dirty; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; @@ -71,14 +77,23 @@ struct nv40_context { unsigned fp_samplers; unsigned vp_samplers; + struct { + struct pipe_scissor_state scissor; + } pipe_state; + + struct { + unsigned scissor_enabled; + struct nouveau_stateobj *scissor; + } state; + struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; struct nouveau_stateobj *so_vtxbuf; struct nouveau_stateobj *so_blend; + struct nv40_rasterizer_state *rasterizer; struct nouveau_stateobj *so_rast; struct nouveau_stateobj *so_zsa; struct nouveau_stateobj *so_bcol; - struct nouveau_stateobj *so_scissor; struct nouveau_stateobj *so_viewport; struct nouveau_stateobj *so_stipple; diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 713f31dbb12..bb9b6b139f0 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -278,12 +278,12 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_rasterizer_state *rsso = MALLOC(sizeof(*rsso)); struct nouveau_stateobj *so = so_new(32, 0); /*XXX: ignored: * light_twoside * offset_cw/ccw -nohw - * scissor * point_smooth -nohw * multisample * offset_units / offset_scale @@ -362,24 +362,29 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, 0); } - return (void *)so; + rsso->so = so; + rsso->pipe = *cso; + return (void *)rsso; } static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_rasterizer_state *rsso = hwcso; - so_ref(hwcso, &nv40->so_rast); + so_ref(rsso->so, &nv40->so_rast); + nv40->rasterizer = rsso; nv40->dirty |= NV40_NEW_RAST; } static void nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nouveau_stateobj *so = hwcso; + struct nv40_rasterizer_state *rsso = hwcso; - so_ref(NULL, &so); + so_ref(NULL, &rsso->so); + free(rsso); } static void * @@ -723,14 +728,8 @@ nv40_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(3, 0); - so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); - so_data (so, ((s->maxx - s->minx) << 16) | s->minx); - so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - - so_ref(so, &nv40->so_scissor); - so_ref(NULL, &so); + nv40->pipe_state.scissor = *s; nv40->dirty |= NV40_NEW_SCISSOR; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index a10c9955480..74306fe22b3 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -22,9 +22,68 @@ nv40_state_emit_dummy_relocs(struct nv40_context *nv40) so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); } +static boolean +nv40_state_scissor_validate(struct nv40_context *nv40) +{ + struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct pipe_scissor_state *s = &nv40->pipe_state.scissor; + struct nouveau_stateobj *so; + + if (nv40->state.scissor && + (rast->scissor == 0 && nv40->state.scissor_enabled == 0)) + return FALSE; + + so = so_new(3, 0); + so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); + if (rast->scissor) { + so_data (so, ((s->maxx - s->minx) << 16) | s->minx); + so_data (so, ((s->maxy - s->miny) << 16) | s->miny); + } else { + so_data (so, 4096 << 16); + so_data (so, 4096 << 16); + } + + so_ref(so, &nv40->state.scissor); + so_ref(NULL, &so); + return TRUE; +} + +struct nv40_state_atom { + boolean (*validate)(struct nv40_context *nv40); + struct { + unsigned pipe; + unsigned hw; + } dirty; +}; + +static struct nv40_state_atom states[] = { + { + .validate = nv40_state_scissor_validate, + .dirty = { + .pipe = NV40_NEW_SCISSOR | NV40_NEW_RAST, + .hw = NV40_NEW_SCISSOR, + } + } +}; + +static void +nv40_state_validate(struct nv40_context *nv40) +{ + unsigned i; + + for (i = 0; i < sizeof(states) / sizeof(states[0]); i++) { + if (nv40->dirty & states[i].dirty.pipe) { + if (states[i].validate(nv40)) + nv40->hw_dirty |= states[i].dirty.hw; + } + } +} + void nv40_emit_hw_state(struct nv40_context *nv40) { + nv40_state_validate(nv40); + if (nv40->dirty & NV40_NEW_FB) so_emit(nv40->nvws, nv40->so_framebuffer); @@ -40,8 +99,10 @@ nv40_emit_hw_state(struct nv40_context *nv40) if (nv40->dirty & NV40_NEW_BCOL) so_emit(nv40->nvws, nv40->so_bcol); - if (nv40->dirty & NV40_NEW_SCISSOR) - so_emit(nv40->nvws, nv40->so_scissor); + if (nv40->hw_dirty & NV40_NEW_SCISSOR) { + so_emit(nv40->nvws, nv40->state.scissor); + nv40->hw_dirty &= ~NV40_NEW_SCISSOR; + } if (nv40->dirty & NV40_NEW_VIEWPORT) so_emit(nv40->nvws, nv40->so_viewport); -- cgit v1.2.3 From 12e0aa7b1d587b7c30897762d2f8f368a4a7d453 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 14:12:58 +1100 Subject: nv40: similar changes to polygon stipple as were done for scissor --- src/gallium/drivers/nv40/nv40_context.h | 18 ++++++++--- src/gallium/drivers/nv40/nv40_state.c | 9 +----- src/gallium/drivers/nv40/nv40_state_emit.c | 52 ++++++++++++++++++++++++++---- 3 files changed, 61 insertions(+), 18 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index c4523112dbc..7d5806f5f73 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -59,6 +59,18 @@ struct nv40_rasterizer_state { struct nouveau_stateobj *so; }; +struct nv40_state { + struct { + unsigned enabled; + struct nouveau_stateobj *so; + } scissor; + + struct { + unsigned enabled; + struct nouveau_stateobj *so; + } stipple; +}; + struct nv40_context { struct pipe_context pipe; struct nouveau_winsys *nvws; @@ -79,12 +91,10 @@ struct nv40_context { struct { struct pipe_scissor_state scissor; + unsigned stipple[32]; } pipe_state; - struct { - unsigned scissor_enabled; - struct nouveau_stateobj *scissor; - } state; + struct nv40_state state; struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index bb9b6b139f0..a36efd37f6a 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -711,15 +711,8 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(33, 0); - unsigned i; - so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); - for (i = 0; i < 32; i++) - so_data(so, stipple->stipple[i]); - - so_ref(so, &nv40->so_stipple); - so_ref(NULL, &so); + memcpy(nv40->pipe_state.stipple, stipple->stipple, 4 * 32); nv40->dirty |= NV40_NEW_STIPPLE; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 74306fe22b3..244c6838f34 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -29,8 +29,8 @@ nv40_state_scissor_validate(struct nv40_context *nv40) struct pipe_scissor_state *s = &nv40->pipe_state.scissor; struct nouveau_stateobj *so; - if (nv40->state.scissor && - (rast->scissor == 0 && nv40->state.scissor_enabled == 0)) + if (nv40->state.scissor.so && + (rast->scissor == 0 && nv40->state.scissor.enabled == 0)) return FALSE; so = so_new(3, 0); @@ -43,7 +43,38 @@ nv40_state_scissor_validate(struct nv40_context *nv40) so_data (so, 4096 << 16); } - so_ref(so, &nv40->state.scissor); + so_ref(so, &nv40->state.scissor.so); + so_ref(NULL, &so); + return TRUE; +} + +static boolean +nv40_state_stipple_validate(struct nv40_context *nv40) +{ + struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct nouveau_grobj *curie = nv40->hw->curie; + struct nouveau_stateobj *so; + + if (nv40->state.stipple.so && (rast->poly_stipple_enable == 0 && + nv40->state.stipple.enabled == 0)) + return FALSE; + + if (rast->poly_stipple_enable) { + unsigned i; + + so = so_new(35, 0); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, 1); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, nv40->pipe_state.stipple[i]); + } else { + so = so_new(2, 0); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, 0); + } + + so_ref(so, &nv40->state.stipple.so); so_ref(NULL, &so); return TRUE; } @@ -63,6 +94,13 @@ static struct nv40_state_atom states[] = { .pipe = NV40_NEW_SCISSOR | NV40_NEW_RAST, .hw = NV40_NEW_SCISSOR, } + }, + { + .validate = nv40_state_stipple_validate, + .dirty = { + .pipe = NV40_NEW_STIPPLE | NV40_NEW_RAST, + .hw = NV40_NEW_STIPPLE, + } } }; @@ -100,15 +138,17 @@ nv40_emit_hw_state(struct nv40_context *nv40) so_emit(nv40->nvws, nv40->so_bcol); if (nv40->hw_dirty & NV40_NEW_SCISSOR) { - so_emit(nv40->nvws, nv40->state.scissor); + so_emit(nv40->nvws, nv40->state.scissor.so); nv40->hw_dirty &= ~NV40_NEW_SCISSOR; } if (nv40->dirty & NV40_NEW_VIEWPORT) so_emit(nv40->nvws, nv40->so_viewport); - if (nv40->dirty & NV40_NEW_STIPPLE) - so_emit(nv40->nvws, nv40->so_stipple); + if (nv40->hw_dirty & NV40_NEW_STIPPLE) { + so_emit(nv40->nvws, nv40->state.stipple.so); + nv40->hw_dirty &= ~NV40_NEW_STIPPLE; + } if (nv40->dirty & NV40_NEW_FRAGPROG) { nv40_fragprog_bind(nv40, nv40->fragprog.current); -- cgit v1.2.3 From 037570fa5a6cf83d3aaaa6cdacc10eb0b5da45bf Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 15:17:39 +1100 Subject: nv40: obey polygon offset state Almost sure this isn't entirely correct. However, I'm not sure what gallium expects yet, and this fixes some bugs, so it'll do for now. --- src/gallium/drivers/nv40/nv40_state.c | 39 +++++++++++++++++++++++++++-------- 1 file changed, 30 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index a36efd37f6a..fec8f946c35 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -280,31 +280,30 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nv40_rasterizer_state *rsso = MALLOC(sizeof(*rsso)); struct nouveau_stateobj *so = so_new(32, 0); + struct nouveau_grobj *curie = nv40->hw->curie; /*XXX: ignored: * light_twoside - * offset_cw/ccw -nohw * point_smooth -nohw * multisample - * offset_units / offset_scale */ - so_method(so, nv40->hw->curie, NV40TCL_SHADE_MODEL, 1); + so_method(so, curie, NV40TCL_SHADE_MODEL, 1); so_data (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT : NV40TCL_SHADE_MODEL_SMOOTH); - so_method(so, nv40->hw->curie, NV40TCL_LINE_WIDTH, 2); + so_method(so, curie, NV40TCL_LINE_WIDTH, 2); so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); so_data (so, cso->line_smooth ? 1 : 0); - so_method(so, nv40->hw->curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); + so_method(so, curie, NV40TCL_LINE_STIPPLE_ENABLE, 2); so_data (so, cso->line_stipple_enable ? 1 : 0); so_data (so, (cso->line_stipple_pattern << 16) | cso->line_stipple_factor); - so_method(so, nv40->hw->curie, NV40TCL_POINT_SIZE, 1); + so_method(so, curie, NV40TCL_POINT_SIZE, 1); so_data (so, fui(cso->point_size)); - so_method(so, nv40->hw->curie, NV40TCL_POLYGON_MODE_FRONT, 6); + so_method(so, curie, NV40TCL_POLYGON_MODE_FRONT, 6); if (cso->front_winding == PIPE_WINDING_CCW) { so_data(so, nvgl_polygon_mode(cso->fill_ccw)); so_data(so, nvgl_polygon_mode(cso->fill_cw)); @@ -345,10 +344,32 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, cso->poly_smooth ? 1 : 0); so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); - so_method(so, nv40->hw->curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, cso->poly_stipple_enable ? 1 : 0); - so_method(so, nv40->hw->curie, NV40TCL_POINT_SPRITE, 1); + so_method(so, curie, 0x0a60, 3); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT)) + so_data(so, 1); + else + so_data(so, 0); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_LINE) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_LINE)) + so_data(so, 1); + else + so_data(so, 0); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_FILL) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_FILL)) + so_data(so, 1); + else + so_data(so, 0); + if (cso->offset_cw || cso->offset_ccw) { + so_method(so, curie, NV40TCL_POLYGON_OFFSET_FACTOR, 2); + so_data (so, fui(cso->offset_scale)); + so_data (so, fui(cso->offset_units * 2)); + } + + so_method(so, curie, NV40TCL_POINT_SPRITE, 1); if (cso->point_sprite) { unsigned psctl = (1 << 0), i; -- cgit v1.2.3 From f911235f64d610e57da88487133d0483c7a094e7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 15:31:55 +1100 Subject: nouveau: header update --- src/gallium/drivers/nouveau/nouveau_class.h | 4 ++-- src/gallium/drivers/nv40/nv40_state.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 5998945677b..c80461038b7 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -4926,9 +4926,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VIEWPORT_SCALE_Y 0x00000a34 #define NV40TCL_VIEWPORT_SCALE_Z 0x00000a38 #define NV40TCL_VIEWPORT_SCALE_W 0x00000a3c -#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a60 #define NV40TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV40TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV40TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a68 #define NV40TCL_DEPTH_FUNC 0x00000a6c #define NV40TCL_DEPTH_FUNC_NEVER 0x00000200 #define NV40TCL_DEPTH_FUNC_LESS 0x00000201 diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index fec8f946c35..d7379e90905 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -347,7 +347,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, cso->poly_stipple_enable ? 1 : 0); - so_method(so, curie, 0x0a60, 3); + so_method(so, curie, NV40TCL_POLYGON_OFFSET_POINT_ENABLE, 3); if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) || (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT)) so_data(so, 1); -- cgit v1.2.3 From 112ba3355a3fa53768efb9a9fb0eeb677bd28d47 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 16:26:33 +1100 Subject: nv40: until gallium is fixed we'll need a fallback for user clip planes --- src/gallium/drivers/nv40/nv40_context.h | 13 +++++++++++ src/gallium/drivers/nv40/nv40_state.c | 4 ++++ src/gallium/drivers/nv40/nv40_state_emit.c | 37 ++++++++++++++++++++++-------- 3 files changed, 45 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 7d5806f5f73..e9ed7ea3fb4 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -34,6 +34,9 @@ #define NV40_NEW_VERTPROG (1 << 9) #define NV40_NEW_FRAGPROG (1 << 10) #define NV40_NEW_ARRAYS (1 << 11) +#define NV40_NEW_UCP (1 << 12) + +#define NV40_FALLBACK_TNL (1 << 0) struct nv40_channel_context { struct nouveau_winsys *nvws; @@ -92,9 +95,11 @@ struct nv40_context { struct { struct pipe_scissor_state scissor; unsigned stipple[32]; + struct pipe_clip_state clip; } pipe_state; struct nv40_state state; + unsigned fallback; struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; @@ -131,6 +136,14 @@ nv40_context(struct pipe_context *pipe) return (struct nv40_context *)pipe; } +struct nv40_state_entry { + boolean (*validate)(struct nv40_context *nv40); + struct { + unsigned pipe; + unsigned hw; + } dirty; +}; + extern void nv40_init_state_functions(struct nv40_context *nv40); extern void nv40_init_surface_functions(struct nv40_context *nv40); extern void nv40_init_miptree_functions(struct nv40_context *nv40); diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index d7379e90905..c203b002403 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -561,6 +561,10 @@ static void nv40_set_clip_state(struct pipe_context *pipe, const struct pipe_clip_state *clip) { + struct nv40_context *nv40 = nv40_context(pipe); + + nv40->pipe_state.clip = *clip; + nv40->dirty |= NV40_NEW_UCP; } static void diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 244c6838f34..ce52a3863e7 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -79,15 +79,15 @@ nv40_state_stipple_validate(struct nv40_context *nv40) return TRUE; } -struct nv40_state_atom { - boolean (*validate)(struct nv40_context *nv40); - struct { - unsigned pipe; - unsigned hw; - } dirty; -}; +static boolean +nv40_state_clip_validate(struct nv40_context *nv40) +{ + if (nv40->pipe_state.clip.nr) + nv40->fallback |= NV40_FALLBACK_TNL; + return FALSE; +} -static struct nv40_state_atom states[] = { +static struct nv40_state_entry states[] = { { .validate = nv40_state_scissor_validate, .dirty = { @@ -101,13 +101,23 @@ static struct nv40_state_atom states[] = { .pipe = NV40_NEW_STIPPLE | NV40_NEW_RAST, .hw = NV40_NEW_STIPPLE, } + }, + { + .validate = nv40_state_clip_validate, + .dirty = { + .pipe = NV40_NEW_UCP, + .hw = 0, + } } }; static void nv40_state_validate(struct nv40_context *nv40) { - unsigned i; + unsigned i, last_fallback; + + last_fallback = nv40->fallback; + nv40->fallback = 0; for (i = 0; i < sizeof(states) / sizeof(states[0]); i++) { if (nv40->dirty & states[i].dirty.pipe) { @@ -115,6 +125,15 @@ nv40_state_validate(struct nv40_context *nv40) nv40->hw_dirty |= states[i].dirty.hw; } } + + if (nv40->fallback & NV40_FALLBACK_TNL && + !(last_fallback & NV40_FALLBACK_TNL)) { + NOUVEAU_ERR("XXX: hwtnl->swtnl\n"); + } else + if (last_fallback & NV40_FALLBACK_TNL && + !(nv40->fallback & NV40_FALLBACK_TNL)) { + NOUVEAU_ERR("XXX: swtnl->hwtnl\n"); + } } void -- cgit v1.2.3 From 56045da083d6530a56a2a7585e3121df0b07bac4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 16:38:27 +1100 Subject: nv40: move some things around --- src/gallium/drivers/nv40/Makefile | 3 + src/gallium/drivers/nv40/nv40_context.h | 3 + src/gallium/drivers/nv40/nv40_state_clip.c | 18 +++++ src/gallium/drivers/nv40/nv40_state_emit.c | 99 ++++----------------------- src/gallium/drivers/nv40/nv40_state_scissor.c | 35 ++++++++++ src/gallium/drivers/nv40/nv40_state_stipple.c | 40 +++++++++++ 6 files changed, 114 insertions(+), 84 deletions(-) create mode 100644 src/gallium/drivers/nv40/nv40_state_clip.c create mode 100644 src/gallium/drivers/nv40/nv40_state_scissor.c create mode 100644 src/gallium/drivers/nv40/nv40_state_stipple.c diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile index 2a9de4a2dcb..12b8eef2590 100644 --- a/src/gallium/drivers/nv40/Makefile +++ b/src/gallium/drivers/nv40/Makefile @@ -13,6 +13,9 @@ DRIVER_SOURCES = \ nv40_query.c \ nv40_state.c \ nv40_state_emit.c \ + nv40_state_clip.c \ + nv40_state_scissor.c \ + nv40_state_stipple.c \ nv40_surface.c \ nv40_vbo.c \ nv40_vertprog.c diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index e9ed7ea3fb4..432204b8252 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -174,6 +174,9 @@ extern void nv40_fragtex_bind(struct nv40_context *); /* nv40_state.c and friends */ extern void nv40_emit_hw_state(struct nv40_context *nv40); extern void nv40_state_tex_update(struct nv40_context *nv40); +extern struct nv40_state_entry nv40_state_clip; +extern struct nv40_state_entry nv40_state_scissor; +extern struct nv40_state_entry nv40_state_stipple; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_state_clip.c b/src/gallium/drivers/nv40/nv40_state_clip.c new file mode 100644 index 00000000000..19f1c3b36de --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_clip.c @@ -0,0 +1,18 @@ +#include "nv40_context.h" + +static boolean +nv40_state_clip_validate(struct nv40_context *nv40) +{ + if (nv40->pipe_state.clip.nr) + nv40->fallback |= NV40_FALLBACK_TNL; + + return FALSE; +} + +struct nv40_state_entry nv40_state_clip = { + .validate = nv40_state_clip_validate, + .dirty = { + .pipe = NV40_NEW_UCP, + .hw = 0 + } +}; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index ce52a3863e7..b5d0d68d6b1 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -22,63 +22,6 @@ nv40_state_emit_dummy_relocs(struct nv40_context *nv40) so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); } -static boolean -nv40_state_scissor_validate(struct nv40_context *nv40) -{ - struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; - struct pipe_scissor_state *s = &nv40->pipe_state.scissor; - struct nouveau_stateobj *so; - - if (nv40->state.scissor.so && - (rast->scissor == 0 && nv40->state.scissor.enabled == 0)) - return FALSE; - - so = so_new(3, 0); - so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); - if (rast->scissor) { - so_data (so, ((s->maxx - s->minx) << 16) | s->minx); - so_data (so, ((s->maxy - s->miny) << 16) | s->miny); - } else { - so_data (so, 4096 << 16); - so_data (so, 4096 << 16); - } - - so_ref(so, &nv40->state.scissor.so); - so_ref(NULL, &so); - return TRUE; -} - -static boolean -nv40_state_stipple_validate(struct nv40_context *nv40) -{ - struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; - struct nouveau_grobj *curie = nv40->hw->curie; - struct nouveau_stateobj *so; - - if (nv40->state.stipple.so && (rast->poly_stipple_enable == 0 && - nv40->state.stipple.enabled == 0)) - return FALSE; - - if (rast->poly_stipple_enable) { - unsigned i; - - so = so_new(35, 0); - so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); - so_data (so, 1); - so_method(so, curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); - for (i = 0; i < 32; i++) - so_data(so, nv40->pipe_state.stipple[i]); - } else { - so = so_new(2, 0); - so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); - so_data (so, 0); - } - - so_ref(so, &nv40->state.stipple.so); - so_ref(NULL, &so); - return TRUE; -} - static boolean nv40_state_clip_validate(struct nv40_context *nv40) { @@ -87,43 +30,31 @@ nv40_state_clip_validate(struct nv40_context *nv40) return FALSE; } -static struct nv40_state_entry states[] = { - { - .validate = nv40_state_scissor_validate, - .dirty = { - .pipe = NV40_NEW_SCISSOR | NV40_NEW_RAST, - .hw = NV40_NEW_SCISSOR, - } - }, - { - .validate = nv40_state_stipple_validate, - .dirty = { - .pipe = NV40_NEW_STIPPLE | NV40_NEW_RAST, - .hw = NV40_NEW_STIPPLE, - } - }, - { - .validate = nv40_state_clip_validate, - .dirty = { - .pipe = NV40_NEW_UCP, - .hw = 0, - } - } +static struct nv40_state_entry *render_states[] = { + &nv40_state_clip, + &nv40_state_scissor, + &nv40_state_stipple, + NULL }; static void nv40_state_validate(struct nv40_context *nv40) { - unsigned i, last_fallback; + struct nv40_state_entry **states = render_states; + unsigned last_fallback; last_fallback = nv40->fallback; nv40->fallback = 0; - for (i = 0; i < sizeof(states) / sizeof(states[0]); i++) { - if (nv40->dirty & states[i].dirty.pipe) { - if (states[i].validate(nv40)) - nv40->hw_dirty |= states[i].dirty.hw; + while (*states) { + struct nv40_state_entry *e = *states; + + if (nv40->dirty & e->dirty.pipe) { + if (e->validate(nv40)) + nv40->hw_dirty |= e->dirty.hw; } + + states++; } if (nv40->fallback & NV40_FALLBACK_TNL && diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c new file mode 100644 index 00000000000..556b820e581 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -0,0 +1,35 @@ +#include "nv40_context.h" + +static boolean +nv40_state_scissor_validate(struct nv40_context *nv40) +{ + struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct pipe_scissor_state *s = &nv40->pipe_state.scissor; + struct nouveau_stateobj *so; + + if (nv40->state.scissor.so && + (rast->scissor == 0 && nv40->state.scissor.enabled == 0)) + return FALSE; + + so = so_new(3, 0); + so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); + if (rast->scissor) { + so_data (so, ((s->maxx - s->minx) << 16) | s->minx); + so_data (so, ((s->maxy - s->miny) << 16) | s->miny); + } else { + so_data (so, 4096 << 16); + so_data (so, 4096 << 16); + } + + so_ref(so, &nv40->state.scissor.so); + so_ref(NULL, &so); + return TRUE; +} + +struct nv40_state_entry nv40_state_scissor = { + .validate = nv40_state_scissor_validate, + .dirty = { + .pipe = NV40_NEW_SCISSOR | NV40_NEW_RAST, + .hw = NV40_NEW_SCISSOR + } +}; diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c new file mode 100644 index 00000000000..52462a0b502 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_stipple.c @@ -0,0 +1,40 @@ +#include "nv40_context.h" + +static boolean +nv40_state_stipple_validate(struct nv40_context *nv40) +{ + struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct nouveau_grobj *curie = nv40->hw->curie; + struct nouveau_stateobj *so; + + if (nv40->state.stipple.so && (rast->poly_stipple_enable == 0 && + nv40->state.stipple.enabled == 0)) + return FALSE; + + if (rast->poly_stipple_enable) { + unsigned i; + + so = so_new(35, 0); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, 1); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, nv40->pipe_state.stipple[i]); + } else { + so = so_new(2, 0); + so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, 0); + } + + so_ref(so, &nv40->state.stipple.so); + so_ref(NULL, &so); + return TRUE; +} + +struct nv40_state_entry nv40_state_stipple = { + .validate = nv40_state_stipple_validate, + .dirty = { + .pipe = NV40_NEW_STIPPLE | NV40_NEW_RAST, + .hw = NV40_NEW_STIPPLE + } +}; -- cgit v1.2.3 From 4a9cb97bbf6961cc4106c4c54e59296a74e889e9 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 17:07:59 +1100 Subject: nv40: get fragprog onto new state mechanism --- src/gallium/drivers/nv40/nv40_context.h | 17 ++-- src/gallium/drivers/nv40/nv40_fragprog.c | 127 +++++++++++++++++------------ src/gallium/drivers/nv40/nv40_fragtex.c | 2 +- src/gallium/drivers/nv40/nv40_state.c | 5 +- src/gallium/drivers/nv40/nv40_state.h | 1 - src/gallium/drivers/nv40/nv40_state_emit.c | 26 +++--- 6 files changed, 97 insertions(+), 81 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 432204b8252..ce0933fea4a 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -37,6 +37,7 @@ #define NV40_NEW_UCP (1 << 12) #define NV40_FALLBACK_TNL (1 << 0) +#define NV40_FALLBACK_RAST (1 << 1) struct nv40_channel_context { struct nouveau_winsys *nvws; @@ -72,6 +73,8 @@ struct nv40_state { unsigned enabled; struct nouveau_stateobj *so; } stipple; + + struct nouveau_stateobj *fragprog; }; struct nv40_context { @@ -96,6 +99,8 @@ struct nv40_context { struct pipe_scissor_state scissor; unsigned stipple[32]; struct pipe_clip_state clip; + struct nv40_fragment_program *fragprog; + struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; } pipe_state; struct nv40_state state; @@ -119,13 +124,6 @@ struct nv40_context { struct pipe_buffer *constant_buf; } vertprog; - struct { - struct nv40_fragment_program *active; - - struct nv40_fragment_program *current; - struct pipe_buffer *constant_buf; - } fragprog; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; @@ -161,10 +159,6 @@ extern void nv40_vertprog_destroy(struct nv40_context *, struct nv40_vertex_program *); /* nv40_fragprog.c */ -extern void nv40_fragprog_translate(struct nv40_context *, - struct nv40_fragment_program *); -extern void nv40_fragprog_bind(struct nv40_context *, - struct nv40_fragment_program *); extern void nv40_fragprog_destroy(struct nv40_context *, struct nv40_fragment_program *); @@ -177,6 +171,7 @@ extern void nv40_state_tex_update(struct nv40_context *nv40); extern struct nv40_state_entry nv40_state_clip; extern struct nv40_state_entry nv40_state_scissor; extern struct nv40_state_entry nv40_state_stipple; +extern struct nv40_state_entry nv40_state_fragprog; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 07a418c1e9f..bfc75eb4620 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -668,7 +668,7 @@ nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, return TRUE; } -void +static void nv40_fragprog_translate(struct nv40_context *nv40, struct nv40_fragment_program *fp) { @@ -750,72 +750,66 @@ nv40_fragprog_translate(struct nv40_context *nv40, fp->insn[fpc->inst_offset + 3] = 0x00000000; fp->translated = TRUE; - fp->on_hw = FALSE; out_err: tgsi_parse_free(&parse); free(fpc); } -void -nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) +static void +nv40_fragprog_upload(struct nv40_context *nv40, + struct nv40_fragment_program *fp) { struct pipe_winsys *ws = nv40->pipe.winsys; - struct nouveau_stateobj *so; + const uint32_t le = 1; + uint32_t *map; int i; - if (!fp->translated) { - nv40_fragprog_translate(nv40, fp); - if (!fp->translated) - assert(0); - } + map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); - if (fp->nr_consts) { - float *map = ws->buffer_map(ws, nv40->fragprog.constant_buf, - PIPE_BUFFER_USAGE_CPU_READ); - for (i = 0; i < fp->nr_consts; i++) { - struct nv40_fragment_program_data *fpd = &fp->consts[i]; - uint32_t *p = &fp->insn[fpd->offset]; - uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; +#if 0 + for (i = 0; i < fp->insn_len; i++) { + NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + } +#endif - if (!memcmp(p, cb, 4 * sizeof(float))) - continue; - memcpy(p, cb, 4 * sizeof(float)); - fp->on_hw = 0; + if ((*(const uint8_t *)&le)) { + for (i = 0; i < fp->insn_len; i++) { + map[i] = fp->insn[i]; + } + } else { + /* Weird swapping for big-endian chips */ + for (i = 0; i < fp->insn_len; i++) { + map[i] = ((fp->insn[i] & 0xffff) << 16) | + ((fp->insn[i] >> 16) & 0xffff); } - ws->buffer_unmap(ws, nv40->fragprog.constant_buf); } - if (!fp->on_hw) { - const uint32_t le = 1; - uint32_t *map; - - if (!fp->buffer) - fp->buffer = ws->buffer_create(ws, 0x100, 0, - fp->insn_len * 4); - map = ws->buffer_map(ws, fp->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); + ws->buffer_unmap(ws, fp->buffer); +} -#if 0 - for (i = 0; i < fp->insn_len; i++) { - NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); - } -#endif +static boolean +nv40_fragprog_validate(struct nv40_context *nv40) +{ + struct nv40_fragment_program *fp = nv40->pipe_state.fragprog; + struct pipe_buffer *constbuf = + nv40->pipe_state.constbuf[PIPE_SHADER_FRAGMENT]; + struct pipe_winsys *ws = nv40->pipe.winsys; + struct nouveau_stateobj *so; + unsigned new_program = FALSE; + int i; - if ((*(const uint8_t *)&le)) { - for (i = 0; i < fp->insn_len; i++) { - map[i] = fp->insn[i]; - } - } else { - /* Weird swapping for big-endian chips */ - for (i = 0; i < fp->insn_len; i++) { - map[i] = ((fp->insn[i] & 0xffff) << 16) | - ((fp->insn[i] >> 16) & 0xffff); - } - } + if (fp->translated) + goto update_constants; - ws->buffer_unmap(ws, fp->buffer); - fp->on_hw = TRUE; + nv40_fragprog_translate(nv40, fp); + if (!fp->translated) { + nv40->fallback |= NV40_FALLBACK_RAST; + return FALSE; } + new_program = TRUE; + + fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); + nv40_fragprog_upload(nv40, fp); so = so_new(4, 1); so_method(so, nv40->hw->curie, NV40TCL_FP_ADDRESS, 1); @@ -824,12 +818,33 @@ nv40_fragprog_bind(struct nv40_context *nv40, struct nv40_fragment_program *fp) NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); so_data (so, fp->fp_control); - - so_emit(nv40->nvws, so); so_ref(so, &fp->so); so_ref(NULL, &so); - nv40->fragprog.active = fp; +update_constants: + if (fp->nr_consts) { + boolean new_consts = FALSE; + float *map; + + map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ); + for (i = 0; i < fp->nr_consts; i++) { + struct nv40_fragment_program_data *fpd = &fp->consts[i]; + uint32_t *p = &fp->insn[fpd->offset]; + uint32_t *cb = (uint32_t *)&map[fpd->index * 4]; + + if (!memcmp(p, cb, 4 * sizeof(float))) + continue; + memcpy(p, cb, 4 * sizeof(float)); + new_consts = TRUE; + } + ws->buffer_unmap(ws, constbuf); + + if (new_consts) + nv40_fragprog_upload(nv40, fp); + } + + so_ref(fp->so, &nv40->state.fragprog); + return new_program; } void @@ -840,3 +855,11 @@ nv40_fragprog_destroy(struct nv40_context *nv40, free(fp->insn); } +struct nv40_state_entry nv40_state_fragprog = { + .validate = nv40_fragprog_validate, + .dirty = { + .pipe = NV40_NEW_FRAGPROG, + .hw = NV40_NEW_FRAGPROG + } +}; + diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 5af5fbe7465..811f3098ba7 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -125,7 +125,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) void nv40_fragtex_bind(struct nv40_context *nv40) { - struct nv40_fragment_program *fp = nv40->fragprog.active; + struct nv40_fragment_program *fp = nv40->pipe_state.fragprog; unsigned samplers, unit; samplers = nv40->fp_samplers & ~fp->samplers; diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index c203b002403..2886c6b0dc4 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -523,9 +523,8 @@ static void nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_fragment_program *fp = hwcso; - nv40->fragprog.current = fp; + nv40->pipe_state.fragprog = hwcso; nv40->dirty |= NV40_NEW_FRAGPROG; } @@ -578,7 +577,7 @@ nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, nv40->dirty |= NV40_NEW_VERTPROG; } else if (shader == PIPE_SHADER_FRAGMENT) { - nv40->fragprog.constant_buf = buf->buffer; + nv40->pipe_state.constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; nv40->dirty |= NV40_NEW_FRAGPROG; } } diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index e82ab9de98a..2701294a077 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -50,7 +50,6 @@ struct nv40_fragment_program { const struct pipe_shader_state *pipe; boolean translated; - boolean on_hw; unsigned samplers; uint32_t *insn; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index b5d0d68d6b1..e10e178432e 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -19,21 +19,14 @@ nv40_state_emit_dummy_relocs(struct nv40_context *nv40) continue; so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); } - so_emit_reloc_markers(nv40->nvws, nv40->fragprog.active->so); -} - -static boolean -nv40_state_clip_validate(struct nv40_context *nv40) -{ - if (nv40->pipe_state.clip.nr) - nv40->fallback |= NV40_FALLBACK_TNL; - return FALSE; + so_emit_reloc_markers(nv40->nvws, nv40->state.fragprog); } static struct nv40_state_entry *render_states[] = { &nv40_state_clip, &nv40_state_scissor, &nv40_state_stipple, + &nv40_state_fragprog, NULL }; @@ -65,6 +58,15 @@ nv40_state_validate(struct nv40_context *nv40) !(nv40->fallback & NV40_FALLBACK_TNL)) { NOUVEAU_ERR("XXX: swtnl->hwtnl\n"); } + + if (nv40->fallback & NV40_FALLBACK_RAST && + !(last_fallback & NV40_FALLBACK_RAST)) { + NOUVEAU_ERR("XXX: hwrast->swrast\n"); + } else + if (last_fallback & NV40_FALLBACK_RAST && + !(nv40->fallback & NV40_FALLBACK_RAST)) { + NOUVEAU_ERR("XXX: swrast->hwrast\n"); + } } void @@ -100,10 +102,8 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->hw_dirty &= ~NV40_NEW_STIPPLE; } - if (nv40->dirty & NV40_NEW_FRAGPROG) { - nv40_fragprog_bind(nv40, nv40->fragprog.current); - /*XXX: clear NV40_NEW_FRAGPROG if no new program uploaded */ - } + if (nv40->hw_dirty & NV40_NEW_FRAGPROG) + so_emit(nv40->nvws, nv40->state.fragprog); if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { nv40_fragtex_bind(nv40); -- cgit v1.2.3 From 9a5dd26fe22c37b85787130a2b724e6e8ef4f553 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 17:36:24 +1100 Subject: nv40: and vertprog.. --- src/gallium/drivers/nv40/nv40_context.h | 14 ++------ src/gallium/drivers/nv40/nv40_fragprog.c | 10 +++--- src/gallium/drivers/nv40/nv40_state.c | 5 ++- src/gallium/drivers/nv40/nv40_state.h | 1 + src/gallium/drivers/nv40/nv40_state_emit.c | 3 +- src/gallium/drivers/nv40/nv40_vbo.c | 2 +- src/gallium/drivers/nv40/nv40_vertprog.c | 57 +++++++++++++++++++++--------- 7 files changed, 55 insertions(+), 37 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index ce0933fea4a..28a0274d771 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -75,6 +75,7 @@ struct nv40_state { } stipple; struct nouveau_stateobj *fragprog; + struct nouveau_stateobj *vertprog; }; struct nv40_context { @@ -99,6 +100,7 @@ struct nv40_context { struct pipe_scissor_state scissor; unsigned stipple[32]; struct pipe_clip_state clip; + struct nv40_vertex_program *vertprog; struct nv40_fragment_program *fragprog; struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; } pipe_state; @@ -117,13 +119,6 @@ struct nv40_context { struct nouveau_stateobj *so_viewport; struct nouveau_stateobj *so_stipple; - struct { - struct nv40_vertex_program *active; - - struct nv40_vertex_program *current; - struct pipe_buffer *constant_buf; - } vertprog; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; @@ -151,10 +146,6 @@ extern void nv40_init_query_functions(struct nv40_context *nv40); extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); /* nv40_vertprog.c */ -extern void nv40_vertprog_translate(struct nv40_context *, - struct nv40_vertex_program *); -extern void nv40_vertprog_bind(struct nv40_context *, - struct nv40_vertex_program *); extern void nv40_vertprog_destroy(struct nv40_context *, struct nv40_vertex_program *); @@ -172,6 +163,7 @@ extern struct nv40_state_entry nv40_state_clip; extern struct nv40_state_entry nv40_state_scissor; extern struct nv40_state_entry nv40_state_stipple; extern struct nv40_state_entry nv40_state_fragprog; +extern struct nv40_state_entry nv40_state_vertprog; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index bfc75eb4620..77ac8ab2c61 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -795,7 +795,6 @@ nv40_fragprog_validate(struct nv40_context *nv40) nv40->pipe_state.constbuf[PIPE_SHADER_FRAGMENT]; struct pipe_winsys *ws = nv40->pipe.winsys; struct nouveau_stateobj *so; - unsigned new_program = FALSE; int i; if (fp->translated) @@ -806,7 +805,6 @@ nv40_fragprog_validate(struct nv40_context *nv40) nv40->fallback |= NV40_FALLBACK_RAST; return FALSE; } - new_program = TRUE; fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); nv40_fragprog_upload(nv40, fp); @@ -843,8 +841,12 @@ update_constants: nv40_fragprog_upload(nv40, fp); } - so_ref(fp->so, &nv40->state.fragprog); - return new_program; + if (fp->so != nv40->state.fragprog) { + so_ref(fp->so, &nv40->state.fragprog); + return TRUE; + } + + return FALSE; } void diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 2886c6b0dc4..8ffbb131f73 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -491,9 +491,8 @@ static void nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_vertex_program *vp = hwcso; - nv40->vertprog.current = vp; + nv40->pipe_state.vertprog = hwcso; nv40->dirty |= NV40_NEW_VERTPROG; } @@ -573,7 +572,7 @@ nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, struct nv40_context *nv40 = nv40_context(pipe); if (shader == PIPE_SHADER_VERTEX) { - nv40->vertprog.constant_buf = buf->buffer; + nv40->pipe_state.constbuf[PIPE_SHADER_VERTEX] = buf->buffer; nv40->dirty |= NV40_NEW_VERTPROG; } else if (shader == PIPE_SHADER_FRAGMENT) { diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index 2701294a077..e5217fe91c4 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -39,6 +39,7 @@ struct nv40_vertex_program { uint32_t ir; uint32_t or; + struct nouveau_stateobj *so; }; struct nv40_fragment_program_data { diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index e10e178432e..e702b103236 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -27,6 +27,7 @@ static struct nv40_state_entry *render_states[] = { &nv40_state_scissor, &nv40_state_stipple, &nv40_state_fragprog, + &nv40_state_vertprog, NULL }; @@ -116,7 +117,7 @@ nv40_emit_hw_state(struct nv40_context *nv40) } if (nv40->dirty & NV40_NEW_VERTPROG) { - nv40_vertprog_bind(nv40, nv40->vertprog.current); + so_emit(nv40->nvws, nv40->state.vertprog); nv40->dirty &= ~NV40_NEW_VERTPROG; } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index fa827ef0c5d..3bfcb264db1 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -101,7 +101,7 @@ static void nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned ib_format) { - struct nv40_vertex_program *vp = nv40->vertprog.active; + struct nv40_vertex_program *vp = nv40->pipe_state.vertprog; struct nouveau_stateobj *vtxbuf, *vtxfmt; unsigned inputs, hw, num_hw; unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 9f4738b8308..4a15e51eb33 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -558,7 +558,7 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) return TRUE; } -void +static void nv40_vertprog_translate(struct nv40_context *nv40, struct nv40_vertex_program *vp) { @@ -631,24 +631,32 @@ out_err: free(vpc); } -void -nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) +static boolean +nv40_vertprog_validate(struct nv40_context *nv40) { + struct nv40_vertex_program *vp = nv40->pipe_state.vertprog; + struct pipe_buffer *constbuf = + nv40->pipe_state.constbuf[PIPE_SHADER_VERTEX]; struct nouveau_winsys *nvws = nv40->nvws; struct pipe_winsys *ws = nv40->pipe.winsys; boolean upload_code = FALSE, upload_data = FALSE; int i; /* Translate TGSI shader into hw bytecode */ + if (vp->translated) + goto check_gpu_resources; + + nv40_vertprog_translate(nv40, vp); if (!vp->translated) { - nv40_vertprog_translate(nv40, vp); - if (!vp->translated) - assert(0); + nv40->fallback |= NV40_FALLBACK_TNL; + return FALSE; } +check_gpu_resources: /* Allocate hw vtxprog exec slots */ if (!vp->exec) { struct nouveau_resource *heap = nv40->hw->vp_exec_heap; + struct nouveau_stateobj *so; uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { @@ -663,6 +671,15 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) assert(0); } + so = so_new(5, 0); + so_method(so, nv40->hw->curie, NV40TCL_VP_START_FROM_ID, 1); + so_data (so, vp->exec->start); + so_method(so, nv40->hw->curie, NV40TCL_VP_ATTRIB_EN, 2); + so_data (so, vp->ir); + so_data (so, vp->or); + so_ref(so, &vp->so); + so_ref(NULL, &so); + upload_code = TRUE; } @@ -725,8 +742,8 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) if (vp->nr_consts) { float *map = NULL; - if (nv40->vertprog.constant_buf) { - map = ws->buffer_map(ws, nv40->vertprog.constant_buf, + if (constbuf) { + map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ); } @@ -747,9 +764,8 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) OUT_RINGp ((uint32_t *)vpd->value, 4); } - if (map) { - ws->buffer_unmap(ws, nv40->vertprog.constant_buf); - } + if (constbuf) + ws->buffer_unmap(ws, constbuf); } /* Upload vtxprog */ @@ -770,13 +786,12 @@ nv40_vertprog_bind(struct nv40_context *nv40, struct nv40_vertex_program *vp) } } - BEGIN_RING(curie, NV40TCL_VP_START_FROM_ID, 1); - OUT_RING (vp->exec->start); - BEGIN_RING(curie, NV40TCL_VP_ATTRIB_EN, 2); - OUT_RING (vp->ir); - OUT_RING (vp->or); + if (vp->so != nv40->state.vertprog) { + so_ref(vp->so, &nv40->state.vertprog); + return TRUE; + } - nv40->vertprog.active = vp; + return FALSE; } void @@ -788,3 +803,11 @@ nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp) free(vp->insns); } +struct nv40_state_entry nv40_state_vertprog = { + .validate = nv40_vertprog_validate, + .dirty = { + .pipe = NV40_NEW_VERTPROG, + .hw = NV40_NEW_VERTPROG + } +}; + -- cgit v1.2.3 From c0f9cab905f3f54cc01bf947665f8a731b8cb347 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 18 Feb 2008 18:23:12 +1100 Subject: nouveau: cleanups + fixes --- src/gallium/drivers/nouveau/nouveau_stateobj.h | 21 +++++++++------------ src/gallium/drivers/nv40/nv40_state_emit.c | 8 +++++--- 2 files changed, 14 insertions(+), 15 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index 07c31b014a5..459cc7d77ae 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -45,22 +45,19 @@ so_new(unsigned push, unsigned reloc) static INLINE void so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) { - struct nouveau_stateobj *so; - - so = *pso; - if (so) { - if (--so->refcount <= 0) { - free(so->push); - free(so->reloc); - free(so); - } - *pso = NULL; - } + struct nouveau_stateobj *so = *pso; if (ref) { ref->refcount++; - *pso = ref; } + + if (so && --so->refcount <= 0) { + free(so->push); + free(so->reloc); + free(so); + } + + *pso = ref; } static INLINE void diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index e702b103236..e8230111bb4 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -103,8 +103,10 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->hw_dirty &= ~NV40_NEW_STIPPLE; } - if (nv40->hw_dirty & NV40_NEW_FRAGPROG) + if (nv40->hw_dirty & NV40_NEW_FRAGPROG) { so_emit(nv40->nvws, nv40->state.fragprog); + nv40->hw_dirty &= ~NV40_NEW_FRAGPROG; + } if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { nv40_fragtex_bind(nv40); @@ -116,9 +118,9 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->dirty &= ~NV40_NEW_FRAGPROG; } - if (nv40->dirty & NV40_NEW_VERTPROG) { + if (nv40->hw_dirty & NV40_NEW_VERTPROG) { so_emit(nv40->nvws, nv40->state.vertprog); - nv40->dirty &= ~NV40_NEW_VERTPROG; + nv40->hw_dirty &= ~NV40_NEW_VERTPROG; } nv40->dirty_samplers = 0; -- cgit v1.2.3 From 46c3d0918dd7a47f69c21e4eb1a3fd2a2fbe6223 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 20 Feb 2008 16:21:28 +1100 Subject: nv40: keep track of generated context state vs current channel state --- src/gallium/drivers/nv40/nv40_context.h | 26 ++++++----- src/gallium/drivers/nv40/nv40_fragprog.c | 6 +-- src/gallium/drivers/nv40/nv40_state_emit.c | 67 ++++++++++----------------- src/gallium/drivers/nv40/nv40_state_scissor.c | 8 ++-- src/gallium/drivers/nv40/nv40_state_stipple.c | 8 ++-- src/gallium/drivers/nv40/nv40_vertprog.c | 6 +-- 6 files changed, 53 insertions(+), 68 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 28a0274d771..2dd137ead0f 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -22,6 +22,15 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); +enum nv40_state_index { + NV40_STATE_CLIP = 1ULL, + NV40_STATE_SCISSOR = 2ULL, + NV40_STATE_STIPPLE = 3ULL, + NV40_STATE_FRAGPROG = 4ULL, + NV40_STATE_VERTPROG = 5ULL, + NV40_STATE_MAX = 6ULL +}; + #define NV40_NEW_BLEND (1 << 0) #define NV40_NEW_RAST (1 << 1) #define NV40_NEW_ZSA (1 << 2) @@ -56,6 +65,9 @@ struct nv40_channel_context { /* Vtxprog resources */ struct nouveau_resource *vp_exec_heap; struct nouveau_resource *vp_data_heap; + + /* Current 3D state of channel */ + struct nouveau_stateobj *state[NV40_STATE_MAX]; }; struct nv40_rasterizer_state { @@ -64,18 +76,10 @@ struct nv40_rasterizer_state { }; struct nv40_state { - struct { - unsigned enabled; - struct nouveau_stateobj *so; - } scissor; - - struct { - unsigned enabled; - struct nouveau_stateobj *so; - } stipple; + unsigned scissor_enabled; + unsigned stipple_enabled; - struct nouveau_stateobj *fragprog; - struct nouveau_stateobj *vertprog; + struct nouveau_stateobj *hw[NV40_STATE_MAX]; }; struct nv40_context { diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 77ac8ab2c61..db2613ef8bb 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -841,8 +841,8 @@ update_constants: nv40_fragprog_upload(nv40, fp); } - if (fp->so != nv40->state.fragprog) { - so_ref(fp->so, &nv40->state.fragprog); + if (fp->so != nv40->state.hw[NV40_STATE_FRAGPROG]) { + so_ref(fp->so, &nv40->state.hw[NV40_STATE_FRAGPROG]); return TRUE; } @@ -861,7 +861,7 @@ struct nv40_state_entry nv40_state_fragprog = { .validate = nv40_fragprog_validate, .dirty = { .pipe = NV40_NEW_FRAGPROG, - .hw = NV40_NEW_FRAGPROG + .hw = NV40_STATE_FRAGPROG } }; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index e8230111bb4..58beb72389a 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -1,27 +1,6 @@ #include "nv40_context.h" #include "nv40_state.h" -/* Emit relocs for every referenced buffer. - * - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. These relocs appear in the push buffer as - * NOPs, and will only be turned into state changes if a buffer - * actually moves. - */ -static void -nv40_state_emit_dummy_relocs(struct nv40_context *nv40) -{ - unsigned i; - - so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); - for (i = 0; i < 16; i++) { - if (!(nv40->fp_samplers & (1 << i))) - continue; - so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); - } - so_emit_reloc_markers(nv40->nvws, nv40->state.fragprog); -} - static struct nv40_state_entry *render_states[] = { &nv40_state_clip, &nv40_state_scissor, @@ -45,7 +24,7 @@ nv40_state_validate(struct nv40_context *nv40) if (nv40->dirty & e->dirty.pipe) { if (e->validate(nv40)) - nv40->hw_dirty |= e->dirty.hw; + nv40->hw_dirty |= (1 << e->dirty.hw); } states++; @@ -70,6 +49,28 @@ nv40_state_validate(struct nv40_context *nv40) } } +static void +nv40_state_emit(struct nv40_context *nv40) +{ + unsigned i; + + while (nv40->hw_dirty) { + unsigned idx = ffs(nv40->hw_dirty) - 1; + nv40->hw_dirty &= ~(1 << idx); + + so_ref (nv40->state.hw[idx], &nv40->hw->state[idx]); + so_emit(nv40->nvws, nv40->hw->state[idx]); + } + + so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); + for (i = 0; i < 16; i++) { + if (!(nv40->fp_samplers & (1 << i))) + continue; + so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); + } + so_emit_reloc_markers(nv40->nvws, nv40->state.hw[NV40_STATE_FRAGPROG]); +} + void nv40_emit_hw_state(struct nv40_context *nv40) { @@ -90,24 +91,9 @@ nv40_emit_hw_state(struct nv40_context *nv40) if (nv40->dirty & NV40_NEW_BCOL) so_emit(nv40->nvws, nv40->so_bcol); - if (nv40->hw_dirty & NV40_NEW_SCISSOR) { - so_emit(nv40->nvws, nv40->state.scissor.so); - nv40->hw_dirty &= ~NV40_NEW_SCISSOR; - } - if (nv40->dirty & NV40_NEW_VIEWPORT) so_emit(nv40->nvws, nv40->so_viewport); - if (nv40->hw_dirty & NV40_NEW_STIPPLE) { - so_emit(nv40->nvws, nv40->state.stipple.so); - nv40->hw_dirty &= ~NV40_NEW_STIPPLE; - } - - if (nv40->hw_dirty & NV40_NEW_FRAGPROG) { - so_emit(nv40->nvws, nv40->state.fragprog); - nv40->hw_dirty &= ~NV40_NEW_FRAGPROG; - } - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { nv40_fragtex_bind(nv40); @@ -118,14 +104,9 @@ nv40_emit_hw_state(struct nv40_context *nv40) nv40->dirty &= ~NV40_NEW_FRAGPROG; } - if (nv40->hw_dirty & NV40_NEW_VERTPROG) { - so_emit(nv40->nvws, nv40->state.vertprog); - nv40->hw_dirty &= ~NV40_NEW_VERTPROG; - } + nv40_state_emit(nv40); nv40->dirty_samplers = 0; nv40->dirty = 0; - - nv40_state_emit_dummy_relocs(nv40); } diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c index 556b820e581..dc7b6d3a9d8 100644 --- a/src/gallium/drivers/nv40/nv40_state_scissor.c +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -7,8 +7,8 @@ nv40_state_scissor_validate(struct nv40_context *nv40) struct pipe_scissor_state *s = &nv40->pipe_state.scissor; struct nouveau_stateobj *so; - if (nv40->state.scissor.so && - (rast->scissor == 0 && nv40->state.scissor.enabled == 0)) + if (nv40->state.hw[NV40_STATE_SCISSOR] && + (rast->scissor == 0 && nv40->state.scissor_enabled == 0)) return FALSE; so = so_new(3, 0); @@ -21,7 +21,7 @@ nv40_state_scissor_validate(struct nv40_context *nv40) so_data (so, 4096 << 16); } - so_ref(so, &nv40->state.scissor.so); + so_ref(so, &nv40->state.hw[NV40_STATE_SCISSOR]); so_ref(NULL, &so); return TRUE; } @@ -30,6 +30,6 @@ struct nv40_state_entry nv40_state_scissor = { .validate = nv40_state_scissor_validate, .dirty = { .pipe = NV40_NEW_SCISSOR | NV40_NEW_RAST, - .hw = NV40_NEW_SCISSOR + .hw = NV40_STATE_SCISSOR } }; diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c index 52462a0b502..1b0b194432e 100644 --- a/src/gallium/drivers/nv40/nv40_state_stipple.c +++ b/src/gallium/drivers/nv40/nv40_state_stipple.c @@ -7,8 +7,8 @@ nv40_state_stipple_validate(struct nv40_context *nv40) struct nouveau_grobj *curie = nv40->hw->curie; struct nouveau_stateobj *so; - if (nv40->state.stipple.so && (rast->poly_stipple_enable == 0 && - nv40->state.stipple.enabled == 0)) + if (nv40->state.hw[NV40_STATE_STIPPLE] && + (rast->poly_stipple_enable == 0 && nv40->state.stipple_enabled == 0)) return FALSE; if (rast->poly_stipple_enable) { @@ -26,7 +26,7 @@ nv40_state_stipple_validate(struct nv40_context *nv40) so_data (so, 0); } - so_ref(so, &nv40->state.stipple.so); + so_ref(so, &nv40->state.hw[NV40_STATE_STIPPLE]); so_ref(NULL, &so); return TRUE; } @@ -35,6 +35,6 @@ struct nv40_state_entry nv40_state_stipple = { .validate = nv40_state_stipple_validate, .dirty = { .pipe = NV40_NEW_STIPPLE | NV40_NEW_RAST, - .hw = NV40_NEW_STIPPLE + .hw = NV40_STATE_STIPPLE, } }; diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 4a15e51eb33..8a2d2336974 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -786,8 +786,8 @@ check_gpu_resources: } } - if (vp->so != nv40->state.vertprog) { - so_ref(vp->so, &nv40->state.vertprog); + if (vp->so != nv40->state.hw[NV40_STATE_VERTPROG]) { + so_ref(vp->so, &nv40->state.hw[NV40_STATE_VERTPROG]); return TRUE; } @@ -807,7 +807,7 @@ struct nv40_state_entry nv40_state_vertprog = { .validate = nv40_vertprog_validate, .dirty = { .pipe = NV40_NEW_VERTPROG, - .hw = NV40_NEW_VERTPROG + .hw = NV40_STATE_VERTPROG, } }; -- cgit v1.2.3 From 9cd10d7618a226fe46395b08beb19e420bc14a4f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 20 Feb 2008 17:14:41 +1100 Subject: nv40: almost there.. --- src/gallium/drivers/nv40/Makefile | 6 ++- src/gallium/drivers/nv40/nv40_context.h | 66 +++++++++++++++++++----- src/gallium/drivers/nv40/nv40_state.c | 62 +++++++++------------- src/gallium/drivers/nv40/nv40_state_blend.c | 41 +++++++++++++++ src/gallium/drivers/nv40/nv40_state_emit.c | 20 ++----- src/gallium/drivers/nv40/nv40_state_rasterizer.c | 17 ++++++ src/gallium/drivers/nv40/nv40_state_scissor.c | 2 +- src/gallium/drivers/nv40/nv40_state_stipple.c | 2 +- src/gallium/drivers/nv40/nv40_state_viewport.c | 30 +++++++++++ src/gallium/drivers/nv40/nv40_state_zsa.c | 17 ++++++ 10 files changed, 194 insertions(+), 69 deletions(-) create mode 100644 src/gallium/drivers/nv40/nv40_state_blend.c create mode 100644 src/gallium/drivers/nv40/nv40_state_rasterizer.c create mode 100644 src/gallium/drivers/nv40/nv40_state_viewport.c create mode 100644 src/gallium/drivers/nv40/nv40_state_zsa.c diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile index 12b8eef2590..82295cbefca 100644 --- a/src/gallium/drivers/nv40/Makefile +++ b/src/gallium/drivers/nv40/Makefile @@ -12,10 +12,14 @@ DRIVER_SOURCES = \ nv40_miptree.c \ nv40_query.c \ nv40_state.c \ - nv40_state_emit.c \ + nv40_state_blend.c \ nv40_state_clip.c \ + nv40_state_emit.c \ + nv40_state_rasterizer.c \ nv40_state_scissor.c \ nv40_state_stipple.c \ + nv40_state_viewport.c \ + nv40_state_zsa.c \ nv40_surface.c \ nv40_vbo.c \ nv40_vertprog.c diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 2dd137ead0f..d71fe11a4ab 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -23,12 +23,38 @@ fprintf(stderr, "nouveau: "fmt, ##args); enum nv40_state_index { - NV40_STATE_CLIP = 1ULL, - NV40_STATE_SCISSOR = 2ULL, - NV40_STATE_STIPPLE = 3ULL, - NV40_STATE_FRAGPROG = 4ULL, - NV40_STATE_VERTPROG = 5ULL, - NV40_STATE_MAX = 6ULL + NV40_STATE_FB = 0, + NV40_STATE_VIEWPORT = 1, + NV40_STATE_BLEND = 2, + NV40_STATE_RAST = 3, + NV40_STATE_ZSA = 4, + NV40_STATE_BCOL = 5, + NV40_STATE_CLIP = 6, + NV40_STATE_SCISSOR = 7, + NV40_STATE_STIPPLE = 8, + NV40_STATE_FRAGPROG = 9, + NV40_STATE_VERTPROG = 10, + NV40_STATE_FRAGTEX0 = 11, + NV40_STATE_FRAGTEX1 = 12, + NV40_STATE_FRAGTEX2 = 13, + NV40_STATE_FRAGTEX3 = 14, + NV40_STATE_FRAGTEX4 = 15, + NV40_STATE_FRAGTEX5 = 16, + NV40_STATE_FRAGTEX6 = 17, + NV40_STATE_FRAGTEX7 = 18, + NV40_STATE_FRAGTEX8 = 19, + NV40_STATE_FRAGTEX9 = 20, + NV40_STATE_FRAGTEX10 = 21, + NV40_STATE_FRAGTEX11 = 22, + NV40_STATE_FRAGTEX12 = 23, + NV40_STATE_FRAGTEX13 = 24, + NV40_STATE_FRAGTEX14 = 25, + NV40_STATE_FRAGTEX15 = 26, + NV40_STATE_VERTTEX0 = 27, + NV40_STATE_VERTTEX1 = 28, + NV40_STATE_VERTTEX2 = 29, + NV40_STATE_VERTTEX3 = 30, + NV40_STATE_MAX = 31 }; #define NV40_NEW_BLEND (1 << 0) @@ -75,6 +101,17 @@ struct nv40_rasterizer_state { struct nouveau_stateobj *so; }; +struct nv40_zsa_state { + struct pipe_depth_stencil_alpha_state pipe; + struct nouveau_stateobj *so; +}; + +struct nv40_blend_state { + struct pipe_blend_state pipe; + struct nouveau_stateobj *so; +}; + + struct nv40_state { unsigned scissor_enabled; unsigned stipple_enabled; @@ -107,6 +144,11 @@ struct nv40_context { struct nv40_vertex_program *vertprog; struct nv40_fragment_program *fragprog; struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + struct nv40_rasterizer_state *rasterizer; + struct nv40_zsa_state *zsa; + struct nv40_blend_state *blend; + struct pipe_blend_color blend_colour; + struct pipe_viewport_state viewport; } pipe_state; struct nv40_state state; @@ -115,13 +157,6 @@ struct nv40_context { struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; struct nouveau_stateobj *so_vtxbuf; - struct nouveau_stateobj *so_blend; - struct nv40_rasterizer_state *rasterizer; - struct nouveau_stateobj *so_rast; - struct nouveau_stateobj *so_zsa; - struct nouveau_stateobj *so_bcol; - struct nouveau_stateobj *so_viewport; - struct nouveau_stateobj *so_stipple; struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; @@ -164,10 +199,15 @@ extern void nv40_fragtex_bind(struct nv40_context *); extern void nv40_emit_hw_state(struct nv40_context *nv40); extern void nv40_state_tex_update(struct nv40_context *nv40); extern struct nv40_state_entry nv40_state_clip; +extern struct nv40_state_entry nv40_state_rasterizer; extern struct nv40_state_entry nv40_state_scissor; extern struct nv40_state_entry nv40_state_stipple; extern struct nv40_state_entry nv40_state_fragprog; extern struct nv40_state_entry nv40_state_vertprog; +extern struct nv40_state_entry nv40_state_blend; +extern struct nv40_state_entry nv40_state_blend_colour; +extern struct nv40_state_entry nv40_state_zsa; +extern struct nv40_state_entry nv40_state_viewport; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 8ffbb131f73..41631ef2dd6 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -11,6 +11,7 @@ nv40_blend_state_create(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_grobj *curie = nv40->hw->curie; + struct nv40_blend_state *bso = MALLOC(sizeof(*bso)); struct nouveau_stateobj *so = so_new(16, 0); if (cso->blend_enable) { @@ -46,7 +47,9 @@ nv40_blend_state_create(struct pipe_context *pipe, so_method(so, curie, NV40TCL_DITHER_ENABLE, 1); so_data (so, cso->dither ? 1 : 0); - return (void *)so; + bso->so = so; + bso->pipe = *cso; + return (void *)bso; } static void @@ -54,16 +57,17 @@ nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_ref(hwcso, &nv40->so_blend); + nv40->pipe_state.blend = hwcso; nv40->dirty |= NV40_NEW_BLEND; } static void nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nouveau_stateobj *so = hwcso; + struct nv40_blend_state *bso = hwcso; - so_ref(NULL, &so); + so_ref(NULL, &bso->so); + FREE(bso); } @@ -260,7 +264,7 @@ nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, static void nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + FREE(hwcso); } static void @@ -392,10 +396,8 @@ static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_rasterizer_state *rsso = hwcso; - so_ref(rsso->so, &nv40->so_rast); - nv40->rasterizer = rsso; + nv40->pipe_state.rasterizer = hwcso; nv40->dirty |= NV40_NEW_RAST; } @@ -405,7 +407,7 @@ nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) struct nv40_rasterizer_state *rsso = hwcso; so_ref(NULL, &rsso->so); - free(rsso); + FREE(rsso); } static void * @@ -413,6 +415,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_zsa_state *zsaso = MALLOC(sizeof(*zsaso)); struct nouveau_stateobj *so = so_new(32, 0); so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); @@ -455,7 +458,9 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, 0); } - return (void *)so; + zsaso->so = so; + zsaso->pipe = *cso; + return (void *)zsaso; } static void @@ -463,16 +468,17 @@ nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - so_ref(hwcso, &nv40->so_zsa); + nv40->pipe_state.zsa = hwcso; nv40->dirty |= NV40_NEW_ZSA; } static void nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nouveau_stateobj *so = hwcso; + struct nv40_zsa_state *zsaso = hwcso; - so_ref(NULL, &so); + so_ref(NULL, &zsaso->so); + FREE(zsaso); } static void * @@ -503,7 +509,7 @@ nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv40_vertex_program *vp = hwcso; nv40_vertprog_destroy(nv40, vp); - free(vp); + FREE(vp); } static void * @@ -534,7 +540,7 @@ nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv40_fragment_program *fp = hwcso; nv40_fragprog_destroy(nv40, fp); - free(fp); + FREE(fp); } static void @@ -542,16 +548,8 @@ nv40_set_blend_color(struct pipe_context *pipe, const struct pipe_blend_color *bcol) { struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(2, 0); - - so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); - so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0))); - so_ref(so, &nv40->so_bcol); - so_ref(NULL, &so); + nv40->pipe_state.blend_colour = *bcol; nv40->dirty |= NV40_NEW_BCOL; } @@ -754,20 +752,8 @@ nv40_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) { struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_stateobj *so = so_new(9, 0); - - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); - so_data (so, fui(vpt->translate[0])); - so_data (so, fui(vpt->translate[1])); - so_data (so, fui(vpt->translate[2])); - so_data (so, fui(vpt->translate[3])); - so_data (so, fui(vpt->scale[0])); - so_data (so, fui(vpt->scale[1])); - so_data (so, fui(vpt->scale[2])); - so_data (so, fui(vpt->scale[3])); - - so_ref(so, &nv40->so_viewport); - so_ref(NULL, &so); + + nv40->pipe_state.viewport = *vpt; nv40->dirty |= NV40_NEW_VIEWPORT; } diff --git a/src/gallium/drivers/nv40/nv40_state_blend.c b/src/gallium/drivers/nv40/nv40_state_blend.c new file mode 100644 index 00000000000..b12f8b03dd8 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_blend.c @@ -0,0 +1,41 @@ +#include "nv40_context.h" + +static boolean +nv40_state_blend_validate(struct nv40_context *nv40) +{ + so_ref(nv40->pipe_state.blend->so, &nv40->state.hw[NV40_STATE_BLEND]); + return TRUE; +} + +struct nv40_state_entry nv40_state_blend = { + .validate = nv40_state_blend_validate, + .dirty = { + .pipe = NV40_NEW_BLEND, + .hw = NV40_STATE_BLEND + } +}; + +static boolean +nv40_state_blend_colour_validate(struct nv40_context *nv40) +{ + struct nouveau_stateobj *so = so_new(2, 0); + struct pipe_blend_color *bcol = &nv40->pipe_state.blend_colour; + + so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); + so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0))); + + so_ref(so, &nv40->state.hw[NV40_STATE_BCOL]); + so_ref(NULL, &so); + return TRUE; +} + +struct nv40_state_entry nv40_state_blend_colour = { + .validate = nv40_state_blend_colour_validate, + .dirty = { + .pipe = NV40_NEW_BCOL, + .hw = NV40_STATE_BCOL + } +}; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 58beb72389a..65d7e2978a8 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -2,11 +2,16 @@ #include "nv40_state.h" static struct nv40_state_entry *render_states[] = { + &nv40_state_rasterizer, &nv40_state_clip, &nv40_state_scissor, &nv40_state_stipple, &nv40_state_fragprog, &nv40_state_vertprog, + &nv40_state_blend, + &nv40_state_blend_colour, + &nv40_state_zsa, + &nv40_state_viewport, NULL }; @@ -79,21 +84,6 @@ nv40_emit_hw_state(struct nv40_context *nv40) if (nv40->dirty & NV40_NEW_FB) so_emit(nv40->nvws, nv40->so_framebuffer); - if (nv40->dirty & NV40_NEW_BLEND) - so_emit(nv40->nvws, nv40->so_blend); - - if (nv40->dirty & NV40_NEW_RAST) - so_emit(nv40->nvws, nv40->so_rast); - - if (nv40->dirty & NV40_NEW_ZSA) - so_emit(nv40->nvws, nv40->so_zsa); - - if (nv40->dirty & NV40_NEW_BCOL) - so_emit(nv40->nvws, nv40->so_bcol); - - if (nv40->dirty & NV40_NEW_VIEWPORT) - so_emit(nv40->nvws, nv40->so_viewport); - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { nv40_fragtex_bind(nv40); diff --git a/src/gallium/drivers/nv40/nv40_state_rasterizer.c b/src/gallium/drivers/nv40/nv40_state_rasterizer.c new file mode 100644 index 00000000000..59b35d1d50a --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_rasterizer.c @@ -0,0 +1,17 @@ +#include "nv40_context.h" + +static boolean +nv40_state_rasterizer_validate(struct nv40_context *nv40) +{ + so_ref(nv40->pipe_state.rasterizer->so, + &nv40->state.hw[NV40_STATE_RAST]); + return TRUE; +} + +struct nv40_state_entry nv40_state_rasterizer = { + .validate = nv40_state_rasterizer_validate, + .dirty = { + .pipe = NV40_NEW_RAST, + .hw = NV40_STATE_RAST + } +}; diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c index dc7b6d3a9d8..2871fa2516d 100644 --- a/src/gallium/drivers/nv40/nv40_state_scissor.c +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -3,7 +3,7 @@ static boolean nv40_state_scissor_validate(struct nv40_context *nv40) { - struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct pipe_rasterizer_state *rast = &nv40->pipe_state.rasterizer->pipe; struct pipe_scissor_state *s = &nv40->pipe_state.scissor; struct nouveau_stateobj *so; diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c index 1b0b194432e..bd163582a39 100644 --- a/src/gallium/drivers/nv40/nv40_state_stipple.c +++ b/src/gallium/drivers/nv40/nv40_state_stipple.c @@ -3,7 +3,7 @@ static boolean nv40_state_stipple_validate(struct nv40_context *nv40) { - struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct pipe_rasterizer_state *rast = &nv40->pipe_state.rasterizer->pipe; struct nouveau_grobj *curie = nv40->hw->curie; struct nouveau_stateobj *so; diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c new file mode 100644 index 00000000000..79fcc31a8bf --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -0,0 +1,30 @@ +#include "nv40_context.h" + +static boolean +nv40_state_viewport_validate(struct nv40_context *nv40) +{ + struct nouveau_stateobj *so = so_new(9, 0); + struct pipe_viewport_state *vpt = &nv40->pipe_state.viewport; + + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(vpt->translate[0])); + so_data (so, fui(vpt->translate[1])); + so_data (so, fui(vpt->translate[2])); + so_data (so, fui(vpt->translate[3])); + so_data (so, fui(vpt->scale[0])); + so_data (so, fui(vpt->scale[1])); + so_data (so, fui(vpt->scale[2])); + so_data (so, fui(vpt->scale[3])); + + so_ref(so, &nv40->state.hw[NV40_STATE_VIEWPORT]); + so_ref(NULL, &so); + return TRUE; +} + +struct nv40_state_entry nv40_state_viewport = { + .validate = nv40_state_viewport_validate, + .dirty = { + .pipe = NV40_NEW_VIEWPORT, + .hw = NV40_STATE_VIEWPORT + } +}; diff --git a/src/gallium/drivers/nv40/nv40_state_zsa.c b/src/gallium/drivers/nv40/nv40_state_zsa.c new file mode 100644 index 00000000000..061a3555cbe --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_zsa.c @@ -0,0 +1,17 @@ +#include "nv40_context.h" + +static boolean +nv40_state_zsa_validate(struct nv40_context *nv40) +{ + so_ref(nv40->pipe_state.zsa->so, + &nv40->state.hw[NV40_STATE_ZSA]); + return TRUE; +} + +struct nv40_state_entry nv40_state_zsa = { + .validate = nv40_state_zsa_validate, + .dirty = { + .pipe = NV40_NEW_ZSA, + .hw = NV40_STATE_ZSA + } +}; -- cgit v1.2.3 From 759fa5fcc8038af4845a6d9c57b75933ef26559c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 20 Feb 2008 17:22:40 +1100 Subject: nv40: fb state --- src/gallium/drivers/nv40/Makefile | 1 + src/gallium/drivers/nv40/nv40_context.h | 3 +- src/gallium/drivers/nv40/nv40_state.c | 140 +------------------------- src/gallium/drivers/nv40/nv40_state_emit.c | 6 +- src/gallium/drivers/nv40/nv40_state_fb.c | 156 +++++++++++++++++++++++++++++ 5 files changed, 162 insertions(+), 144 deletions(-) create mode 100644 src/gallium/drivers/nv40/nv40_state_fb.c diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile index 82295cbefca..fd002b54e78 100644 --- a/src/gallium/drivers/nv40/Makefile +++ b/src/gallium/drivers/nv40/Makefile @@ -15,6 +15,7 @@ DRIVER_SOURCES = \ nv40_state_blend.c \ nv40_state_clip.c \ nv40_state_emit.c \ + nv40_state_fb.c \ nv40_state_rasterizer.c \ nv40_state_scissor.c \ nv40_state_stipple.c \ diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index d71fe11a4ab..69062a8a202 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -149,12 +149,12 @@ struct nv40_context { struct nv40_blend_state *blend; struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; + struct pipe_framebuffer_state framebuffer; } pipe_state; struct nv40_state state; unsigned fallback; - struct nouveau_stateobj *so_framebuffer; struct nouveau_stateobj *so_fragtex[16]; struct nouveau_stateobj *so_vtxbuf; @@ -208,6 +208,7 @@ extern struct nv40_state_entry nv40_state_blend; extern struct nv40_state_entry nv40_state_blend_colour; extern struct nv40_state_entry nv40_state_zsa; extern struct nv40_state_entry nv40_state_viewport; +extern struct nv40_state_entry nv40_state_framebuffer; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 41631ef2dd6..84818d6729e 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -584,146 +584,8 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv40_context *nv40 = nv40_context(pipe); - struct pipe_surface *rt[4], *zeta; - uint32_t rt_enable, rt_format, w, h; - int i, colour_format = 0, zeta_format = 0; - struct nouveau_stateobj *so = so_new(64, 10); - unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; - - rt_enable = 0; - for (i = 0; i < 4; i++) { - if (!fb->cbufs[i]) - continue; - - if (colour_format) { - assert(w == fb->cbufs[i]->width); - assert(h == fb->cbufs[i]->height); - assert(colour_format == fb->cbufs[i]->format); - } else { - w = fb->cbufs[i]->width; - h = fb->cbufs[i]->height; - colour_format = fb->cbufs[i]->format; - rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); - rt[i] = fb->cbufs[i]; - } - } - - if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | - NV40TCL_RT_ENABLE_COLOR3)) - rt_enable |= NV40TCL_RT_ENABLE_MRT; - - if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - - zeta_format = fb->zsbuf->format; - zeta = fb->zsbuf; - } - - rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; - - switch (colour_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case 0: - rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; - break; - case PIPE_FORMAT_R5G6B5_UNORM: - rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; - break; - default: - assert(0); - } - - switch (zeta_format) { - case PIPE_FORMAT_Z16_UNORM: - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; - break; - case PIPE_FORMAT_Z24S8_UNORM: - case 0: - rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; - break; - default: - assert(0); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); - so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); - so_data (so, rt[0]->pitch * rt[0]->cpp); - so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); - so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); - so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_data (so, rt[1]->pitch * rt[1]->cpp); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); - so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); - so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); - so_data (so, rt[2]->pitch * rt[2]->cpp); - } - - if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); - so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); - so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); - so_data (so, rt[3]->pitch * rt[3]->cpp); - } - - if (zeta_format) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); - so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv40->nvws->channel->vram->handle, - nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); - so_reloc (so, zeta->buffer, zeta->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); - so_data (so, zeta->pitch * zeta->cpp); - } - so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); - so_data (so, rt_enable); - so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); - so_data (so, (w << 16) | 0); - so_data (so, (h << 16) | 0); - so_data (so, rt_format); - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); - so_data (so, (w << 16) | 0); - so_data (so, (h << 16) | 0); - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); - so_data (so, ((w - 1) << 16) | 0); - so_data (so, ((h - 1) << 16) | 0); - - so_ref(so, &nv40->so_framebuffer); - so_ref(NULL, &so); + nv40->pipe_state.framebuffer = *fb; nv40->dirty |= NV40_NEW_FB; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 65d7e2978a8..a9ca71c5e93 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -2,6 +2,7 @@ #include "nv40_state.h" static struct nv40_state_entry *render_states[] = { + &nv40_state_framebuffer, &nv40_state_rasterizer, &nv40_state_clip, &nv40_state_scissor, @@ -67,7 +68,7 @@ nv40_state_emit(struct nv40_context *nv40) so_emit(nv40->nvws, nv40->hw->state[idx]); } - so_emit_reloc_markers(nv40->nvws, nv40->so_framebuffer); + so_emit_reloc_markers(nv40->nvws, nv40->state.hw[NV40_STATE_FB]); for (i = 0; i < 16; i++) { if (!(nv40->fp_samplers & (1 << i))) continue; @@ -81,9 +82,6 @@ nv40_emit_hw_state(struct nv40_context *nv40) { nv40_state_validate(nv40); - if (nv40->dirty & NV40_NEW_FB) - so_emit(nv40->nvws, nv40->so_framebuffer); - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { nv40_fragtex_bind(nv40); diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c new file mode 100644 index 00000000000..d3032f1be5a --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -0,0 +1,156 @@ +#include "nv40_context.h" + +static boolean +nv40_state_framebuffer_validate(struct nv40_context *nv40) +{ + struct pipe_framebuffer_state *fb = &nv40->pipe_state.framebuffer; + struct pipe_surface *rt[4], *zeta; + uint32_t rt_enable, rt_format, w, h; + int i, colour_format = 0, zeta_format = 0; + struct nouveau_stateobj *so = so_new(64, 10); + unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; + + rt_enable = 0; + for (i = 0; i < 4; i++) { + if (!fb->cbufs[i]) + continue; + + if (colour_format) { + assert(w == fb->cbufs[i]->width); + assert(h == fb->cbufs[i]->height); + assert(colour_format == fb->cbufs[i]->format); + } else { + w = fb->cbufs[i]->width; + h = fb->cbufs[i]->height; + colour_format = fb->cbufs[i]->format; + rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); + rt[i] = fb->cbufs[i]; + } + } + + if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 | + NV40TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV40TCL_RT_ENABLE_MRT; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + switch (zeta_format) { + case PIPE_FORMAT_Z16_UNORM: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16; + break; + case PIPE_FORMAT_Z24S8_UNORM: + case 0: + rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8; + break; + default: + assert(0); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); + so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); + so_data (so, rt[0]->pitch * rt[0]->cpp); + so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); + so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); + so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, rt[1]->pitch * rt[1]->cpp); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); + so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); + so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); + so_data (so, rt[2]->pitch * rt[2]->cpp); + } + + if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); + so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); + so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); + so_data (so, rt[3]->pitch * rt[3]->cpp); + } + + if (zeta_format) { + so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); + so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv40->nvws->channel->vram->handle, + nv40->nvws->channel->gart->handle); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); + so_reloc (so, zeta->buffer, zeta->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); + so_data (so, zeta->pitch * zeta->cpp); + } + + so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); + so_data (so, rt_enable); + so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_data (so, rt_format); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_data (so, ((w - 1) << 16) | 0); + so_data (so, ((h - 1) << 16) | 0); + + so_ref(so, &nv40->state.hw[NV40_STATE_FB]); + so_ref(NULL, &so); + return TRUE; +} + +struct nv40_state_entry nv40_state_framebuffer = { + .validate = nv40_state_framebuffer_validate, + .dirty = { + .pipe = NV40_NEW_FB, + .hw = NV40_STATE_FB + } +}; -- cgit v1.2.3 From 73e0e567dea3cf4e1591acb3e894eecef812f367 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 22 Feb 2008 12:36:48 +1100 Subject: nouveau: fix build --- configs/default | 2 +- src/gallium/drivers/nv40/nv40_miptree.c | 38 +++++++++++++++++++++++++++++++++ src/gallium/drivers/nv40/nv40_surface.c | 31 --------------------------- 3 files changed, 39 insertions(+), 32 deletions(-) diff --git a/configs/default b/configs/default index 48ddd29282b..af9ff6f4261 100644 --- a/configs/default +++ b/configs/default @@ -70,7 +70,7 @@ PROGRAM_DIRS = demos redbook samples glsl xdemos # Gallium directories and GALLIUM_AUXILIARY_DIRS = draw cso_cache pipebuffer tgsi rtasm util GALLIUM_AUXILIARIES = $(foreach DIR,$(GALLIUM_AUXILIARY_DIRS),$(TOP)/src/gallium/auxiliary/$(DIR)/lib$(DIR).a) -GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple failover +GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv30 nv40 nv50 failover GALLIUM_DRIVERS = $(foreach DIR,$(GALLIUM_DRIVER_DIRS),$(TOP)/src/gallium/drivers/$(DIR)/lib$(DIR).a) GALLIUM_WINSYS_DIRS = xlib diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 92e6b3a43df..5e1c7ade31b 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -63,6 +63,7 @@ nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) if (!mt) return NULL; mt->base = *pt; + mt->base.refcount = 1; nv40_miptree_layout(mt); mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, @@ -95,10 +96,47 @@ nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) } } +static void +nv40_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) +{ +} + +static struct pipe_surface * +nv40_miptree_surface(struct pipe_context *pipe, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_winsys *ws = pipe->winsys; + struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv40mt->level[level].pitch / ps->cpp; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv40mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv40mt->level[level].image_offset[zslice]; + } else { + ps->offset = nv40mt->level[level].image_offset[0]; + } + + return ps; +} + void nv40_init_miptree_functions(struct nv40_context *nv40) { nv40->pipe.texture_create = nv40_miptree_create; nv40->pipe.texture_release = nv40_miptree_release; + nv40->pipe.texture_update = nv40_miptree_update; + nv40->pipe.get_tex_surface = nv40_miptree_surface; } diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c index 9726ab4e4dc..df5d7abdbfc 100644 --- a/src/gallium/drivers/nv40/nv40_surface.c +++ b/src/gallium/drivers/nv40/nv40_surface.c @@ -73,36 +73,6 @@ nv40_surface_format_supported(struct pipe_context *pipe, return FALSE; } -static struct pipe_surface * -nv40_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) -{ - struct pipe_winsys *ws = pipe->winsys; - struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; - struct pipe_surface *ps; - - ps = ws->surface_alloc(ws); - if (!ps) - return NULL; - pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); - ps->format = pt->format; - ps->cpp = pt->cpp; - ps->width = pt->width[level]; - ps->height = pt->height[level]; - ps->pitch = nv40mt->level[level].pitch / ps->cpp; - - if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv40mt->level[level].image_offset[face]; - } else - if (pt->target == PIPE_TEXTURE_3D) { - ps->offset = nv40mt->level[level].image_offset[zslice]; - } else { - ps->offset = nv40mt->level[level].image_offset[0]; - } - - return ps; -} - static void nv40_surface_copy(struct pipe_context *pipe, unsigned do_flip, struct pipe_surface *dest, unsigned destx, unsigned desty, @@ -131,7 +101,6 @@ void nv40_init_surface_functions(struct nv40_context *nv40) { nv40->pipe.is_format_supported = nv40_surface_format_supported; - nv40->pipe.get_tex_surface = nv40_get_tex_surface; nv40->pipe.surface_copy = nv40_surface_copy; nv40->pipe.surface_fill = nv40_surface_fill; } -- cgit v1.2.3 From 5b2ff28a2fd3bb0ca9df569edcaf80e8141ccaa1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 22 Feb 2008 13:32:51 +1100 Subject: nv40: rework fragment texture state --- src/gallium/drivers/nv40/nv40_context.h | 2 +- src/gallium/drivers/nv40/nv40_fragtex.c | 36 +++++++++++++++++++++--------- src/gallium/drivers/nv40/nv40_state.c | 2 ++ src/gallium/drivers/nv40/nv40_state_emit.c | 22 +++++++----------- 4 files changed, 36 insertions(+), 26 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 69062a8a202..cbc798fbd6e 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -155,7 +155,6 @@ struct nv40_context { struct nv40_state state; unsigned fallback; - struct nouveau_stateobj *so_fragtex[16]; struct nouveau_stateobj *so_vtxbuf; struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; @@ -209,6 +208,7 @@ extern struct nv40_state_entry nv40_state_blend_colour; extern struct nv40_state_entry nv40_state_zsa; extern struct nv40_state_entry nv40_state_viewport; extern struct nv40_state_entry nv40_state_framebuffer; +extern struct nv40_state_entry nv40_state_fragtex; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 811f3098ba7..826d4a9478a 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -52,7 +52,7 @@ nv40_fragtex_format(uint pipe_format) } -static void +static struct nouveau_stateobj * nv40_fragtex_build(struct nv40_context *nv40, int unit) { struct nv40_sampler_state *ps = nv40->tex_sampler[unit]; @@ -90,7 +90,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) break; default: NOUVEAU_ERR("Unknown target %d\n", pt->target); - return; + return NULL; } if (swizzled) { @@ -117,15 +117,14 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_method(so, nv40->hw->curie, NV40TCL_TEX_SIZE1(unit), 1); so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); - so_emit(nv40->nvws, so); - so_ref (so, &nv40->so_fragtex[unit]); - so_ref (NULL, &so); + return so; } -void -nv40_fragtex_bind(struct nv40_context *nv40) +static boolean +nv40_fragtex_validate(struct nv40_context *nv40) { struct nv40_fragment_program *fp = nv40->pipe_state.fragprog; + struct nouveau_stateobj *so; unsigned samplers, unit; samplers = nv40->fp_samplers & ~fp->samplers; @@ -133,9 +132,12 @@ nv40_fragtex_bind(struct nv40_context *nv40) unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - so_ref(NULL, &nv40->so_fragtex[unit]); - BEGIN_RING(curie, NV40TCL_TEX_ENABLE(unit), 1); - OUT_RING (0); + so = so_new(2, 0); + so_method(so, nv40->hw->curie, NV40TCL_TEX_ENABLE(unit), 1); + so_data (so, 0); + so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); + so_ref(NULL, &so); + nv40->hw_dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); } samplers = nv40->dirty_samplers & fp->samplers; @@ -143,9 +145,21 @@ nv40_fragtex_bind(struct nv40_context *nv40) unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - nv40_fragtex_build(nv40, unit); + so = nv40_fragtex_build(nv40, unit); + so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); + so_ref(NULL, &so); + nv40->hw_dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); } nv40->fp_samplers = fp->samplers; + return FALSE; } +struct nv40_state_entry nv40_state_fragtex = { + .validate = nv40_fragtex_validate, + .dirty = { + .pipe = NV40_NEW_SAMPLER | NV40_NEW_FRAGPROG, + .hw = 0 + } +}; + diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 84818d6729e..74cbabb023e 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -259,6 +259,7 @@ nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, nv40->tex_sampler[unit] = ps; nv40->dirty_samplers |= (1 << unit); + nv40->dirty |= NV40_NEW_SAMPLER; } static void @@ -275,6 +276,7 @@ nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, nv40->tex_miptree[unit] = (struct nv40_miptree *)miptree; nv40->dirty_samplers |= (1 << unit); + nv40->dirty |= NV40_NEW_SAMPLER; } static void * diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index a9ca71c5e93..6d87b7b52b2 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -8,6 +8,7 @@ static struct nv40_state_entry *render_states[] = { &nv40_state_scissor, &nv40_state_stipple, &nv40_state_fragprog, + &nv40_state_fragtex, &nv40_state_vertprog, &nv40_state_blend, &nv40_state_blend_colour, @@ -35,6 +36,7 @@ nv40_state_validate(struct nv40_context *nv40) states++; } + nv40->dirty = 0; if (nv40->fallback & NV40_FALLBACK_TNL && !(last_fallback & NV40_FALLBACK_TNL)) { @@ -72,7 +74,8 @@ nv40_state_emit(struct nv40_context *nv40) for (i = 0; i < 16; i++) { if (!(nv40->fp_samplers & (1 << i))) continue; - so_emit_reloc_markers(nv40->nvws, nv40->so_fragtex[i]); + so_emit_reloc_markers(nv40->nvws, + nv40->state.hw[NV40_STATE_FRAGTEX0+i]); } so_emit_reloc_markers(nv40->nvws, nv40->state.hw[NV40_STATE_FRAGPROG]); } @@ -81,20 +84,11 @@ void nv40_emit_hw_state(struct nv40_context *nv40) { nv40_state_validate(nv40); - - if (nv40->dirty_samplers || (nv40->dirty & NV40_NEW_FRAGPROG)) { - nv40_fragtex_bind(nv40); - - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (1); - nv40->dirty &= ~NV40_NEW_FRAGPROG; - } - nv40_state_emit(nv40); - nv40->dirty_samplers = 0; - nv40->dirty = 0; + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (1); } -- cgit v1.2.3 From c2e36bdd1a58ba6f58c4e72db1f7f64e8bd05901 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 22 Feb 2008 13:55:18 +1100 Subject: nv40: move hw_dirty --- src/gallium/drivers/nv40/nv40_context.h | 5 ++--- src/gallium/drivers/nv40/nv40_fragtex.c | 9 +++++---- src/gallium/drivers/nv40/nv40_state_emit.c | 24 ++++++++++++------------ 3 files changed, 19 insertions(+), 19 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index cbc798fbd6e..c533b9e0ef9 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -115,7 +115,9 @@ struct nv40_blend_state { struct nv40_state { unsigned scissor_enabled; unsigned stipple_enabled; + unsigned fp_samplers; + unsigned dirty; struct nouveau_stateobj *hw[NV40_STATE_MAX]; }; @@ -129,13 +131,10 @@ struct nv40_context { int chipset; unsigned dirty; - unsigned hw_dirty; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; unsigned dirty_samplers; - unsigned fp_samplers; - unsigned vp_samplers; struct { struct pipe_scissor_state scissor; diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 826d4a9478a..3d27a9bf134 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -124,10 +124,11 @@ static boolean nv40_fragtex_validate(struct nv40_context *nv40) { struct nv40_fragment_program *fp = nv40->pipe_state.fragprog; + struct nv40_state *state = &nv40->state; struct nouveau_stateobj *so; unsigned samplers, unit; - samplers = nv40->fp_samplers & ~fp->samplers; + samplers = state->fp_samplers & ~fp->samplers; while (samplers) { unit = ffs(samplers) - 1; samplers &= ~(1 << unit); @@ -137,7 +138,7 @@ nv40_fragtex_validate(struct nv40_context *nv40) so_data (so, 0); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); so_ref(NULL, &so); - nv40->hw_dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); + state->dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); } samplers = nv40->dirty_samplers & fp->samplers; @@ -148,10 +149,10 @@ nv40_fragtex_validate(struct nv40_context *nv40) so = nv40_fragtex_build(nv40, unit); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); so_ref(NULL, &so); - nv40->hw_dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); + state->dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); } - nv40->fp_samplers = fp->samplers; + nv40->state.fp_samplers = fp->samplers; return FALSE; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 6d87b7b52b2..af09ed47d6e 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -31,7 +31,7 @@ nv40_state_validate(struct nv40_context *nv40) if (nv40->dirty & e->dirty.pipe) { if (e->validate(nv40)) - nv40->hw_dirty |= (1 << e->dirty.hw); + nv40->state.dirty |= (1 << e->dirty.hw); } states++; @@ -60,24 +60,24 @@ nv40_state_validate(struct nv40_context *nv40) static void nv40_state_emit(struct nv40_context *nv40) { - unsigned i; + struct nv40_state *state = &nv40->state; + unsigned i, samplers; - while (nv40->hw_dirty) { - unsigned idx = ffs(nv40->hw_dirty) - 1; - nv40->hw_dirty &= ~(1 << idx); + while (state->dirty) { + unsigned idx = ffs(state->dirty) - 1; - so_ref (nv40->state.hw[idx], &nv40->hw->state[idx]); + so_ref (state->hw[idx], &nv40->hw->state[idx]); so_emit(nv40->nvws, nv40->hw->state[idx]); + state->dirty &= ~(1 << idx); } - so_emit_reloc_markers(nv40->nvws, nv40->state.hw[NV40_STATE_FB]); - for (i = 0; i < 16; i++) { - if (!(nv40->fp_samplers & (1 << i))) - continue; + so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FB]); + for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { so_emit_reloc_markers(nv40->nvws, - nv40->state.hw[NV40_STATE_FRAGTEX0+i]); + state->hw[NV40_STATE_FRAGTEX0+i]); + samplers &= ~(1 << i); } - so_emit_reloc_markers(nv40->nvws, nv40->state.hw[NV40_STATE_FRAGPROG]); + so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FRAGPROG]); } void -- cgit v1.2.3 From 7b938431d0ab5ccce1e7e2b1c38e1dcbdc6001e8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 22 Feb 2008 14:46:48 +1100 Subject: nv40: stateobj start out with 0 refcount --- src/gallium/drivers/nouveau/nouveau_stateobj.h | 2 +- src/gallium/drivers/nv40/nv40_fragprog.c | 1 - src/gallium/drivers/nv40/nv40_fragtex.c | 2 -- src/gallium/drivers/nv40/nv40_state.c | 12 ++++++------ src/gallium/drivers/nv40/nv40_state_blend.c | 1 - src/gallium/drivers/nv40/nv40_state_fb.c | 1 - src/gallium/drivers/nv40/nv40_state_scissor.c | 1 - src/gallium/drivers/nv40/nv40_state_stipple.c | 1 - src/gallium/drivers/nv40/nv40_state_viewport.c | 1 - src/gallium/drivers/nv40/nv40_vbo.c | 1 - src/gallium/drivers/nv40/nv40_vertprog.c | 1 - 11 files changed, 7 insertions(+), 17 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index 459cc7d77ae..439c7e4734a 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -32,7 +32,7 @@ so_new(unsigned push, unsigned reloc) struct nouveau_stateobj *so; so = MALLOC(sizeof(struct nouveau_stateobj)); - so->refcount = 1; + so->refcount = 0; so->push = MALLOC(sizeof(unsigned) * push); so->reloc = MALLOC(sizeof(struct nouveau_stateobj_reloc) * reloc); diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index db2613ef8bb..a4a1ea01e03 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -817,7 +817,6 @@ nv40_fragprog_validate(struct nv40_context *nv40) so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); so_data (so, fp->fp_control); so_ref(so, &fp->so); - so_ref(NULL, &so); update_constants: if (fp->nr_consts) { diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 3d27a9bf134..c8a8120f305 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -137,7 +137,6 @@ nv40_fragtex_validate(struct nv40_context *nv40) so_method(so, nv40->hw->curie, NV40TCL_TEX_ENABLE(unit), 1); so_data (so, 0); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); - so_ref(NULL, &so); state->dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); } @@ -148,7 +147,6 @@ nv40_fragtex_validate(struct nv40_context *nv40) so = nv40_fragtex_build(nv40, unit); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); - so_ref(NULL, &so); state->dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); } diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 74cbabb023e..107e60f1798 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -11,7 +11,7 @@ nv40_blend_state_create(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_grobj *curie = nv40->hw->curie; - struct nv40_blend_state *bso = MALLOC(sizeof(*bso)); + struct nv40_blend_state *bso = CALLOC(1, sizeof(*bso)); struct nouveau_stateobj *so = so_new(16, 0); if (cso->blend_enable) { @@ -47,7 +47,7 @@ nv40_blend_state_create(struct pipe_context *pipe, so_method(so, curie, NV40TCL_DITHER_ENABLE, 1); so_data (so, cso->dither ? 1 : 0); - bso->so = so; + so_ref(so, &bso->so); bso->pipe = *cso; return (void *)bso; } @@ -284,7 +284,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_rasterizer_state *rsso = MALLOC(sizeof(*rsso)); + struct nv40_rasterizer_state *rsso = CALLOC(1, sizeof(*rsso)); struct nouveau_stateobj *so = so_new(32, 0); struct nouveau_grobj *curie = nv40->hw->curie; @@ -389,7 +389,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, 0); } - rsso->so = so; + so_ref(so, &rsso->so); rsso->pipe = *cso; return (void *)rsso; } @@ -417,7 +417,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_zsa_state *zsaso = MALLOC(sizeof(*zsaso)); + struct nv40_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso)); struct nouveau_stateobj *so = so_new(32, 0); so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); @@ -460,7 +460,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, 0); } - zsaso->so = so; + so_ref(so, &zsaso->so); zsaso->pipe = *cso; return (void *)zsaso; } diff --git a/src/gallium/drivers/nv40/nv40_state_blend.c b/src/gallium/drivers/nv40/nv40_state_blend.c index b12f8b03dd8..81b927a67a9 100644 --- a/src/gallium/drivers/nv40/nv40_state_blend.c +++ b/src/gallium/drivers/nv40/nv40_state_blend.c @@ -28,7 +28,6 @@ nv40_state_blend_colour_validate(struct nv40_context *nv40) (float_to_ubyte(bcol->color[2]) << 0))); so_ref(so, &nv40->state.hw[NV40_STATE_BCOL]); - so_ref(NULL, &so); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index d3032f1be5a..c3bf4d43a33 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -143,7 +143,6 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) so_data (so, ((h - 1) << 16) | 0); so_ref(so, &nv40->state.hw[NV40_STATE_FB]); - so_ref(NULL, &so); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c index 2871fa2516d..ee797094d39 100644 --- a/src/gallium/drivers/nv40/nv40_state_scissor.c +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -22,7 +22,6 @@ nv40_state_scissor_validate(struct nv40_context *nv40) } so_ref(so, &nv40->state.hw[NV40_STATE_SCISSOR]); - so_ref(NULL, &so); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c index bd163582a39..aad4d179ac4 100644 --- a/src/gallium/drivers/nv40/nv40_state_stipple.c +++ b/src/gallium/drivers/nv40/nv40_state_stipple.c @@ -27,7 +27,6 @@ nv40_state_stipple_validate(struct nv40_context *nv40) } so_ref(so, &nv40->state.hw[NV40_STATE_STIPPLE]); - so_ref(NULL, &so); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c index 79fcc31a8bf..68820d31339 100644 --- a/src/gallium/drivers/nv40/nv40_state_viewport.c +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -17,7 +17,6 @@ nv40_state_viewport_validate(struct nv40_context *nv40) so_data (so, fui(vpt->scale[3])); so_ref(so, &nv40->state.hw[NV40_STATE_VIEWPORT]); - so_ref(NULL, &so); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 3bfcb264db1..5abe4c9af10 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -159,7 +159,6 @@ nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, so_emit(nv40->nvws, vtxfmt); so_emit(nv40->nvws, vtxbuf); so_ref (vtxbuf, &nv40->so_vtxbuf); - so_ref (NULL, &vtxbuf); so_ref (NULL, &vtxfmt); } diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 8a2d2336974..c482964adc0 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -678,7 +678,6 @@ check_gpu_resources: so_data (so, vp->ir); so_data (so, vp->or); so_ref(so, &vp->so); - so_ref(NULL, &so); upload_code = TRUE; } -- cgit v1.2.3 From 0a5ed0667e6bd934a150bf7f784349fa7f595309 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 22 Feb 2008 17:18:27 +0100 Subject: nv30: wrong number of parameters --- src/gallium/drivers/nv30/nv30_state.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 53368561e07..db4dac24c74 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -609,7 +609,7 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 2); + BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 1); OUT_RING (rt[1]->pitch * rt[1]->cpp); nv30->rt[1] = rt[1]->buffer; } -- cgit v1.2.3 From d8a9d850b9d63c7398d596fad2dfd2f05e55ef7d Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 22 Feb 2008 22:21:15 +0100 Subject: nv30: init zeta to NULL, use color pitch if no zeta --- src/gallium/drivers/nv30/nv30_state.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index db4dac24c74..aa3fe7837e4 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -537,7 +537,7 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv30_context *nv30 = nv30_context(pipe); - struct pipe_surface *rt[4], *zeta; + struct pipe_surface *rt[4], *zeta = NULL; uint32_t rt_enable, rt_format, w, h; int i, colour_format = 0, zeta_format = 0; @@ -603,8 +603,15 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { + uint32_t pitch = rt[0]->pitch * rt[0]->cpp; + if (zeta) { + pitch |= (zeta->pitch * zeta->cpp)<<16; + } else { + pitch |= pitch<<16; + } + BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); - OUT_RING ( (rt[0]->pitch * rt[0]->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + OUT_RING ( pitch ); nv30->rt[0] = rt[0]->buffer; } -- cgit v1.2.3 From 14de997d5df48512c751c627ab19d486691f591d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 25 Feb 2008 12:26:48 +1100 Subject: nv40: dump meaningful names for surface formats --- src/gallium/drivers/nv40/nv40_fragtex.c | 3 +++ src/gallium/drivers/nv40/nv40_vbo.c | 6 +++++- 2 files changed, 8 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index c8a8120f305..7adee8858d0 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -41,6 +41,7 @@ static struct nv40_texture_format * nv40_fragtex_format(uint pipe_format) { struct nv40_texture_format *tf = nv40_texture_formats; + char fs[128]; while (tf->defined) { if (tf->pipe == pipe_format) @@ -48,6 +49,8 @@ nv40_fragtex_format(uint pipe_format) tf++; } + pf_sprint_name(fs, pipe_format); + NOUVEAU_ERR("unknown texture format %s\n", fs); return NULL; } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 5abe4c9af10..753c2fe64b9 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -30,9 +30,13 @@ nv40_vbo_type(uint format) case PIPE_FORMAT_TYPE_UNORM: return NV40TCL_VTXFMT_TYPE_UBYTE; default: - NOUVEAU_ERR("Unknown format 0x%08x\n", format); + { + char fs[128]; + pf_sprint_name(fs, format); + NOUVEAU_ERR("Unknown format %s\n", fs); return NV40TCL_VTXFMT_TYPE_FLOAT; } + } } static boolean -- cgit v1.2.3 From 4058a9012764ce3bc7b90d03c4d79d020540f8e4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 25 Feb 2008 13:29:16 +1100 Subject: nv40: construct vbo state the same way as the rest --- src/gallium/drivers/nv40/nv40_context.h | 11 +- src/gallium/drivers/nv40/nv40_fragtex.c | 4 +- src/gallium/drivers/nv40/nv40_state.c | 2 - src/gallium/drivers/nv40/nv40_state_emit.c | 10 +- src/gallium/drivers/nv40/nv40_vbo.c | 246 ++++++++++++++--------------- 5 files changed, 137 insertions(+), 136 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index c533b9e0ef9..110d9d7ab70 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -54,7 +54,9 @@ enum nv40_state_index { NV40_STATE_VERTTEX1 = 28, NV40_STATE_VERTTEX2 = 29, NV40_STATE_VERTTEX3 = 30, - NV40_STATE_MAX = 31 + NV40_STATE_VTXBUF = 31, + NV40_STATE_VTXFMT = 32, + NV40_STATE_MAX = 33 }; #define NV40_NEW_BLEND (1 << 0) @@ -117,7 +119,7 @@ struct nv40_state { unsigned stipple_enabled; unsigned fp_samplers; - unsigned dirty; + uint64_t dirty; struct nouveau_stateobj *hw[NV40_STATE_MAX]; }; @@ -149,13 +151,13 @@ struct nv40_context { struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; + struct pipe_buffer *idxbuf; + unsigned idxbuf_format; } pipe_state; struct nv40_state state; unsigned fallback; - struct nouveau_stateobj *so_vtxbuf; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; @@ -208,6 +210,7 @@ extern struct nv40_state_entry nv40_state_zsa; extern struct nv40_state_entry nv40_state_viewport; extern struct nv40_state_entry nv40_state_framebuffer; extern struct nv40_state_entry nv40_state_fragtex; +extern struct nv40_state_entry nv40_state_vbo; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 7adee8858d0..ed47d707b2f 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -140,7 +140,7 @@ nv40_fragtex_validate(struct nv40_context *nv40) so_method(so, nv40->hw->curie, NV40TCL_TEX_ENABLE(unit), 1); so_data (so, 0); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); - state->dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); + state->dirty |= (1ULL << (NV40_STATE_FRAGTEX0 + unit)); } samplers = nv40->dirty_samplers & fp->samplers; @@ -150,7 +150,7 @@ nv40_fragtex_validate(struct nv40_context *nv40) so = nv40_fragtex_build(nv40, unit); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); - state->dirty |= (1 << (NV40_STATE_FRAGTEX0 + unit)); + state->dirty |= (1ULL << (NV40_STATE_FRAGTEX0 + unit)); } nv40->state.fp_samplers = fp->samplers; diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 107e60f1798..e6f2754dc51 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -628,7 +628,6 @@ nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, struct nv40_context *nv40 = nv40_context(pipe); nv40->vtxbuf[index] = *vb; - nv40->dirty |= NV40_NEW_ARRAYS; } @@ -639,7 +638,6 @@ nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, struct nv40_context *nv40 = nv40_context(pipe); nv40->vtxelt[index] = *ve; - nv40->dirty |= NV40_NEW_ARRAYS; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index af09ed47d6e..bb2ce0f7221 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -14,6 +14,7 @@ static struct nv40_state_entry *render_states[] = { &nv40_state_blend_colour, &nv40_state_zsa, &nv40_state_viewport, + &nv40_state_vbo, NULL }; @@ -31,7 +32,7 @@ nv40_state_validate(struct nv40_context *nv40) if (nv40->dirty & e->dirty.pipe) { if (e->validate(nv40)) - nv40->state.dirty |= (1 << e->dirty.hw); + nv40->state.dirty |= (1ULL << e->dirty.hw); } states++; @@ -64,20 +65,21 @@ nv40_state_emit(struct nv40_context *nv40) unsigned i, samplers; while (state->dirty) { - unsigned idx = ffs(state->dirty) - 1; + unsigned idx = ffsll(state->dirty) - 1; so_ref (state->hw[idx], &nv40->hw->state[idx]); so_emit(nv40->nvws, nv40->hw->state[idx]); - state->dirty &= ~(1 << idx); + state->dirty &= ~(1ULL << idx); } so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FRAGTEX0+i]); - samplers &= ~(1 << i); + samplers &= ~(1ULL << i); } so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FRAGPROG]); + so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_VTXBUF]); } void diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 753c2fe64b9..b5faf06291c 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -39,6 +39,43 @@ nv40_vbo_type(uint format) } } +static boolean +nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, + unsigned ib_size) +{ + unsigned type; + + if (!ib) { + nv40->pipe_state.idxbuf = NULL; + nv40->pipe_state.idxbuf_format = 0xdeadbeef; + return FALSE; + } + + /* No support for 8bit indices, no support at all on 0x4497 chips */ + if (nv40->hw->curie->grclass == NV44TCL || ib_size == 1) + return FALSE; + + switch (ib_size) { + case 2: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; + break; + case 4: + type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; + break; + default: + return FALSE; + } + + if (ib != nv40->pipe_state.idxbuf || + type != nv40->pipe_state.idxbuf_format) { + nv40->dirty |= NV40_NEW_ARRAYS; + nv40->pipe_state.idxbuf = ib; + nv40->pipe_state.idxbuf_format = type; + } + + return TRUE; +} + static boolean nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, struct pipe_vertex_element *ve, @@ -101,104 +138,15 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, return TRUE; } -static void -nv40_vbo_arrays_update(struct nv40_context *nv40, struct pipe_buffer *ib, - unsigned ib_format) -{ - struct nv40_vertex_program *vp = nv40->pipe_state.vertprog; - struct nouveau_stateobj *vtxbuf, *vtxfmt; - unsigned inputs, hw, num_hw; - unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; - - inputs = vp->ir; - for (hw = 0; hw < 16 && inputs; hw++) { - if (inputs & (1 << hw)) { - num_hw = hw; - inputs &= ~(1 << hw); - } - } - num_hw++; - - vtxbuf = so_new(20, 18); - so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); - vtxfmt = so_new(17, 0); - so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); - - inputs = vp->ir; - for (hw = 0; hw < num_hw; hw++) { - struct pipe_vertex_element *ve; - struct pipe_vertex_buffer *vb; - - if (!(inputs & (1 << hw))) { - so_data(vtxbuf, 0); - so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); - continue; - } - - ve = &nv40->vtxelt[hw]; - vb = &nv40->vtxbuf[ve->vertex_buffer_index]; - - if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { - so_data(vtxbuf, 0); - so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); - continue; - } - - so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, - vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, - 0, NV40TCL_VTXBUF_ADDRESS_DMA1); - so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_ncomp(ve->src_format) << - NV40TCL_VTXFMT_SIZE_SHIFT) | - nv40_vbo_type(ve->src_format))); - } - - if (ib) { - so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); - so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); - so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, - 0, NV40TCL_IDXBUF_FORMAT_DMA1); - } - - so_emit(nv40->nvws, vtxfmt); - so_emit(nv40->nvws, vtxbuf); - so_ref (vtxbuf, &nv40->so_vtxbuf); - so_ref (NULL, &vtxfmt); -} - -static boolean -nv40_vbo_validate_state(struct nv40_context *nv40, - struct pipe_buffer *ib, unsigned ib_format) -{ - unsigned vdn = nv40->dirty & NV40_NEW_ARRAYS; - - nv40_emit_hw_state(nv40); - if (vdn || ib) { - nv40_vbo_arrays_update(nv40, ib, ib_format); - nv40->dirty &= ~NV40_NEW_ARRAYS; - } - - so_emit_reloc_markers(nv40->nvws, nv40->so_vtxbuf); - - BEGIN_RING(curie, 0x1710, 1); - OUT_RING (0); /* vtx cache flush */ - - return TRUE; -} - boolean nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); unsigned nr; - boolean ret; - ret = nv40_vbo_validate_state(nv40, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } + nv40_vbo_set_idxbuf(nv40, NULL, 0); + nv40_emit_hw_state(nv40); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -305,14 +253,9 @@ nv40_draw_elements_inline(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); struct pipe_winsys *ws = pipe->winsys; - boolean ret; void *map; - ret = nv40_vbo_validate_state(nv40, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } + nv40_emit_hw_state(nv40); map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); if (!ib) { @@ -348,30 +291,12 @@ nv40_draw_elements_inline(struct pipe_context *pipe, static boolean nv40_draw_elements_vbo(struct pipe_context *pipe, - struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); - unsigned nr, type; - boolean ret; - - switch (ib_size) { - case 2: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; - break; - case 4: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; - break; - default: - NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); - return FALSE; - } + unsigned nr; - ret = nv40_vbo_validate_state(nv40, ib, type); - if (!ret) { - NOUVEAU_ERR("failed state validation\n"); - return FALSE; - } + nv40_emit_hw_state(nv40); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -409,19 +334,92 @@ nv40_draw_elements(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - /* 0x4497 doesn't support real index buffers, and there doesn't appear - * to be support on any chipset for 8-bit indices. - */ - if (nv40->hw->curie->grclass == NV44TCL || indexSize == 1) { + if (nv40_vbo_set_idxbuf(nv40, indexBuffer, indexSize)) { + nv40_draw_elements_vbo(pipe, mode, start, count); + } else { nv40_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); - } else { - nv40_draw_elements_vbo(pipe, indexBuffer, indexSize, - mode, start, count); } pipe->flush(pipe, 0); return TRUE; } +static boolean +nv40_vbo_validate(struct nv40_context *nv40) +{ + struct nv40_vertex_program *vp = nv40->pipe_state.vertprog; + struct nouveau_stateobj *vtxbuf, *vtxfmt; + struct pipe_buffer *ib = nv40->pipe_state.idxbuf; + unsigned ib_format = nv40->pipe_state.idxbuf_format; + unsigned inputs, hw, num_hw; + unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; + + inputs = vp->ir; + for (hw = 0; hw < 16 && inputs; hw++) { + if (inputs & (1 << hw)) { + num_hw = hw; + inputs &= ~(1 << hw); + } + } + num_hw++; + + vtxbuf = so_new(20, 18); + so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + vtxfmt = so_new(17, 0); + so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); + + inputs = vp->ir; + for (hw = 0; hw < num_hw; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + + if (!(inputs & (1 << hw))) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; + } + + ve = &nv40->vtxelt[hw]; + vb = &nv40->vtxbuf[ve->vertex_buffer_index]; + + if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; + } + + so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, + vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + 0, NV40TCL_VTXBUF_ADDRESS_DMA1); + so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | + (nv40_vbo_ncomp(ve->src_format) << + NV40TCL_VTXFMT_SIZE_SHIFT) | + nv40_vbo_type(ve->src_format))); + } + + if (ib) { + so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, + 0, NV40TCL_IDXBUF_FORMAT_DMA1); + } + + so_method(vtxbuf, nv40->hw->curie, 0x1710, 1); + so_data (vtxbuf, 0); + + so_ref(vtxbuf, &nv40->state.hw[NV40_STATE_VTXBUF]); + nv40->state.dirty |= (1ULL << NV40_STATE_VTXBUF); + so_ref(vtxfmt, &nv40->state.hw[NV40_STATE_VTXFMT]); + nv40->state.dirty |= (1ULL << NV40_STATE_VTXFMT); + return FALSE; +} + +struct nv40_state_entry nv40_state_vbo = { + .validate = nv40_vbo_validate, + .dirty = { + .pipe = NV40_NEW_ARRAYS, + .hw = 0, + } +}; -- cgit v1.2.3 From 026e2fd3c6eb87a010a9c90341e8a77b09376b5b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 25 Feb 2008 13:33:08 +1100 Subject: nv40: remove pipe_state struct now. --- src/gallium/drivers/nv40/nv40_context.h | 41 +++++++++++------------- src/gallium/drivers/nv40/nv40_fragprog.c | 4 +-- src/gallium/drivers/nv40/nv40_fragtex.c | 2 +- src/gallium/drivers/nv40/nv40_state.c | 26 +++++++-------- src/gallium/drivers/nv40/nv40_state_blend.c | 4 +-- src/gallium/drivers/nv40/nv40_state_clip.c | 2 +- src/gallium/drivers/nv40/nv40_state_fb.c | 2 +- src/gallium/drivers/nv40/nv40_state_rasterizer.c | 2 +- src/gallium/drivers/nv40/nv40_state_scissor.c | 4 +-- src/gallium/drivers/nv40/nv40_state_stipple.c | 4 +-- src/gallium/drivers/nv40/nv40_state_viewport.c | 2 +- src/gallium/drivers/nv40/nv40_state_zsa.c | 2 +- src/gallium/drivers/nv40/nv40_vbo.c | 18 +++++------ src/gallium/drivers/nv40/nv40_vertprog.c | 4 +-- 14 files changed, 57 insertions(+), 60 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 110d9d7ab70..16cc053ad9f 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -132,32 +132,29 @@ struct nv40_context { int chipset; - unsigned dirty; + /* HW state derived from pipe states */ + struct nv40_state state; + unsigned fallback; + /* Context state */ + unsigned dirty; + struct pipe_scissor_state scissor; + unsigned stipple[32]; + struct pipe_clip_state clip; + struct nv40_vertex_program *vertprog; + struct nv40_fragment_program *fragprog; + struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + struct nv40_rasterizer_state *rasterizer; + struct nv40_zsa_state *zsa; + struct nv40_blend_state *blend; + struct pipe_blend_color blend_colour; + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state framebuffer; + struct pipe_buffer *idxbuf; + unsigned idxbuf_format; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; unsigned dirty_samplers; - - struct { - struct pipe_scissor_state scissor; - unsigned stipple[32]; - struct pipe_clip_state clip; - struct nv40_vertex_program *vertprog; - struct nv40_fragment_program *fragprog; - struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; - struct nv40_rasterizer_state *rasterizer; - struct nv40_zsa_state *zsa; - struct nv40_blend_state *blend; - struct pipe_blend_color blend_colour; - struct pipe_viewport_state viewport; - struct pipe_framebuffer_state framebuffer; - struct pipe_buffer *idxbuf; - unsigned idxbuf_format; - } pipe_state; - - struct nv40_state state; - unsigned fallback; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; }; diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index a4a1ea01e03..2a8abb32a30 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -790,9 +790,9 @@ nv40_fragprog_upload(struct nv40_context *nv40, static boolean nv40_fragprog_validate(struct nv40_context *nv40) { - struct nv40_fragment_program *fp = nv40->pipe_state.fragprog; + struct nv40_fragment_program *fp = nv40->fragprog; struct pipe_buffer *constbuf = - nv40->pipe_state.constbuf[PIPE_SHADER_FRAGMENT]; + nv40->constbuf[PIPE_SHADER_FRAGMENT]; struct pipe_winsys *ws = nv40->pipe.winsys; struct nouveau_stateobj *so; int i; diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index ed47d707b2f..6be8378c087 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -126,7 +126,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) static boolean nv40_fragtex_validate(struct nv40_context *nv40) { - struct nv40_fragment_program *fp = nv40->pipe_state.fragprog; + struct nv40_fragment_program *fp = nv40->fragprog; struct nv40_state *state = &nv40->state; struct nouveau_stateobj *so; unsigned samplers, unit; diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index e6f2754dc51..24335fbc44d 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -57,7 +57,7 @@ nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.blend = hwcso; + nv40->blend = hwcso; nv40->dirty |= NV40_NEW_BLEND; } @@ -399,7 +399,7 @@ nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.rasterizer = hwcso; + nv40->rasterizer = hwcso; nv40->dirty |= NV40_NEW_RAST; } @@ -470,7 +470,7 @@ nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.zsa = hwcso; + nv40->zsa = hwcso; nv40->dirty |= NV40_NEW_ZSA; } @@ -500,7 +500,7 @@ nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.vertprog = hwcso; + nv40->vertprog = hwcso; nv40->dirty |= NV40_NEW_VERTPROG; } @@ -531,7 +531,7 @@ nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.fragprog = hwcso; + nv40->fragprog = hwcso; nv40->dirty |= NV40_NEW_FRAGPROG; } @@ -551,7 +551,7 @@ nv40_set_blend_color(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.blend_colour = *bcol; + nv40->blend_colour = *bcol; nv40->dirty |= NV40_NEW_BCOL; } @@ -561,7 +561,7 @@ nv40_set_clip_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.clip = *clip; + nv40->clip = *clip; nv40->dirty |= NV40_NEW_UCP; } @@ -572,11 +572,11 @@ nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, struct nv40_context *nv40 = nv40_context(pipe); if (shader == PIPE_SHADER_VERTEX) { - nv40->pipe_state.constbuf[PIPE_SHADER_VERTEX] = buf->buffer; + nv40->constbuf[PIPE_SHADER_VERTEX] = buf->buffer; nv40->dirty |= NV40_NEW_VERTPROG; } else if (shader == PIPE_SHADER_FRAGMENT) { - nv40->pipe_state.constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; + nv40->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; nv40->dirty |= NV40_NEW_FRAGPROG; } } @@ -587,7 +587,7 @@ nv40_set_framebuffer_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.framebuffer = *fb; + nv40->framebuffer = *fb; nv40->dirty |= NV40_NEW_FB; } @@ -597,7 +597,7 @@ nv40_set_polygon_stipple(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - memcpy(nv40->pipe_state.stipple, stipple->stipple, 4 * 32); + memcpy(nv40->stipple, stipple->stipple, 4 * 32); nv40->dirty |= NV40_NEW_STIPPLE; } @@ -607,7 +607,7 @@ nv40_set_scissor_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.scissor = *s; + nv40->scissor = *s; nv40->dirty |= NV40_NEW_SCISSOR; } @@ -617,7 +617,7 @@ nv40_set_viewport_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - nv40->pipe_state.viewport = *vpt; + nv40->viewport = *vpt; nv40->dirty |= NV40_NEW_VIEWPORT; } diff --git a/src/gallium/drivers/nv40/nv40_state_blend.c b/src/gallium/drivers/nv40/nv40_state_blend.c index 81b927a67a9..dd09830aa3d 100644 --- a/src/gallium/drivers/nv40/nv40_state_blend.c +++ b/src/gallium/drivers/nv40/nv40_state_blend.c @@ -3,7 +3,7 @@ static boolean nv40_state_blend_validate(struct nv40_context *nv40) { - so_ref(nv40->pipe_state.blend->so, &nv40->state.hw[NV40_STATE_BLEND]); + so_ref(nv40->blend->so, &nv40->state.hw[NV40_STATE_BLEND]); return TRUE; } @@ -19,7 +19,7 @@ static boolean nv40_state_blend_colour_validate(struct nv40_context *nv40) { struct nouveau_stateobj *so = so_new(2, 0); - struct pipe_blend_color *bcol = &nv40->pipe_state.blend_colour; + struct pipe_blend_color *bcol = &nv40->blend_colour; so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | diff --git a/src/gallium/drivers/nv40/nv40_state_clip.c b/src/gallium/drivers/nv40/nv40_state_clip.c index 19f1c3b36de..93e690161f3 100644 --- a/src/gallium/drivers/nv40/nv40_state_clip.c +++ b/src/gallium/drivers/nv40/nv40_state_clip.c @@ -3,7 +3,7 @@ static boolean nv40_state_clip_validate(struct nv40_context *nv40) { - if (nv40->pipe_state.clip.nr) + if (nv40->clip.nr) nv40->fallback |= NV40_FALLBACK_TNL; return FALSE; diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index c3bf4d43a33..3d0ab920030 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -3,7 +3,7 @@ static boolean nv40_state_framebuffer_validate(struct nv40_context *nv40) { - struct pipe_framebuffer_state *fb = &nv40->pipe_state.framebuffer; + struct pipe_framebuffer_state *fb = &nv40->framebuffer; struct pipe_surface *rt[4], *zeta; uint32_t rt_enable, rt_format, w, h; int i, colour_format = 0, zeta_format = 0; diff --git a/src/gallium/drivers/nv40/nv40_state_rasterizer.c b/src/gallium/drivers/nv40/nv40_state_rasterizer.c index 59b35d1d50a..9ecda5990f0 100644 --- a/src/gallium/drivers/nv40/nv40_state_rasterizer.c +++ b/src/gallium/drivers/nv40/nv40_state_rasterizer.c @@ -3,7 +3,7 @@ static boolean nv40_state_rasterizer_validate(struct nv40_context *nv40) { - so_ref(nv40->pipe_state.rasterizer->so, + so_ref(nv40->rasterizer->so, &nv40->state.hw[NV40_STATE_RAST]); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c index ee797094d39..09ffc49f965 100644 --- a/src/gallium/drivers/nv40/nv40_state_scissor.c +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -3,8 +3,8 @@ static boolean nv40_state_scissor_validate(struct nv40_context *nv40) { - struct pipe_rasterizer_state *rast = &nv40->pipe_state.rasterizer->pipe; - struct pipe_scissor_state *s = &nv40->pipe_state.scissor; + struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; + struct pipe_scissor_state *s = &nv40->scissor; struct nouveau_stateobj *so; if (nv40->state.hw[NV40_STATE_SCISSOR] && diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c index aad4d179ac4..001c396d747 100644 --- a/src/gallium/drivers/nv40/nv40_state_stipple.c +++ b/src/gallium/drivers/nv40/nv40_state_stipple.c @@ -3,7 +3,7 @@ static boolean nv40_state_stipple_validate(struct nv40_context *nv40) { - struct pipe_rasterizer_state *rast = &nv40->pipe_state.rasterizer->pipe; + struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; struct nouveau_grobj *curie = nv40->hw->curie; struct nouveau_stateobj *so; @@ -19,7 +19,7 @@ nv40_state_stipple_validate(struct nv40_context *nv40) so_data (so, 1); so_method(so, curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32); for (i = 0; i < 32; i++) - so_data(so, nv40->pipe_state.stipple[i]); + so_data(so, nv40->stipple[i]); } else { so = so_new(2, 0); so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c index 68820d31339..9616be5052a 100644 --- a/src/gallium/drivers/nv40/nv40_state_viewport.c +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -4,7 +4,7 @@ static boolean nv40_state_viewport_validate(struct nv40_context *nv40) { struct nouveau_stateobj *so = so_new(9, 0); - struct pipe_viewport_state *vpt = &nv40->pipe_state.viewport; + struct pipe_viewport_state *vpt = &nv40->viewport; so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); so_data (so, fui(vpt->translate[0])); diff --git a/src/gallium/drivers/nv40/nv40_state_zsa.c b/src/gallium/drivers/nv40/nv40_state_zsa.c index 061a3555cbe..fb760677c88 100644 --- a/src/gallium/drivers/nv40/nv40_state_zsa.c +++ b/src/gallium/drivers/nv40/nv40_state_zsa.c @@ -3,7 +3,7 @@ static boolean nv40_state_zsa_validate(struct nv40_context *nv40) { - so_ref(nv40->pipe_state.zsa->so, + so_ref(nv40->zsa->so, &nv40->state.hw[NV40_STATE_ZSA]); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index b5faf06291c..1653ebf2a7e 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -46,8 +46,8 @@ nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned type; if (!ib) { - nv40->pipe_state.idxbuf = NULL; - nv40->pipe_state.idxbuf_format = 0xdeadbeef; + nv40->idxbuf = NULL; + nv40->idxbuf_format = 0xdeadbeef; return FALSE; } @@ -66,11 +66,11 @@ nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, return FALSE; } - if (ib != nv40->pipe_state.idxbuf || - type != nv40->pipe_state.idxbuf_format) { + if (ib != nv40->idxbuf || + type != nv40->idxbuf_format) { nv40->dirty |= NV40_NEW_ARRAYS; - nv40->pipe_state.idxbuf = ib; - nv40->pipe_state.idxbuf_format = type; + nv40->idxbuf = ib; + nv40->idxbuf_format = type; } return TRUE; @@ -348,10 +348,10 @@ nv40_draw_elements(struct pipe_context *pipe, static boolean nv40_vbo_validate(struct nv40_context *nv40) { - struct nv40_vertex_program *vp = nv40->pipe_state.vertprog; + struct nv40_vertex_program *vp = nv40->vertprog; struct nouveau_stateobj *vtxbuf, *vtxfmt; - struct pipe_buffer *ib = nv40->pipe_state.idxbuf; - unsigned ib_format = nv40->pipe_state.idxbuf_format; + struct pipe_buffer *ib = nv40->idxbuf; + unsigned ib_format = nv40->idxbuf_format; unsigned inputs, hw, num_hw; unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index c482964adc0..d3ed57b1998 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -634,9 +634,9 @@ out_err: static boolean nv40_vertprog_validate(struct nv40_context *nv40) { - struct nv40_vertex_program *vp = nv40->pipe_state.vertprog; + struct nv40_vertex_program *vp = nv40->vertprog; struct pipe_buffer *constbuf = - nv40->pipe_state.constbuf[PIPE_SHADER_VERTEX]; + nv40->constbuf[PIPE_SHADER_VERTEX]; struct nouveau_winsys *nvws = nv40->nvws; struct pipe_winsys *ws = nv40->pipe.winsys; boolean upload_code = FALSE, upload_data = FALSE; -- cgit v1.2.3 From 84cc07dc89c0ebce4ad55b4b3684d4420a202683 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 29 Feb 2008 15:03:57 +1100 Subject: nouveau: implement pipe_screen Untested on NV3x/NV5x. Quite possibly broken. --- src/gallium/drivers/nouveau/nouveau_winsys.h | 15 +- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.c | 86 +----------- src/gallium/drivers/nv30/nv30_context.h | 2 +- src/gallium/drivers/nv30/nv30_miptree.c | 47 ++++++- src/gallium/drivers/nv30/nv30_screen.c | 151 +++++++++++++++++++++ src/gallium/drivers/nv30/nv30_screen.h | 20 +++ src/gallium/drivers/nv30/nv30_surface.c | 72 ---------- src/gallium/drivers/nv40/Makefile | 1 + src/gallium/drivers/nv40/nv40_context.c | 85 +----------- src/gallium/drivers/nv40/nv40_context.h | 2 + src/gallium/drivers/nv40/nv40_miptree.c | 25 ++-- src/gallium/drivers/nv40/nv40_screen.c | 151 +++++++++++++++++++++ src/gallium/drivers/nv40/nv40_screen.h | 20 +++ src/gallium/drivers/nv40/nv40_surface.c | 41 ------ src/gallium/drivers/nv50/Makefile | 1 + src/gallium/drivers/nv50/nv50_context.c | 91 +------------ src/gallium/drivers/nv50/nv50_context.h | 2 + src/gallium/drivers/nv50/nv50_miptree.c | 29 +++- src/gallium/drivers/nv50/nv50_screen.c | 117 ++++++++++++++++ src/gallium/drivers/nv50/nv50_screen.h | 20 +++ src/gallium/drivers/nv50/nv50_surface.c | 10 -- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 16 ++- .../winsys/dri/nouveau/nouveau_winsys_softpipe.c | 30 ++-- 24 files changed, 623 insertions(+), 412 deletions(-) create mode 100644 src/gallium/drivers/nv30/nv30_screen.c create mode 100644 src/gallium/drivers/nv30/nv30_screen.h create mode 100644 src/gallium/drivers/nv40/nv40_screen.c create mode 100644 src/gallium/drivers/nv40/nv40_screen.h create mode 100644 src/gallium/drivers/nv50/nv50_screen.c create mode 100644 src/gallium/drivers/nv50/nv50_screen.h diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index b5e470cfaa4..98d95e94a58 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -49,13 +49,22 @@ struct nouveau_winsys { unsigned, unsigned, unsigned, unsigned, unsigned); }; +extern struct pipe_screen * +nv30_screen_create(struct pipe_winsys *ws, unsigned chipset); + extern struct pipe_context * -nv30_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); +nv30_create(struct pipe_screen *, struct nouveau_winsys *); + +extern struct pipe_screen * +nv40_screen_create(struct pipe_winsys *ws, unsigned chipset); extern struct pipe_context * -nv40_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); +nv40_create(struct pipe_screen *, struct nouveau_winsys *); + +extern struct pipe_screen * +nv50_screen_create(struct pipe_winsys *ws, unsigned chipset); extern struct pipe_context * -nv50_create(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); +nv50_create(struct pipe_screen *, struct nouveau_winsys *); #endif diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index b7c252fc986..3f80fb87c9b 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -11,6 +11,7 @@ DRIVER_SOURCES = \ nv30_fragtex.c \ nv30_miptree.c \ nv30_query.c \ + nv30_screen.c \ nv30_state.c \ nv30_state_emit.c \ nv30_surface.c \ diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index e9afeb80176..b8452e23b16 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -4,80 +4,7 @@ #include "pipe/p_util.h" #include "nv30_context.h" - -static const char * -nv30_get_name(struct pipe_context *pipe) -{ - struct nv30_context *nv30 = nv30_context(pipe); - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", nv30->chipset); - return buffer; -} - -static const char * -nv30_get_vendor(struct pipe_context *pipe) -{ - return "nouveau"; -} - -static int -nv30_get_param(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 16; - case PIPE_CAP_NPOT_TEXTURES: - return 0; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 1; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 1; - case PIPE_CAP_POINT_SPRITE: - return 1; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 2; - case PIPE_CAP_OCCLUSION_QUERY: - return 1; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 1; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 13; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 10; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 13; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv30_get_paramf(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} +#include "nv30_screen.h" static void nv30_flush(struct pipe_context *pipe, unsigned flags) @@ -338,9 +265,10 @@ nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) #define NV35TCL_CHIPSET_3X_MASK 0x000001e0 struct pipe_context * -nv30_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, - unsigned chipset) +nv30_create(struct pipe_screen *screen, struct nouveau_winsys *nvws) { + struct pipe_winsys *pipe_winsys = screen->winsys; + unsigned chipset = nv30_screen(screen)->chipset; struct nv30_context *nv30; int rankine_class = 0, ret; @@ -404,12 +332,9 @@ nv30_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, /* Pipe context setup */ nv30->pipe.winsys = pipe_winsys; + nv30->pipe.screen = screen; nv30->pipe.destroy = nv30_destroy; - nv30->pipe.get_name = nv30_get_name; - nv30->pipe.get_vendor = nv30_get_vendor; - nv30->pipe.get_param = nv30_get_param; - nv30->pipe.get_paramf = nv30_get_paramf; nv30->pipe.draw_arrays = nv30_draw_arrays; nv30->pipe.draw_elements = nv30_draw_elements; @@ -420,7 +345,6 @@ nv30_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, nv30_init_query_functions(nv30); nv30_init_surface_functions(nv30); nv30_init_state_functions(nv30); - nv30_init_miptree_functions(nv30); nv30->draw = draw_create(); assert(nv30->draw); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index d6d16ee8686..c63847a087c 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -91,7 +91,7 @@ nv30_context(struct pipe_context *pipe) extern void nv30_init_state_functions(struct nv30_context *nv30); extern void nv30_init_surface_functions(struct nv30_context *nv30); -extern void nv30_init_miptree_functions(struct nv30_context *nv30); +extern void nv30_init_miptree_functions(struct pipe_screen *screen); extern void nv30_init_query_functions(struct nv30_context *nv30); /* nv30_draw.c */ diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 5fb89f4cfdc..23bcef08ebc 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -4,6 +4,7 @@ #include "pipe/p_inlines.h" #include "nv30_context.h" +#include "nv30_screen.h" static void nv30_miptree_layout(struct nv30_miptree *nv30mt) @@ -54,9 +55,9 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) } static void -nv30_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) +nv30_miptree_create(struct pipe_screen *screen, struct pipe_texture **pt) { - struct pipe_winsys *ws = pipe->winsys; + struct pipe_winsys *ws = screen->winsys; struct nv30_miptree *nv30mt; nv30mt = realloc(*pt, sizeof(struct nv30_miptree)); @@ -77,9 +78,9 @@ nv30_miptree_create(struct pipe_context *pipe, struct pipe_texture **pt) } static void -nv30_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +nv30_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) { - struct pipe_winsys *ws = pipe->winsys; + struct pipe_winsys *ws = screen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -96,10 +97,42 @@ nv30_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) } } +static struct pipe_surface * +nv30_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv30mt->level[level].pitch / ps->cpp; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv30mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv30mt->level[level].image_offset[zslice]; + } else { + ps->offset = nv30mt->level[level].image_offset[0]; + } + + return ps; +} void -nv30_init_miptree_functions(struct nv30_context *nv30) +nv30_init_miptree_functions(struct pipe_screen *screen) { - nv30->pipe.texture_create = nv30_miptree_create; - nv30->pipe.texture_release = nv30_miptree_release; + struct nv30_screen *nv30screen = nv30_screen(screen); + + nv30screen->screen.texture_create = nv30_miptree_create; + nv30screen->screen.texture_release = nv30_miptree_release; + nv30screen->screen.get_tex_surface = nv30_miptree_surface_get; } diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c new file mode 100644 index 00000000000..6d64025528f --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -0,0 +1,151 @@ +#include "pipe/p_screen.h" +#include "pipe/p_util.h" + +#include "nv30_context.h" +#include "nv30_screen.h" + +static const char * +nv30_screen_get_name(struct pipe_screen *screen) +{ + struct nv30_screen *nv30screen = nv30_screen(screen); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv30screen->chipset); + return buffer; +} + +static const char * +nv30_screen_get_vendor(struct pipe_screen *screen) +{ + return "nouveau"; +} + +static int +nv30_screen_get_param(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 16; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 1; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 2; + case PIPE_CAP_OCCLUSION_QUERY: + return 1; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 1; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv30_screen_get_paramf(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static boolean +nv30_screen_is_format_supported(struct pipe_screen *screen, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static void +nv30_screen_destroy(struct pipe_screen *screen) +{ + FREE(screen); +} + +struct pipe_screen * +nv30_screen_create(struct pipe_winsys *winsys, unsigned chipset) +{ + struct nv30_screen *nv30screen = CALLOC_STRUCT(nv30_screen); + + if (!nv30screen) + return NULL; + + nv30screen->chipset = chipset; + + nv30screen->screen.winsys = winsys; + + nv30screen->screen.destroy = nv30_screen_destroy; + + nv30screen->screen.get_name = nv30_screen_get_name; + nv30screen->screen.get_vendor = nv30_screen_get_vendor; + nv30screen->screen.get_param = nv30_screen_get_param; + nv30screen->screen.get_paramf = nv30_screen_get_paramf; + nv30screen->screen.is_format_supported = + nv30_screen_is_format_supported; + + nv30_init_miptree_functions(&nv30screen->screen); + return &nv30screen->screen; +} + diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h new file mode 100644 index 00000000000..e55242fbf7f --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_screen.h @@ -0,0 +1,20 @@ +#ifndef __NV30_SCREEN_H__ +#define __NV30_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv30_screen { + struct pipe_screen screen; + unsigned chipset; +}; + +static INLINE struct nv30_screen * +nv30_screen(struct pipe_screen *screen) +{ + return (struct nv30_screen *)screen; +} + +extern struct pipe_screen * +nv30_screen_create(struct pipe_winsys *winsys, unsigned chipset); + +#endif diff --git a/src/gallium/drivers/nv30/nv30_surface.c b/src/gallium/drivers/nv30/nv30_surface.c index 974965679f5..b20a3dd4c15 100644 --- a/src/gallium/drivers/nv30/nv30_surface.c +++ b/src/gallium/drivers/nv30/nv30_surface.c @@ -33,76 +33,6 @@ #include "pipe/p_inlines.h" #include "util/p_tile.h" -static boolean -nv30_surface_format_supported(struct pipe_context *pipe, - enum pipe_format format, uint type) -{ - switch (type) { - case PIPE_SURFACE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return TRUE; - default: - break; - } - break; - case PIPE_TEXTURE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A1R5G5B5_UNORM: - case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: - case PIPE_FORMAT_Z16_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - } - break; - default: - assert(0); - }; - - return FALSE; -} - -static struct pipe_surface * -nv30_get_tex_surface(struct pipe_context *pipe, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) -{ - struct pipe_winsys *ws = pipe->winsys; - struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; - struct pipe_surface *ps; - - ps = ws->surface_alloc(ws); - if (!ps) - return NULL; - pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); - ps->format = pt->format; - ps->cpp = pt->cpp; - ps->width = pt->width[level]; - ps->height = pt->height[level]; - ps->pitch = nv30mt->level[level].pitch / ps->cpp; - - if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv30mt->level[level].image_offset[face]; - } else - if (pt->target == PIPE_TEXTURE_3D) { - ps->offset = nv30mt->level[level].image_offset[zslice]; - } else { - ps->offset = nv30mt->level[level].image_offset[0]; - } - - return ps; -} - static void nv30_surface_copy(struct pipe_context *pipe, unsigned do_flip, struct pipe_surface *dest, unsigned destx, unsigned desty, @@ -130,8 +60,6 @@ nv30_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, void nv30_init_surface_functions(struct nv30_context *nv30) { - nv30->pipe.is_format_supported = nv30_surface_format_supported; - nv30->pipe.get_tex_surface = nv30_get_tex_surface; nv30->pipe.surface_copy = nv30_surface_copy; nv30->pipe.surface_fill = nv30_surface_fill; } diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile index fd002b54e78..3369a21574f 100644 --- a/src/gallium/drivers/nv40/Makefile +++ b/src/gallium/drivers/nv40/Makefile @@ -11,6 +11,7 @@ DRIVER_SOURCES = \ nv40_fragtex.c \ nv40_miptree.c \ nv40_query.c \ + nv40_screen.c \ nv40_state.c \ nv40_state_blend.c \ nv40_state_clip.c \ diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 8b5cc693de0..a7f64c6e9e5 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -4,85 +4,12 @@ #include "pipe/p_util.h" #include "nv40_context.h" +#include "nv40_screen.h" #define NV4X_GRCLASS4097_CHIPSETS 0x00000baf #define NV4X_GRCLASS4497_CHIPSETS 0x00005450 #define NV6X_GRCLASS4497_CHIPSETS 0x00000088 -static const char * -nv40_get_name(struct pipe_context *pipe) -{ - struct nv40_context *nv40 = nv40_context(pipe); - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", nv40->chipset); - return buffer; -} - -static const char * -nv40_get_vendor(struct pipe_context *pipe) -{ - return "nouveau"; -} - -static int -nv40_get_param(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 16; - case PIPE_CAP_NPOT_TEXTURES: - return 1; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 1; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 1; - case PIPE_CAP_POINT_SPRITE: - return 1; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 4; - case PIPE_CAP_OCCLUSION_QUERY: - return 1; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 1; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 13; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 10; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 13; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv40_get_paramf(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 16.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} - static void nv40_flush(struct pipe_context *pipe, unsigned flags) { @@ -269,10 +196,11 @@ nv40_destroy(struct pipe_context *pipe) } struct pipe_context * -nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) +nv40_create(struct pipe_screen *pscreen, struct nouveau_winsys *nvws) { + struct pipe_winsys *ws = pscreen->winsys; struct nv40_context *nv40; + unsigned chipset = nv40_screen(pscreen)->chipset; nv40 = CALLOC(1, sizeof(struct nv40_context)); if (!nv40) @@ -288,11 +216,8 @@ nv40_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, nv40->nvws = nvws; nv40->pipe.winsys = ws; + nv40->pipe.screen = pscreen; nv40->pipe.destroy = nv40_destroy; - nv40->pipe.get_name = nv40_get_name; - nv40->pipe.get_vendor = nv40_get_vendor; - nv40->pipe.get_param = nv40_get_param; - nv40->pipe.get_paramf = nv40_get_paramf; nv40->pipe.draw_arrays = nv40_draw_arrays; nv40->pipe.draw_elements = nv40_draw_elements; nv40->pipe.clear = nv40_clear; diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 16cc053ad9f..3ddfbd43f60 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -178,6 +178,8 @@ extern void nv40_init_surface_functions(struct nv40_context *nv40); extern void nv40_init_miptree_functions(struct nv40_context *nv40); extern void nv40_init_query_functions(struct nv40_context *nv40); +extern void nv40_screen_init_miptree_functions(struct pipe_screen *pscreen); + /* nv40_draw.c */ extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 5e1c7ade31b..94ba05b7104 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -54,9 +54,9 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) } static struct pipe_texture * -nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) +nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { - struct pipe_winsys *ws = pipe->winsys; + struct pipe_winsys *ws = pscreen->winsys; struct nv40_miptree *mt; mt = MALLOC(sizeof(struct nv40_miptree)); @@ -64,6 +64,8 @@ nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) return NULL; mt->base = *pt; mt->base.refcount = 1; + mt->base.screen = pscreen; + nv40_miptree_layout(mt); mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, @@ -77,9 +79,9 @@ nv40_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) } static void -nv40_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) { - struct pipe_winsys *ws = pipe->winsys; + struct pipe_winsys *ws = pscreen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -102,10 +104,10 @@ nv40_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) } static struct pipe_surface * -nv40_miptree_surface(struct pipe_context *pipe, struct pipe_texture *pt, +nv40_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { - struct pipe_winsys *ws = pipe->winsys; + struct pipe_winsys *ws = pscreen->winsys; struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; struct pipe_surface *ps; @@ -134,9 +136,14 @@ nv40_miptree_surface(struct pipe_context *pipe, struct pipe_texture *pt, void nv40_init_miptree_functions(struct nv40_context *nv40) { - nv40->pipe.texture_create = nv40_miptree_create; - nv40->pipe.texture_release = nv40_miptree_release; nv40->pipe.texture_update = nv40_miptree_update; - nv40->pipe.get_tex_surface = nv40_miptree_surface; +} + +void +nv40_screen_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv40_miptree_create; + pscreen->texture_release = nv40_miptree_release; + pscreen->get_tex_surface = nv40_miptree_surface; } diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c new file mode 100644 index 00000000000..1941598c641 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -0,0 +1,151 @@ +#include "pipe/p_screen.h" +#include "pipe/p_util.h" + +#include "nv40_context.h" +#include "nv40_screen.h" + +static const char * +nv40_screen_get_name(struct pipe_screen *pscreen) +{ + struct nv40_screen *screen = nv40_screen(pscreen); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", screen->chipset); + return buffer; +} + +static const char * +nv40_screen_get_vendor(struct pipe_screen *pscreen) +{ + return "nouveau"; +} + +static int +nv40_screen_get_param(struct pipe_screen *pscreen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 16; + case PIPE_CAP_NPOT_TEXTURES: + return 1; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 1; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 4; + case PIPE_CAP_OCCLUSION_QUERY: + return 1; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 1; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv40_screen_get_paramf(struct pipe_screen *pscreen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 16.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static boolean +nv40_screen_surface_format_supported(struct pipe_screen *pscreen, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_Z16_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static void +nv40_screen_destroy(struct pipe_screen *pscreen) +{ + FREE(pscreen); +} + +struct pipe_screen * +nv40_screen_create(struct pipe_winsys *ws, unsigned chipset) +{ + struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen); + + if (!screen) + return NULL; + + screen->chipset = chipset; + + screen->pipe.winsys = ws; + screen->pipe.destroy = nv40_screen_destroy; + + screen->pipe.get_name = nv40_screen_get_name; + screen->pipe.get_vendor = nv40_screen_get_vendor; + screen->pipe.get_param = nv40_screen_get_param; + screen->pipe.get_paramf = nv40_screen_get_paramf; + + screen->pipe.is_format_supported = nv40_screen_surface_format_supported; + + nv40_screen_init_miptree_functions(&screen->pipe); + + return &screen->pipe; +} + diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h new file mode 100644 index 00000000000..b30a6c5ad57 --- /dev/null +++ b/src/gallium/drivers/nv40/nv40_screen.h @@ -0,0 +1,20 @@ +#ifndef __NV40_SCREEN_H__ +#define __NV40_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv40_screen { + struct pipe_screen pipe; + unsigned chipset; +}; + +static INLINE struct nv40_screen * +nv40_screen(struct pipe_screen *screen) +{ + return (struct nv40_screen *)screen; +} + +extern struct pipe_screen * +nv40_screen_create(struct pipe_winsys *winsys, unsigned chipset); + +#endif diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c index df5d7abdbfc..e8a60116964 100644 --- a/src/gallium/drivers/nv40/nv40_surface.c +++ b/src/gallium/drivers/nv40/nv40_surface.c @@ -33,46 +33,6 @@ #include "pipe/p_inlines.h" #include "util/p_tile.h" -static boolean -nv40_surface_format_supported(struct pipe_context *pipe, - enum pipe_format format, uint type) -{ - switch (type) { - case PIPE_SURFACE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return TRUE; - default: - break; - } - break; - case PIPE_TEXTURE: - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A1R5G5B5_UNORM: - case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: - case PIPE_FORMAT_Z16_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - } - break; - default: - assert(0); - }; - - return FALSE; -} - static void nv40_surface_copy(struct pipe_context *pipe, unsigned do_flip, struct pipe_surface *dest, unsigned destx, unsigned desty, @@ -100,7 +60,6 @@ nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, void nv40_init_surface_functions(struct nv40_context *nv40) { - nv40->pipe.is_format_supported = nv40_surface_format_supported; nv40->pipe.surface_copy = nv40_surface_copy; nv40->pipe.surface_fill = nv40_surface_fill; } diff --git a/src/gallium/drivers/nv50/Makefile b/src/gallium/drivers/nv50/Makefile index 68eb49ff2a3..1c0b82887a1 100644 --- a/src/gallium/drivers/nv50/Makefile +++ b/src/gallium/drivers/nv50/Makefile @@ -9,6 +9,7 @@ DRIVER_SOURCES = \ nv50_draw.c \ nv50_miptree.c \ nv50_query.c \ + nv50_screen.c \ nv50_state.c \ nv50_surface.c \ nv50_vbo.c diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 3c5a54bfd38..98022809a6c 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -4,85 +4,7 @@ #include "pipe/p_util.h" #include "nv50_context.h" - -static boolean -nv50_is_format_supported(struct pipe_context *pipe, enum pipe_format format, - uint type) -{ - return FALSE; -} - -static const char * -nv50_get_name(struct pipe_context *pipe) -{ - struct nv50_context *nv50 = (struct nv50_context *)pipe; - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", nv50->chipset); - return buffer; -} - -static const char * -nv50_get_vendor(struct pipe_context *pipe) -{ - return "nouveau"; -} - -static int -nv50_get_param(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 32; - case PIPE_CAP_NPOT_TEXTURES: - return 0; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 1; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 0; - case PIPE_CAP_POINT_SPRITE: - return 0; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 8; - case PIPE_CAP_OCCLUSION_QUERY: - return 0; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 0; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 13; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 10; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 13; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv50_get_paramf(struct pipe_context *pipe, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} +#include "nv50_screen.h" static void nv50_flush(struct pipe_context *pipe, unsigned flags) @@ -134,9 +56,10 @@ nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) #define GRCLASS5097_CHIPSETS 0x00000000 #define GRCLASS8297_CHIPSETS 0x00000010 struct pipe_context * -nv50_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, - unsigned chipset) +nv50_create(struct pipe_screen *pscreen, struct nouveau_winsys *nvws) { + struct pipe_winsys *pipe_winsys = pscreen->winsys; + unsigned chipset = nv50_screen(pscreen)->chipset; struct nv50_context *nv50; int tesla_class, ret; @@ -173,13 +96,9 @@ nv50_create(struct pipe_winsys *pipe_winsys, struct nouveau_winsys *nvws, } nv50->pipe.winsys = pipe_winsys; + nv50->pipe.screen = pscreen; nv50->pipe.destroy = nv50_destroy; - nv50->pipe.is_format_supported = nv50_is_format_supported; - nv50->pipe.get_name = nv50_get_name; - nv50->pipe.get_vendor = nv50_get_vendor; - nv50->pipe.get_param = nv50_get_param; - nv50->pipe.get_paramf = nv50_get_paramf; nv50->pipe.draw_arrays = nv50_draw_arrays; nv50->pipe.draw_elements = nv50_draw_elements; diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index b99254f6191..a529bf3c3e3 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -38,6 +38,8 @@ extern void nv50_init_surface_functions(struct nv50_context *nv50); extern void nv50_init_state_functions(struct nv50_context *nv50); extern void nv50_init_query_functions(struct nv50_context *nv50); +extern void nv50_screen_init_miptree_functions(struct pipe_screen *pscreen); + /* nv50_draw.c */ extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50); diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 0c034ed4387..720d33fda91 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -1,25 +1,46 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" +#include "pipe/p_screen.h" #include "nv50_context.h" static struct pipe_texture * -nv50_miptree_create(struct pipe_context *pipe, const struct pipe_texture *pt) +nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { NOUVEAU_ERR("unimplemented\n"); return NULL; } static void -nv50_miptree_release(struct pipe_context *pipe, struct pipe_texture **pt) +nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) { NOUVEAU_ERR("unimplemented\n"); } +static struct pipe_surface * +nv50_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + NOUVEAU_ERR("unimplemented\n"); + return NULL; +} + +void +nv50_screen_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv50_miptree_create; + pscreen->texture_release = nv50_miptree_release; + pscreen->get_tex_surface = nv50_miptree_surface; +} + +static void +nv50_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) +{ +} + void nv50_init_miptree_functions(struct nv50_context *nv50) { - nv50->pipe.texture_create = nv50_miptree_create; - nv50->pipe.texture_release = nv50_miptree_release; + nv50->pipe.texture_update = nv50_miptree_update; } diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c new file mode 100644 index 00000000000..8bf82eb0bc0 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -0,0 +1,117 @@ +#include "pipe/p_screen.h" +#include "pipe/p_util.h" + +#include "nv50_context.h" +#include "nv50_screen.h" + +static boolean +nv50_screen_is_format_supported(struct pipe_screen *pscreen, + enum pipe_format format, uint type) +{ + return FALSE; +} + +static const char * +nv50_screen_get_name(struct pipe_screen *pscreen) +{ + struct nv50_screen *screen = nv50_screen(pscreen); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", screen->chipset); + return buffer; +} + +static const char * +nv50_screen_get_vendor(struct pipe_screen *pscreen) +{ + return "nouveau"; +} + +static int +nv50_screen_get_param(struct pipe_screen *pscreen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 32; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 1; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 0; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 8; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 13; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 13; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv50_screen_get_paramf(struct pipe_screen *pscreen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 16.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static void +nv50_screen_destroy(struct pipe_screen *pscreen) +{ + FREE(pscreen); +} + +struct pipe_screen * +nv50_screen_create(struct pipe_winsys *ws, unsigned chipset) +{ + struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen); + + if (!screen) + return NULL; + + screen->chipset = chipset; + + screen->pipe.winsys = ws; + + screen->pipe.destroy = nv50_screen_destroy; + + screen->pipe.get_name = nv50_screen_get_name; + screen->pipe.get_vendor = nv50_screen_get_vendor; + screen->pipe.get_param = nv50_screen_get_param; + screen->pipe.get_paramf = nv50_screen_get_paramf; + + screen->pipe.is_format_supported = nv50_screen_is_format_supported; + + nv50_screen_init_miptree_functions(&screen->pipe); + + return &screen->pipe; +} + diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h new file mode 100644 index 00000000000..45ebbb8051b --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -0,0 +1,20 @@ +#ifndef __NV50_SCREEN_H__ +#define __NV50_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv50_screen { + struct pipe_screen pipe; + unsigned chipset; +}; + +static INLINE struct nv50_screen * +nv50_screen(struct pipe_screen *screen) +{ + return (struct nv50_screen *)screen; +} + +extern struct pipe_screen * +nv50_screen_create(struct pipe_winsys *winsys, unsigned chipset); + +#endif diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index ca92ff02b89..39cf675a57c 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -33,15 +33,6 @@ #include "pipe/p_inlines.h" #include "util/p_tile.h" -static struct pipe_surface * -nv50_get_tex_surface(struct pipe_context *pipe, - struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) -{ - NOUVEAU_ERR("unimplemented\n"); - return NULL; -} - static void nv50_surface_copy(struct pipe_context *pipe, unsigned flip, struct pipe_surface *dest, unsigned destx, unsigned desty, @@ -69,7 +60,6 @@ nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, void nv50_init_surface_functions(struct nv50_context *nv50) { - nv50->pipe.get_tex_surface = nv50_get_tex_surface; nv50->pipe.surface_copy = nv50_surface_copy; nv50->pipe.surface_fill = nv50_surface_fill; } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 2ca05d84c60..1d758e29e75 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -72,23 +72,29 @@ struct pipe_context * nouveau_pipe_create(struct nouveau_context *nv) { struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); - struct pipe_context *(*hw_create)(struct pipe_winsys *, - struct nouveau_winsys *, - unsigned); + struct pipe_screen *(*hws_create)(struct pipe_winsys *, + unsigned chipset); + struct pipe_context *(*hw_create)(struct pipe_screen *, + struct nouveau_winsys *); + struct pipe_winsys *ws; + struct pipe_screen *pscreen; if (!nvws) return NULL; switch (nv->chipset & 0xf0) { case 0x30: + hws_create = nv30_screen_create; hw_create = nv30_create; break; case 0x40: case 0x60: + hws_create = nv40_screen_create; hw_create = nv40_create; break; case 0x50: case 0x80: + hws_create = nv50_screen_create; hw_create = nv50_create; break; default: @@ -119,6 +125,8 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->surface_copy = nouveau_pipe_surface_copy; nvws->surface_fill = nouveau_pipe_surface_fill; - return hw_create(nouveau_create_pipe_winsys(nv), nvws, nv->chipset); + ws = nouveau_create_pipe_winsys(nv); + pscreen = hws_create(ws, nv->chipset); + return hw_create(pscreen, nvws); } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c index 0e1b4273d1e..704f6c77506 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c @@ -61,23 +61,25 @@ nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) return FALSE; } - - struct pipe_context * nouveau_create_softpipe(struct nouveau_context *nv) { - struct nouveau_softpipe_winsys *nvsws; - - nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); - - /* Fill in this struct with callbacks that softpipe will need to - * communicate with the window system, buffer manager, etc. - */ - nvsws->sws.is_format_supported = nouveau_is_format_supported; - nvsws->nv = nv; + struct nouveau_softpipe_winsys *nvsws; + struct pipe_screen *pscreen; + struct pipe_winsys *ws; + + ws = nouveau_create_pipe_winsys(nv); + if (!ws) + return NULL; + pscreen = softpipe_create_screen(ws); + + nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); + if (!nvsws) + return NULL; + + nvsws->sws.is_format_supported = nouveau_is_format_supported; + nvsws->nv = nv; - /* Create the softpipe context: - */ - return softpipe_create(nouveau_create_pipe_winsys(nv), &nvsws->sws); + return softpipe_create(pscreen, ws, &nvsws->sws); } -- cgit v1.2.3 From baaae562f02563c5966b857c61b3eae7341950e3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 29 Feb 2008 22:54:40 +1100 Subject: nouveau: hand nouveau_winsys in with pipe_screen init --- src/gallium/drivers/nouveau/nouveau_winsys.h | 15 +++++++++------ src/gallium/drivers/nv30/nv30_context.c | 3 ++- src/gallium/drivers/nv30/nv30_screen.c | 4 +++- src/gallium/drivers/nv30/nv30_screen.h | 5 ++--- src/gallium/drivers/nv40/nv40_context.c | 3 ++- src/gallium/drivers/nv40/nv40_screen.c | 4 +++- src/gallium/drivers/nv40/nv40_screen.h | 5 ++--- src/gallium/drivers/nv50/nv50_context.c | 3 ++- src/gallium/drivers/nv50/nv50_screen.c | 4 +++- src/gallium/drivers/nv50/nv50_screen.h | 5 ++--- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 8 ++++---- 11 files changed, 34 insertions(+), 25 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 98d95e94a58..11ca7e80ddb 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -50,21 +50,24 @@ struct nouveau_winsys { }; extern struct pipe_screen * -nv30_screen_create(struct pipe_winsys *ws, unsigned chipset); +nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, + unsigned chipset); extern struct pipe_context * -nv30_create(struct pipe_screen *, struct nouveau_winsys *); +nv30_create(struct pipe_screen *); extern struct pipe_screen * -nv40_screen_create(struct pipe_winsys *ws, unsigned chipset); +nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, + unsigned chipset); extern struct pipe_context * -nv40_create(struct pipe_screen *, struct nouveau_winsys *); +nv40_create(struct pipe_screen *); extern struct pipe_screen * -nv50_screen_create(struct pipe_winsys *ws, unsigned chipset); +nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, + unsigned chipset); extern struct pipe_context * -nv50_create(struct pipe_screen *, struct nouveau_winsys *); +nv50_create(struct pipe_screen *); #endif diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index b8452e23b16..522fb132269 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -265,9 +265,10 @@ nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) #define NV35TCL_CHIPSET_3X_MASK 0x000001e0 struct pipe_context * -nv30_create(struct pipe_screen *screen, struct nouveau_winsys *nvws) +nv30_create(struct pipe_screen *screen) { struct pipe_winsys *pipe_winsys = screen->winsys; + struct nouveau_winsys *nvws = nv30_screen(screen)->nvws; unsigned chipset = nv30_screen(screen)->chipset; struct nv30_context *nv30; int rankine_class = 0, ret; diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 6d64025528f..39f2ac1af5c 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -125,7 +125,8 @@ nv30_screen_destroy(struct pipe_screen *screen) } struct pipe_screen * -nv30_screen_create(struct pipe_winsys *winsys, unsigned chipset) +nv30_screen_create(struct pipe_winsys *winsys, struct nouveau_winsys *nvws, + unsigned chipset) { struct nv30_screen *nv30screen = CALLOC_STRUCT(nv30_screen); @@ -133,6 +134,7 @@ nv30_screen_create(struct pipe_winsys *winsys, unsigned chipset) return NULL; nv30screen->chipset = chipset; + nv30screen->nvws = nvws; nv30screen->screen.winsys = winsys; diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h index e55242fbf7f..f878f81e11f 100644 --- a/src/gallium/drivers/nv30/nv30_screen.h +++ b/src/gallium/drivers/nv30/nv30_screen.h @@ -5,6 +5,8 @@ struct nv30_screen { struct pipe_screen screen; + + struct nouveau_winsys *nvws; unsigned chipset; }; @@ -14,7 +16,4 @@ nv30_screen(struct pipe_screen *screen) return (struct nv30_screen *)screen; } -extern struct pipe_screen * -nv30_screen_create(struct pipe_winsys *winsys, unsigned chipset); - #endif diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index a7f64c6e9e5..679c2ddc6b5 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -196,11 +196,12 @@ nv40_destroy(struct pipe_context *pipe) } struct pipe_context * -nv40_create(struct pipe_screen *pscreen, struct nouveau_winsys *nvws) +nv40_create(struct pipe_screen *pscreen) { struct pipe_winsys *ws = pscreen->winsys; struct nv40_context *nv40; unsigned chipset = nv40_screen(pscreen)->chipset; + struct nouveau_winsys *nvws = nv40_screen(pscreen)->nvws; nv40 = CALLOC(1, sizeof(struct nv40_context)); if (!nv40) diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 1941598c641..66e84b68908 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -125,7 +125,8 @@ nv40_screen_destroy(struct pipe_screen *pscreen) } struct pipe_screen * -nv40_screen_create(struct pipe_winsys *ws, unsigned chipset) +nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) { struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen); @@ -133,6 +134,7 @@ nv40_screen_create(struct pipe_winsys *ws, unsigned chipset) return NULL; screen->chipset = chipset; + screen->nvws = nvws; screen->pipe.winsys = ws; screen->pipe.destroy = nv40_screen_destroy; diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h index b30a6c5ad57..88b8fed26cf 100644 --- a/src/gallium/drivers/nv40/nv40_screen.h +++ b/src/gallium/drivers/nv40/nv40_screen.h @@ -5,6 +5,8 @@ struct nv40_screen { struct pipe_screen pipe; + + struct nouveau_winsys *nvws; unsigned chipset; }; @@ -14,7 +16,4 @@ nv40_screen(struct pipe_screen *screen) return (struct nv40_screen *)screen; } -extern struct pipe_screen * -nv40_screen_create(struct pipe_winsys *winsys, unsigned chipset); - #endif diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 98022809a6c..e5054e34f6b 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -56,9 +56,10 @@ nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) #define GRCLASS5097_CHIPSETS 0x00000000 #define GRCLASS8297_CHIPSETS 0x00000010 struct pipe_context * -nv50_create(struct pipe_screen *pscreen, struct nouveau_winsys *nvws) +nv50_create(struct pipe_screen *pscreen) { struct pipe_winsys *pipe_winsys = pscreen->winsys; + struct nouveau_winsys *nvws = nv50_screen(pscreen)->nvws; unsigned chipset = nv50_screen(pscreen)->chipset; struct nv50_context *nv50; int tesla_class, ret; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 8bf82eb0bc0..f091779e3b0 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -90,7 +90,8 @@ nv50_screen_destroy(struct pipe_screen *pscreen) } struct pipe_screen * -nv50_screen_create(struct pipe_winsys *ws, unsigned chipset) +nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, + unsigned chipset) { struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen); @@ -98,6 +99,7 @@ nv50_screen_create(struct pipe_winsys *ws, unsigned chipset) return NULL; screen->chipset = chipset; + screen->nvws = nvws; screen->pipe.winsys = ws; diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index 45ebbb8051b..d664816a031 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -5,6 +5,8 @@ struct nv50_screen { struct pipe_screen pipe; + + struct nouveau_winsys *nvws; unsigned chipset; }; @@ -14,7 +16,4 @@ nv50_screen(struct pipe_screen *screen) return (struct nv50_screen *)screen; } -extern struct pipe_screen * -nv50_screen_create(struct pipe_winsys *winsys, unsigned chipset); - #endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 1d758e29e75..529f5771812 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -73,9 +73,9 @@ nouveau_pipe_create(struct nouveau_context *nv) { struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); struct pipe_screen *(*hws_create)(struct pipe_winsys *, + struct nouveau_winsys *, unsigned chipset); - struct pipe_context *(*hw_create)(struct pipe_screen *, - struct nouveau_winsys *); + struct pipe_context *(*hw_create)(struct pipe_screen *); struct pipe_winsys *ws; struct pipe_screen *pscreen; @@ -126,7 +126,7 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->surface_fill = nouveau_pipe_surface_fill; ws = nouveau_create_pipe_winsys(nv); - pscreen = hws_create(ws, nv->chipset); - return hw_create(pscreen, nvws); + pscreen = hws_create(ws, nvws, nv->chipset); + return hw_create(pscreen); } -- cgit v1.2.3 From 17f6db9d0197657cd753249ef60355c6fd983032 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 29 Feb 2008 23:08:01 +1100 Subject: nv40: move "channel context" stuff into nv40_screen --- src/gallium/drivers/nv40/nv40_context.c | 168 +------------------------ src/gallium/drivers/nv40/nv40_context.h | 29 +---- src/gallium/drivers/nv40/nv40_fragprog.c | 4 +- src/gallium/drivers/nv40/nv40_fragtex.c | 6 +- src/gallium/drivers/nv40/nv40_query.c | 10 +- src/gallium/drivers/nv40/nv40_screen.c | 128 ++++++++++++++++++- src/gallium/drivers/nv40/nv40_screen.h | 15 +++ src/gallium/drivers/nv40/nv40_state.c | 16 +-- src/gallium/drivers/nv40/nv40_state_blend.c | 2 +- src/gallium/drivers/nv40/nv40_state_emit.c | 4 +- src/gallium/drivers/nv40/nv40_state_fb.c | 34 ++--- src/gallium/drivers/nv40/nv40_state_scissor.c | 2 +- src/gallium/drivers/nv40/nv40_state_stipple.c | 2 +- src/gallium/drivers/nv40/nv40_state_viewport.c | 2 +- src/gallium/drivers/nv40/nv40_vbo.c | 10 +- src/gallium/drivers/nv40/nv40_vertprog.c | 8 +- 16 files changed, 203 insertions(+), 237 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 679c2ddc6b5..084829ce281 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -6,10 +6,6 @@ #include "nv40_context.h" #include "nv40_screen.h" -#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf -#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 -#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 - static void nv40_flush(struct pipe_context *pipe, unsigned flags) { @@ -24,7 +20,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) } if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv40->hw->sync, 0); + nvws->notifier_reset(nv40->screen->sync, 0); BEGIN_RING(curie, 0x104, 1); OUT_RING (0); BEGIN_RING(curie, 0x100, 1); @@ -34,149 +30,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) FIRE_RING(); if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv40->hw->sync, 0, 0, 2000); -} - -static void -nv40_channel_takedown(struct nv40_channel_context *cnv40) -{ - struct nouveau_winsys *nvws = cnv40->nvws; - - nvws->res_free(&cnv40->vp_exec_heap); - nvws->res_free(&cnv40->vp_data_heap); - nvws->res_free(&cnv40->query_heap); - nvws->notifier_free(&cnv40->query); - nvws->notifier_free(&cnv40->sync); - nvws->grobj_free(&cnv40->curie); - free(cnv40); -} - -static struct nv40_channel_context * -nv40_channel_init(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) -{ - struct nv40_channel_context *cnv40 = NULL; - struct nouveau_stateobj *so; - unsigned curie_class = 0; - int ret; - - switch (chipset & 0xf0) { - case 0x40: - if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) - curie_class = NV40TCL; - else - if (NV4X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) - curie_class = NV44TCL; - break; - case 0x60: - if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) - curie_class = NV44TCL; - break; - default: - break; - } - - if (!curie_class) { - NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset); - return NULL; - } - - cnv40 = CALLOC(1, sizeof(struct nv40_channel_context)); - if (!cnv40) - return NULL; - cnv40->chipset = chipset; - cnv40->nvws = nvws; - - /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &cnv40->sync); - if (ret) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv40_channel_takedown(cnv40); - return NULL; - } - - /* Query objects */ - ret = nvws->notifier_alloc(nvws, 32, &cnv40->query); - if (ret) { - NOUVEAU_ERR("Error initialising query objects: %d\n", ret); - nv40_channel_takedown(cnv40); - return NULL; - } - - ret = nvws->res_init(&cnv40->query_heap, 0, 32); - if (ret) { - NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); - nv40_channel_takedown(cnv40); - return NULL; - } - - /* Vtxprog resources */ - if (nvws->res_init(&cnv40->vp_exec_heap, 0, 512) || - nvws->res_init(&cnv40->vp_data_heap, 0, 256)) { - nv40_channel_takedown(cnv40); - return NULL; - } - - /* 3D object */ - ret = nvws->grobj_alloc(nvws, curie_class, &cnv40->curie); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - /* Static curie initialisation */ - so = so_new(128, 0); - so_method(so, cnv40->curie, NV40TCL_DMA_NOTIFY, 1); - so_data (so, cnv40->sync->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_TEXTURE0, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->gart->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_COLOR1, 1); - so_data (so, nvws->channel->vram->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_COLOR0, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->vram->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_VTXBUF0, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->gart->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_FENCE, 2); - so_data (so, 0); - so_data (so, cnv40->query->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_UNK01AC, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->vram->handle); - so_method(so, cnv40->curie, NV40TCL_DMA_COLOR2, 2); - so_data (so, nvws->channel->vram->handle); - so_data (so, nvws->channel->vram->handle); - - so_method(so, cnv40->curie, 0x1ea4, 3); - so_data (so, 0x00000010); - so_data (so, 0x01000100); - so_data (so, 0xff800006); - - /* vtxprog output routing */ - so_method(so, cnv40->curie, 0x1fc4, 1); - so_data (so, 0x06144321); - so_method(so, cnv40->curie, 0x1fc8, 2); - so_data (so, 0xedcba987); - so_data (so, 0x00000021); - so_method(so, cnv40->curie, 0x1fd0, 1); - so_data (so, 0x00171615); - so_method(so, cnv40->curie, 0x1fd4, 1); - so_data (so, 0x001b1a19); - - so_method(so, cnv40->curie, 0x1ef8, 1); - so_data (so, 0x0020ffff); - so_method(so, cnv40->curie, 0x1d64, 1); - so_data (so, 0x00d30000); - so_method(so, cnv40->curie, 0x1e94, 1); - so_data (so, 0x00000001); - - so_emit(nvws, so); - so_ref(NULL, &so); - nvws->push_flush(nvws->channel, 0); - - return cnv40; + nvws->notifier_wait(nv40->screen->sync, 0, 0, 2000); } static void @@ -186,32 +40,22 @@ nv40_destroy(struct pipe_context *pipe) if (nv40->draw) draw_destroy(nv40->draw); - - if (nv40->hw) { - if (--nv40->hw->refcount == 0) - nv40_channel_takedown(nv40->hw); - } - free(nv40); } struct pipe_context * nv40_create(struct pipe_screen *pscreen) { + struct nv40_screen *screen = nv40_screen(pscreen); struct pipe_winsys *ws = pscreen->winsys; struct nv40_context *nv40; - unsigned chipset = nv40_screen(pscreen)->chipset; - struct nouveau_winsys *nvws = nv40_screen(pscreen)->nvws; + unsigned chipset = screen->chipset; + struct nouveau_winsys *nvws = screen->nvws; nv40 = CALLOC(1, sizeof(struct nv40_context)); if (!nv40) return NULL; - - nv40->hw = nv40_channel_init(ws, nvws, chipset); - if (!nv40->hw) { - nv40_destroy(&nv40->pipe); - return NULL; - } + nv40->screen = screen; nv40->chipset = chipset; nv40->nvws = nvws; diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 3ddfbd43f60..3b669594dc3 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -11,7 +11,7 @@ #include "nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv40_channel_context *ctx = nv40->hw + struct nv40_screen *ctx = nv40->screen #include "nouveau/nouveau_push.h" #include "nouveau/nouveau_stateobj.h" @@ -59,6 +59,8 @@ enum nv40_state_index { NV40_STATE_MAX = 33 }; +#include "nv40_screen.h" + #define NV40_NEW_BLEND (1 << 0) #define NV40_NEW_RAST (1 << 1) #define NV40_NEW_ZSA (1 << 2) @@ -76,28 +78,6 @@ enum nv40_state_index { #define NV40_FALLBACK_TNL (1 << 0) #define NV40_FALLBACK_RAST (1 << 1) -struct nv40_channel_context { - struct nouveau_winsys *nvws; - unsigned refcount; - - unsigned chipset; - - /* HW graphics objects */ - struct nouveau_grobj *curie; - struct nouveau_notifier *sync; - - /* Query object resources */ - struct nouveau_notifier *query; - struct nouveau_resource *query_heap; - - /* Vtxprog resources */ - struct nouveau_resource *vp_exec_heap; - struct nouveau_resource *vp_data_heap; - - /* Current 3D state of channel */ - struct nouveau_stateobj *state[NV40_STATE_MAX]; -}; - struct nv40_rasterizer_state { struct pipe_rasterizer_state pipe; struct nouveau_stateobj *so; @@ -125,9 +105,10 @@ struct nv40_state { struct nv40_context { struct pipe_context pipe; + struct nouveau_winsys *nvws; + struct nv40_screen *screen; - struct nv40_channel_context *hw; struct draw_context *draw; int chipset; diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 2a8abb32a30..3c4ea7e99ed 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -810,11 +810,11 @@ nv40_fragprog_validate(struct nv40_context *nv40) nv40_fragprog_upload(nv40, fp); so = so_new(4, 1); - so_method(so, nv40->hw->curie, NV40TCL_FP_ADDRESS, 1); + so_method(so, nv40->screen->curie, NV40TCL_FP_ADDRESS, 1); so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1); - so_method(so, nv40->hw->curie, NV40TCL_FP_CONTROL, 1); + so_method(so, nv40->screen->curie, NV40TCL_FP_CONTROL, 1); so_data (so, fp->fp_control); so_ref(so, &fp->so); diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 6be8378c087..436f954ceca 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -106,7 +106,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) txs = tf->swizzle; so = so_new(16, 2); - so_method(so, nv40->hw->curie, NV40TCL_TEX_OFFSET(unit), 8); + so_method(so, nv40->screen->curie, NV40TCL_TEX_OFFSET(unit), 8); so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1); @@ -117,7 +117,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); so_data (so, ps->bcol); - so_method(so, nv40->hw->curie, NV40TCL_TEX_SIZE1(unit), 1); + so_method(so, nv40->screen->curie, NV40TCL_TEX_SIZE1(unit), 1); so_data (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp); return so; @@ -137,7 +137,7 @@ nv40_fragtex_validate(struct nv40_context *nv40) samplers &= ~(1 << unit); so = so_new(2, 0); - so_method(so, nv40->hw->curie, NV40TCL_TEX_ENABLE(unit), 1); + so_method(so, nv40->screen->curie, NV40TCL_TEX_ENABLE(unit), 1); so_data (so, 0); so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]); state->dirty |= (1ULL << (NV40_STATE_FRAGTEX0 + unit)); diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c index 8bca2788b93..03178456246 100644 --- a/src/gallium/drivers/nv40/nv40_query.c +++ b/src/gallium/drivers/nv40/nv40_query.c @@ -45,9 +45,9 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - if (nv40->nvws->res_alloc(nv40->hw->query_heap, 1, NULL, &q->object)) + if (nv40->nvws->res_alloc(nv40->screen->query_heap, 1, NULL, &q->object)) assert(0); - nv40->nvws->notifier_reset(nv40->hw->query, q->object->start); + nv40->nvws->notifier_reset(nv40->screen->query, q->object->start); BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1); OUT_RING (1); @@ -82,17 +82,17 @@ nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq, if (!q->ready) { unsigned status; - status = nvws->notifier_status(nv40->hw->query, + status = nvws->notifier_status(nv40->screen->query, q->object->start); if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { if (wait == FALSE) return FALSE; - nvws->notifier_wait(nv40->hw->query, q->object->start, + nvws->notifier_wait(nv40->screen->query, q->object->start, NV_NOTIFY_STATE_STATUS_COMPLETED, 0); } - q->result = nvws->notifier_retval(nv40->hw->query, + q->result = nvws->notifier_retval(nv40->screen->query, q->object->start); q->ready = TRUE; nvws->res_free(&q->object); diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 66e84b68908..268ca83ce0d 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -4,6 +4,10 @@ #include "nv40_context.h" #include "nv40_screen.h" +#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf +#define NV4X_GRCLASS4497_CHIPSETS 0x00005450 +#define NV6X_GRCLASS4497_CHIPSETS 0x00000088 + static const char * nv40_screen_get_name(struct pipe_screen *pscreen) { @@ -121,6 +125,16 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, static void nv40_screen_destroy(struct pipe_screen *pscreen) { + struct nv40_screen *screen = nv40_screen(pscreen); + struct nouveau_winsys *nvws = screen->nvws; + + nvws->res_free(&screen->vp_exec_heap); + nvws->res_free(&screen->vp_data_heap); + nvws->res_free(&screen->query_heap); + nvws->notifier_free(&screen->query); + nvws->notifier_free(&screen->sync); + nvws->grobj_free(&screen->curie); + FREE(pscreen); } @@ -129,13 +143,125 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, unsigned chipset) { struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen); + struct nouveau_stateobj *so; + unsigned curie_class; + int ret; if (!screen) return NULL; - screen->chipset = chipset; screen->nvws = nvws; + /* 3D object */ + switch (chipset & 0xf0) { + case 0x40: + if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV40TCL; + else + if (NV4X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV44TCL; + break; + case 0x60: + if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f))) + curie_class = NV44TCL; + break; + default: + break; + } + + if (!curie_class) { + NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset); + return NULL; + } + + ret = nvws->grobj_alloc(nvws, curie_class, &screen->curie); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv40_screen_destroy(&screen->pipe); + return NULL; + } + + /* Query objects */ + ret = nvws->notifier_alloc(nvws, 32, &screen->query); + if (ret) { + NOUVEAU_ERR("Error initialising query objects: %d\n", ret); + nv40_screen_destroy(&screen->pipe); + return NULL; + } + + ret = nvws->res_init(&screen->query_heap, 0, 32); + if (ret) { + NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); + nv40_screen_destroy(&screen->pipe); + return NULL; + } + + /* Vtxprog resources */ + if (nvws->res_init(&screen->vp_exec_heap, 0, 512) || + nvws->res_init(&screen->vp_data_heap, 0, 256)) { + nv40_screen_destroy(&screen->pipe); + return NULL; + } + + /* Static curie initialisation */ + so = so_new(128, 0); + so_method(so, screen->curie, NV40TCL_DMA_NOTIFY, 1); + so_data (so, screen->sync->handle); + so_method(so, screen->curie, NV40TCL_DMA_TEXTURE0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, screen->curie, NV40TCL_DMA_COLOR1, 1); + so_data (so, nvws->channel->vram->handle); + so_method(so, screen->curie, NV40TCL_DMA_COLOR0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, screen->curie, NV40TCL_DMA_VTXBUF0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, screen->curie, NV40TCL_DMA_FENCE, 2); + so_data (so, 0); + so_data (so, screen->query->handle); + so_method(so, screen->curie, NV40TCL_DMA_UNK01AC, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, screen->curie, NV40TCL_DMA_COLOR2, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + + so_method(so, screen->curie, 0x1ea4, 3); + so_data (so, 0x00000010); + so_data (so, 0x01000100); + so_data (so, 0xff800006); + + /* vtxprog output routing */ + so_method(so, screen->curie, 0x1fc4, 1); + so_data (so, 0x06144321); + so_method(so, screen->curie, 0x1fc8, 2); + so_data (so, 0xedcba987); + so_data (so, 0x00000021); + so_method(so, screen->curie, 0x1fd0, 1); + so_data (so, 0x00171615); + so_method(so, screen->curie, 0x1fd4, 1); + so_data (so, 0x001b1a19); + + so_method(so, screen->curie, 0x1ef8, 1); + so_data (so, 0x0020ffff); + so_method(so, screen->curie, 0x1d64, 1); + so_data (so, 0x00d30000); + so_method(so, screen->curie, 0x1e94, 1); + so_data (so, 0x00000001); + + so_emit(nvws, so); + so_ref(NULL, &so); + nvws->push_flush(nvws->channel, 0); + screen->pipe.winsys = ws; screen->pipe.destroy = nv40_screen_destroy; diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h index 88b8fed26cf..9f9668dbb6b 100644 --- a/src/gallium/drivers/nv40/nv40_screen.h +++ b/src/gallium/drivers/nv40/nv40_screen.h @@ -8,6 +8,21 @@ struct nv40_screen { struct nouveau_winsys *nvws; unsigned chipset; + + /* HW graphics objects */ + struct nouveau_grobj *curie; + struct nouveau_notifier *sync; + + /* Query object resources */ + struct nouveau_notifier *query; + struct nouveau_resource *query_heap; + + /* Vtxprog resources */ + struct nouveau_resource *vp_exec_heap; + struct nouveau_resource *vp_data_heap; + + /* Current 3D state of channel */ + struct nouveau_stateobj *state[NV40_STATE_MAX]; }; static INLINE struct nv40_screen * diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 24335fbc44d..caa2f9df0c6 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -10,7 +10,7 @@ nv40_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_grobj *curie = nv40->hw->curie; + struct nouveau_grobj *curie = nv40->screen->curie; struct nv40_blend_state *bso = CALLOC(1, sizeof(*bso)); struct nouveau_stateobj *so = so_new(16, 0); @@ -286,7 +286,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nv40_rasterizer_state *rsso = CALLOC(1, sizeof(*rsso)); struct nouveau_stateobj *so = so_new(32, 0); - struct nouveau_grobj *curie = nv40->hw->curie; + struct nouveau_grobj *curie = nv40->screen->curie; /*XXX: ignored: * light_twoside @@ -420,18 +420,18 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nv40_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso)); struct nouveau_stateobj *so = so_new(32, 0); - so_method(so, nv40->hw->curie, NV40TCL_DEPTH_FUNC, 3); + so_method(so, nv40->screen->curie, NV40TCL_DEPTH_FUNC, 3); so_data (so, nvgl_comparison_op(cso->depth.func)); so_data (so, cso->depth.writemask ? 1 : 0); so_data (so, cso->depth.enabled ? 1 : 0); - so_method(so, nv40->hw->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_method(so, nv40->screen->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); so_data (so, cso->alpha.enabled ? 1 : 0); so_data (so, nvgl_comparison_op(cso->alpha.func)); so_data (so, float_to_ubyte(cso->alpha.ref)); if (cso->stencil[0].enabled) { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_method(so, nv40->screen->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); so_data (so, cso->stencil[0].enabled ? 1 : 0); so_data (so, cso->stencil[0].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[0].func)); @@ -441,12 +441,12 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); } else { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_method(so, nv40->screen->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); so_data (so, 0); } if (cso->stencil[1].enabled) { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_method(so, nv40->screen->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); so_data (so, cso->stencil[1].enabled ? 1 : 0); so_data (so, cso->stencil[1].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[1].func)); @@ -456,7 +456,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); } else { - so_method(so, nv40->hw->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_method(so, nv40->screen->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); so_data (so, 0); } diff --git a/src/gallium/drivers/nv40/nv40_state_blend.c b/src/gallium/drivers/nv40/nv40_state_blend.c index dd09830aa3d..95e6d7394f4 100644 --- a/src/gallium/drivers/nv40/nv40_state_blend.c +++ b/src/gallium/drivers/nv40/nv40_state_blend.c @@ -21,7 +21,7 @@ nv40_state_blend_colour_validate(struct nv40_context *nv40) struct nouveau_stateobj *so = so_new(2, 0); struct pipe_blend_color *bcol = &nv40->blend_colour; - so_method(so, nv40->hw->curie, NV40TCL_BLEND_COLOR, 1); + so_method(so, nv40->screen->curie, NV40TCL_BLEND_COLOR, 1); so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | (float_to_ubyte(bcol->color[0]) << 16) | (float_to_ubyte(bcol->color[1]) << 8) | diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index bb2ce0f7221..221503617c3 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -67,8 +67,8 @@ nv40_state_emit(struct nv40_context *nv40) while (state->dirty) { unsigned idx = ffsll(state->dirty) - 1; - so_ref (state->hw[idx], &nv40->hw->state[idx]); - so_emit(nv40->nvws, nv40->hw->state[idx]); + so_ref (state->hw[idx], &nv40->screen->state[idx]); + so_emit(nv40->nvws, nv40->screen->state[idx]); state->dirty &= ~(1ULL << idx); } diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index 3d0ab920030..71795ab1824 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -72,73 +72,73 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) } if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR0, 1); + so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR0, 1); so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR0_PITCH, 2); + so_method(so, nv40->screen->curie, NV40TCL_COLOR0_PITCH, 2); so_data (so, rt[0]->pitch * rt[0]->cpp); so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR1, 1); + so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR1, 1); so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR1_OFFSET, 2); + so_method(so, nv40->screen->curie, NV40TCL_COLOR1_OFFSET, 2); so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); so_data (so, rt[1]->pitch * rt[1]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR2, 1); + so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR2, 1); so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR2_OFFSET, 1); + so_method(so, nv40->screen->curie, NV40TCL_COLOR2_OFFSET, 1); so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_COLOR2_PITCH, 1); + so_method(so, nv40->screen->curie, NV40TCL_COLOR2_PITCH, 1); so_data (so, rt[2]->pitch * rt[2]->cpp); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_COLOR3, 1); + so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR3, 1); so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_COLOR3_OFFSET, 1); + so_method(so, nv40->screen->curie, NV40TCL_COLOR3_OFFSET, 1); so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_COLOR3_PITCH, 1); + so_method(so, nv40->screen->curie, NV40TCL_COLOR3_PITCH, 1); so_data (so, rt[3]->pitch * rt[3]->cpp); } if (zeta_format) { - so_method(so, nv40->hw->curie, NV40TCL_DMA_ZETA, 1); + so_method(so, nv40->screen->curie, NV40TCL_DMA_ZETA, 1); so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); - so_method(so, nv40->hw->curie, NV40TCL_ZETA_OFFSET, 1); + so_method(so, nv40->screen->curie, NV40TCL_ZETA_OFFSET, 1); so_reloc (so, zeta->buffer, zeta->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv40->hw->curie, NV40TCL_ZETA_PITCH, 1); + so_method(so, nv40->screen->curie, NV40TCL_ZETA_PITCH, 1); so_data (so, zeta->pitch * zeta->cpp); } - so_method(so, nv40->hw->curie, NV40TCL_RT_ENABLE, 1); + so_method(so, nv40->screen->curie, NV40TCL_RT_ENABLE, 1); so_data (so, rt_enable); - so_method(so, nv40->hw->curie, NV40TCL_RT_HORIZ, 3); + so_method(so, nv40->screen->curie, NV40TCL_RT_HORIZ, 3); so_data (so, (w << 16) | 0); so_data (so, (h << 16) | 0); so_data (so, rt_format); - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_HORIZ, 2); + so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_HORIZ, 2); so_data (so, (w << 16) | 0); so_data (so, (h << 16) | 0); - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); so_data (so, ((w - 1) << 16) | 0); so_data (so, ((h - 1) << 16) | 0); diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c index 09ffc49f965..9e9eadc5116 100644 --- a/src/gallium/drivers/nv40/nv40_state_scissor.c +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -12,7 +12,7 @@ nv40_state_scissor_validate(struct nv40_context *nv40) return FALSE; so = so_new(3, 0); - so_method(so, nv40->hw->curie, NV40TCL_SCISSOR_HORIZ, 2); + so_method(so, nv40->screen->curie, NV40TCL_SCISSOR_HORIZ, 2); if (rast->scissor) { so_data (so, ((s->maxx - s->minx) << 16) | s->minx); so_data (so, ((s->maxy - s->miny) << 16) | s->miny); diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c index 001c396d747..b51024ad9b2 100644 --- a/src/gallium/drivers/nv40/nv40_state_stipple.c +++ b/src/gallium/drivers/nv40/nv40_state_stipple.c @@ -4,7 +4,7 @@ static boolean nv40_state_stipple_validate(struct nv40_context *nv40) { struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe; - struct nouveau_grobj *curie = nv40->hw->curie; + struct nouveau_grobj *curie = nv40->screen->curie; struct nouveau_stateobj *so; if (nv40->state.hw[NV40_STATE_STIPPLE] && diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c index 9616be5052a..3a325339072 100644 --- a/src/gallium/drivers/nv40/nv40_state_viewport.c +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -6,7 +6,7 @@ nv40_state_viewport_validate(struct nv40_context *nv40) struct nouveau_stateobj *so = so_new(9, 0); struct pipe_viewport_state *vpt = &nv40->viewport; - so_method(so, nv40->hw->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); so_data (so, fui(vpt->translate[0])); so_data (so, fui(vpt->translate[1])); so_data (so, fui(vpt->translate[2])); diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 1653ebf2a7e..bedc8c6d4ef 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -52,7 +52,7 @@ nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, } /* No support for 8bit indices, no support at all on 0x4497 chips */ - if (nv40->hw->curie->grclass == NV44TCL || ib_size == 1) + if (nv40->screen->curie->grclass == NV44TCL || ib_size == 1) return FALSE; switch (ib_size) { @@ -365,9 +365,9 @@ nv40_vbo_validate(struct nv40_context *nv40) num_hw++; vtxbuf = so_new(20, 18); - so_method(vtxbuf, nv40->hw->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + so_method(vtxbuf, nv40->screen->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); vtxfmt = so_new(17, 0); - so_method(vtxfmt, nv40->hw->curie, NV40TCL_VTXFMT(0), num_hw); + so_method(vtxfmt, nv40->screen->curie, NV40TCL_VTXFMT(0), num_hw); inputs = vp->ir; for (hw = 0; hw < num_hw; hw++) { @@ -399,13 +399,13 @@ nv40_vbo_validate(struct nv40_context *nv40) } if (ib) { - so_method(vtxbuf, nv40->hw->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_method(vtxbuf, nv40->screen->curie, NV40TCL_IDXBUF_ADDRESS, 2); so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, 0, NV40TCL_IDXBUF_FORMAT_DMA1); } - so_method(vtxbuf, nv40->hw->curie, 0x1710, 1); + so_method(vtxbuf, nv40->screen->curie, 0x1710, 1); so_data (vtxbuf, 0); so_ref(vtxbuf, &nv40->state.hw[NV40_STATE_VTXBUF]); diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index d3ed57b1998..5b7a343e55d 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -655,7 +655,7 @@ nv40_vertprog_validate(struct nv40_context *nv40) check_gpu_resources: /* Allocate hw vtxprog exec slots */ if (!vp->exec) { - struct nouveau_resource *heap = nv40->hw->vp_exec_heap; + struct nouveau_resource *heap = nv40->screen->vp_exec_heap; struct nouveau_stateobj *so; uint vplen = vp->nr_insns; @@ -672,9 +672,9 @@ check_gpu_resources: } so = so_new(5, 0); - so_method(so, nv40->hw->curie, NV40TCL_VP_START_FROM_ID, 1); + so_method(so, nv40->screen->curie, NV40TCL_VP_START_FROM_ID, 1); so_data (so, vp->exec->start); - so_method(so, nv40->hw->curie, NV40TCL_VP_ATTRIB_EN, 2); + so_method(so, nv40->screen->curie, NV40TCL_VP_ATTRIB_EN, 2); so_data (so, vp->ir); so_data (so, vp->or); so_ref(so, &vp->so); @@ -684,7 +684,7 @@ check_gpu_resources: /* Allocate hw vtxprog const slots */ if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv40->hw->vp_data_heap; + struct nouveau_resource *heap = nv40->screen->vp_data_heap; if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { while (heap->next && heap->size < vp->nr_consts) { -- cgit v1.2.3 From 0a12e4587ccf2c4fa71e93bb00b4582deb99a82c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 2 Mar 2008 14:09:57 +1100 Subject: nouveau: turn pushbuf macros into inline functions --- src/gallium/winsys/dri/nouveau/nouveau_context.c | 93 +++++++--- src/gallium/winsys/dri/nouveau/nouveau_context.h | 35 ++-- src/gallium/winsys/dri/nouveau/nouveau_local.h | 133 ++++++++------ .../winsys/dri/nouveau/nouveau_swapbuffers.c | 2 +- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 13 +- src/gallium/winsys/dri/nouveau/nv04_surface.c | 192 ++++++++++++--------- src/gallium/winsys/dri/nouveau/nv50_surface.c | 187 +++++++++++--------- 7 files changed, 396 insertions(+), 259 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index 01fada5b89b..7915afae226 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -21,6 +21,75 @@ static const struct dri_debug_control debug_control[] = { int __nouveau_debug = 0; #endif +static void +nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) +{ + nouveau_grobj_free(&nvc->NvNull); + nouveau_grobj_free(&nvc->NvCtxSurf2D); + nouveau_grobj_free(&nvc->NvImageBlit); + nouveau_grobj_free(&nvc->NvGdiRect); + nouveau_grobj_free(&nvc->NvM2MF); + nouveau_grobj_free(&nvc->Nv2D); + + nouveau_notifier_free(&nvc->sync_notifier); + + nouveau_channel_free(&nvc->channel); + + FREE(nvc); +} + +static struct nouveau_channel_context * +nouveau_channel_context_create(struct nouveau_device *nvdev, unsigned chipset) +{ + struct nouveau_channel_context *nvc; + int ret; + + nvc = CALLOC_STRUCT(nouveau_channel_context); + if (!nvc) + return NULL; + nvc->chipset = chipset; + + if ((ret = nouveau_channel_alloc(nvdev, 0x8003d001, 0x8003d002, + &nvc->channel))) { + NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + if ((ret = nouveau_grobj_alloc(nvc->channel, 0x00000000, 0x30, + &nvc->NvNull))) { + NOUVEAU_ERR("Error creating NULL object: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + nvc->next_handle = 0x80000000; + + if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1, + &nvc->sync_notifier))) { + NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + switch (chipset) { + case 0x50: + case 0x80: + ret = nouveau_surface_channel_create_nv50(nvc); + break; + default: + ret = nouveau_surface_channel_create_nv04(nvc); + break; + } + + if (ret) { + NOUVEAU_ERR("Error initialising surface objects: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + return nvc; +} + GLboolean nouveau_context_create(const __GLcontextModes *glVis, __DRIcontextPrivate *driContextPriv, @@ -45,13 +114,6 @@ nouveau_context_create(const __GLcontextModes *glVis, return GL_FALSE; } - if ((ret = nouveau_channel_alloc(nv_screen->device, - 0x8003d001, 0x8003d002, - &nv->channel))) { - NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); - return GL_FALSE; - } - driContextPriv->driverPrivate = (void *)nv; nv->nv_screen = nv_screen; nv->dri_screen = driScrnPriv; @@ -101,16 +163,9 @@ nouveau_context_create(const __GLcontextModes *glVis, nv->frontbuffer = fb_surf; } - if ((ret = nouveau_grobj_alloc(nv->channel, 0x00000000, 0x30, - &nv->NvNull))) { - NOUVEAU_ERR("Error creating NULL object: %d\n", ret); - return GL_FALSE; - } - nv->next_handle = 0x80000000; - - if ((ret = nouveau_notifier_alloc(nv->channel, nv->next_handle++, 1, - &nv->sync_notifier))) { - NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); + nv->nvc = nouveau_channel_context_create(&nvdev->base, nv->chipset); + if (!nv->nvc) { + NOUVEAU_ERR("Failed initialising GPU channel context\n"); return GL_FALSE; } @@ -152,9 +207,7 @@ nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) st_flush(nv->st, PIPE_FLUSH_WAIT); st_destroy_context(nv->st); - nouveau_grobj_free(&nv->NvCtxSurf2D); - nouveau_grobj_free(&nv->NvImageBlit); - nouveau_channel_free(&nv->channel); + nouveau_channel_context_destroy(nv->nvc); free(nv); } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index 5805f969ba8..736c8d8bef3 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -13,6 +13,25 @@ struct nouveau_framebuffer { struct st_framebuffer *stfb; }; +struct nouveau_channel_context { + unsigned chipset; + + struct nouveau_channel *channel; + + struct nouveau_notifier *sync_notifier; + + struct nouveau_grobj *NvNull; + struct nouveau_grobj *NvCtxSurf2D; + struct nouveau_grobj *NvImageBlit; + struct nouveau_grobj *NvGdiRect; + struct nouveau_grobj *NvM2MF; + struct nouveau_grobj *Nv2D; + + uint32_t next_handle; + uint32_t next_subchannel; + uint32_t next_sequence; +}; + struct nouveau_context { struct st_context *st; @@ -31,17 +50,7 @@ struct nouveau_context { struct pipe_surface *frontbuffer; /* Hardware context */ - struct nouveau_channel *channel; - struct nouveau_notifier *sync_notifier; - struct nouveau_grobj *NvNull; - struct nouveau_grobj *NvCtxSurf2D; - struct nouveau_grobj *NvImageBlit; - struct nouveau_grobj *NvGdiRect; - struct nouveau_grobj *NvM2MF; - struct nouveau_grobj *Nv2D; - uint32_t next_handle; - uint32_t next_subchannel; - uint32_t next_sequence; + struct nouveau_channel_context *nvc; /* pipe_surface accel */ struct pipe_surface *surf_src, *surf_dst; @@ -80,6 +89,10 @@ extern int __nouveau_debug; extern void LOCK_HARDWARE(struct nouveau_context *); extern void UNLOCK_HARDWARE(struct nouveau_context *); +extern int +nouveau_surface_channel_create_nv04(struct nouveau_channel_context *); +extern int +nouveau_surface_channel_create_nv50(struct nouveau_channel_context *); extern int nouveau_surface_init_nv04(struct nouveau_context *); extern int nouveau_surface_init_nv50(struct nouveau_context *); diff --git a/src/gallium/winsys/dri/nouveau/nouveau_local.h b/src/gallium/winsys/dri/nouveau/nouveau_local.h index 59febca2929..f7d91fc7487 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_local.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_local.h @@ -1,8 +1,11 @@ #ifndef __NOUVEAU_LOCAL_H__ #define __NOUVEAU_LOCAL_H__ +#include "pipe/p_compiler.h" #include +struct pipe_buffer; + /* Debug output */ #define NOUVEAU_MSG(fmt, args...) do { \ fprintf(stdout, "nouveau: "fmt, ##args); \ @@ -24,66 +27,90 @@ #define NOUVEAU_DMA_TIMEOUT 2000 /* Push buffer access macros */ -#define OUT_RING(data) do { \ - (*nv->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - memcpy(nv->channel->pushbuf->cur, (src), (size)<<2); \ - nv->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define FIRE_RING() do { \ - nouveau_pushbuf_flush(nv->channel, 0); \ -} while(0) - -#define BEGIN_RING_GR(obj,mthd,size) do { \ - if (nv->channel->pushbuf->remaining < ((size) + 1)) \ - nouveau_pushbuf_flush(nv->channel, ((size) + 1)); \ - OUT_RING(((obj)->subc << 13) | ((size) << 18) | (mthd)); \ - nv->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - BEGIN_RING_GR(nv->obj, (mthd), (size)); \ -} while(0) - -#define BIND_RING(o,s) do { \ - nv->o->subc = (s); \ - BEGIN_RING(o, 0x0000, 1); \ - OUT_RING (nv->o->handle); \ -} while(0) - -#define OUT_RELOC(buf,data,flags,vor,tor) do { \ - nouveau_pipe_emit_reloc(nv->channel, nv->channel->pushbuf->cur++, \ - buf, (data), (flags), (vor), (tor)); \ -} while(0) +static INLINE void +OUT_RING(struct nouveau_channel *chan, unsigned data) +{ + *(chan->pushbuf->cur++) = (data); +} + +static INLINE void +OUT_RINGp(struct nouveau_channel *chan, uint32_t *data, unsigned size) +{ + memcpy(chan->pushbuf->cur, data, size * 4); + chan->pushbuf->cur += size; +} + +static INLINE void +OUT_RINGf(struct nouveau_channel *chan, float f) +{ + union { uint32_t i; float f; } c; + c.f = f; + OUT_RING(chan, c.i); +} + +static INLINE void +BEGIN_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, + unsigned mthd, unsigned size) +{ + if (chan->pushbuf->remaining < (size + 1)) + nouveau_pushbuf_flush(chan, (size + 1)); + OUT_RING(chan, (gr->subc << 13) | (size << 18) | mthd); + chan->pushbuf->remaining -= (size + 1); +} + +static INLINE void +FIRE_RING(struct nouveau_channel *chan) +{ + nouveau_pushbuf_flush(chan, 0); +} + +static INLINE void +BIND_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, unsigned subc) +{ + gr->subc = subc; + BEGIN_RING(chan, gr, 0x0000, 1); + OUT_RING (chan, gr->handle); +} + +static INLINE void +OUT_RELOC(struct nouveau_channel *chan, struct pipe_buffer *buf, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + nouveau_pipe_emit_reloc(chan, chan->pushbuf->cur++, buf, + data, flags, vor, tor); +} /* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) +static INLINE void +OUT_RELOCd(struct nouveau_channel *chan, struct pipe_buffer *buf, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + OUT_RELOC(chan, buf, data, flags | NOUVEAU_BO_OR, vor, tor); +} /* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - nv->channel->vram->handle, nv->channel->gart->handle); \ -} while(0) +static INLINE void +OUT_RELOCo(struct nouveau_channel *chan, struct pipe_buffer *buf, + unsigned flags) +{ + OUT_RELOC(chan, buf, 0, flags | NOUVEAU_BO_OR, + chan->vram->handle, chan->gart->handle); +} /* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) +static INLINE void +OUT_RELOCl(struct nouveau_channel *chan, struct pipe_buffer *buf, + unsigned delta, unsigned flags) +{ + OUT_RELOC(chan, buf, delta, flags | NOUVEAU_BO_LOW, 0, 0); +} /* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) +static INLINE void +OUT_RELOCh(struct nouveau_channel *chan, struct pipe_buffer *buf, + unsigned delta, unsigned flags) +{ + OUT_RELOC(chan, buf, delta, flags | NOUVEAU_BO_HIGH, 0, 0); +} #endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c index 91bf243f424..70e0104e83b 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c @@ -42,7 +42,7 @@ nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, nv->surface_copy(nv, dx, dy, sx, sy, w, h); } - FIRE_RING(); + FIRE_RING(nv->nvc->channel); UNLOCK_HARDWARE(nv); if (nv->last_stamp != dPriv->lastStamp) { diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 529f5771812..50d7549b1b7 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -11,7 +11,7 @@ nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, { struct nouveau_context *nv = nvws->nv; - return nouveau_notifier_alloc(nv->channel, nv->next_handle++, + return nouveau_notifier_alloc(nv->nvc->channel, nv->nvc->next_handle++, count, notify); } @@ -20,17 +20,16 @@ nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, struct nouveau_grobj **grobj) { struct nouveau_context *nv = nvws->nv; + struct nouveau_channel *chan = nv->nvc->channel; int ret; - ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, + ret = nouveau_grobj_alloc(chan, nv->nvc->next_handle++, grclass, grobj); if (ret) return ret; - (*grobj)->subc = nv->next_subchannel++; - assert((*grobj)->subc <= 7); - BEGIN_RING_GR(*grobj, 0x0000, 1); - OUT_RING ((*grobj)->handle); + assert(nv->nvc->next_subchannel < 7); + BIND_RING(chan, *grobj, nv->nvc->next_subchannel++); return 0; } @@ -103,7 +102,7 @@ nouveau_pipe_create(struct nouveau_context *nv) } nvws->nv = nv; - nvws->channel = nv->channel; + nvws->channel = nv->nvc->channel; nvws->res_init = nouveau_resource_init; nvws->res_alloc = nouveau_resource_alloc; diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c index fe1ea4ed70f..cdcd71eaad6 100644 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -30,6 +30,7 @@ static void nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, unsigned sx, unsigned sy, unsigned w, unsigned h) { + struct nouveau_channel *chan = nv->nvc->channel; struct pipe_surface *dst = nv->surf_dst; struct pipe_surface *src = nv->surf_src; unsigned dst_offset, src_offset; @@ -40,17 +41,18 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, while (h) { int count = (h > 2047) ? 2047 : h; - BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); - OUT_RELOCl(src->buffer, src_offset, NOUVEAU_BO_VRAM | + BEGIN_RING(chan, nv->nvc->NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + OUT_RELOCl(chan, src->buffer, src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCl(dst->buffer, dst_offset, NOUVEAU_BO_VRAM | + OUT_RELOCl(chan, dst->buffer, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); - OUT_RING (src->pitch * src->cpp); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (w * src->cpp); - OUT_RING (count); - OUT_RING (0x0101); - OUT_RING (0); + OUT_RING (chan, src->pitch * src->cpp); + OUT_RING (chan, dst->pitch * dst->cpp); + OUT_RING (chan, w * src->cpp); + OUT_RING (chan, count); + OUT_RING (chan, 0x0101); + OUT_RING (chan, 0); h -= count; src_offset += src->pitch * src->cpp * count; @@ -62,16 +64,19 @@ static void nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, unsigned sx, unsigned sy, unsigned w, unsigned h) { - BEGIN_RING(NvImageBlit, 0x0300, 3); - OUT_RING ((sy << 16) | sx); - OUT_RING ((dy << 16) | dx); - OUT_RING (( h << 16) | w); + struct nouveau_channel *chan = nv->nvc->channel; + + BEGIN_RING(chan, nv->nvc->NvImageBlit, 0x0300, 3); + OUT_RING (chan, (sy << 16) | sx); + OUT_RING (chan, (dy << 16) | dx); + OUT_RING (chan, ( h << 16) | w); } static int nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, struct pipe_surface *src) { + struct nouveau_channel *chan = nv->nvc->channel; int format; if (src->cpp != dst->cpp) @@ -81,12 +86,12 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, * to NV_MEMORY_TO_MEMORY_FORMAT in this case. */ if ((src->offset & 63) || (dst->offset & 63)) { - BEGIN_RING(NvM2MF, + BEGIN_RING(nv->nvc->channel, nv->nvc->NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | - NOUVEAU_BO_WR); + OUT_RELOCo(chan, src->buffer, NOUVEAU_BO_GART | + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_GART | + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); nv->surface_copy = nv04_surface_copy_m2mf; nv->surf_dst = dst; @@ -101,15 +106,20 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, } nv->surface_copy = nv04_surface_copy_blit; - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (format); - OUT_RING (((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); - OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, + NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(chan, src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, + NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (chan, format); + OUT_RING (chan, ((dst->pitch * dst->cpp) << 16) | + (src->pitch * src->cpp)); + OUT_RELOCl(chan, src->buffer, src->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); return 0; } @@ -117,7 +127,7 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, static void nv04_surface_copy_done(struct nouveau_context *nv) { - FIRE_RING(); + FIRE_RING(nv->nvc->channel); } static int @@ -125,6 +135,9 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, unsigned dx, unsigned dy, unsigned w, unsigned h, unsigned value) { + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *surf2d = nv->nvc->NvCtxSurf2D; + struct nouveau_grobj *rect = nv->nvc->NvGdiRect; int cs2d_format, gdirect_format; if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { @@ -137,86 +150,99 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, return 1; } - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (cs2d_format); - OUT_RING (((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); - OUT_RING (gdirect_format); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); - OUT_RING (value); - BEGIN_RING(NvGdiRect, + BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (chan, cs2d_format); + OUT_RING (chan, ((dst->pitch * dst->cpp) << 16) | + (dst->pitch * dst->cpp)); + OUT_RELOCl(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); + OUT_RING (chan, gdirect_format); + BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); + OUT_RING (chan, value); + BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); - OUT_RING ((dx << 16) | dy); - OUT_RING (( w << 16) | h); + OUT_RING (chan, (dx << 16) | dy); + OUT_RING (chan, ( w << 16) | h); - FIRE_RING(); + FIRE_RING(chan); return 0; } int -nouveau_surface_init_nv04(struct nouveau_context *nv) +nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) { + struct nouveau_channel *chan = nvc->channel; unsigned class; int ret; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, 0x39, - &nv->NvM2MF))) { + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, 0x39, + &nvc->NvM2MF))) { NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); return 1; } - BIND_RING (NvM2MF, nv->next_subchannel++); - BEGIN_RING(NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - - class = nv->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : - NV10_CONTEXT_SURFACES_2D; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvCtxSurf2D))) { + BIND_RING (chan, nvc->NvM2MF, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + + class = nvc->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : + NV10_CONTEXT_SURFACES_2D; + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvCtxSurf2D))) { NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); return 1; } - BIND_RING (NvCtxSurf2D, nv->next_subchannel++); - BEGIN_RING(NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RING (nv->channel->vram->handle); - OUT_RING (nv->channel->vram->handle); - - class = nv->chipset < 0x10 ? NV04_IMAGE_BLIT : - NV12_IMAGE_BLIT; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvImageBlit))) { + BIND_RING (chan, nvc->NvCtxSurf2D, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvCtxSurf2D, + NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RING (chan, nvc->channel->vram->handle); + OUT_RING (chan, nvc->channel->vram->handle); + + class = nvc->chipset < 0x10 ? NV04_IMAGE_BLIT : + NV12_IMAGE_BLIT; + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvImageBlit))) { NOUVEAU_ERR("Error creating blit object: %d\n", ret); return 1; } - BIND_RING (NvImageBlit, nv->next_subchannel++); - BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); - OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); - OUT_RING (NV04_IMAGE_BLIT_OPERATION_SRCCOPY); + BIND_RING (chan, nvc->NvImageBlit, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); + OUT_RING (chan, nvc->NvCtxSurf2D->handle); + BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); + OUT_RING (chan, NV04_IMAGE_BLIT_OPERATION_SRCCOPY); class = NV04_GDI_RECTANGLE_TEXT; - if ((ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, class, - &nv->NvGdiRect))) { + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvGdiRect))) { NOUVEAU_ERR("Error creating rect object: %d\n", ret); return 1; } - BIND_RING (NvGdiRect, nv->next_subchannel++); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); - OUT_RING (nv->NvCtxSurf2D->handle); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); - OUT_RING (NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); - BEGIN_RING(NvGdiRect, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); - OUT_RING (NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); + BIND_RING (chan, nvc->NvGdiRect, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); + OUT_RING (chan, nvc->NvCtxSurf2D->handle); + BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); + OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); + BEGIN_RING(chan, nvc->NvGdiRect, + NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); + OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); + return 0; +} + +int +nouveau_surface_init_nv04(struct nouveau_context *nv) +{ nv->surface_copy_prep = nv04_surface_copy_prep; nv->surface_copy = nv04_surface_copy_blit; nv->surface_copy_done = nv04_surface_copy_done; diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c index 15a10028613..5d74fb8d2bc 100644 --- a/src/gallium/winsys/dri/nouveau/nv50_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv50_surface.c @@ -19,6 +19,8 @@ static int nv50_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, struct pipe_surface *src) { + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *eng2d = nv->nvc->Nv2D; int surf_format; assert(src->cpp == dst->cpp); @@ -26,34 +28,38 @@ nv50_surface_copy_prep(struct nouveau_context *nv, surf_format = nv50_format(dst->cpp); assert(surf_format >= 0); - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RELOCo(src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); - OUT_RING (0); - OUT_RING (0); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - - BEGIN_RING(Nv2D, NV50_2D_SRC_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_SRC_PITCH, 5); - OUT_RING (src->pitch * src->cpp); - OUT_RING (src->pitch); - OUT_RING (src->height); - OUT_RELOCh(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(src->buffer, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RELOCo(chan, src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 2); + OUT_RING (chan, surf_format); + OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, NV50_2D_DST_PITCH, 5); + OUT_RING (chan, dst->pitch * dst->cpp); + OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->height); + OUT_RELOCh(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->height); + + BEGIN_RING(chan, eng2d, NV50_2D_SRC_FORMAT, 2); + OUT_RING (chan, surf_format); + OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, NV50_2D_SRC_PITCH, 5); + OUT_RING (chan, src->pitch * src->cpp); + OUT_RING (chan, src->pitch); + OUT_RING (chan, src->height); + OUT_RELOCh(chan, src->buffer, src->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(chan, src->buffer, src->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); return 0; } @@ -62,27 +68,30 @@ static void nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, unsigned sx, unsigned sy, unsigned w, unsigned h) { - BEGIN_RING(Nv2D, 0x0110, 1); - OUT_RING (0); - BEGIN_RING(Nv2D, NV50_2D_BLIT_DST_X, 12); - OUT_RING (dx); - OUT_RING (dy); - OUT_RING (w); - OUT_RING (h); - OUT_RING (0); - OUT_RING (1); - OUT_RING (0); - OUT_RING (1); - OUT_RING (0); - OUT_RING (sx); - OUT_RING (0); - OUT_RING (sy); + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *eng2d = nv->nvc->Nv2D; + + BEGIN_RING(chan, eng2d, 0x0110, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, eng2d, NV50_2D_BLIT_DST_X, 12); + OUT_RING (chan, dx); + OUT_RING (chan, dy); + OUT_RING (chan, w); + OUT_RING (chan, h); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, sx); + OUT_RING (chan, 0); + OUT_RING (chan, sy); } static void nv50_surface_copy_done(struct nouveau_context *nv) { - FIRE_RING(); + FIRE_RING(nv->nvc->channel); } static int @@ -90,6 +99,8 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, unsigned dx, unsigned dy, unsigned w, unsigned h, unsigned value) { + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *eng2d = nv->nvc->Nv2D; int surf_format, rect_format; surf_format = nv50_format(dst->cpp); @@ -100,57 +111,65 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, if (rect_format < 0) return 1; - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY1, 1); - OUT_RELOCo(dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_DST_FORMAT, 2); - OUT_RING (surf_format); - OUT_RING (1); - BEGIN_RING(Nv2D, NV50_2D_DST_PITCH, 5); - OUT_RING (dst->pitch * dst->cpp); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - OUT_RELOCh(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(dst->buffer, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(Nv2D, NV50_2D_CLIP_X, 4); - OUT_RING (0); - OUT_RING (0); - OUT_RING (dst->pitch); - OUT_RING (dst->height); - - BEGIN_RING(Nv2D, 0x0580, 3); - OUT_RING (4); - OUT_RING (rect_format); - OUT_RING (value); - - BEGIN_RING(Nv2D, NV50_2D_RECT_X1, 4); - OUT_RING (dx); - OUT_RING (dy); - OUT_RING (dx + w); - OUT_RING (dy + h); - - FIRE_RING(); + BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY1, 1); + OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 2); + OUT_RING (chan, surf_format); + OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, NV50_2D_DST_PITCH, 5); + OUT_RING (chan, dst->pitch * dst->cpp); + OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->height); + OUT_RELOCh(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(chan, dst->buffer, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->height); + + BEGIN_RING(chan, eng2d, 0x0580, 3); + OUT_RING (chan, 4); + OUT_RING (chan, rect_format); + OUT_RING (chan, value); + + BEGIN_RING(chan, eng2d, NV50_2D_RECT_X1, 4); + OUT_RING (chan, dx); + OUT_RING (chan, dy); + OUT_RING (chan, dx + w); + OUT_RING (chan, dy + h); + + FIRE_RING(chan); return 0; } int -nouveau_surface_init_nv50(struct nouveau_context *nv) +nouveau_surface_channel_create_nv50(struct nouveau_channel_context *nvc) { int ret; - ret = nouveau_grobj_alloc(nv->channel, nv->next_handle++, NV50_2D, - &nv->Nv2D); + ret = nouveau_grobj_alloc(nvc->channel, nvc->next_handle++, NV50_2D, + &nvc->Nv2D); if (ret) return ret; - BIND_RING (Nv2D, 0); - BEGIN_RING(Nv2D, NV50_2D_DMA_NOTIFY, 1); - OUT_RING (nv->sync_notifier->handle); - BEGIN_RING(Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RING (nv->channel->vram->handle); - OUT_RING (nv->channel->vram->handle); - BEGIN_RING(Nv2D, NV50_2D_OPERATION, 1); - OUT_RING (NV50_2D_OPERATION_SRCCOPY); + BIND_RING (nvc->channel, nvc->Nv2D, 0); + BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_DMA_NOTIFY, 1); + OUT_RING (nvc->channel, nvc->sync_notifier->handle); + BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); + OUT_RING (nvc->channel, nvc->channel->vram->handle); + OUT_RING (nvc->channel, nvc->channel->vram->handle); + BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_OPERATION, 1); + OUT_RING (nvc->channel, NV50_2D_OPERATION_SRCCOPY); + + return 0; +} +int +nouveau_surface_init_nv50(struct nouveau_context *nv) +{ nv->surface_copy_prep = nv50_surface_copy_prep; nv->surface_copy = nv50_surface_copy; nv->surface_copy_done = nv50_surface_copy_done; -- cgit v1.2.3 From b560ed2444383b9634786fe742b8cb6f5cdfc781 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 2 Mar 2008 14:56:42 +1100 Subject: nouveau: enable multi-context/single-channel support for nv40 --- src/gallium/drivers/nouveau/nouveau_winsys.h | 6 +-- src/gallium/drivers/nv30/nv30_context.c | 2 +- src/gallium/drivers/nv40/nv40_context.c | 3 +- src/gallium/drivers/nv40/nv40_context.h | 1 + src/gallium/drivers/nv40/nv40_screen.h | 2 + src/gallium/drivers/nv40/nv40_state_emit.c | 10 ++++ src/gallium/drivers/nv50/nv50_context.c | 2 +- src/gallium/winsys/dri/nouveau/nouveau_context.c | 63 +++++++++++++++++++++--- src/gallium/winsys/dri/nouveau/nouveau_context.h | 8 +++ src/gallium/winsys/dri/nouveau/nouveau_screen.h | 2 + src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 10 ++-- 11 files changed, 94 insertions(+), 15 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 11ca7e80ddb..44c8bb919bb 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -54,20 +54,20 @@ nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, unsigned chipset); extern struct pipe_context * -nv30_create(struct pipe_screen *); +nv30_create(struct pipe_screen *, unsigned pctx_id); extern struct pipe_screen * nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, unsigned chipset); extern struct pipe_context * -nv40_create(struct pipe_screen *); +nv40_create(struct pipe_screen *, unsigned pctx_id); extern struct pipe_screen * nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, unsigned chipset); extern struct pipe_context * -nv50_create(struct pipe_screen *); +nv50_create(struct pipe_screen *, unsigned pctx_id); #endif diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 522fb132269..28d3f057cc4 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -265,7 +265,7 @@ nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) #define NV35TCL_CHIPSET_3X_MASK 0x000001e0 struct pipe_context * -nv30_create(struct pipe_screen *screen) +nv30_create(struct pipe_screen *screen, unsigned pctx_id) { struct pipe_winsys *pipe_winsys = screen->winsys; struct nouveau_winsys *nvws = nv30_screen(screen)->nvws; diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 084829ce281..203c843a013 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -44,7 +44,7 @@ nv40_destroy(struct pipe_context *pipe) } struct pipe_context * -nv40_create(struct pipe_screen *pscreen) +nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) { struct nv40_screen *screen = nv40_screen(pscreen); struct pipe_winsys *ws = pscreen->winsys; @@ -56,6 +56,7 @@ nv40_create(struct pipe_screen *pscreen) if (!nv40) return NULL; nv40->screen = screen; + nv40->pctx_id = pctx_id; nv40->chipset = chipset; nv40->nvws = nvws; diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 3b669594dc3..e118776306b 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -108,6 +108,7 @@ struct nv40_context { struct nouveau_winsys *nvws; struct nv40_screen *screen; + unsigned pctx_id; struct draw_context *draw; diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h index 9f9668dbb6b..3ea78aadfd9 100644 --- a/src/gallium/drivers/nv40/nv40_screen.h +++ b/src/gallium/drivers/nv40/nv40_screen.h @@ -9,6 +9,8 @@ struct nv40_screen { struct nouveau_winsys *nvws; unsigned chipset; + unsigned cur_pctx; + /* HW graphics objects */ struct nouveau_grobj *curie; struct nouveau_notifier *sync; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 221503617c3..a95e2472e2b 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -62,8 +62,18 @@ static void nv40_state_emit(struct nv40_context *nv40) { struct nv40_state *state = &nv40->state; + struct nv40_screen *screen = nv40->screen; unsigned i, samplers; + if (nv40->pctx_id != screen->cur_pctx) { + for (i = 0; i < NV40_STATE_MAX; i++) { + if (screen->state[i] != state->hw[i] && state->hw[i]) + state->dirty |= (1ULL << i); + } + + screen->cur_pctx = nv40->pctx_id; + } + while (state->dirty) { unsigned idx = ffsll(state->dirty) - 1; diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index e5054e34f6b..c937b8de6d4 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -56,7 +56,7 @@ nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) #define GRCLASS5097_CHIPSETS 0x00000000 #define GRCLASS8297_CHIPSETS 0x00000010 struct pipe_context * -nv50_create(struct pipe_screen *pscreen) +nv50_create(struct pipe_screen *pscreen, unsigned pctx_id) { struct pipe_winsys *pipe_winsys = pscreen->winsys; struct nouveau_winsys *nvws = nv50_screen(pscreen)->nvws; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index 7915afae226..dc852c9f498 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -4,6 +4,7 @@ #include "utils.h" #include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" #include "pipe/p_defines.h" #include "pipe/p_context.h" @@ -101,7 +102,8 @@ nouveau_context_create(const __GLcontextModes *glVis, struct nouveau_device_priv *nvdev; struct pipe_context *pipe = NULL; struct st_context *st_share = NULL; - int ret; + struct nouveau_channel_context *nvc = NULL; + int i, ret; if (sharedContextPrivate) { st_share = ((struct nouveau_context *)sharedContextPrivate)->st; @@ -163,12 +165,56 @@ nouveau_context_create(const __GLcontextModes *glVis, nv->frontbuffer = fb_surf; } - nv->nvc = nouveau_channel_context_create(&nvdev->base, nv->chipset); - if (!nv->nvc) { - NOUVEAU_ERR("Failed initialising GPU channel context\n"); - return GL_FALSE; + /* Attempt to share a single channel between multiple contexts from + * a single process. + */ + nvc = nv_screen->nvc; + if (!nvc && st_share) { + struct nouveau_context *snv = st_share->pipe->priv; + if (snv) { + nvc = snv->nvc; + } + } + + /*XXX: temporary - disable multi-context/single-channel on non-NV4x */ + switch (nv->chipset & 0xf0) { + case 0x40: + case 0x60: + break; + default: + nvc = NULL; + break; + } + + if (!nvc) { + nvc = nouveau_channel_context_create(&nvdev->base, nv->chipset); + if (!nvc) { + NOUVEAU_ERR("Failed initialising GPU context\n"); + return GL_FALSE; + } + nv_screen->nvc = nvc; + } + + nvc->refcount++; + nv->nvc = nvc; + + /* Find a free slot for a pipe context, allocate a new one if needed */ + nv->pctx_id = -1; + for (i = 0; i < nvc->nr_pctx; i++) { + if (nvc->pctx[i] == NULL) { + nv->pctx_id = i; + break; + } } + if (nv->pctx_id < 0) { + nv->pctx_id = nvc->nr_pctx++; + nvc->pctx = + realloc(nvc->pctx, + sizeof(struct pipe_context *) * nvc->nr_pctx); + } + + /* Create pipe */ if (nv->chipset < 0x50) ret = nouveau_surface_init_nv04(nv); else @@ -201,13 +247,18 @@ void nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) { struct nouveau_context *nv = driContextPriv->driverPrivate; + struct nouveau_channel_context *nvc = nv->nvc; assert(nv); st_flush(nv->st, PIPE_FLUSH_WAIT); st_destroy_context(nv->st); - nouveau_channel_context_destroy(nv->nvc); + if (nv->pctx_id >= 0) { + nvc->pctx[nv->pctx_id] = NULL; + if (--nvc->refcount <= 0) + nouveau_channel_context_destroy(nvc); + } free(nv); } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index 736c8d8bef3..92f551855a4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -14,6 +14,13 @@ struct nouveau_framebuffer { }; struct nouveau_channel_context { + struct pipe_screen *pscreen; + int refcount; + + unsigned cur_pctx; + unsigned nr_pctx; + struct pipe_context **pctx; + unsigned chipset; struct nouveau_channel *channel; @@ -51,6 +58,7 @@ struct nouveau_context { /* Hardware context */ struct nouveau_channel_context *nvc; + int pctx_id; /* pipe_surface accel */ struct pipe_surface *surf_src, *surf_dst; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.h b/src/gallium/winsys/dri/nouveau/nouveau_screen.h index 019823bd44d..e9da2026904 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.h @@ -14,6 +14,8 @@ struct nouveau_screen { uint32_t front_pitch; uint32_t front_cpp; uint32_t front_height; + + void *nvc; }; #endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 50d7549b1b7..87619bdcfb4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -70,11 +70,12 @@ nouveau_pipe_emit_reloc(struct nouveau_channel *chan, void *ptr, struct pipe_context * nouveau_pipe_create(struct nouveau_context *nv) { + struct nouveau_channel_context *nvc = nv->nvc; struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); struct pipe_screen *(*hws_create)(struct pipe_winsys *, struct nouveau_winsys *, unsigned chipset); - struct pipe_context *(*hw_create)(struct pipe_screen *); + struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned); struct pipe_winsys *ws; struct pipe_screen *pscreen; @@ -125,7 +126,10 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->surface_fill = nouveau_pipe_surface_fill; ws = nouveau_create_pipe_winsys(nv); - pscreen = hws_create(ws, nvws, nv->chipset); - return hw_create(pscreen); + + if (!nvc->pscreen) + nvc->pscreen = hws_create(ws, nvws, nv->chipset); + nvc->pctx[nv->pctx_id] = hw_create(nvc->pscreen, nv->pctx_id); + return nvc->pctx[nv->pctx_id]; } -- cgit v1.2.3 From 59d4b7cc626704dbbd9c817019ec2dd9183322ad Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 2 Mar 2008 15:28:24 +1100 Subject: nv40: fix segv when app "skips" texture units. --- src/gallium/drivers/nv40/nv40_state_emit.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index a95e2472e2b..9f268640e00 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -67,7 +67,7 @@ nv40_state_emit(struct nv40_context *nv40) if (nv40->pctx_id != screen->cur_pctx) { for (i = 0; i < NV40_STATE_MAX; i++) { - if (screen->state[i] != state->hw[i] && state->hw[i]) + if (state->hw[i] && screen->state[i] != state->hw[i]) state->dirty |= (1ULL << i); } @@ -84,6 +84,8 @@ nv40_state_emit(struct nv40_context *nv40) so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { + if (!(samplers & (1 << i))) + continue; so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FRAGTEX0+i]); samplers &= ~(1ULL << i); -- cgit v1.2.3 From 578b5cd9a030189bcba5c3e86080e1e26eb6e108 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 2 Mar 2008 16:35:47 +1100 Subject: nouveau: fix potential crash --- src/gallium/winsys/dri/nouveau/nouveau_context.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index dc852c9f498..2e54729aa97 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -256,8 +256,10 @@ nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) if (nv->pctx_id >= 0) { nvc->pctx[nv->pctx_id] = NULL; - if (--nvc->refcount <= 0) + if (--nvc->refcount <= 0) { nouveau_channel_context_destroy(nvc); + nv->nv_screen->nvc = NULL; + } } free(nv); -- cgit v1.2.3 From a5966c8b28702bba1e4eb4bb4aec2247c90fcfa2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 2 Mar 2008 16:39:24 +1100 Subject: nouveau: silence some warnings --- src/gallium/winsys/dri/nouveau/nouveau_local.h | 6 ++++-- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 3 +-- 2 files changed, 5 insertions(+), 4 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_local.h b/src/gallium/winsys/dri/nouveau/nouveau_local.h index f7d91fc7487..8a52f26714f 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_local.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_local.h @@ -2,6 +2,7 @@ #define __NOUVEAU_LOCAL_H__ #include "pipe/p_compiler.h" +#include "nouveau_winsys_pipe.h" #include struct pipe_buffer; @@ -76,8 +77,9 @@ static INLINE void OUT_RELOC(struct nouveau_channel *chan, struct pipe_buffer *buf, unsigned data, unsigned flags, unsigned vor, unsigned tor) { - nouveau_pipe_emit_reloc(chan, chan->pushbuf->cur++, buf, - data, flags, vor, tor); + nouveau_pushbuf_emit_reloc(chan, chan->pushbuf->cur++, + nouveau_buffer(buf)->bo, + data, flags, vor, tor); } /* Raw data + flags depending on FB/TT buffer */ diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 87619bdcfb4..dc7c4c3d71d 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -58,7 +58,7 @@ nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, return 0; } -int +static int nouveau_pipe_emit_reloc(struct nouveau_channel *chan, void *ptr, struct pipe_buffer *buf, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor) @@ -77,7 +77,6 @@ nouveau_pipe_create(struct nouveau_context *nv) unsigned chipset); struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned); struct pipe_winsys *ws; - struct pipe_screen *pscreen; if (!nvws) return NULL; -- cgit v1.2.3 From 57b8711aebdce9bc21bf3311c50dbfb0f9ad6d42 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 2 Mar 2008 16:48:15 +1100 Subject: nv40: nuke debug --- src/gallium/drivers/nv40/nv40_fragprog.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 3c4ea7e99ed..d981a02a639 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -416,8 +416,6 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, ai = fsrc->SrcRegister.Index; src[i] = tgsi_src(fpc, fsrc); } else { - NOUVEAU_MSG("extra src attr %d\n", - fsrc->SrcRegister.Index); src[i] = temp(fpc); arith(fpc, 0, MOV, src[i], MASK_ALL, tgsi_src(fpc, fsrc), none, none); -- cgit v1.2.3 From 1de15ad83e5a6902ac57212a3df63bb9b829bc20 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 3 Mar 2008 00:01:44 +1100 Subject: nv40: re-do vtxbuf format code --- src/gallium/drivers/nv40/nv40_vbo.c | 79 +++++++++++++++++++++++-------------- 1 file changed, 50 insertions(+), 29 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index bedc8c6d4ef..f16afc23b84 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -9,34 +9,53 @@ #include "nouveau/nouveau_pushbuf.h" static INLINE int -nv40_vbo_ncomp(uint format) +nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) { - int ncomp = 0; - - if (pf_size_x(format)) ncomp++; - if (pf_size_y(format)) ncomp++; - if (pf_size_z(format)) ncomp++; - if (pf_size_w(format)) ncomp++; - - return ncomp; -} - -static INLINE int -nv40_vbo_type(uint format) -{ - switch (pf_type(format)) { - case PIPE_FORMAT_TYPE_FLOAT: - return NV40TCL_VTXFMT_TYPE_FLOAT; - case PIPE_FORMAT_TYPE_UNORM: - return NV40TCL_VTXFMT_TYPE_UBYTE; + char fs[128]; + + switch (pipe) { + case PIPE_FORMAT_R32_FLOAT: + case PIPE_FORMAT_R32G32_FLOAT: + case PIPE_FORMAT_R32G32B32_FLOAT: + case PIPE_FORMAT_R32G32B32A32_FLOAT: + *fmt = NV40TCL_VTXFMT_TYPE_FLOAT; + break; + case PIPE_FORMAT_R8_UNORM: + case PIPE_FORMAT_R8G8_UNORM: + case PIPE_FORMAT_R8G8B8_UNORM: + case PIPE_FORMAT_R8G8B8A8_UNORM: + *fmt = NV40TCL_VTXFMT_TYPE_UBYTE; + break; default: - { - char fs[128]; - pf_sprint_name(fs, format); + pf_sprint_name(fs, pipe); NOUVEAU_ERR("Unknown format %s\n", fs); - return NV40TCL_VTXFMT_TYPE_FLOAT; + return 1; } + + switch (pipe) { + case PIPE_FORMAT_R8_UNORM: + case PIPE_FORMAT_R32_FLOAT: + *ncomp = 1; + break; + case PIPE_FORMAT_R8G8_UNORM: + case PIPE_FORMAT_R32G32_FLOAT: + *ncomp = 2; + break; + case PIPE_FORMAT_R8G8B8_UNORM: + case PIPE_FORMAT_R32G32B32_FLOAT: + *ncomp = 3; + break; + case PIPE_FORMAT_R8G8B8A8_UNORM: + case PIPE_FORMAT_R32G32B32A32_FLOAT: + *ncomp = 4; + break; + default: + pf_sprint_name(fs, pipe); + NOUVEAU_ERR("Unknown format %s\n", fs); + return 1; } + + return 0; } static boolean @@ -82,11 +101,11 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, struct pipe_vertex_buffer *vb) { struct pipe_winsys *ws = nv40->pipe.winsys; - int type, ncomp; + unsigned type, ncomp; void *map; - type = nv40_vbo_type(ve->src_format); - ncomp = nv40_vbo_ncomp(ve->src_format); + if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp)) + return FALSE; map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); map += vb->buffer_offset + ve->src_offset; @@ -373,6 +392,7 @@ nv40_vbo_validate(struct nv40_context *nv40) for (hw = 0; hw < num_hw; hw++) { struct pipe_vertex_element *ve; struct pipe_vertex_buffer *vb; + unsigned type, ncomp; if (!(inputs & (1 << hw))) { so_data(vtxbuf, 0); @@ -389,13 +409,14 @@ nv40_vbo_validate(struct nv40_context *nv40) continue; } + if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp)) + assert(0); + so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, 0, NV40TCL_VTXBUF_ADDRESS_DMA1); so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) | - (nv40_vbo_ncomp(ve->src_format) << - NV40TCL_VTXFMT_SIZE_SHIFT) | - nv40_vbo_type(ve->src_format))); + (ncomp << NV40TCL_VTXFMT_SIZE_SHIFT) | type)); } if (ib) { -- cgit v1.2.3 From 82f22d9e147ed55c2ca513ebc2d069e197d36ea8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 6 Mar 2008 11:52:25 +1100 Subject: nv30: a couple of vtxprog fixes --- src/gallium/drivers/nv30/nv30_context.c | 2 +- src/gallium/drivers/nv30/nv30_vertprog.c | 17 ++++++++++++++--- 2 files changed, 15 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 28d3f057cc4..0124f9af0fc 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -319,7 +319,7 @@ nv30_create(struct pipe_screen *screen, unsigned pctx_id) } /* Vtxprog resources */ - if (nvws->res_init(&nv30->vertprog.exec_heap, 0, 512) || + if (nvws->res_init(&nv30->vertprog.exec_heap, 0, 256) || nvws->res_init(&nv30->vertprog.data_heap, 0, 256)) { nv30_destroy(&nv30->pipe); return NULL; diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 4a8269d5dd7..548b6907d02 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -225,6 +225,18 @@ nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, // hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; // hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); + if (dst.type == NV30SR_OUTPUT) { + if (slot) + hw[3] |= (mask << NV30_VP_INST_SDEST_WRITEMASK_SHIFT); + else + hw[3] |= (mask << NV30_VP_INST_VDEST_WRITEMASK_SHIFT); + } else { + if (slot) + hw[3] |= (mask << NV30_VP_INST_STEMP_WRITEMASK_SHIFT); + else + hw[3] |= (mask << NV30_VP_INST_VTEMP_WRITEMASK_SHIFT); + } + emit_dst(vpc, hw, slot, dst); emit_src(vpc, hw, 0, s0); emit_src(vpc, hw, 1, s1); @@ -752,7 +764,7 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) } #endif BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (/*vp->exec->start*/0); + OUT_RING (vp->exec->start); for (i = 0; i < vp->nr_insns; i++) { BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); OUT_RINGp (vp->insns[i].data, 4); @@ -760,8 +772,7 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) } BEGIN_RING(rankine, NV34TCL_VP_START_FROM_ID, 1); -// OUT_RING (vp->exec->start); - OUT_RING (0); + OUT_RING (vp->exec->start); nv30->vertprog.active = vp; } -- cgit v1.2.3 From 66ba021e9fa2fa932cb9be8fc2fb8272baf51fc6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 7 Mar 2008 22:48:26 +1100 Subject: nouveau: another "argh gallium fscks with assert" fix --- src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c index 7d5eddb92ff..fd9a5c5a96d 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c @@ -193,7 +193,8 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) nouveau_fence_ref(NULL, &fence); /* Allocate space for next push buffer */ - assert(!nouveau_pushbuf_space(chan, min)); + ret = nouveau_pushbuf_space(chan, min); + assert(!ret); return 0; } -- cgit v1.2.3 From ae0e047ba4e05d25d6e0b9b0574e36c7e8ccd510 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Mar 2008 14:27:22 +1100 Subject: nv30: put the card into vtxprog mode + small cleanups/fixes --- src/gallium/drivers/nv30/nv30_context.c | 122 ++----------------------------- src/gallium/drivers/nv30/nv30_vertprog.c | 5 ++ 2 files changed, 10 insertions(+), 117 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 0124f9af0fc..b3906e28e3f 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -133,129 +133,17 @@ nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) BEGIN_RING(rankine, NV34TCL_RC_ENABLE, 1); OUT_RING (0); - /* Attempt to setup a known state.. Probably missing a heap of - * stuff here.. - */ - BEGIN_RING(rankine, NV34TCL_STENCIL_FRONT_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_DEPTH_WRITE_ENABLE, 2); - OUT_RING (0); /* wr disable */ - OUT_RING (0); /* test disable */ - BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); - OUT_RING (0x01010101); /* TR,TR,TR,TR */ - BEGIN_RING(rankine, NV34TCL_CULL_FACE_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 5); - OUT_RING (0); /* Blend enable */ - OUT_RING (0); /* Blend src */ - OUT_RING (0); /* Blend dst */ - OUT_RING (0x00000000); /* Blend colour */ - OUT_RING (0x8006); /* FUNC_ADD */ - BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (0); - OUT_RING (0x1503 /*GL_COPY*/); - BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); - OUT_RING (1); - BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); - OUT_RING (0x1d01 /*GL_SMOOTH*/); - BEGIN_RING(rankine, NV34TCL_POLYGON_OFFSET_FACTOR,2); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 2); - OUT_RING (0x1b02 /*GL_FILL*/); - OUT_RING (0x1b02 /*GL_FILL*/); - /* - Disable texture units - * - Set fragprog to MOVR result.color, fragment.color */ - for (i=0;i<16;i++) { - BEGIN_RING(rankine, - NV34TCL_TX_ENABLE(i), 1); - OUT_RING (0); - } - /* Polygon stipple */ - BEGIN_RING(rankine, - NV34TCL_POLYGON_STIPPLE_PATTERN(0), 0x20); - for (i=0;i<0x20;i++) - OUT_RING (0xFFFFFFFF); - - int w=4096; - int h=4096; - int pitch=4096*4; - BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 5); - OUT_RING (w<<16); - OUT_RING (h<<16); - OUT_RING (0x148); /* format */ - OUT_RING (pitch << 16 | pitch); - OUT_RING (0x0); - BEGIN_RING(rankine, 0x0a00, 2); - OUT_RING ((w<<16) | 0); - OUT_RING ((h<<16) | 0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING ((w-1)<<16); - OUT_RING ((h-1)<<16); - BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); - OUT_RING (w<<16); - OUT_RING (h<<16); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); - OUT_RING (w<<16); - OUT_RING (h<<16); - - BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - - BEGIN_RING(rankine, NV34TCL_MODELVIEW_MATRIX(0), 16); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - - BEGIN_RING(rankine, NV34TCL_PROJECTION_MATRIX(0), 16); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); + BEGIN_RING(rankine, NV34TCL_DEPTH_RANGE_NEAR, 2); OUT_RINGf (0.0); OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - - BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); - OUT_RING (4096<<16); - OUT_RING (4096<<16); BEGIN_RING(rankine, NV34TCL_MULTISAMPLE_CONTROL, 1); OUT_RING (0xffff0000); + /* enables use of vp rather than fixed-function somehow */ + BEGIN_RING(rankine, 0x1e94, 1); + OUT_RING (0x13); + FIRE_RING (); return TRUE; } diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 548b6907d02..96bc4b5ef9a 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -193,6 +193,11 @@ emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); + + /*XXX: no way this is entirely correct, someone needs to + * figure out what exactly it is. + */ + hw[3] |= 0x800; break; default: assert(0); -- cgit v1.2.3 From 1fb3c94f03e07a80bb7a93777d4fef5173da71ca Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 02:20:40 +1100 Subject: nv50: some forgotten changes --- src/gallium/drivers/nv50/nv50_draw.c | 45 +++++++++++++++--------- src/gallium/drivers/nv50/nv50_screen.c | 2 ++ src/gallium/winsys/dri/nouveau/nouveau_context.c | 2 +- 3 files changed, 32 insertions(+), 17 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c index c6ed6838c6a..790408c6df8 100644 --- a/src/gallium/drivers/nv50/nv50_draw.c +++ b/src/gallium/drivers/nv50/nv50_draw.c @@ -3,53 +3,66 @@ #include "nv50_context.h" -struct nv50_draw_stage { - struct draw_stage draw; +struct nv50_render_stage { + struct draw_stage stage; struct nv50_context *nv50; }; +static INLINE struct nv50_render_stage * +nv50_render_stage(struct draw_stage *stage) +{ + return (struct nv50_render_stage *)stage; +} + static void -nv50_draw_point(struct draw_stage *draw, struct prim_header *prim) +nv50_render_point(struct draw_stage *stage, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv50_draw_line(struct draw_stage *draw, struct prim_header *prim) +nv50_render_line(struct draw_stage *stage, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv50_draw_tri(struct draw_stage *draw, struct prim_header *prim) +nv50_render_tri(struct draw_stage *stage, struct prim_header *prim) { NOUVEAU_ERR("\n"); } static void -nv50_draw_flush(struct draw_stage *draw, unsigned flags) +nv50_render_flush(struct draw_stage *stage, unsigned flags) { } static void -nv50_draw_reset_stipple_counter(struct draw_stage *draw) +nv50_render_reset_stipple_counter(struct draw_stage *stage) { NOUVEAU_ERR("\n"); } +static void +nv50_render_destroy(struct draw_stage *stage) +{ + free(stage); +} + struct draw_stage * nv50_draw_render_stage(struct nv50_context *nv50) { - struct nv50_draw_stage *nv50draw = CALLOC_STRUCT(nv50_draw_stage); + struct nv50_render_stage *rs = CALLOC_STRUCT(nv50_render_stage); - nv50draw->nv50 = nv50; - nv50draw->draw.draw = nv50->draw; - nv50draw->draw.point = nv50_draw_point; - nv50draw->draw.line = nv50_draw_line; - nv50draw->draw.tri = nv50_draw_tri; - nv50draw->draw.flush = nv50_draw_flush; - nv50draw->draw.reset_stipple_counter = nv50_draw_reset_stipple_counter; + rs->nv50 = nv50; + rs->stage.draw = nv50->draw; + rs->stage.destroy = nv50_render_destroy; + rs->stage.point = nv50_render_point; + rs->stage.line = nv50_render_line; + rs->stage.tri = nv50_render_tri; + rs->stage.flush = nv50_render_flush; + rs->stage.reset_stipple_counter = nv50_render_reset_stipple_counter; - return &nv50draw->draw; + return &rs->stage; } diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index f091779e3b0..77ceb678f21 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -77,6 +77,8 @@ nv50_screen_get_paramf(struct pipe_screen *pscreen, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index 2e54729aa97..8dac08a5d21 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -72,7 +72,7 @@ nouveau_channel_context_create(struct nouveau_device *nvdev, unsigned chipset) return NULL; } - switch (chipset) { + switch (chipset & 0xf0) { case 0x50: case 0x80: ret = nouveau_surface_channel_create_nv50(nvc); -- cgit v1.2.3 From b2e48f848496d5e315e536688c8c33dfb1fab7eb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 02:39:13 +1100 Subject: nv50: convert to hwctx-in-screen as nv40 is --- src/gallium/drivers/nv50/nv50_context.c | 61 ++++----------------------------- src/gallium/drivers/nv50/nv50_context.h | 11 +++--- src/gallium/drivers/nv50/nv50_screen.c | 49 +++++++++++++++++++++++++- src/gallium/drivers/nv50/nv50_screen.h | 5 +++ src/gallium/drivers/nv50/nv50_surface.c | 4 +-- 5 files changed, 67 insertions(+), 63 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index c937b8de6d4..980d066c844 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -10,10 +10,11 @@ static void nv50_flush(struct pipe_context *pipe, unsigned flags) { struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; + struct nv50_screen *screen = nv50->screen; + struct nouveau_winsys *nvws = screen->nvws; if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv50->sync, 0); + nvws->notifier_reset(screen->sync, 0); BEGIN_RING(tesla, 0x104, 1); OUT_RING (0); BEGIN_RING(tesla, 0x100, 1); @@ -23,7 +24,7 @@ nv50_flush(struct pipe_context *pipe, unsigned flags) FIRE_RING(); if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv50->sync, 0, 0, 2000); + nvws->notifier_wait(screen->sync, 0, 0, 2000); } static void @@ -35,66 +36,18 @@ nv50_destroy(struct pipe_context *pipe) free(nv50); } -static boolean -nv50_init_hwctx(struct nv50_context *nv50, int tesla_class) -{ - struct nouveau_winsys *nvws = nv50->nvws; - int ret; - - if ((ret = nvws->grobj_alloc(nvws, tesla_class, &nv50->tesla))) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - BEGIN_RING(tesla, NV50TCL_DMA_NOTIFY, 1); - OUT_RING (nv50->sync->handle); - - FIRE_RING (); - return TRUE; -} - -#define GRCLASS5097_CHIPSETS 0x00000000 -#define GRCLASS8297_CHIPSETS 0x00000010 struct pipe_context * nv50_create(struct pipe_screen *pscreen, unsigned pctx_id) { struct pipe_winsys *pipe_winsys = pscreen->winsys; - struct nouveau_winsys *nvws = nv50_screen(pscreen)->nvws; - unsigned chipset = nv50_screen(pscreen)->chipset; + struct nv50_screen *screen = nv50_screen(pscreen); struct nv50_context *nv50; - int tesla_class, ret; - - if ((chipset & 0xf0) != 0x50 && (chipset & 0xf0) != 0x80) { - NOUVEAU_ERR("Not a G8x chipset\n"); - return NULL; - } - - if (GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) { - tesla_class = 0x5097; - } else - if (GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) { - tesla_class = 0x8297; - } else { - NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset); - return NULL; - } nv50 = CALLOC_STRUCT(nv50_context); if (!nv50) return NULL; - nv50->chipset = chipset; - nv50->nvws = nvws; - - if ((ret = nvws->notifier_alloc(nvws, 1, &nv50->sync))) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - free(nv50); - return NULL; - } - - if (!nv50_init_hwctx(nv50, tesla_class)) { - free(nv50); - return NULL; - } + nv50->screen = screen; + nv50->pctx_id = pctx_id; nv50->pipe.winsys = pipe_winsys; nv50->pipe.screen = pscreen; diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index a529bf3c3e3..e2656bd539b 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -11,10 +11,11 @@ #include "nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv50_context *ctx = nv50 + struct nv50_screen *ctx = nv50->screen #include "nouveau/nouveau_push.h" #include "nv50_state.h" +#include "nv50_screen.h" #define NOUVEAU_ERR(fmt, args...) \ fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); @@ -23,13 +24,11 @@ struct nv50_context { struct pipe_context pipe; - struct nouveau_winsys *nvws; - struct draw_context *draw; + struct nv50_screen *screen; + unsigned pctx_id; - int chipset; - struct nouveau_grobj *tesla; - struct nouveau_notifier *sync; + struct draw_context *draw; }; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 77ceb678f21..f8fd11dc5ff 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -4,6 +4,11 @@ #include "nv50_context.h" #include "nv50_screen.h" +#include "nouveau/nouveau_stateobj.h" + +#define GRCLASS5097_CHIPSETS 0x00000000 +#define GRCLASS8297_CHIPSETS 0x00000010 + static boolean nv50_screen_is_format_supported(struct pipe_screen *pscreen, enum pipe_format format, uint type) @@ -96,13 +101,55 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, unsigned chipset) { struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen); + struct nouveau_stateobj *so; + unsigned tesla_class = 0, ret; if (!screen) return NULL; - screen->chipset = chipset; screen->nvws = nvws; + /* 3D object */ + if ((chipset & 0xf0) != 0x50 && (chipset & 0xf0) != 0x80) { + NOUVEAU_ERR("Not a G8x chipset\n"); + nv50_screen_destroy(&screen->pipe); + return NULL; + } + + if (GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) { + tesla_class = 0x5097; + } else + if (GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) { + tesla_class = 0x8297; + } else { + NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset); + nv50_screen_destroy(&screen->pipe); + return NULL; + } + + ret = nvws->grobj_alloc(nvws, tesla_class, &screen->tesla); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + nv50_screen_destroy(&screen->pipe); + return NULL; + } + + /* Sync notifier */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv50_screen_destroy(&screen->pipe); + return NULL; + } + + /* Static tesla init */ + so = so_new(128, 0); + so_method(so, screen->tesla, NV50TCL_DMA_NOTIFY, 1); + so_data (so, screen->sync->handle); + so_emit(nvws, so); + so_ref(NULL, &so); + nvws->push_flush(nvws->channel, 0); + screen->pipe.winsys = ws; screen->pipe.destroy = nv50_screen_destroy; diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index d664816a031..6e4120d6693 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -8,6 +8,11 @@ struct nv50_screen { struct nouveau_winsys *nvws; unsigned chipset; + + unsigned cur_pctx; + + struct nouveau_grobj *tesla; + struct nouveau_notifier *sync; }; static INLINE struct nv50_screen * diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index 39cf675a57c..4e294198b07 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -40,7 +40,7 @@ nv50_surface_copy(struct pipe_context *pipe, unsigned flip, unsigned width, unsigned height) { struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; + struct nouveau_winsys *nvws = nv50->screen->nvws; nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, width, height); @@ -52,7 +52,7 @@ nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, unsigned height, unsigned value) { struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nouveau_winsys *nvws = nv50->nvws; + struct nouveau_winsys *nvws = nv50->screen->nvws; nvws->surface_fill(nvws, dest, destx, desty, width, height, value); } -- cgit v1.2.3 From 3250bacd2411d3f1af50135599380b2140238535 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 02:56:10 +1100 Subject: nv50: create blend stateobj --- src/gallium/drivers/nv50/nv50_context.h | 15 ++++++++ src/gallium/drivers/nv50/nv50_screen.c | 10 ++++++ src/gallium/drivers/nv50/nv50_state.c | 62 ++++++++++++++++++++++++++++++++- 3 files changed, 86 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index e2656bd539b..6096818d40d 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -22,6 +22,13 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); +#define NV50_NEW_BLEND (1 << 0) + +struct nv50_blend_stateobj { + struct pipe_blend_state pipe; + struct nouveau_stateobj *so; +}; + struct nv50_context { struct pipe_context pipe; @@ -29,8 +36,16 @@ struct nv50_context { unsigned pctx_id; struct draw_context *draw; + + unsigned dirty; + struct nv50_blend_stateobj *blend; }; +static INLINE struct nv50_context * +nv50_context(struct pipe_context *pipe) +{ + return (struct nv50_context *)pipe; +} extern void nv50_init_miptree_functions(struct nv50_context *nv50); extern void nv50_init_surface_functions(struct nv50_context *nv50); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index f8fd11dc5ff..62c23c790ca 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -103,6 +103,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen); struct nouveau_stateobj *so; unsigned tesla_class = 0, ret; + int i; if (!screen) return NULL; @@ -146,6 +147,15 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, so = so_new(128, 0); so_method(so, screen->tesla, NV50TCL_DMA_NOTIFY, 1); so_data (so, screen->sync->handle); + so_method(so, screen->tesla, NV50TCL_DMA_IN_MEMORY0(0), + NV50TCL_DMA_IN_MEMORY0__SIZE); + for (i = 0; i < NV50TCL_DMA_IN_MEMORY0__SIZE; i++) + so_data(so, nvws->channel->vram->handle); + so_method(so, screen->tesla, NV50TCL_DMA_IN_MEMORY1(0), + NV50TCL_DMA_IN_MEMORY1__SIZE); + for (i = 0; i < NV50TCL_DMA_IN_MEMORY1__SIZE; i++) + so_data(so, nvws->channel->vram->handle); + so_emit(nvws, so); so_ref(NULL, &so); nvws->push_flush(nvws->channel, 0); diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 99dcab51b26..7a01100fd74 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -5,21 +5,81 @@ #include "nv50_context.h" #include "nv50_state.h" +#include "nouveau/nouveau_stateobj.h" + static void * nv50_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { - return NULL; + struct nouveau_stateobj *so = so_new(64, 0); + struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla; + struct nv50_blend_stateobj *bso = CALLOC_STRUCT(nv50_blend_stateobj); + unsigned cmask = 0, i; + + /*XXX ignored: + * - dither + */ + + if (cso->blend_enable == 0) { + so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8); + for (i = 0; i < 8; i++) + so_data(so, 0); + } else { + so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8); + for (i = 0; i < 8; i++) + so_data(so, 1); + so_method(so, tesla, NV50TCL_BLEND_EQUATION_RGB, 5); + so_data (so, nvgl_blend_eqn(cso->rgb_func)); + so_data (so, nvgl_blend_func(cso->rgb_src_factor)); + so_data (so, nvgl_blend_func(cso->rgb_dst_factor)); + so_data (so, nvgl_blend_eqn(cso->alpha_func)); + so_data (so, nvgl_blend_func(cso->alpha_src_factor)); + so_method(so, tesla, NV50TCL_BLEND_FUNC_DST_ALPHA, 1); + so_data (so, nvgl_blend_func(cso->alpha_dst_factor)); + } + + if (cso->logicop_enable == 0 ) { + so_method(so, tesla, NV50TCL_LOGIC_OP_ENABLE, 1); + so_data (so, 0); + } else { + so_method(so, tesla, NV50TCL_LOGIC_OP_ENABLE, 2); + so_data (so, 1); + so_data (so, nvgl_logicop_func(cso->logicop_func)); + } + + if (cso->colormask & PIPE_MASK_R) + cmask |= (1 << 0); + if (cso->colormask & PIPE_MASK_G) + cmask |= (1 << 4); + if (cso->colormask & PIPE_MASK_B) + cmask |= (1 << 8); + if (cso->colormask & PIPE_MASK_A) + cmask |= (1 << 12); + so_method(so, tesla, NV50TCL_COLOR_MASK(0), 8); + for (i = 0; i < 8; i++) + so_data(so, cmask); + + bso->pipe = *cso; + so_ref(so, &bso->so); + return (void *)bso; } static void nv50_blend_state_bind(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->blend = hwcso; + nv50->dirty |= NV50_NEW_BLEND; } static void nv50_blend_state_delete(struct pipe_context *pipe, void *hwcso) { + struct nv50_blend_stateobj *bso = hwcso; + + so_ref(NULL, &bso->so); + FREE(bso); } static void * -- cgit v1.2.3 From 9c29512154992f95c11939615ddcbef185c6a96c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 02:59:20 +1100 Subject: nv50: crappy state validate/emit function Just for testing stateobjs to make sure they don't hang the engine. --- src/gallium/drivers/nv50/Makefile | 1 + src/gallium/drivers/nv50/nv50_context.h | 2 ++ src/gallium/drivers/nv50/nv50_state_validate.c | 14 ++++++++++++++ src/gallium/drivers/nv50/nv50_vbo.c | 8 ++++++++ 4 files changed, 25 insertions(+) create mode 100644 src/gallium/drivers/nv50/nv50_state_validate.c diff --git a/src/gallium/drivers/nv50/Makefile b/src/gallium/drivers/nv50/Makefile index 1c0b82887a1..a62a4d961bc 100644 --- a/src/gallium/drivers/nv50/Makefile +++ b/src/gallium/drivers/nv50/Makefile @@ -11,6 +11,7 @@ DRIVER_SOURCES = \ nv50_query.c \ nv50_screen.c \ nv50_state.c \ + nv50_state_validate.c \ nv50_surface.c \ nv50_vbo.c diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 6096818d40d..98b9aba0796 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -70,4 +70,6 @@ extern boolean nv50_draw_elements(struct pipe_context *pipe, extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue); +extern boolean nv50_state_validate(struct nv50_context *nv50); + #endif diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c new file mode 100644 index 00000000000..a89d1526c58 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -0,0 +1,14 @@ +#include "nv50_context.h" +#include "nouveau/nouveau_stateobj.h" + +boolean +nv50_state_validate(struct nv50_context *nv50) +{ + struct nouveau_winsys *nvws = nv50->screen->nvws; + + if (nv50->dirty & NV50_NEW_BLEND) + so_emit(nvws, nv50->blend->so); + + return TRUE; +} + diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 6c0dc23a439..b01ce1d42c5 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -9,6 +9,10 @@ boolean nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50_state_validate(nv50); + NOUVEAU_ERR("unimplemented\n"); return TRUE; } @@ -18,6 +22,10 @@ nv50_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50_state_validate(nv50); + NOUVEAU_ERR("unimplemented\n"); return TRUE; } -- cgit v1.2.3 From 9911ca2226cf87e2cc4ffc32e40bccaf0a4e5745 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 03:21:48 +1100 Subject: nv50: ouch, next_subchannel++ :) --- src/gallium/winsys/dri/nouveau/nv50_surface.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c index 5d74fb8d2bc..5c6733e44a3 100644 --- a/src/gallium/winsys/dri/nouveau/nv50_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv50_surface.c @@ -155,7 +155,7 @@ nouveau_surface_channel_create_nv50(struct nouveau_channel_context *nvc) &nvc->Nv2D); if (ret) return ret; - BIND_RING (nvc->channel, nvc->Nv2D, 0); + BIND_RING (nvc->channel, nvc->Nv2D, nvc->next_subchannel++); BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_DMA_NOTIFY, 1); OUT_RING (nvc->channel, nvc->sync_notifier->handle); BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); -- cgit v1.2.3 From 06bd7d78b979df66915b161157f2b6b1c09ad285 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 03:41:05 +1100 Subject: nv50: depth_stencil_alpha stateobj --- src/gallium/drivers/nv50/nv50_context.h | 9 +++- src/gallium/drivers/nv50/nv50_state.c | 70 +++++++++++++++++++++++++- src/gallium/drivers/nv50/nv50_state_validate.c | 3 ++ 3 files changed, 80 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 98b9aba0796..93027648940 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -22,13 +22,19 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV50_NEW_BLEND (1 << 0) +#define NV50_NEW_BLEND (1 << 0) +#define NV50_NEW_ZSA (1 << 1) struct nv50_blend_stateobj { struct pipe_blend_state pipe; struct nouveau_stateobj *so; }; +struct nv50_zsa_stateobj { + struct pipe_depth_stencil_alpha_state pipe; + struct nouveau_stateobj *so; +}; + struct nv50_context { struct pipe_context pipe; @@ -39,6 +45,7 @@ struct nv50_context { unsigned dirty; struct nv50_blend_stateobj *blend; + struct nv50_zsa_stateobj *zsa; }; static INLINE struct nv50_context * diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 7a01100fd74..8fd4f774e27 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -127,17 +127,85 @@ static void * nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { - return NULL; + struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla; + struct nv50_zsa_stateobj *zsa = CALLOC_STRUCT(nv50_zsa_stateobj); + struct nouveau_stateobj *so = so_new(64, 0); + + so_method(so, tesla, NV50TCL_DEPTH_WRITE_ENABLE, 1); + so_data (so, cso->depth.writemask ? 1 : 0); + if (cso->depth.enabled) { + so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1); + so_data (so, 1); + so_method(so, tesla, NV50TCL_DEPTH_TEST_FUNC, 1); + so_data (so, nvgl_comparison_op(cso->depth.func)); + } else { + so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1); + so_data (so, 0); + } + + if (cso->stencil[0].enabled) { + so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 5); + so_data (so, 1); + so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); + so_data (so, nvgl_comparison_op(cso->stencil[0].func)); + so_method(so, tesla, NV50TCL_STENCIL_FRONT_FUNC_REF, 3); + so_data (so, cso->stencil[0].ref_value); + so_data (so, cso->stencil[0].write_mask); + so_data (so, cso->stencil[0].value_mask); + } else { + so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 1); + so_data (so, 0); + } + + if (cso->stencil[1].enabled) { + so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 8); + so_data (so, 1); + so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); + so_data (so, nvgl_comparison_op(cso->stencil[1].func)); + so_data (so, cso->stencil[1].ref_value); + so_data (so, cso->stencil[1].write_mask); + so_data (so, cso->stencil[1].value_mask); + } else { + so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 1); + so_data (so, 0); + } + + if (cso->alpha.enabled) { + so_method(so, tesla, NV50TCL_ALPHA_TEST_ENABLE, 1); + so_data (so, 1); + so_method(so, tesla, NV50TCL_ALPHA_TEST_REF, 2); + so_data (so, fui(cso->alpha.ref)); + so_data (so, nvgl_comparison_op(cso->alpha.func)); + } else { + so_method(so, tesla, NV50TCL_ALPHA_TEST_ENABLE, 1); + so_data (so, 0); + } + + zsa->pipe = *cso; + so_ref(so, &zsa->so); + return (void *)zsa; } static void nv50_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->zsa = hwcso; + nv50->dirty |= NV50_NEW_ZSA; } static void nv50_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { + struct nv50_zsa_stateobj *zsa = hwcso; + + so_ref(NULL, &zsa->so); + FREE(zsa); } static void * diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index a89d1526c58..786987c76ad 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -8,7 +8,10 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & NV50_NEW_BLEND) so_emit(nvws, nv50->blend->so); + if (nv50->dirty & NV50_NEW_ZSA) + so_emit(nvws, nv50->zsa->so); + nv50->dirty = 0; return TRUE; } -- cgit v1.2.3 From 2fee5f76483feb301546b24c26eea699732ffb57 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 03:54:53 +1100 Subject: nv50: scissor/viewport/blend colour/stipple --- src/gallium/drivers/nv50/nv50_context.h | 12 +++++-- src/gallium/drivers/nv50/nv50_state.c | 16 +++++++++ src/gallium/drivers/nv50/nv50_state_validate.c | 49 ++++++++++++++++++++++++++ 3 files changed, 75 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 93027648940..406ad2ca634 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -22,8 +22,12 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV50_NEW_BLEND (1 << 0) -#define NV50_NEW_ZSA (1 << 1) +#define NV50_NEW_BLEND (1 << 0) +#define NV50_NEW_ZSA (1 << 1) +#define NV50_NEW_BLEND_COLOUR (1 << 2) +#define NV50_NEW_STIPPLE (1 << 3) +#define NV50_NEW_SCISSOR (1 << 4) +#define NV50_NEW_VIEWPORT (1 << 5) struct nv50_blend_stateobj { struct pipe_blend_state pipe; @@ -46,6 +50,10 @@ struct nv50_context { unsigned dirty; struct nv50_blend_stateobj *blend; struct nv50_zsa_stateobj *zsa; + struct pipe_blend_color blend_colour; + struct pipe_poly_stipple stipple; + struct pipe_scissor_state scissor; + struct pipe_viewport_state viewport; }; static INLINE struct nv50_context * diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 8fd4f774e27..40e23f46094 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -246,6 +246,10 @@ static void nv50_set_blend_color(struct pipe_context *pipe, const struct pipe_blend_color *bcol) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->blend_colour = *bcol; + nv50->dirty |= NV50_NEW_BLEND_COLOUR; } static void @@ -270,18 +274,30 @@ static void nv50_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->stipple = *stipple; + nv50->dirty |= NV50_NEW_STIPPLE; } static void nv50_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->scissor = *s; + nv50->dirty |= NV50_NEW_SCISSOR; } static void nv50_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->viewport = *vpt; + nv50->dirty |= NV50_NEW_VIEWPORT; } static void diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 786987c76ad..8920b6f49b0 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -5,12 +5,61 @@ boolean nv50_state_validate(struct nv50_context *nv50) { struct nouveau_winsys *nvws = nv50->screen->nvws; + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nouveau_stateobj *so; + unsigned i; if (nv50->dirty & NV50_NEW_BLEND) so_emit(nvws, nv50->blend->so); + if (nv50->dirty & NV50_NEW_ZSA) so_emit(nvws, nv50->zsa->so); + if (nv50->dirty & NV50_NEW_BLEND_COLOUR) { + so = so_new(5, 0); + so_method(so, tesla, NV50TCL_BLEND_COLOR(0), 8); + so_data (so, fui(nv50->blend_colour.color[3])); + so_data (so, fui(nv50->blend_colour.color[0])); + so_data (so, fui(nv50->blend_colour.color[1])); + so_data (so, fui(nv50->blend_colour.color[2])); + so_emit(nvws, so); + so_ref(NULL, &so); + } + + if (nv50->dirty & NV50_NEW_STIPPLE) { + so = so_new(33, 0); + so_method(so, tesla, NV50TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, nv50->stipple.stipple[i]); + so_emit(nvws, so); + so_ref(NULL, &so); + } + + if (nv50->dirty & NV50_NEW_SCISSOR) { + so = so_new(3, 0); + so_method(so, tesla, NV50TCL_SCISSOR_HORIZ, 2); + so_data (so, (nv50->scissor.maxx << 16) | + nv50->scissor.minx); + so_data (so, (nv50->scissor.maxy << 16) | + nv50->scissor.miny); + so_emit(nvws, so); + so_ref(NULL, &so); + } + + if (nv50->dirty & NV50_NEW_VIEWPORT) { + so = so_new(8, 0); + so_method(so, tesla, NV50TCL_VIEWPORT_UNK0(0), 3); + so_data (so, fui(nv50->viewport.translate[0])); + so_data (so, fui(nv50->viewport.translate[1])); + so_data (so, fui(nv50->viewport.translate[2])); + so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3); + so_data (so, fui(nv50->viewport.scale[0])); + so_data (so, fui(nv50->viewport.scale[1])); + so_data (so, fui(nv50->viewport.scale[2])); + so_emit(nvws, so); + so_ref(NULL, &so); + } + nv50->dirty = 0; return TRUE; } -- cgit v1.2.3 From cd85dc1e5dfa37cb9bee696e5e18332e3f1d65a1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 04:29:58 +1100 Subject: nv50: rasterizer stateobj --- src/gallium/drivers/nv50/nv50_context.h | 7 ++ src/gallium/drivers/nv50/nv50_state.c | 120 ++++++++++++++++++++++++- src/gallium/drivers/nv50/nv50_state_validate.c | 3 + 3 files changed, 129 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 406ad2ca634..d839cf701eb 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -28,6 +28,7 @@ #define NV50_NEW_STIPPLE (1 << 3) #define NV50_NEW_SCISSOR (1 << 4) #define NV50_NEW_VIEWPORT (1 << 5) +#define NV50_NEW_RASTERIZER (1 << 6) struct nv50_blend_stateobj { struct pipe_blend_state pipe; @@ -39,6 +40,11 @@ struct nv50_zsa_stateobj { struct nouveau_stateobj *so; }; +struct nv50_rasterizer_stateobj { + struct pipe_rasterizer_state pipe; + struct nouveau_stateobj *so; +}; + struct nv50_context { struct pipe_context pipe; @@ -50,6 +56,7 @@ struct nv50_context { unsigned dirty; struct nv50_blend_stateobj *blend; struct nv50_zsa_stateobj *zsa; + struct nv50_rasterizer_stateobj *rasterizer; struct pipe_blend_color blend_colour; struct pipe_poly_stipple stipple; struct pipe_scissor_state scissor; diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 40e23f46094..088df9e6756 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -110,17 +110,135 @@ static void * nv50_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { - return NULL; + struct nouveau_stateobj *so = so_new(64, 0); + struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla; + struct nv50_rasterizer_stateobj *rso = + CALLOC_STRUCT(nv50_rasterizer_stateobj); + unsigned i; + + /*XXX: ignored + * - light_twosize + * - point_smooth + * - multisample + * - point_sprite / sprite_coord_mode + */ + + so_method(so, tesla, NV50TCL_SHADE_MODEL, 1); + so_data (so, cso->flatshade ? NV50TCL_SHADE_MODEL_FLAT : + NV50TCL_SHADE_MODEL_SMOOTH); + + so_method(so, tesla, NV50TCL_LINE_WIDTH, 1); + so_data (so, fui(cso->line_width)); + so_method(so, tesla, NV50TCL_LINE_SMOOTH_ENABLE, 1); + so_data (so, cso->line_smooth ? 1 : 0); + if (cso->line_stipple_enable) { + so_method(so, tesla, NV50TCL_LINE_STIPPLE_ENABLE, 1); + so_data (so, 1); + so_method(so, tesla, NV50TCL_LINE_STIPPLE_PATTERN, 1); + so_data (so, (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor); + } else { + so_method(so, tesla, NV50TCL_LINE_STIPPLE_ENABLE, 1); + so_data (so, 0); + } + + so_method(so, tesla, NV50TCL_POINT_SIZE, 1); + so_data (so, fui(cso->point_size)); + + so_method(so, tesla, NV50TCL_POLYGON_MODE_FRONT, 3); + if (cso->front_winding == PIPE_WINDING_CCW) { + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + } else { + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + } + so_data(so, cso->poly_smooth ? 1 : 0); + + so_method(so, tesla, NV50TCL_CULL_FACE_ENABLE, 3); + so_data (so, cso->cull_mode != PIPE_WINDING_NONE); + if (cso->front_winding == PIPE_WINDING_CCW) { + so_data(so, NV50TCL_FRONT_FACE_CCW); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV50TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_CW: + so_data(so, NV50TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV50TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, NV50TCL_CULL_FACE_BACK); + break; + } + } else { + so_data(so, NV50TCL_FRONT_FACE_CW); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV50TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_CW: + so_data(so, NV50TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV50TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, NV50TCL_CULL_FACE_BACK); + break; + } + } + + so_method(so, tesla, NV50TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, cso->poly_stipple_enable ? 1 : 0); + + so_method(so, tesla, NV50TCL_POLYGON_OFFSET_POINT_ENABLE, 3); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT)) + so_data(so, 1); + else + so_data(so, 0); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_LINE) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_LINE)) + so_data(so, 1); + else + so_data(so, 0); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_FILL) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_FILL)) + so_data(so, 1); + else + so_data(so, 0); + + if (cso->offset_cw || cso->offset_ccw) { + so_method(so, tesla, NV50TCL_POLYGON_OFFSET_FACTOR, 1); + so_data (so, fui(cso->offset_scale)); + so_method(so, tesla, NV50TCL_POLYGON_OFFSET_UNITS, 1); + so_data (so, fui(cso->offset_units * 2)); + } + + rso->pipe = *cso; + so_ref(so, &rso->so); + return (void *)rso; } static void nv50_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->rasterizer = hwcso; + nv50->dirty |= NV50_NEW_RASTERIZER; } static void nv50_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { + struct nv50_rasterizer_stateobj *rso = hwcso; + + so_ref(NULL, &rso->so); + FREE(rso); } static void * diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 8920b6f49b0..8921bfaf887 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -15,6 +15,9 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & NV50_NEW_ZSA) so_emit(nvws, nv50->zsa->so); + if (nv50->dirty & NV50_NEW_RASTERIZER) + so_emit(nvws, nv50->rasterizer->so); + if (nv50->dirty & NV50_NEW_BLEND_COLOUR) { so = so_new(5, 0); so_method(so, tesla, NV50TCL_BLEND_COLOR(0), 8); -- cgit v1.2.3 From cd9ed05aec9d1d9614973165fd13647ba2e1b8c7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 12 Mar 2008 04:50:53 +1100 Subject: nv50: start on fb state --- src/gallium/drivers/nv50/nv50_context.h | 2 + src/gallium/drivers/nv50/nv50_state.c | 4 ++ src/gallium/drivers/nv50/nv50_state_validate.c | 93 ++++++++++++++++++++++++++ 3 files changed, 99 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index d839cf701eb..f532fa6bfb3 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -29,6 +29,7 @@ #define NV50_NEW_SCISSOR (1 << 4) #define NV50_NEW_VIEWPORT (1 << 5) #define NV50_NEW_RASTERIZER (1 << 6) +#define NV50_NEW_FRAMEBUFFER (1 << 7) struct nv50_blend_stateobj { struct pipe_blend_state pipe; @@ -61,6 +62,7 @@ struct nv50_context { struct pipe_poly_stipple stipple; struct pipe_scissor_state scissor; struct pipe_viewport_state viewport; + struct pipe_framebuffer_state framebuffer; }; static INLINE struct nv50_context * diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 088df9e6756..aa65fd482e4 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -386,6 +386,10 @@ static void nv50_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->framebuffer = *fb; + nv50->dirty |= NV50_NEW_FRAMEBUFFER; } static void diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 8921bfaf887..68d419f9fc8 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -1,6 +1,96 @@ #include "nv50_context.h" #include "nouveau/nouveau_stateobj.h" +static void +nv50_state_validate_fb(struct nv50_context *nv50) +{ + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nouveau_stateobj *so = so_new(128, 18); + struct pipe_framebuffer_state *fb = &nv50->framebuffer; + unsigned i, w, h, gw = 0; + + for (i = 0; i < fb->num_cbufs; i++) { + if (!gw) { + w = fb->cbufs[i]->width; + h = fb->cbufs[i]->height; + gw = 1; + } else { + assert(w != fb->cbufs[i]->width); + assert(h != fb->cbufs[i]->height); + } + + so_method(so, tesla, NV50TCL_RT_HORIZ(i), 2); + so_data (so, fb->cbufs[i]->width); + so_data (so, fb->cbufs[i]->height); + + so_method(so, tesla, NV50TCL_RT_ADDRESS_HIGH(i), 5); + so_reloc (so, fb->cbufs[i]->buffer, fb->cbufs[i]->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, fb->cbufs[i]->buffer, fb->cbufs[i]->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0); + switch (fb->cbufs[i]->format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + so_data(so, 0xcf); + break; + case PIPE_FORMAT_R5G6B5_UNORM: + so_data(so, 0xe8); + break; + default: + { + char fmt[128]; + pf_sprint_name(fmt, fb->cbufs[i]->format); + NOUVEAU_ERR("AIIII unknown format %s\n", fmt); + } + so_data(so, 0xe6); + break; + } + so_data(so, 0x00000040); + so_data(so, 0x00000000); + } + + if (fb->zsbuf) { + if (!gw) { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + gw = 1; + } else { + assert(w != fb->zsbuf->width); + assert(h != fb->zsbuf->height); + } + + so_method(so, tesla, NV50TCL_RT_ADDRESS_HIGH(i), 5); + so_reloc (so, fb->zsbuf->buffer, fb->zsbuf->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, fb->zsbuf->buffer, fb->zsbuf->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0); + switch (fb->zsbuf->format) { + case PIPE_FORMAT_Z24S8_UNORM: + so_data(so, 0x16); + break; + default: + { + char fmt[128]; + pf_sprint_name(fmt, fb->zsbuf->format); + NOUVEAU_ERR("AIIII unknown format %s\n", fmt); + } + so_data(so, 0x16); + break; + } + so_data(so, 0x00000040); + so_data(so, 0x00000000); + } + + so_method(so, tesla, NV50TCL_VIEWPORT_HORIZ, 2); + so_data (so, w << 16); + so_data (so, h << 16); + so_method(so, tesla, 0xff0, 2); + so_data (so, w << 16); + so_data (so, h << 16); + + so_emit(nv50->screen->nvws, so); + so_ref(NULL, &so); +} + boolean nv50_state_validate(struct nv50_context *nv50) { @@ -9,6 +99,9 @@ nv50_state_validate(struct nv50_context *nv50) struct nouveau_stateobj *so; unsigned i; + if (nv50->dirty & NV50_NEW_FRAMEBUFFER) + nv50_state_validate_fb(nv50); + if (nv50->dirty & NV50_NEW_BLEND) so_emit(nvws, nv50->blend->so); -- cgit v1.2.3 From 169912b71a4242389301890ef303046d49ce71df Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Tue, 11 Mar 2008 19:22:02 +0100 Subject: nv30: silence some warnings --- src/gallium/drivers/nv30/nv30_fragprog.c | 2 +- src/gallium/drivers/nv30/nv30_state.c | 2 +- src/gallium/drivers/nv30/nv30_vbo.c | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 09ad555c324..0a2ba04f958 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -378,7 +378,7 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, { const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); struct nv30_sreg src[3], dst, tmp; - int mask, sat, unit; + int mask, sat, unit = 0; int ai = -1, ci = -1; int i; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index aa3fe7837e4..319d53fccac 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -538,7 +538,7 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); struct pipe_surface *rt[4], *zeta = NULL; - uint32_t rt_enable, rt_format, w, h; + uint32_t rt_enable, rt_format, w = 0, h = 0; int i, colour_format = 0, zeta_format = 0; rt_enable = 0; diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index 9e00cdac3fb..a62462f7bc5 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -102,7 +102,7 @@ nv30_vbo_arrays_update(struct nv30_context *nv30) { struct nv30_vertex_program *vp = nv30->vertprog.active; uint32_t inputs, vtxfmt[16]; - int hw, num_hw; + int hw, num_hw = 0; nv30->vb_enable = 0; -- cgit v1.2.3 From 6bd5e5ce00b1870a8d94337cc10faa8134cbefd5 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Wed, 12 Mar 2008 18:40:37 +0100 Subject: nv30: line up the miptree creation to latest changes. --- src/gallium/drivers/nv30/nv30_miptree.c | 30 ++++++++++++++++-------------- 1 file changed, 16 insertions(+), 14 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 23bcef08ebc..19945e9ab87 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -54,27 +54,29 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) nv30mt->total_size = offset; } -static void -nv30_miptree_create(struct pipe_screen *screen, struct pipe_texture **pt) +static struct pipe_texture * +nv30_miptree_create(struct pipe_screen *screen, struct pipe_texture *pt) { struct pipe_winsys *ws = screen->winsys; - struct nv30_miptree *nv30mt; + struct nv30_miptree *mt; - nv30mt = realloc(*pt, sizeof(struct nv30_miptree)); - if (!nv30mt) - return; - *pt = NULL; + mt = MALLOC(sizeof(struct nv30_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + mt->base.refcount = 1; + mt->base.screen = screen; - nv30_miptree_layout(nv30mt); + nv30_miptree_layout(mt); - nv30mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - nv30mt->total_size); - if (!nv30mt->buffer) { - free(nv30mt); - return; + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + free(mt); + return NULL; } - *pt = &nv30mt->base; + return &mt->base; } static void -- cgit v1.2.3 From 830b4709f0ac27915450b53b622a8886264d8c8c Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Wed, 12 Mar 2008 18:43:29 +0100 Subject: nouveau: update to latest reg header. --- src/gallium/drivers/nouveau/nouveau_class.h | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index c80461038b7..61c3d97fd4c 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -1747,7 +1747,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_DMA_NOTIFY 0x00000180 #define NV10TCL_DMA_IN_MEMORY0 0x00000184 #define NV10TCL_DMA_IN_MEMORY1 0x00000188 -#define NV10TCL_DISPLAY_LIST 0x0000018c +#define NV10TCL_DMA_VTXBUF0 0x0000018c #define NV10TCL_DMA_IN_MEMORY2 0x00000194 #define NV10TCL_DMA_IN_MEMORY3 0x00000198 #define NV10TCL_VIEWPORT_HORIZ 0x00000200 @@ -2866,11 +2866,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_VERTEX_BEGIN_END_QUADS 0x00000008 #define NV10TCL_VERTEX_BEGIN_END_QUAD_STRIP 0x00000009 #define NV10TCL_VERTEX_BEGIN_END_POLYGON 0x0000000a -#define NV10TCL_DRAW_INDEX 0x00000e00 -#define NV10TCL_DRAW_INDEX_I0_SHIFT 0 -#define NV10TCL_DRAW_INDEX_I0_MASK 0x0000ffff -#define NV10TCL_DRAW_INDEX_I1_SHIFT 24 -#define NV10TCL_DRAW_INDEX_I1_MASK 0xff000000 +#define NV10TCL_VB_ELEMENT_U16 0x00000e00 +#define NV10TCL_VB_ELEMENT_U16_I0_SHIFT 0 +#define NV10TCL_VB_ELEMENT_U16_I0_MASK 0x0000ffff +#define NV10TCL_VB_ELEMENT_U16_I1_SHIFT 24 +#define NV10TCL_VB_ELEMENT_U16_I1_MASK 0xff000000 #define NV10TCL_VERTEX_BUFFER_BEGIN_END 0x000013fc #define NV10TCL_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 #define NV10TCL_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 @@ -4099,9 +4099,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VIEWPORT_SCALE_Y 0x00000a34 #define NV34TCL_VIEWPORT_SCALE_Z 0x00000a38 #define NV34TCL_VIEWPORT_SCALE_W 0x00000a3c -#define NV34TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a60 +#define NV34TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a60 #define NV34TCL_POLYGON_OFFSET_LINE_ENABLE 0x00000a64 -#define NV34TCL_POLYGON_OFFSET_POINT_ENABLE 0x00000a68 +#define NV34TCL_POLYGON_OFFSET_FILL_ENABLE 0x00000a68 #define NV34TCL_DEPTH_FUNC 0x00000a6c #define NV34TCL_DEPTH_FUNC_NEVER 0x00000200 #define NV34TCL_DEPTH_FUNC_LESS 0x00000201 -- cgit v1.2.3 From 3b2a9b01a0c9b80573556a21e9db11b6f64eff8e Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Wed, 12 Mar 2008 19:06:22 +0100 Subject: nv30: debug dumps vp constants --- src/gallium/drivers/nv30/nv30_vertprog.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 96bc4b5ef9a..75f7351261d 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -751,6 +751,11 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); OUT_RING (i + vp->data->start); OUT_RINGp ((uint32_t *)vpd->value, 4); +#if 0 + NOUVEAU_MSG("VP const %d: %f %f %f %f\n", + i, vpd->value[0], vpd->value[1], + vpd->value[2], vpd->value[3]); +#endif } if (map) { @@ -762,10 +767,9 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) if (upload_code) { #if 0 for (i = 0; i < vp->nr_insns; i++) { - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]); - NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]); + NOUVEAU_MSG("VP inst %d: 0x%08x 0x%08x 0x%08x 0x%08x\n", + i, vp->insns[i].data[0], vp->insns[i].data[1], + vp->insns[i].data[2], vp->insns[i].data[3]); } #endif BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); -- cgit v1.2.3 From bd4fe0e87c1b979973d9a76aa48de5fbbb8d52b7 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Thu, 13 Mar 2008 00:42:50 +0100 Subject: nouveau: update to latest nouveau_class.h --- src/gallium/drivers/nouveau/nouveau_class.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 61c3d97fd4c..1b015507903 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -2869,8 +2869,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_VB_ELEMENT_U16 0x00000e00 #define NV10TCL_VB_ELEMENT_U16_I0_SHIFT 0 #define NV10TCL_VB_ELEMENT_U16_I0_MASK 0x0000ffff -#define NV10TCL_VB_ELEMENT_U16_I1_SHIFT 24 -#define NV10TCL_VB_ELEMENT_U16_I1_MASK 0xff000000 +#define NV10TCL_VB_ELEMENT_U16_I1_SHIFT 16 +#define NV10TCL_VB_ELEMENT_U16_I1_MASK 0xffff0000 +#define NV10TCL_VB_ELEMENT_U32 0x00001100 #define NV10TCL_VERTEX_BUFFER_BEGIN_END 0x000013fc #define NV10TCL_VERTEX_BUFFER_BEGIN_END_STOP 0x00000000 #define NV10TCL_VERTEX_BUFFER_BEGIN_END_POINTS 0x00000001 -- cgit v1.2.3 From 9a4938d7033101122b627786273ff37229b5558a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 13 Mar 2008 12:36:35 +1100 Subject: nouveau: match interface changes --- src/gallium/drivers/nv30/nv30_fragprog.c | 11 ++++---- src/gallium/drivers/nv30/nv30_state.c | 30 +++++++++++--------- src/gallium/drivers/nv30/nv30_state.h | 4 +-- src/gallium/drivers/nv30/nv30_vertprog.c | 4 +-- src/gallium/drivers/nv40/nv40_context.h | 2 ++ src/gallium/drivers/nv40/nv40_fragprog.c | 11 ++++---- src/gallium/drivers/nv40/nv40_state.c | 47 +++++++++++++++++++++++--------- src/gallium/drivers/nv40/nv40_state.h | 4 +-- src/gallium/drivers/nv40/nv40_vertprog.c | 4 +-- src/gallium/drivers/nv50/nv50_state.c | 11 ++++---- 10 files changed, 76 insertions(+), 52 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 0a2ba04f958..6f61d36f4e8 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -583,15 +583,14 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); break; case TGSI_OPCODE_TEX: - if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == - TGSI_EXTSWIZZLE_W) { - tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); - } else - tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); break; case TGSI_OPCODE_TXB: tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); break; + case TGSI_OPCODE_TXP: + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + break; case TGSI_OPCODE_XPD: tmp = temp(fpc); arith(fpc, 0, MUL, tmp, mask, @@ -683,7 +682,7 @@ nv30_fragprog_translate(struct nv30_context *nv30, fpc->high_temp = -1; fpc->num_regs = 2; - tgsi_parse_init(&parse, fp->pipe->tokens); + tgsi_parse_init(&parse, fp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&parse)) { tgsi_parse_token(&parse); diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 319d53fccac..951a32bc81e 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -231,14 +231,15 @@ nv30_sampler_state_create(struct pipe_context *pipe, } static void -nv30_sampler_state_bind(struct pipe_context *pipe, unsigned unit, - void *hwcso) +nv30_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) { struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_sampler_state *ps = hwcso; + unsigned unit; - nv30->tex_sampler[unit] = ps; - nv30->dirty_samplers |= (1 << unit); + for (unit = 0; unit < nr; unit++) { + nv30->tex_sampler[unit] = sampler[unit]; + nv30->dirty_samplers |= (1 << unit); + } } static void @@ -248,13 +249,16 @@ nv30_sampler_state_delete(struct pipe_context *pipe, void *hwcso) } static void -nv30_set_sampler_texture(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *miptree) +nv30_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **miptree) { struct nv30_context *nv30 = nv30_context(pipe); + unsigned unit; - nv30->tex_miptree[unit] = (struct nv30_miptree *)miptree; - nv30->dirty_samplers |= (1 << unit); + for (unit = 0; unit < nr; unit++) { + nv30->tex_miptree[unit] = (struct nv30_miptree *)miptree[unit]; + nv30->dirty_samplers |= (1 << unit); + } } static void * @@ -440,7 +444,7 @@ nv30_vp_state_create(struct pipe_context *pipe, struct nv30_vertex_program *vp; vp = CALLOC(1, sizeof(struct nv30_vertex_program)); - vp->pipe = cso; + vp->pipe = *cso; return (void *)vp; } @@ -472,7 +476,7 @@ nv30_fp_state_create(struct pipe_context *pipe, struct nv30_fragment_program *fp; fp = CALLOC(1, sizeof(struct nv30_fragment_program)); - fp->pipe = cso; + fp->pipe = *cso; return (void *)fp; } @@ -709,9 +713,9 @@ nv30_init_state_functions(struct nv30_context *nv30) nv30->pipe.delete_blend_state = nv30_blend_state_delete; nv30->pipe.create_sampler_state = nv30_sampler_state_create; - nv30->pipe.bind_sampler_state = nv30_sampler_state_bind; + nv30->pipe.bind_sampler_states = nv30_sampler_state_bind; nv30->pipe.delete_sampler_state = nv30_sampler_state_delete; - nv30->pipe.set_sampler_texture = nv30_set_sampler_texture; + nv30->pipe.set_sampler_textures = nv30_set_sampler_texture; nv30->pipe.create_rasterizer_state = nv30_rasterizer_state_create; nv30->pipe.bind_rasterizer_state = nv30_rasterizer_state_bind; diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index 233600f69ab..117520dd13c 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -60,7 +60,7 @@ struct nv30_vertex_program_data { }; struct nv30_vertex_program { - const struct pipe_shader_state *pipe; + struct pipe_shader_state pipe; boolean translated; struct nv30_vertex_program_exec *insns; @@ -84,7 +84,7 @@ struct nv30_fragment_program_data { }; struct nv30_fragment_program { - const struct pipe_shader_state *pipe; + struct pipe_shader_state pipe; boolean translated; boolean on_hw; diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 75f7351261d..e9b62ff48b4 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -541,7 +541,7 @@ nv30_vertprog_prepare(struct nv30_vpc *vpc) struct tgsi_parse_context p; int nr_imm = 0; - tgsi_parse_init(&p, vpc->vp->pipe->tokens); + tgsi_parse_init(&p, vpc->vp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&p)) { const union tgsi_full_token *tok = &p.FullToken; @@ -582,7 +582,7 @@ nv30_vertprog_translate(struct nv30_context *nv30, return; } - tgsi_parse_init(&parse, vp->pipe->tokens); + tgsi_parse_init(&parse, vp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&parse)) { tgsi_parse_token(&parse); diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index e118776306b..100c678187a 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -136,6 +136,8 @@ struct nv40_context { unsigned idxbuf_format; struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned nr_samplers; + unsigned nr_textures; unsigned dirty_samplers; struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index d981a02a639..953f9cd9082 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -580,15 +580,14 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); break; case TGSI_OPCODE_TEX: - if (finst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide == - TGSI_EXTSWIZZLE_W) { - tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); - } else - tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); + tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none); break; case TGSI_OPCODE_TXB: tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none); break; + case TGSI_OPCODE_TXP: + tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none); + break; case TGSI_OPCODE_XPD: tmp = temp(fpc); arith(fpc, 0, MUL, tmp, mask, @@ -680,7 +679,7 @@ nv40_fragprog_translate(struct nv40_context *nv40, fpc->high_temp = -1; fpc->num_regs = 2; - tgsi_parse_init(&parse, fp->pipe->tokens); + tgsi_parse_init(&parse, fp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&parse)) { tgsi_parse_token(&parse); diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index caa2f9df0c6..321d5de0415 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -1,6 +1,7 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" +#include "pipe/p_inlines.h" #include "nv40_context.h" #include "nv40_state.h" @@ -251,14 +252,22 @@ nv40_sampler_state_create(struct pipe_context *pipe, } static void -nv40_sampler_state_bind(struct pipe_context *pipe, unsigned unit, - void *hwcso) +nv40_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_sampler_state *ps = hwcso; + unsigned unit; - nv40->tex_sampler[unit] = ps; - nv40->dirty_samplers |= (1 << unit); + for (unit = 0; unit < nr; unit++) { + nv40->tex_sampler[unit] = sampler[unit]; + nv40->dirty_samplers |= (1 << unit); + } + + for (unit = nr; unit < nv40->nr_samplers; unit++) { + nv40->tex_sampler[unit] = NULL; + nv40->dirty_samplers |= (1 << unit); + } + + nv40->nr_samplers = nr; nv40->dirty |= NV40_NEW_SAMPLER; } @@ -269,13 +278,25 @@ nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso) } static void -nv40_set_sampler_texture(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *miptree) +nv40_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **miptree) { struct nv40_context *nv40 = nv40_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + pipe_texture_reference((struct pipe_texture **) + &nv40->tex_miptree[unit], miptree[unit]); + nv40->dirty_samplers |= (1 << unit); + } + + for (unit = nr; unit < nv40->nr_textures; unit++) { + pipe_texture_reference((struct pipe_texture **) + &nv40->tex_miptree[unit], NULL); + nv40->dirty_samplers |= (1 << unit); + } - nv40->tex_miptree[unit] = (struct nv40_miptree *)miptree; - nv40->dirty_samplers |= (1 << unit); + nv40->nr_textures = nr; nv40->dirty |= NV40_NEW_SAMPLER; } @@ -490,7 +511,7 @@ nv40_vp_state_create(struct pipe_context *pipe, struct nv40_vertex_program *vp; vp = CALLOC(1, sizeof(struct nv40_vertex_program)); - vp->pipe = cso; + vp->pipe = *cso; return (void *)vp; } @@ -521,7 +542,7 @@ nv40_fp_state_create(struct pipe_context *pipe, struct nv40_fragment_program *fp; fp = CALLOC(1, sizeof(struct nv40_fragment_program)); - fp->pipe = cso; + fp->pipe = *cso; return (void *)fp; } @@ -649,9 +670,9 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.delete_blend_state = nv40_blend_state_delete; nv40->pipe.create_sampler_state = nv40_sampler_state_create; - nv40->pipe.bind_sampler_state = nv40_sampler_state_bind; + nv40->pipe.bind_sampler_states = nv40_sampler_state_bind; nv40->pipe.delete_sampler_state = nv40_sampler_state_delete; - nv40->pipe.set_sampler_texture = nv40_set_sampler_texture; + nv40->pipe.set_sampler_textures = nv40_set_sampler_texture; nv40->pipe.create_rasterizer_state = nv40_rasterizer_state_create; nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind; diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index e5217fe91c4..a02ea0c8781 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -23,7 +23,7 @@ struct nv40_vertex_program_data { }; struct nv40_vertex_program { - const struct pipe_shader_state *pipe; + struct pipe_shader_state pipe; boolean translated; struct nv40_vertex_program_exec *insns; @@ -48,7 +48,7 @@ struct nv40_fragment_program_data { }; struct nv40_fragment_program { - const struct pipe_shader_state *pipe; + struct pipe_shader_state pipe; boolean translated; unsigned samplers; diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 5b7a343e55d..3d730c1a321 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -535,7 +535,7 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) struct tgsi_parse_context p; int nr_imm = 0; - tgsi_parse_init(&p, vpc->vp->pipe->tokens); + tgsi_parse_init(&p, vpc->vp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&p)) { const union tgsi_full_token *tok = &p.FullToken; @@ -576,7 +576,7 @@ nv40_vertprog_translate(struct nv40_context *nv40, return; } - tgsi_parse_init(&parse, vp->pipe->tokens); + tgsi_parse_init(&parse, vp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&parse)) { tgsi_parse_token(&parse); diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index aa65fd482e4..b096a2583d7 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -90,8 +90,7 @@ nv50_sampler_state_create(struct pipe_context *pipe, } static void -nv50_sampler_state_bind(struct pipe_context *pipe, unsigned unit, - void *hwcso) +nv50_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) { } @@ -101,8 +100,8 @@ nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) } static void -nv50_set_sampler_texture(struct pipe_context *pipe, unsigned unit, - struct pipe_texture *pt) +nv50_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **pt) { } @@ -442,9 +441,9 @@ nv50_init_state_functions(struct nv50_context *nv50) nv50->pipe.delete_blend_state = nv50_blend_state_delete; nv50->pipe.create_sampler_state = nv50_sampler_state_create; - nv50->pipe.bind_sampler_state = nv50_sampler_state_bind; + nv50->pipe.bind_sampler_states = nv50_sampler_state_bind; nv50->pipe.delete_sampler_state = nv50_sampler_state_delete; - nv50->pipe.set_sampler_texture = nv50_set_sampler_texture; + nv50->pipe.set_sampler_textures = nv50_set_sampler_texture; nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create; nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; -- cgit v1.2.3 From 1cec61e441ad5b4b1ac8d1abcaa7535bc1827eb3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 13 Mar 2008 18:08:22 +1100 Subject: nouveau: NV9X is basically a G80, fix issue with NV6X being detected as G80. --- src/gallium/drivers/nv50/nv50_screen.c | 29 +++++++++++----- src/gallium/winsys/dri/nouveau/nouveau_context.c | 42 ++++++++++++++++++++---- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 1 + 3 files changed, 57 insertions(+), 15 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 62c23c790ca..721c6421d1f 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -6,8 +6,9 @@ #include "nouveau/nouveau_stateobj.h" -#define GRCLASS5097_CHIPSETS 0x00000000 -#define GRCLASS8297_CHIPSETS 0x00000010 +#define NV5X_GRCLASS5097_CHIPSETS 0x00000001 +#define NV8X_GRCLASS8297_CHIPSETS 0x00000010 +#define NV9X_GRCLASS8297_CHIPSETS 0x00000004 static boolean nv50_screen_is_format_supported(struct pipe_screen *pscreen, @@ -117,12 +118,24 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, return NULL; } - if (GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) { - tesla_class = 0x5097; - } else - if (GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) { - tesla_class = 0x8297; - } else { + switch (chipset & 0xf0) { + case 0x50: + if (NV5X_GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f))) + tesla_class = 0x5097; + break; + case 0x80: + if (NV8X_GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) + tesla_class = 0x8297; + break; + case 0x90: + if (NV9X_GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f))) + tesla_class = 0x8297; + break; + default: + break; + } + + if (tesla_class == 0) { NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset); nv50_screen_destroy(&screen->pipe); return NULL; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index 8dac08a5d21..8b20b3689c6 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -75,6 +75,7 @@ nouveau_channel_context_create(struct nouveau_device *nvdev, unsigned chipset) switch (chipset & 0xf0) { case 0x50: case 0x80: + case 0x90: ret = nouveau_surface_channel_create_nv50(nvc); break; default: @@ -109,6 +110,7 @@ nouveau_context_create(const __GLcontextModes *glVis, st_share = ((struct nouveau_context *)sharedContextPrivate)->st; } + /* Check for supported arch */ if ((ret = nouveau_device_get_param(nv_screen->device, NOUVEAU_GETPARAM_CHIPSET_ID, &nv->chipset))) { @@ -116,6 +118,22 @@ nouveau_context_create(const __GLcontextModes *glVis, return GL_FALSE; } + switch (nv->chipset & 0xf0) { + case 0x30: + /* NV30 */ + case 0x40: + case 0x60: + /* NV40 */ + case 0x50: + case 0x80: + case 0x90: + /* G80 */ + break; + default: + NOUVEAU_ERR("Unsupported chipset: NV%02x\n", nv->chipset); + return GL_FALSE; + } + driContextPriv->driverPrivate = (void *)nv; nv->nv_screen = nv_screen; nv->dri_screen = driScrnPriv; @@ -176,10 +194,15 @@ nouveau_context_create(const __GLcontextModes *glVis, } } - /*XXX: temporary - disable multi-context/single-channel on non-NV4x */ + /*XXX: temporary - disable multi-context/single-channel on pre-NV4x */ switch (nv->chipset & 0xf0) { case 0x40: case 0x60: + /* NV40 class */ + case 0x50: + case 0x80: + case 0x90: + /* G80 class */ break; default: nvc = NULL; @@ -215,12 +238,17 @@ nouveau_context_create(const __GLcontextModes *glVis, } /* Create pipe */ - if (nv->chipset < 0x50) - ret = nouveau_surface_init_nv04(nv); - else - ret = nouveau_surface_init_nv50(nv); - if (ret) { - return GL_FALSE; + switch (nv->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + if (nouveau_surface_init_nv50(nv)) + return GL_FALSE; + break; + default: + if (nouveau_surface_init_nv04(nv)) + return GL_FALSE; + break; } if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index dc7c4c3d71d..64e84fb68e4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -93,6 +93,7 @@ nouveau_pipe_create(struct nouveau_context *nv) break; case 0x50: case 0x80: + case 0x90: hws_create = nv50_screen_create; hw_create = nv50_create; break; -- cgit v1.2.3 From 9a3320e0791a4a03f5f4b7a6f9c3b0d9d78655b3 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sat, 15 Mar 2008 04:42:48 +0100 Subject: nv30: only 2 RTs. --- src/gallium/drivers/nv30/nv30_state.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 951a32bc81e..722626db1fa 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -546,7 +546,7 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, int i, colour_format = 0, zeta_format = 0; rt_enable = 0; - for (i = 0; i < 4; i++) { + for (i = 0; i < 2; i++) { if (!fb->cbufs[i]) continue; -- cgit v1.2.3 From 509044609d5121b2a09d64bd24d7aa37e3744a77 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sat, 15 Mar 2008 04:43:12 +0100 Subject: nouveau: latest header. --- src/gallium/drivers/nouveau/nouveau_class.h | 44 ++++++++++++++++++----------- 1 file changed, 28 insertions(+), 16 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 1b015507903..7392490df01 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -1750,22 +1750,34 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_DMA_VTXBUF0 0x0000018c #define NV10TCL_DMA_IN_MEMORY2 0x00000194 #define NV10TCL_DMA_IN_MEMORY3 0x00000198 -#define NV10TCL_VIEWPORT_HORIZ 0x00000200 -#define NV10TCL_VIEWPORT_HORIZ_X_SHIFT 0 -#define NV10TCL_VIEWPORT_HORIZ_X_MASK 0x0000ffff -#define NV10TCL_VIEWPORT_HORIZ_W_SHIFT 16 -#define NV10TCL_VIEWPORT_HORIZ_W_MASK 0xffff0000 -#define NV10TCL_VIEWPORT_VERT 0x00000204 -#define NV10TCL_VIEWPORT_VERT_Y_SHIFT 0 -#define NV10TCL_VIEWPORT_VERT_Y_MASK 0x0000ffff -#define NV10TCL_VIEWPORT_VERT_H_SHIFT 16 -#define NV10TCL_VIEWPORT_VERT_H_MASK 0xffff0000 -#define NV10TCL_BUFFER_FORMAT 0x00000208 -#define NV10TCL_BUFFER_PITCH 0x0000020c -#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_SHIFT 0 -#define NV10TCL_BUFFER_PITCH_COLOR_PITCH_MASK 0x0000ffff -#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_SHIFT 16 -#define NV10TCL_BUFFER_PITCH_ZETA_PITCH_MASK 0xffff0000 +#define NV10TCL_RT_HORIZ 0x00000200 +#define NV10TCL_RT_HORIZ_X_SHIFT 0 +#define NV10TCL_RT_HORIZ_X_MASK 0x0000ffff +#define NV10TCL_RT_HORIZ_W_SHIFT 16 +#define NV10TCL_RT_HORIZ_W_MASK 0xffff0000 +#define NV10TCL_RT_VERT 0x00000204 +#define NV10TCL_RT_VERT_Y_SHIFT 0 +#define NV10TCL_RT_VERT_Y_MASK 0x0000ffff +#define NV10TCL_RT_VERT_H_SHIFT 16 +#define NV10TCL_RT_VERT_H_MASK 0xffff0000 +#define NV10TCL_RT_FORMAT 0x00000208 +#define NV10TCL_RT_FORMAT_TYPE_SHIFT 8 +#define NV10TCL_RT_FORMAT_TYPE_MASK 0x00000f00 +#define NV10TCL_RT_FORMAT_TYPE_LINEAR 0x00000100 +#define NV10TCL_RT_FORMAT_TYPE_SWIZZLED 0x00000200 +#define NV10TCL_RT_FORMAT_COLOR_SHIFT 0 +#define NV10TCL_RT_FORMAT_COLOR_MASK 0x0000001f +#define NV10TCL_RT_FORMAT_COLOR_R5G6B5 0x00000003 +#define NV10TCL_RT_FORMAT_COLOR_X8R8G8B8 0x00000005 +#define NV10TCL_RT_FORMAT_COLOR_A8R8G8B8 0x00000008 +#define NV10TCL_RT_FORMAT_COLOR_B8 0x00000009 +#define NV10TCL_RT_FORMAT_COLOR_X8B8G8R8 0x0000000f +#define NV10TCL_RT_FORMAT_COLOR_A8B8G8R8 0x00000010 +#define NV10TCL_RT_PITCH 0x0000020c +#define NV10TCL_RT_PITCH_COLOR_PITCH_SHIFT 0 +#define NV10TCL_RT_PITCH_COLOR_PITCH_MASK 0x0000ffff +#define NV10TCL_RT_PITCH_ZETA_PITCH_SHIFT 16 +#define NV10TCL_RT_PITCH_ZETA_PITCH_MASK 0xffff0000 #define NV10TCL_COLOR_OFFSET 0x00000210 #define NV10TCL_ZETA_OFFSET 0x00000214 #define NV10TCL_TX_OFFSET(x) (0x00000218+((x)*4)) -- cgit v1.2.3 From d493203045b214770473f8afeaa610542fe42c2a Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Sat, 15 Mar 2008 05:37:57 +0100 Subject: nv10. --- configs/default | 2 +- configs/linux-dri | 2 +- src/gallium/drivers/nouveau/nouveau_winsys.h | 7 + src/gallium/drivers/nv10/Makefile | 28 + src/gallium/drivers/nv10/nv10_clear.c | 12 + src/gallium/drivers/nv10/nv10_context.c | 325 +++++++++++ src/gallium/drivers/nv10/nv10_context.h | 127 +++++ src/gallium/drivers/nv10/nv10_fragprog.c | 22 + src/gallium/drivers/nv10/nv10_fragtex.c | 145 +++++ src/gallium/drivers/nv10/nv10_miptree.c | 134 +++++ src/gallium/drivers/nv10/nv10_prim_vbuf.c | 224 ++++++++ src/gallium/drivers/nv10/nv10_screen.c | 150 +++++ src/gallium/drivers/nv10/nv10_screen.h | 19 + src/gallium/drivers/nv10/nv10_state.c | 697 +++++++++++++++++++++++ src/gallium/drivers/nv10/nv10_state.h | 135 +++++ src/gallium/drivers/nv10/nv10_state_emit.c | 69 +++ src/gallium/drivers/nv10/nv10_surface.c | 65 +++ src/gallium/drivers/nv10/nv10_vbo.c | 72 +++ src/gallium/winsys/dri/nouveau/Makefile | 1 + src/gallium/winsys/dri/nouveau/nouveau_context.c | 3 + src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 5 + 21 files changed, 2242 insertions(+), 2 deletions(-) create mode 100644 src/gallium/drivers/nv10/Makefile create mode 100644 src/gallium/drivers/nv10/nv10_clear.c create mode 100644 src/gallium/drivers/nv10/nv10_context.c create mode 100644 src/gallium/drivers/nv10/nv10_context.h create mode 100644 src/gallium/drivers/nv10/nv10_fragprog.c create mode 100644 src/gallium/drivers/nv10/nv10_fragtex.c create mode 100644 src/gallium/drivers/nv10/nv10_miptree.c create mode 100644 src/gallium/drivers/nv10/nv10_prim_vbuf.c create mode 100644 src/gallium/drivers/nv10/nv10_screen.c create mode 100644 src/gallium/drivers/nv10/nv10_screen.h create mode 100644 src/gallium/drivers/nv10/nv10_state.c create mode 100644 src/gallium/drivers/nv10/nv10_state.h create mode 100644 src/gallium/drivers/nv10/nv10_state_emit.c create mode 100644 src/gallium/drivers/nv10/nv10_surface.c create mode 100644 src/gallium/drivers/nv10/nv10_vbo.c diff --git a/configs/default b/configs/default index 49d3737ec7c..d010721a103 100644 --- a/configs/default +++ b/configs/default @@ -70,7 +70,7 @@ PROGRAM_DIRS = demos redbook samples glsl xdemos # Gallium directories and GALLIUM_AUXILIARY_DIRS = draw cso_cache pipebuffer tgsi rtasm util sct GALLIUM_AUXILIARIES = $(foreach DIR,$(GALLIUM_AUXILIARY_DIRS),$(TOP)/src/gallium/auxiliary/$(DIR)/lib$(DIR).a) -GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv30 nv40 nv50 failover +GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv10 nv30 nv40 nv50 failover GALLIUM_DRIVERS = $(foreach DIR,$(GALLIUM_DRIVER_DIRS),$(TOP)/src/gallium/drivers/$(DIR)/lib$(DIR).a) GALLIUM_WINSYS_DIRS = xlib diff --git a/configs/linux-dri b/configs/linux-dri index 67e60cbd4c5..bf7881937e4 100644 --- a/configs/linux-dri +++ b/configs/linux-dri @@ -65,4 +65,4 @@ GALLIUM_WINSYS_DIRS = dri # gamma are missing because they have not been converted to use the new # interface. -DRI_DIRS = intel +DRI_DIRS = nouveau diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 44c8bb919bb..e4b20478a0c 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -49,6 +49,13 @@ struct nouveau_winsys { unsigned, unsigned, unsigned, unsigned, unsigned); }; +extern struct pipe_screen * +nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, + unsigned chipset); + +extern struct pipe_context * +nv10_create(struct pipe_screen *, unsigned pctx_id); + extern struct pipe_screen * nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, unsigned chipset); diff --git a/src/gallium/drivers/nv10/Makefile b/src/gallium/drivers/nv10/Makefile new file mode 100644 index 00000000000..4ba7ce586d6 --- /dev/null +++ b/src/gallium/drivers/nv10/Makefile @@ -0,0 +1,28 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv10 + +DRIVER_SOURCES = \ + nv10_clear.c \ + nv10_context.c \ + nv10_fragprog.c \ + nv10_fragtex.c \ + nv10_miptree.c \ + nv10_prim_vbuf.c \ + nv10_screen.c \ + nv10_state.c \ + nv10_state_emit.c \ + nv10_surface.c \ + nv10_vbo.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/drivers/nv10/nv10_clear.c b/src/gallium/drivers/nv10/nv10_clear.c new file mode 100644 index 00000000000..be7e09cf4b0 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv10_context.h" + +void +nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c new file mode 100644 index 00000000000..2599acf2861 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -0,0 +1,325 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static void +nv10_flush(struct pipe_context *pipe, unsigned flags) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + if (flags & PIPE_FLUSH_TEXTURE_CACHE) { + BEGIN_RING(celsius, 0x1fd8, 1); + OUT_RING (2); + BEGIN_RING(celsius, 0x1fd8, 1); + OUT_RING (1); + } + + if (flags & PIPE_FLUSH_WAIT) { + nvws->notifier_reset(nv10->sync, 0); + BEGIN_RING(celsius, 0x104, 1); + OUT_RING (0); + BEGIN_RING(celsius, 0x100, 1); + OUT_RING (0); + } + + FIRE_RING(); + + if (flags & PIPE_FLUSH_WAIT) + nvws->notifier_wait(nv10->sync, 0, 0, 2000); +} + +static void +nv10_destroy(struct pipe_context *pipe) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + if (nv10->draw) + draw_destroy(nv10->draw); + + nvws->res_free(&nv10->vertprog.exec_heap); + nvws->res_free(&nv10->vertprog.data_heap); + + nvws->notifier_free(&nv10->sync); + + nvws->grobj_free(&nv10->celsius); + + free(nv10); +} + +static boolean +nv10_init_hwctx(struct nv10_context *nv10, int celsius_class) +{ + struct nouveau_winsys *nvws = nv10->nvws; + int ret; + int i; + + ret = nvws->grobj_alloc(nvws, celsius_class, &nv10->celsius); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); + OUT_RING (nv10->sync->handle); + BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY2, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 2); + OUT_RING (0); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1); + OUT_RING ((0x7ff<<16)|0x800); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1); + OUT_RING ((0x7ff<<16)|0x800); + + for (i=1;i<8;i++) { + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(i), 1); + OUT_RING (0); + } + + BEGIN_RING(celsius, 0x290, 1); + OUT_RING ((0x10<<16)|1); + BEGIN_RING(celsius, 0x3f4, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + if (celsius_class != NV10TCL) { + /* For nv11, nv17 */ + BEGIN_RING(celsius, 0x120, 3); + OUT_RING (0); + OUT_RING (1); + OUT_RING (2); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + } + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + /* Set state */ + BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 2); + OUT_RING (0x207); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_RC_OUT_ALPHA(0), 6); + OUT_RING (0x00000c00); + OUT_RING (0); + OUT_RING (0x00000c00); + OUT_RING (0x18000000); + OUT_RING (0x300c0000); + OUT_RING (0x00001c80); + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2); + OUT_RING (1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VERTEX_WEIGHT_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_SRC, 4); + OUT_RING (1); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x8006); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 8); + OUT_RING (0xff); + OUT_RING (0x207); + OUT_RING (0); + OUT_RING (0xff); + OUT_RING (0x1e00); + OUT_RING (0x1e00); + OUT_RING (0x1e00); + OUT_RING (0x1d01); + BEGIN_RING(celsius, NV10TCL_NORMALIZE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LIGHT_MODEL, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_COLOR_CONTROL, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ENABLED_LIGHTS, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + OUT_RING (0x201); + BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_FACTOR, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (8); + BEGIN_RING(celsius, NV10TCL_POINT_PARAMETERS_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LINE_WIDTH, 1); + OUT_RING (8); + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (0x1b02); + OUT_RING (0x1b02); + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (0x405); + OUT_RING (0x901); + BEGIN_RING(celsius, NV10TCL_POLYGON_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_CLIP_PLANE_ENABLE(0), 8); + for (i=0;i<8;i++) { + OUT_RING (0); + } + BEGIN_RING(celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3); + OUT_RING (0x3fc00000); /* -1.50 */ + OUT_RING (0xbdb8aa0a); /* -0.09 */ + OUT_RING (0); /* 0.00 */ + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_FOG_MODE, 2); + OUT_RING (0x802); + OUT_RING (2); + /* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when + * using texturing, except when using the texture matrix + */ + BEGIN_RING(celsius, NV10TCL_VIEW_MATRIX_ENABLE, 1); + OUT_RING (6); + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (0x01010101); + + /* Set vertex component */ + BEGIN_RING(celsius, NV10TCL_VERTEX_COL_4F_R, 4); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_COL2_3F_R, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VERTEX_NOR_3F_X, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_TX0_4F_S, 4); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_TX1_4F_S, 4); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_FOG_1F, 1); + OUT_RINGf (0.0); + BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1); + OUT_RING (1); + + + FIRE_RING (); + return TRUE; +} + +struct pipe_context * +nv10_create(struct pipe_screen *screen, unsigned pctx_id) +{ + struct pipe_winsys *pipe_winsys = screen->winsys; + struct nouveau_winsys *nvws = nv10_screen(screen)->nvws; + unsigned chipset = nv10_screen(screen)->chipset; + struct nv10_context *nv10; + int celsius_class = 0, ret; + + if (chipset>=0x20) + celsius_class=NV11TCL; + else if (chipset>=0x17) + celsius_class=NV17TCL; + else if (chipset>=0x11) + celsius_class=NV11TCL; + else + celsius_class=NV10TCL; + + nv10 = CALLOC_STRUCT(nv10_context); + if (!nv10) + return NULL; + nv10->chipset = chipset; + nv10->nvws = nvws; + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &nv10->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv10_destroy(&nv10->pipe); + return NULL; + } + + /* Vtxprog resources */ + if (nvws->res_init(&nv10->vertprog.exec_heap, 0, 512) || + nvws->res_init(&nv10->vertprog.data_heap, 0, 256)) { + nv10_destroy(&nv10->pipe); + return NULL; + } + + /* Static celsius initialisation */ + if (!nv10_init_hwctx(nv10, celsius_class)) { + nv10_destroy(&nv10->pipe); + return NULL; + } + + /* Pipe context setup */ + nv10->pipe.winsys = pipe_winsys; + + nv10->pipe.destroy = nv10_destroy; + + nv10->pipe.draw_arrays = nv10_draw_arrays; + nv10->pipe.draw_elements = nv10_draw_elements; + nv10->pipe.clear = nv10_clear; + + nv10->pipe.flush = nv10_flush; + + nv10_init_surface_functions(nv10); + nv10_init_state_functions(nv10); + + nv10->draw = draw_create(); + assert(nv10->draw); + draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10)); + + return &nv10->pipe; +} + diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h new file mode 100644 index 00000000000..8269d6121f0 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -0,0 +1,127 @@ +#ifndef __NV10_CONTEXT_H__ +#define __NV10_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv10_context *ctx = nv10 +#include "nouveau/nouveau_push.h" + +#include "nv10_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV10_NEW_VERTPROG (1 << 1) +#define NV10_NEW_FRAGPROG (1 << 2) +#define NV10_NEW_ARRAYS (1 << 3) +#define NV10_NEW_VBO (1 << 4) + +struct nv10_context { + struct pipe_context pipe; + struct nouveau_winsys *nvws; + + struct draw_context *draw; + + int chipset; + struct nouveau_grobj *celsius; + struct nouveau_notifier *sync; + + uint32_t dirty; + + struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv10_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + uint32_t rt_enable; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; + + struct { + struct pipe_buffer *buffer; + uint32_t format; + } tex[16]; + + unsigned vb_enable; + struct { + struct pipe_buffer *buffer; + unsigned delta; + } vb[16]; + + struct vertex_info vertex_info; + struct { + + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + + struct nv10_vertex_program *active; + + struct nv10_vertex_program *current; + struct pipe_buffer *constant_buf; + } vertprog; + + struct { + struct nv10_fragment_program *active; + + struct nv10_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; +}; + +static INLINE struct nv10_context * +nv10_context(struct pipe_context *pipe) +{ + return (struct nv10_context *)pipe; +} + +extern void nv10_init_state_functions(struct nv10_context *nv10); +extern void nv10_init_surface_functions(struct nv10_context *nv10); +extern void nv10_init_miptree_functions(struct pipe_screen *screen); + +/* nv10_clear.c */ +extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +/* nv10_draw.c */ +extern struct draw_stage *nv10_draw_render_stage(struct nv10_context *nv10); + +/* nv10_fragprog.c */ +extern void nv10_fragprog_bind(struct nv10_context *, + struct nv10_fragment_program *); +extern void nv10_fragprog_destroy(struct nv10_context *, + struct nv10_fragment_program *); + +/* nv10_fragtex.c */ +extern void nv10_fragtex_bind(struct nv10_context *); + +/* nv10_prim_vbuf.c */ +struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ); + +/* nv10_state.c and friends */ +extern void nv10_emit_hw_state(struct nv10_context *nv10); +extern void nv10_state_tex_update(struct nv10_context *nv10); + +/* nv10_vbo.c */ +extern boolean nv10_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv10_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count); + + +#endif diff --git a/src/gallium/drivers/nv10/nv10_fragprog.c b/src/gallium/drivers/nv10/nv10_fragprog.c new file mode 100644 index 00000000000..2a63c8a704e --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_fragprog.c @@ -0,0 +1,22 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_util.h" + +#include "nv10_context.h" + +void +nv10_fragprog_bind(struct nv10_context *nv10, struct nv10_fragment_program *fp) +{ +} + +void +nv10_fragprog_destroy(struct nv10_context *nv10, + struct nv10_fragment_program *fp) +{ +} + diff --git a/src/gallium/drivers/nv10/nv10_fragtex.c b/src/gallium/drivers/nv10/nv10_fragtex.c new file mode 100644 index 00000000000..a4bf1082845 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_fragtex.c @@ -0,0 +1,145 @@ +#include "nv10_context.h" + +static INLINE int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + +#define _(m,tf) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV10TCL_TX_FORMAT_FORMAT_##tf, \ +} + +struct nv10_texture_format { + boolean defined; + uint pipe; + int format; +}; + +static struct nv10_texture_format +nv10_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8), + _(A1R5G5B5_UNORM, A1R5G5B5), + _(A4R4G4B4_UNORM, A4R4G4B4), + _(U_L8 , L8 ), + _(U_A8 , A8 ), + _(U_A8_L8 , A8L8 ), +// _(RGB_DXT1 , DXT1, ), +// _(RGBA_DXT1 , DXT1, ), +// _(RGBA_DXT3 , DXT3, ), +// _(RGBA_DXT5 , DXT5, ), + {}, +}; + +static struct nv10_texture_format * +nv10_fragtex_format(uint pipe_format) +{ + struct nv10_texture_format *tf = nv10_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv10_fragtex_build(struct nv10_context *nv10, int unit) +{ + struct nv10_sampler_state *ps = nv10->tex_sampler[unit]; + struct nv10_miptree *nv10mt = nv10->tex_miptree[unit]; + struct pipe_texture *pt = &nv10mt->base; + struct nv10_texture_format *tf; + uint32_t txf, txs, txp; + + tf = nv10_fragtex_format(pt->format); + if (!tf || !tf->defined) { + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); + return; + } + + txf = tf->format << 8; + txf |= (pt->last_level + 1) << 16; + txf |= log2i(pt->width[0]) << 20; + txf |= log2i(pt->height[0]) << 24; + txf |= log2i(pt->depth[0]) << 28; + txf |= 8; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE: + txf |= NV10TCL_TX_FORMAT_CUBE_MAP; + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= (2<<4); + break; + case PIPE_TEXTURE_1D: + txf |= (1<<4); + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(unit), 8); + OUT_RELOCl(nv10mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv10mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RING (ps->wrap); + OUT_RING (0x40000000); /* enable */ + OUT_RING (txs); + OUT_RING (ps->filt | 0x2000 /* magic */); + OUT_RING ((pt->width[0] << 16) | pt->height[0]); + OUT_RING (ps->bcol); +} + +void +nv10_fragtex_bind(struct nv10_context *nv10) +{ + struct nv10_fragment_program *fp = nv10->fragprog.active; + unsigned samplers, unit; + + samplers = nv10->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(celsius, NV10TCL_TX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv10->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv10_fragtex_build(nv10, unit); + } + + nv10->fp_samplers = fp->samplers; +} + diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c new file mode 100644 index 00000000000..7b7f39b80c9 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -0,0 +1,134 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static void +nv10_miptree_layout(struct nv10_miptree *nv10mt) +{ + struct pipe_texture *pt = &nv10mt->base; + boolean swizzled = FALSE; + uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint offset = 0; + int nr_faces, l, f; + + if (pt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else { + nr_faces = 1; + } + + for (l = 0; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->depth[l] = depth; + + if (swizzled) + nv10mt->level[l].pitch = pt->width[l] * pt->cpp; + else + nv10mt->level[l].pitch = pt->width[0] * pt->cpp; + nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; + + nv10mt->level[l].image_offset = + CALLOC(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + depth = MAX2(1, depth >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = 0; l <= pt->last_level; l++) { + nv10mt->level[l].image_offset[f] = offset; + offset += nv10mt->level[l].pitch * pt->height[l]; + } + } + + nv10mt->total_size = offset; +} + +static struct pipe_texture * +nv10_miptree_create(struct pipe_screen *screen, struct pipe_texture *pt) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv10_miptree *mt; + + mt = MALLOC(sizeof(struct nv10_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + mt->base.refcount = 1; + mt->base.screen = screen; + + nv10_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + free(mt); + return NULL; + } + + return &mt->base; +} + +static void +nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = screen->winsys; + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt; + int l; + + pipe_buffer_reference(ws, &nv10mt->buffer, NULL); + for (l = 0; l <= mt->last_level; l++) { + if (nv10mt->level[l].image_offset) + free(nv10mt->level[l].image_offset); + } + free(nv10mt); + } +} + +static struct pipe_surface * +nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv10mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = nv10mt->level[level].pitch / ps->cpp; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv10mt->level[level].image_offset[face]; + } else { + ps->offset = nv10mt->level[level].image_offset[0]; + } + + return ps; +} +void +nv10_init_miptree_functions(struct pipe_screen *screen) +{ + struct nv10_screen *nv10screen = nv10_screen(screen); + + nv10screen->screen.texture_create = nv10_miptree_create; + nv10screen->screen.texture_release = nv10_miptree_release; + nv10screen->screen.get_tex_surface = nv10_miptree_surface_get; +} + diff --git a/src/gallium/drivers/nv10/nv10_prim_vbuf.c b/src/gallium/drivers/nv10/nv10_prim_vbuf.c new file mode 100644 index 00000000000..15268912234 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_prim_vbuf.c @@ -0,0 +1,224 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * \file + * Build post-transformation, post-clipping vertex buffers and element + * lists by hooking into the end of the primitive pipeline and + * manipulating the vertex_id field in the vertex headers. + * + * XXX: work in progress + * + * \author José Fonseca + * \author Keith Whitwell + */ + + +#include "draw/draw_vbuf.h" +#include "pipe/p_debug.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" +#include "pipe/p_winsys.h" + +#include "nv10_context.h" +#include "nv10_state.h" + + +/** + * Primitive renderer for nv10. + */ +struct nv10_vbuf_render { + struct vbuf_render base; + + struct nv10_context *nv10; + + /** Vertex buffer */ + struct pipe_buffer* buffer; + + /** Vertex size in bytes */ + unsigned vertex_size; + + /** Hardware primitive */ + unsigned hwprim; +}; + + +/** + * Basically a cast wrapper. + */ +static INLINE struct nv10_vbuf_render * +nv10_vbuf_render( struct vbuf_render *render ) +{ + assert(render); + return (struct nv10_vbuf_render *)render; +} + + +static const struct vertex_info * +nv10_vbuf_render_get_vertex_info( struct vbuf_render *render ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + return &nv10->vertex_info; +} + + +static void * +nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, + ushort vertex_size, + ushort nr_vertices ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + struct pipe_winsys *winsys = nv10->pipe.winsys; + size_t size = (size_t)vertex_size * (size_t)nr_vertices; + + assert(!nv10_render->buffer); + nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); + + nv10->dirty |= NV10_NEW_VBO; + + return winsys->buffer_map(winsys, + nv10_render->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); +} + + +static void +nv10_vbuf_render_set_primitive( struct vbuf_render *render, + unsigned prim ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + nv10_render->hwprim = prim + 1; +} + + +static void +nv10_vbuf_render_draw( struct vbuf_render *render, + const ushort *indices, + uint nr_indices) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + int push, i; + + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); + OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + + BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING(nv10_render->hwprim); + + if (nr_indices & 1) { + BEGIN_RING(celsius, NV10TCL_VB_ELEMENT_U32, 1); + OUT_RING (indices[0]); + indices++; nr_indices--; + } + + while (nr_indices) { + // XXX too big ? + push = MIN2(nr_indices, 2047 * 2); + + BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((indices[i+1] << 16) | indices[i]); + + nr_indices -= push; + indices += push; + } + + BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING (0); +} + + +static void +nv10_vbuf_render_release_vertices( struct vbuf_render *render, + void *vertices, + unsigned vertex_size, + unsigned vertices_used ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + struct pipe_winsys *winsys = nv10->pipe.winsys; + + assert(nv10_render->buffer); + winsys->buffer_unmap(winsys, nv10_render->buffer); + pipe_buffer_reference(winsys, &nv10_render->buffer, NULL); +} + + +static void +nv10_vbuf_render_destroy( struct vbuf_render *render ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + FREE(nv10_render); +} + + +/** + * Create a new primitive render. + */ +static struct vbuf_render * +nv10_vbuf_render_create( struct nv10_context *nv10 ) +{ + struct nv10_vbuf_render *nv10_render = CALLOC_STRUCT(nv10_vbuf_render); + + nv10_render->nv10 = nv10; + + nv10_render->base.max_vertex_buffer_bytes = 1024*1024; + nv10_render->base.max_indices = 64*1024; + nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info; + nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices; + nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive; + nv10_render->base.draw = nv10_vbuf_render_draw; + nv10_render->base.release_vertices = nv10_vbuf_render_release_vertices; + nv10_render->base.destroy = nv10_vbuf_render_destroy; + + return &nv10_render->base; +} + + +/** + * Create a new primitive vbuf/render stage. + */ +struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ) +{ + struct vbuf_render *render; + struct draw_stage *stage; + + render = nv10_vbuf_render_create(nv10); + if(!render) + return NULL; + + stage = draw_vbuf_stage( nv10->draw, render ); + if(!stage) { + render->destroy(render); + return NULL; + } + + return stage; +} diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c new file mode 100644 index 00000000000..4d8e0b05ccd --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -0,0 +1,150 @@ +#include "pipe/p_screen.h" +#include "pipe/p_util.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static const char * +nv10_screen_get_name(struct pipe_screen *screen) +{ + struct nv10_screen *nv10screen = nv10_screen(screen); + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", nv10screen->chipset); + return buffer; +} + +static const char * +nv10_screen_get_vendor(struct pipe_screen *screen) +{ + return "nouveau"; +} + +static int +nv10_screen_get_param(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 2; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 0; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 1; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 12; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 0; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 12; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv10_screen_get_paramf(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 2.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + case PIPE_CAP_BITMAP_TEXCOORD_BIAS: + return 0.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static boolean +nv10_screen_is_format_supported(struct pipe_screen *screen, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_U_L8: + case PIPE_FORMAT_U_A8: + case PIPE_FORMAT_U_I8: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static void +nv10_screen_destroy(struct pipe_screen *screen) +{ + FREE(screen); +} + +struct pipe_screen * +nv10_screen_create(struct pipe_winsys *winsys, struct nouveau_winsys *nvws, + unsigned chipset) +{ + struct nv10_screen *nv10screen = CALLOC_STRUCT(nv10_screen); + + if (!nv10screen) + return NULL; + + nv10screen->chipset = chipset; + nv10screen->nvws = nvws; + + nv10screen->screen.winsys = winsys; + + nv10screen->screen.destroy = nv10_screen_destroy; + + nv10screen->screen.get_name = nv10_screen_get_name; + nv10screen->screen.get_vendor = nv10_screen_get_vendor; + nv10screen->screen.get_param = nv10_screen_get_param; + nv10screen->screen.get_paramf = nv10_screen_get_paramf; + nv10screen->screen.is_format_supported = + nv10_screen_is_format_supported; + + nv10_init_miptree_functions(&nv10screen->screen); + return &nv10screen->screen; +} + diff --git a/src/gallium/drivers/nv10/nv10_screen.h b/src/gallium/drivers/nv10/nv10_screen.h new file mode 100644 index 00000000000..ac8c04072ab --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_screen.h @@ -0,0 +1,19 @@ +#ifndef __NV10_SCREEN_H__ +#define __NV10_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv10_screen { + struct pipe_screen screen; + + struct nouveau_winsys *nvws; + unsigned chipset; +}; + +static INLINE struct nv10_screen * +nv10_screen(struct pipe_screen *screen) +{ + return (struct nv10_screen *)screen; +} + +#endif diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c new file mode 100644 index 00000000000..313230fe889 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -0,0 +1,697 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" + + +#include "nv10_context.h" +#include "nv10_state.h" + +static void nv10_vertex_layout(struct pipe_context* pipe) +{ + struct nv10_context *nv10 = nv10_context(pipe); + const struct pipe_shader_state *fs = nv10->fragprog.current->pipe; + uint32_t src = 0; + int i; + struct vertex_info vinfo; + + memset(&vinfo, 0, sizeof(vinfo)); + + for (i = 0; i < fs->num_inputs; i++) { + switch (fs->input_semantic_name[i]) { + case TGSI_SEMANTIC_POSITION: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + case TGSI_SEMANTIC_COLOR: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + default: + case TGSI_SEMANTIC_GENERIC: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + case TGSI_SEMANTIC_FOG: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + } + } + draw_compute_vertex_size(&vinfo); +} + +static void * +nv10_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv10_blend_state *cb; + + cb = malloc(sizeof(struct nv10_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + + cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + + cb->d_enable = cso->dither ? 1 : 0; + + return (void *)cb; +} + +static void +nv10_blend_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_blend_state *cb = hwcso; + + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); + OUT_RING (cb->d_enable); + + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); + OUT_RING (cb->b_enable); + OUT_RING (cb->b_srcfunc); + OUT_RING (cb->b_dstfunc); + + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (cb->c_mask); +} + +static void +nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + case PIPE_TEX_WRAP_MIRROR_CLAMP: + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; + break; + } + + return ret >> NV10TCL_TX_FORMAT_WRAP_S_SHIFT; +} + +static void * +nv10_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv10_sampler_state *ps; + uint32_t filter = 0; + + ps = malloc(sizeof(struct nv10_sampler_state)); + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy > 1.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + +/* if (cso->max_anisotropy >= 16.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_16X; + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_12X; + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_10X; + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_6X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_4X; + } else { + ps->en |= NV10TCL_TX_ENABLE_ANISO_2X; + }*/ + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MAGNIFY_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV10TCL_TX_FILTER_MAGNIFY_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST; + break; + } + break; + } + + ps->filt = filter; + +/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER; + break; + case PIPE_FUNC_GREATER: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER; + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL; + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL; + break; + case PIPE_FUNC_LESS: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS; + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL; + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL; + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS; + break; + default: + break; + } + }*/ + + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) +{ + struct nv10_context *nv10 = nv10_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv10->tex_sampler[unit] = sampler[unit]; + nv10->dirty_samplers |= (1 << unit); + } +} + +static void +nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **miptree) +{ + struct nv10_context *nv10 = nv10_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit]; + nv10->dirty_samplers |= (1 << unit); + } +} + +static void * +nv10_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv10_rasterizer_state *rs; + int i; + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + rs = malloc(sizeof(struct nv10_rasterizer_state)); + + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; + + rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; + rs->line_smooth_en = cso->line_smooth ? 1 : 0; + + rs->point_size = *(uint32_t*)&cso->point_size; + + rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; + + if (cso->front_winding == PIPE_WINDING_CCW) { + rs->front_face = NV10TCL_FRONT_FACE_CCW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + } else { + rs->front_face = NV10TCL_FRONT_FACE_CW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + } + + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CCW) + rs->cull_face = NV10TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV10TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_CW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CW) + rs->cull_face = NV10TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV10TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_BOTH: + rs->cull_face_en = 1; + rs->cull_face = NV10TCL_CULL_FACE_FRONT_AND_BACK; + break; + case PIPE_WINDING_NONE: + default: + rs->cull_face_en = 0; + rs->cull_face = 0; + break; + } + + if (cso->point_sprite) { + rs->point_sprite = (1 << 0); + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + rs->point_sprite |= (1 << (8 + i)); + } + } else { + rs->point_sprite = 0; + } + + return (void *)rs; +} + +static void +nv10_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_rasterizer_state *rs = hwcso; + + BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); + OUT_RING (rs->shade_model); + OUT_RING (rs->line_width); + + + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (rs->point_size); + + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (rs->poly_mode_front); + OUT_RING (rs->poly_mode_back); + + + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (rs->cull_face); + OUT_RING (rs->front_face); + + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); + OUT_RING (rs->line_smooth_en); + OUT_RING (rs->poly_smooth_en); + + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (rs->cull_face_en); + +/* BEGIN_RING(celsius, NV10TCL_POINT_SPRITE, 1); + OUT_RING (rs->point_sprite);*/ +} + +static void +nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv10_depth_stencil_alpha_state *hw; + + hw = malloc(sizeof(struct nv10_depth_stencil_alpha_state)); + + hw->depth.func = nvgl_comparison_op(cso->depth.func); + hw->depth.write_enable = cso->depth.writemask ? 1 : 0; + hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + + hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0; + hw->stencil.wmask = cso->stencil[0].write_mask; + hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func); + hw->stencil.ref = cso->stencil[0].ref_value; + hw->stencil.vmask = cso->stencil[0].value_mask; + hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op); + hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op); + hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op); + + hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; + hw->alpha.func = nvgl_comparison_op(cso->alpha.func); + hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + + return (void *)hw; +} + +static void +nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_depth_stencil_alpha_state *hw = hwcso; + + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 3); + OUT_RINGp ((uint32_t *)&hw->depth, 3); + BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); + OUT_RING (hw->stencil.enable); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); + OUT_RINGp ((uint32_t *)&(hw->stencil.wmask), 7); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 3); + OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); +} + +static void +nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv10_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv10_vertex_program *vp; + + vp = CALLOC(1, sizeof(struct nv10_vertex_program)); + vp->pipe = cso; + + return (void *)vp; +} + +static void +nv10_vp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_vertex_program *vp = hwcso; + + nv10->vertprog.current = vp; + nv10->dirty |= NV10_NEW_VERTPROG; +} + +static void +nv10_vp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_vertex_program *vp = hwcso; + + //nv10_vertprog_destroy(nv10, vp); + free(vp); +} + +static void * +nv10_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv10_fragment_program *fp; + + fp = CALLOC(1, sizeof(struct nv10_fragment_program)); + fp->pipe = cso; + + return (void *)fp; +} + +static void +nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_fragment_program *fp = hwcso; + + nv10->fragprog.current = fp; + nv10->dirty |= NV10_NEW_FRAGPROG; +} + +static void +nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_fragment_program *fp = hwcso; + + nv10_fragprog_destroy(nv10, fp); + free(fp); +} + +static void +nv10_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); + OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0)); +} + +static void +nv10_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + if (shader == PIPE_SHADER_VERTEX) { + nv10->vertprog.constant_buf = buf->buffer; + nv10->dirty |= NV10_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv10->fragprog.constant_buf = buf->buffer; + nv10->dirty |= NV10_NEW_FRAGPROG; + } +} + +static void +nv10_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct pipe_surface *rt, *zeta; + uint32_t rt_format, w, h; + int i, colour_format = 0, zeta_format = 0; + + w = fb->cbufs[0]->width; + h = fb->cbufs[0]->height; + colour_format = fb->cbufs[0]->format; + rt = fb->cbufs[0]; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + nv10->rt[0] = rt->buffer; + + if (zeta_format) + { + nv10->zeta = zeta->buffer; + } + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((w - 1) << 16) | 0); + OUT_RING (((h - 1) << 16) | 0); +} + +static void +nv10_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + NOUVEAU_ERR("line stipple hahaha\n"); +} + +static void +nv10_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + // XXX +/* BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ +} + +static void +nv10_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv10_context *nv10 = nv10_context(pipe); + +/* OUT_RINGf (vpt->translate[0]); + OUT_RINGf (vpt->translate[1]); + OUT_RINGf (vpt->translate[2]); + OUT_RINGf (vpt->translate[3]);*/ + BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); + OUT_RINGf (vpt->scale[0]); + OUT_RINGf (vpt->scale[1]); + OUT_RINGf (vpt->scale[2]); + OUT_RINGf (vpt->scale[3]); +} + +static void +nv10_set_vertex_buffer(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_buffer *vb) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->vtxbuf[index] = *vb; + + nv10->dirty |= NV10_NEW_ARRAYS; +} + +static void +nv10_set_vertex_element(struct pipe_context *pipe, unsigned index, + const struct pipe_vertex_element *ve) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->vtxelt[index] = *ve; + + nv10->dirty |= NV10_NEW_ARRAYS; +} + +void +nv10_init_state_functions(struct nv10_context *nv10) +{ + nv10->pipe.create_blend_state = nv10_blend_state_create; + nv10->pipe.bind_blend_state = nv10_blend_state_bind; + nv10->pipe.delete_blend_state = nv10_blend_state_delete; + + nv10->pipe.create_sampler_state = nv10_sampler_state_create; + nv10->pipe.bind_sampler_states = nv10_sampler_state_bind; + nv10->pipe.delete_sampler_state = nv10_sampler_state_delete; + nv10->pipe.set_sampler_textures = nv10_set_sampler_texture; + + nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create; + nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind; + nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete; + + nv10->pipe.create_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_create; + nv10->pipe.bind_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_bind; + nv10->pipe.delete_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_delete; + + nv10->pipe.create_vs_state = nv10_vp_state_create; + nv10->pipe.bind_vs_state = nv10_vp_state_bind; + nv10->pipe.delete_vs_state = nv10_vp_state_delete; + + nv10->pipe.create_fs_state = nv10_fp_state_create; + nv10->pipe.bind_fs_state = nv10_fp_state_bind; + nv10->pipe.delete_fs_state = nv10_fp_state_delete; + + nv10->pipe.set_blend_color = nv10_set_blend_color; + nv10->pipe.set_clip_state = nv10_set_clip_state; + nv10->pipe.set_constant_buffer = nv10_set_constant_buffer; + nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state; + nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple; + nv10->pipe.set_scissor_state = nv10_set_scissor_state; + nv10->pipe.set_viewport_state = nv10_set_viewport_state; + + nv10->pipe.set_vertex_buffer = nv10_set_vertex_buffer; + nv10->pipe.set_vertex_element = nv10_set_vertex_element; +} + diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h new file mode 100644 index 00000000000..d4b703c7adb --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_state.h @@ -0,0 +1,135 @@ +#ifndef __NV10_STATE_H__ +#define __NV10_STATE_H__ + +#include "pipe/p_state.h" + +struct nv10_blend_state { + uint32_t b_enable; + uint32_t b_srcfunc; + uint32_t b_dstfunc; + + uint32_t c_mask; + + uint32_t d_enable; +}; + +struct nv10_sampler_state { + uint32_t wrap; + uint32_t en; + uint32_t filt; + uint32_t bcol; +}; + +struct nv10_rasterizer_state { + uint32_t shade_model; + + uint32_t line_width; + uint32_t line_smooth_en; + + uint32_t point_size; + + uint32_t poly_smooth_en; + + uint32_t poly_mode_front; + uint32_t poly_mode_back; + + uint32_t front_face; + uint32_t cull_face; + uint32_t cull_face_en; + + uint32_t point_sprite; +}; + +struct nv10_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv10_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + +struct nv10_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + struct nv10_vertex_program_exec *insns; + unsigned nr_insns; + struct nv10_vertex_program_data *consts; + unsigned nr_consts; + + struct nouveau_resource *exec; + unsigned exec_start; + struct nouveau_resource *data; + unsigned data_start; + unsigned data_start_min; + + uint32_t ir; + uint32_t or; +}; + +struct nv10_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv10_fragment_program { + const struct pipe_shader_state *pipe; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv10_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + uint32_t fp_reg_control; +}; + + +struct nv10_depth_stencil_alpha_state { + struct { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; + } depth; + + struct { + uint32_t enable; + uint32_t wmask; + uint32_t func; + uint32_t ref; + uint32_t vmask; + uint32_t fail; + uint32_t zfail; + uint32_t zpass; + } stencil; + + struct { + uint32_t enabled; + uint32_t func; + uint32_t ref; + } alpha; +}; + +struct nv10_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +#endif diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c new file mode 100644 index 00000000000..1d104e2f913 --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -0,0 +1,69 @@ +#include "nv10_context.h" +#include "nv10_state.h" + +void +nv10_emit_hw_state(struct nv10_context *nv10) +{ + int i; + + if (nv10->dirty & NV10_NEW_FRAGPROG) { + nv10_fragprog_bind(nv10, nv10->fragprog.current); + /*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */ + } + + if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) { + nv10_fragtex_bind(nv10); + nv10->dirty &= ~NV10_NEW_FRAGPROG; + } + + if (nv10->dirty & NV10_NEW_VERTPROG) { + //nv10_vertprog_bind(nv10, nv10->vertprog.current); + nv10->dirty &= ~NV10_NEW_VERTPROG; + } + + if (nv10->dirty & NV10_NEW_VBO) { + + } + + nv10->dirty_samplers = 0; + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv10_vbo.c + */ + + /* Render target */ +// XXX figre out who's who for NV10TCL_DMA_* and fill accordingly +// BEGIN_RING(celsius, NV10TCL_DMA_COLOR0, 1); +// OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + if (nv10->zeta) { +// XXX +// BEGIN_RING(celsius, NV10TCL_DMA_ZETA, 1); +// OUT_RELOCo(nv10->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + /* XXX for when we allocate LMA on nv17 */ +/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_OFFSET, 1); + OUT_RELOCl(nv10->zeta+...);*/ + } + + /* Texture images */ + for (i = 0; i < 2; i++) { + if (!(nv10->fp_samplers & (1 << i))) + continue; + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 2); + OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + // XXX +/* OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0, + NV10TCL_TX_FORMAT_DMA1);*/ + } +} + diff --git a/src/gallium/drivers/nv10/nv10_surface.c b/src/gallium/drivers/nv10/nv10_surface.c new file mode 100644 index 00000000000..2e230ebbecc --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_surface.c @@ -0,0 +1,65 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv10_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/p_tile.h" + +static void +nv10_surface_copy(struct pipe_context *pipe, unsigned do_flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv10_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv10_init_surface_functions(struct nv10_context *nv10) +{ + nv10->pipe.surface_copy = nv10_surface_copy; + nv10->pipe.surface_fill = nv10_surface_fill; +} diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c new file mode 100644 index 00000000000..025ad6de4bd --- /dev/null +++ b/src/gallium/drivers/nv10/nv10_vbo.c @@ -0,0 +1,72 @@ +#include "draw/draw_context.h" +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_pushbuf.h" + +boolean nv10_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count) +{ + struct nv10_context *nv10 = nv10_context( pipe ); + struct draw_context *draw = nv10->draw; + unsigned i; + + /* + * Map vertex buffers + */ + for (i = 0; i < PIPE_ATTRIB_MAX; i++) { + if (nv10->vtxbuf[i].buffer) { + void *buf + = pipe->winsys->buffer_map(pipe->winsys, + nv10->vtxbuf[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(draw, i, buf); + } + } + /* Map index buffer, if present */ + if (indexBuffer) { + void *mapped_indexes + = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes); + } + else { + /* no index/element buffer */ + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + /* draw! */ + draw_arrays(nv10->draw, prim, start, count); + + /* + * unmap vertex/index buffers + */ + for (i = 0; i < PIPE_ATTRIB_MAX; i++) { + if (nv10->vtxbuf[i].buffer) { + pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer); + draw_set_mapped_vertex_buffer(draw, i, NULL); + } + } + if (indexBuffer) { + pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + return TRUE; +} + +boolean nv10_draw_arrays( struct pipe_context *pipe, + unsigned prim, unsigned start, unsigned count) +{ + return nv10_draw_elements(pipe, NULL, 0, prim, start, count); +} + + + diff --git a/src/gallium/winsys/dri/nouveau/Makefile b/src/gallium/winsys/dri/nouveau/Makefile index b463f218fd5..f637b2cebfc 100644 --- a/src/gallium/winsys/dri/nouveau/Makefile +++ b/src/gallium/winsys/dri/nouveau/Makefile @@ -8,6 +8,7 @@ MINIGLX_SOURCES = PIPE_DRIVERS = \ $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ + $(TOP)/src/gallium/drivers/nv10/libnv10.a \ $(TOP)/src/gallium/drivers/nv30/libnv30.a \ $(TOP)/src/gallium/drivers/nv40/libnv40.a \ $(TOP)/src/gallium/drivers/nv50/libnv50.a diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index 8b20b3689c6..336dd65847b 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -119,6 +119,9 @@ nouveau_context_create(const __GLcontextModes *glVis, } switch (nv->chipset & 0xf0) { + case 0x10: + case 0x20: + /* NV10 */ case 0x30: /* NV30 */ case 0x40: diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 64e84fb68e4..6c85aab9f5f 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -82,6 +82,11 @@ nouveau_pipe_create(struct nouveau_context *nv) return NULL; switch (nv->chipset & 0xf0) { + case 0x10: + case 0x20: + hws_create = nv10_screen_create; + hw_create = nv10_create; + break; case 0x30: hws_create = nv30_screen_create; hw_create = nv30_create; -- cgit v1.2.3 From 5e17088ee3d0ddfa8871d92d262bb5242bdd92bd Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 15 Mar 2008 16:45:49 +0100 Subject: cso_context_destroy calls bind_state functions with NULL parameter --- src/gallium/drivers/nv30/nv30_state.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 722626db1fa..80dfd9c5c0a 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -40,6 +40,10 @@ nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso) struct nv30_context *nv30 = nv30_context(pipe); struct nv30_blend_state *cb = hwcso; + if (!hwcso) { + return; + } + BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); OUT_RING (cb->d_enable); @@ -236,6 +240,10 @@ nv30_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) struct nv30_context *nv30 = nv30_context(pipe); unsigned unit; + if (!sampler) { + return; + } + for (unit = 0; unit < nr; unit++) { nv30->tex_sampler[unit] = sampler[unit]; nv30->dirty_samplers |= (1 << unit); @@ -346,6 +354,10 @@ nv30_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) struct nv30_context *nv30 = nv30_context(pipe); struct nv30_rasterizer_state *rs = hwcso; + if (!hwcso) { + return; + } + BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); OUT_RING (rs->shade_model); @@ -422,6 +434,10 @@ nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) struct nv30_context *nv30 = nv30_context(pipe); struct nv30_depth_stencil_alpha_state *hw = hwcso; + if (!hwcso) { + return; + } + BEGIN_RING(rankine, NV34TCL_DEPTH_FUNC, 3); OUT_RINGp ((uint32_t *)&hw->depth, 3); BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 16); @@ -455,6 +471,10 @@ nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso) struct nv30_context *nv30 = nv30_context(pipe); struct nv30_vertex_program *vp = hwcso; + if (!hwcso) { + return; + } + nv30->vertprog.current = vp; nv30->dirty |= NV30_NEW_VERTPROG; } @@ -487,6 +507,10 @@ nv30_fp_state_bind(struct pipe_context *pipe, void *hwcso) struct nv30_context *nv30 = nv30_context(pipe); struct nv30_fragment_program *fp = hwcso; + if (!hwcso) { + return; + } + nv30->fragprog.current = fp; nv30->dirty |= NV30_NEW_FRAGPROG; } -- cgit v1.2.3 From 7d2c63e90983088f1e2f49543caf0468aa91111f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 16 Mar 2008 12:55:02 +1100 Subject: nv10: fix build after merge --- src/gallium/drivers/nv10/nv10_state.c | 8 +++++--- src/gallium/drivers/nv10/nv10_state.h | 2 ++ 2 files changed, 7 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 313230fe889..d7c445f1b39 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -10,15 +10,15 @@ static void nv10_vertex_layout(struct pipe_context* pipe) { struct nv10_context *nv10 = nv10_context(pipe); - const struct pipe_shader_state *fs = nv10->fragprog.current->pipe; + struct nv10_fragment_program *fp = nv10->fragprog.current; uint32_t src = 0; int i; struct vertex_info vinfo; memset(&vinfo, 0, sizeof(vinfo)); - for (i = 0; i < fs->num_inputs; i++) { - switch (fs->input_semantic_name[i]) { + for (i = 0; i < fp->info.num_inputs; i++) { + switch (fp->info.input_semantic_name[i]) { case TGSI_SEMANTIC_POSITION: draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); break; @@ -476,6 +476,8 @@ nv10_fp_state_create(struct pipe_context *pipe, fp = CALLOC(1, sizeof(struct nv10_fragment_program)); fp->pipe = cso; + + tgsi_scan_shader(cso->tokens, &fp->info); return (void *)fp; } diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h index d4b703c7adb..9bda8a7d6a3 100644 --- a/src/gallium/drivers/nv10/nv10_state.h +++ b/src/gallium/drivers/nv10/nv10_state.h @@ -2,6 +2,7 @@ #define __NV10_STATE_H__ #include "pipe/p_state.h" +#include "tgsi/util/tgsi_scan.h" struct nv10_blend_state { uint32_t b_enable; @@ -77,6 +78,7 @@ struct nv10_fragment_program_data { struct nv10_fragment_program { const struct pipe_shader_state *pipe; + struct tgsi_shader_info info; boolean translated; boolean on_hw; -- cgit v1.2.3 From e1cf3f00e546f814effd25e9ccd072c941366444 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 13 Mar 2008 18:29:56 +1100 Subject: nv40: simple swtnl path (half broken, but getting there) --- src/gallium/drivers/nv40/nv40_context.c | 6 +- src/gallium/drivers/nv40/nv40_context.h | 25 +- src/gallium/drivers/nv40/nv40_draw.c | 334 +++++++++++++++++++++++-- src/gallium/drivers/nv40/nv40_fragprog.c | 3 +- src/gallium/drivers/nv40/nv40_shader.h | 2 + src/gallium/drivers/nv40/nv40_state.c | 27 +- src/gallium/drivers/nv40/nv40_state.h | 4 + src/gallium/drivers/nv40/nv40_state_clip.c | 8 +- src/gallium/drivers/nv40/nv40_state_emit.c | 130 +++++++--- src/gallium/drivers/nv40/nv40_state_viewport.c | 45 +++- src/gallium/drivers/nv40/nv40_vbo.c | 22 +- src/gallium/drivers/nv40/nv40_vertprog.c | 16 +- 12 files changed, 531 insertions(+), 91 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 203c843a013..58627443b80 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -74,8 +74,12 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) nv40_init_state_functions(nv40); nv40_init_miptree_functions(nv40); + /* Create, configure, and install fallback swtnl path */ nv40->draw = draw_create(); - assert(nv40->draw); + draw_wide_point_threshold(nv40->draw, 9999999.0); + draw_wide_line_threshold(nv40->draw, 9999999.0); + draw_enable_line_stipple(nv40->draw, FALSE); + draw_enable_point_sprites(nv40->draw, FALSE); draw_set_rasterize_stage(nv40->draw, nv40_draw_render_stage(nv40)); return &nv40->pipe; diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 100c678187a..02ca20b8014 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -116,7 +116,20 @@ struct nv40_context { /* HW state derived from pipe states */ struct nv40_state state; - unsigned fallback; + struct { + struct nv40_vertex_program *vertprog; + + unsigned nr_attribs; + unsigned hw[PIPE_MAX_SHADER_INPUTS]; + unsigned draw[PIPE_MAX_SHADER_INPUTS]; + unsigned emit[PIPE_MAX_SHADER_INPUTS]; + } swtnl; + + enum { + HW, SWTNL, SWRAST + } render_mode; + unsigned fallback_swtnl; + unsigned fallback_swrast; /* Context state */ unsigned dirty; @@ -166,6 +179,10 @@ extern void nv40_screen_init_miptree_functions(struct pipe_screen *pscreen); /* nv40_draw.c */ extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40); +extern boolean nv40_draw_elements_swtnl(struct pipe_context *pipe, + struct pipe_buffer *idxbuf, + unsigned ib_size, unsigned mode, + unsigned start, unsigned count); /* nv40_vertprog.c */ extern void nv40_vertprog_destroy(struct nv40_context *, @@ -179,8 +196,9 @@ extern void nv40_fragprog_destroy(struct nv40_context *, extern void nv40_fragtex_bind(struct nv40_context *); /* nv40_state.c and friends */ -extern void nv40_emit_hw_state(struct nv40_context *nv40); -extern void nv40_state_tex_update(struct nv40_context *nv40); +extern boolean nv40_state_validate(struct nv40_context *nv40); +extern boolean nv40_state_validate_swtnl(struct nv40_context *nv40); +extern void nv40_state_emit(struct nv40_context *nv40); extern struct nv40_state_entry nv40_state_clip; extern struct nv40_state_entry nv40_state_rasterizer; extern struct nv40_state_entry nv40_state_scissor; @@ -194,6 +212,7 @@ extern struct nv40_state_entry nv40_state_viewport; extern struct nv40_state_entry nv40_state_framebuffer; extern struct nv40_state_entry nv40_state_fragtex; extern struct nv40_state_entry nv40_state_vbo; +extern struct nv40_state_entry nv40_state_vtxfmt; /* nv40_vbo.c */ extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index a39bb85e99b..ce0e0bc6f23 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -1,62 +1,350 @@ -#include "draw/draw_private.h" #include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" + +#include "draw/draw_context.h" +#include "draw/draw_vertex.h" +#include "draw/draw_private.h" #include "nv40_context.h" +#define NV40_SHADER_NO_FUCKEDNESS +#include "nv40_shader.h" + +/* Simple, but crappy, swtnl path, hopefully we wont need to hit this very + * often at all. Uses "quadro style" vertex submission + a fixed vertex + * layout to avoid the need to generate a vertex program or vtxfmt. + */ -struct nv40_draw_stage { - struct draw_stage draw; +struct nv40_render_stage { + struct draw_stage stage; struct nv40_context *nv40; + unsigned prim; }; +static INLINE struct nv40_render_stage * +nv40_render_stage(struct draw_stage *stage) +{ + return (struct nv40_render_stage *)stage; +} + +static INLINE void +nv40_render_vertex(struct nv40_context *nv40, const struct vertex_header *v) +{ + unsigned i; + + for (i = 0; i < nv40->swtnl.nr_attribs; i++) { + unsigned idx = nv40->swtnl.draw[i]; + unsigned hw = nv40->swtnl.hw[i]; + + switch (nv40->swtnl.emit[i]) { + case EMIT_OMIT: + break; + case EMIT_1F: + BEGIN_RING(curie, 0x1e40 + (hw * 4), 1); + OUT_RING (fui(v->data[idx][0])); + break; + case EMIT_2F: + BEGIN_RING(curie, NV40TCL_VTX_ATTR_2F_X(hw), 2); + OUT_RING (fui(v->data[idx][0])); + OUT_RING (fui(v->data[idx][1])); + break; + case EMIT_3F: + BEGIN_RING(curie, NV40TCL_VTX_ATTR_3F_X(hw), 3); + OUT_RING (fui(v->data[idx][0])); + OUT_RING (fui(v->data[idx][1])); + OUT_RING (fui(v->data[idx][2])); + break; + case EMIT_4F: + BEGIN_RING(curie, NV40TCL_VTX_ATTR_4F_X(hw), 4); + OUT_RING (fui(v->data[idx][0])); + OUT_RING (fui(v->data[idx][1])); + OUT_RING (fui(v->data[idx][2])); + OUT_RING (fui(v->data[idx][3])); + break; + case EMIT_4UB: + BEGIN_RING(curie, 0x1940 + (hw * 4), 1); + OUT_RING (pack_ub4(float_to_ubyte(v->data[idx][0]), + float_to_ubyte(v->data[idx][1]), + float_to_ubyte(v->data[idx][2]), + float_to_ubyte(v->data[idx][3]))); + break; + default: + assert(0); + break; + } + } +} + +static INLINE void +nv40_render_prim(struct draw_stage *stage, struct prim_header *prim, + unsigned mode, unsigned count) +{ + struct nv40_render_stage *rs = nv40_render_stage(stage); + struct nv40_context *nv40 = rs->nv40; + struct nouveau_pushbuf *pb = nv40->nvws->channel->pushbuf; + unsigned i; + + /* Ensure there's room for 4xfloat32 + potentially 3 begin/end */ + if (pb->remaining < ((count * 20) + 6)) { + if (rs->prim != NV40TCL_BEGIN_END_STOP) { + NOUVEAU_ERR("AIII, missed flush\n"); + assert(0); + } + FIRE_RING(); + nv40_state_emit(nv40); + } + + /* Switch primitive modes if necessary */ + if (rs->prim != mode) { + if (rs->prim != NV40TCL_BEGIN_END_STOP) { + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (NV40TCL_BEGIN_END_STOP); + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (mode); + rs->prim = mode; + } + + /* Emit vertex data */ + for (i = 0; i < count; i++) + nv40_render_vertex(nv40, prim->v[i]); + + /* If it's likely we'll need to empty the push buffer soon, finish + * off the primitive now. + */ + if (pb->remaining < ((count * 20) + 6)) { + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (NV40TCL_BEGIN_END_STOP); + rs->prim = NV40TCL_BEGIN_END_STOP; + } +} + static void -nv40_draw_point(struct draw_stage *draw, struct prim_header *prim) +nv40_render_point(struct draw_stage *draw, struct prim_header *prim) { - NOUVEAU_ERR("\n"); + nv40_render_prim(draw, prim, NV40TCL_BEGIN_END_POINTS, 1); } static void -nv40_draw_line(struct draw_stage *draw, struct prim_header *prim) +nv40_render_line(struct draw_stage *draw, struct prim_header *prim) { - NOUVEAU_ERR("\n"); + nv40_render_prim(draw, prim, NV40TCL_BEGIN_END_LINES, 2); } static void -nv40_draw_tri(struct draw_stage *draw, struct prim_header *prim) +nv40_render_tri(struct draw_stage *draw, struct prim_header *prim) { - NOUVEAU_ERR("\n"); + nv40_render_prim(draw, prim, NV40TCL_BEGIN_END_TRIANGLES, 3); } static void -nv40_draw_flush(struct draw_stage *draw, unsigned flags) +nv40_render_flush(struct draw_stage *draw, unsigned flags) { + struct nv40_render_stage *rs = nv40_render_stage(draw); + struct nv40_context *nv40 = rs->nv40; + + if (rs->prim != NV40TCL_BEGIN_END_STOP) { + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (NV40TCL_BEGIN_END_STOP); + rs->prim = NV40TCL_BEGIN_END_STOP; + } } static void -nv40_draw_reset_stipple_counter(struct draw_stage *draw) +nv40_render_reset_stipple_counter(struct draw_stage *draw) { - NOUVEAU_ERR("\n"); } static void -nv40_draw_destroy(struct draw_stage *draw) +nv40_render_destroy(struct draw_stage *draw) { free(draw); } +static INLINE void +emit_mov(struct nv40_vertex_program *vp, + unsigned dst, unsigned src, unsigned vor, unsigned mask) +{ + struct nv40_vertex_program_exec *inst; + + vp->insns = realloc(vp->insns, + sizeof(struct nv40_vertex_program_exec) * + ++vp->nr_insns); + inst = &vp->insns[vp->nr_insns - 1]; + + inst->data[0] = 0x401f9c6c; + inst->data[1] = 0x0040000d | (src << 8); + inst->data[2] = 0x8106c083; + inst->data[3] = 0x6041ff80 | (dst << 2) | (mask << 13); + inst->const_index = -1; + inst->has_branch_offset = FALSE; + + vp->ir |= (1 << src); + if (vor != ~0) + vp->or |= (1 << vor); +} + +static struct nv40_vertex_program * +create_drawvp(struct nv40_context *nv40) +{ + struct nv40_vertex_program *vp = CALLOC_STRUCT(nv40_vertex_program); + unsigned i; + + emit_mov(vp, NV40_VP_INST_DEST_POS, 0, ~0, 0xf); + emit_mov(vp, NV40_VP_INST_DEST_COL0, 3, 0, 0xf); + emit_mov(vp, NV40_VP_INST_DEST_COL1, 4, 1, 0xf); + emit_mov(vp, NV40_VP_INST_DEST_BFC0, 3, 2, 0xf); + emit_mov(vp, NV40_VP_INST_DEST_BFC1, 4, 3, 0xf); + emit_mov(vp, NV40_VP_INST_DEST_FOGC, 5, 4, 0x8); + for (i = 0; i < 8; i++) + emit_mov(vp, NV40_VP_INST_DEST_TC(i), 8 + i, 14 + i, 0xf); + + vp->insns[vp->nr_insns - 1].data[3] |= 1; + vp->translated = TRUE; + return vp; +} + struct draw_stage * nv40_draw_render_stage(struct nv40_context *nv40) { - struct nv40_draw_stage *nv40draw = CALLOC_STRUCT(nv40_draw_stage); + struct nv40_render_stage *render = CALLOC_STRUCT(nv40_render_stage); + + if (!nv40->swtnl.vertprog) + nv40->swtnl.vertprog = create_drawvp(nv40); + + render->nv40 = nv40; + render->stage.draw = nv40->draw; + render->stage.point = nv40_render_point; + render->stage.line = nv40_render_line; + render->stage.tri = nv40_render_tri; + render->stage.flush = nv40_render_flush; + render->stage.reset_stipple_counter = nv40_render_reset_stipple_counter; + render->stage.destroy = nv40_render_destroy; + + return &render->stage; +} + +boolean +nv40_draw_elements_swtnl(struct pipe_context *pipe, + struct pipe_buffer *idxbuf, unsigned idxbuf_size, + unsigned mode, unsigned start, unsigned count) +{ + struct nv40_context *nv40 = nv40_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + unsigned i; + void *map; + + if (!nv40_state_validate_swtnl(nv40)) + return FALSE; + nv40_state_emit(nv40); - nv40draw->nv40 = nv40; - nv40draw->draw.draw = nv40->draw; - nv40draw->draw.point = nv40_draw_point; - nv40draw->draw.line = nv40_draw_line; - nv40draw->draw.tri = nv40_draw_tri; - nv40draw->draw.flush = nv40_draw_flush; - nv40draw->draw.reset_stipple_counter = nv40_draw_reset_stipple_counter; - nv40draw->draw.destroy = nv40_draw_destroy; + for (i = 0; i < PIPE_ATTRIB_MAX; i++) { + if (!nv40->vtxbuf[i].buffer) + continue; + map = ws->buffer_map(ws, nv40->vtxbuf[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(nv40->draw, i, map); + } - return &nv40draw->draw; + if (idxbuf) { + map = ws->buffer_map(ws, idxbuf, PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_element_buffer(nv40->draw, idxbuf_size, map); + } else { + draw_set_mapped_element_buffer(nv40->draw, 0, NULL); + } + + if (nv40->constbuf[PIPE_SHADER_VERTEX]) { + map = ws->buffer_map(ws, nv40->constbuf[PIPE_SHADER_VERTEX], + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_constant_buffer(nv40->draw, map); + } + + draw_arrays(nv40->draw, mode, start, count); + + for (i = 0; i < PIPE_ATTRIB_MAX; i++) { + if (!nv40->vtxbuf[i].buffer) + continue; + ws->buffer_unmap(ws, nv40->vtxbuf[i].buffer); + } + + if (idxbuf) + ws->buffer_unmap(ws, idxbuf); + + if (nv40->constbuf[PIPE_SHADER_VERTEX]) + ws->buffer_unmap(ws, nv40->constbuf[PIPE_SHADER_VERTEX]); + + draw_flush(nv40->draw); + pipe->flush(pipe, 0); + + return TRUE; } +static INLINE void +emit_attrib(struct nv40_context *nv40, unsigned hw, unsigned emit, + unsigned semantic, unsigned index) +{ + unsigned draw_out = draw_find_vs_output(nv40->draw, semantic, index); + unsigned a = nv40->swtnl.nr_attribs++; + + nv40->swtnl.hw[a] = hw; + nv40->swtnl.emit[a] = emit; + nv40->swtnl.draw[a] = draw_out; +} + +static boolean +nv40_state_vtxfmt_validate(struct nv40_context *nv40) +{ + struct nv40_fragment_program *fp = nv40->fragprog; + unsigned colour = 0, texcoords = 0, fog = 0, i; + + /* Determine needed fragprog inputs */ + for (i = 0; i < fp->info.num_inputs; i++) { + switch (fp->info.input_semantic_name[i]) { + case TGSI_SEMANTIC_POSITION: + break; + case TGSI_SEMANTIC_COLOR: + colour |= (1 << fp->info.input_semantic_index[i]); + break; + case TGSI_SEMANTIC_GENERIC: + texcoords |= (1 << fp->info.input_semantic_index[i]); + break; + case TGSI_SEMANTIC_FOG: + fog = 1; + break; + default: + assert(0); + } + } + + nv40->swtnl.nr_attribs = 0; + + /* Map draw vtxprog output to hw attribute IDs */ + for (i = 0; i < 2; i++) { + if (!(colour & (1 << i))) + continue; + emit_attrib(nv40, 3 + i, EMIT_4UB, TGSI_SEMANTIC_COLOR, i); + } + + for (i = 0; i < 8; i++) { + if (!(texcoords & (1 << i))) + continue; + emit_attrib(nv40, 8 + i, EMIT_4F, TGSI_SEMANTIC_GENERIC, i); + } + + if (fog) { + emit_attrib(nv40, 5, EMIT_1F, TGSI_SEMANTIC_FOG, 0); + } + + emit_attrib(nv40, 0, EMIT_4F, TGSI_SEMANTIC_POSITION, 0); + + return FALSE; +} + +struct nv40_state_entry nv40_state_vtxfmt = { + .validate = nv40_state_vtxfmt_validate, + .dirty = { + .pipe = NV40_NEW_ARRAYS | NV40_NEW_FRAGPROG, + .hw = 0 + } +}; + diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 953f9cd9082..82dbcd3eef2 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -797,9 +797,10 @@ nv40_fragprog_validate(struct nv40_context *nv40) if (fp->translated) goto update_constants; + nv40->fallback_swrast &= ~NV40_NEW_FRAGPROG; nv40_fragprog_translate(nv40, fp); if (!fp->translated) { - nv40->fallback |= NV40_FALLBACK_RAST; + nv40->fallback_swrast |= NV40_NEW_FRAGPROG; return FALSE; } diff --git a/src/gallium/drivers/nv40/nv40_shader.h b/src/gallium/drivers/nv40/nv40_shader.h index 5909c70713c..854dccf5486 100644 --- a/src/gallium/drivers/nv40/nv40_shader.h +++ b/src/gallium/drivers/nv40/nv40_shader.h @@ -476,6 +476,7 @@ # define NV40_FP_SWIZZLE_W 3 #define NV40_FP_REG_NEGATE (1 << 17) +#ifndef NV40_SHADER_NO_FUCKEDNESS #define NV40SR_NONE 0 #define NV40SR_OUTPUT 1 #define NV40SR_INPUT 2 @@ -550,5 +551,6 @@ nv40_sr_scale(struct nv40_sreg src, int scale) src.dst_scale = scale; return src; } +#endif #endif diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 321d5de0415..3eafbece309 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -3,6 +3,8 @@ #include "pipe/p_util.h" #include "pipe/p_inlines.h" +#include "draw/draw_context.h" + #include "nv40_context.h" #include "nv40_state.h" @@ -345,7 +347,7 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); break; default: - so_data(so, 0); + so_data(so, NV40TCL_CULL_FACE_BACK); break; } so_data(so, NV40TCL_FRONT_FACE_CCW); @@ -363,13 +365,13 @@ nv40_rasterizer_state_create(struct pipe_context *pipe, so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK); break; default: - so_data(so, 0); + so_data(so, NV40TCL_CULL_FACE_BACK); break; } so_data(so, NV40TCL_FRONT_FACE_CW); } so_data(so, cso->poly_smooth ? 1 : 0); - so_data(so, cso->cull_mode != PIPE_WINDING_NONE ? 1 : 0); + so_data(so, (cso->cull_mode != PIPE_WINDING_NONE) ? 1 : 0); so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1); so_data (so, cso->poly_stipple_enable ? 1 : 0); @@ -419,6 +421,9 @@ static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_rasterizer_state *rsso = hwcso; + + draw_set_rasterizer_state(nv40->draw, &rsso->pipe); nv40->rasterizer = hwcso; nv40->dirty |= NV40_NEW_RAST; @@ -508,10 +513,12 @@ static void * nv40_vp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { + struct nv40_context *nv40 = nv40_context(pipe); struct nv40_vertex_program *vp; vp = CALLOC(1, sizeof(struct nv40_vertex_program)); vp->pipe = *cso; + vp->draw = draw_create_vertex_shader(nv40->draw, &vp->pipe); return (void *)vp; } @@ -520,6 +527,9 @@ static void nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); + struct nv40_vertex_program *vp = hwcso; + + draw_bind_vertex_shader(nv40->draw, vp ? vp->draw : NULL); nv40->vertprog = hwcso; nv40->dirty |= NV40_NEW_VERTPROG; @@ -531,6 +541,7 @@ nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv40_context *nv40 = nv40_context(pipe); struct nv40_vertex_program *vp = hwcso; + draw_delete_vertex_shader(nv40->draw, vp->draw); nv40_vertprog_destroy(nv40, vp); FREE(vp); } @@ -544,6 +555,8 @@ nv40_fp_state_create(struct pipe_context *pipe, fp = CALLOC(1, sizeof(struct nv40_fragment_program)); fp->pipe = *cso; + tgsi_scan_shader(fp->pipe.tokens, &fp->info); + return (void *)fp; } @@ -582,6 +595,8 @@ nv40_set_clip_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); + draw_set_clip_state(nv40->draw, clip); + nv40->clip = *clip; nv40->dirty |= NV40_NEW_UCP; } @@ -638,6 +653,8 @@ nv40_set_viewport_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); + draw_set_viewport_state(nv40->draw, vpt); + nv40->viewport = *vpt; nv40->dirty |= NV40_NEW_VIEWPORT; } @@ -648,6 +665,8 @@ nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, { struct nv40_context *nv40 = nv40_context(pipe); + draw_set_vertex_buffer(nv40->draw, index, vb); + nv40->vtxbuf[index] = *vb; nv40->dirty |= NV40_NEW_ARRAYS; } @@ -658,6 +677,8 @@ nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, { struct nv40_context *nv40 = nv40_context(pipe); + draw_set_vertex_element(nv40->draw, index, ve); + nv40->vtxelt[index] = *ve; nv40->dirty |= NV40_NEW_ARRAYS; } diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index a02ea0c8781..ab2866eb7a5 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -2,6 +2,7 @@ #define __NV40_STATE_H__ #include "pipe/p_state.h" +#include "tgsi/util/tgsi_scan.h" struct nv40_sampler_state { uint32_t fmt; @@ -25,6 +26,8 @@ struct nv40_vertex_program_data { struct nv40_vertex_program { struct pipe_shader_state pipe; + struct draw_vertex_shader *draw; + boolean translated; struct nv40_vertex_program_exec *insns; unsigned nr_insns; @@ -49,6 +52,7 @@ struct nv40_fragment_program_data { struct nv40_fragment_program { struct pipe_shader_state pipe; + struct tgsi_shader_info info; boolean translated; unsigned samplers; diff --git a/src/gallium/drivers/nv40/nv40_state_clip.c b/src/gallium/drivers/nv40/nv40_state_clip.c index 93e690161f3..c52390f9edb 100644 --- a/src/gallium/drivers/nv40/nv40_state_clip.c +++ b/src/gallium/drivers/nv40/nv40_state_clip.c @@ -3,8 +3,12 @@ static boolean nv40_state_clip_validate(struct nv40_context *nv40) { - if (nv40->clip.nr) - nv40->fallback |= NV40_FALLBACK_TNL; + + if (nv40->render_mode == HW) { + nv40->fallback_swtnl &= ~NV40_NEW_UCP; + if (nv40->clip.nr) + nv40->fallback_swtnl |= NV40_NEW_UCP; + } return FALSE; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 9f268640e00..056238cc836 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -1,5 +1,6 @@ #include "nv40_context.h" #include "nv40_state.h" +#include "draw/draw_context.h" static struct nv40_state_entry *render_states[] = { &nv40_state_framebuffer, @@ -18,15 +19,27 @@ static struct nv40_state_entry *render_states[] = { NULL }; +static struct nv40_state_entry *swtnl_states[] = { + &nv40_state_framebuffer, + &nv40_state_rasterizer, + &nv40_state_clip, + &nv40_state_scissor, + &nv40_state_stipple, + &nv40_state_fragprog, + &nv40_state_fragtex, + &nv40_state_vertprog, + &nv40_state_blend, + &nv40_state_blend_colour, + &nv40_state_zsa, + &nv40_state_viewport, + &nv40_state_vtxfmt, + NULL +}; + static void -nv40_state_validate(struct nv40_context *nv40) +nv40_state_do_validate(struct nv40_context *nv40, + struct nv40_state_entry **states) { - struct nv40_state_entry **states = render_states; - unsigned last_fallback; - - last_fallback = nv40->fallback; - nv40->fallback = 0; - while (*states) { struct nv40_state_entry *e = *states; @@ -38,32 +51,15 @@ nv40_state_validate(struct nv40_context *nv40) states++; } nv40->dirty = 0; - - if (nv40->fallback & NV40_FALLBACK_TNL && - !(last_fallback & NV40_FALLBACK_TNL)) { - NOUVEAU_ERR("XXX: hwtnl->swtnl\n"); - } else - if (last_fallback & NV40_FALLBACK_TNL && - !(nv40->fallback & NV40_FALLBACK_TNL)) { - NOUVEAU_ERR("XXX: swtnl->hwtnl\n"); - } - - if (nv40->fallback & NV40_FALLBACK_RAST && - !(last_fallback & NV40_FALLBACK_RAST)) { - NOUVEAU_ERR("XXX: hwrast->swrast\n"); - } else - if (last_fallback & NV40_FALLBACK_RAST && - !(nv40->fallback & NV40_FALLBACK_RAST)) { - NOUVEAU_ERR("XXX: swrast->hwrast\n"); - } } -static void +void nv40_state_emit(struct nv40_context *nv40) { struct nv40_state *state = &nv40->state; struct nv40_screen *screen = nv40->screen; unsigned i, samplers; + uint64 states; if (nv40->pctx_id != screen->cur_pctx) { for (i = 0; i < NV40_STATE_MAX; i++) { @@ -74,14 +70,24 @@ nv40_state_emit(struct nv40_context *nv40) screen->cur_pctx = nv40->pctx_id; } - while (state->dirty) { - unsigned idx = ffsll(state->dirty) - 1; + for (i = 0, states = state->dirty; states; i++) { + if (!(states & (1ULL << i))) + continue; + so_ref (state->hw[i], &nv40->screen->state[i]); + so_emit(nv40->nvws, nv40->screen->state[i]); + states &= ~(1ULL << i); + } - so_ref (state->hw[idx], &nv40->screen->state[idx]); - so_emit(nv40->nvws, nv40->screen->state[idx]); - state->dirty &= ~(1ULL << idx); + if (state->dirty & ((1ULL << NV40_STATE_FRAGPROG) | + (1ULL << NV40_STATE_FRAGTEX0))) { + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (2); + BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); + OUT_RING (1); } + state->dirty = 0; + so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { if (!(samplers & (1 << i))) @@ -91,18 +97,62 @@ nv40_state_emit(struct nv40_context *nv40) samplers &= ~(1ULL << i); } so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FRAGPROG]); - so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_VTXBUF]); + if (state->hw[NV40_STATE_VTXBUF] && nv40->render_mode == HW) + so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_VTXBUF]); } -void -nv40_emit_hw_state(struct nv40_context *nv40) +boolean +nv40_state_validate(struct nv40_context *nv40) { - nv40_state_validate(nv40); - nv40_state_emit(nv40); + boolean was_sw = nv40->fallback_swtnl ? TRUE : FALSE; + + if (nv40->render_mode != HW) { + /* Don't even bother trying to go back to hw if none + * of the states that caused swtnl previously have changed. + */ + if ((nv40->fallback_swtnl & nv40->dirty) + != nv40->fallback_swtnl) + return FALSE; + + /* Attempt to go to hwtnl again */ + nv40->pipe.flush(&nv40->pipe, 0); + nv40->dirty |= (NV40_NEW_VIEWPORT | + NV40_NEW_VERTPROG | + NV40_NEW_ARRAYS | + NV40_NEW_UCP); + nv40->render_mode = HW; + } + + nv40_state_do_validate(nv40, render_states); + if (nv40->fallback_swtnl || nv40->fallback_swrast) + return FALSE; + + if (was_sw) + NOUVEAU_ERR("swtnl->hw\n"); + + return TRUE; +} + +boolean +nv40_state_validate_swtnl(struct nv40_context *nv40) +{ + /* Setup for swtnl */ + if (nv40->render_mode == HW) { + NOUVEAU_ERR("hw->swtnl 0x%08x\n", nv40->fallback_swtnl); + nv40->pipe.flush(&nv40->pipe, 0); + nv40->dirty |= (NV40_NEW_VIEWPORT | + NV40_NEW_VERTPROG | + NV40_NEW_ARRAYS | + NV40_NEW_UCP); + nv40->render_mode = SWTNL; + } + + nv40_state_do_validate(nv40, swtnl_states); + if (nv40->fallback_swrast) { + NOUVEAU_ERR("swtnl->swrast 0x%08x\n", nv40->fallback_swrast); + return FALSE; + } - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1); - OUT_RING (1); + return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c index 3a325339072..9e5c7a72a7f 100644 --- a/src/gallium/drivers/nv40/nv40_state_viewport.c +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -3,18 +3,43 @@ static boolean nv40_state_viewport_validate(struct nv40_context *nv40) { - struct nouveau_stateobj *so = so_new(9, 0); + struct nouveau_stateobj *so = so_new(11, 0); struct pipe_viewport_state *vpt = &nv40->viewport; - so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); - so_data (so, fui(vpt->translate[0])); - so_data (so, fui(vpt->translate[1])); - so_data (so, fui(vpt->translate[2])); - so_data (so, fui(vpt->translate[3])); - so_data (so, fui(vpt->scale[0])); - so_data (so, fui(vpt->scale[1])); - so_data (so, fui(vpt->scale[2])); - so_data (so, fui(vpt->scale[3])); + if (nv40->render_mode == HW) { + so_method(so, nv40->screen->curie, + NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(vpt->translate[0])); + so_data (so, fui(vpt->translate[1])); + so_data (so, fui(vpt->translate[2])); + so_data (so, fui(vpt->translate[3])); + so_data (so, fui(vpt->scale[0])); + so_data (so, fui(vpt->scale[1])); + so_data (so, fui(vpt->scale[2])); + so_data (so, fui(vpt->scale[3])); + so_method(so, nv40->screen->curie, 0x1d78, 1); + so_data (so, 1); + } else { + so_method(so, nv40->screen->curie, + NV40TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(1.0)); + so_data (so, fui(1.0)); + so_data (so, fui(1.0)); + so_data (so, fui(0.0)); + /* Not entirely certain what this is yet. The DDX uses this + * value also as it fixes rendering when you pass + * pre-transformed vertices to the GPU. My best gusss is that + * this bypasses some culling/clipping stage. Might be worth + * noting that points/lines are uneffected by whatever this + * value fixes, only filled polygons are effected. + */ + so_method(so, nv40->screen->curie, 0x1d78, 1); + so_data (so, 0x110); + } so_ref(so, &nv40->state.hw[NV40_STATE_VIEWPORT]); return TRUE; diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index f16afc23b84..fad423fdf85 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -8,6 +8,8 @@ #include "nouveau/nouveau_channel.h" #include "nouveau/nouveau_pushbuf.h" +#define FORCE_SWTNL 0 + static INLINE int nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) { @@ -165,7 +167,11 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned nr; nv40_vbo_set_idxbuf(nv40, NULL, 0); - nv40_emit_hw_state(nv40); + if (FORCE_SWTNL || !nv40_state_validate(nv40)) { + return nv40_draw_elements_swtnl(pipe, NULL, 0, + mode, start, count); + } + nv40_state_emit(nv40); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -274,7 +280,7 @@ nv40_draw_elements_inline(struct pipe_context *pipe, struct pipe_winsys *ws = pipe->winsys; void *map; - nv40_emit_hw_state(nv40); + nv40_state_emit(nv40); map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); if (!ib) { @@ -315,7 +321,7 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); unsigned nr; - nv40_emit_hw_state(nv40); + nv40_state_emit(nv40); BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (nvgl_primitive(mode)); @@ -352,8 +358,16 @@ nv40_draw_elements(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); + boolean idxbuf; + + idxbuf = nv40_vbo_set_idxbuf(nv40, indexBuffer, indexSize); + if (FORCE_SWTNL || !nv40_state_validate(nv40)) { + return nv40_draw_elements_swtnl(pipe, NULL, 0, + mode, start, count); + } + nv40_state_emit(nv40); - if (nv40_vbo_set_idxbuf(nv40, indexBuffer, indexSize)) { + if (idxbuf) { nv40_draw_elements_vbo(pipe, mode, start, count); } else { nv40_draw_elements_inline(pipe, indexBuffer, indexSize, diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 3d730c1a321..9f1ee575ce9 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -634,21 +634,29 @@ out_err: static boolean nv40_vertprog_validate(struct nv40_context *nv40) { - struct nv40_vertex_program *vp = nv40->vertprog; - struct pipe_buffer *constbuf = - nv40->constbuf[PIPE_SHADER_VERTEX]; struct nouveau_winsys *nvws = nv40->nvws; struct pipe_winsys *ws = nv40->pipe.winsys; + struct nv40_vertex_program *vp; + struct pipe_buffer *constbuf; boolean upload_code = FALSE, upload_data = FALSE; int i; + if (nv40->render_mode == HW) { + vp = nv40->vertprog; + constbuf = nv40->constbuf[PIPE_SHADER_VERTEX]; + } else { + vp = nv40->swtnl.vertprog; + constbuf = NULL; + } + /* Translate TGSI shader into hw bytecode */ if (vp->translated) goto check_gpu_resources; + nv40->fallback_swtnl &= ~NV40_NEW_VERTPROG; nv40_vertprog_translate(nv40, vp); if (!vp->translated) { - nv40->fallback |= NV40_FALLBACK_TNL; + nv40->fallback_swtnl |= NV40_NEW_VERTPROG; return FALSE; } -- cgit v1.2.3 From b2f01b0777f27a093849f299490b377ab8aab2fb Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Mon, 17 Mar 2008 03:14:11 +0100 Subject: nouveau: latest header. --- src/gallium/drivers/nouveau/nouveau_class.h | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 7392490df01..dc202086d2c 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -1784,6 +1784,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_TX_OFFSET__SIZE 0x00000002 #define NV10TCL_TX_FORMAT(x) (0x00000220+((x)*4)) #define NV10TCL_TX_FORMAT__SIZE 0x00000002 +#define NV10TCL_TX_FORMAT_DMA0 (1 << 0) +#define NV10TCL_TX_FORMAT_DMA1 (1 << 1) #define NV10TCL_TX_FORMAT_CUBE_MAP (1 << 2) #define NV10TCL_TX_FORMAT_FORMAT_SHIFT 7 #define NV10TCL_TX_FORMAT_FORMAT_MASK 0x00000780 @@ -5360,12 +5362,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50_2D_RECT_Y1 0x00000604 #define NV50_2D_RECT_X2 0x00000608 #define NV50_2D_RECT_Y2 0x0000060c -#define NV50_2D_BLIT_DST_X 0x000008b0 -#define NV50_2D_BLIT_DST_Y 0x000008b4 -#define NV50_2D_BLIT_DST_W 0x000008b8 -#define NV50_2D_BLIT_DST_H 0x000008bc -#define NV50_2D_BLIT_SRC_X 0x000008d4 -#define NV50_2D_BLIT_SRC_Y 0x000008dc #define NV50_2D_SIFC_UNK0800 0x00000800 #define NV50_2D_SIFC_FORMAT 0x00000804 #define NV50_2D_SIFC_FORMAT_32BPP 0x000000cf @@ -5384,6 +5380,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50_2D_SIFC_UNK0858 0x00000858 #define NV50_2D_SIFC_DST_Y 0x0000085c #define NV50_2D_SIFC_DATA 0x00000860 +#define NV50_2D_BLIT_DST_X 0x000008b0 +#define NV50_2D_BLIT_DST_Y 0x000008b4 +#define NV50_2D_BLIT_DST_W 0x000008b8 +#define NV50_2D_BLIT_DST_H 0x000008bc +#define NV50_2D_BLIT_SRC_X 0x000008d4 +#define NV50_2D_BLIT_SRC_Y 0x000008dc #define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 -- cgit v1.2.3 From 5c15b1276ed56064382197d73d9d357201e5f71f Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Mon, 17 Mar 2008 03:32:07 +0100 Subject: nv10: fixes. --- src/gallium/drivers/nv10/nv10_context.h | 3 ++- src/gallium/drivers/nv10/nv10_state_emit.c | 10 +++++----- 2 files changed, 7 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 8269d6121f0..386138556e5 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -47,11 +47,12 @@ struct nv10_context { uint32_t rt_enable; struct pipe_buffer *rt[4]; struct pipe_buffer *zeta; + uint32_t lma_offset; struct { struct pipe_buffer *buffer; uint32_t format; - } tex[16]; + } tex[2]; unsigned vb_enable; struct { diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index 1d104e2f913..8bf0bd2d683 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -49,21 +49,21 @@ nv10_emit_hw_state(struct nv10_context *nv10) OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); /* XXX for when we allocate LMA on nv17 */ /* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_OFFSET, 1); - OUT_RELOCl(nv10->zeta+...);*/ + OUT_RELOCl(nv10->zeta+lma_offset);*/ } /* Texture images */ for (i = 0; i < 2; i++) { if (!(nv10->fp_samplers & (1 << i))) continue; - BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 2); + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 1); OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - // XXX -/* OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, + BEGIN_RING(celsius, NV10TCL_TX_FORMAT(i), 1); + OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0, - NV10TCL_TX_FORMAT_DMA1);*/ + NV10TCL_TX_FORMAT_DMA1); } } -- cgit v1.2.3 From 767cd2ed6e97ae09526b15728495f361d5e22cb2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Mar 2008 13:49:03 +1100 Subject: nv40: workaround main swtnl breakage Not sure where the real bug is here yet, but for now this gives us correct rendering in far more cases than previously. --- src/gallium/drivers/nv40/nv40_draw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index ce0e0bc6f23..eb4f2395c48 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -335,7 +335,7 @@ nv40_state_vtxfmt_validate(struct nv40_context *nv40) emit_attrib(nv40, 5, EMIT_1F, TGSI_SEMANTIC_FOG, 0); } - emit_attrib(nv40, 0, EMIT_4F, TGSI_SEMANTIC_POSITION, 0); + emit_attrib(nv40, 0, EMIT_3F, TGSI_SEMANTIC_POSITION, 0); return FALSE; } -- cgit v1.2.3 From f93386de0fb281e79633c3bf57060f660abdfade Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Mar 2008 14:22:36 +1100 Subject: nouveau: create fence object when allocating pushbuf, instead of submit --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 3 ++- src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 4 +++- src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 14 ++++++-------- 3 files changed, 11 insertions(+), 10 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index ad587bafdff..40d4667c477 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -376,9 +376,10 @@ nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, int nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) + uint32_t flags) { struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_fence *fence = nouveau_pushbuf(chan->pushbuf)->fence; int ret; assert(bo->map == NULL); diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h index 37e404fc6cb..1c9b5799f96 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h @@ -127,6 +127,8 @@ struct nouveau_pushbuf_bo { struct nouveau_pushbuf_priv { struct nouveau_pushbuf base; + struct nouveau_fence *fence; + unsigned nop_jump; unsigned start; unsigned size; @@ -288,7 +290,7 @@ nouveau_bo_unmap(struct nouveau_bo *); extern int nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, - struct nouveau_fence *fence, uint32_t flags); + uint32_t flags); extern int nouveau_resource_init(struct nouveau_resource **heap, unsigned start, diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c index fd9a5c5a96d..a2b9321b15b 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c @@ -56,6 +56,10 @@ nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min) nvpb->base.remaining = nvpb->size; nvpb->base.cur = &nvchan->pushbuf[nvpb->start]; + /* Create a new fence object for this "frame" */ + nouveau_fence_ref(NULL, &nvpb->fence); + nouveau_fence_new(chan, &nvpb->fence); + return 0; } @@ -128,7 +132,6 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) struct nouveau_channel_priv *nvchan = nouveau_channel(chan); struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; struct nouveau_pushbuf_bo *pbbo; - struct nouveau_fence *fence = NULL; int ret; if (nvpb->base.remaining == nvpb->size) @@ -139,10 +142,6 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) nvchan->dma->free -= nvpb->size; assert(nvchan->dma->cur <= nvchan->dma->max); - ret = nouveau_fence_new(chan, &fence); - if (ret) - return ret; - nvchan->dma = &nvchan->dma_bufmgr; nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 | (nvchan->dma->base + (nvchan->dma->cur << 2)); @@ -153,7 +152,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) struct nouveau_pushbuf_reloc *r; struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; - ret = nouveau_bo_validate(chan, bo, fence, pbbo->flags); + ret = nouveau_bo_validate(chan, bo, pbbo->flags); assert (ret == 0); if (bo->offset == nouveau_bo(bo)->offset && @@ -188,9 +187,8 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) nvchan->dma = &nvchan->dma_master; /* Fence + kickoff */ - nouveau_fence_emit(fence); + nouveau_fence_emit(nvpb->fence); FIRE_RING_CH(chan); - nouveau_fence_ref(NULL, &fence); /* Allocate space for next push buffer */ ret = nouveau_pushbuf_space(chan, min); -- cgit v1.2.3 From 6b3269900101a4e4745f95028bfc0c7cfced12a8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 17 Mar 2008 23:05:46 +1100 Subject: nv40: a few more fp opcodes --- src/gallium/drivers/nv40/nv40_fragprog.c | 50 ++++++++++++++++++++++++++++++-- 1 file changed, 48 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 82dbcd3eef2..4fb28a01ea2 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -468,6 +468,34 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, case TGSI_OPCODE_COS: arith(fpc, sat, COS, dst, mask, src[0], none, none); break; + case TGSI_OPCODE_DDX: + if (mask & (MASK_Z | MASK_W)) { + tmp = temp(fpc); + arith(fpc, sat, DDX, tmp, MASK_X | MASK_Y, + swz(src[0], Z, W, Z, W), none, none); + arith(fpc, 0, MOV, tmp, MASK_Z | MASK_W, + swz(tmp, X, Y, X, Y), none, none); + arith(fpc, sat, DDX, tmp, MASK_X | MASK_Y, src[0], + none, none); + arith(fpc, 0, MOV, dst, mask, tmp, none, none); + } else { + arith(fpc, sat, DDX, dst, mask, src[0], none, none); + } + break; + case TGSI_OPCODE_DDY: + if (mask & (MASK_Z | MASK_W)) { + tmp = temp(fpc); + arith(fpc, sat, DDY, tmp, MASK_X | MASK_Y, + swz(src[0], Z, W, Z, W), none, none); + arith(fpc, 0, MOV, tmp, MASK_Z | MASK_W, + swz(tmp, X, Y, X, Y), none, none); + arith(fpc, sat, DDY, tmp, MASK_X | MASK_Y, src[0], + none, none); + arith(fpc, 0, MOV, dst, mask, tmp, none, none); + } else { + arith(fpc, sat, DDY, dst, mask, src[0], none, none); + } + break; case TGSI_OPCODE_DP3: arith(fpc, sat, DP3, dst, mask, src[0], src[1], none); break; @@ -567,15 +595,33 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, swz(src[0], X, X, X, X), none, none); } break; - case TGSI_OPCODE_SIN: - arith(fpc, sat, SIN, dst, mask, src[0], none, none); + case TGSI_OPCODE_SEQ: + arith(fpc, sat, SEQ, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SFL: + arith(fpc, sat, SFL, dst, mask, src[0], src[1], none); break; case TGSI_OPCODE_SGE: arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); break; + case TGSI_OPCODE_SGT: + arith(fpc, sat, SGT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SIN: + arith(fpc, sat, SIN, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_SLE: + arith(fpc, sat, SLE, dst, mask, src[0], src[1], none); + break; case TGSI_OPCODE_SLT: arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); break; + case TGSI_OPCODE_SNE: + arith(fpc, sat, SNE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_STR: + arith(fpc, sat, STR, dst, mask, src[0], src[1], none); + break; case TGSI_OPCODE_SUB: arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none); break; -- cgit v1.2.3 From a33da10b6c999a8ea348789ea13d2147f117a722 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 17 Mar 2008 18:13:18 +0100 Subject: nv30: only 2 render targets --- src/gallium/drivers/nv30/nv30_context.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index c63847a087c..21cffc6d274 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -48,7 +48,7 @@ struct nv30_context { unsigned vp_samplers; uint32_t rt_enable; - struct pipe_buffer *rt[4]; + struct pipe_buffer *rt[2]; struct pipe_buffer *zeta; struct { -- cgit v1.2.3 From 3394ba65b10a1ec01345c37b7888e18dcfdbe808 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 17 Mar 2008 19:03:38 +0100 Subject: nv30: another 2 rt, and set viewport tx origin, so we render at the proper place \o/ --- src/gallium/drivers/nv30/nv30_state.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 80dfd9c5c0a..b0055892aeb 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -565,7 +565,7 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv30_context *nv30 = nv30_context(pipe); - struct pipe_surface *rt[4], *zeta = NULL; + struct pipe_surface *rt[2], *zeta = NULL; uint32_t rt_enable, rt_format, w = 0, h = 0; int i, colour_format = 0, zeta_format = 0; @@ -667,6 +667,8 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); OUT_RING (((w - 1) << 16) | 0); OUT_RING (((h - 1) << 16) | 0); + BEGIN_RING(rankine, NV34TCL_VIEWPORT_TX_ORIGIN, 1); + OUT_RING (0); } static void -- cgit v1.2.3 From a514aeb77899816d82c5b31f3bf2206d82d68893 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Tue, 18 Mar 2008 13:20:59 +0100 Subject: nv30: update miptree stuff for texturing --- src/gallium/drivers/nv30/nv30_context.c | 1 + src/gallium/drivers/nv30/nv30_context.h | 4 +++- src/gallium/drivers/nv30/nv30_miptree.c | 22 ++++++++++++++++------ src/gallium/drivers/nv30/nv30_screen.c | 3 ++- 4 files changed, 22 insertions(+), 8 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index b3906e28e3f..cdd662a9f19 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -234,6 +234,7 @@ nv30_create(struct pipe_screen *screen, unsigned pctx_id) nv30_init_query_functions(nv30); nv30_init_surface_functions(nv30); nv30_init_state_functions(nv30); + nv30_init_miptree_functions(nv30); nv30->draw = draw_create(); assert(nv30->draw); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 21cffc6d274..0ee6cfdb339 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -91,9 +91,11 @@ nv30_context(struct pipe_context *pipe) extern void nv30_init_state_functions(struct nv30_context *nv30); extern void nv30_init_surface_functions(struct nv30_context *nv30); -extern void nv30_init_miptree_functions(struct pipe_screen *screen); +extern void nv30_init_miptree_functions(struct nv30_context *nv30); extern void nv30_init_query_functions(struct nv30_context *nv30); +extern void nv30_screen_init_miptree_functions(struct pipe_screen *pscreen); + /* nv30_draw.c */ extern struct draw_stage *nv30_draw_render_stage(struct nv30_context *nv30); diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 19945e9ab87..f5659353ea5 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -55,7 +55,7 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) } static struct pipe_texture * -nv30_miptree_create(struct pipe_screen *screen, struct pipe_texture *pt) +nv30_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) { struct pipe_winsys *ws = screen->winsys; struct nv30_miptree *mt; @@ -99,6 +99,11 @@ nv30_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) } } +static void +nv30_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) +{ +} + static struct pipe_surface * nv30_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) @@ -128,13 +133,18 @@ nv30_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, return ps; } + void -nv30_init_miptree_functions(struct pipe_screen *screen) +nv30_init_miptree_functions(struct nv30_context *nv30) { - struct nv30_screen *nv30screen = nv30_screen(screen); + nv30->pipe.texture_update = nv30_miptree_update; +} - nv30screen->screen.texture_create = nv30_miptree_create; - nv30screen->screen.texture_release = nv30_miptree_release; - nv30screen->screen.get_tex_surface = nv30_miptree_surface_get; +void +nv30_screen_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv30_miptree_create; + pscreen->texture_release = nv30_miptree_release; + pscreen->get_tex_surface = nv30_miptree_surface_get; } diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 39f2ac1af5c..3ca50e4fbf4 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -147,7 +147,8 @@ nv30_screen_create(struct pipe_winsys *winsys, struct nouveau_winsys *nvws, nv30screen->screen.is_format_supported = nv30_screen_is_format_supported; - nv30_init_miptree_functions(&nv30screen->screen); + nv30_screen_init_miptree_functions(&nv30screen->screen); + return &nv30screen->screen; } -- cgit v1.2.3 From e4b5008de46659f13fca7dcd1b587504e9d4484c Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Tue, 18 Mar 2008 13:34:26 +0100 Subject: nv30: hmm, no buffer for texture --- src/gallium/drivers/nv30/nv30_state_emit.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 70b98836f0f..2cb4692e176 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -64,6 +64,8 @@ nv30_emit_hw_state(struct nv30_context *nv30) for (i = 0; i < 16; i++) { if (!(nv30->fp_samplers & (1 << i))) continue; + if (!nv30->tex[i].buffer) + continue; BEGIN_RING(rankine, NV34TCL_TX_OFFSET(i), 2); OUT_RELOCl(nv30->tex[i].buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); -- cgit v1.2.3 From 75e714d476eb573a6e06585341ce693434eb237e Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Tue, 18 Mar 2008 21:50:01 +0100 Subject: nv30: disable emit texture image in state_emit, already done in fragtex --- src/gallium/drivers/nv30/nv30_context.h | 4 ++-- src/gallium/drivers/nv30/nv30_state_emit.c | 6 +++--- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 0ee6cfdb339..c1cc3eca1e3 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -51,10 +51,10 @@ struct nv30_context { struct pipe_buffer *rt[2]; struct pipe_buffer *zeta; - struct { + /*struct { struct pipe_buffer *buffer; uint32_t format; - } tex[16]; + } tex[16];*/ unsigned vb_enable; struct { diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 2cb4692e176..eda2fd45f56 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -60,12 +60,11 @@ nv30_emit_hw_state(struct nv30_context *nv30) OUT_RING(0);*/ } - /* Texture images */ + /* Texture images, emitted in nv30_fragtex_build */ +#if 0 for (i = 0; i < 16; i++) { if (!(nv30->fp_samplers & (1 << i))) continue; - if (!nv30->tex[i].buffer) - continue; BEGIN_RING(rankine, NV34TCL_TX_OFFSET(i), 2); OUT_RELOCl(nv30->tex[i].buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); @@ -74,6 +73,7 @@ nv30_emit_hw_state(struct nv30_context *nv30) NOUVEAU_BO_OR, NV34TCL_TX_FORMAT_DMA0, NV34TCL_TX_FORMAT_DMA1); } +#endif /* Fragment program */ BEGIN_RING(rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); -- cgit v1.2.3 From 9575e35807c89c0b8a745671bc2dcd54d96379ff Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 19 Mar 2008 00:06:15 +1100 Subject: nouveau: pass nvws to nvws->push_* functions, rather than nouveau_channel* --- src/gallium/drivers/nouveau/nouveau_push.h | 7 ++- src/gallium/drivers/nouveau/nouveau_stateobj.h | 10 ++-- src/gallium/drivers/nouveau/nouveau_winsys.h | 4 +- src/gallium/drivers/nv40/nv40_dma.h | 66 ------------------------- src/gallium/drivers/nv40/nv40_screen.c | 2 +- src/gallium/drivers/nv50/nv50_screen.c | 2 +- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 15 ++++-- 7 files changed, 23 insertions(+), 83 deletions(-) delete mode 100644 src/gallium/drivers/nv40/nv40_dma.h diff --git a/src/gallium/drivers/nouveau/nouveau_push.h b/src/gallium/drivers/nouveau/nouveau_push.h index c5c5d988d58..225c17744ad 100644 --- a/src/gallium/drivers/nouveau/nouveau_push.h +++ b/src/gallium/drivers/nouveau/nouveau_push.h @@ -27,7 +27,7 @@ #define BEGIN_RING(obj,mthd,size) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + pc->nvws->push_flush(pc->nvws, ((size) + 1)); \ OUT_RING((pc->obj->subc << 13) | ((size) << 18) | (mthd)); \ pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ } while(0) @@ -38,13 +38,12 @@ #define FIRE_RING() do { \ NOUVEAU_PUSH_CONTEXT(pc); \ - pc->nvws->push_flush(pc->nvws->channel, 0); \ + pc->nvws->push_flush(pc->nvws, 0); \ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ - pc->nvws->push_reloc(pc->nvws->channel, \ - pc->nvws->channel->pushbuf->cur++, \ + pc->nvws->push_reloc(pc->nvws, pc->nvws->channel->pushbuf->cur++, \ (bo), (data), (flags), (vor), (tor)); \ } while(0) diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index 439c7e4734a..d465223748a 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -99,14 +99,14 @@ so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) nr = so->cur - so->push; if (pb->remaining < nr) - nvws->push_flush(nvws->channel, nr); + nvws->push_flush(nvws, nr); pb->remaining -= nr; memcpy(pb->cur, so->push, nr * 4); for (i = 0; i < so->cur_reloc; i++) { struct nouveau_stateobj_reloc *r = &so->reloc[i]; - nvws->push_reloc(nvws->channel, pb->cur + r->offset, r->bo, + nvws->push_reloc(nvws, pb->cur + r->offset, r->bo, r->data, r->flags, r->vor, r->tor); } pb->cur += nr; @@ -120,17 +120,17 @@ so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) i = so->cur_reloc << 1; if (nvws->channel->pushbuf->remaining < i) - nvws->push_flush(nvws->channel, i); + nvws->push_flush(nvws, i); nvws->channel->pushbuf->remaining -= i; for (i = 0; i < so->cur_reloc; i++) { struct nouveau_stateobj_reloc *r = &so->reloc[i]; - nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->packet, + nvws->push_reloc(nvws, pb->cur++, r->bo, r->packet, (r->flags & (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART)) | NOUVEAU_BO_DUMMY, 0, 0); - nvws->push_reloc(nvws->channel, pb->cur++, r->bo, r->data, + nvws->push_reloc(nvws, pb->cur++, r->bo, r->data, r->flags | NOUVEAU_BO_DUMMY, r->vor, r->tor); } } diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index e4b20478a0c..7fa7cc0910c 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -24,10 +24,10 @@ struct nouveau_winsys { struct nouveau_resource **); void (*res_free)(struct nouveau_resource **); - int (*push_reloc)(struct nouveau_channel *, void *ptr, + int (*push_reloc)(struct nouveau_winsys *, void *ptr, struct pipe_buffer *, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor); - int (*push_flush)(struct nouveau_channel *, unsigned size); + int (*push_flush)(struct nouveau_winsys *, unsigned size); int (*grobj_alloc)(struct nouveau_winsys *, int grclass, struct nouveau_grobj **); diff --git a/src/gallium/drivers/nv40/nv40_dma.h b/src/gallium/drivers/nv40/nv40_dma.h deleted file mode 100644 index 1fb82677689..00000000000 --- a/src/gallium/drivers/nv40/nv40_dma.h +++ /dev/null @@ -1,66 +0,0 @@ -#ifndef __NV40_DMA_H__ -#define __NV40_DMA_H__ - -#include "pipe/nouveau/nouveau_winsys.h" - -#define OUT_RING(data) do { \ - (*nv40->nvws->channel->pushbuf->cur++) = (data); \ -} while(0) - -#define OUT_RINGp(src,size) do { \ - memcpy(nv40->nvws->channel->pushbuf->cur, (src), (size) * 4); \ - nv40->nvws->channel->pushbuf->cur += (size); \ -} while(0) - -#define OUT_RINGf(data) do { \ - union { float v; uint32_t u; } c; \ - c.v = (data); \ - OUT_RING(c.u); \ -} while(0) - -#define BEGIN_RING(obj,mthd,size) do { \ - if (nv40->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - nv40->nvws->push_flush(nv40->nvws->channel, ((size) + 1)); \ - OUT_RING((nv40->obj->subc << 13) | ((size) << 18) | (mthd)); \ - nv40->nvws->channel->pushbuf->remaining -= ((size) + 1); \ -} while(0) - -#define BEGIN_RING_NI(obj,mthd,size) do { \ - BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ -} while(0) - -#define FIRE_RING() do { \ - nv40->nvws->push_flush(nv40->nvws->channel, 0); \ -} while(0) - -#define OUT_RELOC(bo,data,flags,vor,tor) do { \ - nv40->nvws->push_reloc(nv40->nvws->channel, \ - nv40->nvws->channel->pushbuf->cur, \ - (struct nouveau_bo *)(bo), \ - (data), (flags), (vor), (tor)); \ - OUT_RING(0); \ -} while(0) - -/* Raw data + flags depending on FB/TT buffer */ -#define OUT_RELOCd(bo,data,flags,vor,tor) do { \ - OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor)); \ -} while(0) - -/* FB/TT object handle */ -#define OUT_RELOCo(bo,flags) do { \ - OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR, \ - nv40->nvws->channel->vram->handle, \ - nv40->nvws->channel->gart->handle); \ -} while(0) - -/* Low 32-bits of offset */ -#define OUT_RELOCl(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0); \ -} while(0) - -/* High 32-bits of offset */ -#define OUT_RELOCh(bo,delta,flags) do { \ - OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0); \ -} while(0) - -#endif diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 268ca83ce0d..bf30fbeca19 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -260,7 +260,7 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, so_emit(nvws, so); so_ref(NULL, &so); - nvws->push_flush(nvws->channel, 0); + nvws->push_flush(nvws, 0); screen->pipe.winsys = ws; screen->pipe.destroy = nv40_screen_destroy; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 721c6421d1f..ff4aca81a5a 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -171,7 +171,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, so_emit(nvws, so); so_ref(NULL, &so); - nvws->push_flush(nvws->channel, 0); + nvws->push_flush(nvws, 0); screen->pipe.winsys = ws; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 6c85aab9f5f..bf1afce5d9d 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -59,14 +59,21 @@ nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, } static int -nouveau_pipe_emit_reloc(struct nouveau_channel *chan, void *ptr, +nouveau_pipe_push_reloc(struct nouveau_winsys *nvws, void *ptr, struct pipe_buffer *buf, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor) { - return nouveau_pushbuf_emit_reloc(chan, ptr, nouveau_buffer(buf)->bo, + return nouveau_pushbuf_emit_reloc(nvws->channel, ptr, + nouveau_buffer(buf)->bo, data, flags, vor, tor); } +static int +nouveau_pipe_push_flush(struct nouveau_winsys *nvws, unsigned size) +{ + return nouveau_pushbuf_flush(nvws->channel, size); +} + struct pipe_context * nouveau_pipe_create(struct nouveau_context *nv) { @@ -114,8 +121,8 @@ nouveau_pipe_create(struct nouveau_context *nv) nvws->res_alloc = nouveau_resource_alloc; nvws->res_free = nouveau_resource_free; - nvws->push_reloc = nouveau_pipe_emit_reloc; - nvws->push_flush = nouveau_pushbuf_flush; + nvws->push_reloc = nouveau_pipe_push_reloc; + nvws->push_flush = nouveau_pipe_push_flush; nvws->grobj_alloc = nouveau_pipe_grobj_alloc; nvws->grobj_free = nouveau_grobj_free; -- cgit v1.2.3 From 176df85568992a5d99aab7f0b1e382d41459aa13 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 19 Mar 2008 21:52:19 +1100 Subject: nv40: "implement" noise opcodes in fragprog For the moment, we do what NVIDIA does and return 0 unconditionally. This isn't correct, but it's an implementation at least. --- src/gallium/drivers/nv40/nv40_fragprog.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 4fb28a01ea2..33aac37d569 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -554,6 +554,12 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, case TGSI_OPCODE_MUL: arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); break; + case TGSI_OPCODE_NOISE1: + case TGSI_OPCODE_NOISE2: + case TGSI_OPCODE_NOISE3: + case TGSI_OPCODE_NOISE4: + arith(fpc, sat, SFL, dst, mask, none, none, none); + break; case TGSI_OPCODE_POW: tmp = temp(fpc); arith(fpc, 0, LG2, tmp, MASK_X, -- cgit v1.2.3 From 1d9049c4df24d47446218f4032b891b817af0d00 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 00:18:18 +1100 Subject: nouveau: match gallium API changes --- src/gallium/drivers/nv30/nv30_miptree.c | 3 ++- src/gallium/drivers/nv40/nv40_miptree.c | 3 ++- src/gallium/drivers/nv50/nv50_miptree.c | 3 ++- 3 files changed, 6 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index f5659353ea5..6fdcf42a243 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -100,7 +100,8 @@ nv30_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) } static void -nv30_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) +nv30_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, + uint face, uint levels) { } diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 94ba05b7104..0dff9b3ad6d 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -99,7 +99,8 @@ nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) } static void -nv40_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) +nv40_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, + uint face, uint levels) { } diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 720d33fda91..7474c657650 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -35,7 +35,8 @@ nv50_screen_init_miptree_functions(struct pipe_screen *pscreen) } static void -nv50_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt) +nv50_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, + uint face, uint levels) { } -- cgit v1.2.3 From 95e8cad9a38181052790b34837daa6717e0c5171 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 12:57:05 +1100 Subject: nv40: use new pipe_framebuffer width/height fields --- src/gallium/drivers/nv40/nv40_state_fb.c | 16 +++------------- 1 file changed, 3 insertions(+), 13 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index 71795ab1824..107b4400283 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -5,10 +5,12 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) { struct pipe_framebuffer_state *fb = &nv40->framebuffer; struct pipe_surface *rt[4], *zeta; - uint32_t rt_enable, rt_format, w, h; + uint32_t rt_enable, rt_format; int i, colour_format = 0, zeta_format = 0; struct nouveau_stateobj *so = so_new(64, 10); unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; + unsigned w = fb->width; + unsigned h = fb->height; rt_enable = 0; for (i = 0; i < 4; i++) { @@ -16,12 +18,8 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) continue; if (colour_format) { - assert(w == fb->cbufs[i]->width); - assert(h == fb->cbufs[i]->height); assert(colour_format == fb->cbufs[i]->format); } else { - w = fb->cbufs[i]->width; - h = fb->cbufs[i]->height; colour_format = fb->cbufs[i]->format; rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i); rt[i] = fb->cbufs[i]; @@ -33,14 +31,6 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) rt_enable |= NV40TCL_RT_ENABLE_MRT; if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - zeta_format = fb->zsbuf->format; zeta = fb->zsbuf; } -- cgit v1.2.3 From 5d2577e576635559da202d0d062601e404843b2c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 17:38:24 +1100 Subject: nouveau: try combining prev and next resources on free Fixes some cases where we end up with a list of many unused chunks that are too small to be useful. --- src/gallium/winsys/dri/nouveau/nouveau_resource.c | 29 +++++++++++++---------- 1 file changed, 17 insertions(+), 12 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_resource.c b/src/gallium/winsys/dri/nouveau/nouveau_resource.c index 5d9d578b4fe..3bbcb5c45e0 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_resource.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_resource.c @@ -88,6 +88,22 @@ nouveau_resource_free(struct nouveau_resource **res) if (!res || !*res) return; r = *res; + *res = NULL; + + r->in_use = 0; + + if (r->next && !r->next->in_use) { + struct nouveau_resource *new = r->next; + + new->prev = r->prev; + if (r->prev) + r->prev->next = new; + new->size += r->size; + new->start = r->start; + + free(r); + r = new; + } if (r->prev && !r->prev->in_use) { r->prev->next = r->next; @@ -95,17 +111,6 @@ nouveau_resource_free(struct nouveau_resource **res) r->next->prev = r->prev; r->prev->size += r->size; free(r); - } else - if (r->next && !r->next->in_use) { - r->next->prev = r->prev; - if (r->prev) - r->prev->next = r->next; - r->next->size += r->size; - r->next->start = r->start; - free(r); - } else { - r->in_use = 0; } - - *res = NULL; + } -- cgit v1.2.3 From 735ecf5de8587f8e52fe6ffc7f3f17403d140e23 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 17:40:01 +1100 Subject: nouveau: potentially delay buffer destruction Rather than spinning waiting for the GPU to finish with the buffer, add a callback on the buffer's fence object instead. Gives a sizable performance increase in OA. --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index 40d4667c477..5dbb7d03744 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -206,6 +206,18 @@ nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, return 0; } +static void +nouveau_bo_del_cb(void *priv) +{ + struct nouveau_bo_priv *nvbo = priv; + + nouveau_fence_ref(NULL, &nvbo->fence); + nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem && !nvbo->user) + free(nvbo->sysmem); + free(nvbo); +} + void nouveau_bo_del(struct nouveau_bo **bo) { @@ -220,11 +232,9 @@ nouveau_bo_del(struct nouveau_bo **bo) return; if (nvbo->fence) - nouveau_fence_wait(&nvbo->fence); - nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); - if (nvbo->sysmem && !nvbo->user) - free(nvbo->sysmem); - free(nvbo); + nouveau_fence_signal_cb(nvbo->fence, nouveau_bo_del_cb, nvbo); + else + nouveau_bo_del_cb(nvbo); } int -- cgit v1.2.3 From 83e94189c424303fee4218a9d9380fc73b7ba318 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 18:15:58 +1100 Subject: nouveau: use saved fence pointer, not head of unsignalled list It's possible for the unsignalled list head to change during fence_flush. Fixes valgrind complaint exposed by a previous commit. --- src/gallium/winsys/dri/nouveau/nouveau_fence.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_fence.c b/src/gallium/winsys/dri/nouveau/nouveau_fence.c index 7714e6f2485..e7b0b4ff079 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_fence.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_fence.c @@ -172,14 +172,13 @@ nouveau_fence_flush(struct nouveau_channel *chan) nvfence = nouveau_fence(nvchan->fence_head); if (nvfence->sequence > sequence) break; - nouveau_fence_del_unsignalled(&nvfence->base); nvfence->signalled = 1; if (nvfence->signal_cb) { struct nouveau_fence *fence = NULL; - nouveau_fence_ref(nvchan->fence_head, &fence); + nouveau_fence_ref(&nvfence->base, &fence); while (nvfence->signal_cb) { struct nouveau_fence_cb *cb; -- cgit v1.2.3 From 1a4dcde808dbd5daa58f939361d9a9b539f81b50 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 18:28:07 +1100 Subject: nouveau: do a retry if initial buffer alloc fails --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index 5dbb7d03744..46c0759dbb4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -369,8 +369,13 @@ nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, int ret; ret = nouveau_bo_set_status(bo, flags); - if (ret) - return ret; + if (ret) { + nouveau_fence_flush(chan); + + ret = nouveau_bo_set_status(bo, flags); + if (ret) + return ret; + } if (nvbo->user) nouveau_bo_upload(nvbo); -- cgit v1.2.3 From ebde8d3a1276f5c72d39936efabe72b5325f8e98 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 21:17:57 +1100 Subject: nv40: use num_cbufs --- src/gallium/drivers/nv40/nv40_state_fb.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index 107b4400283..93b79377c92 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -13,10 +13,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) unsigned h = fb->height; rt_enable = 0; - for (i = 0; i < 4; i++) { - if (!fb->cbufs[i]) - continue; - + for (i = 0; i < fb->num_cbufs; i++) { if (colour_format) { assert(colour_format == fb->cbufs[i]->format); } else { -- cgit v1.2.3 From 32162871396f65e8afdd90c602b1ccd01233c2e2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 21:58:17 +1100 Subject: nv40: align each level to 64 pixels --- src/gallium/drivers/nv40/nv40_miptree.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 0dff9b3ad6d..01b3952fe66 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -12,7 +12,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) boolean swizzled = FALSE; uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; uint offset = 0; - int nr_faces, l, f; + int nr_faces, l, f, pitch; if (pt->target == PIPE_TEXTURE_CUBE) { nr_faces = 6; @@ -22,18 +22,18 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) } else { nr_faces = 1; } - + + pitch = pt->width[0]; for (l = 0; l <= pt->last_level; l++) { pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; if (swizzled) - nv40mt->level[l].pitch = pt->width[l] * pt->cpp; - else - nv40mt->level[l].pitch = pt->width[0] * pt->cpp; - nv40mt->level[l].pitch = (nv40mt->level[l].pitch + 63) & ~63; + pitch = pt->width[l]; + pitch = (pitch + 63) & ~63; + nv40mt->level[l].pitch = pitch * pt->cpp; nv40mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); @@ -49,6 +49,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) offset += nv40mt->level[l].pitch * pt->height[l]; } } + NOUVEAU_ERR("\n"); nv40mt->total_size = offset; } -- cgit v1.2.3 From 46b8dd9c16755e97ae547c0a1823e338f7a7c791 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 22:01:22 +1100 Subject: nv40: s/free/FREE/ --- src/gallium/drivers/nv40/nv40_context.c | 2 +- src/gallium/drivers/nv40/nv40_draw.c | 2 +- src/gallium/drivers/nv40/nv40_fragprog.c | 4 ++-- src/gallium/drivers/nv40/nv40_miptree.c | 6 +++--- src/gallium/drivers/nv40/nv40_query.c | 2 +- src/gallium/drivers/nv40/nv40_vertprog.c | 8 ++++---- 6 files changed, 12 insertions(+), 12 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 58627443b80..7fcf8b86197 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -40,7 +40,7 @@ nv40_destroy(struct pipe_context *pipe) if (nv40->draw) draw_destroy(nv40->draw); - free(nv40); + FREE(nv40); } struct pipe_context * diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index eb4f2395c48..7c5e0df5b85 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -158,7 +158,7 @@ nv40_render_reset_stipple_counter(struct draw_stage *draw) static void nv40_render_destroy(struct draw_stage *draw) { - free(draw); + FREE(draw); } static INLINE void diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 33aac37d569..2d82f86ef5c 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -801,7 +801,7 @@ nv40_fragprog_translate(struct nv40_context *nv40, fp->translated = TRUE; out_err: tgsi_parse_free(&parse); - free(fpc); + FREE(fpc); } static void @@ -903,7 +903,7 @@ nv40_fragprog_destroy(struct nv40_context *nv40, struct nv40_fragment_program *fp) { if (fp->insn_len) - free(fp->insn); + FREE(fp->insn); } struct nv40_state_entry nv40_state_fragprog = { diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 01b3952fe66..215b91fcd08 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -72,7 +72,7 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, mt->total_size); if (!mt->buffer) { - free(mt); + FREE(mt); return NULL; } @@ -93,9 +93,9 @@ nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) pipe_buffer_reference(ws, &nv40mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv40mt->level[l].image_offset) - free(nv40mt->level[l].image_offset); + FREE(nv40mt->level[l].image_offset); } - free(nv40mt); + FREE(nv40mt); } } diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c index 03178456246..a0a30724066 100644 --- a/src/gallium/drivers/nv40/nv40_query.c +++ b/src/gallium/drivers/nv40/nv40_query.c @@ -34,7 +34,7 @@ nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) if (q->object) nv40->nvws->res_free(&q->object); - free(q); + FREE(q); } static void diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 9f1ee575ce9..385c8aa0780 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -572,7 +572,7 @@ nv40_vertprog_translate(struct nv40_context *nv40, vpc->high_temp = -1; if (!nv40_vertprog_prepare(vpc)) { - free(vpc); + FREE(vpc); return; } @@ -628,7 +628,7 @@ nv40_vertprog_translate(struct nv40_context *nv40, vp->translated = TRUE; out_err: tgsi_parse_free(&parse); - free(vpc); + FREE(vpc); } static boolean @@ -805,9 +805,9 @@ void nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp) { if (vp->nr_consts) - free(vp->consts); + FREE(vp->consts); if (vp->nr_insns) - free(vp->insns); + FREE(vp->insns); } struct nv40_state_entry nv40_state_vertprog = { -- cgit v1.2.3 From 0c91f5991dee7827ea915214a2a6973c2d6a7257 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 22:02:49 +1100 Subject: nv40: oops --- src/gallium/drivers/nv40/nv40_miptree.c | 1 - 1 file changed, 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 215b91fcd08..502edc16298 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -49,7 +49,6 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) offset += nv40mt->level[l].pitch * pt->height[l]; } } - NOUVEAU_ERR("\n"); nv40mt->total_size = offset; } -- cgit v1.2.3 From 308d7b171179f40b767b6590f71f969473ade25c Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 22:09:29 +1100 Subject: nv40: swtnl fallback on unsupported array format --- src/gallium/drivers/nv40/nv40_vbo.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index fad423fdf85..4a5b8119390 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -423,8 +423,12 @@ nv40_vbo_validate(struct nv40_context *nv40) continue; } - if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp)) - assert(0); + if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp)) { + nv40->fallback_swtnl |= NV40_NEW_ARRAYS; + so_ref(NULL, &vtxbuf); + so_ref(NULL, &vtxfmt); + return FALSE; + } so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, -- cgit v1.2.3 From 19b57690ad251e2b0714abe3a20893722bd99d54 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 22:23:19 +1100 Subject: nv40: call semi-magic 0x1d88 method --- src/gallium/drivers/nv40/nv40_state_fb.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index 93b79377c92..88baf61ffb3 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -128,6 +128,8 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2); so_data (so, ((w - 1) << 16) | 0); so_data (so, ((h - 1) << 16) | 0); + so_method(so, nv40->screen->curie, 0x1d88, 1); + so_data (so, (1 << 12) | h); so_ref(so, &nv40->state.hw[NV40_STATE_FB]); return TRUE; -- cgit v1.2.3 From 3a0dd2e6d162fad6f98f337ee4f6b5dada1e37f3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 23:03:25 +1100 Subject: nv40: ensure vb relocs don't end up on list for swtnl Avoids bo code bailing out because of mapped buffers being validated. --- src/gallium/drivers/nv40/nv40_draw.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index 7c5e0df5b85..d05e5ad1936 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -236,6 +236,7 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe, if (!nv40_state_validate_swtnl(nv40)) return FALSE; + nv40->dirty &= ~(1ULL << NV40_STATE_VTXBUF); nv40_state_emit(nv40); for (i = 0; i < PIPE_ATTRIB_MAX; i++) { -- cgit v1.2.3 From 3be8785e08128bc2821c0cdff97f7adbb46c745b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 23:09:01 +1100 Subject: nv40: add 16-bit SSCALED vb formats --- src/gallium/drivers/nv40/nv40_vbo.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 4a5b8119390..4fae10f74b4 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -28,6 +28,12 @@ nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) case PIPE_FORMAT_R8G8B8A8_UNORM: *fmt = NV40TCL_VTXFMT_TYPE_UBYTE; break; + case PIPE_FORMAT_R16_SSCALED: + case PIPE_FORMAT_R16G16_SSCALED: + case PIPE_FORMAT_R16G16B16_SSCALED: + case PIPE_FORMAT_R16G16B16A16_SSCALED: + *fmt = 5; + break; default: pf_sprint_name(fs, pipe); NOUVEAU_ERR("Unknown format %s\n", fs); @@ -37,18 +43,22 @@ nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) switch (pipe) { case PIPE_FORMAT_R8_UNORM: case PIPE_FORMAT_R32_FLOAT: + case PIPE_FORMAT_R16_SSCALED: *ncomp = 1; break; case PIPE_FORMAT_R8G8_UNORM: case PIPE_FORMAT_R32G32_FLOAT: + case PIPE_FORMAT_R16G16_SSCALED: *ncomp = 2; break; case PIPE_FORMAT_R8G8B8_UNORM: case PIPE_FORMAT_R32G32B32_FLOAT: + case PIPE_FORMAT_R16G16B16_SSCALED: *ncomp = 3; break; case PIPE_FORMAT_R8G8B8A8_UNORM: case PIPE_FORMAT_R32G32B32A32_FLOAT: + case PIPE_FORMAT_R16G16B16A16_SSCALED: *ncomp = 4; break; default: -- cgit v1.2.3 From 75b85fd33abe143d9cca6f8405f0a4243b6a5ddb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 21 Mar 2008 23:24:16 +1100 Subject: nv40: fix bug in query code --- src/gallium/drivers/nv40/nv40_query.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c index a0a30724066..15961591b96 100644 --- a/src/gallium/drivers/nv40/nv40_query.c +++ b/src/gallium/drivers/nv40/nv40_query.c @@ -45,6 +45,15 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq) assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); + /* Happens when end_query() is called, then another begin_query() + * without querying the result in-between. For now we'll wait for + * the existing query to notify completion, but it could be better. + */ + if (q->object) { + uint64 tmp; + pipe->get_query_result(pipe, pq, 1, &tmp); + } + if (nv40->nvws->res_alloc(nv40->screen->query_heap, 1, NULL, &q->object)) assert(0); nv40->nvws->notifier_reset(nv40->screen->query, q->object->start); -- cgit v1.2.3 From 62a8e7685f0567052f50a2b9aaa64054e5dfa0c6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 14:29:35 +1100 Subject: nv40: add dxtn formats (disabled) --- src/gallium/drivers/nv40/nv40_fragtex.c | 8 ++++---- src/gallium/drivers/nv40/nv40_screen.c | 6 ++++++ 2 files changed, 10 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 436f954ceca..980705f94ad 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -30,10 +30,10 @@ nv40_texture_formats[] = { _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), -// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W), + _(DXT1_RGBA , DXT1 , S1, S1, S1, S1, X, Y, Z, W), + _(DXT3_RGBA , DXT3 , S1, S1, S1, S1, X, Y, Z, W), + _(DXT5_RGBA , DXT5 , S1, S1, S1, S1, X, Y, Z, W), {}, }; diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index bf30fbeca19..75b965bb9d1 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -110,6 +110,12 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_U_A8_L8: case PIPE_FORMAT_Z16_UNORM: case PIPE_FORMAT_Z24S8_UNORM: +#if 0 /* state tracker not up to the task just yet. */ + case PIPE_FORMAT_DXT1_RGB: + case PIPE_FORMAT_DXT1_RGBA: + case PIPE_FORMAT_DXT3_RGBA: + case PIPE_FORMAT_DXT5_RGBA: +#endif return TRUE; default: break; -- cgit v1.2.3 From 01cb2cd93efe7ad94d7fd36aa5a776c2e3ab4c7d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 16:11:03 +1100 Subject: nv40: workaround fp result.data[0] clobberage Temporary, the fp reg handling will get reworked at some point in the near future. But before that, there's a few bugs to find. --- src/gallium/drivers/nv40/nv40_fragprog.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 2d82f86ef5c..4e425366985 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -58,7 +58,7 @@ temp(struct nv40_fpc *fpc) int idx; idx = fpc->temp_temp_count++; - idx += fpc->high_temp + 1; + idx += fpc->high_temp + 2; return nv40_sr(NV40SR_TEMP, idx); } @@ -817,7 +817,9 @@ nv40_fragprog_upload(struct nv40_context *nv40, #if 0 for (i = 0; i < fp->insn_len; i++) { + fflush(stdout); fflush(stderr); NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + fflush(stdout); fflush(stderr); } #endif -- cgit v1.2.3 From ba223e91df8f372a983e99c453947e4340d7d884 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 16:26:42 +1100 Subject: nv40: fp: fix multiple refs to a single const withing an instruction --- src/gallium/drivers/nv40/nv40_fragprog.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 4e425366985..4cdbf962ab4 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -41,6 +41,7 @@ struct nv40_fpc { uint colour_id; unsigned inst_offset; + unsigned have_const; struct { int pipe; @@ -113,7 +114,11 @@ emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src) sr |= (src.index << NV40_FP_REG_SRC_SHIFT); break; case NV40SR_CONST: - grow_insns(fpc, 4); + if (!fpc->have_const) { + grow_insns(fpc, 4); + fpc->have_const = 1; + } + hw = &fp->insn[fpc->inst_offset]; if (fpc->consts[src.index].pipe >= 0) { struct nv40_fragment_program_data *fpd; @@ -190,6 +195,7 @@ nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op, uint32_t *hw; fpc->inst_offset = fp->insn_len; + fpc->have_const = 0; grow_insns(fpc, 4); hw = &fp->insn[fpc->inst_offset]; memset(hw, 0, sizeof(uint32_t) * 4); -- cgit v1.2.3 From 3158035154915ae11bebca045e3f0ce3b0e264ee Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 16:31:31 +1100 Subject: nv40: catch fp extra-const cases where both const and immd have same idx --- src/gallium/drivers/nv40/nv40_fragprog.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 4cdbf962ab4..c854ae89f74 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -384,7 +384,7 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); struct nv40_sreg src[3], dst, tmp; int mask, sat, unit; - int ai = -1, ci = -1; + int ai = -1, ci = -1, ii = -1; int i; if (finst->Instruction.Opcode == TGSI_OPCODE_END) @@ -428,8 +428,8 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, } break; case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { + if ((ci == -1 && ii == -1) || + ci == fsrc->SrcRegister.Index) { ci = fsrc->SrcRegister.Index; src[i] = tgsi_src(fpc, fsrc); } else { @@ -438,6 +438,17 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, tgsi_src(fpc, fsrc), none, none); } break; + case TGSI_FILE_IMMEDIATE: + if ((ci == -1 && ii == -1) || + ii == fsrc->SrcRegister.Index) { + ii = fsrc->SrcRegister.Index; + src[i] = tgsi_src(fpc, fsrc); + } else { + src[i] = temp(fpc); + arith(fpc, 0, MOV, src[i], MASK_ALL, + tgsi_src(fpc, fsrc), none, none); + } + break; case TGSI_FILE_TEMPORARY: /* handled above */ break; -- cgit v1.2.3 From 4dfcf912964f7c14c877abde04255a3896fb903d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 18:11:49 +1100 Subject: nv40: remove some badness from fp temp/result allocation Should hopefully be OK now (on the fragprog size) for MRT. The hack from a commit 01cb2cd93efe7ad94d7fd36aa5a776c2e3ab4c7d is no longer needed. --- src/gallium/drivers/nv40/nv40_fragprog.c | 158 +++++++++++++++++++++---------- 1 file changed, 107 insertions(+), 51 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index c854ae89f74..6d8a29ed9c0 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -33,12 +33,12 @@ struct nv40_fpc { uint attrib_map[PIPE_MAX_SHADER_INPUTS]; - int high_temp; - int temp_temp_count; - int num_regs; + unsigned r_temps; + unsigned r_temps_discard; + struct nv40_sreg r_result[PIPE_MAX_SHADER_OUTPUTS]; + struct nv40_sreg *r_temp; - uint depth_id; - uint colour_id; + int num_regs; unsigned inst_offset; unsigned have_const; @@ -56,13 +56,26 @@ struct nv40_fpc { static INLINE struct nv40_sreg temp(struct nv40_fpc *fpc) { - int idx; + int idx = ffs(~fpc->r_temps) - 1; - idx = fpc->temp_temp_count++; - idx += fpc->high_temp + 2; + if (idx < 0) { + NOUVEAU_ERR("out of temps!!\n"); + assert(0); + return nv40_sr(NV40SR_TEMP, 0); + } + + fpc->r_temps |= (1 << idx); + fpc->r_temps_discard |= (1 << idx); return nv40_sr(NV40SR_TEMP, idx); } +static INLINE void +release_temps(struct nv40_fpc *fpc) +{ + fpc->r_temps &= ~fpc->r_temps_discard; + fpc->r_temps_discard = 0; +} + static INLINE struct nv40_sreg constant(struct nv40_fpc *fpc, int pipe, float vals[4]) { @@ -254,18 +267,11 @@ tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) src = fpc->imm[fsrc->SrcRegister.Index]; break; case TGSI_FILE_TEMPORARY: - src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index + 1); - if (fpc->high_temp < src.index) - fpc->high_temp = src.index; + src = fpc->r_temp[fsrc->SrcRegister.Index]; break; - /* This is clearly insane, but gallium hands us shaders like this. - * Luckily fragprog results are just temp regs.. - */ + /* NV40 fragprog result regs are just temps, so this is simple */ case TGSI_FILE_OUTPUT: - if (fsrc->SrcRegister.Index == fpc->colour_id) - return nv40_sr(NV40SR_OUTPUT, 0); - else - return nv40_sr(NV40SR_OUTPUT, 1); + src = fpc->r_result[fsrc->SrcRegister.Index]; break; default: NOUVEAU_ERR("bad src file\n"); @@ -283,20 +289,11 @@ tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc) static INLINE struct nv40_sreg tgsi_dst(struct nv40_fpc *fpc, const struct tgsi_full_dst_register *fdst) { - int idx; - switch (fdst->DstRegister.File) { case TGSI_FILE_OUTPUT: - if (fdst->DstRegister.Index == fpc->colour_id) - return nv40_sr(NV40SR_OUTPUT, 0); - else - return nv40_sr(NV40SR_OUTPUT, 1); - break; + return fpc->r_result[fdst->DstRegister.Index]; case TGSI_FILE_TEMPORARY: - idx = fdst->DstRegister.Index + 1; - if (fpc->high_temp < idx) - fpc->high_temp = idx; - return nv40_sr(NV40SR_TEMP, idx); + return fpc->r_temp[fdst->DstRegister.Index]; case TGSI_FILE_NULL: return nv40_sr(NV40SR_NONE, 0); default: @@ -390,7 +387,6 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, if (finst->Instruction.Opcode == TGSI_OPCODE_END) return TRUE; - fpc->temp_temp_count = 0; for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fsrc; @@ -670,6 +666,7 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, return FALSE; } + release_temps(fpc); return TRUE; } @@ -719,45 +716,50 @@ static boolean nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, const struct tgsi_full_declaration *fdec) { + unsigned idx = fdec->u.DeclarationRange.First; + unsigned hw; + switch (fdec->Semantic.SemanticName) { case TGSI_SEMANTIC_POSITION: - fpc->depth_id = fdec->u.DeclarationRange.First; + hw = 1; break; case TGSI_SEMANTIC_COLOR: - fpc->colour_id = fdec->u.DeclarationRange.First; + switch (idx) { + case 0: hw = 0; break; + case 1: hw = 2; break; + case 2: hw = 3; break; + case 3: hw = 4; break; + default: + NOUVEAU_ERR("bad rcol index\n"); + return FALSE; + } break; default: NOUVEAU_ERR("bad output semantic\n"); return FALSE; } + fpc->r_result[idx] = nv40_sr(NV40SR_OUTPUT, hw); + fpc->r_temps |= (1 << hw); return TRUE; } -static void -nv40_fragprog_translate(struct nv40_context *nv40, - struct nv40_fragment_program *fp) +static boolean +nv40_fragprog_prepare(struct nv40_fpc *fpc) { - struct tgsi_parse_context parse; - struct nv40_fpc *fpc = NULL; + struct tgsi_parse_context p; + int high_temp = -1, i; - fpc = CALLOC(1, sizeof(struct nv40_fpc)); - if (!fpc) - return; - fpc->fp = fp; - fpc->high_temp = -1; - fpc->num_regs = 2; + tgsi_parse_init(&p, fpc->fp->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; - tgsi_parse_init(&parse, fp->pipe.tokens); - - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { + tgsi_parse_token(&p); + switch(tok->Token.Type) { case TGSI_TOKEN_TYPE_DECLARATION: { const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; + fdec = &p.FullToken.FullDeclaration; switch (fdec->Declaration.File) { case TGSI_FILE_INPUT: if (!nv40_fragprog_parse_decl_attrib(fpc, fdec)) @@ -767,6 +769,12 @@ nv40_fragprog_translate(struct nv40_context *nv40, if (!nv40_fragprog_parse_decl_output(fpc, fdec)) goto out_err; break; + case TGSI_FILE_TEMPORARY: + if (fdec->u.DeclarationRange.Last > high_temp) { + high_temp = + fdec->u.DeclarationRange.Last; + } + break; default: break; } @@ -777,7 +785,7 @@ nv40_fragprog_translate(struct nv40_context *nv40, struct tgsi_full_immediate *imm; float vals[4]; - imm = &parse.FullToken.FullImmediate; + imm = &p.FullToken.FullImmediate; assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); assert(fpc->nr_imm < MAX_IMM); @@ -788,6 +796,52 @@ nv40_fragprog_translate(struct nv40_context *nv40, fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); } break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (++high_temp) { + fpc->r_temp = CALLOC(high_temp, sizeof(struct nv40_sreg)); + for (i = 0; i < high_temp; i++) + fpc->r_temp[i] = temp(fpc); + fpc->r_temps_discard = 0; + } + + return TRUE; + +out_err: + if (fpc->r_temp) + FREE(fpc->r_temp); + tgsi_parse_free(&p); + return FALSE; +} + +static void +nv40_fragprog_translate(struct nv40_context *nv40, + struct nv40_fragment_program *fp) +{ + struct tgsi_parse_context parse; + struct nv40_fpc *fpc = NULL; + + fpc = CALLOC(1, sizeof(struct nv40_fpc)); + if (!fpc) + return; + fpc->fp = fp; + fpc->num_regs = 2; + + if (!nv40_fragprog_prepare(fpc)) { + FREE(fpc); + return; + } + + tgsi_parse_init(&parse, fp->pipe.tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { case TGSI_TOKEN_TYPE_INSTRUCTION: { const struct tgsi_full_instruction *finst; @@ -818,6 +872,8 @@ nv40_fragprog_translate(struct nv40_context *nv40, fp->translated = TRUE; out_err: tgsi_parse_free(&parse); + if (fpc->r_temp) + FREE(fpc->r_temp); FREE(fpc); } -- cgit v1.2.3 From e1ad8c232ba31985a6f9e5b76279f2f131312d1d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 18:20:04 +1100 Subject: nv40: fix fp depth write --- src/gallium/drivers/nv40/nv40_fragprog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 6d8a29ed9c0..87bca41cf64 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -724,7 +724,7 @@ nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, hw = 1; break; case TGSI_SEMANTIC_COLOR: - switch (idx) { + switch (fdec->Semantic.SemanticIndex) { case 0: hw = 0; break; case 1: hw = 2; break; case 2: hw = 3; break; -- cgit v1.2.3 From 17491ea27ffa6b48e31c0ad6ad6f795dd000c476 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 23 Mar 2008 19:08:59 +1100 Subject: nv40: vp reg changes similar to recent fp changes --- src/gallium/drivers/nv40/nv40_vertprog.c | 106 +++++++++++++++++++++++++------ 1 file changed, 86 insertions(+), 20 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 385c8aa0780..e5ce8943752 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -42,10 +42,11 @@ struct nv40_vpc { struct nv40_vertex_program_exec *vpi; - unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; - - int high_temp; - int temp_temp_count; + unsigned r_temps; + unsigned r_temps_discard; + struct nv40_sreg r_result[PIPE_MAX_SHADER_OUTPUTS]; + struct nv40_sreg *r_address; + struct nv40_sreg *r_temp; struct nv40_sreg *imm; unsigned nr_imm; @@ -54,13 +55,26 @@ struct nv40_vpc { static struct nv40_sreg temp(struct nv40_vpc *vpc) { - int idx; + int idx = ffs(~vpc->r_temps) - 1; + + if (idx < 0) { + NOUVEAU_ERR("out of temps!!\n"); + assert(0); + return nv40_sr(NV40SR_TEMP, 0); + } - idx = vpc->temp_temp_count++; - idx += vpc->high_temp + 1; + vpc->r_temps |= (1 << idx); + vpc->r_temps_discard |= (1 << idx); return nv40_sr(NV40SR_TEMP, idx); } +static INLINE void +release_temps(struct nv40_vpc *vpc) +{ + vpc->r_temps &= ~vpc->r_temps_discard; + vpc->r_temps_discard = 0; +} + static struct nv40_sreg constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w) { @@ -257,9 +271,7 @@ tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) { src = vpc->imm[fsrc->SrcRegister.Index]; break; case TGSI_FILE_TEMPORARY: - if (vpc->high_temp < fsrc->SrcRegister.Index) - vpc->high_temp = fsrc->SrcRegister.Index; - src = nv40_sr(NV40SR_TEMP, fsrc->SrcRegister.Index); + src = vpc->r_temp[fsrc->SrcRegister.Index]; break; default: NOUVEAU_ERR("bad src file\n"); @@ -281,14 +293,13 @@ tgsi_dst(struct nv40_vpc *vpc, const struct tgsi_full_dst_register *fdst) { switch (fdst->DstRegister.File) { case TGSI_FILE_OUTPUT: - dst = nv40_sr(NV40SR_OUTPUT, - vpc->output_map[fdst->DstRegister.Index]); - + dst = vpc->r_result[fdst->DstRegister.Index]; break; case TGSI_FILE_TEMPORARY: - dst = nv40_sr(NV40SR_TEMP, fdst->DstRegister.Index); - if (vpc->high_temp < dst.index) - vpc->high_temp = dst.index; + dst = vpc->r_temp[fdst->DstRegister.Index]; + break; + case TGSI_FILE_ADDRESS: + dst = vpc->r_address[fdst->DstRegister.Index]; break; default: NOUVEAU_ERR("bad dst file\n"); @@ -323,7 +334,6 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, if (finst->Instruction.Opcode == TGSI_OPCODE_END) return TRUE; - vpc->temp_temp_count = 0; for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { const struct tgsi_full_src_register *fsrc; @@ -471,6 +481,7 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, return FALSE; } + release_temps(vpc); return TRUE; } @@ -478,6 +489,7 @@ static boolean nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, const struct tgsi_full_declaration *fdec) { + unsigned idx = fdec->u.DeclarationRange.First; int hw; switch (fdec->Semantic.SemanticName) { @@ -525,7 +537,7 @@ nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, return FALSE; } - vpc->output_map[fdec->u.DeclarationRange.First] = hw; + vpc->r_result[idx] = nv40_sr(NV40SR_OUTPUT, hw); return TRUE; } @@ -533,7 +545,7 @@ static boolean nv40_vertprog_prepare(struct nv40_vpc *vpc) { struct tgsi_parse_context p; - int nr_imm = 0; + int high_temp = -1, high_addr = -1, nr_imm = 0, i; tgsi_parse_init(&p, vpc->vp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&p)) { @@ -544,6 +556,48 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) case TGSI_TOKEN_TYPE_IMMEDIATE: nr_imm++; break; + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + + fdec = &p.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_TEMPORARY: + if (fdec->u.DeclarationRange.Last > high_temp) { + high_temp = + fdec->u.DeclarationRange.Last; + } + break; +#if 0 /* this would be nice.. except gallium doesn't track it */ + case TGSI_FILE_ADDRESS: + if (fdec->u.DeclarationRange.Last > high_addr) { + high_addr = + fdec->u.DeclarationRange.Last; + } + break; +#endif + default: + break; + } + } + break; +#if 1 /* yay, parse instructions looking for address regs instead */ + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + const struct tgsi_full_dst_register *fdst; + + finst = &p.FullToken.FullInstruction; + fdst = &finst->FullDstRegisters[0]; + + if (fdst->DstRegister.File == TGSI_FILE_ADDRESS) { + if (fdst->DstRegister.Index > high_addr) + high_addr = fdst->DstRegister.Index; + } + + } + break; +#endif default: break; } @@ -555,6 +609,19 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) assert(vpc->imm); } + if (++high_temp) { + vpc->r_temp = CALLOC(high_temp, sizeof(struct nv40_sreg)); + for (i = 0; i < high_temp; i++) + vpc->r_temp[i] = temp(vpc); + } + + if (++high_addr) { + vpc->r_address = CALLOC(high_addr, sizeof(struct nv40_sreg)); + for (i = 0; i < high_addr; i++) + vpc->r_address[i] = temp(vpc); + } + + vpc->r_temps_discard = 0; return TRUE; } @@ -569,7 +636,6 @@ nv40_vertprog_translate(struct nv40_context *nv40, if (!vpc) return; vpc->vp = vp; - vpc->high_temp = -1; if (!nv40_vertprog_prepare(vpc)) { FREE(vpc); -- cgit v1.2.3 From 46a7f297a1477b9b59a5a11bf090db0ecbdf1ed7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 24 Mar 2008 13:01:35 +1100 Subject: nouveau: fix infinite wait on space for userbuf. A recent commit removed a bug which essentially caused a sync after each use of a user buffer. In fixing it, the scratch area can now become fragmented under some circumstances leading to nouveau_bo_tmp() waiting forever for a large enough block of free space. This fixes ppracer, gl-117 and probably a heap of other apps. --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index 46c0759dbb4..792eaaa79e5 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -88,6 +88,26 @@ nouveau_bo_tmp_del(void *priv) nouveau_resource_free(&r); } +static unsigned +nouveau_bo_tmp_max(struct nouveau_device_priv *nvdev) +{ + struct nouveau_resource *r = nvdev->sa_heap; + unsigned max = 0; + + while (r) { + if (r->in_use && !nouveau_fence(r->priv)->emitted) { + r = r->next; + continue; + } + + if (max < r->size) + max = r->size; + r = r->next; + } + + return max; +} + static struct nouveau_resource * nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, struct nouveau_fence *fence) @@ -103,6 +123,11 @@ nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, assert(ref); while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) { + if (nouveau_bo_tmp_max(nvdev) < size) { + nouveau_fence_ref(NULL, &ref); + return NULL; + } + nouveau_fence_flush(chan); } nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r); -- cgit v1.2.3 From 799d3bce06e998e51ad8df6bcadeb41bd061801e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 24 Mar 2008 13:41:40 +1100 Subject: nv40: respect rasterizer cso bypass_clipping flag --- src/gallium/drivers/nv40/nv40_context.h | 1 + src/gallium/drivers/nv40/nv40_state_viewport.c | 16 ++++++++++++++-- 2 files changed, 15 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 02ca20b8014..b50f6f8fefc 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -97,6 +97,7 @@ struct nv40_blend_state { struct nv40_state { unsigned scissor_enabled; unsigned stipple_enabled; + unsigned viewport_bypass; unsigned fp_samplers; uint64_t dirty; diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c index 9e5c7a72a7f..1b6248e5b8e 100644 --- a/src/gallium/drivers/nv40/nv40_state_viewport.c +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -5,8 +5,20 @@ nv40_state_viewport_validate(struct nv40_context *nv40) { struct nouveau_stateobj *so = so_new(11, 0); struct pipe_viewport_state *vpt = &nv40->viewport; + unsigned bypass; - if (nv40->render_mode == HW) { + if (nv40->render_mode == HW && !nv40->rasterizer->pipe.bypass_clipping) + bypass = 0; + else + bypass = 1; + + if (nv40->state.hw[NV40_STATE_VIEWPORT] && + (bypass || !(nv40->dirty & NV40_NEW_VIEWPORT)) && + nv40->state.viewport_bypass == bypass) + return FALSE; + nv40->state.viewport_bypass = bypass; + + if (!bypass) { so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); so_data (so, fui(vpt->translate[0])); @@ -48,7 +60,7 @@ nv40_state_viewport_validate(struct nv40_context *nv40) struct nv40_state_entry nv40_state_viewport = { .validate = nv40_state_viewport_validate, .dirty = { - .pipe = NV40_NEW_VIEWPORT, + .pipe = NV40_NEW_VIEWPORT | NV40_NEW_RAST, .hw = NV40_STATE_VIEWPORT } }; -- cgit v1.2.3 From 5ce37d42b3268102caf8225be4ca18418bfab7c4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 24 Mar 2008 13:45:38 +1100 Subject: nv40: maintain pipe_surface status field --- src/gallium/drivers/nv40/nv40_clear.c | 1 + src/gallium/drivers/nv40/nv40_state_emit.c | 8 ++++++++ 2 files changed, 9 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_clear.c b/src/gallium/drivers/nv40/nv40_clear.c index 2c4e8f01fda..59efd620e32 100644 --- a/src/gallium/drivers/nv40/nv40_clear.c +++ b/src/gallium/drivers/nv40/nv40_clear.c @@ -9,4 +9,5 @@ nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); + ps->status = PIPE_SURFACE_STATUS_CLEAR; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 056238cc836..a9a9abc9220 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -40,6 +40,14 @@ static void nv40_state_do_validate(struct nv40_context *nv40, struct nv40_state_entry **states) { + const struct pipe_framebuffer_state *fb = &nv40->framebuffer; + unsigned i; + + for (i = 0; i < fb->num_cbufs; i++) + fb->cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED; + if (fb->zsbuf) + fb->zsbuf->status = PIPE_SURFACE_STATUS_DEFINED; + while (*states) { struct nv40_state_entry *e = *states; -- cgit v1.2.3 From 601b018a9a6143c634239d5bb51616724c2e593d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 25 Mar 2008 12:12:26 +1100 Subject: nouveau: refcount buffers on validate list. --- src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 1 - src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 8 ++++++-- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h index 1c9b5799f96..d5a0e25b573 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h @@ -108,7 +108,6 @@ nouveau_fence_flush(struct nouveau_channel *); struct nouveau_pushbuf_reloc { uint64_t next; - uint64_t handle; uint32_t *ptr; uint32_t flags; uint32_t data; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c index a2b9321b15b..2e3ac5492fc 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c @@ -157,11 +157,13 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) if (bo->offset == nouveau_bo(bo)->offset && bo->flags == nouveau_bo(bo)->flags) { + while ((r = ptr_to_pbrel(pbbo->relocs))) { pbbo->relocs = r->next; free(r); } + nouveau_bo_del(&bo); nvpb->buffers = pbbo->next; free(pbbo); continue; @@ -175,6 +177,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) free(r); } + nouveau_bo_del(&bo); nvpb->buffers = pbbo->next; free(pbbo); } @@ -202,6 +205,7 @@ nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) { struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); struct nouveau_pushbuf_bo *pbbo = ptr_to_pbbo(nvpb->buffers); + struct nouveau_bo *ref = NULL; while (pbbo) { if (pbbo->handle == bo->handle) @@ -214,7 +218,8 @@ nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) nvpb->buffers = pbbo_to_ptr(pbbo); nvpb->nr_buffers++; - pbbo->handle = bo_to_ptr(bo); + nouveau_bo_ref(bo->device, bo->handle, &ref); + pbbo->handle = bo_to_ptr(ref); pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; pbbo->relocs = 0; pbbo->nr_relocs = 0; @@ -244,7 +249,6 @@ nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, pbbo->flags |= (flags & NOUVEAU_BO_RDWR); pbbo->flags &= (flags | NOUVEAU_BO_RDWR); - r->handle = bo_to_ptr(r); r->ptr = ptr; r->flags = flags; r->data = data; -- cgit v1.2.3 From ce64778ed1f436d81178862dc0032dfd16b4b7de Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 25 Mar 2008 12:57:36 +1100 Subject: nv40: respect do_flip in surface_copy() --- src/gallium/drivers/nv40/nv40_surface.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c index e8a60116964..c0d135eb36d 100644 --- a/src/gallium/drivers/nv40/nv40_surface.c +++ b/src/gallium/drivers/nv40/nv40_surface.c @@ -42,8 +42,20 @@ nv40_surface_copy(struct pipe_context *pipe, unsigned do_flip, struct nv40_context *nv40 = nv40_context(pipe); struct nouveau_winsys *nvws = nv40->nvws; - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); + if (do_flip) { + /*XXX: This dodgyness will do for now for correctness. But, + * need to investigate whether the 2D engine is able to + * manage a flip (perhaps SIFM?), if not, use the 3D engine + */ + desty += height; + while (height--) { + nvws->surface_copy(nvws, dest, destx, desty--, src, + srcx, srcy++, width, 1); + } + } else { + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); + } } static void -- cgit v1.2.3 From 3017999d9bee8f9d2ef170c1bb6926aab8e08393 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Mar 2008 19:00:07 +1000 Subject: nv40: vp const/immd fix --- src/gallium/drivers/nv40/nv40_vertprog.c | 20 ++++++++++++++------ 1 file changed, 14 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index e5ce8943752..41885b9d4a5 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -328,7 +328,7 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, struct nv40_sreg src[3], dst, tmp; struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); int mask; - int ai = -1, ci = -1; + int ai = -1, ci = -1, ii = -1; int i; if (finst->Instruction.Opcode == TGSI_OPCODE_END) @@ -358,12 +358,9 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, tgsi_src(vpc, fsrc), none, none); } break; - /*XXX: index comparison is broken now that consts come from - * two different register files. - */ case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { + if ((ci == -1 && ii == -1) || + ci == fsrc->SrcRegister.Index) { ci = fsrc->SrcRegister.Index; src[i] = tgsi_src(vpc, fsrc); } else { @@ -372,6 +369,17 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, tgsi_src(vpc, fsrc), none, none); } break; + case TGSI_FILE_IMMEDIATE: + if ((ci == -1 && ii == -1) || + ii == fsrc->SrcRegister.Index) { + ii = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; case TGSI_FILE_TEMPORARY: /* handled above */ break; -- cgit v1.2.3 From 68395f6726183a0776e324b900e429449ede2b22 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Mar 2008 19:08:59 +1000 Subject: nv40: vp 1/0/- swz --- src/gallium/drivers/nv40/nv40_vertprog.c | 74 +++++++++++++++++++++++++++++++- 1 file changed, 73 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 41885b9d4a5..40ef7174a41 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -4,13 +4,13 @@ #include "pipe/p_shader_tokens.h" #include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_util.h" #include "nv40_context.h" #include "nv40_state.h" /* TODO (at least...): * 1. Indexed consts + ARL - * 2. Arb. swz/negation * 3. NV_vp11, NV_vp2, NV_vp3 features * - extra arith opcodes * - branching @@ -321,6 +321,66 @@ tgsi_mask(uint tgsi) return mask; } +static boolean +src_native_swz(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc, + struct nv40_sreg *src) +{ + const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); + struct nv40_sreg tgsi = tgsi_src(vpc, fsrc); + uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0; + uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX, + fsrc->SrcRegisterExtSwz.NegateY, + fsrc->SrcRegisterExtSwz.NegateZ, + fsrc->SrcRegisterExtSwz.NegateW }; + uint c; + + for (c = 0; c < 4; c++) { + switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + mask |= tgsi_mask(1 << c); + break; + case TGSI_EXTSWIZZLE_ZERO: + zero_mask |= tgsi_mask(1 << c); + tgsi.swz[c] = SWZ_X; + break; + case TGSI_EXTSWIZZLE_ONE: + one_mask |= tgsi_mask(1 << c); + tgsi.swz[c] = SWZ_X; + break; + default: + assert(0); + } + + if (!tgsi.negate && neg[c]) + neg_mask |= tgsi_mask(1 << c); + } + + if (mask == MASK_ALL && !neg_mask) + return TRUE; + + *src = temp(vpc); + + if (mask) + arith(vpc, 0, OP_MOV, *src, mask, tgsi, none, none); + + if (zero_mask) + arith(vpc, 0, OP_SFL, *src, zero_mask, *src, none, none); + + if (one_mask) + arith(vpc, 0, OP_STR, *src, one_mask, *src, none, none); + + if (neg_mask) { + struct nv40_sreg one = temp(vpc); + arith(vpc, 0, OP_STR, one, neg_mask, one, none, none); + arith(vpc, 0, OP_MUL, *src, neg_mask, *src, neg(one), none); + } + + return FALSE; +} + static boolean nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, const struct tgsi_full_instruction *finst) @@ -347,6 +407,18 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc; fsrc = &finst->FullSrcRegisters[i]; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + case TGSI_FILE_CONSTANT: + case TGSI_FILE_TEMPORARY: + if (!src_native_swz(vpc, fsrc, &src[i])) + continue; + break; + default: + break; + } + switch (fsrc->SrcRegister.File) { case TGSI_FILE_INPUT: if (ai == -1 || ai == fsrc->SrcRegister.Index) { -- cgit v1.2.3 From bbefb541ad94382debb0f7a8daa636729799a31a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 30 Mar 2008 20:32:22 +1000 Subject: nouveau: adapt to recent gallium changes --- src/gallium/drivers/nouveau/nouveau_push.h | 8 ++--- src/gallium/drivers/nouveau/nouveau_stateobj.h | 4 +-- src/gallium/drivers/nouveau/nouveau_winsys.h | 3 +- src/gallium/drivers/nv10/nv10_context.c | 19 +++------- src/gallium/drivers/nv10/nv10_state.c | 18 +++++----- src/gallium/drivers/nv30/nv30_context.c | 19 +++------- src/gallium/drivers/nv30/nv30_query.c | 2 +- src/gallium/drivers/nv30/nv30_state.c | 18 +++++----- src/gallium/drivers/nv30/nv30_vbo.c | 4 +-- src/gallium/drivers/nv40/nv40_context.c | 17 ++------- src/gallium/drivers/nv40/nv40_draw.c | 4 +-- src/gallium/drivers/nv40/nv40_query.c | 2 +- src/gallium/drivers/nv40/nv40_screen.c | 2 +- src/gallium/drivers/nv40/nv40_state.c | 20 +++++------ src/gallium/drivers/nv40/nv40_state_emit.c | 4 +-- src/gallium/drivers/nv40/nv40_vbo.c | 4 +-- src/gallium/drivers/nv50/nv50_context.c | 18 ++-------- src/gallium/drivers/nv50/nv50_screen.c | 2 +- src/gallium/drivers/nv50/nv50_state.c | 12 +++---- src/gallium/winsys/dri/nouveau/nouveau_context.c | 4 +-- src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 12 ++++++- .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 42 ++++++++++++++++++++++ 22 files changed, 121 insertions(+), 117 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_push.h b/src/gallium/drivers/nouveau/nouveau_push.h index 225c17744ad..54ef1c1291e 100644 --- a/src/gallium/drivers/nouveau/nouveau_push.h +++ b/src/gallium/drivers/nouveau/nouveau_push.h @@ -27,7 +27,7 @@ #define BEGIN_RING(obj,mthd,size) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - pc->nvws->push_flush(pc->nvws, ((size) + 1)); \ + pc->nvws->push_flush(pc->nvws, ((size) + 1), NULL); \ OUT_RING((pc->obj->subc << 13) | ((size) << 18) | (mthd)); \ pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ } while(0) @@ -36,9 +36,9 @@ BEGIN_RING(obj, (mthd) | 0x40000000, (size)); \ } while(0) -#define FIRE_RING() do { \ +#define FIRE_RING(fence) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ - pc->nvws->push_flush(pc->nvws, 0); \ + pc->nvws->push_flush(pc->nvws, 0, fence); \ } while(0) #define OUT_RELOC(bo,data,flags,vor,tor) do { \ @@ -73,7 +73,7 @@ #define OUT_RELOCm(bo, flags, obj, mthd, size) do { \ NOUVEAU_PUSH_CONTEXT(pc); \ if (pc->nvws->channel->pushbuf->remaining < ((size) + 1)) \ - pc->nvws->push_flush(pc->nvws->channel, ((size) + 1)); \ + pc->nvws->push_flush(pc->nvws->channel, ((size) + 1), NULL); \ OUT_RELOCd((bo), (pc->obj->subc << 13) | ((size) << 18) | (mthd), \ (flags), 0, 0); \ pc->nvws->channel->pushbuf->remaining -= ((size) + 1); \ diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index d465223748a..d501b76b51e 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -99,7 +99,7 @@ so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) nr = so->cur - so->push; if (pb->remaining < nr) - nvws->push_flush(nvws, nr); + nvws->push_flush(nvws, nr, NULL); pb->remaining -= nr; memcpy(pb->cur, so->push, nr * 4); @@ -120,7 +120,7 @@ so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) i = so->cur_reloc << 1; if (nvws->channel->pushbuf->remaining < i) - nvws->push_flush(nvws, i); + nvws->push_flush(nvws, i, NULL); nvws->channel->pushbuf->remaining -= i; for (i = 0; i < so->cur_reloc; i++) { diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 7fa7cc0910c..2a5305f7ce4 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -27,7 +27,8 @@ struct nouveau_winsys { int (*push_reloc)(struct nouveau_winsys *, void *ptr, struct pipe_buffer *, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor); - int (*push_flush)(struct nouveau_winsys *, unsigned size); + int (*push_flush)(struct nouveau_winsys *, unsigned size, + struct pipe_fence_handle **fence); int (*grobj_alloc)(struct nouveau_winsys *, int grclass, struct nouveau_grobj **); diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index 2599acf2861..14042fb2fbb 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -7,10 +7,10 @@ #include "nv10_screen.h" static void -nv10_flush(struct pipe_context *pipe, unsigned flags) +nv10_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) { struct nv10_context *nv10 = nv10_context(pipe); - struct nouveau_winsys *nvws = nv10->nvws; if (flags & PIPE_FLUSH_TEXTURE_CACHE) { BEGIN_RING(celsius, 0x1fd8, 1); @@ -19,18 +19,7 @@ nv10_flush(struct pipe_context *pipe, unsigned flags) OUT_RING (1); } - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv10->sync, 0); - BEGIN_RING(celsius, 0x104, 1); - OUT_RING (0); - BEGIN_RING(celsius, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv10->sync, 0, 0, 2000); + FIRE_RING(fence); } static void @@ -253,7 +242,7 @@ nv10_init_hwctx(struct nv10_context *nv10, int celsius_class) OUT_RING (1); - FIRE_RING (); + FIRE_RING (NULL); return TRUE; } diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index d7c445f1b39..1ff01de1060 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -633,24 +633,22 @@ nv10_set_viewport_state(struct pipe_context *pipe, } static void -nv10_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) +nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *vb) { struct nv10_context *nv10 = nv10_context(pipe); - nv10->vtxbuf[index] = *vb; - + memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); nv10->dirty |= NV10_NEW_ARRAYS; } static void -nv10_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) +nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *ve) { struct nv10_context *nv10 = nv10_context(pipe); - nv10->vtxelt[index] = *ve; - + memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); nv10->dirty |= NV10_NEW_ARRAYS; } @@ -693,7 +691,7 @@ nv10_init_state_functions(struct nv10_context *nv10) nv10->pipe.set_scissor_state = nv10_set_scissor_state; nv10->pipe.set_viewport_state = nv10_set_viewport_state; - nv10->pipe.set_vertex_buffer = nv10_set_vertex_buffer; - nv10->pipe.set_vertex_element = nv10_set_vertex_element; + nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers; + nv10->pipe.set_vertex_elements = nv10_set_vertex_elements; } diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index cdd662a9f19..1e729d789bd 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -7,10 +7,10 @@ #include "nv30_screen.h" static void -nv30_flush(struct pipe_context *pipe, unsigned flags) +nv30_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) { struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_winsys *nvws = nv30->nvws; if (flags & PIPE_FLUSH_TEXTURE_CACHE) { BEGIN_RING(rankine, 0x1fd8, 1); @@ -19,18 +19,7 @@ nv30_flush(struct pipe_context *pipe, unsigned flags) OUT_RING (1); } - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv30->sync, 0); - BEGIN_RING(rankine, 0x104, 1); - OUT_RING (0); - BEGIN_RING(rankine, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv30->sync, 0, 0, 2000); + FIRE_RING(fence); } static void @@ -144,7 +133,7 @@ nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) BEGIN_RING(rankine, 0x1e94, 1); OUT_RING (0x13); - FIRE_RING (); + FIRE_RING (NULL); return TRUE; } diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c index 71fdcfa24df..e19cb455dce 100644 --- a/src/gallium/drivers/nv30/nv30_query.c +++ b/src/gallium/drivers/nv30/nv30_query.c @@ -67,7 +67,7 @@ nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq) BEGIN_RING(rankine, NV34TCL_QUERY_GET, 1); OUT_RING ((0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) | ((q->object->start * 32) << NV34TCL_QUERY_GET_OFFSET_SHIFT)); - FIRE_RING(); + FIRE_RING(NULL); } static boolean diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index b0055892aeb..983638adccb 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -710,24 +710,22 @@ nv30_set_viewport_state(struct pipe_context *pipe, } static void -nv30_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) +nv30_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *vb) { struct nv30_context *nv30 = nv30_context(pipe); - nv30->vtxbuf[index] = *vb; - + memcpy(nv30->vtxbuf, vb, sizeof(*vb) * count); nv30->dirty |= NV30_NEW_ARRAYS; } static void -nv30_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) +nv30_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *ve) { struct nv30_context *nv30 = nv30_context(pipe); - nv30->vtxelt[index] = *ve; - + memcpy(nv30->vtxelt, ve, sizeof(*ve) * count); nv30->dirty |= NV30_NEW_ARRAYS; } @@ -770,7 +768,7 @@ nv30_init_state_functions(struct nv30_context *nv30) nv30->pipe.set_scissor_state = nv30_set_scissor_state; nv30->pipe.set_viewport_state = nv30_set_viewport_state; - nv30->pipe.set_vertex_buffer = nv30_set_vertex_buffer; - nv30->pipe.set_vertex_element = nv30_set_vertex_element; + nv30->pipe.set_vertex_buffers = nv30_set_vertex_buffers; + nv30->pipe.set_vertex_elements = nv30_set_vertex_elements; } diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index a62462f7bc5..b18a407ec5a 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -229,7 +229,7 @@ nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); OUT_RING (0); - pipe->flush(pipe, 0); + pipe->flush(pipe, 0, NULL); return TRUE; } @@ -418,7 +418,7 @@ nv30_draw_elements(struct pipe_context *pipe, mode, start, count); } - pipe->flush(pipe, 0); + pipe->flush(pipe, 0, NULL); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 7fcf8b86197..f9c93f7a2d7 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -7,10 +7,10 @@ #include "nv40_screen.h" static void -nv40_flush(struct pipe_context *pipe, unsigned flags) +nv40_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) { struct nv40_context *nv40 = nv40_context(pipe); - struct nouveau_winsys *nvws = nv40->nvws; if (flags & PIPE_FLUSH_TEXTURE_CACHE) { BEGIN_RING(curie, 0x1fd8, 1); @@ -19,18 +19,7 @@ nv40_flush(struct pipe_context *pipe, unsigned flags) OUT_RING (1); } - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(nv40->screen->sync, 0); - BEGIN_RING(curie, 0x104, 1); - OUT_RING (0); - BEGIN_RING(curie, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(nv40->screen->sync, 0, 0, 2000); + FIRE_RING(fence); } static void diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index d05e5ad1936..9cd8fa6a497 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -89,7 +89,7 @@ nv40_render_prim(struct draw_stage *stage, struct prim_header *prim, NOUVEAU_ERR("AIII, missed flush\n"); assert(0); } - FIRE_RING(); + FIRE_RING(NULL); nv40_state_emit(nv40); } @@ -275,7 +275,7 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe, ws->buffer_unmap(ws, nv40->constbuf[PIPE_SHADER_VERTEX]); draw_flush(nv40->draw); - pipe->flush(pipe, 0); + pipe->flush(pipe, 0, NULL); return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c index 15961591b96..57f39cfab0c 100644 --- a/src/gallium/drivers/nv40/nv40_query.c +++ b/src/gallium/drivers/nv40/nv40_query.c @@ -75,7 +75,7 @@ nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq) BEGIN_RING(curie, NV40TCL_QUERY_GET, 1); OUT_RING ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) | ((q->object->start * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT)); - FIRE_RING(); + FIRE_RING(NULL); } static boolean diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 75b965bb9d1..e98005f749f 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -266,7 +266,7 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, so_emit(nvws, so); so_ref(NULL, &so); - nvws->push_flush(nvws, 0); + nvws->push_flush(nvws, 0, NULL); screen->pipe.winsys = ws; screen->pipe.destroy = nv40_screen_destroy; diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 3eafbece309..1417c95e758 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -660,26 +660,26 @@ nv40_set_viewport_state(struct pipe_context *pipe, } static void -nv40_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) +nv40_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *vb) { struct nv40_context *nv40 = nv40_context(pipe); - draw_set_vertex_buffer(nv40->draw, index, vb); + draw_set_vertex_buffers(nv40->draw, count, vb); - nv40->vtxbuf[index] = *vb; + memcpy(nv40->vtxbuf, vb, sizeof(*vb) * count); nv40->dirty |= NV40_NEW_ARRAYS; } static void -nv40_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) +nv40_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *ve) { struct nv40_context *nv40 = nv40_context(pipe); - draw_set_vertex_element(nv40->draw, index, ve); + draw_set_vertex_elements(nv40->draw, count, ve); - nv40->vtxelt[index] = *ve; + memcpy(nv40->vtxelt, ve, sizeof(*ve) * count); nv40->dirty |= NV40_NEW_ARRAYS; } @@ -722,7 +722,7 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.set_scissor_state = nv40_set_scissor_state; nv40->pipe.set_viewport_state = nv40_set_viewport_state; - nv40->pipe.set_vertex_buffer = nv40_set_vertex_buffer; - nv40->pipe.set_vertex_element = nv40_set_vertex_element; + nv40->pipe.set_vertex_buffers = nv40_set_vertex_buffers; + nv40->pipe.set_vertex_elements = nv40_set_vertex_elements; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index a9a9abc9220..74feb6d4bfc 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -123,7 +123,7 @@ nv40_state_validate(struct nv40_context *nv40) return FALSE; /* Attempt to go to hwtnl again */ - nv40->pipe.flush(&nv40->pipe, 0); + nv40->pipe.flush(&nv40->pipe, 0, NULL); nv40->dirty |= (NV40_NEW_VIEWPORT | NV40_NEW_VERTPROG | NV40_NEW_ARRAYS | @@ -147,7 +147,7 @@ nv40_state_validate_swtnl(struct nv40_context *nv40) /* Setup for swtnl */ if (nv40->render_mode == HW) { NOUVEAU_ERR("hw->swtnl 0x%08x\n", nv40->fallback_swtnl); - nv40->pipe.flush(&nv40->pipe, 0); + nv40->pipe.flush(&nv40->pipe, 0, NULL); nv40->dirty |= (NV40_NEW_VIEWPORT | NV40_NEW_VERTPROG | NV40_NEW_ARRAYS | diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 4fae10f74b4..b66bf26afb3 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -209,7 +209,7 @@ nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); OUT_RING (0); - pipe->flush(pipe, 0); + pipe->flush(pipe, 0, NULL); return TRUE; } @@ -384,7 +384,7 @@ nv40_draw_elements(struct pipe_context *pipe, mode, start, count); } - pipe->flush(pipe, 0); + pipe->flush(pipe, 0, NULL); return TRUE; } diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 980d066c844..e822d863945 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -7,24 +7,12 @@ #include "nv50_screen.h" static void -nv50_flush(struct pipe_context *pipe, unsigned flags) +nv50_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) { struct nv50_context *nv50 = (struct nv50_context *)pipe; - struct nv50_screen *screen = nv50->screen; - struct nouveau_winsys *nvws = screen->nvws; - if (flags & PIPE_FLUSH_WAIT) { - nvws->notifier_reset(screen->sync, 0); - BEGIN_RING(tesla, 0x104, 1); - OUT_RING (0); - BEGIN_RING(tesla, 0x100, 1); - OUT_RING (0); - } - - FIRE_RING(); - - if (flags & PIPE_FLUSH_WAIT) - nvws->notifier_wait(screen->sync, 0, 0, 2000); + FIRE_RING(fence); } static void diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index ff4aca81a5a..586373a5c4b 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -171,7 +171,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, so_emit(nvws, so); so_ref(NULL, &so); - nvws->push_flush(nvws, 0); + nvws->push_flush(nvws, 0, NULL); screen->pipe.winsys = ws; diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index b096a2583d7..a614ea03358 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -422,14 +422,14 @@ nv50_set_viewport_state(struct pipe_context *pipe, } static void -nv50_set_vertex_buffer(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_buffer *vb) +nv50_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *vb) { } static void -nv50_set_vertex_element(struct pipe_context *pipe, unsigned index, - const struct pipe_vertex_element *ve) +nv50_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *ve) { } @@ -472,7 +472,7 @@ nv50_init_state_functions(struct nv50_context *nv50) nv50->pipe.set_scissor_state = nv50_set_scissor_state; nv50->pipe.set_viewport_state = nv50_set_viewport_state; - nv50->pipe.set_vertex_buffer = nv50_set_vertex_buffer; - nv50->pipe.set_vertex_element = nv50_set_vertex_element; + nv50->pipe.set_vertex_buffers = nv50_set_vertex_buffers; + nv50->pipe.set_vertex_elements = nv50_set_vertex_elements; } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index 336dd65847b..cf1d83b18f4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -282,7 +282,7 @@ nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) assert(nv); - st_flush(nv->st, PIPE_FLUSH_WAIT); + st_finish(nv->st); st_destroy_context(nv->st); if (nv->pctx_id >= 0) { @@ -337,7 +337,7 @@ nouveau_context_unbind(__DRIcontextPrivate *driContextPriv) struct nouveau_context *nv = driContextPriv->driverPrivate; (void)nv; - st_flush(nv->st, 0); + st_flush(nv->st, 0, NULL); return GL_TRUE; } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index bf1afce5d9d..60fdbb8dfdf 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -69,8 +69,18 @@ nouveau_pipe_push_reloc(struct nouveau_winsys *nvws, void *ptr, } static int -nouveau_pipe_push_flush(struct nouveau_winsys *nvws, unsigned size) +nouveau_pipe_push_flush(struct nouveau_winsys *nvws, unsigned size, + struct pipe_fence_handle **fence) { + if (fence) { + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(pb); + struct nouveau_fence *ref = NULL; + + nouveau_fence_ref(nvpb->fence, &ref); + *fence = (struct pipe_fence_handle *)ref; + } + return nouveau_pushbuf_flush(nvws->channel, size); } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index e1a9271395b..453b3623f4a 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -164,6 +164,44 @@ nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) nouveau_bo_unmap(nvbuf->bo); } +static INLINE struct nouveau_fence * +nouveau_pipe_fence(struct pipe_fence_handle *pfence) +{ + return (struct nouveau_fence *)pfence; +} + +static void +nouveau_pipe_fence_reference(struct pipe_winsys *ws, + struct pipe_fence_handle **ptr, + struct pipe_fence_handle *pfence) +{ + nouveau_fence_ref((void *)pfence, (void *)ptr); +} + +static int +nouveau_pipe_fence_signalled(struct pipe_winsys *ws, + struct pipe_fence_handle *pfence, unsigned flag) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)ws; + struct nouveau_fence *fence = nouveau_pipe_fence(pfence); + + if (nouveau_fence(fence)->signalled == 0) + nouveau_fence_flush(nvpws->nv->nvc->channel); + + return !nouveau_fence(fence)->signalled; +} + +static int +nouveau_pipe_fence_finish(struct pipe_winsys *ws, + struct pipe_fence_handle *pfence, unsigned flag) +{ + struct nouveau_fence *fence = nouveau_pipe_fence(pfence); + struct nouveau_fence *ref = NULL; + + nouveau_fence_ref(fence, &ref); + return nouveau_fence_wait(&ref); +} + struct pipe_winsys * nouveau_create_pipe_winsys(struct nouveau_context *nv) { @@ -189,6 +227,10 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws->buffer_map = nouveau_pipe_bo_map; pws->buffer_unmap = nouveau_pipe_bo_unmap; + pws->fence_reference = nouveau_pipe_fence_reference; + pws->fence_signalled = nouveau_pipe_fence_signalled; + pws->fence_finish = nouveau_pipe_fence_finish; + pws->get_name = nouveau_get_name; return &nvpws->pws; -- cgit v1.2.3 From 833b1fb152851ba0d4fa2a5ba4702ee98d9bc217 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 31 Mar 2008 05:13:06 +1000 Subject: nv40: mark fp dirty even when only consts updated Fixes arbfplight "sticking". --- src/gallium/drivers/nv40/nv40_fragprog.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 87bca41cf64..4b7667e0381 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -919,6 +919,7 @@ nv40_fragprog_validate(struct nv40_context *nv40) nv40->constbuf[PIPE_SHADER_FRAGMENT]; struct pipe_winsys *ws = nv40->pipe.winsys; struct nouveau_stateobj *so; + boolean new_consts = FALSE; int i; if (fp->translated) @@ -945,7 +946,6 @@ nv40_fragprog_validate(struct nv40_context *nv40) update_constants: if (fp->nr_consts) { - boolean new_consts = FALSE; float *map; map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ); @@ -965,7 +965,7 @@ update_constants: nv40_fragprog_upload(nv40, fp); } - if (fp->so != nv40->state.hw[NV40_STATE_FRAGPROG]) { + if (new_consts || fp->so != nv40->state.hw[NV40_STATE_FRAGPROG]) { so_ref(fp->so, &nv40->state.hw[NV40_STATE_FRAGPROG]); return TRUE; } -- cgit v1.2.3 From 7b389f8d2f307fa0714494f2a43e9141cc04ed3e Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sun, 30 Mar 2008 21:52:36 +0200 Subject: nv30: use FREE macro --- src/gallium/drivers/nv30/nv30_context.c | 2 +- src/gallium/drivers/nv30/nv30_draw.c | 2 +- src/gallium/drivers/nv30/nv30_fragprog.c | 4 ++-- src/gallium/drivers/nv30/nv30_miptree.c | 6 +++--- src/gallium/drivers/nv30/nv30_query.c | 2 +- src/gallium/drivers/nv30/nv30_state.c | 4 ++-- src/gallium/drivers/nv30/nv30_vertprog.c | 8 ++++---- 7 files changed, 14 insertions(+), 14 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 1e729d789bd..b8e8b86d99a 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -41,7 +41,7 @@ nv30_destroy(struct pipe_context *pipe) nvws->grobj_free(&nv30->rankine); - free(nv30); + FREE(nv30); } static boolean diff --git a/src/gallium/drivers/nv30/nv30_draw.c b/src/gallium/drivers/nv30/nv30_draw.c index 59a72657996..8ec0835225b 100644 --- a/src/gallium/drivers/nv30/nv30_draw.c +++ b/src/gallium/drivers/nv30/nv30_draw.c @@ -40,7 +40,7 @@ nv30_draw_reset_stipple_counter(struct draw_stage *draw) static void nv30_draw_destroy(struct draw_stage *draw) { - free(draw); + FREE(draw); } struct draw_stage * diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 6f61d36f4e8..51000bd6fcd 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -753,7 +753,7 @@ nv30_fragprog_translate(struct nv30_context *nv30, fp->on_hw = FALSE; out_err: tgsi_parse_free(&parse); - free(fpc); + FREE(fpc); } void @@ -829,6 +829,6 @@ nv30_fragprog_destroy(struct nv30_context *nv30, struct nv30_fragment_program *fp) { if (fp->insn_len) - free(fp->insn); + FREE(fp->insn); } diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 6fdcf42a243..afb05fdd2bf 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -72,7 +72,7 @@ nv30_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, mt->total_size); if (!mt->buffer) { - free(mt); + FREE(mt); return NULL; } @@ -93,9 +93,9 @@ nv30_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) pipe_buffer_reference(ws, &nv30mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv30mt->level[l].image_offset) - free(nv30mt->level[l].image_offset); + FREE(nv30mt->level[l].image_offset); } - free(nv30mt); + FREE(nv30mt); } } diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c index e19cb455dce..0c2d941562c 100644 --- a/src/gallium/drivers/nv30/nv30_query.c +++ b/src/gallium/drivers/nv30/nv30_query.c @@ -35,7 +35,7 @@ nv30_query_destroy(struct pipe_context *pipe, struct pipe_query *pq) if (q->object) nv30->nvws->res_free(&q->object); - free(q); + FREE(q); } static void diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 983638adccb..620038fa647 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -486,7 +486,7 @@ nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv30_vertex_program *vp = hwcso; nv30_vertprog_destroy(nv30, vp); - free(vp); + FREE(vp); } static void * @@ -522,7 +522,7 @@ nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv30_fragment_program *fp = hwcso; nv30_fragprog_destroy(nv30, fp); - free(fp); + FREE(fp); } static void diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index e9b62ff48b4..fe1a467565d 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -578,7 +578,7 @@ nv30_vertprog_translate(struct nv30_context *nv30, vpc->high_temp = -1; if (!nv30_vertprog_prepare(vpc)) { - free(vpc); + FREE(vpc); return; } @@ -634,7 +634,7 @@ nv30_vertprog_translate(struct nv30_context *nv30, vp->translated = TRUE; out_err: tgsi_parse_free(&parse); - free(vpc); + FREE(vpc); } void @@ -790,8 +790,8 @@ void nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) { if (vp->nr_consts) - free(vp->consts); + FREE(vp->consts); if (vp->nr_insns) - free(vp->insns); + FREE(vp->insns); } -- cgit v1.2.3 From 6806519a0b2a8af0c950f71705b02e13876d460f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 31 Mar 2008 06:35:43 +1000 Subject: nv40: track current scissor enable/disable state Not sure how this was forgotten :) Anyhow, fixes gearbox/bzflag/xmoto, probably other things that use scissored clears / hit the nasty clear-with-quad path. --- src/gallium/drivers/nv40/nv40_state_scissor.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c index 9e9eadc5116..285239ef419 100644 --- a/src/gallium/drivers/nv40/nv40_state_scissor.c +++ b/src/gallium/drivers/nv40/nv40_state_scissor.c @@ -10,10 +10,11 @@ nv40_state_scissor_validate(struct nv40_context *nv40) if (nv40->state.hw[NV40_STATE_SCISSOR] && (rast->scissor == 0 && nv40->state.scissor_enabled == 0)) return FALSE; + nv40->state.scissor_enabled = rast->scissor; so = so_new(3, 0); so_method(so, nv40->screen->curie, NV40TCL_SCISSOR_HORIZ, 2); - if (rast->scissor) { + if (nv40->state.scissor_enabled) { so_data (so, ((s->maxx - s->minx) << 16) | s->minx); so_data (so, ((s->maxy - s->miny) << 16) | s->miny); } else { -- cgit v1.2.3 From 169faae6db9a289c5a2d9430d85c36ac36abd218 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 31 Mar 2008 09:00:25 +1000 Subject: nv40: support vp clip distance regs, unused currently. --- src/gallium/drivers/nv40/nv40_state.h | 1 + src/gallium/drivers/nv40/nv40_vertprog.c | 106 ++++++++++++++++++++++++++++++- 2 files changed, 104 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index ab2866eb7a5..e018464c9f8 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -42,6 +42,7 @@ struct nv40_vertex_program { uint32_t ir; uint32_t or; + uint32_t clip_ctrl; struct nouveau_stateobj *so; }; diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 40ef7174a41..5906280e5e1 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -37,6 +37,8 @@ #define neg(s) nv40_sr_neg((s)) #define abs(s) nv40_sr_abs((s)) +#define NV40_VP_INST_DEST_CLIP(n) ((~0 - 6) + (n)) + struct nv40_vpc { struct nv40_vertex_program *vp; @@ -200,6 +202,36 @@ emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst) case NV40_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; case NV40_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; case NV40_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + case NV40_VP_INST_DEST_CLIP(0): + vp->or |= (1 << 6); + vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE0; + dst.index = NV40_VP_INST_DEST_FOGC; + break; + case NV40_VP_INST_DEST_CLIP(1): + vp->or |= (1 << 7); + vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE1; + dst.index = NV40_VP_INST_DEST_FOGC; + break; + case NV40_VP_INST_DEST_CLIP(2): + vp->or |= (1 << 8); + vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE2; + dst.index = NV40_VP_INST_DEST_FOGC; + break; + case NV40_VP_INST_DEST_CLIP(3): + vp->or |= (1 << 9); + vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE3; + dst.index = NV40_VP_INST_DEST_PSZ; + break; + case NV40_VP_INST_DEST_CLIP(4): + vp->or |= (1 << 10); + vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE4; + dst.index = NV40_VP_INST_DEST_PSZ; + break; + case NV40_VP_INST_DEST_CLIP(5): + vp->or |= (1 << 11); + vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE5; + dst.index = NV40_VP_INST_DEST_PSZ; + break; default: break; } @@ -391,6 +423,11 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, int ai = -1, ci = -1, ii = -1; int i; + struct { + struct nv40_sreg dst; + unsigned c, m; + } clip; + if (finst->Instruction.Opcode == TGSI_OPCODE_END) return TRUE; @@ -464,6 +501,51 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + /* If writing to clip distance regs, need to modify instruction to + * change which component is written to. On NV40 the clip regs + * are the unused components (yzw) of FOGC/PSZ + */ + clip.dst = none; + if (dst.type == NV40SR_OUTPUT && + dst.index >= NV40_VP_INST_DEST_CLIP(0) && + dst.index <= NV40_VP_INST_DEST_CLIP(5)) { + unsigned n = dst.index - NV40_VP_INST_DEST_CLIP(0); + unsigned c[] = { SWZ_Y, SWZ_Z, SWZ_W, SWZ_Y, SWZ_Z, SWZ_W }; + unsigned m[] = + { MASK_Y, MASK_Z, MASK_W, MASK_Y, MASK_Z, MASK_W }; + + /* Some instructions we can get away with swizzling and/or + * changing the writemask. Others, we'll use a temp reg. + */ + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_DST: + case TGSI_OPCODE_EXP: + case TGSI_OPCODE_LIT: + case TGSI_OPCODE_LOG: + case TGSI_OPCODE_XPD: + clip.dst = dst; + clip.c = c[n]; + clip.m = m[n]; + dst = temp(vpc); + break; + case TGSI_OPCODE_DP3: + case TGSI_OPCODE_DP4: + case TGSI_OPCODE_DPH: + case TGSI_OPCODE_POW: + case TGSI_OPCODE_RCP: + case TGSI_OPCODE_RSQ: + mask = m[n]; + break; + default: + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + src[i] = nv40_sr_swz(src[i], + c[n], c[n], c[n], c[n]); + } + mask = m[n]; + break; + } + } + switch (finst->Instruction.Opcode) { case TGSI_OPCODE_ABS: arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); @@ -561,6 +643,12 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, return FALSE; } + if (clip.dst.type != NV40SR_NONE) { + arith(vpc, 0, OP_MOV, clip.dst, clip.m, + nv40_sr_swz(dst, clip.c, clip.c, clip.c, clip.c), + none, none); + } + release_temps(vpc); return TRUE; } @@ -612,6 +700,15 @@ nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, return FALSE; } break; +#if 0 + case TGSI_SEMANTIC_CLIP: + if (fdec->Semantic.SemanticIndex >= 6) { + NOUVEAU_ERR("bad clip distance index\n"); + return FALSE; + } + hw = NV40_VP_INST_DEST_CLIP(fdec->Semantic.SemanticIndex); + break; +#endif default: NOUVEAU_ERR("bad output semantic\n"); return FALSE; @@ -782,6 +879,7 @@ nv40_vertprog_validate(struct nv40_context *nv40) { struct nouveau_winsys *nvws = nv40->nvws; struct pipe_winsys *ws = nv40->pipe.winsys; + struct nouveau_grobj *curie = nv40->screen->curie; struct nv40_vertex_program *vp; struct pipe_buffer *constbuf; boolean upload_code = FALSE, upload_data = FALSE; @@ -825,12 +923,14 @@ check_gpu_resources: assert(0); } - so = so_new(5, 0); - so_method(so, nv40->screen->curie, NV40TCL_VP_START_FROM_ID, 1); + so = so_new(7, 0); + so_method(so, curie, NV40TCL_VP_START_FROM_ID, 1); so_data (so, vp->exec->start); - so_method(so, nv40->screen->curie, NV40TCL_VP_ATTRIB_EN, 2); + so_method(so, curie, NV40TCL_VP_ATTRIB_EN, 2); so_data (so, vp->ir); so_data (so, vp->or); + so_method(so, curie, NV40TCL_CLIP_PLANE_ENABLE, 1); + so_data (so, vp->clip_ctrl); so_ref(so, &vp->so); upload_code = TRUE; -- cgit v1.2.3 From 4ad9dd6179787a46ecb223ab0e59e6b25b9368af Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 31 Mar 2008 09:29:22 +1000 Subject: nouveau: update object header --- src/gallium/drivers/nouveau/nouveau_class.h | 14 ++++++++------ src/gallium/drivers/nv40/nv40_vbo.c | 2 +- 2 files changed, 9 insertions(+), 7 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index dc202086d2c..bf1e622d2c4 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -4249,6 +4249,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f #define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT 0x00000002 #define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE 0x00000004 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_USHORT 0x00000005 #define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT 4 #define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_MASK 0x000000f0 #define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 @@ -4963,12 +4964,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) #define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 #define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 1) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 5) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 9) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 13) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 17) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 21) #define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c #define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) #define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 @@ -4990,6 +4991,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f #define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 #define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 +#define NV40TCL_VTXFMT_TYPE_USHORT 0x00000005 #define NV40TCL_VTXFMT_SIZE_SHIFT 4 #define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 #define NV40TCL_VTXFMT_STRIDE_SHIFT 8 diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index b66bf26afb3..bc53924a676 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -32,7 +32,7 @@ nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) case PIPE_FORMAT_R16G16_SSCALED: case PIPE_FORMAT_R16G16B16_SSCALED: case PIPE_FORMAT_R16G16B16A16_SSCALED: - *fmt = 5; + *fmt = NV40TCL_VTXFMT_TYPE_USHORT; break; default: pf_sprint_name(fs, pipe); -- cgit v1.2.3 From e616d3f3e2178e34e4e7d769b38b0dff4ad615fe Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 31 Mar 2008 09:37:57 +1000 Subject: nv40: fix slight thinko --- src/gallium/drivers/nv40/nv40_vertprog.c | 15 +++++---------- 1 file changed, 5 insertions(+), 10 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 5906280e5e1..08d3f387e09 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -425,7 +425,7 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, struct { struct nv40_sreg dst; - unsigned c, m; + unsigned m; } clip; if (finst->Instruction.Opcode == TGSI_OPCODE_END) @@ -503,14 +503,13 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, /* If writing to clip distance regs, need to modify instruction to * change which component is written to. On NV40 the clip regs - * are the unused components (yzw) of FOGC/PSZ + * are the unused components (yzw) of FOGC/PSZ. */ clip.dst = none; if (dst.type == NV40SR_OUTPUT && dst.index >= NV40_VP_INST_DEST_CLIP(0) && dst.index <= NV40_VP_INST_DEST_CLIP(5)) { unsigned n = dst.index - NV40_VP_INST_DEST_CLIP(0); - unsigned c[] = { SWZ_Y, SWZ_Z, SWZ_W, SWZ_Y, SWZ_Z, SWZ_W }; unsigned m[] = { MASK_Y, MASK_Z, MASK_W, MASK_Y, MASK_Z, MASK_W }; @@ -524,7 +523,6 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, case TGSI_OPCODE_LOG: case TGSI_OPCODE_XPD: clip.dst = dst; - clip.c = c[n]; clip.m = m[n]; dst = temp(vpc); break; @@ -537,10 +535,8 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, mask = m[n]; break; default: - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - src[i] = nv40_sr_swz(src[i], - c[n], c[n], c[n], c[n]); - } + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) + src[i] = swz(src[i], X, X, X, X); mask = m[n]; break; } @@ -645,8 +641,7 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, if (clip.dst.type != NV40SR_NONE) { arith(vpc, 0, OP_MOV, clip.dst, clip.m, - nv40_sr_swz(dst, clip.c, clip.c, clip.c, clip.c), - none, none); + swz(dst, X, X, X, X), none, none); } release_temps(vpc); -- cgit v1.2.3 From 71d7aec2b3bb299f1dff7cdb2e8e96164976113a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 2 Apr 2008 00:40:18 +1000 Subject: nouveau: winsys printf disappeared :) --- src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index 453b3623f4a..b1bf9c521aa 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -20,15 +20,6 @@ nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, nouveau_copy_buffer(dPriv, surf, NULL); } -static void -nouveau_printf(struct pipe_winsys *pws, const char *fmt, ...) -{ - va_list args; - va_start(args, fmt); - vfprintf(stderr, fmt, args); - va_end(args); -} - static const char * nouveau_get_name(struct pipe_winsys *pws) { @@ -215,7 +206,6 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws = &nvpws->pws; pws->flush_frontbuffer = nouveau_flush_frontbuffer; - pws->printf = nouveau_printf; pws->surface_alloc = nouveau_surface_alloc; pws->surface_alloc_storage = nouveau_surface_alloc_storage; -- cgit v1.2.3 From bdf5b23bfd222ade9b3599ebd0f8932a5179431e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 2 Apr 2008 12:54:37 +1000 Subject: nv40: shorten zsa state lines --- src/gallium/drivers/nv40/nv40_state.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 1417c95e758..89f4078b720 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -445,19 +445,20 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nv40_context *nv40 = nv40_context(pipe); struct nv40_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso)); struct nouveau_stateobj *so = so_new(32, 0); + struct nouveau_grobj *curie = nv40->screen->curie; - so_method(so, nv40->screen->curie, NV40TCL_DEPTH_FUNC, 3); + so_method(so, curie, NV40TCL_DEPTH_FUNC, 3); so_data (so, nvgl_comparison_op(cso->depth.func)); so_data (so, cso->depth.writemask ? 1 : 0); so_data (so, cso->depth.enabled ? 1 : 0); - so_method(so, nv40->screen->curie, NV40TCL_ALPHA_TEST_ENABLE, 3); + so_method(so, curie, NV40TCL_ALPHA_TEST_ENABLE, 3); so_data (so, cso->alpha.enabled ? 1 : 0); so_data (so, nvgl_comparison_op(cso->alpha.func)); so_data (so, float_to_ubyte(cso->alpha.ref)); if (cso->stencil[0].enabled) { - so_method(so, nv40->screen->curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); + so_method(so, curie, NV40TCL_STENCIL_FRONT_ENABLE, 8); so_data (so, cso->stencil[0].enabled ? 1 : 0); so_data (so, cso->stencil[0].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[0].func)); @@ -467,12 +468,12 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); } else { - so_method(so, nv40->screen->curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); + so_method(so, curie, NV40TCL_STENCIL_FRONT_ENABLE, 1); so_data (so, 0); } if (cso->stencil[1].enabled) { - so_method(so, nv40->screen->curie, NV40TCL_STENCIL_BACK_ENABLE, 8); + so_method(so, curie, NV40TCL_STENCIL_BACK_ENABLE, 8); so_data (so, cso->stencil[1].enabled ? 1 : 0); so_data (so, cso->stencil[1].write_mask); so_data (so, nvgl_comparison_op(cso->stencil[1].func)); @@ -482,7 +483,7 @@ nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); } else { - so_method(so, nv40->screen->curie, NV40TCL_STENCIL_BACK_ENABLE, 1); + so_method(so, curie, NV40TCL_STENCIL_BACK_ENABLE, 1); so_data (so, 0); } -- cgit v1.2.3 From ae87909d0d261d0f4e888f6a167e6329eb129a87 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 2 Apr 2008 13:04:06 +1000 Subject: nv40: only update draw module state when using swtnl --- src/gallium/drivers/nv40/nv40_context.h | 6 ++++-- src/gallium/drivers/nv40/nv40_state.c | 22 ++++++++++------------ src/gallium/drivers/nv40/nv40_state_emit.c | 20 ++++++++++++++++++++ 3 files changed, 34 insertions(+), 14 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index b50f6f8fefc..525eef8d63d 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -133,7 +133,7 @@ struct nv40_context { unsigned fallback_swrast; /* Context state */ - unsigned dirty; + unsigned dirty, draw_dirty; struct pipe_scissor_state scissor; unsigned stipple[32]; struct pipe_clip_state clip; @@ -153,8 +153,10 @@ struct nv40_context { unsigned nr_samplers; unsigned nr_textures; unsigned dirty_samplers; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + unsigned vtxbuf_nr; struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; + unsigned vtxelt_nr; }; static INLINE struct nv40_context * diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 89f4078b720..5dc2991212d 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -423,10 +423,9 @@ nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) struct nv40_context *nv40 = nv40_context(pipe); struct nv40_rasterizer_state *rsso = hwcso; - draw_set_rasterizer_state(nv40->draw, &rsso->pipe); - nv40->rasterizer = hwcso; nv40->dirty |= NV40_NEW_RAST; + nv40->draw_dirty |= NV40_NEW_RAST; } static void @@ -530,10 +529,9 @@ nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) struct nv40_context *nv40 = nv40_context(pipe); struct nv40_vertex_program *vp = hwcso; - draw_bind_vertex_shader(nv40->draw, vp ? vp->draw : NULL); - nv40->vertprog = hwcso; nv40->dirty |= NV40_NEW_VERTPROG; + nv40->draw_dirty |= NV40_NEW_VERTPROG; } static void @@ -596,10 +594,9 @@ nv40_set_clip_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - draw_set_clip_state(nv40->draw, clip); - nv40->clip = *clip; nv40->dirty |= NV40_NEW_UCP; + nv40->draw_dirty |= NV40_NEW_UCP; } static void @@ -654,10 +651,9 @@ nv40_set_viewport_state(struct pipe_context *pipe, { struct nv40_context *nv40 = nv40_context(pipe); - draw_set_viewport_state(nv40->draw, vpt); - nv40->viewport = *vpt; nv40->dirty |= NV40_NEW_VIEWPORT; + nv40->draw_dirty |= NV40_NEW_VIEWPORT; } static void @@ -666,10 +662,11 @@ nv40_set_vertex_buffers(struct pipe_context *pipe, unsigned count, { struct nv40_context *nv40 = nv40_context(pipe); - draw_set_vertex_buffers(nv40->draw, count, vb); - memcpy(nv40->vtxbuf, vb, sizeof(*vb) * count); + nv40->vtxbuf_nr = count; + nv40->dirty |= NV40_NEW_ARRAYS; + nv40->draw_dirty |= NV40_NEW_ARRAYS; } static void @@ -678,10 +675,11 @@ nv40_set_vertex_elements(struct pipe_context *pipe, unsigned count, { struct nv40_context *nv40 = nv40_context(pipe); - draw_set_vertex_elements(nv40->draw, count, ve); - memcpy(nv40->vtxelt, ve, sizeof(*ve) * count); + nv40->vtxelt_nr = count; + nv40->dirty |= NV40_NEW_ARRAYS; + nv40->draw_dirty |= NV40_NEW_ARRAYS; } void diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 74feb6d4bfc..722b9f31e6d 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -144,6 +144,8 @@ nv40_state_validate(struct nv40_context *nv40) boolean nv40_state_validate_swtnl(struct nv40_context *nv40) { + struct draw_context *draw = nv40->draw; + /* Setup for swtnl */ if (nv40->render_mode == HW) { NOUVEAU_ERR("hw->swtnl 0x%08x\n", nv40->fallback_swtnl); @@ -155,12 +157,30 @@ nv40_state_validate_swtnl(struct nv40_context *nv40) nv40->render_mode = SWTNL; } + if (nv40->draw_dirty & NV40_NEW_VERTPROG) + draw_bind_vertex_shader(draw, nv40->vertprog->draw); + + if (nv40->draw_dirty & NV40_NEW_RAST) + draw_set_rasterizer_state(draw, &nv40->rasterizer->pipe); + + if (nv40->draw_dirty & NV40_NEW_UCP) + draw_set_clip_state(draw, &nv40->clip); + + if (nv40->draw_dirty & NV40_NEW_VIEWPORT) + draw_set_viewport_state(draw, &nv40->viewport); + + if (nv40->draw_dirty & NV40_NEW_ARRAYS) { + draw_set_vertex_buffers(draw, nv40->vtxbuf_nr, nv40->vtxbuf); + draw_set_vertex_elements(draw, nv40->vtxelt_nr, nv40->vtxelt); + } + nv40_state_do_validate(nv40, swtnl_states); if (nv40->fallback_swrast) { NOUVEAU_ERR("swtnl->swrast 0x%08x\n", nv40->fallback_swrast); return FALSE; } + nv40->draw_dirty = 0; return TRUE; } -- cgit v1.2.3 From b1a361ba7a565063200c033e4939e6b28c006b13 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Wed, 2 Apr 2008 05:10:18 +0200 Subject: nv10: fix stuff and things. --- src/gallium/drivers/nouveau/nouveau_class.h | 24 +++- src/gallium/drivers/nv10/nv10_context.c | 92 +++--------- src/gallium/drivers/nv10/nv10_context.h | 31 +++- src/gallium/drivers/nv10/nv10_miptree.c | 28 ++-- src/gallium/drivers/nv10/nv10_prim_vbuf.c | 17 ++- src/gallium/drivers/nv10/nv10_screen.c | 70 ++++++--- src/gallium/drivers/nv10/nv10_screen.h | 6 +- src/gallium/drivers/nv10/nv10_state.c | 155 ++++---------------- src/gallium/drivers/nv10/nv10_state_emit.c | 216 ++++++++++++++++++++++++++-- 9 files changed, 395 insertions(+), 244 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index dc202086d2c..880afe6ce08 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -2804,6 +2804,16 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_VERTEX_WGH_1F 0x00000ce4 #define NV10TCL_EDGEFLAG_ENABLE 0x00000cec #define NV10TCL_VERTEX_ARRAY_VALIDATE 0x00000cf0 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(x) (0x00000d00+((x)*8)) +#define NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET__SIZE 0x00000008 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(x) (0x00000d04+((x)*8)) +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT__SIZE 0x00000008 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_TYPE_SHIFT 0 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_TYPE_MASK 0x0000000f +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_FIELDS_SHIFT 4 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_FIELDS_MASK 0x000000f0 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_STRIDE_SHIFT 8 +#define NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_STRIDE_MASK 0x0000ff00 #define NV10TCL_VERTEX_ARRAY_OFFSET_POS 0x00000d00 #define NV10TCL_VERTEX_ARRAY_FORMAT_POS 0x00000d04 #define NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT 0 @@ -4249,6 +4259,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f #define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT 0x00000002 #define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE 0x00000004 +#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_USHORT 0x00000005 #define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT 4 #define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_MASK 0x000000f0 #define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 @@ -4963,12 +4974,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) #define NV40TCL_VP_UPLOAD_INST__SIZE 0x00000004 #define NV40TCL_CLIP_PLANE_ENABLE 0x00001478 -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 2) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 6) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 10) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 14) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 18) -#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 22) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE0 (1 << 1) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE1 (1 << 5) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE2 (1 << 9) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE3 (1 << 13) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE4 (1 << 17) +#define NV40TCL_CLIP_PLANE_ENABLE_PLANE5 (1 << 21) #define NV40TCL_POLYGON_STIPPLE_ENABLE 0x0000147c #define NV40TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) #define NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 @@ -4990,6 +5001,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VTXFMT_TYPE_MASK 0x0000000f #define NV40TCL_VTXFMT_TYPE_FLOAT 0x00000002 #define NV40TCL_VTXFMT_TYPE_UBYTE 0x00000004 +#define NV40TCL_VTXFMT_TYPE_USHORT 0x00000005 #define NV40TCL_VTXFMT_SIZE_SHIFT 4 #define NV40TCL_VTXFMT_SIZE_MASK 0x000000f0 #define NV40TCL_VTXFMT_STRIDE_SHIFT 8 diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index 14042fb2fbb..42c496b9590 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -12,13 +12,6 @@ nv10_flush(struct pipe_context *pipe, unsigned flags, { struct nv10_context *nv10 = nv10_context(pipe); - if (flags & PIPE_FLUSH_TEXTURE_CACHE) { - BEGIN_RING(celsius, 0x1fd8, 1); - OUT_RING (2); - BEGIN_RING(celsius, 0x1fd8, 1); - OUT_RING (1); - } - FIRE_RING(fence); } @@ -26,36 +19,21 @@ static void nv10_destroy(struct pipe_context *pipe) { struct nv10_context *nv10 = nv10_context(pipe); - struct nouveau_winsys *nvws = nv10->nvws; if (nv10->draw) draw_destroy(nv10->draw); - nvws->res_free(&nv10->vertprog.exec_heap); - nvws->res_free(&nv10->vertprog.data_heap); - - nvws->notifier_free(&nv10->sync); - - nvws->grobj_free(&nv10->celsius); - - free(nv10); + FREE(nv10); } -static boolean -nv10_init_hwctx(struct nv10_context *nv10, int celsius_class) +static void nv10_init_hwctx(struct nv10_context *nv10) { - struct nouveau_winsys *nvws = nv10->nvws; - int ret; + struct nv10_screen *screen = nv10->screen; + struct nouveau_winsys *nvws = screen->nvws; int i; - ret = nvws->grobj_alloc(nvws, celsius_class, &nv10->celsius); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); - OUT_RING (nv10->sync->handle); + OUT_RING (screen->sync->handle); BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2); OUT_RING (nvws->channel->vram->handle); OUT_RING (nvws->channel->gart->handle); @@ -90,7 +68,7 @@ nv10_init_hwctx(struct nv10_context *nv10, int celsius_class) BEGIN_RING(celsius, NV10TCL_NOP, 1); OUT_RING (0); - if (celsius_class != NV10TCL) { + if (nv10->screen->celsius->grclass != NV10TCL) { /* For nv11, nv17 */ BEGIN_RING(celsius, 0x120, 3); OUT_RING (0); @@ -243,72 +221,44 @@ nv10_init_hwctx(struct nv10_context *nv10, int celsius_class) FIRE_RING (NULL); - return TRUE; } struct pipe_context * -nv10_create(struct pipe_screen *screen, unsigned pctx_id) +nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) { - struct pipe_winsys *pipe_winsys = screen->winsys; - struct nouveau_winsys *nvws = nv10_screen(screen)->nvws; - unsigned chipset = nv10_screen(screen)->chipset; + struct nv10_screen *screen = nv10_screen(pscreen); + struct pipe_winsys *ws = pscreen->winsys; struct nv10_context *nv10; - int celsius_class = 0, ret; - - if (chipset>=0x20) - celsius_class=NV11TCL; - else if (chipset>=0x17) - celsius_class=NV17TCL; - else if (chipset>=0x11) - celsius_class=NV11TCL; - else - celsius_class=NV10TCL; - - nv10 = CALLOC_STRUCT(nv10_context); + unsigned chipset = screen->chipset; + struct nouveau_winsys *nvws = screen->nvws; + + nv10 = CALLOC(1, sizeof(struct nv10_context)); if (!nv10) return NULL; + nv10->screen = screen; + nv10->pctx_id = pctx_id; + nv10->chipset = chipset; nv10->nvws = nvws; - /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &nv10->sync); - if (ret) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv10_destroy(&nv10->pipe); - return NULL; - } - - /* Vtxprog resources */ - if (nvws->res_init(&nv10->vertprog.exec_heap, 0, 512) || - nvws->res_init(&nv10->vertprog.data_heap, 0, 256)) { - nv10_destroy(&nv10->pipe); - return NULL; - } - - /* Static celsius initialisation */ - if (!nv10_init_hwctx(nv10, celsius_class)) { - nv10_destroy(&nv10->pipe); - return NULL; - } - - /* Pipe context setup */ - nv10->pipe.winsys = pipe_winsys; - + nv10->pipe.winsys = ws; + nv10->pipe.screen = pscreen; nv10->pipe.destroy = nv10_destroy; - nv10->pipe.draw_arrays = nv10_draw_arrays; nv10->pipe.draw_elements = nv10_draw_elements; nv10->pipe.clear = nv10_clear; - nv10->pipe.flush = nv10_flush; nv10_init_surface_functions(nv10); nv10_init_state_functions(nv10); + nv10_init_miptree_functions(nv10); nv10->draw = draw_create(); assert(nv10->draw); draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10)); + nv10_init_hwctx(nv10); + return &nv10->pipe; } diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 386138556e5..61eb4e6a937 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -11,7 +11,7 @@ #include "nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv10_context *ctx = nv10 + struct nv10_screen *ctx = nv10->screen #include "nouveau/nouveau_push.h" #include "nv10_state.h" @@ -24,17 +24,27 @@ #define NV10_NEW_VERTPROG (1 << 1) #define NV10_NEW_FRAGPROG (1 << 2) #define NV10_NEW_ARRAYS (1 << 3) -#define NV10_NEW_VBO (1 << 4) +#define NV10_NEW_VTXFMT (1 << 4) +#define NV10_NEW_BLEND (1 << 5) +#define NV10_NEW_BLENDCOL (1 << 6) +#define NV10_NEW_RAST (1 << 7) +#define NV10_NEW_DSA (1 << 8) +#define NV10_NEW_VIEWPORT (1 << 9) +#define NV10_NEW_SCISSOR (1 << 9) +#define NV10_NEW_FRAMEBUFFER (1 << 10) + +#include "nv10_screen.h" struct nv10_context { struct pipe_context pipe; + struct nouveau_winsys *nvws; + struct nv10_screen *screen; + unsigned pctx_id; struct draw_context *draw; int chipset; - struct nouveau_grobj *celsius; - struct nouveau_notifier *sync; uint32_t dirty; @@ -49,6 +59,14 @@ struct nv10_context { struct pipe_buffer *zeta; uint32_t lma_offset; + struct nv10_blend_state *blend; + struct pipe_blend_color *blend_color; + struct nv10_rasterizer_state *rast; + struct nv10_depth_stencil_alpha_state *dsa; + struct pipe_viewport_state *viewport; + struct pipe_scissor_state *scissor; + struct pipe_framebuffer_state *framebuffer; + struct { struct pipe_buffer *buffer; uint32_t format; @@ -91,7 +109,9 @@ nv10_context(struct pipe_context *pipe) extern void nv10_init_state_functions(struct nv10_context *nv10); extern void nv10_init_surface_functions(struct nv10_context *nv10); -extern void nv10_init_miptree_functions(struct pipe_screen *screen); +extern void nv10_init_miptree_functions(struct nv10_context *nv10); + +extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen); /* nv10_clear.c */ extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, @@ -111,6 +131,7 @@ extern void nv10_fragtex_bind(struct nv10_context *); /* nv10_prim_vbuf.c */ struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ); +extern void nv10_vtxbuf_bind(struct nv10_context* nv10); /* nv10_state.c and friends */ extern void nv10_emit_hw_state(struct nv10_context *nv10); diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index 7b7f39b80c9..4dfc675a6b9 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -69,7 +69,7 @@ nv10_miptree_create(struct pipe_screen *screen, struct pipe_texture *pt) mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, mt->total_size); if (!mt->buffer) { - free(mt); + FREE(mt); return NULL; } @@ -90,12 +90,19 @@ nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) pipe_buffer_reference(ws, &nv10mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv10mt->level[l].image_offset) - free(nv10mt->level[l].image_offset); + FREE(nv10mt->level[l].image_offset); } - free(nv10mt); + FREE(nv10mt); } } +static void +nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, + uint face, uint levels) +{ +} + + static struct pipe_surface * nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) @@ -122,13 +129,16 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, return ps; } -void -nv10_init_miptree_functions(struct pipe_screen *screen) + +void nv10_init_miptree_functions(struct nv10_context *nv10) { - struct nv10_screen *nv10screen = nv10_screen(screen); + nv10->pipe.texture_update = nv10_miptree_update; +} - nv10screen->screen.texture_create = nv10_miptree_create; - nv10screen->screen.texture_release = nv10_miptree_release; - nv10screen->screen.get_tex_surface = nv10_miptree_surface_get; +void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv10_miptree_create; + pscreen->texture_release = nv10_miptree_release; + pscreen->get_tex_surface = nv10_miptree_surface_get; } diff --git a/src/gallium/drivers/nv10/nv10_prim_vbuf.c b/src/gallium/drivers/nv10/nv10_prim_vbuf.c index 15268912234..bbcfe1a2fdd 100644 --- a/src/gallium/drivers/nv10/nv10_prim_vbuf.c +++ b/src/gallium/drivers/nv10/nv10_prim_vbuf.c @@ -67,6 +67,17 @@ struct nv10_vbuf_render { }; +void nv10_vtxbuf_bind( struct nv10_context* nv10 ) +{ + int i; + for(i = 0; i < 8; i++) { + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1); + OUT_RING(0/*nv10->vtxbuf*/); + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1); + OUT_RING(0/*XXX*/); + } +} + /** * Basically a cast wrapper. */ @@ -100,7 +111,7 @@ nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, assert(!nv10_render->buffer); nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); - nv10->dirty |= NV10_NEW_VBO; + nv10->dirty |= NV10_NEW_ARRAYS; return winsys->buffer_map(winsys, nv10_render->buffer, @@ -139,8 +150,8 @@ nv10_vbuf_render_draw( struct vbuf_render *render, } while (nr_indices) { - // XXX too big ? - push = MIN2(nr_indices, 2047 * 2); + // XXX too big/small ? check the size + push = MIN2(nr_indices, 1200 * 2); BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 4d8e0b05ccd..80676ead1a9 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -116,35 +116,71 @@ nv10_screen_is_format_supported(struct pipe_screen *screen, } static void -nv10_screen_destroy(struct pipe_screen *screen) +nv10_screen_destroy(struct pipe_screen *pscreen) { - FREE(screen); + struct nv10_screen *screen = nv10_screen(pscreen); + struct nouveau_winsys *nvws = screen->nvws; + + nvws->notifier_free(&screen->sync); + nvws->grobj_free(&screen->celsius); + + FREE(pscreen); } struct pipe_screen * -nv10_screen_create(struct pipe_winsys *winsys, struct nouveau_winsys *nvws, +nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, unsigned chipset) { - struct nv10_screen *nv10screen = CALLOC_STRUCT(nv10_screen); + struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen); + unsigned celsius_class; + int ret; - if (!nv10screen) + if (!screen) + return NULL; + screen->chipset = chipset; + screen->nvws = nvws; + + /* 3D object */ + if (chipset>=0x20) + celsius_class=NV11TCL; + else if (chipset>=0x17) + celsius_class=NV17TCL; + else if (chipset>=0x11) + celsius_class=NV11TCL; + else + celsius_class=NV10TCL; + + if (!celsius_class) { + NOUVEAU_ERR("Unknown nv1x chipset: nv%02x\n", chipset); return NULL; + } + + ret = nvws->grobj_alloc(nvws, celsius_class, &screen->celsius); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv10_screen_destroy(&screen->pipe); + return NULL; + } - nv10screen->chipset = chipset; - nv10screen->nvws = nvws; + screen->pipe.winsys = ws; + screen->pipe.destroy = nv10_screen_destroy; - nv10screen->screen.winsys = winsys; + screen->pipe.get_name = nv10_screen_get_name; + screen->pipe.get_vendor = nv10_screen_get_vendor; + screen->pipe.get_param = nv10_screen_get_param; + screen->pipe.get_paramf = nv10_screen_get_paramf; - nv10screen->screen.destroy = nv10_screen_destroy; + screen->pipe.is_format_supported = nv10_screen_is_format_supported; - nv10screen->screen.get_name = nv10_screen_get_name; - nv10screen->screen.get_vendor = nv10_screen_get_vendor; - nv10screen->screen.get_param = nv10_screen_get_param; - nv10screen->screen.get_paramf = nv10_screen_get_paramf; - nv10screen->screen.is_format_supported = - nv10_screen_is_format_supported; + nv10_screen_init_miptree_functions(&screen->pipe); - nv10_init_miptree_functions(&nv10screen->screen); - return &nv10screen->screen; + return &screen->pipe; } diff --git a/src/gallium/drivers/nv10/nv10_screen.h b/src/gallium/drivers/nv10/nv10_screen.h index ac8c04072ab..4192fe11ef4 100644 --- a/src/gallium/drivers/nv10/nv10_screen.h +++ b/src/gallium/drivers/nv10/nv10_screen.h @@ -4,10 +4,14 @@ #include "pipe/p_screen.h" struct nv10_screen { - struct pipe_screen screen; + struct pipe_screen pipe; struct nouveau_winsys *nvws; unsigned chipset; + + /* HW graphics objects */ + struct nouveau_grobj *celsius; + struct nouveau_notifier *sync; }; static INLINE struct nv10_screen * diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 1ff01de1060..12722709f6b 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -35,6 +35,8 @@ static void nv10_vertex_layout(struct pipe_context* pipe) } } draw_compute_vertex_size(&vinfo); + + nv10->dirty |= NV10_NEW_VTXFMT; } static void * @@ -62,27 +64,19 @@ nv10_blend_state_create(struct pipe_context *pipe, } static void -nv10_blend_state_bind(struct pipe_context *pipe, void *hwcso) +nv10_blend_state_bind(struct pipe_context *pipe, void *blend) { struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_blend_state *cb = hwcso; - - BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); - OUT_RING (cb->d_enable); - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); - OUT_RING (cb->b_enable); - OUT_RING (cb->b_srcfunc); - OUT_RING (cb->b_dstfunc); + nv10->blend = (struct nv10_blend_state*)blend; - BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); - OUT_RING (cb->c_mask); + nv10->dirty |= NV10_NEW_BLEND; } static void nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + FREE(hwcso); } @@ -255,7 +249,7 @@ nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) static void nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + FREE(hwcso); } static void @@ -347,43 +341,19 @@ nv10_rasterizer_state_create(struct pipe_context *pipe, } static void -nv10_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) +nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast) { struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_rasterizer_state *rs = hwcso; - - BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); - OUT_RING (rs->shade_model); - OUT_RING (rs->line_width); - - - BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); - OUT_RING (rs->point_size); - - BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); - OUT_RING (rs->poly_mode_front); - OUT_RING (rs->poly_mode_back); - - BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); - OUT_RING (rs->cull_face); - OUT_RING (rs->front_face); + nv10->rast = (struct nv10_rasterizer_state*)rast; - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); - OUT_RING (rs->line_smooth_en); - OUT_RING (rs->poly_smooth_en); - - BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); - OUT_RING (rs->cull_face_en); - -/* BEGIN_RING(celsius, NV10TCL_POINT_SPRITE, 1); - OUT_RING (rs->point_sprite);*/ + nv10->dirty |= NV10_NEW_RAST; } static void nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + FREE(hwcso); } static void * @@ -415,25 +385,19 @@ nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, } static void -nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa) { struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_depth_stencil_alpha_state *hw = hwcso; - - BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&hw->depth, 3); - BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); - OUT_RING (hw->stencil.enable); - BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); - OUT_RINGp ((uint32_t *)&(hw->stencil.wmask), 7); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 3); - OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); + + nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa; + + nv10->dirty |= NV10_NEW_DSA; } static void nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + FREE(hwcso); } static void * @@ -461,11 +425,11 @@ nv10_vp_state_bind(struct pipe_context *pipe, void *hwcso) static void nv10_vp_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nv10_context *nv10 = nv10_context(pipe); + //struct nv10_context *nv10 = nv10_context(pipe); struct nv10_vertex_program *vp = hwcso; //nv10_vertprog_destroy(nv10, vp); - free(vp); + FREE(vp); } static void * @@ -499,7 +463,7 @@ nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv10_fragment_program *fp = hwcso; nv10_fragprog_destroy(nv10, fp); - free(fp); + FREE(fp); } static void @@ -508,11 +472,9 @@ nv10_set_blend_color(struct pipe_context *pipe, { struct nv10_context *nv10 = nv10_context(pipe); - BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); - OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0)); + nv10->blend_color = (struct pipe_blend_color*)bcol; + + nv10->dirty |= NV10_NEW_BLENDCOL; } static void @@ -542,58 +504,10 @@ nv10_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv10_context *nv10 = nv10_context(pipe); - struct pipe_surface *rt, *zeta; - uint32_t rt_format, w, h; - int i, colour_format = 0, zeta_format = 0; - - w = fb->cbufs[0]->width; - h = fb->cbufs[0]->height; - colour_format = fb->cbufs[0]->format; - rt = fb->cbufs[0]; - - if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - - zeta_format = fb->zsbuf->format; - zeta = fb->zsbuf; - } - rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR; + nv10->framebuffer = (struct pipe_framebuffer_state*)fb; - switch (colour_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case 0: - rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8; - break; - case PIPE_FORMAT_R5G6B5_UNORM: - rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5; - break; - default: - assert(0); - } - - BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); - nv10->rt[0] = rt->buffer; - - if (zeta_format) - { - nv10->zeta = zeta->buffer; - } - - BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - OUT_RING (rt_format); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0); - OUT_RING (((h - 1) << 16) | 0); + nv10->dirty |= NV10_NEW_FRAMEBUFFER; } static void @@ -609,10 +523,9 @@ nv10_set_scissor_state(struct pipe_context *pipe, { struct nv10_context *nv10 = nv10_context(pipe); - // XXX -/* BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); - OUT_RING (((s->maxx - s->minx) << 16) | s->minx); - OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ + nv10->scissor = (struct pipe_scissor_state*)s; + + nv10->dirty |= NV10_NEW_SCISSOR; } static void @@ -621,15 +534,9 @@ nv10_set_viewport_state(struct pipe_context *pipe, { struct nv10_context *nv10 = nv10_context(pipe); -/* OUT_RINGf (vpt->translate[0]); - OUT_RINGf (vpt->translate[1]); - OUT_RINGf (vpt->translate[2]); - OUT_RINGf (vpt->translate[3]);*/ - BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); - OUT_RINGf (vpt->scale[0]); - OUT_RINGf (vpt->scale[1]); - OUT_RINGf (vpt->scale[2]); - OUT_RINGf (vpt->scale[3]); + nv10->viewport = (struct pipe_viewport_state*)vpt; + + nv10->dirty |= NV10_NEW_VIEWPORT; } static void diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index 8bf0bd2d683..0c524963896 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -1,14 +1,174 @@ +#include "pipe/p_util.h" + #include "nv10_context.h" #include "nv10_state.h" +static void nv10_state_emit_blend(struct nv10_context* nv10) +{ + struct nv10_blend_state *b = nv10->blend; + + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); + OUT_RING (b->d_enable); + + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); + OUT_RING (b->b_enable); + OUT_RING (b->b_srcfunc); + OUT_RING (b->b_dstfunc); + + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (b->c_mask); +} + +static void nv10_state_emit_blend_color(struct nv10_context* nv10) +{ + struct pipe_blend_color *c = nv10->blend_color; + + BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); + OUT_RING ((float_to_ubyte(c->color[3]) << 24)| + (float_to_ubyte(c->color[0]) << 16)| + (float_to_ubyte(c->color[1]) << 8) | + (float_to_ubyte(c->color[2]) << 0)); +} + +static void nv10_state_emit_rast(struct nv10_context* nv10) +{ + struct nv10_rasterizer_state *r = nv10->rast; + + BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); + OUT_RING (r->shade_model); + OUT_RING (r->line_width); + + + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (r->point_size); + + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (r->poly_mode_front); + OUT_RING (r->poly_mode_back); + + + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (r->cull_face); + OUT_RING (r->front_face); + + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); + OUT_RING (r->line_smooth_en); + OUT_RING (r->poly_smooth_en); + + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (r->cull_face_en); +} + +static void nv10_state_emit_dsa(struct nv10_context* nv10) +{ + struct nv10_depth_stencil_alpha_state *d = nv10->dsa; + + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 3); + OUT_RINGp ((uint32_t *)&d->depth, 3); + BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); + OUT_RING (d->stencil.enable); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); + OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 3); + OUT_RINGp ((uint32_t *)&d->alpha.enabled, 3); +} + +static void nv10_state_emit_viewport(struct nv10_context* nv10) +{ + struct pipe_viewport_state *vpt = nv10->viewport; + +/* OUT_RINGf (vpt->translate[0]); + OUT_RINGf (vpt->translate[1]); + OUT_RINGf (vpt->translate[2]); + OUT_RINGf (vpt->translate[3]);*/ + BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); + OUT_RINGf (vpt->scale[0]); + OUT_RINGf (vpt->scale[1]); + OUT_RINGf (vpt->scale[2]); + OUT_RINGf (vpt->scale[3]); +} + +static void nv10_state_emit_scissor(struct nv10_context* nv10) +{ + // XXX this is so not working +/* struct pipe_scissor_state *s = nv10->scissor; + BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ +} + +static void nv10_state_emit_framebuffer(struct nv10_context* nv10) +{ + struct pipe_framebuffer_state* fb = nv10->framebuffer; + struct pipe_surface *rt, *zeta; + uint32_t rt_format, w, h; + int colour_format = 0, zeta_format = 0; + + w = fb->cbufs[0]->width; + h = fb->cbufs[0]->height; + colour_format = fb->cbufs[0]->format; + rt = fb->cbufs[0]; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + nv10->rt[0] = rt->buffer; + + if (zeta_format) + { + nv10->zeta = zeta->buffer; + } + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((w - 1) << 16) | 0); + OUT_RING (((h - 1) << 16) | 0); +} + void nv10_emit_hw_state(struct nv10_context *nv10) { int i; + if (nv10->dirty & NV10_NEW_VERTPROG) { + //nv10_vertprog_bind(nv10, nv10->vertprog.current); + nv10->dirty &= ~NV10_NEW_VERTPROG; + } + if (nv10->dirty & NV10_NEW_FRAGPROG) { nv10_fragprog_bind(nv10, nv10->fragprog.current); /*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */ + nv10->dirty_samplers |= (1<<10); + nv10->dirty_samplers = 0; } if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) { @@ -16,16 +176,50 @@ nv10_emit_hw_state(struct nv10_context *nv10) nv10->dirty &= ~NV10_NEW_FRAGPROG; } - if (nv10->dirty & NV10_NEW_VERTPROG) { - //nv10_vertprog_bind(nv10, nv10->vertprog.current); - nv10->dirty &= ~NV10_NEW_VERTPROG; + if (nv10->dirty & NV10_NEW_ARRAYS) { + nv10->dirty &= ~NV10_NEW_ARRAYS; + // array state will be put here once it's not emitted at each frame } - if (nv10->dirty & NV10_NEW_VBO) { - + if (nv10->dirty & NV10_NEW_VTXFMT) { + nv10->dirty &= ~NV10_NEW_VTXFMT; + nv10_vtxbuf_bind(nv10); } - nv10->dirty_samplers = 0; + if (nv10->dirty & NV10_NEW_BLEND) { + nv10->dirty &= ~NV10_NEW_BLEND; + nv10_state_emit_blend(nv10); + } + + if (nv10->dirty & NV10_NEW_BLENDCOL) { + nv10->dirty &= ~NV10_NEW_BLENDCOL; + nv10_state_emit_blend_color(nv10); + } + + if (nv10->dirty & NV10_NEW_RAST) { + nv10->dirty &= ~NV10_NEW_RAST; + nv10_state_emit_rast(nv10); + } + + if (nv10->dirty & NV10_NEW_DSA) { + nv10->dirty &= ~NV10_NEW_DSA; + nv10_state_emit_dsa(nv10); + } + + if (nv10->dirty & NV10_NEW_VIEWPORT) { + nv10->dirty &= ~NV10_NEW_VIEWPORT; + nv10_state_emit_viewport(nv10); + } + + if (nv10->dirty & NV10_NEW_SCISSOR) { + nv10->dirty &= ~NV10_NEW_SCISSOR; + nv10_state_emit_scissor(nv10); + } + + if (nv10->dirty & NV10_NEW_FRAMEBUFFER) { + nv10->dirty &= ~NV10_NEW_FRAMEBUFFER; + nv10_state_emit_framebuffer(nv10); + } /* Emit relocs for every referenced buffer. * This is to ensure the bufmgr has an accurate idea of how @@ -48,10 +242,16 @@ nv10_emit_hw_state(struct nv10_context *nv10) BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1); OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); /* XXX for when we allocate LMA on nv17 */ -/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_OFFSET, 1); - OUT_RELOCl(nv10->zeta+lma_offset);*/ +/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1); + OUT_RELOCl(nv10->zeta + lma_offset);*/ } + /* Vertex buffer */ + BEGIN_RING(celsius, NV10TCL_DMA_VTXBUF0, 1); + OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + /* Texture images */ for (i = 0; i < 2; i++) { if (!(nv10->fp_samplers & (1 << i))) -- cgit v1.2.3 From 8f26e975ca6341cb3366a18beb352b5cdcaee2bc Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Wed, 2 Apr 2008 15:09:32 +0200 Subject: nv10: set rasterizer state. --- src/gallium/drivers/nv10/nv10_state.c | 4 ++++ src/gallium/drivers/nv10/nv10_state.h | 2 ++ 2 files changed, 6 insertions(+) diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 12722709f6b..182f6857f5b 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -282,6 +282,8 @@ nv10_rasterizer_state_create(struct pipe_context *pipe, */ rs = malloc(sizeof(struct nv10_rasterizer_state)); + rs->templ = cso; + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; @@ -347,6 +349,8 @@ nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast) nv10->rast = (struct nv10_rasterizer_state*)rast; + draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL)); + nv10->dirty |= NV10_NEW_RAST; } diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h index 9bda8a7d6a3..3ca501d1356 100644 --- a/src/gallium/drivers/nv10/nv10_state.h +++ b/src/gallium/drivers/nv10/nv10_state.h @@ -39,6 +39,8 @@ struct nv10_rasterizer_state { uint32_t cull_face_en; uint32_t point_sprite; + + const struct pipe_rasterizer_state *templ; }; struct nv10_vertex_program_exec { -- cgit v1.2.3 From fb19b3393fbee26f7bb88b572b3d0cc2943f2edc Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Wed, 2 Apr 2008 18:26:49 +0200 Subject: nv10: fix some more state, work on the vertex emission code. --- src/gallium/auxiliary/draw/draw_vertex_fetch.c | 2 + src/gallium/drivers/nv10/nv10_context.h | 21 +++++---- src/gallium/drivers/nv10/nv10_fragtex.c | 4 ++ src/gallium/drivers/nv10/nv10_prim_vbuf.c | 7 ++- src/gallium/drivers/nv10/nv10_state.c | 60 ++++++-------------------- src/gallium/drivers/nv10/nv10_state_emit.c | 39 ++++++++++++++--- src/gallium/drivers/nv10/nv10_vbo.c | 2 + 7 files changed, 68 insertions(+), 67 deletions(-) diff --git a/src/gallium/auxiliary/draw/draw_vertex_fetch.c b/src/gallium/auxiliary/draw/draw_vertex_fetch.c index 11f99babf65..9a9ac4ee451 100644 --- a/src/gallium/auxiliary/draw/draw_vertex_fetch.c +++ b/src/gallium/auxiliary/draw/draw_vertex_fetch.c @@ -508,6 +508,7 @@ void draw_update_vertex_fetch( struct draw_context *draw ) draw->vertex_fetch.nr_attrs = nr_attrs; draw->vertex_fetch.fetch_func = generic_vertex_fetch; + printf("pouet vertex fetch %x\n",draw->vertex_fetch.fetch_func); switch (nr_attrs) { case 2: @@ -524,5 +525,6 @@ void draw_update_vertex_fetch( struct draw_context *draw ) default: break; } + printf("pouet vertex fetch %x\n",draw->vertex_fetch.fetch_func); } diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 61eb4e6a937..8c13d6897f7 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -21,17 +21,16 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV10_NEW_VERTPROG (1 << 1) -#define NV10_NEW_FRAGPROG (1 << 2) -#define NV10_NEW_ARRAYS (1 << 3) -#define NV10_NEW_VTXFMT (1 << 4) -#define NV10_NEW_BLEND (1 << 5) -#define NV10_NEW_BLENDCOL (1 << 6) -#define NV10_NEW_RAST (1 << 7) -#define NV10_NEW_DSA (1 << 8) -#define NV10_NEW_VIEWPORT (1 << 9) -#define NV10_NEW_SCISSOR (1 << 9) -#define NV10_NEW_FRAMEBUFFER (1 << 10) +#define NV10_NEW_VERTPROG (1 << 0) +#define NV10_NEW_FRAGPROG (1 << 1) +#define NV10_NEW_VTXARRAYS (1 << 2) +#define NV10_NEW_BLEND (1 << 3) +#define NV10_NEW_BLENDCOL (1 << 4) +#define NV10_NEW_RAST (1 << 5) +#define NV10_NEW_DSA (1 << 6) +#define NV10_NEW_VIEWPORT (1 << 7) +#define NV10_NEW_SCISSOR (1 << 8) +#define NV10_NEW_FRAMEBUFFER (1 << 9) #include "nv10_screen.h" diff --git a/src/gallium/drivers/nv10/nv10_fragtex.c b/src/gallium/drivers/nv10/nv10_fragtex.c index a4bf1082845..67e0b4bd45d 100644 --- a/src/gallium/drivers/nv10/nv10_fragtex.c +++ b/src/gallium/drivers/nv10/nv10_fragtex.c @@ -72,6 +72,7 @@ nv10_fragtex_format(uint pipe_format) static void nv10_fragtex_build(struct nv10_context *nv10, int unit) { +#if 0 struct nv10_sampler_state *ps = nv10->tex_sampler[unit]; struct nv10_miptree *nv10mt = nv10->tex_miptree[unit]; struct pipe_texture *pt = &nv10mt->base; @@ -115,11 +116,13 @@ nv10_fragtex_build(struct nv10_context *nv10, int unit) OUT_RING (ps->filt | 0x2000 /* magic */); OUT_RING ((pt->width[0] << 16) | pt->height[0]); OUT_RING (ps->bcol); +#endif } void nv10_fragtex_bind(struct nv10_context *nv10) { +#if 0 struct nv10_fragment_program *fp = nv10->fragprog.active; unsigned samplers, unit; @@ -141,5 +144,6 @@ nv10_fragtex_bind(struct nv10_context *nv10) } nv10->fp_samplers = fp->samplers; +#endif } diff --git a/src/gallium/drivers/nv10/nv10_prim_vbuf.c b/src/gallium/drivers/nv10/nv10_prim_vbuf.c index bbcfe1a2fdd..412ee9a23f9 100644 --- a/src/gallium/drivers/nv10/nv10_prim_vbuf.c +++ b/src/gallium/drivers/nv10/nv10_prim_vbuf.c @@ -94,6 +94,9 @@ nv10_vbuf_render_get_vertex_info( struct vbuf_render *render ) { struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); struct nv10_context *nv10 = nv10_render->nv10; + + nv10_emit_hw_state(nv10); + return &nv10->vertex_info; } @@ -111,7 +114,7 @@ nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, assert(!nv10_render->buffer); nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); - nv10->dirty |= NV10_NEW_ARRAYS; + nv10->dirty |= NV10_NEW_VTXARRAYS; return winsys->buffer_map(winsys, nv10_render->buffer, @@ -137,6 +140,8 @@ nv10_vbuf_render_draw( struct vbuf_render *render, struct nv10_context *nv10 = nv10_render->nv10; int push, i; + nv10_emit_hw_state(nv10); + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 182f6857f5b..924cf803acb 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -1,3 +1,4 @@ +#include "draw/draw_context.h" #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" @@ -7,38 +8,6 @@ #include "nv10_context.h" #include "nv10_state.h" -static void nv10_vertex_layout(struct pipe_context* pipe) -{ - struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_fragment_program *fp = nv10->fragprog.current; - uint32_t src = 0; - int i; - struct vertex_info vinfo; - - memset(&vinfo, 0, sizeof(vinfo)); - - for (i = 0; i < fp->info.num_inputs; i++) { - switch (fp->info.input_semantic_name[i]) { - case TGSI_SEMANTIC_POSITION: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); - break; - case TGSI_SEMANTIC_COLOR: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); - break; - default: - case TGSI_SEMANTIC_GENERIC: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); - break; - case TGSI_SEMANTIC_FOG: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); - break; - } - } - draw_compute_vertex_size(&vinfo); - - nv10->dirty |= NV10_NEW_VTXFMT; -} - static void * nv10_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) @@ -406,34 +375,29 @@ nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) static void * nv10_vp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) + const struct pipe_shader_state *templ) { - struct nv10_vertex_program *vp; - - vp = CALLOC(1, sizeof(struct nv10_vertex_program)); - vp->pipe = cso; + struct nv10_context *nv10 = nv10_context(pipe); - return (void *)vp; + return draw_create_vertex_shader(nv10->draw, templ); } static void -nv10_vp_state_bind(struct pipe_context *pipe, void *hwcso) +nv10_vp_state_bind(struct pipe_context *pipe, void *shader) { struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_vertex_program *vp = hwcso; - nv10->vertprog.current = vp; + draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); + nv10->dirty |= NV10_NEW_VERTPROG; } static void -nv10_vp_state_delete(struct pipe_context *pipe, void *hwcso) +nv10_vp_state_delete(struct pipe_context *pipe, void *shader) { - //struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_vertex_program *vp = hwcso; + struct nv10_context *nv10 = nv10_context(pipe); - //nv10_vertprog_destroy(nv10, vp); - FREE(vp); + draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); } static void * @@ -550,7 +514,7 @@ nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, struct nv10_context *nv10 = nv10_context(pipe); memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); - nv10->dirty |= NV10_NEW_ARRAYS; + nv10->dirty |= NV10_NEW_VTXARRAYS; } static void @@ -560,7 +524,7 @@ nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, struct nv10_context *nv10 = nv10_context(pipe); memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); - nv10->dirty |= NV10_NEW_ARRAYS; + nv10->dirty |= NV10_NEW_VTXARRAYS; } void diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index 0c524963896..134e52bd622 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -154,6 +154,35 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10) OUT_RING (((h - 1) << 16) | 0); } +static void nv10_vertex_layout(struct nv10_context *nv10) +{ + struct nv10_fragment_program *fp = nv10->fragprog.current; + uint32_t src = 0; + int i; + struct vertex_info vinfo; + + memset(&vinfo, 0, sizeof(vinfo)); + + for (i = 0; i < fp->info.num_inputs; i++) { + switch (fp->info.input_semantic_name[i]) { + case TGSI_SEMANTIC_POSITION: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + case TGSI_SEMANTIC_COLOR: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + default: + case TGSI_SEMANTIC_GENERIC: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + case TGSI_SEMANTIC_FOG: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + } + } + draw_compute_vertex_size(&vinfo); +} + void nv10_emit_hw_state(struct nv10_context *nv10) { @@ -176,13 +205,9 @@ nv10_emit_hw_state(struct nv10_context *nv10) nv10->dirty &= ~NV10_NEW_FRAGPROG; } - if (nv10->dirty & NV10_NEW_ARRAYS) { - nv10->dirty &= ~NV10_NEW_ARRAYS; - // array state will be put here once it's not emitted at each frame - } - - if (nv10->dirty & NV10_NEW_VTXFMT) { - nv10->dirty &= ~NV10_NEW_VTXFMT; + if (nv10->dirty & NV10_NEW_VTXARRAYS) { + nv10->dirty &= ~NV10_NEW_VTXARRAYS; + nv10_vertex_layout(nv10); nv10_vtxbuf_bind(nv10); } diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c index 025ad6de4bd..0f5902602f6 100644 --- a/src/gallium/drivers/nv10/nv10_vbo.c +++ b/src/gallium/drivers/nv10/nv10_vbo.c @@ -18,6 +18,8 @@ boolean nv10_draw_elements( struct pipe_context *pipe, struct draw_context *draw = nv10->draw; unsigned i; + nv10_emit_hw_state(nv10); + /* * Map vertex buffers */ -- cgit v1.2.3 From d3340cda9c2ee7165961517494f636e3fa361d5f Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Wed, 2 Apr 2008 18:28:44 +0200 Subject: don't push "pouets" --- src/gallium/auxiliary/draw/draw_vertex_fetch.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/gallium/auxiliary/draw/draw_vertex_fetch.c b/src/gallium/auxiliary/draw/draw_vertex_fetch.c index 9a9ac4ee451..11f99babf65 100644 --- a/src/gallium/auxiliary/draw/draw_vertex_fetch.c +++ b/src/gallium/auxiliary/draw/draw_vertex_fetch.c @@ -508,7 +508,6 @@ void draw_update_vertex_fetch( struct draw_context *draw ) draw->vertex_fetch.nr_attrs = nr_attrs; draw->vertex_fetch.fetch_func = generic_vertex_fetch; - printf("pouet vertex fetch %x\n",draw->vertex_fetch.fetch_func); switch (nr_attrs) { case 2: @@ -525,6 +524,5 @@ void draw_update_vertex_fetch( struct draw_context *draw ) default: break; } - printf("pouet vertex fetch %x\n",draw->vertex_fetch.fetch_func); } -- cgit v1.2.3 From 7f21b63a988a041bca120751c795f6f6abf0f2bd Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Thu, 3 Apr 2008 04:07:16 +0200 Subject: nv10: fix more vertex stuff --- src/gallium/drivers/nv10/nv10_context.h | 8 ++++---- src/gallium/drivers/nv10/nv10_state.c | 13 +++++++++++-- src/gallium/drivers/nv10/nv10_state_emit.c | 10 ++++++++-- src/gallium/drivers/nv10/nv10_vbo.c | 2 ++ 4 files changed, 25 insertions(+), 8 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 8c13d6897f7..63d33ef7c91 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -65,6 +65,8 @@ struct nv10_context { struct pipe_viewport_state *viewport; struct pipe_scissor_state *scissor; struct pipe_framebuffer_state *framebuffer; + struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + struct vertex_info vertex_info; struct { struct pipe_buffer *buffer; @@ -77,8 +79,7 @@ struct nv10_context { unsigned delta; } vb[16]; - struct vertex_info vertex_info; - struct { +/* struct { struct nouveau_resource *exec_heap; struct nouveau_resource *data_heap; @@ -86,9 +87,8 @@ struct nv10_context { struct nv10_vertex_program *active; struct nv10_vertex_program *current; - struct pipe_buffer *constant_buf; } vertprog; - +*/ struct { struct nv10_fragment_program *active; diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 924cf803acb..1d9a202dd7c 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -449,6 +449,9 @@ static void nv10_set_clip_state(struct pipe_context *pipe, const struct pipe_clip_state *clip) { + struct nv10_context *nv10 = nv10_context(pipe); + + draw_set_clip_state(nv10->draw, clip); } static void @@ -458,11 +461,11 @@ nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, struct nv10_context *nv10 = nv10_context(pipe); if (shader == PIPE_SHADER_VERTEX) { - nv10->vertprog.constant_buf = buf->buffer; + nv10->constbuf[PIPE_SHADER_VERTEX] = buf->buffer; nv10->dirty |= NV10_NEW_VERTPROG; } else if (shader == PIPE_SHADER_FRAGMENT) { - nv10->fragprog.constant_buf = buf->buffer; + nv10->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; nv10->dirty |= NV10_NEW_FRAGPROG; } } @@ -504,6 +507,8 @@ nv10_set_viewport_state(struct pipe_context *pipe, nv10->viewport = (struct pipe_viewport_state*)vpt; + draw_set_viewport_state(nv10->draw, &nv10->viewport); + nv10->dirty |= NV10_NEW_VIEWPORT; } @@ -515,6 +520,8 @@ nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); nv10->dirty |= NV10_NEW_VTXARRAYS; + + draw_set_vertex_buffers(nv10->draw, count, vb); } static void @@ -525,6 +532,8 @@ nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); nv10->dirty |= NV10_NEW_VTXARRAYS; + + draw_set_vertex_elements(nv10->draw, count, ve); } void diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index 134e52bd622..18566986b06 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -136,8 +136,14 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10) assert(0); } - BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + if (zeta) { + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + } else { + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING ( (rt->pitch * rt->cpp) ); + } + nv10->rt[0] = rt->buffer; if (zeta_format) diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c index 0f5902602f6..3a4f49e1564 100644 --- a/src/gallium/drivers/nv10/nv10_vbo.c +++ b/src/gallium/drivers/nv10/nv10_vbo.c @@ -44,6 +44,8 @@ boolean nv10_draw_elements( struct pipe_context *pipe, draw_set_mapped_element_buffer(draw, 0, NULL); } + draw_set_mapped_constant_buffer(draw, nv10->constbuf[PIPE_SHADER_VERTEX]); + /* draw! */ draw_arrays(nv10->draw, prim, start, count); -- cgit v1.2.3 From 8ed894bd17bd6f426a0d87f7113f23043cda3bc3 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Thu, 3 Apr 2008 04:20:22 +0200 Subject: nv10: emit dummy zeta size when no zbuffer is used. --- src/gallium/drivers/nv10/nv10_state_emit.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index 18566986b06..edc194588d1 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -141,7 +141,7 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10) OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); } else { BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING ( (rt->pitch * rt->cpp) ); + OUT_RING ( (rt->pitch * rt->cpp) | ( (rt->pitch * rt->cpp) << 16) ); } nv10->rt[0] = rt->buffer; -- cgit v1.2.3 From 73322bba5c7102f0e100c9a07273a7a87705cf55 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 28 Mar 2008 23:51:24 +1100 Subject: nv40: need to resubmit buffers if pushbuf gets flushed during draw --- src/gallium/drivers/nouveau/nouveau_util.h | 64 +++++++++++++++ src/gallium/drivers/nv40/nv40_vbo.c | 120 ++++++++++++++++++----------- 2 files changed, 140 insertions(+), 44 deletions(-) create mode 100644 src/gallium/drivers/nouveau/nouveau_util.h diff --git a/src/gallium/drivers/nouveau/nouveau_util.h b/src/gallium/drivers/nouveau/nouveau_util.h new file mode 100644 index 00000000000..c92041ebeba --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_util.h @@ -0,0 +1,64 @@ +#ifndef __NOUVEAU_UTIL_H__ +#define __NOUVEAU_UTIL_H__ + +/* Determine how many vertices can be pushed into the command stream. + * Where the remaining space isn't large enough to represent all verices, + * split the buffer at primitive boundaries. + * + * Returns a count of vertices that can be rendered, and an index to + * restart drawing at after a flush. + */ +static INLINE unsigned +nouveau_vbuf_split(unsigned remaining, unsigned overhead, unsigned vpp, + unsigned mode, unsigned start, unsigned count, + unsigned *restart) +{ + int max, adj = 0; + + max = remaining - overhead; + if (max < 0) + return 0; + + max *= vpp; + if (max >= count) + return count; + + switch (mode) { + case PIPE_PRIM_POINTS: + break; + case PIPE_PRIM_LINES: + max = max & 1; + break; + case PIPE_PRIM_TRIANGLES: + max = max - (max % 3); + break; + case PIPE_PRIM_QUADS: + max = max & 3; + break; + case PIPE_PRIM_LINE_LOOP: + case PIPE_PRIM_LINE_STRIP: + if (max < 2) + max = 0; + adj = 1; + break; + case PIPE_PRIM_POLYGON: + case PIPE_PRIM_TRIANGLE_STRIP: + case PIPE_PRIM_TRIANGLE_FAN: + if (max < 3) + max = 0; + adj = 2; + break; + case PIPE_PRIM_QUAD_STRIP: + if (max < 4) + max = 0; + adj = 3; + break; + default: + assert(0); + } + + *restart = start + max - adj; + return max; +} + +#endif diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index bc53924a676..33d3efb58d0 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -7,6 +7,7 @@ #include "nouveau/nouveau_channel.h" #include "nouveau/nouveau_pushbuf.h" +#include "nouveau/nouveau_util.h" #define FORCE_SWTNL 0 @@ -170,44 +171,60 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, } boolean -nv40_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, - unsigned count) +nv40_draw_arrays(struct pipe_context *pipe, + unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); - unsigned nr; + struct nouveau_channel *chan = nv40->nvws->channel; + unsigned restart; nv40_vbo_set_idxbuf(nv40, NULL, 0); if (FORCE_SWTNL || !nv40_state_validate(nv40)) { return nv40_draw_elements_swtnl(pipe, NULL, 0, mode, start, count); } - nv40_state_emit(nv40); - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + while (count) { + unsigned vc, nr; - nr = (count & 0xff); - if (nr) { - BEGIN_RING(curie, NV40TCL_VB_VERTEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } + nv40_state_emit(nv40); - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256, + mode, start, count, &restart); + if (!vc) { + FIRE_RING(NULL); + continue; + } - nr -= push; + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); - BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; + nr = (vc & 0xff); + if (nr) { + BEGIN_RING(curie, NV40TCL_VB_VERTEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; } - } - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (0); + nr = vc >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; + + BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + count -= vc; + start = restart; + } pipe->flush(pipe, 0, NULL); return TRUE; @@ -329,35 +346,50 @@ nv40_draw_elements_vbo(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { struct nv40_context *nv40 = nv40_context(pipe); - unsigned nr; + struct nouveau_channel *chan = nv40->nvws->channel; + unsigned restart; - nv40_state_emit(nv40); + while (count) { + unsigned nr, vc; - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + nv40_state_emit(nv40); - nr = (count & 0xff); - if (nr) { - BEGIN_RING(curie, NV40TCL_VB_INDEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256, + mode, start, count, &restart); + if (!vc) { + FIRE_RING(NULL); + continue; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (vc & 0xff); + if (nr) { + BEGIN_RING(curie, NV40TCL_VB_INDEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; + nr = vc >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; - nr -= push; + nr -= push; - BEGIN_RING_NI(curie, NV40TCL_VB_INDEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; + BEGIN_RING_NI(curie, NV40TCL_VB_INDEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } } - } - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + count -= vc; + start = restart; + } return TRUE; } -- cgit v1.2.3 From 7e9b83ac0ac59298f1b983e6a9aed3a8f2ccb147 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 29 Mar 2008 00:30:04 +1100 Subject: nv40: convert the inline idxbuf paths also --- src/gallium/drivers/nv40/nv40_vbo.c | 159 +++++++++++++++++++++++++----------- 1 file changed, 111 insertions(+), 48 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 33d3efb58d0..8cc977a19a1 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -232,69 +232,139 @@ nv40_draw_arrays(struct pipe_context *pipe, static INLINE void nv40_draw_elements_u08(struct nv40_context *nv40, void *ib, - unsigned start, unsigned count) + unsigned mode, unsigned start, unsigned count) { - uint8_t *elts = (uint8_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } + struct nouveau_channel *chan = nv40->nvws->channel; while (count) { - push = MIN2(count, 2047 * 2); + uint8_t *elts = (uint8_t *)ib + start; + unsigned vc, push, restart; + + nv40_state_emit(nv40); + + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2, + mode, start, count, &restart); + if (vc == 0) { + FIRE_RING(NULL); + continue; + } + count -= vc; + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + if (vc & 1) { + BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; vc--; + } + + while (vc) { + unsigned i; + + push = MIN2(vc, 2047 * 2); - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); - count -= push; - elts += push; + vc -= push; + elts += push; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + start = restart; } } static INLINE void nv40_draw_elements_u16(struct nv40_context *nv40, void *ib, - unsigned start, unsigned count) + unsigned mode, unsigned start, unsigned count) { - uint16_t *elts = (uint16_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } + struct nouveau_channel *chan = nv40->nvws->channel; while (count) { - push = MIN2(count, 2047 * 2); + uint16_t *elts = (uint16_t *)ib + start; + unsigned vc, push, restart; - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); + nv40_state_emit(nv40); - count -= push; - elts += push; + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2, + mode, start, count, &restart); + if (vc == 0) { + FIRE_RING(NULL); + continue; + } + count -= vc; + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + if (vc & 1) { + BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; vc--; + } + + while (vc) { + unsigned i; + + push = MIN2(vc, 2047 * 2); + + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); + + vc -= push; + elts += push; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + start = restart; } } static INLINE void nv40_draw_elements_u32(struct nv40_context *nv40, void *ib, - unsigned start, unsigned count) + unsigned mode, unsigned start, unsigned count) { - uint32_t *elts = (uint32_t *)ib + start; - int push; + struct nouveau_channel *chan = nv40->nvws->channel; while (count) { - push = MIN2(count, 2047); + uint32_t *elts = (uint32_t *)ib + start; + unsigned vc, push, restart; + + nv40_state_emit(nv40); + + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 5, 1, + mode, start, count, &restart); + if (vc == 0) { + FIRE_RING(NULL); + continue; + } + count -= vc; + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + while (vc) { + push = MIN2(vc, 2047); - BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U32, push); - OUT_RINGp (elts, push); + BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U32, push); + OUT_RINGp (elts, push); - count -= push; - elts += push; + vc -= push; + elts += push; + } + + BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); + OUT_RING (0); + + start = restart; } } @@ -315,29 +385,22 @@ nv40_draw_elements_inline(struct pipe_context *pipe, return FALSE; } - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - switch (ib_size) { case 1: - nv40_draw_elements_u08(nv40, map, start, count); + nv40_draw_elements_u08(nv40, map, mode, start, count); break; case 2: - nv40_draw_elements_u16(nv40, map, start, count); + nv40_draw_elements_u16(nv40, map, mode, start, count); break; case 4: - nv40_draw_elements_u32(nv40, map, start, count); + nv40_draw_elements_u32(nv40, map, mode, start, count); break; default: NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); break; } - BEGIN_RING(curie, NV40TCL_BEGIN_END, 1); - OUT_RING (0); - ws->buffer_unmap(ws, ib); - return TRUE; } -- cgit v1.2.3 From 6fbc50e013f1ac7684d8d63d9433f6dd72b4c1cb Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 3 Apr 2008 08:21:56 +1000 Subject: nv40: static attribs -> stateobj --- src/gallium/drivers/nv40/nv40_context.h | 3 +- src/gallium/drivers/nv40/nv40_state_emit.c | 3 +- src/gallium/drivers/nv40/nv40_vbo.c | 62 ++++++++++++++++-------------- 3 files changed, 38 insertions(+), 30 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 525eef8d63d..a1b5c88a063 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -56,7 +56,8 @@ enum nv40_state_index { NV40_STATE_VERTTEX3 = 30, NV40_STATE_VTXBUF = 31, NV40_STATE_VTXFMT = 32, - NV40_STATE_MAX = 33 + NV40_STATE_VTXATTR = 33, + NV40_STATE_MAX = 34 }; #include "nv40_screen.h" diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 722b9f31e6d..c742e4f421e 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -82,7 +82,8 @@ nv40_state_emit(struct nv40_context *nv40) if (!(states & (1ULL << i))) continue; so_ref (state->hw[i], &nv40->screen->state[i]); - so_emit(nv40->nvws, nv40->screen->state[i]); + if (state->hw[i]) + so_emit(nv40->nvws, nv40->screen->state[i]); states &= ~(1ULL << i); } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 8cc977a19a1..e5ee5f3c470 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -109,11 +109,12 @@ nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, } static boolean -nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, - struct pipe_vertex_element *ve, +nv40_vbo_static_attrib(struct nv40_context *nv40, struct nouveau_stateobj *so, + int attrib, struct pipe_vertex_element *ve, struct pipe_vertex_buffer *vb) { struct pipe_winsys *ws = nv40->pipe.winsys; + struct nouveau_grobj *curie = nv40->screen->curie; unsigned type, ncomp; void *map; @@ -128,31 +129,28 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, int attrib, { float *v = map; - BEGIN_RING(curie, NV40TCL_VTX_ATTR_4F_X(attrib), 4); switch (ncomp) { case 4: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(v[3]); + so_method(so, curie, NV40TCL_VTX_ATTR_4F_X(attrib), 4); + so_data (so, fui(v[0])); + so_data (so, fui(v[1])); + so_data (so, fui(v[2])); + so_data (so, fui(v[3])); break; case 3: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(1.0); + so_method(so, curie, NV40TCL_VTX_ATTR_3F_X(attrib), 3); + so_data (so, fui(v[0])); + so_data (so, fui(v[1])); + so_data (so, fui(v[2])); break; case 2: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(0.0); - OUT_RINGf(1.0); + so_method(so, curie, NV40TCL_VTX_ATTR_2F_X(attrib), 2); + so_data (so, fui(v[0])); + so_data (so, fui(v[1])); break; case 1: - OUT_RINGf(v[0]); - OUT_RINGf(0.0); - OUT_RINGf(0.0); - OUT_RINGf(1.0); + so_method(so, curie, 0x1e40 + (attrib * 4), 1); + so_data (so, fui(v[0])); break; default: ws->buffer_unmap(ws, vb->buffer); @@ -487,7 +485,8 @@ static boolean nv40_vbo_validate(struct nv40_context *nv40) { struct nv40_vertex_program *vp = nv40->vertprog; - struct nouveau_stateobj *vtxbuf, *vtxfmt; + struct nouveau_stateobj *vtxbuf, *vtxfmt, *sattr = NULL; + struct nouveau_grobj *curie = nv40->screen->curie; struct pipe_buffer *ib = nv40->idxbuf; unsigned ib_format = nv40->idxbuf_format; unsigned inputs, hw, num_hw; @@ -503,9 +502,9 @@ nv40_vbo_validate(struct nv40_context *nv40) num_hw++; vtxbuf = so_new(20, 18); - so_method(vtxbuf, nv40->screen->curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + so_method(vtxbuf, curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); vtxfmt = so_new(17, 0); - so_method(vtxfmt, nv40->screen->curie, NV40TCL_VTXFMT(0), num_hw); + so_method(vtxfmt, curie, NV40TCL_VTXFMT(0), num_hw); inputs = vp->ir; for (hw = 0; hw < num_hw; hw++) { @@ -522,10 +521,15 @@ nv40_vbo_validate(struct nv40_context *nv40) ve = &nv40->vtxelt[hw]; vb = &nv40->vtxbuf[ve->vertex_buffer_index]; - if (!vb->pitch && nv40_vbo_static_attrib(nv40, hw, ve, vb)) { - so_data(vtxbuf, 0); - so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); - continue; + if (!vb->pitch) { + if (!sattr) + sattr = so_new(16 * 5, 0); + + if (nv40_vbo_static_attrib(nv40, sattr, hw, ve, vb)) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); + continue; + } } if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp)) { @@ -543,19 +547,21 @@ nv40_vbo_validate(struct nv40_context *nv40) } if (ib) { - so_method(vtxbuf, nv40->screen->curie, NV40TCL_IDXBUF_ADDRESS, 2); + so_method(vtxbuf, curie, NV40TCL_IDXBUF_ADDRESS, 2); so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, 0, NV40TCL_IDXBUF_FORMAT_DMA1); } - so_method(vtxbuf, nv40->screen->curie, 0x1710, 1); + so_method(vtxbuf, curie, 0x1710, 1); so_data (vtxbuf, 0); so_ref(vtxbuf, &nv40->state.hw[NV40_STATE_VTXBUF]); nv40->state.dirty |= (1ULL << NV40_STATE_VTXBUF); so_ref(vtxfmt, &nv40->state.hw[NV40_STATE_VTXFMT]); nv40->state.dirty |= (1ULL << NV40_STATE_VTXFMT); + so_ref(sattr, &nv40->state.hw[NV40_STATE_VTXATTR]); + nv40->state.dirty |= (1ULL << NV40_STATE_VTXATTR); return FALSE; } -- cgit v1.2.3 From 0b57662fa6feb3d4571e4a3bc3a2243547595816 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 3 Apr 2008 09:39:00 +1000 Subject: nv40: remove redundant state_emit() calls --- src/gallium/drivers/nv40/nv40_vbo.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index e5ee5f3c470..8c446936ea1 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -375,8 +375,6 @@ nv40_draw_elements_inline(struct pipe_context *pipe, struct pipe_winsys *ws = pipe->winsys; void *map; - nv40_state_emit(nv40); - map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); if (!ib) { NOUVEAU_ERR("failed mapping ib\n"); @@ -468,7 +466,6 @@ nv40_draw_elements(struct pipe_context *pipe, return nv40_draw_elements_swtnl(pipe, NULL, 0, mode, start, count); } - nv40_state_emit(nv40); if (idxbuf) { nv40_draw_elements_vbo(pipe, mode, start, count); -- cgit v1.2.3 From fbb6cc7842ec8a59b60018233275babc4deb6765 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 4 Apr 2008 12:39:45 +1000 Subject: nouveau: in some cases don't create the buffer in local mem initially. --- src/gallium/drivers/nouveau/nouveau_winsys.h | 5 +++++ src/gallium/drivers/nv10/nv10_screen.c | 3 +++ src/gallium/drivers/nv30/nv30_screen.c | 3 +++ src/gallium/drivers/nv40/nv40_miptree.c | 4 +++- src/gallium/drivers/nv40/nv40_screen.c | 8 ++++++++ src/gallium/drivers/nv50/nv50_screen.c | 3 +++ src/gallium/winsys/dri/nouveau/nouveau_context.c | 11 ++++++++++- src/gallium/winsys/dri/nouveau/nouveau_context.h | 5 +++++ src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 2 ++ .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 22 ++++++++++++++++++++-- 10 files changed, 62 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 2a5305f7ce4..fbde7adc6ad 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -13,6 +13,11 @@ #include "nouveau/nouveau_resource.h" #include "nouveau/nouveau_pushbuf.h" +#define NOUVEAU_CAP_HW_VTXBUF (0xbeef0000) +#define NOUVEAU_CAP_HW_IDXBUF (0xbeef0001) + +#define NOUVEAU_BUFFER_USAGE_TEXTURE (1 << 16) + struct nouveau_winsys { struct nouveau_context *nv; diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 80676ead1a9..45fbde62ea2 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -50,6 +50,9 @@ nv10_screen_get_param(struct pipe_screen *screen, int param) return 0; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 12; + case NOUVEAU_CAP_HW_VTXBUF: + case NOUVEAU_CAP_HW_IDXBUF: + return 0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0; diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 3ca50e4fbf4..c7487b37bc0 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -50,6 +50,9 @@ nv30_screen_get_param(struct pipe_screen *screen, int param) return 10; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; + case NOUVEAU_CAP_HW_VTXBUF: + case NOUVEAU_CAP_HW_IDXBUF: + return 0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0; diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 502edc16298..1b192172232 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -68,7 +68,9 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) nv40_miptree_layout(mt); - mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->buffer = ws->buffer_create(ws, 256, + PIPE_BUFFER_USAGE_PIXEL | + NOUVEAU_BUFFER_USAGE_TEXTURE, mt->total_size); if (!mt->buffer) { FREE(mt); diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index e98005f749f..a408d7262f3 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -27,6 +27,8 @@ nv40_screen_get_vendor(struct pipe_screen *pscreen) static int nv40_screen_get_param(struct pipe_screen *pscreen, int param) { + struct nv40_screen *screen = nv40_screen(pscreen); + switch (param) { case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: return 16; @@ -54,6 +56,12 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param) return 10; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; + case NOUVEAU_CAP_HW_VTXBUF: + return 1; + case NOUVEAU_CAP_HW_IDXBUF: + if (screen->curie->grclass == NV40TCL) + return 1; + return 0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 586373a5c4b..adb724b9b7a 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -63,6 +63,9 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param) return 10; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; + case NOUVEAU_CAP_HW_VTXBUF: + case NOUVEAU_CAP_HW_IDXBUF: + return 0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index cf1d83b18f4..aaeaebd2718 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -7,6 +7,7 @@ #include "state_tracker/st_context.h" #include "pipe/p_defines.h" #include "pipe/p_context.h" +#include "pipe/p_screen.h" #include "nouveau_context.h" #include "nouveau_dri.h" @@ -133,7 +134,7 @@ nouveau_context_create(const __GLcontextModes *glVis, /* G80 */ break; default: - NOUVEAU_ERR("Unsupported chipset: NV%02x\n", nv->chipset); + NOUVEAU_ERR("Unsupported chipset: NV%02x\n", (int)nv->chipset); return GL_FALSE; } @@ -255,9 +256,17 @@ nouveau_context_create(const __GLcontextModes *glVis, } if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { + struct pipe_screen *pscreen; + pipe = nouveau_pipe_create(nv); if (!pipe) NOUVEAU_ERR("Couldn't create hw pipe\n"); + pscreen = nvc->pscreen; + + nv->cap.hw_vertex_buffer = + pscreen->get_param(pscreen, NOUVEAU_CAP_HW_VTXBUF); + nv->cap.hw_index_buffer = + pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF); } if (!pipe) { diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index 92f551855a4..acb58fab448 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -56,6 +56,11 @@ struct nouveau_context { struct nouveau_screen *nv_screen; struct pipe_surface *frontbuffer; + struct { + int hw_vertex_buffer; + int hw_index_buffer; + } cap; + /* Hardware context */ struct nouveau_channel_context *nvc; int pctx_id; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c index 2e3ac5492fc..78919bdee8c 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c @@ -137,6 +137,8 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) if (nvpb->base.remaining == nvpb->size) return 0; + nouveau_fence_flush(chan); + nvpb->size -= nvpb->base.remaining; nvchan->dma->cur += nvpb->size; nvchan->dma->free -= nvpb->size; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index b1bf9c521aa..849e38d22b6 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -79,9 +79,10 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size) { struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; - struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_context *nv = nvpws->nv; + struct nouveau_device *dev = nv->nv_screen->device; struct nouveau_pipe_buffer *nvbuf; - uint32_t flags = 0; + uint32_t flags; nvbuf = calloc(1, sizeof(*nvbuf)); if (!nvbuf) @@ -92,6 +93,23 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, nvbuf->base.size = size; flags = NOUVEAU_BO_LOCAL; + + if (usage & PIPE_BUFFER_USAGE_PIXEL) { + if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE) + flags |= NOUVEAU_BO_GART; + flags |= NOUVEAU_BO_VRAM; + } + + if (usage & PIPE_BUFFER_USAGE_VERTEX) { + if (nv->cap.hw_vertex_buffer) + flags |= NOUVEAU_BO_GART; + } + + if (usage & PIPE_BUFFER_USAGE_INDEX) { + if (nv->cap.hw_index_buffer) + flags |= NOUVEAU_BO_GART; + } + if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { free(nvbuf); return NULL; -- cgit v1.2.3 From a45a12e757a8f4d41daea2a3f632d4772ff69e38 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 4 Apr 2008 13:02:37 +1000 Subject: nv40: have test for hw idxbuf in single place --- src/gallium/drivers/nv40/nv40_vbo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 8c446936ea1..ec88470b317 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -75,6 +75,7 @@ static boolean nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, unsigned ib_size) { + struct pipe_screen *pscreen = &nv40->screen->pipe; unsigned type; if (!ib) { @@ -83,8 +84,7 @@ nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib, return FALSE; } - /* No support for 8bit indices, no support at all on 0x4497 chips */ - if (nv40->screen->curie->grclass == NV44TCL || ib_size == 1) + if (!pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF) || ib_size == 1) return FALSE; switch (ib_size) { -- cgit v1.2.3 From bc67533f29abe578e2306be2a24db392a0c62fc4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 4 Apr 2008 16:14:15 +1000 Subject: nouveau: create swizzled surface + scaled image objects --- src/gallium/winsys/dri/nouveau/nouveau_context.c | 2 + src/gallium/winsys/dri/nouveau/nouveau_context.h | 7 +++- src/gallium/winsys/dri/nouveau/nv04_surface.c | 48 ++++++++++++++++++++++++ 3 files changed, 56 insertions(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index aaeaebd2718..ff08a870db1 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -32,6 +32,8 @@ nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) nouveau_grobj_free(&nvc->NvGdiRect); nouveau_grobj_free(&nvc->NvM2MF); nouveau_grobj_free(&nvc->Nv2D); + nouveau_grobj_free(&nvc->NvSwzSurf); + nouveau_grobj_free(&nvc->NvSIFM); nouveau_notifier_free(&nvc->sync_notifier); diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index acb58fab448..872ef93807a 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -27,11 +27,16 @@ struct nouveau_channel_context { struct nouveau_notifier *sync_notifier; + /* Common */ struct nouveau_grobj *NvNull; + struct nouveau_grobj *NvM2MF; + /* NV04-NV40 */ struct nouveau_grobj *NvCtxSurf2D; + struct nouveau_grobj *NvSwzSurf; struct nouveau_grobj *NvImageBlit; struct nouveau_grobj *NvGdiRect; - struct nouveau_grobj *NvM2MF; + struct nouveau_grobj *NvSIFM; + /* G80 */ struct nouveau_grobj *Nv2D; uint32_t next_handle; diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c index cdcd71eaad6..f61bd1477ff 100644 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -237,6 +237,54 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); + switch (nvc->chipset & 0xf0) { + case 0x00: + case 0x10: + class = NV04_SWIZZLED_SURFACE; + break; + case 0x20: + class = NV20_SWIZZLED_SURFACE; + break; + case 0x30: + class = NV30_SWIZZLED_SURFACE; + break; + case 0x40: + case 0x60: + class = NV40_SWIZZLED_SURFACE; + break; + default: + /* Famous last words: this really can't happen.. */ + assert(0); + break; + } + + ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvSwzSurf); + if (ret) { + NOUVEAU_ERR("Error creating swizzled surface: %d\n", ret); + return 1; + } + + BIND_RING (chan, nvc->NvSwzSurf, nvc->next_subchannel++); + + if (nvc->chipset < 0x10) { + class = NV04_SCALED_IMAGE_FROM_MEMORY; + } else + if (nvc->chipset < 0x40) { + class = NV10_SCALED_IMAGE_FROM_MEMORY; + } else { + class = NV40_SCALED_IMAGE_FROM_MEMORY; + } + + ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvSIFM); + if (ret) { + NOUVEAU_ERR("Error creating scaled image object: %d\n", ret); + return 1; + } + + BIND_RING (chan, nvc->NvSIFM, nvc->next_subchannel++); + return 0; } -- cgit v1.2.3 From 2946a5a012f494bad280a0ecf082d81ed4e89c3b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 4 Apr 2008 20:32:06 +1000 Subject: nv40: kill some warnings --- src/gallium/drivers/nv40/nv40_state.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 5dc2991212d..997beca883a 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -421,7 +421,6 @@ static void nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_rasterizer_state *rsso = hwcso; nv40->rasterizer = hwcso; nv40->dirty |= NV40_NEW_RAST; @@ -527,7 +526,6 @@ static void nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv40_context *nv40 = nv40_context(pipe); - struct nv40_vertex_program *vp = hwcso; nv40->vertprog = hwcso; nv40->dirty |= NV40_NEW_VERTPROG; -- cgit v1.2.3 From 2655f6901289bcfe3835cf28d7b9eefa242045b8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 7 Apr 2008 20:10:40 +1000 Subject: nv40: implement user clip planes It turns out the user planes handed to the driver are already in clip space. Hence, we no longer need to transform incoming vertices before computing the clip distance, and no longer need to change the interface provided by gallium. Yay :) The clip state change handling could be better, but this works. --- src/gallium/drivers/nv40/Makefile | 1 - src/gallium/drivers/nv40/nv40_context.h | 4 - src/gallium/drivers/nv40/nv40_draw.c | 2 +- src/gallium/drivers/nv40/nv40_state.h | 3 + src/gallium/drivers/nv40/nv40_state_clip.c | 22 ---- src/gallium/drivers/nv40/nv40_state_emit.c | 8 +- src/gallium/drivers/nv40/nv40_vertprog.c | 159 +++++++++++++++-------------- 7 files changed, 87 insertions(+), 112 deletions(-) delete mode 100644 src/gallium/drivers/nv40/nv40_state_clip.c diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile index 3369a21574f..9c8eadf7e44 100644 --- a/src/gallium/drivers/nv40/Makefile +++ b/src/gallium/drivers/nv40/Makefile @@ -14,7 +14,6 @@ DRIVER_SOURCES = \ nv40_screen.c \ nv40_state.c \ nv40_state_blend.c \ - nv40_state_clip.c \ nv40_state_emit.c \ nv40_state_fb.c \ nv40_state_rasterizer.c \ diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index a1b5c88a063..436351b6bcc 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -76,9 +76,6 @@ enum nv40_state_index { #define NV40_NEW_ARRAYS (1 << 11) #define NV40_NEW_UCP (1 << 12) -#define NV40_FALLBACK_TNL (1 << 0) -#define NV40_FALLBACK_RAST (1 << 1) - struct nv40_rasterizer_state { struct pipe_rasterizer_state pipe; struct nouveau_stateobj *so; @@ -203,7 +200,6 @@ extern void nv40_fragtex_bind(struct nv40_context *); extern boolean nv40_state_validate(struct nv40_context *nv40); extern boolean nv40_state_validate_swtnl(struct nv40_context *nv40); extern void nv40_state_emit(struct nv40_context *nv40); -extern struct nv40_state_entry nv40_state_clip; extern struct nv40_state_entry nv40_state_rasterizer; extern struct nv40_state_entry nv40_state_scissor; extern struct nv40_state_entry nv40_state_stipple; diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index 9cd8fa6a497..7f008aca3a8 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -236,7 +236,7 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe, if (!nv40_state_validate_swtnl(nv40)) return FALSE; - nv40->dirty &= ~(1ULL << NV40_STATE_VTXBUF); + nv40->state.dirty &= ~(1ULL << NV40_STATE_VTXBUF); nv40_state_emit(nv40); for (i = 0; i < PIPE_ATTRIB_MAX; i++) { diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index e018464c9f8..2b4225deb2e 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -29,6 +29,9 @@ struct nv40_vertex_program { struct draw_vertex_shader *draw; boolean translated; + + struct pipe_clip_state ucp; + struct nv40_vertex_program_exec *insns; unsigned nr_insns; struct nv40_vertex_program_data *consts; diff --git a/src/gallium/drivers/nv40/nv40_state_clip.c b/src/gallium/drivers/nv40/nv40_state_clip.c deleted file mode 100644 index c52390f9edb..00000000000 --- a/src/gallium/drivers/nv40/nv40_state_clip.c +++ /dev/null @@ -1,22 +0,0 @@ -#include "nv40_context.h" - -static boolean -nv40_state_clip_validate(struct nv40_context *nv40) -{ - - if (nv40->render_mode == HW) { - nv40->fallback_swtnl &= ~NV40_NEW_UCP; - if (nv40->clip.nr) - nv40->fallback_swtnl |= NV40_NEW_UCP; - } - - return FALSE; -} - -struct nv40_state_entry nv40_state_clip = { - .validate = nv40_state_clip_validate, - .dirty = { - .pipe = NV40_NEW_UCP, - .hw = 0 - } -}; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index c742e4f421e..864dfc2e0c3 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -5,7 +5,6 @@ static struct nv40_state_entry *render_states[] = { &nv40_state_framebuffer, &nv40_state_rasterizer, - &nv40_state_clip, &nv40_state_scissor, &nv40_state_stipple, &nv40_state_fragprog, @@ -22,7 +21,6 @@ static struct nv40_state_entry *render_states[] = { static struct nv40_state_entry *swtnl_states[] = { &nv40_state_framebuffer, &nv40_state_rasterizer, - &nv40_state_clip, &nv40_state_scissor, &nv40_state_stipple, &nv40_state_fragprog, @@ -127,8 +125,7 @@ nv40_state_validate(struct nv40_context *nv40) nv40->pipe.flush(&nv40->pipe, 0, NULL); nv40->dirty |= (NV40_NEW_VIEWPORT | NV40_NEW_VERTPROG | - NV40_NEW_ARRAYS | - NV40_NEW_UCP); + NV40_NEW_ARRAYS); nv40->render_mode = HW; } @@ -153,8 +150,7 @@ nv40_state_validate_swtnl(struct nv40_context *nv40) nv40->pipe.flush(&nv40->pipe, 0, NULL); nv40->dirty |= (NV40_NEW_VIEWPORT | NV40_NEW_VERTPROG | - NV40_NEW_ARRAYS | - NV40_NEW_UCP); + NV40_NEW_ARRAYS); nv40->render_mode = SWTNL; } diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 08d3f387e09..e10250528e2 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -52,6 +52,8 @@ struct nv40_vpc { struct nv40_sreg *imm; unsigned nr_imm; + + unsigned hpos_idx; }; static struct nv40_sreg @@ -423,11 +425,6 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, int ai = -1, ci = -1, ii = -1; int i; - struct { - struct nv40_sreg dst; - unsigned m; - } clip; - if (finst->Instruction.Opcode == TGSI_OPCODE_END) return TRUE; @@ -501,47 +498,6 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); - /* If writing to clip distance regs, need to modify instruction to - * change which component is written to. On NV40 the clip regs - * are the unused components (yzw) of FOGC/PSZ. - */ - clip.dst = none; - if (dst.type == NV40SR_OUTPUT && - dst.index >= NV40_VP_INST_DEST_CLIP(0) && - dst.index <= NV40_VP_INST_DEST_CLIP(5)) { - unsigned n = dst.index - NV40_VP_INST_DEST_CLIP(0); - unsigned m[] = - { MASK_Y, MASK_Z, MASK_W, MASK_Y, MASK_Z, MASK_W }; - - /* Some instructions we can get away with swizzling and/or - * changing the writemask. Others, we'll use a temp reg. - */ - switch (finst->Instruction.Opcode) { - case TGSI_OPCODE_DST: - case TGSI_OPCODE_EXP: - case TGSI_OPCODE_LIT: - case TGSI_OPCODE_LOG: - case TGSI_OPCODE_XPD: - clip.dst = dst; - clip.m = m[n]; - dst = temp(vpc); - break; - case TGSI_OPCODE_DP3: - case TGSI_OPCODE_DP4: - case TGSI_OPCODE_DPH: - case TGSI_OPCODE_POW: - case TGSI_OPCODE_RCP: - case TGSI_OPCODE_RSQ: - mask = m[n]; - break; - default: - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) - src[i] = swz(src[i], X, X, X, X); - mask = m[n]; - break; - } - } - switch (finst->Instruction.Opcode) { case TGSI_OPCODE_ABS: arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); @@ -639,11 +595,6 @@ nv40_vertprog_parse_instruction(struct nv40_vpc *vpc, return FALSE; } - if (clip.dst.type != NV40SR_NONE) { - arith(vpc, 0, OP_MOV, clip.dst, clip.m, - swz(dst, X, X, X, X), none, none); - } - release_temps(vpc); return TRUE; } @@ -658,6 +609,7 @@ nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, switch (fdec->Semantic.SemanticName) { case TGSI_SEMANTIC_POSITION: hw = NV40_VP_INST_DEST_POS; + vpc->hpos_idx = idx; break; case TGSI_SEMANTIC_COLOR: if (fdec->Semantic.SemanticIndex == 0) { @@ -695,15 +647,6 @@ nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, return FALSE; } break; -#if 0 - case TGSI_SEMANTIC_CLIP: - if (fdec->Semantic.SemanticIndex >= 6) { - NOUVEAU_ERR("bad clip distance index\n"); - return FALSE; - } - hw = NV40_VP_INST_DEST_CLIP(fdec->Semantic.SemanticIndex); - break; -#endif default: NOUVEAU_ERR("bad output semantic\n"); return FALSE; @@ -748,6 +691,10 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) } break; #endif + case TGSI_FILE_OUTPUT: + if (!nv40_vertprog_parse_decl_output(vpc, fdec)) + return FALSE; + break; default: break; } @@ -803,6 +750,8 @@ nv40_vertprog_translate(struct nv40_context *nv40, { struct tgsi_parse_context parse; struct nv40_vpc *vpc = NULL; + struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0); + int i; vpc = CALLOC(1, sizeof(struct nv40_vpc)); if (!vpc) @@ -814,26 +763,21 @@ nv40_vertprog_translate(struct nv40_context *nv40, return; } + /* Redirect post-transform vertex position to a temp if user clip + * planes are enabled. We need to append code the the vtxprog + * to handle clip planes later. + */ + if (vp->ucp.nr) { + vpc->r_result[vpc->hpos_idx] = temp(vpc); + vpc->r_temps_discard = 0; + } + tgsi_parse_init(&parse, vp->pipe.tokens); while (!tgsi_parse_end_of_tokens(&parse)) { tgsi_parse_token(&parse); switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - { - const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; - switch (fdec->Declaration.File) { - case TGSI_FILE_OUTPUT: - if (!nv40_vertprog_parse_decl_output(vpc, fdec)) - goto out_err; - break; - default: - break; - } - } - break; case TGSI_TOKEN_TYPE_IMMEDIATE: { const struct tgsi_full_immediate *imm; @@ -862,6 +806,39 @@ nv40_vertprog_translate(struct nv40_context *nv40, } } + /* Write out HPOS if it was redirected to a temp earlier */ + if (vpc->r_result[vpc->hpos_idx].type != NV40SR_OUTPUT) { + struct nv40_sreg hpos = nv40_sr(NV40SR_OUTPUT, + NV40_VP_INST_DEST_POS); + struct nv40_sreg htmp = vpc->r_result[vpc->hpos_idx]; + + arith(vpc, 0, OP_MOV, hpos, MASK_ALL, htmp, none, none); + } + + /* Insert code to handle user clip planes */ + for (i = 0; i < vp->ucp.nr; i++) { + struct nv40_sreg cdst = nv40_sr(NV40SR_OUTPUT, + NV40_VP_INST_DEST_CLIP(i)); + struct nv40_sreg ceqn = constant(vpc, -1, + nv40->clip.ucp[i][0], + nv40->clip.ucp[i][1], + nv40->clip.ucp[i][2], + nv40->clip.ucp[i][3]); + struct nv40_sreg htmp = vpc->r_result[vpc->hpos_idx]; + unsigned mask; + + switch (i) { + case 0: case 3: mask = MASK_Y; break; + case 1: case 4: mask = MASK_Z; break; + case 2: case 5: mask = MASK_W; break; + default: + NOUVEAU_ERR("invalid clip dist #%d\n", i); + goto out_err; + } + + arith(vpc, 0, OP_DP4, cdst, mask, htmp, ceqn, none); + } + vp->insns[vp->nr_insns - 1].data[3] |= NV40_VP_INST_LAST; vp->translated = TRUE; out_err: @@ -883,6 +860,12 @@ nv40_vertprog_validate(struct nv40_context *nv40) if (nv40->render_mode == HW) { vp = nv40->vertprog; constbuf = nv40->constbuf[PIPE_SHADER_VERTEX]; + + if ((nv40->dirty & NV40_NEW_UCP) || + memcmp(&nv40->clip, &vp->ucp, sizeof(vp->ucp))) { + nv40_vertprog_destroy(nv40, vp); + memcpy(&vp->ucp, &nv40->clip, sizeof(vp->ucp)); + } } else { vp = nv40->swtnl.vertprog; constbuf = NULL; @@ -1045,16 +1028,36 @@ check_gpu_resources: void nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp) { - if (vp->nr_consts) - FREE(vp->consts); - if (vp->nr_insns) + struct nouveau_winsys *nvws = nv40->screen->nvws; + + vp->translated = FALSE; + + if (vp->nr_insns) { FREE(vp->insns); + vp->insns = NULL; + vp->nr_insns = 0; + } + + if (vp->nr_consts) { + FREE(vp->consts); + vp->consts = NULL; + vp->nr_consts = 0; + } + + nvws->res_free(&vp->exec); + vp->exec_start = 0; + nvws->res_free(&vp->data); + vp->data_start = 0; + vp->data_start_min = 0; + + vp->ir = vp->or = vp->clip_ctrl = 0; + so_ref(NULL, &vp->so); } struct nv40_state_entry nv40_state_vertprog = { .validate = nv40_vertprog_validate, .dirty = { - .pipe = NV40_NEW_VERTPROG, + .pipe = NV40_NEW_VERTPROG | NV40_NEW_UCP, .hw = NV40_STATE_VERTPROG, } }; -- cgit v1.2.3 From bdfcce47921cdd808740ee26e6781837351bad98 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 9 Apr 2008 13:58:51 +1000 Subject: nv40: use vb/ve counts rather than shader inputs for related loops --- src/gallium/drivers/nv40/nv40_draw.c | 9 ++------- src/gallium/drivers/nv40/nv40_vbo.c | 25 ++++--------------------- 2 files changed, 6 insertions(+), 28 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index 7f008aca3a8..dd5cc8fc998 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -239,9 +239,7 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe, nv40->state.dirty &= ~(1ULL << NV40_STATE_VTXBUF); nv40_state_emit(nv40); - for (i = 0; i < PIPE_ATTRIB_MAX; i++) { - if (!nv40->vtxbuf[i].buffer) - continue; + for (i = 0; i < nv40->vtxbuf_nr; i++) { map = ws->buffer_map(ws, nv40->vtxbuf[i].buffer, PIPE_BUFFER_USAGE_CPU_READ); draw_set_mapped_vertex_buffer(nv40->draw, i, map); @@ -262,11 +260,8 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe, draw_arrays(nv40->draw, mode, start, count); - for (i = 0; i < PIPE_ATTRIB_MAX; i++) { - if (!nv40->vtxbuf[i].buffer) - continue; + for (i = 0; i < nv40->vtxbuf_nr; i++) ws->buffer_unmap(ws, nv40->vtxbuf[i].buffer); - } if (idxbuf) ws->buffer_unmap(ws, idxbuf); diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index ec88470b317..06374184b12 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -481,40 +481,23 @@ nv40_draw_elements(struct pipe_context *pipe, static boolean nv40_vbo_validate(struct nv40_context *nv40) { - struct nv40_vertex_program *vp = nv40->vertprog; struct nouveau_stateobj *vtxbuf, *vtxfmt, *sattr = NULL; struct nouveau_grobj *curie = nv40->screen->curie; struct pipe_buffer *ib = nv40->idxbuf; unsigned ib_format = nv40->idxbuf_format; - unsigned inputs, hw, num_hw; unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; - - inputs = vp->ir; - for (hw = 0; hw < 16 && inputs; hw++) { - if (inputs & (1 << hw)) { - num_hw = hw; - inputs &= ~(1 << hw); - } - } - num_hw++; + int hw; vtxbuf = so_new(20, 18); - so_method(vtxbuf, curie, NV40TCL_VTXBUF_ADDRESS(0), num_hw); + so_method(vtxbuf, curie, NV40TCL_VTXBUF_ADDRESS(0), nv40->vtxelt_nr); vtxfmt = so_new(17, 0); - so_method(vtxfmt, curie, NV40TCL_VTXFMT(0), num_hw); + so_method(vtxfmt, curie, NV40TCL_VTXFMT(0), nv40->vtxelt_nr); - inputs = vp->ir; - for (hw = 0; hw < num_hw; hw++) { + for (hw = 0; hw < nv40->vtxelt_nr; hw++) { struct pipe_vertex_element *ve; struct pipe_vertex_buffer *vb; unsigned type, ncomp; - if (!(inputs & (1 << hw))) { - so_data(vtxbuf, 0); - so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT); - continue; - } - ve = &nv40->vtxelt[hw]; vb = &nv40->vtxbuf[ve->vertex_buffer_index]; -- cgit v1.2.3 From b1de9c948f9794db97ed7d34b8cbcdccc206ea77 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Apr 2008 21:37:28 +0200 Subject: nv30: Set pipe status on clear --- src/gallium/drivers/nv30/nv30_clear.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/gallium/drivers/nv30/nv30_clear.c b/src/gallium/drivers/nv30/nv30_clear.c index 71f413588ee..8c3ca204d58 100644 --- a/src/gallium/drivers/nv30/nv30_clear.c +++ b/src/gallium/drivers/nv30/nv30_clear.c @@ -9,4 +9,5 @@ nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); + ps->status = PIPE_SURFACE_STATUS_CLEAR; } -- cgit v1.2.3 From 4c498c1b0c38a3ba93649fa4435937d15689bae1 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Apr 2008 21:37:41 +0200 Subject: nv30: Do flip --- src/gallium/drivers/nv30/nv30_surface.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_surface.c b/src/gallium/drivers/nv30/nv30_surface.c index b20a3dd4c15..b22211ac86e 100644 --- a/src/gallium/drivers/nv30/nv30_surface.c +++ b/src/gallium/drivers/nv30/nv30_surface.c @@ -42,8 +42,20 @@ nv30_surface_copy(struct pipe_context *pipe, unsigned do_flip, struct nv30_context *nv30 = nv30_context(pipe); struct nouveau_winsys *nvws = nv30->nvws; - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); + if (do_flip) { + /*XXX: This dodgyness will do for now for correctness. But, + * need to investigate whether the 2D engine is able to + * manage a flip (perhaps SIFM?), if not, use the 3D engine + */ + desty += height; + while (height--) { + nvws->surface_copy(nvws, dest, destx, desty--, src, + srcx, srcy++, width, 1); + } + } else { + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); + } } static void -- cgit v1.2.3 From 186277ee928a7c9ad8a31776f93e502613ad94fd Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Apr 2008 23:39:29 +0200 Subject: nv30: Move some structures and functions from context to screen --- src/gallium/drivers/nv30/nv30_context.c | 185 ++-------------------------- src/gallium/drivers/nv30/nv30_context.h | 71 +++++++++-- src/gallium/drivers/nv30/nv30_query.c | 23 ++-- src/gallium/drivers/nv30/nv30_screen.c | 202 +++++++++++++++++++++++++++---- src/gallium/drivers/nv30/nv30_screen.h | 17 ++- src/gallium/drivers/nv30/nv30_vertprog.c | 31 ++++- 6 files changed, 303 insertions(+), 226 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index b8e8b86d99a..d38713e98ad 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -26,198 +26,36 @@ static void nv30_destroy(struct pipe_context *pipe) { struct nv30_context *nv30 = nv30_context(pipe); - struct nouveau_winsys *nvws = nv30->nvws; if (nv30->draw) draw_destroy(nv30->draw); - - nvws->res_free(&nv30->vertprog.exec_heap); - nvws->res_free(&nv30->vertprog.data_heap); - - nvws->res_free(&nv30->query_heap); - nvws->notifier_free(&nv30->query); - - nvws->notifier_free(&nv30->sync); - - nvws->grobj_free(&nv30->rankine); - FREE(nv30); } -static boolean -nv30_init_hwctx(struct nv30_context *nv30, int rankine_class) -{ - struct nouveau_winsys *nvws = nv30->nvws; - int ret; - int i; - - ret = nvws->grobj_alloc(nvws, rankine_class, &nv30->rankine); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - BEGIN_RING(rankine, NV34TCL_DMA_NOTIFY, 1); - OUT_RING (nv30->sync->handle); - BEGIN_RING(rankine, NV34TCL_DMA_TEXTURE0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(rankine, NV34TCL_DMA_VTXBUF0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); -/* BEGIN_RING(rankine, NV34TCL_DMA_FENCE, 2); - OUT_RING (0); - OUT_RING (nv30->query->handle);*/ - BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY7, 1); - OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(rankine, NV34TCL_DMA_IN_MEMORY8, 1); - OUT_RING (nvws->channel->vram->handle); - - for (i=1; i<8; i++) { - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(i), 1); - OUT_RING (0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_VERT(i), 1); - OUT_RING (0); - } - - BEGIN_RING(rankine, 0x220, 1); - OUT_RING (1); - - BEGIN_RING(rankine, 0x03b0, 1); - OUT_RING (0x00100000); - BEGIN_RING(rankine, 0x1454, 1); - OUT_RING (0); - BEGIN_RING(rankine, 0x1d80, 1); - OUT_RING (3); - BEGIN_RING(rankine, 0x1450, 1); - OUT_RING (0x00030004); - - /* NEW */ - BEGIN_RING(rankine, 0x1e98, 1); - OUT_RING (0); - BEGIN_RING(rankine, 0x17e0, 3); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0x3f800000); - BEGIN_RING(rankine, 0x1f80, 16); - OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); - OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); - OUT_RING (0x0000ffff); - OUT_RING (0); OUT_RING (0); OUT_RING (0); OUT_RING (0); - OUT_RING (0); OUT_RING (0); OUT_RING (0); - - BEGIN_RING(rankine, 0x120, 3); - OUT_RING (0); - OUT_RING (1); - OUT_RING (2); - - BEGIN_RING(rankine, 0x1d88, 1); - OUT_RING (0x00001200); - - BEGIN_RING(rankine, NV34TCL_RC_ENABLE, 1); - OUT_RING (0); - - BEGIN_RING(rankine, NV34TCL_DEPTH_RANGE_NEAR, 2); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - - BEGIN_RING(rankine, NV34TCL_MULTISAMPLE_CONTROL, 1); - OUT_RING (0xffff0000); - - /* enables use of vp rather than fixed-function somehow */ - BEGIN_RING(rankine, 0x1e94, 1); - OUT_RING (0x13); - - FIRE_RING (NULL); - return TRUE; -} - -#define NV30TCL_CHIPSET_3X_MASK 0x00000003 -#define NV34TCL_CHIPSET_3X_MASK 0x00000010 -#define NV35TCL_CHIPSET_3X_MASK 0x000001e0 - struct pipe_context * -nv30_create(struct pipe_screen *screen, unsigned pctx_id) +nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) { - struct pipe_winsys *pipe_winsys = screen->winsys; - struct nouveau_winsys *nvws = nv30_screen(screen)->nvws; - unsigned chipset = nv30_screen(screen)->chipset; + struct nv30_screen *screen = nv30_screen(pscreen); + struct pipe_winsys *ws = pscreen->winsys; struct nv30_context *nv30; - int rankine_class = 0, ret; + unsigned chipset = screen->chipset; + struct nouveau_winsys *nvws = screen->nvws; - if ((chipset & 0xf0) != 0x30) { - NOUVEAU_ERR("Not a NV3X chipset\n"); - return NULL; - } - - if (NV30TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { - rankine_class = 0x0397; - } else if (NV34TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { - rankine_class = 0x0697; - } else if (NV35TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) { - rankine_class = 0x0497; - } else { - NOUVEAU_ERR("Unknown NV3X chipset: NV%02x\n", chipset); - return NULL; - } - - nv30 = CALLOC_STRUCT(nv30_context); + nv30 = CALLOC(1, sizeof(struct nv30_context)); if (!nv30) return NULL; + nv30->screen = screen; + nv30->pctx_id = pctx_id; + nv30->chipset = chipset; nv30->nvws = nvws; - /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &nv30->sync); - if (ret) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Query objects */ - ret = nvws->notifier_alloc(nvws, 32, &nv30->query); - if (ret) { - NOUVEAU_ERR("Error initialising query objects: %d\n", ret); - nv30_destroy(&nv30->pipe); - return NULL; - } - - ret = nvws->res_init(&nv30->query_heap, 0, 32); - if (ret) { - NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Vtxprog resources */ - if (nvws->res_init(&nv30->vertprog.exec_heap, 0, 256) || - nvws->res_init(&nv30->vertprog.data_heap, 0, 256)) { - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Static rankine initialisation */ - if (!nv30_init_hwctx(nv30, rankine_class)) { - nv30_destroy(&nv30->pipe); - return NULL; - } - - /* Pipe context setup */ - nv30->pipe.winsys = pipe_winsys; - nv30->pipe.screen = screen; - + nv30->pipe.winsys = ws; + nv30->pipe.screen = pscreen; nv30->pipe.destroy = nv30_destroy; - nv30->pipe.draw_arrays = nv30_draw_arrays; nv30->pipe.draw_elements = nv30_draw_elements; nv30->pipe.clear = nv30_clear; - nv30->pipe.flush = nv30_flush; nv30_init_query_functions(nv30); @@ -226,7 +64,6 @@ nv30_create(struct pipe_screen *screen, unsigned pctx_id) nv30_init_miptree_functions(nv30); nv30->draw = draw_create(); - assert(nv30->draw); draw_set_rasterize_stage(nv30->draw, nv30_draw_render_stage(nv30)); return &nv30->pipe; diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index c1cc3eca1e3..180969731bf 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -11,8 +11,9 @@ #include "nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv30_context *ctx = nv30 + struct nv30_screen *ctx = nv30->screen #include "nouveau/nouveau_push.h" +#include "nouveau/nouveau_stateobj.h" #include "nv30_state.h" @@ -21,23 +22,70 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV30_NEW_VERTPROG (1 << 1) -#define NV30_NEW_FRAGPROG (1 << 2) -#define NV30_NEW_ARRAYS (1 << 3) +enum nv30_state_index { + NV30_STATE_FB = 0, + NV30_STATE_VIEWPORT = 1, + NV30_STATE_BLEND = 2, + NV30_STATE_RAST = 3, + NV30_STATE_ZSA = 4, + NV30_STATE_BCOL = 5, + NV30_STATE_CLIP = 6, + NV30_STATE_SCISSOR = 7, + NV30_STATE_STIPPLE = 8, + NV30_STATE_FRAGPROG = 9, + NV30_STATE_VERTPROG = 10, + NV30_STATE_FRAGTEX0 = 11, + NV30_STATE_FRAGTEX1 = 12, + NV30_STATE_FRAGTEX2 = 13, + NV30_STATE_FRAGTEX3 = 14, + NV30_STATE_FRAGTEX4 = 15, + NV30_STATE_FRAGTEX5 = 16, + NV30_STATE_FRAGTEX6 = 17, + NV30_STATE_FRAGTEX7 = 18, + NV30_STATE_FRAGTEX8 = 19, + NV30_STATE_FRAGTEX9 = 20, + NV30_STATE_FRAGTEX10 = 21, + NV30_STATE_FRAGTEX11 = 22, + NV30_STATE_FRAGTEX12 = 23, + NV30_STATE_FRAGTEX13 = 24, + NV30_STATE_FRAGTEX14 = 25, + NV30_STATE_FRAGTEX15 = 26, + NV30_STATE_VERTTEX0 = 27, + NV30_STATE_VERTTEX1 = 28, + NV30_STATE_VERTTEX2 = 29, + NV30_STATE_VERTTEX3 = 30, + NV30_STATE_VTXBUF = 31, + NV30_STATE_VTXFMT = 32, + NV30_STATE_VTXATTR = 33, + NV30_STATE_MAX = 34 +}; + +#include "nv30_screen.h" + +#define NV30_NEW_BLEND (1 << 0) +#define NV30_NEW_RAST (1 << 1) +#define NV30_NEW_ZSA (1 << 2) +#define NV30_NEW_SAMPLER (1 << 3) +#define NV30_NEW_FB (1 << 4) +#define NV30_NEW_STIPPLE (1 << 5) +#define NV30_NEW_SCISSOR (1 << 6) +#define NV30_NEW_VIEWPORT (1 << 7) +#define NV30_NEW_BCOL (1 << 8) +#define NV30_NEW_VERTPROG (1 << 9) +#define NV30_NEW_FRAGPROG (1 << 10) +#define NV30_NEW_ARRAYS (1 << 11) +#define NV30_NEW_UCP (1 << 12) struct nv30_context { struct pipe_context pipe; + struct nouveau_winsys *nvws; + struct nv30_screen *screen; + unsigned pctx_id; struct draw_context *draw; int chipset; - struct nouveau_grobj *rankine; - struct nouveau_notifier *sync; - - /* query objects */ - struct nouveau_notifier *query; - struct nouveau_resource *query_heap; uint32_t dirty; @@ -63,9 +111,6 @@ struct nv30_context { } vb[16]; struct { - struct nouveau_resource *exec_heap; - struct nouveau_resource *data_heap; - struct nv30_vertex_program *active; struct nv30_vertex_program *current; diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c index 0c2d941562c..d40d75f2640 100644 --- a/src/gallium/drivers/nv30/nv30_query.c +++ b/src/gallium/drivers/nv30/nv30_query.c @@ -1,5 +1,4 @@ #include "pipe/p_context.h" -#include "pipe/p_util.h" #include "nv30_context.h" @@ -10,7 +9,7 @@ struct nv30_query { uint64_t result; }; -static inline struct nv30_query * +static INLINE struct nv30_query * nv30_query(struct pipe_query *pipe) { return (struct nv30_query *)pipe; @@ -46,9 +45,18 @@ nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq) assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER); - if (nv30->nvws->res_alloc(nv30->query_heap, 1, NULL, &q->object)) + /* Happens when end_query() is called, then another begin_query() + * without querying the result in-between. For now we'll wait for + * the existing query to notify completion, but it could be better. + */ + if (q->object) { + uint64 tmp; + pipe->get_query_result(pipe, pq, 1, &tmp); + } + + if (nv30->nvws->res_alloc(nv30->screen->query_heap, 1, NULL, &q->object)) assert(0); - nv30->nvws->notifier_reset(nv30->query, q->object->start); + nv30->nvws->notifier_reset(nv30->screen->query, q->object->start); BEGIN_RING(rankine, NV34TCL_QUERY_RESET, 1); OUT_RING (1); @@ -83,16 +91,17 @@ nv30_query_result(struct pipe_context *pipe, struct pipe_query *pq, if (!q->ready) { unsigned status; - status = nvws->notifier_status(nv30->query, q->object->start); + status = nvws->notifier_status(nv30->screen->query, + q->object->start); if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) { if (wait == FALSE) return FALSE; - nvws->notifier_wait(nv30->query, q->object->start, + nvws->notifier_wait(nv30->screen->query, q->object->start, NV_NOTIFY_STATE_STATUS_COMPLETED, 0); } - q->result = nvws->notifier_retval(nv30->query, + q->result = nvws->notifier_retval(nv30->screen->query, q->object->start); q->ready = TRUE; nvws->res_free(&q->object); diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index c7487b37bc0..ce6c9ec523a 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -4,24 +4,28 @@ #include "nv30_context.h" #include "nv30_screen.h" +#define NV30TCL_CHIPSET_3X_MASK 0x00000003 +#define NV34TCL_CHIPSET_3X_MASK 0x00000010 +#define NV35TCL_CHIPSET_3X_MASK 0x000001e0 + static const char * -nv30_screen_get_name(struct pipe_screen *screen) +nv30_screen_get_name(struct pipe_screen *pscreen) { - struct nv30_screen *nv30screen = nv30_screen(screen); + struct nv30_screen *screen = nv30_screen(pscreen); static char buffer[128]; - snprintf(buffer, sizeof(buffer), "NV%02X", nv30screen->chipset); + snprintf(buffer, sizeof(buffer), "NV%02X", screen->chipset); return buffer; } static const char * -nv30_screen_get_vendor(struct pipe_screen *screen) +nv30_screen_get_vendor(struct pipe_screen *pscreen) { return "nouveau"; } static int -nv30_screen_get_param(struct pipe_screen *screen, int param) +nv30_screen_get_param(struct pipe_screen *pscreen, int param) { switch (param) { case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: @@ -60,7 +64,7 @@ nv30_screen_get_param(struct pipe_screen *screen, int param) } static float -nv30_screen_get_paramf(struct pipe_screen *screen, int param) +nv30_screen_get_paramf(struct pipe_screen *pscreen, int param) { switch (param) { case PIPE_CAP_MAX_LINE_WIDTH: @@ -82,8 +86,8 @@ nv30_screen_get_paramf(struct pipe_screen *screen, int param) } static boolean -nv30_screen_is_format_supported(struct pipe_screen *screen, - enum pipe_format format, uint type) +nv30_screen_surface_format_supported(struct pipe_screen *pscreen, + enum pipe_format format, uint type) { switch (type) { case PIPE_SURFACE: @@ -122,36 +126,184 @@ nv30_screen_is_format_supported(struct pipe_screen *screen, } static void -nv30_screen_destroy(struct pipe_screen *screen) +nv30_screen_destroy(struct pipe_screen *pscreen) { - FREE(screen); + struct nv30_screen *screen = nv30_screen(pscreen); + struct nouveau_winsys *nvws = screen->nvws; + + nvws->res_free(&screen->vp_exec_heap); + nvws->res_free(&screen->vp_data_heap); + nvws->res_free(&screen->query_heap); + nvws->notifier_free(&screen->query); + nvws->notifier_free(&screen->sync); + nvws->grobj_free(&screen->rankine); + + FREE(pscreen); } struct pipe_screen * -nv30_screen_create(struct pipe_winsys *winsys, struct nouveau_winsys *nvws, +nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, unsigned chipset) { - struct nv30_screen *nv30screen = CALLOC_STRUCT(nv30_screen); + struct nv30_screen *screen = CALLOC_STRUCT(nv30_screen); + struct nouveau_stateobj *so; + unsigned rankine_class = 0; + int ret, i; + + if (!screen) + return NULL; + screen->chipset = chipset; + screen->nvws = nvws; + + /* 3D object */ + switch (chipset & 0xf0) { + case 0x30: + if (NV30TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) + rankine_class = 0x0397; + else + if (NV34TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) + rankine_class = 0x0697; + else + if (NV35TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f))) + rankine_class = 0x0497; + break; + default: + break; + } - if (!nv30screen) + if (!rankine_class) { + NOUVEAU_ERR("Unknown nv3x chipset: nv%02x\n", chipset); return NULL; + } + + ret = nvws->grobj_alloc(nvws, rankine_class, &screen->rankine); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv30_screen_destroy(&screen->pipe); + return NULL; + } + + /* Query objects */ + ret = nvws->notifier_alloc(nvws, 32, &screen->query); + if (ret) { + NOUVEAU_ERR("Error initialising query objects: %d\n", ret); + nv30_screen_destroy(&screen->pipe); + return NULL; + } + + ret = nvws->res_init(&screen->query_heap, 0, 32); + if (ret) { + NOUVEAU_ERR("Error initialising query object heap: %d\n", ret); + nv30_screen_destroy(&screen->pipe); + return NULL; + } + + /* Vtxprog resources */ + if (nvws->res_init(&screen->vp_exec_heap, 0, 256) || + nvws->res_init(&screen->vp_data_heap, 0, 256)) { + nv30_screen_destroy(&screen->pipe); + return NULL; + } + + /* Static rankine initialisation */ + so = so_new(128, 0); + so_method(so, screen->rankine, NV34TCL_DMA_NOTIFY, 1); + so_data (so, screen->sync->handle); + so_method(so, screen->rankine, NV34TCL_DMA_TEXTURE0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); + so_method(so, screen->rankine, NV34TCL_DMA_COLOR1, 1); + so_data (so, nvws->channel->vram->handle); + so_method(so, screen->rankine, NV34TCL_DMA_COLOR0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->vram->handle); + so_method(so, screen->rankine, NV34TCL_DMA_VTXBUF0, 2); + so_data (so, nvws->channel->vram->handle); + so_data (so, nvws->channel->gart->handle); +/* so_method(so, screen->rankine, NV34TCL_DMA_FENCE, 2); + so_data (so, 0); + so_data (so, screen->query->handle);*/ + so_method(so, screen->rankine, NV34TCL_DMA_IN_MEMORY7, 1); + so_data (so, nvws->channel->vram->handle); + so_method(so, screen->rankine, NV34TCL_DMA_IN_MEMORY8, 1); + so_data (so, nvws->channel->vram->handle); + + for (i=1; i<8; i++) { + so_method(so, screen->rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(i), 1); + so_data (so, 0); + so_method(so, screen->rankine, NV34TCL_VIEWPORT_CLIP_VERT(i), 1); + so_data (so, 0); + } + + so_method(so, screen->rankine, 0x220, 1); + so_data (so, 1); + + so_method(so, screen->rankine, 0x03b0, 1); + so_data (so, 0x00100000); + so_method(so, screen->rankine, 0x1454, 1); + so_data (so, 0); + so_method(so, screen->rankine, 0x1d80, 1); + so_data (so, 3); + so_method(so, screen->rankine, 0x1450, 1); + so_data (so, 0x00030004); + + /* NEW */ + so_method(so, screen->rankine, 0x1e98, 1); + so_data (so, 0); + so_method(so, screen->rankine, 0x17e0, 3); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(1.0)); + so_method(so, screen->rankine, 0x1f80, 16); + for (i=0; i<16; i++) { + so_data (so, (i==8) ? 0x0000ffff : 0); + } + + so_method(so, screen->rankine, 0x120, 3); + so_data (so, 0); + so_data (so, 1); + so_data (so, 2); + + so_method(so, screen->rankine, 0x1d88, 1); + so_data (so, 0x00001200); + + so_method(so, screen->rankine, NV34TCL_RC_ENABLE, 1); + so_data (so, 0); + + so_method(so, screen->rankine, NV34TCL_DEPTH_RANGE_NEAR, 2); + so_data (so, fui(0.0)); + so_data (so, fui(1.0)); + + so_method(so, screen->rankine, NV34TCL_MULTISAMPLE_CONTROL, 1); + so_data (so, 0xffff0000); + + /* enables use of vp rather than fixed-function somehow */ + so_method(so, screen->rankine, 0x1e94, 1); + so_data (so, 0x13); - nv30screen->chipset = chipset; - nv30screen->nvws = nvws; + so_emit(nvws, so); + so_ref(NULL, &so); + nvws->push_flush(nvws, 0, NULL); - nv30screen->screen.winsys = winsys; + screen->pipe.winsys = ws; + screen->pipe.destroy = nv30_screen_destroy; - nv30screen->screen.destroy = nv30_screen_destroy; + screen->pipe.get_name = nv30_screen_get_name; + screen->pipe.get_vendor = nv30_screen_get_vendor; + screen->pipe.get_param = nv30_screen_get_param; + screen->pipe.get_paramf = nv30_screen_get_paramf; - nv30screen->screen.get_name = nv30_screen_get_name; - nv30screen->screen.get_vendor = nv30_screen_get_vendor; - nv30screen->screen.get_param = nv30_screen_get_param; - nv30screen->screen.get_paramf = nv30_screen_get_paramf; - nv30screen->screen.is_format_supported = - nv30_screen_is_format_supported; + screen->pipe.is_format_supported = nv30_screen_surface_format_supported; - nv30_screen_init_miptree_functions(&nv30screen->screen); + nv30_screen_init_miptree_functions(&screen->pipe); - return &nv30screen->screen; + return &screen->pipe; } diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h index f878f81e11f..56f8776a173 100644 --- a/src/gallium/drivers/nv30/nv30_screen.h +++ b/src/gallium/drivers/nv30/nv30_screen.h @@ -4,10 +4,25 @@ #include "pipe/p_screen.h" struct nv30_screen { - struct pipe_screen screen; + struct pipe_screen pipe; struct nouveau_winsys *nvws; unsigned chipset; + + /* HW graphics objects */ + struct nouveau_grobj *rankine; + struct nouveau_notifier *sync; + + /* Query object resources */ + struct nouveau_notifier *query; + struct nouveau_resource *query_heap; + + /* Vtxprog resources */ + struct nouveau_resource *vp_exec_heap; + struct nouveau_resource *vp_data_heap; + + /* Current 3D state of channel */ + struct nouveau_stateobj *state[NV30_STATE_MAX]; }; static INLINE struct nv30_screen * diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index fe1a467565d..71aea3a59cb 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -1,7 +1,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi/util/tgsi_parse.h" @@ -654,7 +653,7 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) /* Allocate hw vtxprog exec slots */ if (!vp->exec) { - struct nouveau_resource *heap = nv30->vertprog.exec_heap; + struct nouveau_resource *heap = nv30->screen->vp_exec_heap; uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { @@ -674,7 +673,7 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) /* Allocate hw vtxprog const slots */ if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv30->vertprog.data_heap; + struct nouveau_resource *heap = nv30->screen->vp_data_heap; if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { while (heap->next && heap->size < vp->nr_consts) { @@ -789,9 +788,29 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) void nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) { - if (vp->nr_consts) - FREE(vp->consts); - if (vp->nr_insns) + struct nouveau_winsys *nvws = nv30->screen->nvws; + + vp->translated = FALSE; + + if (vp->nr_insns) { FREE(vp->insns); + vp->insns = NULL; + vp->nr_insns = 0; + } + + if (vp->nr_consts) { + FREE(vp->consts); + vp->consts = NULL; + vp->nr_consts = 0; + } + + nvws->res_free(&vp->exec); + vp->exec_start = 0; + nvws->res_free(&vp->data); + vp->data_start = 0; + vp->data_start_min = 0; + + /* vp->ir = vp->or = vp->clip_ctrl = 0; + so_ref(NULL, &vp->so); */ } -- cgit v1.2.3 From 7f811f2c42937f254ae1b11e5b0ece765a8ea31b Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Apr 2008 22:16:42 +0200 Subject: nv30: Update miptree stuff --- src/gallium/drivers/nv30/nv30_miptree.c | 37 +++++++++++++++++---------------- 1 file changed, 19 insertions(+), 18 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index afb05fdd2bf..10ab46e19a5 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -4,7 +4,6 @@ #include "pipe/p_inlines.h" #include "nv30_context.h" -#include "nv30_screen.h" static void nv30_miptree_layout(struct nv30_miptree *nv30mt) @@ -13,7 +12,7 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) boolean swizzled = FALSE; uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; uint offset = 0; - int nr_faces, l, f; + int nr_faces, l, f, pitch; if (pt->target == PIPE_TEXTURE_CUBE) { nr_faces = 6; @@ -24,17 +23,17 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) nr_faces = 1; } + pitch = pt->width[0]; for (l = 0; l <= pt->last_level; l++) { pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; if (swizzled) - nv30mt->level[l].pitch = pt->width[l] * pt->cpp; - else - nv30mt->level[l].pitch = pt->width[0] * pt->cpp; - nv30mt->level[l].pitch = (nv30mt->level[l].pitch + 63) & ~63; + pitch = pt->width[l]; + pitch = (pitch + 63) & ~63; + nv30mt->level[l].pitch = pitch * pt->cpp; nv30mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); @@ -55,9 +54,9 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) } static struct pipe_texture * -nv30_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) +nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { - struct pipe_winsys *ws = screen->winsys; + struct pipe_winsys *ws = pscreen->winsys; struct nv30_miptree *mt; mt = MALLOC(sizeof(struct nv30_miptree)); @@ -65,24 +64,26 @@ nv30_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) return NULL; mt->base = *pt; mt->base.refcount = 1; - mt->base.screen = screen; + mt->base.screen = pscreen; nv30_miptree_layout(mt); - mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - mt->total_size); + mt->buffer = ws->buffer_create(ws, 256, + PIPE_BUFFER_USAGE_PIXEL | + NOUVEAU_BUFFER_USAGE_TEXTURE, + mt->total_size); if (!mt->buffer) { FREE(mt); return NULL; } - + return &mt->base; } static void -nv30_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) +nv30_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) { - struct pipe_winsys *ws = screen->winsys; + struct pipe_winsys *ws = pscreen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -106,10 +107,10 @@ nv30_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, } static struct pipe_surface * -nv30_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) +nv30_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice) { - struct pipe_winsys *ws = screen->winsys; + struct pipe_winsys *ws = pscreen->winsys; struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; struct pipe_surface *ps; @@ -146,6 +147,6 @@ nv30_screen_init_miptree_functions(struct pipe_screen *pscreen) { pscreen->texture_create = nv30_miptree_create; pscreen->texture_release = nv30_miptree_release; - pscreen->get_tex_surface = nv30_miptree_surface_get; + pscreen->get_tex_surface = nv30_miptree_surface; } -- cgit v1.2.3 From 7d2085bd7d9378703bfff3d974536ff5e9fd9b30 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 15 Apr 2008 13:25:28 +1000 Subject: nouveau: PIPE_ATTRIB_MAX -> PIPE_MAX_ATTRIBS --- src/gallium/drivers/nv10/nv10_context.h | 4 ++-- src/gallium/drivers/nv10/nv10_vbo.c | 4 ++-- src/gallium/drivers/nv30/nv30_context.h | 4 ++-- src/gallium/drivers/nv40/nv40_context.h | 4 ++-- 4 files changed, 8 insertions(+), 8 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 63d33ef7c91..b80f36ad34c 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -96,8 +96,8 @@ struct nv10_context { struct pipe_buffer *constant_buf; } fragprog; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; - struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; }; static INLINE struct nv10_context * diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c index 3a4f49e1564..2a334e137d9 100644 --- a/src/gallium/drivers/nv10/nv10_vbo.c +++ b/src/gallium/drivers/nv10/nv10_vbo.c @@ -23,7 +23,7 @@ boolean nv10_draw_elements( struct pipe_context *pipe, /* * Map vertex buffers */ - for (i = 0; i < PIPE_ATTRIB_MAX; i++) { + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { if (nv10->vtxbuf[i].buffer) { void *buf = pipe->winsys->buffer_map(pipe->winsys, @@ -52,7 +52,7 @@ boolean nv10_draw_elements( struct pipe_context *pipe, /* * unmap vertex/index buffers */ - for (i = 0; i < PIPE_ATTRIB_MAX; i++) { + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { if (nv10->vtxbuf[i].buffer) { pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer); draw_set_mapped_vertex_buffer(draw, i, NULL); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 180969731bf..1a016405e62 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -124,8 +124,8 @@ struct nv30_context { struct pipe_buffer *constant_buf; } fragprog; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; - struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; }; static INLINE struct nv30_context * diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 436351b6bcc..2f10540ff0c 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -151,9 +151,9 @@ struct nv40_context { unsigned nr_samplers; unsigned nr_textures; unsigned dirty_samplers; - struct pipe_vertex_buffer vtxbuf[PIPE_ATTRIB_MAX]; + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; unsigned vtxbuf_nr; - struct pipe_vertex_element vtxelt[PIPE_ATTRIB_MAX]; + struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; unsigned vtxelt_nr; }; -- cgit v1.2.3 From e7bdf047f28ea9b928b3890c37d7d20db1e67e96 Mon Sep 17 00:00:00 2001 From: Peter Winters Date: Sun, 20 Apr 2008 16:07:04 +0200 Subject: nv10: fix random stuff --- src/gallium/drivers/nv10/nv10_context.c | 36 ++++++++++++++++++++++++++--- src/gallium/drivers/nv10/nv10_context.h | 5 +++- src/gallium/drivers/nv10/nv10_prim_vbuf.c | 4 ++-- src/gallium/drivers/nv10/nv10_state.c | 20 +++++++++------- src/gallium/drivers/nv10/nv10_state_emit.c | 37 +++++++++++++++++------------- 5 files changed, 72 insertions(+), 30 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index 42c496b9590..bbd307d5d98 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -12,6 +12,8 @@ nv10_flush(struct pipe_context *pipe, unsigned flags, { struct nv10_context *nv10 = nv10_context(pipe); + draw_flush(nv10->draw); + FIRE_RING(fence); } @@ -31,6 +33,7 @@ static void nv10_init_hwctx(struct nv10_context *nv10) struct nv10_screen *screen = nv10->screen; struct nouveau_winsys *nvws = screen->nvws; int i; + float projectionmatrix[16]; BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); OUT_RING (screen->sync->handle); @@ -93,13 +96,21 @@ static void nv10_init_hwctx(struct nv10_context *nv10) BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_RC_OUT_ALPHA(0), 6); + + BEGIN_RING(celsius, NV10TCL_RC_IN_ALPHA(0), 12); + OUT_RING (0x30141010); + OUT_RING (0); + OUT_RING (0x20040000); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); OUT_RING (0x00000c00); OUT_RING (0); OUT_RING (0x00000c00); OUT_RING (0x18000000); - OUT_RING (0x300c0000); - OUT_RING (0x00001c80); + OUT_RING (0x300e0300); + OUT_RING (0x0c091c80); + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1); OUT_RING (0); BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2); @@ -219,6 +230,25 @@ static void nv10_init_hwctx(struct nv10_context *nv10) BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1); OUT_RING (1); + memset(projectionmatrix, 0, sizeof(projectionmatrix)); + BEGIN_RING(celsius, NV10TCL_PROJECTION_MATRIX(0), 16); + projectionmatrix[0*4+0] = 1.0; + projectionmatrix[1*4+1] = 1.0; + projectionmatrix[2*4+2] = 1.0; + projectionmatrix[3*4+3] = 1.0; + for (i=0;i<16;i++) { + OUT_RINGf (projectionmatrix[i]); + } + + BEGIN_RING(celsius, NV10TCL_DEPTH_RANGE_NEAR, 2); + OUT_RING (0.0); + OUT_RINGf (16777216.0); + + BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); + OUT_RINGf (-2048.0); + OUT_RINGf (-2048.0); + OUT_RINGf (16777215.0 * 0.5); + OUT_RING (0); FIRE_RING (NULL); } diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index b80f36ad34c..1b794c18724 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -65,7 +65,10 @@ struct nv10_context { struct pipe_viewport_state *viewport; struct pipe_scissor_state *scissor; struct pipe_framebuffer_state *framebuffer; - struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + + //struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + float *constbuf[PIPE_SHADER_TYPES][32][4]; + struct vertex_info vertex_info; struct { diff --git a/src/gallium/drivers/nv10/nv10_prim_vbuf.c b/src/gallium/drivers/nv10/nv10_prim_vbuf.c index 412ee9a23f9..930536b9468 100644 --- a/src/gallium/drivers/nv10/nv10_prim_vbuf.c +++ b/src/gallium/drivers/nv10/nv10_prim_vbuf.c @@ -205,8 +205,8 @@ nv10_vbuf_render_create( struct nv10_context *nv10 ) nv10_render->nv10 = nv10; - nv10_render->base.max_vertex_buffer_bytes = 1024*1024; - nv10_render->base.max_indices = 64*1024; + nv10_render->base.max_vertex_buffer_bytes = 16*1024; + nv10_render->base.max_indices = 1024; nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info; nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices; nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive; diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 1d9a202dd7c..4dcb9a31ab1 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -459,14 +459,18 @@ nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, const struct pipe_constant_buffer *buf ) { struct nv10_context *nv10 = nv10_context(pipe); + struct pipe_winsys *ws = pipe->winsys; - if (shader == PIPE_SHADER_VERTEX) { - nv10->constbuf[PIPE_SHADER_VERTEX] = buf->buffer; - nv10->dirty |= NV10_NEW_VERTPROG; - } else - if (shader == PIPE_SHADER_FRAGMENT) { - nv10->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; - nv10->dirty |= NV10_NEW_FRAGPROG; + assert(shader < PIPE_SHADER_TYPES); + assert(index == 0); + + if (buf) { + void *mapped; + if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ))) + { + memcpy(nv10->constbuf[shader], mapped, buf->size); + ws->buffer_unmap(ws, buf->buffer); + } } } @@ -507,7 +511,7 @@ nv10_set_viewport_state(struct pipe_context *pipe, nv10->viewport = (struct pipe_viewport_state*)vpt; - draw_set_viewport_state(nv10->draw, &nv10->viewport); + draw_set_viewport_state(nv10->draw, nv10->viewport); nv10->dirty |= NV10_NEW_VIEWPORT; } diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index edc194588d1..b7ae57d500a 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -63,29 +63,34 @@ static void nv10_state_emit_dsa(struct nv10_context* nv10) { struct nv10_depth_stencil_alpha_state *d = nv10->dsa; - BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&d->depth, 3); + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + OUT_RING (d->depth.func); + + BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + OUT_RING (d->depth.write_enable); + + BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + OUT_RING (d->depth.test_enable); + +#if 0 BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); OUT_RING (d->stencil.enable); BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 3); - OUT_RINGp ((uint32_t *)&d->alpha.enabled, 3); +#endif + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (d->alpha.enabled); + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 1); + OUT_RING (d->alpha.func); + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_REF, 1); + OUT_RING (d->alpha.ref); } static void nv10_state_emit_viewport(struct nv10_context* nv10) { - struct pipe_viewport_state *vpt = nv10->viewport; - -/* OUT_RINGf (vpt->translate[0]); - OUT_RINGf (vpt->translate[1]); - OUT_RINGf (vpt->translate[2]); - OUT_RINGf (vpt->translate[3]);*/ - BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); - OUT_RINGf (vpt->scale[0]); - OUT_RINGf (vpt->scale[1]); - OUT_RINGf (vpt->scale[2]); - OUT_RINGf (vpt->scale[3]); } static void nv10_state_emit_scissor(struct nv10_context* nv10) @@ -100,7 +105,7 @@ static void nv10_state_emit_scissor(struct nv10_context* nv10) static void nv10_state_emit_framebuffer(struct nv10_context* nv10) { struct pipe_framebuffer_state* fb = nv10->framebuffer; - struct pipe_surface *rt, *zeta; + struct pipe_surface *rt, *zeta = NULL; uint32_t rt_format, w, h; int colour_format = 0, zeta_format = 0; -- cgit v1.2.3 From b20acef90695d6e5975f538b6e9cb812b05f0cf6 Mon Sep 17 00:00:00 2001 From: Peter Winters Date: Sun, 20 Apr 2008 22:48:50 +0200 Subject: nv10: enable viewport clipping --- src/gallium/drivers/nv10/nv10_state_emit.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index b7ae57d500a..41422c88825 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -161,8 +161,8 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10) OUT_RING ((h << 16) | 0); OUT_RING (rt_format); BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0); - OUT_RING (((h - 1) << 16) | 0); + OUT_RING (((w - 1) << 16) | 0 | 0x08000800); + OUT_RING (((h - 1) << 16) | 0 | 0x08000800); } static void nv10_vertex_layout(struct nv10_context *nv10) -- cgit v1.2.3 From a3265958994ce4107da2a3954c04b496e29cd8aa Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 23 Apr 2008 12:41:42 +1000 Subject: nouveau: fix build --- src/gallium/drivers/nv30/nv30_draw.c | 2 +- src/gallium/drivers/nv40/nv40_draw.c | 2 +- src/gallium/drivers/nv50/nv50_draw.c | 2 +- src/gallium/drivers/nv50/nv50_state.c | 1 - 4 files changed, 3 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_draw.c b/src/gallium/drivers/nv30/nv30_draw.c index 8ec0835225b..aeeaf58f20a 100644 --- a/src/gallium/drivers/nv30/nv30_draw.c +++ b/src/gallium/drivers/nv30/nv30_draw.c @@ -1,4 +1,4 @@ -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" #include "pipe/p_util.h" #include "nv30_context.h" diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index dd5cc8fc998..a9a939af0c0 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -3,7 +3,7 @@ #include "draw/draw_context.h" #include "draw/draw_vertex.h" -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" #include "nv40_context.h" #define NV40_SHADER_NO_FUCKEDNESS diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c index 790408c6df8..d185d999502 100644 --- a/src/gallium/drivers/nv50/nv50_draw.c +++ b/src/gallium/drivers/nv50/nv50_draw.c @@ -1,4 +1,4 @@ -#include "draw/draw_private.h" +#include "draw/draw_pipe.h" #include "pipe/p_util.h" #include "nv50_context.h" diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index a614ea03358..f42bae0c28d 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -113,7 +113,6 @@ nv50_rasterizer_state_create(struct pipe_context *pipe, struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla; struct nv50_rasterizer_stateobj *rso = CALLOC_STRUCT(nv50_rasterizer_stateobj); - unsigned i; /*XXX: ignored * - light_twosize -- cgit v1.2.3 From 7342688286cc3b7c938af2dfeac22df4fa8c8464 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Wed, 23 Apr 2008 22:38:49 +0200 Subject: nv30: add stuff to init swtnl --- src/gallium/drivers/nv30/nv30_context.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index d38713e98ad..57b0e71dad7 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -63,7 +63,12 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30_init_state_functions(nv30); nv30_init_miptree_functions(nv30); + /* Create, configure, and install fallback swtnl path */ nv30->draw = draw_create(); + draw_wide_point_threshold(nv30->draw, 9999999.0); + draw_wide_line_threshold(nv30->draw, 9999999.0); + draw_enable_line_stipple(nv30->draw, FALSE); + draw_enable_point_sprites(nv30->draw, FALSE); draw_set_rasterize_stage(nv30->draw, nv30_draw_render_stage(nv30)); return &nv30->pipe; -- cgit v1.2.3 From 2193578851b3b5a99c078b28187cf3158f4218f6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 27 Apr 2008 18:12:14 +1000 Subject: nouveau: stub set_edgeflags for all nv pipe drivers --- src/gallium/drivers/nv10/nv10_context.c | 6 ++++++ src/gallium/drivers/nv30/nv30_context.c | 7 +++++++ src/gallium/drivers/nv40/nv40_context.c | 6 ++++++ src/gallium/drivers/nv50/nv50_context.c | 7 +++++++ 4 files changed, 26 insertions(+) diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index bbd307d5d98..a25c082a5d1 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -253,6 +253,11 @@ static void nv10_init_hwctx(struct nv10_context *nv10) FIRE_RING (NULL); } +static void +nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + struct pipe_context * nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) { @@ -274,6 +279,7 @@ nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) nv10->pipe.winsys = ws; nv10->pipe.screen = pscreen; nv10->pipe.destroy = nv10_destroy; + nv10->pipe.set_edgeflags = nv10_set_edgeflags; nv10->pipe.draw_arrays = nv10_draw_arrays; nv10->pipe.draw_elements = nv10_draw_elements; nv10->pipe.clear = nv10_clear; diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 57b0e71dad7..4b0892971ec 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -32,6 +32,12 @@ nv30_destroy(struct pipe_context *pipe) FREE(nv30); } + +static void +nv30_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + struct pipe_context * nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) { @@ -53,6 +59,7 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30->pipe.winsys = ws; nv30->pipe.screen = pscreen; nv30->pipe.destroy = nv30_destroy; + nv30->pipe.set_edgeflags = nv30_set_edgeflags; nv30->pipe.draw_arrays = nv30_draw_arrays; nv30->pipe.draw_elements = nv30_draw_elements; nv30->pipe.clear = nv30_clear; diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index f9c93f7a2d7..f1d4f3c76f1 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -32,6 +32,11 @@ nv40_destroy(struct pipe_context *pipe) FREE(nv40); } +static void +nv40_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + struct pipe_context * nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) { @@ -53,6 +58,7 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) nv40->pipe.winsys = ws; nv40->pipe.screen = pscreen; nv40->pipe.destroy = nv40_destroy; + nv40->pipe.set_edgeflags = nv40_set_edgeflags; nv40->pipe.draw_arrays = nv40_draw_arrays; nv40->pipe.draw_elements = nv40_draw_elements; nv40->pipe.clear = nv40_clear; diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index e822d863945..6eb1878b847 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -24,6 +24,12 @@ nv50_destroy(struct pipe_context *pipe) free(nv50); } + +static void +nv50_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + struct pipe_context * nv50_create(struct pipe_screen *pscreen, unsigned pctx_id) { @@ -42,6 +48,7 @@ nv50_create(struct pipe_screen *pscreen, unsigned pctx_id) nv50->pipe.destroy = nv50_destroy; + nv50->pipe.set_edgeflags = nv50_set_edgeflags; nv50->pipe.draw_arrays = nv50_draw_arrays; nv50->pipe.draw_elements = nv50_draw_elements; nv50->pipe.clear = nv50_clear; -- cgit v1.2.3 From 58b1bcaa094ed07a54bd7e4cdbddbcdcf1f629a7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 28 Apr 2008 15:42:19 +1000 Subject: nv40: do full swtnl fallback when edge flags present. This isn't necessary, with some effort we can do this on the hw. However, until I encounter something "real" that uses them there's not a lot of point. --- src/gallium/drivers/nv40/nv40_context.c | 6 ------ src/gallium/drivers/nv40/nv40_context.h | 1 + src/gallium/drivers/nv40/nv40_state.c | 11 +++++++++++ src/gallium/drivers/nv40/nv40_state_emit.c | 1 + src/gallium/drivers/nv40/nv40_vbo.c | 5 +++++ 5 files changed, 18 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index f1d4f3c76f1..f9c93f7a2d7 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -32,11 +32,6 @@ nv40_destroy(struct pipe_context *pipe) FREE(nv40); } -static void -nv40_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) -{ -} - struct pipe_context * nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) { @@ -58,7 +53,6 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) nv40->pipe.winsys = ws; nv40->pipe.screen = pscreen; nv40->pipe.destroy = nv40_destroy; - nv40->pipe.set_edgeflags = nv40_set_edgeflags; nv40->pipe.draw_arrays = nv40_draw_arrays; nv40->pipe.draw_elements = nv40_draw_elements; nv40->pipe.clear = nv40_clear; diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 2f10540ff0c..24e8cd23379 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -155,6 +155,7 @@ struct nv40_context { unsigned vtxbuf_nr; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; unsigned vtxelt_nr; + const unsigned *edgeflags; }; static INLINE struct nv40_context * diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 997beca883a..2d921d2b8a4 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -680,6 +680,16 @@ nv40_set_vertex_elements(struct pipe_context *pipe, unsigned count, nv40->draw_dirty |= NV40_NEW_ARRAYS; } +static void +nv40_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ + struct nv40_context *nv40 = nv40_context(pipe); + + nv40->edgeflags = bitfield; + nv40->dirty |= NV40_NEW_ARRAYS; + nv40->draw_dirty |= NV40_NEW_ARRAYS; +} + void nv40_init_state_functions(struct nv40_context *nv40) { @@ -719,6 +729,7 @@ nv40_init_state_functions(struct nv40_context *nv40) nv40->pipe.set_scissor_state = nv40_set_scissor_state; nv40->pipe.set_viewport_state = nv40_set_viewport_state; + nv40->pipe.set_edgeflags = nv40_set_edgeflags; nv40->pipe.set_vertex_buffers = nv40_set_vertex_buffers; nv40->pipe.set_vertex_elements = nv40_set_vertex_elements; } diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 864dfc2e0c3..ab88dc416e5 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -167,6 +167,7 @@ nv40_state_validate_swtnl(struct nv40_context *nv40) draw_set_viewport_state(draw, &nv40->viewport); if (nv40->draw_dirty & NV40_NEW_ARRAYS) { + draw_set_edgeflags(draw, nv40->edgeflags); draw_set_vertex_buffers(draw, nv40->vtxbuf_nr, nv40->vtxbuf); draw_set_vertex_elements(draw, nv40->vtxelt_nr, nv40->vtxelt); } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 06374184b12..e5f9bd5668a 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -488,6 +488,11 @@ nv40_vbo_validate(struct nv40_context *nv40) unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; int hw; + if (nv40->edgeflags) { + nv40->fallback_swtnl |= NV40_NEW_ARRAYS; + return FALSE; + } + vtxbuf = so_new(20, 18); so_method(vtxbuf, curie, NV40TCL_VTXBUF_ADDRESS(0), nv40->vtxelt_nr); vtxfmt = so_new(17, 0); -- cgit v1.2.3 From 480ab1b7893290505efba925ea95d3b173aa97d4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 29 Apr 2008 08:56:41 +1000 Subject: nv40: enable DXTn formats GL state tracker capable enough for progs/tests/texcompress2 at least. --- src/gallium/drivers/nv40/nv40_screen.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index a408d7262f3..c64c3b1c39d 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -39,7 +39,7 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param) case PIPE_CAP_GLSL: return 0; case PIPE_CAP_S3TC: - return 0; + return 1; case PIPE_CAP_ANISOTROPIC_FILTER: return 1; case PIPE_CAP_POINT_SPRITE: @@ -118,12 +118,10 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_U_A8_L8: case PIPE_FORMAT_Z16_UNORM: case PIPE_FORMAT_Z24S8_UNORM: -#if 0 /* state tracker not up to the task just yet. */ case PIPE_FORMAT_DXT1_RGB: case PIPE_FORMAT_DXT1_RGBA: case PIPE_FORMAT_DXT3_RGBA: case PIPE_FORMAT_DXT5_RGBA: -#endif return TRUE; default: break; -- cgit v1.2.3 From 612f44266cba78c4e5677a2f992581fdaa17f4e4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 5 May 2008 19:45:21 +1000 Subject: nouveau: bitmap texcoord bias has been removed --- src/gallium/drivers/nv10/nv10_screen.c | 2 -- src/gallium/drivers/nv30/nv30_screen.c | 2 -- src/gallium/drivers/nv40/nv40_screen.c | 2 -- src/gallium/drivers/nv50/nv50_screen.c | 2 -- 4 files changed, 8 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 45fbde62ea2..cf6b2fac84e 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -73,8 +73,6 @@ nv10_screen_get_paramf(struct pipe_screen *screen, int param) return 2.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index ce6c9ec523a..1de45079043 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -77,8 +77,6 @@ nv30_screen_get_paramf(struct pipe_screen *pscreen, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index c64c3b1c39d..7f68539a85c 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -82,8 +82,6 @@ nv40_screen_get_paramf(struct pipe_screen *pscreen, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 16.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index adb724b9b7a..4902f16de3a 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -86,8 +86,6 @@ nv50_screen_get_paramf(struct pipe_screen *pscreen, int param) return 16.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; - case PIPE_CAP_BITMAP_TEXCOORD_BIAS: - return 0.0; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0.0; -- cgit v1.2.3 From 32ed02bcfbe7a2132929658b1a73708ab16af006 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 5 May 2008 19:49:06 +1000 Subject: nv50: report some supported formats to keep the state tracker from asserting. --- src/gallium/drivers/nv50/nv50_screen.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 4902f16de3a..e4ca72c7171 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -14,6 +14,30 @@ static boolean nv50_screen_is_format_supported(struct pipe_screen *pscreen, enum pipe_format format, uint type) { + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_I8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + } + return FALSE; } -- cgit v1.2.3 From 9b0054c7f87e3cc89fc0e60408af41f3e86dfdff Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 5 May 2008 19:52:19 +1000 Subject: nv50: slightly less skeletal texture funcs, prevents fun segfaults --- src/gallium/drivers/nv50/nv50_miptree.c | 62 ++++++++++++++++++++++++++++++--- 1 file changed, 58 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 7474c657650..58584934b18 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -1,29 +1,83 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" -#include "pipe/p_screen.h" +#include "pipe/p_inlines.h" #include "nv50_context.h" +struct nv50_miptree { + struct pipe_texture base; + struct pipe_buffer *buffer; +}; + +static INLINE struct nv50_miptree * +nv50_miptree(struct pipe_texture *pt) +{ + return (struct nv50_miptree *)pt; +} + static struct pipe_texture * nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { + struct pipe_winsys *ws = pscreen->winsys; + struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree); + NOUVEAU_ERR("unimplemented\n"); - return NULL; + + mt->base = *pt; + mt->base.refcount = 1; + mt->base.screen = pscreen; + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + 512*32*4); + if (!mt->buffer) { + FREE(mt); + return NULL; + } + + return &mt->base; } static void -nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) +nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt) { + struct pipe_winsys *ws = pscreen->winsys; + struct pipe_texture *pt = *ppt; + NOUVEAU_ERR("unimplemented\n"); + + *ppt = NULL; + if (--pt->refcount <= 0) { + struct nv50_miptree *mt = nv50_miptree(pt); + + pipe_buffer_reference(ws, &mt->buffer, NULL); + FREE(mt); + } } static struct pipe_surface * nv50_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice) { + struct pipe_winsys *ws = pscreen->winsys; + struct nv50_miptree *mt = nv50_miptree(pt); + struct pipe_surface *ps; + NOUVEAU_ERR("unimplemented\n"); - return NULL; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + + pipe_buffer_reference(ws, &ps->buffer, mt->buffer); + ps->format = pt->format; + ps->cpp = pt->cpp; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->pitch = ps->width; + ps->offset = 0; + + return ps; } void -- cgit v1.2.3 From 83e6df12ea31ef48d251fe7b38acd882145f8f14 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 13 May 2008 12:01:08 +1000 Subject: nouveau: chipset is a device property, remove from nv/nvc. --- src/gallium/winsys/dri/nouveau/nouveau_context.c | 36 ++++++++++-------------- src/gallium/winsys/dri/nouveau/nouveau_context.h | 5 ---- src/gallium/winsys/dri/nouveau/nouveau_device.c | 13 +++++++++ src/gallium/winsys/dri/nouveau/nouveau_device.h | 1 + src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 8 ++++-- src/gallium/winsys/dri/nouveau/nv04_surface.c | 15 +++++----- 6 files changed, 41 insertions(+), 37 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index ff08a870db1..e65b057335c 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -43,7 +43,7 @@ nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) } static struct nouveau_channel_context * -nouveau_channel_context_create(struct nouveau_device *nvdev, unsigned chipset) +nouveau_channel_context_create(struct nouveau_device *dev) { struct nouveau_channel_context *nvc; int ret; @@ -51,9 +51,8 @@ nouveau_channel_context_create(struct nouveau_device *nvdev, unsigned chipset) nvc = CALLOC_STRUCT(nouveau_channel_context); if (!nvc) return NULL; - nvc->chipset = chipset; - if ((ret = nouveau_channel_alloc(nvdev, 0x8003d001, 0x8003d002, + if ((ret = nouveau_channel_alloc(dev, 0x8003d001, 0x8003d002, &nvc->channel))) { NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); nouveau_channel_context_destroy(nvc); @@ -75,7 +74,7 @@ nouveau_channel_context_create(struct nouveau_device *nvdev, unsigned chipset) return NULL; } - switch (chipset & 0xf0) { + switch (dev->chipset & 0xf0) { case 0x50: case 0x80: case 0x90: @@ -103,25 +102,17 @@ nouveau_context_create(const __GLcontextModes *glVis, __DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv; struct nouveau_screen *nv_screen = driScrnPriv->private; struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); - struct nouveau_device_priv *nvdev; struct pipe_context *pipe = NULL; struct st_context *st_share = NULL; struct nouveau_channel_context *nvc = NULL; + struct nouveau_device *dev = nv_screen->device; int i, ret; if (sharedContextPrivate) { st_share = ((struct nouveau_context *)sharedContextPrivate)->st; } - /* Check for supported arch */ - if ((ret = nouveau_device_get_param(nv_screen->device, - NOUVEAU_GETPARAM_CHIPSET_ID, - &nv->chipset))) { - NOUVEAU_ERR("Error determining chipset id: %d\n", ret); - return GL_FALSE; - } - - switch (nv->chipset & 0xf0) { + switch (dev->chipset & 0xf0) { case 0x10: case 0x20: /* NV10 */ @@ -136,7 +127,7 @@ nouveau_context_create(const __GLcontextModes *glVis, /* G80 */ break; default: - NOUVEAU_ERR("Unsupported chipset: NV%02x\n", (int)nv->chipset); + NOUVEAU_ERR("Unsupported chipset: NV%02x\n", dev->chipset); return GL_FALSE; } @@ -144,9 +135,12 @@ nouveau_context_create(const __GLcontextModes *glVis, nv->nv_screen = nv_screen; nv->dri_screen = driScrnPriv; - nvdev = nouveau_device(nv_screen->device); - nvdev->ctx = driContextPriv->hHWContext; - nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; + { + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + nvdev->ctx = driContextPriv->hHWContext; + nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; + } driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, nv->dri_screen->myNum, "nouveau"); @@ -201,7 +195,7 @@ nouveau_context_create(const __GLcontextModes *glVis, } /*XXX: temporary - disable multi-context/single-channel on pre-NV4x */ - switch (nv->chipset & 0xf0) { + switch (dev->chipset & 0xf0) { case 0x40: case 0x60: /* NV40 class */ @@ -216,7 +210,7 @@ nouveau_context_create(const __GLcontextModes *glVis, } if (!nvc) { - nvc = nouveau_channel_context_create(&nvdev->base, nv->chipset); + nvc = nouveau_channel_context_create(dev); if (!nvc) { NOUVEAU_ERR("Failed initialising GPU context\n"); return GL_FALSE; @@ -244,7 +238,7 @@ nouveau_context_create(const __GLcontextModes *glVis, } /* Create pipe */ - switch (nv->chipset & 0xf0) { + switch (dev->chipset & 0xf0) { case 0x50: case 0x80: case 0x90: diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index 872ef93807a..9872d6b6914 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -21,8 +21,6 @@ struct nouveau_channel_context { unsigned nr_pctx; struct pipe_context **pctx; - unsigned chipset; - struct nouveau_channel *channel; struct nouveau_notifier *sync_notifier; @@ -47,9 +45,6 @@ struct nouveau_channel_context { struct nouveau_context { struct st_context *st; - /* Misc HW info */ - uint64_t chipset; - /* DRI stuff */ __DRIscreenPrivate *dri_screen; __DRIdrawablePrivate *dri_drawable; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_device.c b/src/gallium/winsys/dri/nouveau/nouveau_device.c index 409e4415f76..0b452fcd02d 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_device.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_device.c @@ -50,6 +50,19 @@ nouveau_device_open_existing(struct nouveau_device **dev, int close, return ret; } + { + uint64_t value; + + ret = nouveau_device_get_param(&nvdev->base, + NOUVEAU_GETPARAM_CHIPSET_ID, + &value); + if (ret) { + nouveau_device_close((void *)&nvdev); + return ret; + } + nvdev->base.chipset = value; + } + *dev = &nvdev->base; return 0; } diff --git a/src/gallium/winsys/dri/nouveau/nouveau_device.h b/src/gallium/winsys/dri/nouveau/nouveau_device.h index 744a89f74bf..e25e89fedda 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_device.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_device.h @@ -24,6 +24,7 @@ #define __NOUVEAU_DEVICE_H__ struct nouveau_device { + unsigned chipset; }; #endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 60fdbb8dfdf..635fd478a94 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -1,6 +1,7 @@ #include "pipe/p_util.h" #include "nouveau_context.h" +#include "nouveau_screen.h" #include "nouveau_winsys_pipe.h" #include "nouveau/nouveau_winsys.h" @@ -94,11 +95,12 @@ nouveau_pipe_create(struct nouveau_context *nv) unsigned chipset); struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned); struct pipe_winsys *ws; + unsigned chipset = nv->nv_screen->device->chipset; if (!nvws) return NULL; - switch (nv->chipset & 0xf0) { + switch (chipset & 0xf0) { case 0x10: case 0x20: hws_create = nv10_screen_create; @@ -120,7 +122,7 @@ nouveau_pipe_create(struct nouveau_context *nv) hw_create = nv50_create; break; default: - NOUVEAU_ERR("Unknown chipset NV%02x\n", (int)nv->chipset); + NOUVEAU_ERR("Unknown chipset NV%02x\n", chipset); return NULL; } @@ -150,7 +152,7 @@ nouveau_pipe_create(struct nouveau_context *nv) ws = nouveau_create_pipe_winsys(nv); if (!nvc->pscreen) - nvc->pscreen = hws_create(ws, nvws, nv->chipset); + nvc->pscreen = hws_create(ws, nvws, chipset); nvc->pctx[nv->pctx_id] = hw_create(nvc->pscreen, nv->pctx_id); return nvc->pctx[nv->pctx_id]; } diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c index f61bd1477ff..dc167e4acfb 100644 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -179,7 +179,7 @@ int nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) { struct nouveau_channel *chan = nvc->channel; - unsigned class; + unsigned chipset = nvc->channel->device->chipset, class; int ret; if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, 0x39, @@ -192,8 +192,8 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); OUT_RING (chan, nvc->sync_notifier->handle); - class = nvc->chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : - NV10_CONTEXT_SURFACES_2D; + class = chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : + NV10_CONTEXT_SURFACES_2D; if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, &nvc->NvCtxSurf2D))) { NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); @@ -205,8 +205,7 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) OUT_RING (chan, nvc->channel->vram->handle); OUT_RING (chan, nvc->channel->vram->handle); - class = nvc->chipset < 0x10 ? NV04_IMAGE_BLIT : - NV12_IMAGE_BLIT; + class = chipset < 0x10 ? NV04_IMAGE_BLIT : NV12_IMAGE_BLIT; if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, &nvc->NvImageBlit))) { NOUVEAU_ERR("Error creating blit object: %d\n", ret); @@ -237,7 +236,7 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); - switch (nvc->chipset & 0xf0) { + switch (chipset & 0xf0) { case 0x00: case 0x10: class = NV04_SWIZZLED_SURFACE; @@ -267,10 +266,10 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) BIND_RING (chan, nvc->NvSwzSurf, nvc->next_subchannel++); - if (nvc->chipset < 0x10) { + if (chipset < 0x10) { class = NV04_SCALED_IMAGE_FROM_MEMORY; } else - if (nvc->chipset < 0x40) { + if (chipset < 0x40) { class = NV10_SCALED_IMAGE_FROM_MEMORY; } else { class = NV40_SCALED_IMAGE_FROM_MEMORY; -- cgit v1.2.3 From c962ad7cd5dbea12d13997b421a44b16af3c6662 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 13 May 2008 12:06:32 +1000 Subject: nouveau: create objnull during channel creation --- src/gallium/drivers/nouveau/nouveau_channel.h | 1 + src/gallium/winsys/dri/nouveau/nouveau_channel.c | 8 ++++++++ src/gallium/winsys/dri/nouveau/nouveau_context.c | 7 ------- src/gallium/winsys/dri/nouveau/nouveau_context.h | 1 - src/gallium/winsys/dri/nouveau/nouveau_grobj.c | 2 +- 5 files changed, 10 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_channel.h b/src/gallium/drivers/nouveau/nouveau_channel.h index b99de9add86..cd99a676bdc 100644 --- a/src/gallium/drivers/nouveau/nouveau_channel.h +++ b/src/gallium/drivers/nouveau/nouveau_channel.h @@ -29,6 +29,7 @@ struct nouveau_channel { struct nouveau_pushbuf *pushbuf; + struct nouveau_grobj *nullobj; struct nouveau_grobj *vram; struct nouveau_grobj *gart; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_channel.c b/src/gallium/winsys/dri/nouveau/nouveau_channel.c index df80d04add5..3b4dcd1ecf2 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_channel.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_channel.c @@ -85,6 +85,13 @@ nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma, return ret; } + ret = nouveau_grobj_alloc(&nvchan->base, 0x00000000, 0x0030, + &nvchan->base.nullobj); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + nouveau_dma_channel_init(&nvchan->base); nouveau_pushbuf_init(&nvchan->base); @@ -109,6 +116,7 @@ nouveau_channel_free(struct nouveau_channel **chan) nouveau_grobj_free(&nvchan->base.vram); nouveau_grobj_free(&nvchan->base.gart); + nouveau_grobj_free(&nvchan->base.nullobj); cf.channel = nvchan->drm.channel; drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf)); diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index e65b057335c..d9fc3f6ce1f 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -26,7 +26,6 @@ int __nouveau_debug = 0; static void nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) { - nouveau_grobj_free(&nvc->NvNull); nouveau_grobj_free(&nvc->NvCtxSurf2D); nouveau_grobj_free(&nvc->NvImageBlit); nouveau_grobj_free(&nvc->NvGdiRect); @@ -59,12 +58,6 @@ nouveau_channel_context_create(struct nouveau_device *dev) return NULL; } - if ((ret = nouveau_grobj_alloc(nvc->channel, 0x00000000, 0x30, - &nvc->NvNull))) { - NOUVEAU_ERR("Error creating NULL object: %d\n", ret); - nouveau_channel_context_destroy(nvc); - return NULL; - } nvc->next_handle = 0x80000000; if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1, diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index 9872d6b6914..b20107a94c6 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -26,7 +26,6 @@ struct nouveau_channel_context { struct nouveau_notifier *sync_notifier; /* Common */ - struct nouveau_grobj *NvNull; struct nouveau_grobj *NvM2MF; /* NV04-NV40 */ struct nouveau_grobj *NvCtxSurf2D; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_grobj.c b/src/gallium/winsys/dri/nouveau/nouveau_grobj.c index 55dfeb99aa7..51523897d58 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_grobj.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_grobj.c @@ -50,7 +50,7 @@ nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle, ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC, &g, sizeof(g)); if (ret) { - nouveau_grobj_free((void *)&grobj); + nouveau_grobj_free((void *)&nvgrobj); return ret; } -- cgit v1.2.3 From 2f80d4d2a705835b272cf2274ea578fe5bbe1919 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 13 May 2008 12:09:04 +1000 Subject: nouveau: make nouveau_device.h part of public API --- src/gallium/drivers/nouveau/nouveau_device.h | 30 ++++++++++++++++++++++ src/gallium/winsys/dri/nouveau/nouveau_context.h | 1 - src/gallium/winsys/dri/nouveau/nouveau_device.h | 30 ---------------------- src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 2 +- src/gallium/winsys/dri/nouveau/nouveau_screen.c | 1 - src/gallium/winsys/dri/nouveau/nouveau_screen.h | 1 - .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 1 - 7 files changed, 31 insertions(+), 35 deletions(-) create mode 100644 src/gallium/drivers/nouveau/nouveau_device.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_device.h diff --git a/src/gallium/drivers/nouveau/nouveau_device.h b/src/gallium/drivers/nouveau/nouveau_device.h new file mode 100644 index 00000000000..e25e89fedda --- /dev/null +++ b/src/gallium/drivers/nouveau/nouveau_device.h @@ -0,0 +1,30 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DEVICE_H__ +#define __NOUVEAU_DEVICE_H__ + +struct nouveau_device { + unsigned chipset; +}; + +#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h index b20107a94c6..77e2147a2c7 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.h @@ -5,7 +5,6 @@ #include "xmlconfig.h" #include "nouveau/nouveau_winsys.h" -#include "nouveau_device.h" #include "nouveau_drmif.h" #include "nouveau_dma.h" diff --git a/src/gallium/winsys/dri/nouveau/nouveau_device.h b/src/gallium/winsys/dri/nouveau/nouveau_device.h deleted file mode 100644 index e25e89fedda..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_device.h +++ /dev/null @@ -1,30 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_DEVICE_H__ -#define __NOUVEAU_DEVICE_H__ - -struct nouveau_device { - unsigned chipset; -}; - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h index d5a0e25b573..a31c9a514b1 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h @@ -27,7 +27,7 @@ #include #include -#include "nouveau_device.h" +#include "nouveau/nouveau_device.h" #include "nouveau/nouveau_channel.h" #include "nouveau/nouveau_grobj.h" #include "nouveau/nouveau_notifier.h" diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c index f06e1784831..9041275a88f 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.c @@ -7,7 +7,6 @@ #include "state_tracker/st_cb_fbo.h" #include "nouveau_context.h" -#include "nouveau_device.h" #include "nouveau_drm.h" #include "nouveau_dri.h" #include "nouveau_local.h" diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.h b/src/gallium/winsys/dri/nouveau/nouveau_screen.h index e9da2026904..388d6be9bbc 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.h @@ -2,7 +2,6 @@ #define __NOUVEAU_SCREEN_H__ #include "xmlconfig.h" -#include "nouveau_device.h" struct nouveau_screen { __DRIscreenPrivate *driScrnPriv; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index 849e38d22b6..755c8b6997a 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -4,7 +4,6 @@ #include "pipe/p_inlines.h" #include "nouveau_context.h" -#include "nouveau_device.h" #include "nouveau_local.h" #include "nouveau_screen.h" #include "nouveau_swapbuffers.h" -- cgit v1.2.3 From 1ef08564d2a201a422db772a6bb23d1129888304 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 13 May 2008 12:16:35 +1000 Subject: nouveau: remove chipset fields in all nv pipe driver context/screen structs. --- src/gallium/drivers/nouveau/nouveau_winsys.h | 13 +++++-------- src/gallium/drivers/nv10/nv10_context.c | 2 -- src/gallium/drivers/nv10/nv10_context.h | 2 -- src/gallium/drivers/nv10/nv10_screen.c | 8 ++++---- src/gallium/drivers/nv10/nv10_screen.h | 1 - src/gallium/drivers/nv30/nv30_context.c | 2 -- src/gallium/drivers/nv30/nv30_context.h | 2 -- src/gallium/drivers/nv30/nv30_screen.c | 8 ++++---- src/gallium/drivers/nv30/nv30_screen.h | 1 - src/gallium/drivers/nv40/nv40_context.c | 2 -- src/gallium/drivers/nv40/nv40_context.h | 2 -- src/gallium/drivers/nv40/nv40_screen.c | 8 ++++---- src/gallium/drivers/nv40/nv40_screen.h | 1 - src/gallium/drivers/nv50/nv50_screen.c | 8 ++++---- src/gallium/drivers/nv50/nv50_screen.h | 1 - src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 5 ++--- 16 files changed, 23 insertions(+), 43 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index fbde7adc6ad..07a41dcf4a1 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -8,6 +8,7 @@ #include "nouveau/nouveau_bo.h" #include "nouveau/nouveau_channel.h" #include "nouveau/nouveau_class.h" +#include "nouveau/nouveau_device.h" #include "nouveau/nouveau_grobj.h" #include "nouveau/nouveau_notifier.h" #include "nouveau/nouveau_resource.h" @@ -56,29 +57,25 @@ struct nouveau_winsys { }; extern struct pipe_screen * -nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, - unsigned chipset); +nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); extern struct pipe_context * nv10_create(struct pipe_screen *, unsigned pctx_id); extern struct pipe_screen * -nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, - unsigned chipset); +nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); extern struct pipe_context * nv30_create(struct pipe_screen *, unsigned pctx_id); extern struct pipe_screen * -nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, - unsigned chipset); +nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); extern struct pipe_context * nv40_create(struct pipe_screen *, unsigned pctx_id); extern struct pipe_screen * -nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *, - unsigned chipset); +nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); extern struct pipe_context * nv50_create(struct pipe_screen *, unsigned pctx_id); diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index a25c082a5d1..79253f8a2b3 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -264,7 +264,6 @@ nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) struct nv10_screen *screen = nv10_screen(pscreen); struct pipe_winsys *ws = pscreen->winsys; struct nv10_context *nv10; - unsigned chipset = screen->chipset; struct nouveau_winsys *nvws = screen->nvws; nv10 = CALLOC(1, sizeof(struct nv10_context)); @@ -273,7 +272,6 @@ nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) nv10->screen = screen; nv10->pctx_id = pctx_id; - nv10->chipset = chipset; nv10->nvws = nvws; nv10->pipe.winsys = ws; diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 1b794c18724..433d04dc2a8 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -43,8 +43,6 @@ struct nv10_context { struct draw_context *draw; - int chipset; - uint32_t dirty; struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index cf6b2fac84e..fad96058b8e 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -8,9 +8,10 @@ static const char * nv10_screen_get_name(struct pipe_screen *screen) { struct nv10_screen *nv10screen = nv10_screen(screen); + struct nouveau_device *dev = nv10screen->nvws->channel->device; static char buffer[128]; - snprintf(buffer, sizeof(buffer), "NV%02X", nv10screen->chipset); + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); return buffer; } @@ -129,16 +130,15 @@ nv10_screen_destroy(struct pipe_screen *pscreen) } struct pipe_screen * -nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) +nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) { struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen); unsigned celsius_class; + unsigned chipset = nvws->channel->device->chipset; int ret; if (!screen) return NULL; - screen->chipset = chipset; screen->nvws = nvws; /* 3D object */ diff --git a/src/gallium/drivers/nv10/nv10_screen.h b/src/gallium/drivers/nv10/nv10_screen.h index 4192fe11ef4..3f8750a13f7 100644 --- a/src/gallium/drivers/nv10/nv10_screen.h +++ b/src/gallium/drivers/nv10/nv10_screen.h @@ -7,7 +7,6 @@ struct nv10_screen { struct pipe_screen pipe; struct nouveau_winsys *nvws; - unsigned chipset; /* HW graphics objects */ struct nouveau_grobj *celsius; diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 4b0892971ec..7a2fee78751 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -44,7 +44,6 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) struct nv30_screen *screen = nv30_screen(pscreen); struct pipe_winsys *ws = pscreen->winsys; struct nv30_context *nv30; - unsigned chipset = screen->chipset; struct nouveau_winsys *nvws = screen->nvws; nv30 = CALLOC(1, sizeof(struct nv30_context)); @@ -53,7 +52,6 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30->screen = screen; nv30->pctx_id = pctx_id; - nv30->chipset = chipset; nv30->nvws = nvws; nv30->pipe.winsys = ws; diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 1a016405e62..f49450f82d9 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -85,8 +85,6 @@ struct nv30_context { struct draw_context *draw; - int chipset; - uint32_t dirty; struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 1de45079043..9b0ac55ff88 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -12,9 +12,10 @@ static const char * nv30_screen_get_name(struct pipe_screen *pscreen) { struct nv30_screen *screen = nv30_screen(pscreen); + struct nouveau_device *dev = screen->nvws->channel->device; static char buffer[128]; - snprintf(buffer, sizeof(buffer), "NV%02X", screen->chipset); + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); return buffer; } @@ -140,17 +141,16 @@ nv30_screen_destroy(struct pipe_screen *pscreen) } struct pipe_screen * -nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) +nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) { struct nv30_screen *screen = CALLOC_STRUCT(nv30_screen); struct nouveau_stateobj *so; unsigned rankine_class = 0; + unsigned chipset = nvws->channel->device->chipset; int ret, i; if (!screen) return NULL; - screen->chipset = chipset; screen->nvws = nvws; /* 3D object */ diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h index 56f8776a173..816ece94c4a 100644 --- a/src/gallium/drivers/nv30/nv30_screen.h +++ b/src/gallium/drivers/nv30/nv30_screen.h @@ -7,7 +7,6 @@ struct nv30_screen { struct pipe_screen pipe; struct nouveau_winsys *nvws; - unsigned chipset; /* HW graphics objects */ struct nouveau_grobj *rankine; diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index f9c93f7a2d7..d9d9accea8d 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -38,7 +38,6 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) struct nv40_screen *screen = nv40_screen(pscreen); struct pipe_winsys *ws = pscreen->winsys; struct nv40_context *nv40; - unsigned chipset = screen->chipset; struct nouveau_winsys *nvws = screen->nvws; nv40 = CALLOC(1, sizeof(struct nv40_context)); @@ -47,7 +46,6 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) nv40->screen = screen; nv40->pctx_id = pctx_id; - nv40->chipset = chipset; nv40->nvws = nvws; nv40->pipe.winsys = ws; diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 24e8cd23379..77b3da0ab91 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -111,8 +111,6 @@ struct nv40_context { struct draw_context *draw; - int chipset; - /* HW state derived from pipe states */ struct nv40_state state; struct { diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 7f68539a85c..1bf0931e4d9 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -12,9 +12,10 @@ static const char * nv40_screen_get_name(struct pipe_screen *pscreen) { struct nv40_screen *screen = nv40_screen(pscreen); + struct nouveau_device *dev = screen->nvws->channel->device; static char buffer[128]; - snprintf(buffer, sizeof(buffer), "NV%02X", screen->chipset); + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); return buffer; } @@ -149,17 +150,16 @@ nv40_screen_destroy(struct pipe_screen *pscreen) } struct pipe_screen * -nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) +nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) { struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen); struct nouveau_stateobj *so; unsigned curie_class; + unsigned chipset = nvws->channel->device; int ret; if (!screen) return NULL; - screen->chipset = chipset; screen->nvws = nvws; /* 3D object */ diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h index 3ea78aadfd9..c04a1275a00 100644 --- a/src/gallium/drivers/nv40/nv40_screen.h +++ b/src/gallium/drivers/nv40/nv40_screen.h @@ -7,7 +7,6 @@ struct nv40_screen { struct pipe_screen pipe; struct nouveau_winsys *nvws; - unsigned chipset; unsigned cur_pctx; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index e4ca72c7171..d069639dc48 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -45,9 +45,10 @@ static const char * nv50_screen_get_name(struct pipe_screen *pscreen) { struct nv50_screen *screen = nv50_screen(pscreen); + struct nouveau_device *dev = screen->nvws->channel->device; static char buffer[128]; - snprintf(buffer, sizeof(buffer), "NV%02X", screen->chipset); + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); return buffer; } @@ -123,17 +124,16 @@ nv50_screen_destroy(struct pipe_screen *pscreen) } struct pipe_screen * -nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws, - unsigned chipset) +nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) { struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen); struct nouveau_stateobj *so; unsigned tesla_class = 0, ret; + unsigned chipset = nvws->channel->device->chipset; int i; if (!screen) return NULL; - screen->chipset = chipset; screen->nvws = nvws; /* 3D object */ diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index 6e4120d6693..5a2732e37f4 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -7,7 +7,6 @@ struct nv50_screen { struct pipe_screen pipe; struct nouveau_winsys *nvws; - unsigned chipset; unsigned cur_pctx; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c index 635fd478a94..5eabbc88939 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c @@ -91,8 +91,7 @@ nouveau_pipe_create(struct nouveau_context *nv) struct nouveau_channel_context *nvc = nv->nvc; struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); struct pipe_screen *(*hws_create)(struct pipe_winsys *, - struct nouveau_winsys *, - unsigned chipset); + struct nouveau_winsys *); struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned); struct pipe_winsys *ws; unsigned chipset = nv->nv_screen->device->chipset; @@ -152,7 +151,7 @@ nouveau_pipe_create(struct nouveau_context *nv) ws = nouveau_create_pipe_winsys(nv); if (!nvc->pscreen) - nvc->pscreen = hws_create(ws, nvws, chipset); + nvc->pscreen = hws_create(ws, nvws); nvc->pctx[nv->pctx_id] = hw_create(nvc->pscreen, nv->pctx_id); return nvc->pctx[nv->pctx_id]; } -- cgit v1.2.3 From 666ac923f016b1b231c5a8847cbe084321f697ca Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 14 May 2008 00:25:09 +1000 Subject: nouveau: remove winsys-related stuff from nouveau_local.h --- src/gallium/winsys/dri/nouveau/nouveau_local.h | 21 +++++++-------- src/gallium/winsys/dri/nouveau/nv04_surface.c | 36 ++++++++++++++------------ src/gallium/winsys/dri/nouveau/nv50_surface.c | 21 ++++++++------- 3 files changed, 42 insertions(+), 36 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_local.h b/src/gallium/winsys/dri/nouveau/nouveau_local.h index 8a52f26714f..e878a408037 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_local.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_local.h @@ -74,45 +74,44 @@ BIND_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, unsigned subc) } static INLINE void -OUT_RELOC(struct nouveau_channel *chan, struct pipe_buffer *buf, +OUT_RELOC(struct nouveau_channel *chan, struct nouveau_bo *bo, unsigned data, unsigned flags, unsigned vor, unsigned tor) { - nouveau_pushbuf_emit_reloc(chan, chan->pushbuf->cur++, - nouveau_buffer(buf)->bo, + nouveau_pushbuf_emit_reloc(chan, chan->pushbuf->cur++, bo, data, flags, vor, tor); } /* Raw data + flags depending on FB/TT buffer */ static INLINE void -OUT_RELOCd(struct nouveau_channel *chan, struct pipe_buffer *buf, +OUT_RELOCd(struct nouveau_channel *chan, struct nouveau_bo *bo, unsigned data, unsigned flags, unsigned vor, unsigned tor) { - OUT_RELOC(chan, buf, data, flags | NOUVEAU_BO_OR, vor, tor); + OUT_RELOC(chan, bo, data, flags | NOUVEAU_BO_OR, vor, tor); } /* FB/TT object handle */ static INLINE void -OUT_RELOCo(struct nouveau_channel *chan, struct pipe_buffer *buf, +OUT_RELOCo(struct nouveau_channel *chan, struct nouveau_bo *bo, unsigned flags) { - OUT_RELOC(chan, buf, 0, flags | NOUVEAU_BO_OR, + OUT_RELOC(chan, bo, 0, flags | NOUVEAU_BO_OR, chan->vram->handle, chan->gart->handle); } /* Low 32-bits of offset */ static INLINE void -OUT_RELOCl(struct nouveau_channel *chan, struct pipe_buffer *buf, +OUT_RELOCl(struct nouveau_channel *chan, struct nouveau_bo *bo, unsigned delta, unsigned flags) { - OUT_RELOC(chan, buf, delta, flags | NOUVEAU_BO_LOW, 0, 0); + OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_LOW, 0, 0); } /* High 32-bits of offset */ static INLINE void -OUT_RELOCh(struct nouveau_channel *chan, struct pipe_buffer *buf, +OUT_RELOCh(struct nouveau_channel *chan, struct nouveau_bo *bo, unsigned delta, unsigned flags) { - OUT_RELOC(chan, buf, delta, flags | NOUVEAU_BO_HIGH, 0, 0); + OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_HIGH, 0, 0); } #endif diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c index dc167e4acfb..83c790db17a 100644 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -43,10 +43,10 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, BEGIN_RING(chan, nv->nvc->NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); - OUT_RELOCl(chan, src->buffer, src_offset, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCl(chan, dst->buffer, dst_offset, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_WR); + OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src_offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst_offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); OUT_RING (chan, src->pitch * src->cpp); OUT_RING (chan, dst->pitch * dst->cpp); OUT_RING (chan, w * src->cpp); @@ -88,10 +88,10 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, if ((src->offset & 63) || (dst->offset & 63)) { BEGIN_RING(nv->nvc->channel, nv->nvc->NvM2MF, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(chan, src->buffer, NOUVEAU_BO_GART | - NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_GART | - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); nv->surface_copy = nv04_surface_copy_m2mf; nv->surf_dst = dst; @@ -108,17 +108,19 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(chan, src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (chan, format); OUT_RING (chan, ((dst->pitch * dst->cpp) << 16) | (src->pitch * src->cpp)); - OUT_RELOCl(chan, src->buffer, src->offset, + OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(chan, dst->buffer, dst->offset, + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); return 0; @@ -151,15 +153,17 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, } BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (chan, cs2d_format); OUT_RING (chan, ((dst->pitch * dst->cpp) << 16) | (dst->pitch * dst->cpp)); - OUT_RELOCl(chan, dst->buffer, dst->offset, + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, dst->buffer, dst->offset, + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c index 5c6733e44a3..d6da116b774 100644 --- a/src/gallium/winsys/dri/nouveau/nv50_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv50_surface.c @@ -29,8 +29,10 @@ nv50_surface_copy_prep(struct nouveau_context *nv, assert(surf_format >= 0); BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RELOCo(chan, src->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 2); OUT_RING (chan, surf_format); @@ -39,9 +41,9 @@ nv50_surface_copy_prep(struct nouveau_context *nv, OUT_RING (chan, dst->pitch * dst->cpp); OUT_RING (chan, dst->pitch); OUT_RING (chan, dst->height); - OUT_RELOCh(chan, dst->buffer, dst->offset, + OUT_RELOCh(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, dst->buffer, dst->offset, + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); OUT_RING (chan, 0); @@ -56,9 +58,9 @@ nv50_surface_copy_prep(struct nouveau_context *nv, OUT_RING (chan, src->pitch * src->cpp); OUT_RING (chan, src->pitch); OUT_RING (chan, src->height); - OUT_RELOCh(chan, src->buffer, src->offset, + OUT_RELOCh(chan, nouveau_buffer(src->buffer)->bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(chan, src->buffer, src->offset, + OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); return 0; @@ -112,7 +114,8 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, return 1; BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY1, 1); - OUT_RELOCo(chan, dst->buffer, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 2); OUT_RING (chan, surf_format); OUT_RING (chan, 1); @@ -120,9 +123,9 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, OUT_RING (chan, dst->pitch * dst->cpp); OUT_RING (chan, dst->pitch); OUT_RING (chan, dst->height); - OUT_RELOCh(chan, dst->buffer, dst->offset, + OUT_RELOCh(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, dst->buffer, dst->offset, + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); OUT_RING (chan, 0); -- cgit v1.2.3 From 9a01ee4424718e0c3015c1f0477cae63ee63d96b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 14 May 2008 18:35:37 +1000 Subject: nv40: fix typo in one of the previous commits --- src/gallium/drivers/nv40/nv40_screen.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 1bf0931e4d9..0336b84e5f2 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -155,7 +155,7 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen); struct nouveau_stateobj *so; unsigned curie_class; - unsigned chipset = nvws->channel->device; + unsigned chipset = nvws->channel->device->chipset; int ret; if (!screen) -- cgit v1.2.3 From f5599a7a3c4a6335ce79fdbd82e18f08bb0ac8e7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 May 2008 14:25:15 +1000 Subject: nouveau: remove final PIPE_FORMAT_U_* usage --- src/gallium/drivers/nv10/nv10_fragtex.c | 6 +++--- src/gallium/drivers/nv10/nv10_screen.c | 6 +++--- src/gallium/drivers/nv30/nv30_fragtex.c | 8 ++++---- src/gallium/drivers/nv30/nv30_screen.c | 8 ++++---- src/gallium/drivers/nv40/nv40_fragtex.c | 8 ++++---- src/gallium/drivers/nv40/nv40_screen.c | 8 ++++---- 6 files changed, 22 insertions(+), 22 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_fragtex.c b/src/gallium/drivers/nv10/nv10_fragtex.c index 67e0b4bd45d..238634d0bb4 100644 --- a/src/gallium/drivers/nv10/nv10_fragtex.c +++ b/src/gallium/drivers/nv10/nv10_fragtex.c @@ -44,9 +44,9 @@ nv10_texture_formats[] = { _(A8R8G8B8_UNORM, A8R8G8B8), _(A1R5G5B5_UNORM, A1R5G5B5), _(A4R4G4B4_UNORM, A4R4G4B4), - _(U_L8 , L8 ), - _(U_A8 , A8 ), - _(U_A8_L8 , A8L8 ), + _(L8_UNORM , L8 ), + _(A8_UNORM , A8 ), + _(A8L8_UNORM , A8L8 ), // _(RGB_DXT1 , DXT1, ), // _(RGBA_DXT1 , DXT1, ), // _(RGBA_DXT3 , DXT3, ), diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index fad96058b8e..5fe3a030811 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -102,9 +102,9 @@ nv10_screen_is_format_supported(struct pipe_screen *screen, case PIPE_FORMAT_A1R5G5B5_UNORM: case PIPE_FORMAT_A4R4G4B4_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + case PIPE_FORMAT_I8_UNORM: return TRUE; default: break; diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index 45ee6db8d6a..47f3ff047db 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -50,10 +50,10 @@ nv30_texture_formats[] = { _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), // _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), - _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), + _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X), + _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X), + _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y), // _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), // _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), // _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 9b0ac55ff88..bb77776ff15 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -106,10 +106,10 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_A1R5G5B5_UNORM: case PIPE_FORMAT_A4R4G4B4_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + case PIPE_FORMAT_I8_UNORM: + case PIPE_FORMAT_A8L8_UNORM: case PIPE_FORMAT_Z16_UNORM: case PIPE_FORMAT_Z24S8_UNORM: return TRUE; diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 980705f94ad..c79ea8becb3 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -24,10 +24,10 @@ nv40_texture_formats[] = { _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(U_L8 , L8 , S1, S1, S1, ONE, X, X, X, X), - _(U_A8 , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(U_I8 , L8 , S1, S1, S1, S1, X, X, X, X), - _(U_A8_L8 , A8L8 , S1, S1, S1, S1, X, X, X, Y), + _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X), + _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), + _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X), + _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y), _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W), diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 0336b84e5f2..51640533938 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -111,10 +111,10 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_A1R5G5B5_UNORM: case PIPE_FORMAT_A4R4G4B4_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_U_L8: - case PIPE_FORMAT_U_A8: - case PIPE_FORMAT_U_I8: - case PIPE_FORMAT_U_A8_L8: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + case PIPE_FORMAT_I8_UNORM: + case PIPE_FORMAT_A8L8_UNORM: case PIPE_FORMAT_Z16_UNORM: case PIPE_FORMAT_Z24S8_UNORM: case PIPE_FORMAT_DXT1_RGB: -- cgit v1.2.3 From 8f67f98959261d193cb5f3db274b55fb24e2bb1e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 25 May 2008 14:26:18 +1000 Subject: nouveau: missed PCI case --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index 792eaaa79e5..aca16a8b127 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -310,7 +310,8 @@ nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) /* Check current memtype vs requested, if they match do nothing */ if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) return 0; - if ((nvbo->drm.flags & NOUVEAU_MEM_AGP) && (flags & NOUVEAU_BO_GART)) + if ((nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) && + (flags & NOUVEAU_BO_GART)) return 0; if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) return 0; -- cgit v1.2.3 From fa4b2439d4f240a5e573d4ea198b829791d614f4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 27 May 2008 01:22:22 +1000 Subject: nouveau: very quick port to tex-surface changes. probably the last match-gallium-upstream merge for a bit, some cleanup+nv50 work coming RSN... --- src/gallium/drivers/nv10/nv10_context.c | 1 - src/gallium/drivers/nv10/nv10_context.h | 1 - src/gallium/drivers/nv10/nv10_miptree.c | 11 +++++++---- src/gallium/drivers/nv10/nv10_screen.c | 25 +++++++++++++++++++++++++ src/gallium/drivers/nv30/nv30_context.c | 1 - src/gallium/drivers/nv30/nv30_context.h | 1 - src/gallium/drivers/nv30/nv30_miptree.c | 20 ++++++++------------ src/gallium/drivers/nv30/nv30_screen.c | 25 +++++++++++++++++++++++++ src/gallium/drivers/nv40/nv40_context.c | 1 - src/gallium/drivers/nv40/nv40_context.h | 1 - src/gallium/drivers/nv40/nv40_miptree.c | 20 ++++++++------------ src/gallium/drivers/nv40/nv40_screen.c | 25 +++++++++++++++++++++++++ src/gallium/drivers/nv50/nv50_context.c | 1 - src/gallium/drivers/nv50/nv50_context.h | 1 - src/gallium/drivers/nv50/nv50_miptree.c | 25 +++++++++++-------------- src/gallium/drivers/nv50/nv50_screen.c | 25 +++++++++++++++++++++++++ 16 files changed, 134 insertions(+), 50 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index 79253f8a2b3..9fcd0b0fc3a 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -285,7 +285,6 @@ nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) nv10_init_surface_functions(nv10); nv10_init_state_functions(nv10); - nv10_init_miptree_functions(nv10); nv10->draw = draw_create(); assert(nv10->draw); diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 433d04dc2a8..5636dfc9d22 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -109,7 +109,6 @@ nv10_context(struct pipe_context *pipe) extern void nv10_init_state_functions(struct nv10_context *nv10); extern void nv10_init_surface_functions(struct nv10_context *nv10); -extern void nv10_init_miptree_functions(struct nv10_context *nv10); extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen); diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index 4dfc675a6b9..1b9947354d6 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -52,7 +52,7 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) } static struct pipe_texture * -nv10_miptree_create(struct pipe_screen *screen, struct pipe_texture *pt) +nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) { struct pipe_winsys *ws = screen->winsys; struct nv10_miptree *mt; @@ -105,7 +105,8 @@ nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, static struct pipe_surface * nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) + unsigned face, unsigned level, unsigned zslice, + unsigned flags) { struct pipe_winsys *ws = screen->winsys; struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt; @@ -130,9 +131,10 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, return ps; } -void nv10_init_miptree_functions(struct nv10_context *nv10) +static void +nv10_miptree_surface_release(struct pipe_screen *screen, + struct pipe_surface **surface) { - nv10->pipe.texture_update = nv10_miptree_update; } void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) @@ -140,5 +142,6 @@ void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) pscreen->texture_create = nv10_miptree_create; pscreen->texture_release = nv10_miptree_release; pscreen->get_tex_surface = nv10_miptree_surface_get; + pscreen->tex_surface_release = nv10_miptree_surface_release; } diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 5fe3a030811..67787d8e9ce 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -117,6 +117,28 @@ nv10_screen_is_format_supported(struct pipe_screen *screen, return FALSE; } +static void * +nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + static void nv10_screen_destroy(struct pipe_screen *pscreen) { @@ -180,6 +202,9 @@ nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) screen->pipe.is_format_supported = nv10_screen_is_format_supported; + screen->pipe.surface_map = nv10_surface_map; + screen->pipe.surface_unmap = nv10_surface_unmap; + nv10_screen_init_miptree_functions(&screen->pipe); return &screen->pipe; diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 7a2fee78751..b2d9d3f1814 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -66,7 +66,6 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30_init_query_functions(nv30); nv30_init_surface_functions(nv30); nv30_init_state_functions(nv30); - nv30_init_miptree_functions(nv30); /* Create, configure, and install fallback swtnl path */ nv30->draw = draw_create(); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index f49450f82d9..333bd4875c6 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -134,7 +134,6 @@ nv30_context(struct pipe_context *pipe) extern void nv30_init_state_functions(struct nv30_context *nv30); extern void nv30_init_surface_functions(struct nv30_context *nv30); -extern void nv30_init_miptree_functions(struct nv30_context *nv30); extern void nv30_init_query_functions(struct nv30_context *nv30); extern void nv30_screen_init_miptree_functions(struct pipe_screen *pscreen); diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 10ab46e19a5..6078b1865e2 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -100,15 +100,10 @@ nv30_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) } } -static void -nv30_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, - uint face, uint levels) -{ -} - static struct pipe_surface * -nv30_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) +nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice, + unsigned flags) { struct pipe_winsys *ws = pscreen->winsys; struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; @@ -136,10 +131,10 @@ nv30_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, return ps; } -void -nv30_init_miptree_functions(struct nv30_context *nv30) +static void +nv30_miptree_surface_del(struct pipe_screen *pscreen, + struct pipe_surface **psurface) { - nv30->pipe.texture_update = nv30_miptree_update; } void @@ -147,6 +142,7 @@ nv30_screen_init_miptree_functions(struct pipe_screen *pscreen) { pscreen->texture_create = nv30_miptree_create; pscreen->texture_release = nv30_miptree_release; - pscreen->get_tex_surface = nv30_miptree_surface; + pscreen->get_tex_surface = nv30_miptree_surface_new; + pscreen->tex_surface_release = nv30_miptree_surface_del; } diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index bb77776ff15..9c576369891 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -124,6 +124,28 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen, return FALSE; } +static void * +nv30_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv30_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + static void nv30_screen_destroy(struct pipe_screen *pscreen) { @@ -300,6 +322,9 @@ nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) screen->pipe.is_format_supported = nv30_screen_surface_format_supported; + screen->pipe.surface_map = nv30_surface_map; + screen->pipe.surface_unmap = nv30_surface_unmap; + nv30_screen_init_miptree_functions(&screen->pipe); return &screen->pipe; diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index d9d9accea8d..a40f14895f4 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -59,7 +59,6 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) nv40_init_query_functions(nv40); nv40_init_surface_functions(nv40); nv40_init_state_functions(nv40); - nv40_init_miptree_functions(nv40); /* Create, configure, and install fallback swtnl path */ nv40->draw = draw_create(); diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 77b3da0ab91..d8d1891dff0 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -172,7 +172,6 @@ struct nv40_state_entry { extern void nv40_init_state_functions(struct nv40_context *nv40); extern void nv40_init_surface_functions(struct nv40_context *nv40); -extern void nv40_init_miptree_functions(struct nv40_context *nv40); extern void nv40_init_query_functions(struct nv40_context *nv40); extern void nv40_screen_init_miptree_functions(struct pipe_screen *pscreen); diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 1b192172232..23da6e36a3e 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -100,15 +100,10 @@ nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) } } -static void -nv40_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, - uint face, uint levels) -{ -} - static struct pipe_surface * -nv40_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) +nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice, + unsigned flags) { struct pipe_winsys *ws = pscreen->winsys; struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; @@ -136,10 +131,10 @@ nv40_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, return ps; } -void -nv40_init_miptree_functions(struct nv40_context *nv40) +static void +nv40_miptree_surface_del(struct pipe_screen *pscreen, + struct pipe_surface **psurface) { - nv40->pipe.texture_update = nv40_miptree_update; } void @@ -147,6 +142,7 @@ nv40_screen_init_miptree_functions(struct pipe_screen *pscreen) { pscreen->texture_create = nv40_miptree_create; pscreen->texture_release = nv40_miptree_release; - pscreen->get_tex_surface = nv40_miptree_surface; + pscreen->get_tex_surface = nv40_miptree_surface_new; + pscreen->tex_surface_release = nv40_miptree_surface_del; } diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 51640533938..ed0215b4862 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -133,6 +133,28 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, return FALSE; } +static void * +nv40_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv40_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + static void nv40_screen_destroy(struct pipe_screen *pscreen) { @@ -282,6 +304,9 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) screen->pipe.is_format_supported = nv40_screen_surface_format_supported; + screen->pipe.surface_map = nv40_surface_map; + screen->pipe.surface_unmap = nv40_surface_unmap; + nv40_screen_init_miptree_functions(&screen->pipe); return &screen->pipe; diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 6eb1878b847..a225c4bf728 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -55,7 +55,6 @@ nv50_create(struct pipe_screen *pscreen, unsigned pctx_id) nv50->pipe.flush = nv50_flush; - nv50_init_miptree_functions(nv50); nv50_init_surface_functions(nv50); nv50_init_state_functions(nv50); nv50_init_query_functions(nv50); diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index f532fa6bfb3..e68c702deae 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -71,7 +71,6 @@ nv50_context(struct pipe_context *pipe) return (struct nv50_context *)pipe; } -extern void nv50_init_miptree_functions(struct nv50_context *nv50); extern void nv50_init_surface_functions(struct nv50_context *nv50); extern void nv50_init_state_functions(struct nv50_context *nv50); extern void nv50_init_query_functions(struct nv50_context *nv50); diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 58584934b18..ccb916d6acc 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -56,8 +56,9 @@ nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt) } static struct pipe_surface * -nv50_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice) +nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice, + unsigned flags) { struct pipe_winsys *ws = pscreen->winsys; struct nv50_miptree *mt = nv50_miptree(pt); @@ -80,22 +81,18 @@ nv50_miptree_surface(struct pipe_screen *pscreen, struct pipe_texture *pt, return ps; } -void -nv50_screen_init_miptree_functions(struct pipe_screen *pscreen) -{ - pscreen->texture_create = nv50_miptree_create; - pscreen->texture_release = nv50_miptree_release; - pscreen->get_tex_surface = nv50_miptree_surface; -} - static void -nv50_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, - uint face, uint levels) +nv50_miptree_surface_del(struct pipe_screen *pscreen, + struct pipe_surface **psurface) { } void -nv50_init_miptree_functions(struct nv50_context *nv50) +nv50_screen_init_miptree_functions(struct pipe_screen *pscreen) { - nv50->pipe.texture_update = nv50_miptree_update; + pscreen->texture_create = nv50_miptree_create; + pscreen->texture_release = nv50_miptree_release; + pscreen->get_tex_surface = nv50_miptree_surface_new; + pscreen->tex_surface_release = nv50_miptree_surface_del; } + diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index d069639dc48..29c057a1453 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -117,6 +117,28 @@ nv50_screen_get_paramf(struct pipe_screen *pscreen, int param) } } +static void * +nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + static void nv50_screen_destroy(struct pipe_screen *pscreen) { @@ -209,6 +231,9 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) screen->pipe.is_format_supported = nv50_screen_is_format_supported; + screen->pipe.surface_map = nv50_surface_map; + screen->pipe.surface_unmap = nv50_surface_unmap; + nv50_screen_init_miptree_functions(&screen->pipe); return &screen->pipe; -- cgit v1.2.3 From d0c0c0d1e67f92df8866e6218b868b3de954a5e1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 30 May 2008 00:28:29 +1000 Subject: nouveau: oops, more tex-surfaces fallout --- src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index 755c8b6997a..2fab1eab652 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -42,7 +42,8 @@ nouveau_surface_alloc(struct pipe_winsys *ws) static int nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, unsigned width, unsigned height, - enum pipe_format format, unsigned flags) + enum pipe_format format, unsigned flags, + unsigned tex_usage) { unsigned pitch = ((width * pf_get_size(format)) + 63) & ~63; -- cgit v1.2.3 From 524408f1a5b39c8c25a277e41e4eee54fd726b84 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 30 May 2008 00:38:07 +1000 Subject: nouveau: rework buffer validation a bit --- src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 20 +++-- src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 109 ++++++++++++----------- 2 files changed, 69 insertions(+), 60 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h index a31c9a514b1..5829f649cd5 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h @@ -107,7 +107,7 @@ extern void nouveau_fence_flush(struct nouveau_channel *); struct nouveau_pushbuf_reloc { - uint64_t next; + struct nouveau_pushbuf_bo *pbbo; uint32_t *ptr; uint32_t flags; uint32_t data; @@ -116,13 +116,14 @@ struct nouveau_pushbuf_reloc { }; struct nouveau_pushbuf_bo { - uint64_t next; - uint64_t handle; - uint64_t flags; - uint64_t relocs; - int nr_relocs; + struct nouveau_channel *channel; + struct nouveau_bo *bo; + unsigned flags; + unsigned handled; }; +#define NOUVEAU_PUSHBUF_MAX_BUFFERS 1024 +#define NOUVEAU_PUSHBUF_MAX_RELOCS 1024 struct nouveau_pushbuf_priv { struct nouveau_pushbuf base; @@ -132,8 +133,10 @@ struct nouveau_pushbuf_priv { unsigned start; unsigned size; - uint64_t buffers; - int nr_buffers; + struct nouveau_pushbuf_bo *buffers; + unsigned nr_buffers; + struct nouveau_pushbuf_reloc *relocs; + unsigned nr_relocs; }; #define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) @@ -242,6 +245,7 @@ nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, struct nouveau_bo_priv { struct nouveau_bo base; + struct nouveau_pushbuf_bo *pending; struct nouveau_fence *fence; struct nouveau_fence *wr_fence; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c index 78919bdee8c..815046ba85f 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c @@ -97,6 +97,10 @@ nouveau_pushbuf_init(struct nouveau_channel *chan) nouveau_pushbuf_space(chan, 0); chan->pushbuf = &nvchan->pb.base; + nvchan->pb.buffers = calloc(NOUVEAU_PUSHBUF_MAX_BUFFERS, + sizeof(struct nouveau_pushbuf_bo)); + nvchan->pb.relocs = calloc(NOUVEAU_PUSHBUF_MAX_RELOCS, + sizeof(struct nouveau_pushbuf_reloc)); return 0; } @@ -131,8 +135,7 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) { struct nouveau_channel_priv *nvchan = nouveau_channel(chan); struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; - struct nouveau_pushbuf_bo *pbbo; - int ret; + int ret, i; if (nvpb->base.remaining == nvpb->size) return 0; @@ -150,38 +153,47 @@ nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) /* Validate buffers + apply relocations */ nvchan->user_charge = 0; - while ((pbbo = ptr_to_pbbo(nvpb->buffers))) { - struct nouveau_pushbuf_reloc *r; - struct nouveau_bo *bo = &ptr_to_bo(pbbo->handle)->base; - - ret = nouveau_bo_validate(chan, bo, pbbo->flags); - assert (ret == 0); - - if (bo->offset == nouveau_bo(bo)->offset && - bo->flags == nouveau_bo(bo)->flags) { + for (i = 0; i < nvpb->nr_relocs; i++) { + struct nouveau_pushbuf_reloc *r = &nvpb->relocs[i]; + struct nouveau_pushbuf_bo *pbbo = r->pbbo; + struct nouveau_bo *bo = pbbo->bo; + + /* Validated, mem matches presumed, no relocation necessary */ + if (pbbo->handled & 2) { + if (!(pbbo->handled & 1)) + assert(0); + continue; + } - while ((r = ptr_to_pbrel(pbbo->relocs))) { - pbbo->relocs = r->next; - free(r); + /* Not yet validated, do it now */ + if (!(pbbo->handled & 1)) { + ret = nouveau_bo_validate(chan, bo, pbbo->flags); + if (ret) { + assert(0); + return ret; } + pbbo->handled |= 1; - nouveau_bo_del(&bo); - nvpb->buffers = pbbo->next; - free(pbbo); - continue; + if (bo->offset == nouveau_bo(bo)->offset && + bo->flags == nouveau_bo(bo)->flags) { + pbbo->handled |= 2; + continue; + } + bo->offset = nouveau_bo(bo)->offset; + bo->flags = nouveau_bo(bo)->flags; } - bo->offset = nouveau_bo(bo)->offset; - bo->flags = nouveau_bo(bo)->flags; - while ((r = ptr_to_pbrel(pbbo->relocs))) { - *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); - pbbo->relocs = r->next; - free(r); - } + /* Apply the relocation */ + *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); + } + nvpb->nr_relocs = 0; + + /* Dereference all buffers on validate list */ + for (i = 0; i < nvpb->nr_buffers; i++) { + struct nouveau_pushbuf_bo *pbbo = &nvpb->buffers[i]; - nouveau_bo_del(&bo); - nvpb->buffers = pbbo->next; - free(pbbo); + nouveau_bo(pbbo->bo)->pending = NULL; + nouveau_bo_del(&pbbo->bo); } nvpb->nr_buffers = 0; @@ -206,25 +218,21 @@ static struct nouveau_pushbuf_bo * nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) { struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); - struct nouveau_pushbuf_bo *pbbo = ptr_to_pbbo(nvpb->buffers); - struct nouveau_bo *ref = NULL; + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_pushbuf_bo *pbbo; - while (pbbo) { - if (pbbo->handle == bo->handle) - return pbbo; - pbbo = ptr_to_pbbo(pbbo->next); - } + if (nvbo->pending) + return nvbo->pending; - pbbo = malloc(sizeof(struct nouveau_pushbuf_bo)); - pbbo->next = nvpb->buffers; - nvpb->buffers = pbbo_to_ptr(pbbo); - nvpb->nr_buffers++; + if (nvpb->nr_buffers >= NOUVEAU_PUSHBUF_MAX_BUFFERS) + return NULL; + pbbo = nvpb->buffers + nvpb->nr_buffers++; + nvbo->pending = pbbo; - nouveau_bo_ref(bo->device, bo->handle, &ref); - pbbo->handle = bo_to_ptr(ref); + nouveau_bo_ref(bo->device, bo->handle, &pbbo->bo); + pbbo->channel = chan; pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; - pbbo->relocs = 0; - pbbo->nr_relocs = 0; + pbbo->handled = 0; return pbbo; } @@ -233,24 +241,21 @@ nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, struct nouveau_bo *bo, uint32_t data, uint32_t flags, uint32_t vor, uint32_t tor) { + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); struct nouveau_pushbuf_bo *pbbo; struct nouveau_pushbuf_reloc *r; - if (!chan) - return -EINVAL; + if (nvpb->nr_relocs >= NOUVEAU_PUSHBUF_MAX_RELOCS) + return -ENOMEM; pbbo = nouveau_pushbuf_emit_buffer(chan, bo); if (!pbbo) - return -EFAULT; - - r = malloc(sizeof(struct nouveau_pushbuf_reloc)); - r->next = pbbo->relocs; - pbbo->relocs = pbrel_to_ptr(r); - pbbo->nr_relocs++; - + return -ENOMEM; pbbo->flags |= (flags & NOUVEAU_BO_RDWR); pbbo->flags &= (flags | NOUVEAU_BO_RDWR); + r = nvpb->relocs + nvpb->nr_relocs++; + r->pbbo = pbbo; r->ptr = ptr; r->flags = flags; r->data = data; -- cgit v1.2.3 From 8b31d5fc8a5425b01adf80f4873cb816925ee0d1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 30 May 2008 11:00:59 +1000 Subject: nouveau: flush pending pushbuf if buffer on validate list at map/del --- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index aca16a8b127..57f5b7d1ae1 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -256,6 +256,9 @@ nouveau_bo_del(struct nouveau_bo **bo) if (--nvbo->refcount) return; + if (nvbo->pending) + nouveau_pushbuf_flush(nvbo->pending->channel, 0); + if (nvbo->fence) nouveau_fence_signal_cb(nvbo->fence, nouveau_bo_del_cb, nvbo); else @@ -270,6 +273,11 @@ nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) if (!nvbo) return -EINVAL; + if (nvbo->pending && + (nvbo->pending->flags & NOUVEAU_BO_WR || flags & NOUVEAU_BO_WR)) { + nouveau_pushbuf_flush(nvbo->pending->channel, 0); + } + if (flags & NOUVEAU_BO_WR) nouveau_fence_wait(&nvbo->fence); else -- cgit v1.2.3 From 996b549fdbfe772ee56a51858e81e93bccaae5c5 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 30 May 2008 11:04:55 +1000 Subject: nv40: a couple of memory leaks --- src/gallium/drivers/nv40/nv40_state_viewport.c | 3 ++- src/gallium/drivers/nv40/nv40_vertprog.c | 6 ++++++ 2 files changed, 8 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c index 1b6248e5b8e..869a55b4053 100644 --- a/src/gallium/drivers/nv40/nv40_state_viewport.c +++ b/src/gallium/drivers/nv40/nv40_state_viewport.c @@ -3,8 +3,8 @@ static boolean nv40_state_viewport_validate(struct nv40_context *nv40) { - struct nouveau_stateobj *so = so_new(11, 0); struct pipe_viewport_state *vpt = &nv40->viewport; + struct nouveau_stateobj *so; unsigned bypass; if (nv40->render_mode == HW && !nv40->rasterizer->pipe.bypass_clipping) @@ -18,6 +18,7 @@ nv40_state_viewport_validate(struct nv40_context *nv40) return FALSE; nv40->state.viewport_bypass = bypass; + so = so_new(11, 0); if (!bypass) { so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_TRANSLATE_X, 8); diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index e10250528e2..eb14869bfe0 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -843,6 +843,12 @@ nv40_vertprog_translate(struct nv40_context *nv40, vp->translated = TRUE; out_err: tgsi_parse_free(&parse); + if (vpc->r_temp) + FREE(vpc->r_temp); + if (vpc->r_address) + FREE(vpc->r_address); + if (vpc->imm) + FREE(vpc->imm); FREE(vpc); } -- cgit v1.2.3 From cf13cf245e91669eb916a6d51676b0d66ff62ead Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 30 May 2008 14:32:59 +1000 Subject: nouveau: fix warning --- src/gallium/winsys/dri/nouveau/nouveau_context.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index d9fc3f6ce1f..ef5fb7ec0d3 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -99,7 +99,7 @@ nouveau_context_create(const __GLcontextModes *glVis, struct st_context *st_share = NULL; struct nouveau_channel_context *nvc = NULL; struct nouveau_device *dev = nv_screen->device; - int i, ret; + int i; if (sharedContextPrivate) { st_share = ((struct nouveau_context *)sharedContextPrivate)->st; -- cgit v1.2.3 From edea59e8e5a3ef4c6afdcb4f1d32961466be508b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 8 Jun 2008 12:50:47 +1000 Subject: nouveau: kill off surface_alloc_storage() --- .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 23 ---------------------- 1 file changed, 23 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index 2fab1eab652..ba5dd2eea4b 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -39,28 +39,6 @@ nouveau_surface_alloc(struct pipe_winsys *ws) return surf; } -static int -nouveau_surface_alloc_storage(struct pipe_winsys *ws, struct pipe_surface *surf, - unsigned width, unsigned height, - enum pipe_format format, unsigned flags, - unsigned tex_usage) -{ - unsigned pitch = ((width * pf_get_size(format)) + 63) & ~63; - - surf->format = format; - surf->width = width; - surf->height = height; - surf->cpp = pf_get_size(format); - surf->pitch = pitch / surf->cpp; - - surf->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - pitch * height); - if (!surf->buffer) - return 1; - - return 0; -} - static void nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) { @@ -226,7 +204,6 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws->flush_frontbuffer = nouveau_flush_frontbuffer; pws->surface_alloc = nouveau_surface_alloc; - pws->surface_alloc_storage = nouveau_surface_alloc_storage; pws->surface_release = nouveau_surface_release; pws->buffer_create = nouveau_pipe_bo_create; -- cgit v1.2.3 From c11a7ec821d41b91a3825c5abfb4687c98b5bf98 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 8 Jun 2008 03:04:14 -0400 Subject: Initial commit for g3dvl. Initial commit for g3dvl, contains support for basic XvMC features. - Context, surface, block, macroblock creation and deletion - Surface rendering - Frame pictures - Frame based motion compensation - Intra-coded macroblocks - Predicted macroblocks - Bi-directionally predicted macroblocks - Surface display - Color conversion - Scaling --- src/gallium/state_trackers/g3dvl/Makefile | 18 + src/gallium/state_trackers/g3dvl/tests/.gitignore | 6 + src/gallium/state_trackers/g3dvl/tests/Makefile | 42 + .../state_trackers/g3dvl/tests/test_b_rendering.c | 226 ++ .../state_trackers/g3dvl/tests/test_context.c | 22 + .../state_trackers/g3dvl/tests/test_i_rendering.c | 137 ++ .../state_trackers/g3dvl/tests/test_p_rendering.c | 214 ++ .../state_trackers/g3dvl/tests/test_surface.c | 26 + src/gallium/state_trackers/g3dvl/vl_context.c | 2293 ++++++++++++++++++++ src/gallium/state_trackers/g3dvl/vl_context.h | 73 + src/gallium/state_trackers/g3dvl/vl_data.c | 188 ++ src/gallium/state_trackers/g3dvl/vl_data.h | 25 + src/gallium/state_trackers/g3dvl/vl_defs.h | 12 + src/gallium/state_trackers/g3dvl/vl_surface.c | 539 +++++ src/gallium/state_trackers/g3dvl/vl_surface.h | 81 + src/gallium/state_trackers/g3dvl/vl_types.h | 88 + src/gallium/winsys/g3dvl/xsp_winsys.c | 256 +++ src/gallium/winsys/g3dvl/xsp_winsys.h | 11 + src/libXvMC/Makefile | 33 + src/libXvMC/attributes.c | 20 + src/libXvMC/block.c | 85 + src/libXvMC/context.c | 154 ++ src/libXvMC/subpicture.c | 110 + src/libXvMC/surface.c | 373 ++++ src/libXvMC/tests/.gitignore | 5 + src/libXvMC/tests/Makefile | 25 + src/libXvMC/tests/test_blocks.c | 95 + src/libXvMC/tests/test_context.c | 94 + src/libXvMC/tests/test_rendering.c | 153 ++ src/libXvMC/tests/test_surface.c | 72 + src/libXvMC/tests/testlib.c | 87 + src/libXvMC/tests/testlib.h | 38 + 32 files changed, 5601 insertions(+) create mode 100644 src/gallium/state_trackers/g3dvl/Makefile create mode 100644 src/gallium/state_trackers/g3dvl/tests/.gitignore create mode 100644 src/gallium/state_trackers/g3dvl/tests/Makefile create mode 100644 src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c create mode 100644 src/gallium/state_trackers/g3dvl/tests/test_context.c create mode 100644 src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c create mode 100644 src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c create mode 100644 src/gallium/state_trackers/g3dvl/tests/test_surface.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_context.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_context.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_data.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_data.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_defs.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_surface.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_surface.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_types.h create mode 100644 src/gallium/winsys/g3dvl/xsp_winsys.c create mode 100644 src/gallium/winsys/g3dvl/xsp_winsys.h create mode 100644 src/libXvMC/Makefile create mode 100644 src/libXvMC/attributes.c create mode 100644 src/libXvMC/block.c create mode 100644 src/libXvMC/context.c create mode 100644 src/libXvMC/subpicture.c create mode 100644 src/libXvMC/surface.c create mode 100644 src/libXvMC/tests/.gitignore create mode 100644 src/libXvMC/tests/Makefile create mode 100644 src/libXvMC/tests/test_blocks.c create mode 100644 src/libXvMC/tests/test_context.c create mode 100644 src/libXvMC/tests/test_rendering.c create mode 100644 src/libXvMC/tests/test_surface.c create mode 100644 src/libXvMC/tests/testlib.c create mode 100644 src/libXvMC/tests/testlib.h diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile new file mode 100644 index 00000000000..a0d85fbcc8f --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -0,0 +1,18 @@ +TARGET = libg3dvl.a +OBJECTS = vl_context.o vl_data.o vl_surface.o +GALLIUMDIR = ../.. + +CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary + +############################################# + +.PHONY = all clean + +all: ${TARGET} + +${TARGET}: ${OBJECTS} + ar rcs $@ $^ + +clean: + rm -rf ${OBJECTS} ${TARGET} + diff --git a/src/gallium/state_trackers/g3dvl/tests/.gitignore b/src/gallium/state_trackers/g3dvl/tests/.gitignore new file mode 100644 index 00000000000..939666da9ab --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/.gitignore @@ -0,0 +1,6 @@ +test_context +test_surface +test_i_rendering +test_p_rendering +test_b_rendering + diff --git a/src/gallium/state_trackers/g3dvl/tests/Makefile b/src/gallium/state_trackers/g3dvl/tests/Makefile new file mode 100644 index 00000000000..8f983593c3a --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/Makefile @@ -0,0 +1,42 @@ +GALLIUMDIR = ../../.. + +CFLAGS += -g -Wall -Werror \ + -I${GALLIUMDIR}/state_trackers/g3dvl \ + -I${GALLIUMDIR}/winsys/g3dvl \ + -I${GALLIUMDIR}/include \ + -I${GALLIUMDIR}/auxiliary \ + -I${GALLIUMDIR}/drivers +LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ + -L${GALLIUMDIR}/drivers/softpipe \ + -L${GALLIUMDIR}/auxiliary/tgsi \ + -L${GALLIUMDIR}/auxiliary/draw \ + -L${GALLIUMDIR}/auxiliary/util \ + -L${GALLIUMDIR}/auxiliary/translate \ + -L${GALLIUMDIR}/auxiliary/cso_cache \ + -L${GALLIUMDIR}/auxiliary/rtasm +LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lX11 -lm + +############################################# + +.PHONY = all clean + +all: test_context test_surface test_i_rendering test_p_rendering test_b_rendering + +test_context: test_context.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_surface: test_surface.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_i_rendering: test_i_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_p_rendering: test_p_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_b_rendering: test_b_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +clean: + rm -rf *.o test_context test_surface test_i_rendering test_p_rendering test_b_rendering + diff --git a/src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c new file mode 100644 index 00000000000..b78cc851ae4 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c @@ -0,0 +1,226 @@ +#include +#include +#include +#include +#include + +static const unsigned short ycbcr16x16_420[8*8*6] = +{ + 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, + + 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, + + 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E +}; + +static const signed short ycbcr16x16_420_2[8*8*6] = +{ + -0x00A5,-0x00A5,-0x00A5,-0x0072,-0x00A5,-0x0072,-0x0072,-0x0072, + -0x0072,-0x00A5,-0x0072,-0x0072,-0x00A5,-0x0072,-0x0072,-0x0072, + -0x0072,-0x00A5,-0x0072,-0x0072,-0x00A5,-0x00A5,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x00A5, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x00A5,-0x00A5, + -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x00A5,-0x00A5,-0x00A5, + -0x0072,-0x0072,-0x0072,-0x0072,-0x00A5,-0x00A5,-0x00A5,-0x00A5, + + -0x004F,-0x004F,-0x004F,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, + -0x00B2,-0x00B2,-0x004F,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, + -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, + -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, + -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, + -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, + -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x004F, + -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x004F,-0x004F, + + -0x003E,-0x003E,-0x003E,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, + -0x0060,-0x0060,-0x003E,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, + -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, + -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, + -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, + -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, + -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x003E, + -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x003E,-0x003E +}; + +int main(int argc, char **argv) +{ + const unsigned int video_width = 32, video_height = 32; + const unsigned int window_width = video_width * 2, window_height = video_height * 2; + int quit = 0; + Display *display; + Window root, window; + Pixmap framebuffer; + XEvent event; + struct pipe_context *pipe; + struct VL_CONTEXT *ctx; + struct VL_SURFACE *sfc, *past_sfc, *future_sfc; + struct VL_MOTION_VECTOR motion_vector[2] = + { + { + {0, 0}, {0, 0} + }, + { + {0, 0}, {0, 0} + } + }; + + display = XOpenDisplay(NULL); + root = XDefaultRootWindow(display); + window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); + framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + + pipe = create_pipe_context(display); + vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); + vlCreateSurface(ctx, &sfc); + vlCreateSurface(ctx, &past_sfc); + vlCreateSurface(ctx, &future_sfc); + + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); + vlRenderBMacroBlock + ( + VL_FRAME_PICTURE, + VL_FIELD_FIRST, + 0, + 0, + VL_FRAME_MC, + motion_vector, + 0x3F, + VL_DCT_FRAME_CODED, + (short*)ycbcr16x16_420_2, + past_sfc, + future_sfc, + sfc + ); + vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); + + puts("Press any key to continue..."); + + while (!quit) + { + XNextEvent(display, &event); + switch (event.type) + { + case Expose: + { + XCopyArea + ( + display, + framebuffer, + window, + XDefaultGC(display, XDefaultScreen(display)), + 0, + 0, + window_width, + window_height, + 0, + 0 + ); + break; + } + case KeyPress: + { + quit = 1; + break; + } + } + } + + vlDestroySurface(sfc); + vlDestroySurface(past_sfc); + vlDestroySurface(future_sfc); + vlDestroyContext(ctx); + + XFreePixmap(display, framebuffer); + XDestroyWindow(display, window); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/tests/test_context.c b/src/gallium/state_trackers/g3dvl/tests/test_context.c new file mode 100644 index 00000000000..2002977ee24 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/test_context.c @@ -0,0 +1,22 @@ +#include +#include + +int main(int argc, char **argv) +{ + const unsigned int video_width = 32, video_height = 32; + + Display *display; + struct pipe_context *pipe; + struct VL_CONTEXT *ctx; + + display = XOpenDisplay(NULL); + pipe = create_pipe_context(display); + + vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); + vlDestroyContext(ctx); + + XCloseDisplay(display); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c new file mode 100644 index 00000000000..1f964711308 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c @@ -0,0 +1,137 @@ +#include +#include +#include +#include +#include + +static const unsigned short ycbcr16x16_420[8*8*6] = +{ + 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, + + 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, + + 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E +}; + +int main(int argc, char **argv) +{ + const unsigned int video_width = 32, video_height = 32; + const unsigned int window_width = video_width * 2, window_height = video_height * 2; + int quit = 0; + Display *display; + Window root, window; + Pixmap framebuffer; + XEvent event; + struct pipe_context *pipe; + struct VL_CONTEXT *ctx; + struct VL_SURFACE *sfc; + + display = XOpenDisplay(NULL); + root = XDefaultRootWindow(display); + window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); + framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + + pipe = create_pipe_context(display); + vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); + vlCreateSurface(ctx, &sfc); + + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); + vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); + + puts("Press any key to continue..."); + + while (!quit) + { + XNextEvent(display, &event); + switch (event.type) + { + case Expose: + { + XCopyArea + ( + display, + framebuffer, + window, + XDefaultGC(display, XDefaultScreen(display)), + 0, + 0, + window_width, + window_height, + 0, + 0 + ); + break; + } + case KeyPress: + { + quit = 1; + break; + } + } + } + + vlDestroySurface(sfc); + vlDestroyContext(ctx); + + XFreePixmap(display, framebuffer); + XDestroyWindow(display, window); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c new file mode 100644 index 00000000000..2203349784f --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c @@ -0,0 +1,214 @@ +#include +#include +#include +#include +#include + +static const unsigned short ycbcr16x16_420[8*8*6] = +{ + 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, + + 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, + + 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E +}; + +static const signed short ycbcr16x16_420_2[8*8*6] = +{ + -51,-51,-51, 0,-51, 0, 0, 0, + 0,-51, 0, 0,-51, 0, 0, 0, + 0,-51, 0, 0,-51,-51, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 99, 99, 99, 0, 0, 0, 0, 0, + 0, 0, 99, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 33, 33, 33, 0, 0, 0, 0, 0, + 0, 0, 33, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0 +}; + +int main(int argc, char **argv) +{ + const unsigned int video_width = 32, video_height = 32; + const unsigned int window_width = video_width * 2, window_height = video_height * 2; + int quit = 0; + Display *display; + Window root, window; + Pixmap framebuffer; + XEvent event; + struct pipe_context *pipe; + struct VL_CONTEXT *ctx; + struct VL_SURFACE *sfc, *ref_sfc; + struct VL_MOTION_VECTOR motion_vector = + { + {0, 0}, {0, 0} + }; + + display = XOpenDisplay(NULL); + root = XDefaultRootWindow(display); + window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); + framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + + pipe = create_pipe_context(display); + vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); + vlCreateSurface(ctx, &sfc); + vlCreateSurface(ctx, &ref_sfc); + + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderPMacroBlock + ( + VL_FRAME_PICTURE, + VL_FIELD_FIRST, + 0, + 0, + VL_FRAME_MC, + &motion_vector, + 0x3F, + VL_DCT_FRAME_CODED, + (short*)ycbcr16x16_420_2, + ref_sfc, + sfc + ); + vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); + + puts("Press any key to continue..."); + + while (!quit) + { + XNextEvent(display, &event); + switch (event.type) + { + case Expose: + { + XCopyArea + ( + display, + framebuffer, + window, + XDefaultGC(display, XDefaultScreen(display)), + 0, + 0, + window_width, + window_height, + 0, + 0 + ); + break; + } + case KeyPress: + { + quit = 1; + break; + } + } + } + + vlDestroySurface(sfc); + vlDestroySurface(ref_sfc); + vlDestroyContext(ctx); + + XFreePixmap(display, framebuffer); + XDestroyWindow(display, window); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/tests/test_surface.c b/src/gallium/state_trackers/g3dvl/tests/test_surface.c new file mode 100644 index 00000000000..4d1946396af --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/test_surface.c @@ -0,0 +1,26 @@ +#include +#include +#include + +int main(int argc, char **argv) +{ + const unsigned int video_width = 32, video_height = 32; + + Display *display; + struct pipe_context *pipe; + struct VL_CONTEXT *ctx; + struct VL_SURFACE *sfc; + + display = XOpenDisplay(NULL); + pipe = create_pipe_context(display); + + vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); + vlCreateSurface(ctx, &sfc); + vlDestroySurface(sfc); + vlDestroyContext(ctx); + + XCloseDisplay(display); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c new file mode 100644 index 00000000000..7193f7cceaf --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -0,0 +1,2293 @@ +#include "vl_context.h" +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "vl_data.h" + +static int vlInitIDCT(struct VL_CONTEXT *context) +{ + assert(context); + + + + return 0; +} + +static int vlDestroyIDCT(struct VL_CONTEXT *context) +{ + assert(context); + + + + return 0; +} + +static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 50; + const unsigned int num_attribs = 4; + const unsigned int semantic_names[4] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC + }; + const unsigned int semantic_indexes[4] = {0, 1, 2, 3}; + const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant inputs */ + /* C[0] scales the normalized MB to cover 16x16 pixels, + C[1] translates the macroblock into position on the surface */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 1; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, c1 ; Translate vertex into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* + mov o1, i1 ; Move texcoords to output + mov o2, i2 + mov o3, i3 + */ + for (i = 1; i < num_attribs; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.mc.i_vs = pipe->create_vs_state(pipe, &vs); + + return 0; +} + +static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 50; + const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (texcoords) */ + for (i = 0; i < 3; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = i + 1; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare output (color) */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare samplers */ + for (i = 0; i < 3; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* + tex2d o0.x, i0, s0 ; Read texel from luma texture into .x channel + tex2d o0.y, i1, s1 ; Read texel from chroma Cb texture into .y channel + tex2d o0.z, i2, s2 ; Read texel from chroma Cr texture into .z channel + */ + for (i = 0; i < 3; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + fs.tokens = tokens; + + context->states.mc.i_fs = pipe->create_fs_state(pipe, &fs); + + return 0; +} + +static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int num_attribs = 5; + const unsigned int semantic_names[5] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC + }; + const unsigned int semantic_indexes[5] = {0, 1, 2, 3, 4}; + const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant inputs */ + /* C[0] scales the normalized MB to cover 16x16 pixels, + C[1] translates the macroblock into position on the surface + C[2] translates the ref surface texcoords to the ref macroblock */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 2; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, c1 ; Translate vertex into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* + mov o1, i1 ; Move luma & chroma texcoords to output + mov o2, i2 + mov o3, i3 + */ + for (i = 1; i < num_attribs - 1; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i4, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o4, t0, c2 ; Translate texcoords into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 4; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.mc.p_vs = pipe->create_vs_state(pipe, &vs); + + return 0; +} + +static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (texcoords) */ + for (i = 0; i < 4; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = i + 1; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant input */ + /* C[0] is a multiplier to use when concatenating differential into a single channel + C[0] is a bias to get differential back to -1,1 range*/ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 1; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare output */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare samplers */ + for (i = 0; i < 4; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* + tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + mov t1.x, t0.w ; Move high part from .w channel to .x + tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + mov t1.y, t0.w ; Move high part from .w channel to .y + tex2d t0.zw, i2, s2 ; Read texel from chroma Cr texture into .z and .w channels + mov t1.z, t0.w ; Move high part from .w channel to .z + */ + for (i = 0; i < 3; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* tex2d t1, i3, s3 ; Read texel from ref macroblock */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + fs.tokens = tokens; + + context->states.mc.p_fs = pipe->create_fs_state(pipe, &fs); + + return 0; +} + +static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int num_attribs = 6; + const unsigned int semantic_names[6] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC + }; + const unsigned int semantic_indexes[6] = {0, 1, 2, 3, 4, 5}; + const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant inputs */ + /* C[0] scales the normalized MB to cover 16x16 pixels, + C[1] translates the macroblock into position on the surface + C[2] translates the past surface texcoords to the ref macroblock + C[3] translates the future surface texcoords to the ref macroblock */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 3; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, c1 ; Translate vertex into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* + mov o1, i1 ; Move luma & chroma texcoords to output + mov o2, i2 + mov o3, i3 + */ + for (i = 1; i < num_attribs - 1; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i4, c0 ; Scale normalized coords to window coords + add o4, t0, c2 ; Translate texcoords into position + mul t1, i5, c0 ; Repeat for the future surface + add o5, t1, c3 */ + for (i = 0; i < 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i + 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i + 4; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = i + 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.mc.b_vs = pipe->create_vs_state(pipe, &vs); + + return 0; +} + +static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (texcoords) */ + for (i = 0; i < 5; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = i + 1; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant input */ + /* C[0] is a multiplier to use when concatenating differential into a single channel + C[0] is a bias to get differential back to -1,1 range*/ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 1; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare output */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare samplers */ + for (i = 0; i < 5; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* + tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + mov t1.x, t0.w ; Move high part from .w channel to .x + tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + mov t1.y, t0.w ; Move high part from .w channel to .y + tex2d t0.zw, i2, s2 ; Read texel from chroma Cr texture into .z and .w channels + mov t1.z, t0.w ; Move high part from .w channel to .z + */ + for (i = 0; i < 3; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* tex2d t1, i3, s3 ; Read texel from past macroblock + tex2d t2, i4, s4 ; Read texel from future macroblock */ + for (i = 0; i < 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = i + 1; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i + 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* add t0, t0, t1 ; Add past and differential to form partial output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, t2 ; Add future and differential to form final output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + fs.tokens = tokens; + + context->states.mc.b_fs = pipe->create_fs_state(pipe, &fs); + + return 0; +} + +int vlCreateDataBufsMC(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + /* Create our vertex buffer and vertex buffer element */ + context->states.mc.vertex_bufs[0].pitch = sizeof(struct VL_VERTEX2F); + context->states.mc.vertex_bufs[0].max_index = 23; + context->states.mc.vertex_bufs[0].buffer_offset = 0; + context->states.mc.vertex_bufs[0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct VL_VERTEX2F) * 24 + ); + + context->states.mc.vertex_buf_elems[0].src_offset = 0; + context->states.mc.vertex_buf_elems[0].vertex_buffer_index = 0; + context->states.mc.vertex_buf_elems[0].nr_components = 2; + context->states.mc.vertex_buf_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Create our texcoord buffers and texcoord buffer elements */ + /* TODO: Should be able to use 1 texcoord buf for chroma textures, 1 buf for ref surfaces */ + for (i = 1; i < 6; ++i) + { + context->states.mc.vertex_bufs[i].pitch = sizeof(struct VL_TEXCOORD2F); + context->states.mc.vertex_bufs[i].max_index = 23; + context->states.mc.vertex_bufs[i].buffer_offset = 0; + context->states.mc.vertex_bufs[i].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct VL_TEXCOORD2F) * 24 + ); + + context->states.mc.vertex_buf_elems[i].src_offset = 0; + context->states.mc.vertex_buf_elems[i].vertex_buffer_index = i; + context->states.mc.vertex_buf_elems[i].nr_components = 2; + context->states.mc.vertex_buf_elems[i].src_format = PIPE_FORMAT_R32G32_FLOAT; + } + + /* Fill buffers */ + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_chroma_420_texcoords, + sizeof(struct VL_VERTEX2F) * 24 + ); + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_luma_texcoords, + sizeof(struct VL_TEXCOORD2F) * 24 + ); + /* TODO: Accomodate 422, 444 */ + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[2].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_chroma_420_texcoords, + sizeof(struct VL_TEXCOORD2F) * 24 + ); + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[3].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_chroma_420_texcoords, + sizeof(struct VL_TEXCOORD2F) * 24 + ); + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[4].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_ref_surface_texcoords, + sizeof(struct VL_TEXCOORD2F) * 24 + ); + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[5].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_ref_surface_texcoords, + sizeof(struct VL_TEXCOORD2F) * 24 + ); + + for (i = 0; i < 6; ++i) + pipe->winsys->buffer_unmap(pipe->winsys, context->states.mc.vertex_bufs[i].buffer); + + /* Create our constant buffer */ + context->states.mc.vs_const_buf.size = sizeof(struct VL_MC_VS_CONSTS); + context->states.mc.vs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + context->states.mc.vs_const_buf.size + ); + + context->states.mc.fs_const_buf.size = sizeof(struct VL_MC_FS_CONSTS); + context->states.mc.fs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + context->states.mc.fs_const_buf.size + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.mc.fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + &vl_mc_fs_consts, + sizeof(struct VL_MC_FS_CONSTS) + ); + + pipe->winsys->buffer_unmap(pipe->winsys, context->states.mc.fs_const_buf.buffer); + + return 0; +} + +static int vlInitMC(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + struct pipe_sampler_state sampler; + struct pipe_texture template; + unsigned int filters[5]; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + context->states.mc.viewport.scale[0] = context->video_width; + context->states.mc.viewport.scale[1] = context->video_height; + context->states.mc.viewport.scale[2] = 1; + context->states.mc.viewport.scale[3] = 1; + context->states.mc.viewport.translate[0] = 0; + context->states.mc.viewport.translate[1] = 0; + context->states.mc.viewport.translate[2] = 0; + context->states.mc.viewport.translate[3] = 0; + + context->states.mc.render_target.width = context->video_width; + context->states.mc.render_target.height = context->video_height; + context->states.mc.render_target.num_cbufs = 1; + /* FB for MC stage is a VL_SURFACE, set in vlSetRenderSurface() */ + /*context->states.mc.render_target.cbufs[0] = ;*/ + context->states.mc.render_target.zsbuf = NULL; + + filters[0] = PIPE_TEX_FILTER_NEAREST; + filters[1] = context->video_format == VL_FORMAT_YCBCR_444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + filters[2] = context->video_format == VL_FORMAT_YCBCR_444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + filters[3] = PIPE_TEX_FILTER_NEAREST; + filters[4] = PIPE_TEX_FILTER_NEAREST; + + for (i = 0; i < 5; ++i) + { + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = filters[i]; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = filters[i]; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.shadow_ambient = ;*/ + /*sampler.lod_bias = ;*/ + /*sampler.min_lod = ;*/ + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + context->states.mc.samplers[i] = pipe->create_sampler_state(pipe, &sampler); + } + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + template.format = PIPE_FORMAT_A8L8_UNORM; + template.last_level = 0; + template.width[0] = 8; + template.height[0] = 8 * 4; + template.depth[0] = 1; + template.compressed = 0; + template.cpp = 2; + context->states.mc.textures[0] = pipe->screen->texture_create(pipe->screen, &template); + + if (context->video_format == VL_FORMAT_YCBCR_420) + template.height[0] = 8; + else if (context->video_format == VL_FORMAT_YCBCR_422) + template.height[0] = 8 * 2; + else if (context->video_format == VL_FORMAT_YCBCR_444) + template.height[0] = 8 * 4; + else + assert(0); + + context->states.mc.textures[1] = pipe->screen->texture_create(pipe->screen, &template); + context->states.mc.textures[2] = pipe->screen->texture_create(pipe->screen, &template); + + /* textures[3] & textures[4] are assigned from VL_SURFACEs for P and B macroblocks at render time */ + + vlCreateVertexShaderIMC(context); + vlCreateFragmentShaderIMC(context); + vlCreateVertexShaderPMC(context); + vlCreateFragmentShaderPMC(context); + vlCreateVertexShaderBMC(context); + vlCreateFragmentShaderBMC(context); + vlCreateDataBufsMC(context); + + return 0; +} + +static int vlDestroyMC(struct VL_CONTEXT *context) +{ + unsigned int i; + + assert(context); + + for (i = 0; i < 5; ++i) + { + context->pipe->delete_sampler_state(context->pipe, context->states.mc.samplers[i]); + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vertex_bufs[i].buffer); + } + + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vertex_bufs[5].buffer); + + /* Textures 3 & 4 are not created directly, no need to release them here */ + for (i = 0; i < 3; ++i) + pipe_texture_release(&context->states.mc.textures[i]); + + context->pipe->delete_vs_state(context->pipe, context->states.mc.i_vs); + context->pipe->delete_fs_state(context->pipe, context->states.mc.i_fs); + context->pipe->delete_vs_state(context->pipe, context->states.mc.p_vs); + context->pipe->delete_fs_state(context->pipe, context->states.mc.p_fs); + context->pipe->delete_vs_state(context->pipe, context->states.mc.b_vs); + context->pipe->delete_fs_state(context->pipe, context->states.mc.b_fs); + + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vs_const_buf.buffer); + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.fs_const_buf.buffer); + + return 0; +} + +static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 50; + const unsigned int num_attribs = 2; + const unsigned int semantic_names[2] = {TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC}; + const unsigned int semantic_indexes[2] = {0, 1}; + const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = semantic_names[i]; + decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* MOV instructions */ + /* mov o0, i0 + mov o1, i1 */ + for (i = 0; i < num_attribs; i++) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* END instruction */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.csc.vertex_shader = pipe->create_vs_state(pipe, &vs); + + return 0; +} + +static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 50; + const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare TEX[0] input */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 1; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare constant input */ + /* Constants include bias vector, 4x4 csc matrix, total 5 vectors */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 4; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare output */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare sampler */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* TEX instruction */ + /* tex2d t0, i0, s0 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* SUB instruction */ + /* sub t0, t0, c0 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* DP4 instruction */ + /* dp4 o0.x, t0, c1 + dp4 o0.y, t0, c2 + dp4 o0.z, t0, c3 + dp4 o0.w, t0, c4 */ + for (i = 0; i < 4; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_DP4; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = i + 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* END instruction */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + fs.tokens = tokens; + + context->states.csc.fragment_shader = pipe->create_fs_state(pipe, &fs); + + return 0; +} + +static int vlCreateDataBufsCSC(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + + assert(context); + + pipe = context->pipe; + + /* + Create our vertex buffer and vertex buffer element + VB contains 4 vertices that render a quad covering the entire window + to display a rendered surface + Quad is rendered as a tri strip + */ + context->states.csc.vertex_bufs[0].pitch = sizeof(struct VL_VERTEX2F); + context->states.csc.vertex_bufs[0].max_index = 3; + context->states.csc.vertex_bufs[0].buffer_offset = 0; + context->states.csc.vertex_bufs[0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct VL_VERTEX2F) * 4 + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.csc.vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_surface_vertex_positions, + sizeof(struct VL_VERTEX2F) * 4 + ); + + pipe->winsys->buffer_unmap(pipe->winsys, context->states.csc.vertex_bufs[0].buffer); + + context->states.csc.vertex_buf_elems[0].src_offset = 0; + context->states.csc.vertex_buf_elems[0].vertex_buffer_index = 0; + context->states.csc.vertex_buf_elems[0].nr_components = 2; + context->states.csc.vertex_buf_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* + Create our texcoord buffer and texcoord buffer element + Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices + */ + context->states.csc.vertex_bufs[1].pitch = sizeof(struct VL_TEXCOORD2F); + context->states.csc.vertex_bufs[1].max_index = 3; + context->states.csc.vertex_bufs[1].buffer_offset = 0; + context->states.csc.vertex_bufs[1].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct VL_TEXCOORD2F) * 4 + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.csc.vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + vl_surface_texcoords, + sizeof(struct VL_TEXCOORD2F) * 4 + ); + + pipe->winsys->buffer_unmap(pipe->winsys, context->states.csc.vertex_bufs[1].buffer); + + context->states.csc.vertex_buf_elems[1].src_offset = 0; + context->states.csc.vertex_buf_elems[1].vertex_buffer_index = 1; + context->states.csc.vertex_buf_elems[1].nr_components = 2; + context->states.csc.vertex_buf_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* + Create our fragment shader's constant buffer + Const buffer contains the color conversion matrix and bias vectors + */ + context->states.csc.fs_const_buf.size = sizeof(struct VL_CSC_FS_CONSTS); + context->states.csc.fs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + context->states.csc.fs_const_buf.size + ); + + /* + TODO: Refactor this into a seperate function, + allow changing the csc matrix at runtime to switch between regular & full versions + */ + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, context->states.csc.fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + &vl_csc_fs_consts_601, + sizeof(struct VL_CSC_FS_CONSTS) + ); + + pipe->winsys->buffer_unmap(pipe->winsys, context->states.csc.fs_const_buf.buffer); + + return 0; +} + +static int vlInitCSC(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + struct pipe_sampler_state sampler; + + assert(context); + + pipe = context->pipe; + + /* Delay creating the FB until vlPutSurface() so we know window size */ + context->states.csc.framebuffer.num_cbufs = 1; + context->states.csc.framebuffer.cbufs[0] = NULL; + context->states.csc.framebuffer.zsbuf = NULL; + + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.shadow_ambient = ;*/ + /*sampler.lod_bias = ;*/ + /*sampler.min_lod = ;*/ + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + context->states.csc.sampler = pipe->create_sampler_state(pipe, &sampler); + + vlCreateVertexShaderCSC(context); + vlCreateFragmentShaderCSC(context); + vlCreateDataBufsCSC(context); + + return 0; +} + +static int vlDestroyCSC(struct VL_CONTEXT *context) +{ + assert(context); + + /* + Since we create the final FB when we display our first surface, + it may not be created if vlPutSurface() is never called + */ + if (context->states.csc.framebuffer.cbufs[0]) + context->pipe->winsys->surface_release(context->pipe->winsys, &context->states.csc.framebuffer.cbufs[0]); + context->pipe->delete_sampler_state(context->pipe, context->states.csc.sampler); + context->pipe->delete_vs_state(context->pipe, context->states.csc.vertex_shader); + context->pipe->delete_fs_state(context->pipe, context->states.csc.fragment_shader); + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vertex_bufs[0].buffer); + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vertex_bufs[1].buffer); + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.fs_const_buf.buffer); + + return 0; +} + +static int vlInitCommon(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + struct pipe_rasterizer_state rast; + struct pipe_blend_state blend; + struct pipe_depth_stencil_alpha_state dsa; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + rast.flatshade = 1; + rast.light_twoside = 0; + rast.front_winding = PIPE_WINDING_CCW; + rast.cull_mode = PIPE_WINDING_CW; + rast.fill_cw = PIPE_POLYGON_MODE_FILL; + rast.fill_ccw = PIPE_POLYGON_MODE_FILL; + rast.offset_cw = 0; + rast.offset_ccw = 0; + rast.scissor = 0; + rast.poly_smooth = 0; + rast.point_sprite = 0; + rast.point_size_per_vertex = 0; + rast.multisample = 0; + rast.line_smooth = 0; + rast.line_stipple_enable = 0; + rast.line_stipple_factor = 0; + rast.line_stipple_pattern = 0; + rast.line_last_pixel = 0; + /* Don't need clipping, but viewport mapping done here */ + rast.bypass_clipping = 0; + rast.bypass_vs = 0; + rast.origin_lower_left = 0; + rast.line_width = 1; + rast.point_size = 1; + rast.offset_units = 1; + rast.offset_scale = 1; + /*rast.sprite_coord_mode[i] = ;*/ + context->states.common.raster = pipe->create_rasterizer_state(pipe, &rast); + pipe->bind_rasterizer_state(pipe, context->states.common.raster); + + blend.blend_enable = 0; + blend.rgb_func = PIPE_BLEND_ADD; + blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE; + blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE; + blend.alpha_func = PIPE_BLEND_ADD; + blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE; + blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE; + blend.logicop_enable = 0; + blend.logicop_func = PIPE_LOGICOP_CLEAR; + /* Needed to allow color writes to FB, even if blending disabled */ + blend.colormask = PIPE_MASK_RGBA; + blend.dither = 0; + context->states.common.blend = pipe->create_blend_state(pipe, &blend); + pipe->bind_blend_state(pipe, context->states.common.blend); + + dsa.depth.enabled = 0; + dsa.depth.writemask = 0; + dsa.depth.func = PIPE_FUNC_ALWAYS; + dsa.depth.occlusion_count = 0; + for (i = 0; i < 2; ++i) + { + dsa.stencil[i].enabled = 0; + dsa.stencil[i].func = PIPE_FUNC_ALWAYS; + dsa.stencil[i].fail_op = PIPE_STENCIL_OP_KEEP; + dsa.stencil[i].zpass_op = PIPE_STENCIL_OP_KEEP; + dsa.stencil[i].zfail_op = PIPE_STENCIL_OP_KEEP; + dsa.stencil[i].ref_value = 0; + dsa.stencil[i].value_mask = 0; + dsa.stencil[i].write_mask = 0; + } + dsa.alpha.enabled = 0; + dsa.alpha.func = PIPE_FUNC_ALWAYS; + dsa.alpha.ref = 0; + context->states.common.dsa = pipe->create_depth_stencil_alpha_state(pipe, &dsa); + pipe->bind_depth_stencil_alpha_state(pipe, context->states.common.dsa); + + return 0; +} + +static int vlDestroyCommon(struct VL_CONTEXT *context) +{ + assert(context); + + context->pipe->delete_blend_state(context->pipe, context->states.common.blend); + context->pipe->delete_rasterizer_state(context->pipe, context->states.common.raster); + context->pipe->delete_depth_stencil_alpha_state(context->pipe, context->states.common.dsa); + + return 0; +} + +static int vlInit(struct VL_CONTEXT *context) +{ + assert(context); + + vlInitCommon(context); + vlInitCSC(context); + vlInitMC(context); + vlInitIDCT(context); + + return 0; +} + +static int vlDestroy(struct VL_CONTEXT *context) +{ + assert(context); + + /* Must unbind shaders before we can delete them for some reason */ + context->pipe->bind_vs_state(context->pipe, NULL); + context->pipe->bind_fs_state(context->pipe, NULL); + + vlDestroyCommon(context); + vlDestroyCSC(context); + vlDestroyMC(context); + vlDestroyIDCT(context); + + return 0; +} + +int vlCreateContext +( + Display *display, + struct pipe_context *pipe, + unsigned int video_width, + unsigned int video_height, + enum VL_FORMAT video_format, + struct VL_CONTEXT **context +) +{ + struct VL_CONTEXT *ctx; + + assert(display); + assert(pipe); + assert(context); + + ctx = calloc(1, sizeof(struct VL_CONTEXT)); + + ctx->display = display; + ctx->pipe = pipe; + ctx->video_width = video_width; + ctx->video_height = video_height; + ctx->video_format = video_format; + + vlInit(ctx); + + /* Since we only change states in vlPutSurface() we need to start in render mode */ + vlBeginRender(ctx); + + *context = ctx; + + return 0; +} + +int vlDestroyContext(struct VL_CONTEXT *context) +{ + assert(context); + + vlDestroy(context); + + free(context); + + return 0; +} + +int vlBeginRender(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + + assert(context); + + pipe = context->pipe; + + /* Frame buffer set in vlRender*Macroblock() */ + /* Shaders, samplers, textures, VBs, VB elements set in vlRender*Macroblock() */ + pipe->set_viewport_state(pipe, &context->states.mc.viewport); + pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &context->states.mc.vs_const_buf); + pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &context->states.mc.fs_const_buf); + + return 0; +} + +int vlEndRender(struct VL_CONTEXT *context) +{ + struct pipe_context *pipe; + + assert(context); + + pipe = context->pipe; + + pipe->set_framebuffer_state(pipe, &context->states.csc.framebuffer); + pipe->set_viewport_state(pipe, &context->states.csc.viewport); + pipe->bind_sampler_states(pipe, 1, (void**)&context->states.csc.sampler); + /* Source texture set in vlPutSurface() */ + pipe->bind_vs_state(pipe, context->states.csc.vertex_shader); + pipe->bind_fs_state(pipe, context->states.csc.fragment_shader); + pipe->set_vertex_buffers(pipe, 2, context->states.csc.vertex_bufs); + pipe->set_vertex_elements(pipe, 2, context->states.csc.vertex_buf_elems); + pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &context->states.csc.fs_const_buf); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h new file mode 100644 index 00000000000..0aeba184ccb --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_context.h @@ -0,0 +1,73 @@ +#ifndef vl_context_h +#define vl_context_h + +#include +#include +#include "vl_types.h" + +struct pipe_context; + +struct VL_CONTEXT +{ + Display *display; + struct pipe_context *pipe; + unsigned int video_width; + unsigned int video_height; + enum VL_FORMAT video_format; + + struct + { + struct + { + struct pipe_rasterizer_state *raster; + struct pipe_depth_stencil_alpha_state *dsa; + struct pipe_blend_state *blend; + } common; + + struct + { + } idct; + + struct + { + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state render_target; + struct pipe_sampler_state *samplers[5]; + struct pipe_texture *textures[5]; + struct pipe_shader_state *i_vs, *p_vs, *b_vs; + struct pipe_shader_state *i_fs, *p_fs, *b_fs; + struct pipe_vertex_buffer vertex_bufs[6]; + struct pipe_vertex_element vertex_buf_elems[6]; + struct pipe_constant_buffer vs_const_buf, fs_const_buf; + } mc; + + struct + { + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state framebuffer; + struct pipe_sampler_state *sampler; + struct pipe_shader_state *vertex_shader, *fragment_shader; + struct pipe_vertex_buffer vertex_bufs[2]; + struct pipe_vertex_element vertex_buf_elems[2]; + struct pipe_constant_buffer fs_const_buf; + } csc; + } states; +}; + +int vlCreateContext +( + Display *display, + struct pipe_context *pipe, + unsigned int video_width, + unsigned int video_height, + enum VL_FORMAT video_format, + struct VL_CONTEXT **context +); + +int vlDestroyContext(struct VL_CONTEXT *context); + +int vlBeginRender(struct VL_CONTEXT *context); +int vlEndRender(struct VL_CONTEXT *context); + +#endif + diff --git a/src/gallium/state_trackers/g3dvl/vl_data.c b/src/gallium/state_trackers/g3dvl/vl_data.c new file mode 100644 index 00000000000..c04163276d8 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_data.c @@ -0,0 +1,188 @@ +#include "vl_data.h" + +/* + * Represents 8 triangles (4 quads, 1 per block) in noormalized coords + * that render a macroblock. + * Need to be scaled to cover mbW*mbH macroblock pixels and translated into + * position on target surface. + */ +const struct VL_VERTEX2F vl_mb_vertex_positions[24] = +{ + {0.0f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.0f}, + {0.5f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.5f}, + + {0.5f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.0f}, + {1.0f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.5f}, + + {0.0f, 0.5f}, {0.0f, 1.0f}, {0.5f, 0.5f}, + {0.5f, 0.5f}, {0.0f, 1.0f}, {0.5f, 1.0f}, + + {0.5f, 0.5f}, {0.5f, 1.0f}, {1.0f, 0.5f}, + {1.0f, 0.5f}, {0.5f, 1.0f}, {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above for rendering 4 luma blocks arranged + * in a bW*(bH*4) texture. First luma block located at 0,0->bW,bH; second at + * 0,bH->bW,2bH; third at 0,2bH->bW,3bH; fourth at 0,3bH->bW,4bH. + */ +const struct VL_TEXCOORD2F vl_luma_texcoords[24] = +{ + {0.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.0f}, + {1.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.25f}, + + {0.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.25f}, + {1.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.5f}, + + {0.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.5f}, + {1.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.75f}, + + {0.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 0.75f}, + {1.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above for rendering 1 chroma block. + * Straight forward 0,0->1,1 mapping so we can reuse the MB pos vectors. + */ +const struct VL_TEXCOORD2F *vl_chroma_420_texcoords = (const struct VL_TEXCOORD2F*)vl_mb_vertex_positions; + +/* + * Represents texcoords for the above for rendering 2 chroma blocks arranged + * in a bW*(bH*2) texture. First chroma block located at 0,0->bW,bH; second at + * 0,bH->bW,2bH. We can render this with 0,0->1,1 mapping. + * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. + */ +const struct VL_TEXCOORD2F *vl_chroma_422_texcoords = (const struct VL_TEXCOORD2F*)vl_mb_vertex_positions; + +/* + * Represents texcoords for the above for rendering 4 chroma blocks. + * Same case as 4 luma blocks. + */ +const struct VL_TEXCOORD2F *vl_chroma_444_texcoords = vl_luma_texcoords; + +/* + * Represents texcoords for the above for rendering a predicted macroblock. + * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. + * Texcoords need to be translated to cover source macroblock on the + * past/future surface. + */ + const struct VL_TEXCOORD2F *vl_ref_surface_texcoords = (const struct VL_TEXCOORD2F*)vl_mb_vertex_positions; + +/* + * Represents 2 triangles in a strip in normalized coords. + * Used to render the surface onto the frame buffer. + */ +const struct VL_VERTEX2F vl_surface_vertex_positions[4] = +{ + {0.0f, 0.0f}, + {0.0f, 1.0f}, + {1.0f, 0.0f}, + {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above. We can use the position values directly. + */ +const struct VL_TEXCOORD2F *vl_surface_texcoords = (const struct VL_TEXCOORD2F*)vl_surface_vertex_positions; + +/* + * Used when rendering P and B macroblocks, multiplier is applied to the A channel, + * which is then added to the L channel, then the bias is subtracted from that to + * get back the differential. The differential is then added to the samples from the + * reference surface(s). + */ +const struct VL_MC_FS_CONSTS vl_mc_fs_consts = +{ + {256.0f, 256.0f, 256.0f, 0.0f}, + {256.0f / 255.0f, 256.0f / 255.0f, 256.0f / 255.0f, 0.0f} +}; + +/* + * Identity color conversion constants, for debugging + */ +const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_identity = +{ + { + 0.0f, 0.0f, 0.0f, 0.0f + }, + { + 1.0f, 0.0f, 0.0f, 0.0f, + 0.0f, 1.0f, 0.0f, 0.0f, + 0.0f, 0.0f, 1.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [16,235] + */ +const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601 = +{ + { + 0.0f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.0f, 0.0f, 1.371f, 0.0f, + 1.0f, -0.336f, -0.698f, 0.0f, + 1.0f, 1.732f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [0,255] + */ +const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601_full = +{ + { + 0.062745098f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.164f, 0.0f, 1.596f, 0.0f, + 1.164f, -0.391f, -0.813f, 0.0f, + 1.164f, 2.018f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [16,235] + */ +const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709 = +{ + { + 0.0f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.0f, 0.0f, 1.540f, 0.0f, + 1.0f, -0.183f, -0.459f, 0.0f, + 1.0f, 1.816f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [0,255] + */ +const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709_full = +{ + { + 0.062745098f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.164f, 0.0f, 1.793f, 0.0f, + 1.164f, -0.213f, -0.534f, 0.0f, + 1.164f, 2.115f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + diff --git a/src/gallium/state_trackers/g3dvl/vl_data.h b/src/gallium/state_trackers/g3dvl/vl_data.h new file mode 100644 index 00000000000..67a0a749909 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_data.h @@ -0,0 +1,25 @@ +#ifndef vl_data_h +#define vl_data_h + +#include "vl_types.h" + +extern const struct VL_VERTEX2F vl_mb_vertex_positions[24]; +extern const struct VL_TEXCOORD2F vl_luma_texcoords[24]; +extern const struct VL_TEXCOORD2F *vl_chroma_420_texcoords; +extern const struct VL_TEXCOORD2F *vl_chroma_422_texcoords; +extern const struct VL_TEXCOORD2F *vl_chroma_444_texcoords; +extern const struct VL_TEXCOORD2F *vl_ref_surface_texcoords; + +extern const struct VL_VERTEX2F vl_surface_vertex_positions[4]; +extern const struct VL_TEXCOORD2F *vl_surface_texcoords; + +extern const struct VL_MC_FS_CONSTS vl_mc_fs_consts; + +extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_identity; +extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601; +extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601_full; +extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709; +extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709_full; + +#endif + diff --git a/src/gallium/state_trackers/g3dvl/vl_defs.h b/src/gallium/state_trackers/g3dvl/vl_defs.h new file mode 100644 index 00000000000..e668a7a10e0 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_defs.h @@ -0,0 +1,12 @@ +#ifndef vl_defs_h +#define vl_defs_h + +#define VL_BLOCK_WIDTH 8 +#define VL_BLOCK_HEIGHT 8 +#define VL_BLOCK_SIZE (VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT) +#define VL_MACROBLOCK_WIDTH 16 +#define VL_MACROBLOCK_HEIGHT 16 +#define VL_MACROBLOCK_SIZE (VL_MACROBLOCK_WIDTH * VL_MACROBLOCK_HEIGHT) + +#endif + diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c new file mode 100644 index 00000000000..e58e434dab1 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -0,0 +1,539 @@ +#include "vl_surface.h" +#include +#include +#include +#include +#include +#include +#include "vl_context.h" +#include "vl_defs.h" + +static int vlGrabBlocks +( + struct VL_CONTEXT *context, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + enum VL_SAMPLE_TYPE sample_type, + short *blocks +) +{ + struct pipe_surface *tex_surface; + short *texels; + unsigned int b, x, y, y2; + + assert(context); + assert(blocks); + + tex_surface = context->pipe->screen->get_tex_surface + ( + context->pipe->screen, + context->states.mc.textures[0], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); + + texels = pipe_surface_map(tex_surface, 0); + + for (b = 0; b < 4; ++b) + { + if ((coded_block_pattern >> b) & 1) + { + if (dct_type == VL_DCT_FRAME_CODED) + { + if (sample_type == VL_FULL_SAMPLE) + { + for (y = VL_BLOCK_HEIGHT * b; y < VL_BLOCK_HEIGHT * (b + 1); ++y) + memcpy + ( + texels + y * tex_surface->pitch, + blocks + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + } + else + { + for (y = VL_BLOCK_HEIGHT * b; y < VL_BLOCK_HEIGHT * (b + 1); ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + texels[y * tex_surface->pitch + x] = + blocks[y * VL_BLOCK_WIDTH + x] + 0x100; + } + } + else + { + if (sample_type == VL_FULL_SAMPLE) + { + for + ( + y = VL_BLOCK_HEIGHT * (b % 2), y2 = VL_BLOCK_HEIGHT * b; + y < VL_BLOCK_HEIGHT * ((b % 2) + 1); + y += 2, ++y2 + ) + memcpy + ( + texels + y * tex_surface->pitch, + blocks + y2 * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + for + ( + y = VL_BLOCK_HEIGHT * ((b % 2) + 2); + y < VL_BLOCK_HEIGHT * (((b % 2) + 2) + 1); + y += 2, ++y2 + ) + memcpy + ( + texels + y * tex_surface->pitch, + blocks + y2 * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + } + else + { + for + ( + y = VL_BLOCK_HEIGHT * (b % 2), y2 = VL_BLOCK_HEIGHT * b; + y < VL_BLOCK_HEIGHT * ((b % 2) + 1); + y += 2, ++y2 + ) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + texels[y * tex_surface->pitch + x] = + blocks[y2 * VL_BLOCK_WIDTH + x] + 0x100; + for + ( + y = VL_BLOCK_HEIGHT * ((b % 2) + 2); + y < VL_BLOCK_HEIGHT * (((b % 2) + 2) + 1); + y += 2, ++y2 + ) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + texels[y * tex_surface->pitch + x] = + blocks[y2 * VL_BLOCK_WIDTH + x] + 0x100; + } + } + } + else + { + for (y = VL_BLOCK_HEIGHT * b; y < VL_BLOCK_HEIGHT * (b + 1); ++y) + { + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + texels[y * tex_surface->pitch + x] = 0x100; + } + } + } + + pipe_surface_unmap(tex_surface); + + /* TODO: Implement 422, 444 */ + for (b = 0; b < 2; ++b) + { + tex_surface = context->pipe->screen->get_tex_surface + ( + context->pipe->screen, + context->states.mc.textures[b + 1], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); + + texels = pipe_surface_map(tex_surface, 0); + + if ((coded_block_pattern >> (b + 4)) & 1) + { + if (sample_type == VL_FULL_SAMPLE) + { + for (y = 0; y < tex_surface->height; ++y) + memcpy + ( + texels + y * tex_surface->pitch, + blocks + VL_BLOCK_SIZE * (b + 4) + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + } + else + { + for (y = 0; y < tex_surface->height; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + texels[y * tex_surface->pitch + x] = + blocks[VL_BLOCK_SIZE * (b + 4) + y * VL_BLOCK_WIDTH + x] + 0x100; + } + } + else + { + for (y = 0; y < tex_surface->height; ++y) + { + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + texels[y * tex_surface->pitch + x] = 0x100; + } + } + + pipe_surface_unmap(tex_surface); + } + + return 0; +} + +int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface) +{ + struct pipe_context *pipe; + struct pipe_texture template; + struct VL_SURFACE *sfc; + + assert(context); + assert(surface); + + pipe = context->pipe; + + sfc = calloc(1, sizeof(struct VL_SURFACE)); + + sfc->context = context; + sfc->width = context->video_width; + sfc->height = context->video_height; + sfc->format = context->video_format; + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + template.format = PIPE_FORMAT_A8R8G8B8_UNORM; + template.last_level = 0; + template.width[0] = sfc->width; + template.height[0] = sfc->height; + template.depth[0] = 1; + template.compressed = 0; + template.cpp = 4; + + sfc->texture = pipe->screen->texture_create(pipe->screen, &template); + + *surface = sfc; + + return 0; +} + +int vlDestroySurface(struct VL_SURFACE *surface) +{ + assert(surface); + pipe_texture_release(&surface->texture); + free(surface); + + return 0; +} + +int vlRenderIMacroBlock +( + enum VL_PICTURE picture_type, + enum VL_FIELD_ORDER field_order, + unsigned int mbx, + unsigned int mby, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + short *blocks, + struct VL_SURFACE *surface +) +{ + struct pipe_context *pipe; + struct VL_MC_VS_CONSTS *vscbdata; + + assert(blocks); + assert(surface); + + /* TODO: Implement interlaced rendering */ + /*assert(picture_type == VL_FRAME_PICTURE);*/ + if (picture_type != VL_FRAME_PICTURE) + { + /*fprintf(stderr, "field picture (I) unimplemented, ignoring\n");*/ + return 0; + } + + pipe = surface->context->pipe; + + vscbdata = pipe->winsys->buffer_map + ( + pipe->winsys, + surface->context->states.mc.vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vscbdata->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; + vscbdata->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; + vscbdata->scale.z = 1.0f; + vscbdata->scale.w = 1.0f; + vscbdata->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; + vscbdata->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; + vscbdata->mb_pos_trans.z = 0.0f; + vscbdata->mb_pos_trans.w = 0.0f; + + pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); + + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_FULL_SAMPLE, blocks); + + surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE + ); + pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); + pipe->set_sampler_textures(pipe, 3, surface->context->states.mc.textures); + pipe->bind_sampler_states(pipe, 3, (void**)surface->context->states.mc.samplers); + pipe->set_vertex_buffers(pipe, 4, surface->context->states.mc.vertex_bufs); + pipe->set_vertex_elements(pipe, 4, surface->context->states.mc.vertex_buf_elems); + pipe->bind_vs_state(pipe, surface->context->states.mc.i_vs); + pipe->bind_fs_state(pipe, surface->context->states.mc.i_fs); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + return 0; +} + +int vlRenderPMacroBlock +( + enum VL_PICTURE picture_type, + enum VL_FIELD_ORDER field_order, + unsigned int mbx, + unsigned int mby, + enum VL_MC_TYPE mc_type, + struct VL_MOTION_VECTOR *motion_vector, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + short *blocks, + struct VL_SURFACE *ref_surface, + struct VL_SURFACE *surface +) +{ + struct pipe_context *pipe; + struct VL_MC_VS_CONSTS *vscbdata; + + assert(motion_vectors); + assert(blocks); + assert(ref_surface); + assert(surface); + + /* TODO: Implement interlaced rendering */ + /*assert(picture_type == VL_FRAME_PICTURE);*/ + if (picture_type != VL_FRAME_PICTURE) + { + /*fprintf(stderr, "field picture (P) unimplemented, ignoring\n");*/ + return 0; + } + /* TODO: Implement field based motion compensation */ + /*assert(mc_type == VL_FRAME_MC);*/ + if (mc_type != VL_FRAME_MC) + { + /*fprintf(stderr, "field MC (P) unimplemented, ignoring\n");*/ + return 0; + } + + pipe = surface->context->pipe; + + vscbdata = pipe->winsys->buffer_map + ( + pipe->winsys, + surface->context->states.mc.vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vscbdata->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; + vscbdata->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; + vscbdata->scale.z = 1.0f; + vscbdata->scale.w = 1.0f; + vscbdata->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; + vscbdata->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; + vscbdata->mb_pos_trans.z = 0.0f; + vscbdata->mb_pos_trans.w = 0.0f; + vscbdata->mb_tc_trans[0].x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector->top_field.x * 0.5f) / (float)surface->width; + vscbdata->mb_tc_trans[0].y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector->top_field.y * 0.5f) / (float)surface->height; + vscbdata->mb_tc_trans[0].z = 0.0f; + vscbdata->mb_tc_trans[0].w = 0.0f; + + pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); + + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); + + surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE + ); + pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); + + surface->context->states.mc.textures[3] = ref_surface->texture; + pipe->set_sampler_textures(pipe, 4, surface->context->states.mc.textures); + pipe->bind_sampler_states(pipe, 4, (void**)surface->context->states.mc.samplers); + pipe->set_vertex_buffers(pipe, 5, surface->context->states.mc.vertex_bufs); + pipe->set_vertex_elements(pipe, 5, surface->context->states.mc.vertex_buf_elems); + pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs); + pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + return 0; +} + +int vlRenderBMacroBlock +( + enum VL_PICTURE picture_type, + enum VL_FIELD_ORDER field_order, + unsigned int mbx, + unsigned int mby, + enum VL_MC_TYPE mc_type, + struct VL_MOTION_VECTOR *motion_vector, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + short *blocks, + struct VL_SURFACE *past_surface, + struct VL_SURFACE *future_surface, + struct VL_SURFACE *surface +) +{ + struct pipe_context *pipe; + struct VL_MC_VS_CONSTS *vscbdata; + + assert(motion_vectors); + assert(blocks); + assert(ref_surface); + assert(surface); + + /* TODO: Implement interlaced rendering */ + /*assert(picture_type == VL_FRAME_PICTURE);*/ + if (picture_type != VL_FRAME_PICTURE) + { + /*fprintf(stderr, "field picture (B) unimplemented, ignoring\n");*/ + return 0; + } + /* TODO: Implement field based motion compensation */ + /*assert(mc_type == VL_FRAME_MC);*/ + if (mc_type != VL_FRAME_MC) + { + /*fprintf(stderr, "field MC (B) unimplemented, ignoring\n");*/ + return 0; + } + + pipe = surface->context->pipe; + + vscbdata = pipe->winsys->buffer_map + ( + pipe->winsys, + surface->context->states.mc.vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vscbdata->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; + vscbdata->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; + vscbdata->scale.z = 1.0f; + vscbdata->scale.w = 1.0f; + vscbdata->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; + vscbdata->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; + vscbdata->mb_pos_trans.z = 0.0f; + vscbdata->mb_pos_trans.w = 0.0f; + vscbdata->mb_tc_trans[0].x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[0].top_field.x * 0.5f) / (float)surface->width; + vscbdata->mb_tc_trans[0].y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[0].top_field.y * 0.5f) / (float)surface->height; + vscbdata->mb_tc_trans[0].z = 0.0f; + vscbdata->mb_tc_trans[0].w = 0.0f; + vscbdata->mb_tc_trans[1].x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[1].top_field.x * 0.5f) / (float)surface->width; + vscbdata->mb_tc_trans[1].y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[1].top_field.y * 0.5f) / (float)surface->height; + vscbdata->mb_tc_trans[1].z = 0.0f; + vscbdata->mb_tc_trans[1].w = 0.0f; + + pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); + + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); + + surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE + ); + pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); + + surface->context->states.mc.textures[3] = past_surface->texture; + surface->context->states.mc.textures[4] = future_surface->texture; + pipe->set_sampler_textures(pipe, 5, surface->context->states.mc.textures); + pipe->bind_sampler_states(pipe, 5, (void**)surface->context->states.mc.samplers); + pipe->set_vertex_buffers(pipe, 6, surface->context->states.mc.vertex_bufs); + pipe->set_vertex_elements(pipe, 6, surface->context->states.mc.vertex_buf_elems); + pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs); + pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + return 0; +} + +int vlPutSurface +( + struct VL_SURFACE *surface, + Drawable drawable, + unsigned int srcx, + unsigned int srcy, + unsigned int srcw, + unsigned int srch, + unsigned int destx, + unsigned int desty, + unsigned int destw, + unsigned int desth, + enum VL_PICTURE picture_type +) +{ + unsigned int create_fb = 0; + struct pipe_context *pipe; + + assert(surface); + + pipe = surface->context->pipe; + + if (!surface->context->states.csc.framebuffer.cbufs[0]) + create_fb = 1; + else if + ( + surface->context->states.csc.framebuffer.width != destw || + surface->context->states.csc.framebuffer.height != desth + ) + { + pipe->winsys->surface_release + ( + pipe->winsys, + &surface->context->states.csc.framebuffer.cbufs[0] + ); + + create_fb = 1; + } + + if (create_fb) + { + surface->context->states.csc.viewport.scale[0] = destw; + surface->context->states.csc.viewport.scale[1] = desth; + surface->context->states.csc.viewport.scale[2] = 1; + surface->context->states.csc.viewport.scale[3] = 1; + surface->context->states.csc.viewport.translate[0] = 0; + surface->context->states.csc.viewport.translate[1] = 0; + surface->context->states.csc.viewport.translate[2] = 0; + surface->context->states.csc.viewport.translate[3] = 0; + + surface->context->states.csc.framebuffer.width = destw; + surface->context->states.csc.framebuffer.height = desth; + surface->context->states.csc.framebuffer.cbufs[0] = pipe->winsys->surface_alloc(pipe->winsys); + pipe->winsys->surface_alloc_storage + ( + pipe->winsys, + surface->context->states.csc.framebuffer.cbufs[0], + destw, + desth, + PIPE_FORMAT_A8R8G8B8_UNORM, + PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE, + 0 + ); + } + + vlEndRender(surface->context); + + pipe->set_sampler_textures(pipe, 1, &surface->texture); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); + pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + pipe->winsys->flush_frontbuffer + ( + pipe->winsys, + surface->context->states.csc.framebuffer.cbufs[0], + &drawable + ); + + vlBeginRender(surface->context); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h new file mode 100644 index 00000000000..9f56b77e1e7 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_surface.h @@ -0,0 +1,81 @@ +#ifndef vl_surface_h +#define vl_surface_h + +#include +#include "vl_types.h" + +struct pipe_texture; + +struct VL_SURFACE +{ + struct VL_CONTEXT *context; + unsigned int width; + unsigned int height; + enum VL_FORMAT format; + struct pipe_texture *texture; +}; + +int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface); + +int vlDestroySurface(struct VL_SURFACE *surface); + +int vlRenderIMacroBlock +( + enum VL_PICTURE picture_type, + enum VL_FIELD_ORDER field_order, + unsigned int mbx, + unsigned int mby, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + short *blocks, + struct VL_SURFACE *surface +); + +int vlRenderPMacroBlock +( + enum VL_PICTURE picture_type, + enum VL_FIELD_ORDER field_order, + unsigned int mbx, + unsigned int mby, + enum VL_MC_TYPE mc_type, + struct VL_MOTION_VECTOR *motion_vector, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + short *blocks, + struct VL_SURFACE *ref_surface, + struct VL_SURFACE *surface +); + +int vlRenderBMacroBlock +( + enum VL_PICTURE picture_type, + enum VL_FIELD_ORDER field_order, + unsigned int mbx, + unsigned int mby, + enum VL_MC_TYPE mc_type, + struct VL_MOTION_VECTOR *motion_vector, + unsigned int coded_block_pattern, + enum VL_DCT_TYPE dct_type, + short *blocks, + struct VL_SURFACE *past_surface, + struct VL_SURFACE *future_surface, + struct VL_SURFACE *surface +); + +int vlPutSurface +( + struct VL_SURFACE *surface, + Drawable drawable, + unsigned int srcx, + unsigned int srcy, + unsigned int srcw, + unsigned int srch, + unsigned int destx, + unsigned int desty, + unsigned int destw, + unsigned int desth, + enum VL_PICTURE picture_type +); + +#endif + diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h new file mode 100644 index 00000000000..7040b745036 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_types.h @@ -0,0 +1,88 @@ +#ifndef vl_types_h +#define vl_types_h + +enum VL_FORMAT +{ + VL_FORMAT_YCBCR_420, + VL_FORMAT_YCBCR_422, + VL_FORMAT_YCBCR_444 +}; + +enum VL_PICTURE +{ + VL_TOP_FIELD, + VL_BOTTOM_FIELD, + VL_FRAME_PICTURE +}; + +enum VL_FIELD_ORDER +{ + VL_FIELD_FIRST, + VL_FIELD_SECOND +}; + +enum VL_DCT_TYPE +{ + VL_DCT_FIELD_CODED, + VL_DCT_FRAME_CODED +}; + +enum VL_SAMPLE_TYPE +{ + VL_FULL_SAMPLE, + VL_DIFFERENCE_SAMPLE +}; + +enum VL_MC_TYPE +{ + VL_FIELD_MC, + VL_FRAME_MC +}; + +struct VL_VERTEX4F +{ + float x, y, z, w; +}; + +struct VL_VERTEX2F +{ + float x, y; +}; + +struct VL_TEXCOORD2F +{ + float s, t; +}; + +struct VL_MC_VS_CONSTS +{ + struct VL_VERTEX4F scale; + struct VL_VERTEX4F mb_pos_trans; + struct VL_VERTEX4F mb_tc_trans[2]; +}; + +struct VL_MC_FS_CONSTS +{ + struct VL_VERTEX4F multiplier; + struct VL_VERTEX4F bias; +}; + +struct VL_CSC_FS_CONSTS +{ + struct VL_VERTEX4F bias; + float matrix[16]; +}; + +struct VL_MOTION_VECTOR +{ + struct + { + int x, y; + } top_field, bottom_field; +}; + +struct VL_CONTEXT; +struct VL_SURFACE; + +#endif + diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c new file mode 100644 index 00000000000..eb4c9085d6d --- /dev/null +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -0,0 +1,256 @@ +#include "xsp_winsys.h" +#include +#include +#include +#include +#include +#include + +/* pipe_winsys implementation */ + +struct xsp_pipe_winsys +{ + struct pipe_winsys base; + Display *display; + XImage fbimage; +}; + +struct xsp_buffer +{ + struct pipe_buffer base; + boolean is_user_buffer; + void *data; + void *mapped_data; +}; + +static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size) +{ + struct xsp_buffer *buffer; + + assert(pws); + + buffer = calloc(1, sizeof(struct xsp_buffer)); + buffer->base.refcount = 1; + buffer->base.alignment = alignment; + buffer->base.usage = usage; + buffer->base.size = size; + buffer->data = align_malloc(size, alignment); + + return (struct pipe_buffer*)buffer; +} + +static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size) +{ + struct xsp_buffer *buffer; + + assert(pws); + + buffer = calloc(1, sizeof(struct xsp_buffer)); + buffer->base.refcount = 1; + buffer->base.size = size; + buffer->is_user_buffer = TRUE; + buffer->data = data; + + return (struct pipe_buffer*)buffer; +} + +static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags) +{ + struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; + + assert(pws); + assert(buffer); + + xsp_buf->mapped_data = xsp_buf->data; + return xsp_buf->mapped_data; +} + +static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer) +{ + struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; + + assert(pws); + assert(buffer); + + xsp_buf->mapped_data = NULL; +} + +static void xsp_buffer_destroy(struct pipe_winsys *pws, struct pipe_buffer *buffer) +{ + struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; + + assert(pws); + assert(buffer); + + if (!xsp_buf->is_user_buffer) + align_free(xsp_buf->data); + + free(xsp_buf); +} + +static struct pipe_surface* xsp_surface_alloc(struct pipe_winsys *pws) +{ + struct pipe_surface *surface; + + assert(pws); + + surface = calloc(1, sizeof(struct pipe_surface)); + surface->refcount = 1; + surface->winsys = pws; + + return surface; +} + +static int xsp_surface_alloc_storage +( + struct pipe_winsys *pws, + struct pipe_surface *surface, + unsigned width, + unsigned height, + enum pipe_format format, + unsigned flags, + unsigned tex_usage +) +{ + const unsigned int ALIGNMENT = 1; + + assert(pws); + assert(surface); + + surface->width = width; + surface->height = height; + surface->format = format; + surface->cpp = pf_get_size(format); + surface->pitch = width; + surface->usage = flags; + surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->pitch * surface->cpp * height); + + return 0; +} + +static void xsp_surface_release(struct pipe_winsys *pws, struct pipe_surface **surface) +{ + struct pipe_surface *s; + + assert(pws); + assert(surface); + assert(*surface); + + s = *surface; + + s->refcount--; + + if (s->refcount == 0) + { + pipe_buffer_reference(pws, &s->buffer, NULL); + free(s); + } + + *surface = NULL; +} + +static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private) +{ + struct xsp_pipe_winsys *xsp_winsys; + + assert(pws); + assert(surface); + assert(context_private); + + xsp_winsys = (struct xsp_pipe_winsys*)pws; + + xsp_winsys->fbimage.width = surface->width; + xsp_winsys->fbimage.height = surface->height; + xsp_winsys->fbimage.bytes_per_line = surface->width * (xsp_winsys->fbimage.bits_per_pixel >> 3); + xsp_winsys->fbimage.data = pipe_surface_map(surface, 0); + + XPutImage + ( + xsp_winsys->display, + *(Drawable*)context_private, + XDefaultGC(xsp_winsys->display, XDefaultScreen(xsp_winsys->display)), + &xsp_winsys->fbimage, + 0, + 0, + 0, + 0, + surface->width, + surface->height + ); + XFlush(xsp_winsys->display); + pipe_surface_unmap(surface); +} + +static const char* xsp_get_name(struct pipe_winsys *pws) +{ + assert(pws); + return "X11 SoftPipe"; +} + +/* softpipe_winsys implementation */ + +static boolean xsp_is_format_supported(struct softpipe_winsys *spws, enum pipe_format format) +{ + assert(spws); + + /* TODO: Test that 'format' is equal to our output window's format */ + return TRUE; +} + +/* Show starts here */ + +struct pipe_context* create_pipe_context(Display *display) +{ + struct xsp_pipe_winsys *xsp_winsys; + struct softpipe_winsys *sp_winsys; + struct pipe_screen *p_screen; + struct pipe_context *p_context; + + assert(display); + + xsp_winsys = calloc(1, sizeof(struct xsp_pipe_winsys)); + xsp_winsys->base.buffer_create = xsp_buffer_create; + xsp_winsys->base.user_buffer_create = xsp_user_buffer_create; + xsp_winsys->base.buffer_map = xsp_buffer_map; + xsp_winsys->base.buffer_unmap = xsp_buffer_unmap; + xsp_winsys->base.buffer_destroy = xsp_buffer_destroy; + xsp_winsys->base.surface_alloc = xsp_surface_alloc; + xsp_winsys->base.surface_alloc_storage = xsp_surface_alloc_storage; + xsp_winsys->base.surface_release = xsp_surface_release; + xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer; + xsp_winsys->base.get_name = xsp_get_name; + xsp_winsys->display = display; + + { + /* XXX: Can't use the returned XImage* directly, + since we don't have control over winsys destruction + and we wouldn't be able to free it */ + XImage *template = XCreateImage + ( + display, + XDefaultVisual(display, XDefaultScreen(display)), + XDefaultDepth(display, XDefaultScreen(display)), + ZPixmap, + 0, + NULL, + 0, /* Don't know the width and height until flush_frontbuffer */ + 0, + 32, + 0 + ); + + memcpy(&xsp_winsys->fbimage, template, sizeof(XImage)); + XInitImage(&xsp_winsys->fbimage); + + XDestroyImage(template); + } + + sp_winsys = calloc(1, sizeof(struct softpipe_winsys)); + sp_winsys->is_format_supported = xsp_is_format_supported; + + p_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); + p_context = softpipe_create(p_screen, (struct pipe_winsys*)xsp_winsys, sp_winsys); + + return p_context; +} + diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.h b/src/gallium/winsys/g3dvl/xsp_winsys.h new file mode 100644 index 00000000000..47a6ac6e251 --- /dev/null +++ b/src/gallium/winsys/g3dvl/xsp_winsys.h @@ -0,0 +1,11 @@ +#ifndef xsp_winsys_h +#define xsp_winsys_h + +#include + +struct pipe_context; + +struct pipe_context* create_pipe_context(Display *display); + +#endif + diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile new file mode 100644 index 00000000000..b2f41e176d4 --- /dev/null +++ b/src/libXvMC/Makefile @@ -0,0 +1,33 @@ +TARGET = libXvMCg3dvl.so +SONAME = libXvMCg3dvl.so.1 +GALLIUMDIR = ../gallium +OBJECTS = block.o surface.o context.o subpicture.o attributes.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + +CFLAGS += -g -fPIC -Wall -Werror \ + -I${GALLIUMDIR}/state_trackers/g3dvl \ + -I${GALLIUMDIR}/winsys/g3dvl \ + -I${GALLIUMDIR}/include \ + -I${GALLIUMDIR}/auxiliary \ + -I${GALLIUMDIR}/drivers +LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ + -L${GALLIUMDIR}/drivers/softpipe \ + -L${GALLIUMDIR}/auxiliary/tgsi \ + -L${GALLIUMDIR}/auxiliary/draw \ + -L${GALLIUMDIR}/auxiliary/translate \ + -L${GALLIUMDIR}/auxiliary/cso_cache \ + -L${GALLIUMDIR}/auxiliary/util \ + -L${GALLIUMDIR}/auxiliary/rtasm +LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil + +############################################# + +.PHONY = all clean + +all: ${TARGET} + +${TARGET}: ${OBJECTS} + $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ $^ ${LIBS} + +clean: + rm -rf ${OBJECTS} ${TARGET} + diff --git a/src/libXvMC/attributes.c b/src/libXvMC/attributes.c new file mode 100644 index 00000000000..674524b8b82 --- /dev/null +++ b/src/libXvMC/attributes.c @@ -0,0 +1,20 @@ +#include +#include +#include +#include + +XvAttribute* XvMCQueryAttributes(Display *display, XvMCContext *context, int *number) +{ + return NULL; +} + +Status XvMCSetAttribute(Display *display, XvMCContext *context, Atom attribute, int value) +{ + return BadImplementation; +} + +Status XvMCGetAttribute(Display *display, XvMCContext *context, Atom attribute, int *value) +{ + return BadImplementation; +} + diff --git a/src/libXvMC/block.c b/src/libXvMC/block.c new file mode 100644 index 00000000000..b56348d4645 --- /dev/null +++ b/src/libXvMC/block.c @@ -0,0 +1,85 @@ +#include +#include +#include +#include +#include + +/* + * XvMC defines 64 element blocks (8x8 elements). + * Elements are 8 bits when they represent color values, + * 9 bits when they reprecent DCT coefficients, we + * store them in 2 bytes in either case. DCT coefficients + * can be signed or unsigned, at our option. + */ +#define BLOCK_SIZE (64 * 2) + +Status XvMCCreateBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks) +{ + struct vl_context *vl_ctx; + + assert(display); + + if (!context) + return XvMCBadContext; + if (num_blocks == 0) + return BadValue; + + assert(blocks); + + vl_ctx = context->privData; + assert(display == vl_ctx->display); + + blocks->context_id = context->context_id; + blocks->num_blocks = num_blocks; + blocks->blocks = malloc(BLOCK_SIZE * num_blocks); + /* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */ + blocks->privData = display; + + return Success; +} + +Status XvMCDestroyBlocks(Display *display, XvMCBlockArray *blocks) +{ + assert(display); + assert(blocks); + assert(display == blocks->privData); + free(blocks->blocks); + + return Success; +} + +Status XvMCCreateMacroBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks) +{ + struct vl_context *vl_ctx; + + assert(display); + + if (!context) + return XvMCBadContext; + if (num_blocks == 0) + return BadValue; + + assert(blocks); + + vl_ctx = context->privData; + assert(display == vl_ctx->display); + + blocks->context_id = context->context_id; + blocks->num_blocks = num_blocks; + blocks->macro_blocks = malloc(sizeof(XvMCMacroBlock) * num_blocks); + /* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */ + blocks->privData = display; + + return Success; +} + +Status XvMCDestroyMacroBlocks(Display *display, XvMCMacroBlockArray *blocks) +{ + assert(display); + assert(blocks); + assert(display == blocks->privData); + free(blocks->macro_blocks); + + return Success; +} + diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c new file mode 100644 index 00000000000..c835a6acf7a --- /dev/null +++ b/src/libXvMC/context.c @@ -0,0 +1,154 @@ +#include +#include +#include +#include +#include + +static Status Validate(Display *display, XvPortID port, int surface_type_id, unsigned int width, unsigned int height, int flags, int *chroma_format) +{ + unsigned int found_port = 0; + unsigned int found_surface = 0; + XvAdaptorInfo *adaptor_info; + unsigned int num_adaptors; + int num_types; + unsigned int max_width, max_height; + Status ret; + unsigned int i, j, k; + + assert(display && chroma_format); + + ret = XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info); + if (ret != Success) + return ret; + + /* Scan through all adaptors looking for this port and surface */ + for (i = 0; i < num_adaptors && !found_port; ++i) + { + /* Scan through all ports of this adaptor looking for our port */ + for (j = 0; j < adaptor_info[i].num_ports && !found_port; ++j) + { + /* If this is our port, scan through all its surfaces looking for our surface */ + if (adaptor_info[i].base_id + j == port) + { + XvMCSurfaceInfo *surface_info; + + found_port = 1; + surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types); + + if (surface_info) + { + for (k = 0; k < num_types && !found_surface; ++k) + { + if (surface_info[k].surface_type_id == surface_type_id) + { + found_surface = 1; + max_width = surface_info[k].max_width; + max_height = surface_info[k].max_height; + *chroma_format = surface_info[k].chroma_format; + } + } + + XFree(surface_info); + } + else + { + XvFreeAdaptorInfo(adaptor_info); + return BadAlloc; + } + } + } + } + + XvFreeAdaptorInfo(adaptor_info); + + if (!found_port) + return XvBadPort; + if (!found_surface) + return BadMatch; + if (width > max_width || height > max_height) + return BadValue; + if (flags != XVMC_DIRECT && flags != 0) + return BadValue; + + return Success; +} + +static enum VL_FORMAT FormatToVL(int xvmc_format) +{ + enum VL_FORMAT vl_format; + + switch (xvmc_format) + { + case XVMC_CHROMA_FORMAT_420: + { + vl_format = VL_FORMAT_YCBCR_420; + break; + } + case XVMC_CHROMA_FORMAT_422: + { + vl_format = VL_FORMAT_YCBCR_422; + break; + } + case XVMC_CHROMA_FORMAT_444: + { + vl_format = VL_FORMAT_YCBCR_444; + break; + } + default: + assert(0); + } + + return vl_format; +} + +Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, int width, int height, int flags, XvMCContext *context) +{ + int chroma_format; + Status ret; + struct VL_CONTEXT *vl_ctx; + struct pipe_context *pipe; + + assert(display); + + if (!context) + return XvMCBadContext; + + ret = Validate(display, port, surface_type_id, width, height, flags, &chroma_format); + if (ret != Success) + return ret; + + pipe = create_pipe_context(display); + + assert(pipe); + + vlCreateContext(display, pipe, width, height, FormatToVL(chroma_format), &vl_ctx); + + context->context_id = XAllocID(display); + context->surface_type_id = surface_type_id; + context->width = width; + context->height = height; + context->flags = flags; + context->port = port; + context->privData = vl_ctx; + + return Success; +} + +Status XvMCDestroyContext(Display *display, XvMCContext *context) +{ + struct VL_CONTEXT *vl_ctx; + + assert(display); + + if (!context) + return XvMCBadContext; + + vl_ctx = context->privData; + + assert(display == vl_ctx->display); + + vlDestroyContext(vl_ctx); + + return Success; +} + diff --git a/src/libXvMC/subpicture.c b/src/libXvMC/subpicture.c new file mode 100644 index 00000000000..38d93438334 --- /dev/null +++ b/src/libXvMC/subpicture.c @@ -0,0 +1,110 @@ +#include +#include +#include +#include + +Status XvMCCreateSubpicture +( + Display *display, + XvMCContext *context, + XvMCSubpicture *subpicture, + unsigned short width, + unsigned short height, + int xvimage_id +) +{ + return BadImplementation; +} + +Status XvMCClearSubpicture +( + Display *display, + XvMCSubpicture *subpicture, + short x, + short y, + unsigned short width, + unsigned short height, + unsigned int color +) +{ + return BadImplementation; +} + +Status XvMCCompositeSubpicture +( + Display *display, + XvMCSubpicture *subpicture, + XvImage *image, + short srcx, + short srcy, + unsigned short width, + unsigned short height, + short dstx, + short dsty +) +{ + return BadImplementation; +} + +Status XvMCDestroySubpicture(Display *display, XvMCSubpicture *subpicture) +{ + return BadImplementation; +} + +Status XvMCSetSubpicturePalette(Display *display, XvMCSubpicture *subpicture, unsigned char *palette) +{ + return BadImplementation; +} + +Status XvMCBlendSubpicture +( + Display *display, + XvMCSurface *target_surface, + XvMCSubpicture *subpicture, + short subx, + short suby, + unsigned short subw, + unsigned short subh, + short surfx, + short surfy, + unsigned short surfw, + unsigned short surfh +) +{ + return BadImplementation; +} + +Status XvMCBlendSubpicture2 +( + Display *display, + XvMCSurface *source_surface, + XvMCSurface *target_surface, + XvMCSubpicture *subpicture, + short subx, + short suby, + unsigned short subw, + unsigned short subh, + short surfx, + short surfy, + unsigned short surfw, + unsigned short surfh +) +{ + return BadImplementation; +} + +Status XvMCSyncSubpicture(Display *display, XvMCSubpicture *subpicture) +{ + return BadImplementation; +} + +Status XvMCFlushSubpicture(Display *display, XvMCSubpicture *subpicture) +{ + return BadImplementation; +} + +Status XvMCGetSubpictureStatus(Display *display, XvMCSubpicture *subpicture, int *status) +{ + return BadImplementation; +} + diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c new file mode 100644 index 00000000000..e4602d82042 --- /dev/null +++ b/src/libXvMC/surface.c @@ -0,0 +1,373 @@ +#include +#include +#include +#include +#include + +static enum VL_PICTURE PictureToVL(int xvmc_pic) +{ + enum VL_PICTURE vl_pic; + + switch (xvmc_pic) + { + case XVMC_TOP_FIELD: + { + vl_pic = VL_TOP_FIELD; + break; + } + case XVMC_BOTTOM_FIELD: + { + vl_pic = VL_BOTTOM_FIELD; + break; + } + case XVMC_FRAME_PICTURE: + { + vl_pic = VL_FRAME_PICTURE; + break; + } + default: + assert(0); + } + + return vl_pic; +} + +Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *surface) +{ + struct VL_CONTEXT *vl_ctx; + struct VL_SURFACE *vl_sfc; + + assert(display); + + if (!context) + return XvMCBadContext; + if (!surface) + return XvMCBadSurface; + + vl_ctx = context->privData; + + assert(display == vl_ctx->display); + + vlCreateSurface(vl_ctx, &vl_sfc); + + surface->surface_id = XAllocID(display); + surface->context_id = context->context_id; + surface->surface_type_id = context->surface_type_id; + surface->width = context->width; + surface->height = context->height; + surface->privData = vl_sfc; + + return Success; +} + +Status XvMCRenderSurface +( + Display *display, + XvMCContext *context, + unsigned int picture_structure, + XvMCSurface *target_surface, + XvMCSurface *past_surface, + XvMCSurface *future_surface, + unsigned int flags, + unsigned int num_macroblocks, + unsigned int first_macroblock, + XvMCMacroBlockArray *macroblocks, + XvMCBlockArray *blocks +) +{ + struct VL_CONTEXT *vl_ctx; + struct VL_SURFACE *target_vl_surface; + struct VL_SURFACE *past_vl_surface; + struct VL_SURFACE *future_vl_surface; + unsigned int i; + + assert(display); + + if (!context) + return XvMCBadContext; + if (!target_surface) + return XvMCBadSurface; + + if + ( + picture_structure != XVMC_TOP_FIELD && + picture_structure != XVMC_BOTTOM_FIELD && + picture_structure != XVMC_FRAME_PICTURE + ) + return BadValue; + if (future_surface && !past_surface) + return BadMatch; + + vl_ctx = context->privData; + + assert(display == vl_ctx->display); + + target_vl_surface = target_surface->privData; + past_vl_surface = past_surface ? past_surface->privData : NULL; + future_vl_surface = future_surface ? future_surface->privData : NULL; + + assert(vl_ctx == target_vl_surface->context); + assert(!past_vl_surface || vl_ctx == past_vl_surface->context); + assert(!future_vl_surface || vl_ctx == future_vl_surface->context); + + assert(macroblocks); + assert(blocks); + + assert(macroblocks->context_id == context->context_id); + assert(blocks->context_id == context->context_id); + + assert(flags == 0 || flags == XVMC_SECOND_FIELD); + + for (i = first_macroblock; i < first_macroblock + num_macroblocks; ++i) + if (macroblocks->macro_blocks[i].macroblock_type & XVMC_MB_TYPE_INTRA) + vlRenderIMacroBlock + ( + PictureToVL(picture_structure), + flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, + macroblocks->macro_blocks[i].x, + macroblocks->macro_blocks[i].y, + macroblocks->macro_blocks[i].coded_block_pattern, + macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, + blocks->blocks + (macroblocks->macro_blocks[i].index * 64), + target_vl_surface + ); + else if + ( + (macroblocks->macro_blocks[i].macroblock_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) + == XVMC_MB_TYPE_MOTION_FORWARD + ) + { + struct VL_MOTION_VECTOR motion_vector = + { + { + macroblocks->macro_blocks[i].PMV[0][0][0], + macroblocks->macro_blocks[i].PMV[0][0][1], + }, + { + macroblocks->macro_blocks[i].PMV[1][0][0], + macroblocks->macro_blocks[i].PMV[1][0][1], + } + }; + + vlRenderPMacroBlock + ( + PictureToVL(picture_structure), + flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, + macroblocks->macro_blocks[i].x, + macroblocks->macro_blocks[i].y, + macroblocks->macro_blocks[i].motion_type == XVMC_PREDICTION_FRAME ? VL_FIELD_MC : VL_FRAME_MC, + &motion_vector, + macroblocks->macro_blocks[i].coded_block_pattern, + macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, + blocks->blocks + (macroblocks->macro_blocks[i].index * 64), + past_vl_surface, + target_vl_surface + ); + } + else if + ( + (macroblocks->macro_blocks[i].macroblock_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) + == XVMC_MB_TYPE_MOTION_BACKWARD + ) + { + struct VL_MOTION_VECTOR motion_vector = + { + { + macroblocks->macro_blocks[i].PMV[0][1][0], + macroblocks->macro_blocks[i].PMV[0][1][1], + }, + { + macroblocks->macro_blocks[i].PMV[1][1][0], + macroblocks->macro_blocks[i].PMV[1][1][1], + } + }; + + vlRenderPMacroBlock + ( + PictureToVL(picture_structure), + flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, + macroblocks->macro_blocks[i].x, + macroblocks->macro_blocks[i].y, + macroblocks->macro_blocks[i].motion_type == XVMC_PREDICTION_FRAME ? VL_FIELD_MC : VL_FRAME_MC, + &motion_vector, + macroblocks->macro_blocks[i].coded_block_pattern, + macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, + blocks->blocks + (macroblocks->macro_blocks[i].index * 64), + future_vl_surface, + target_vl_surface + ); + } + else if + ( + (macroblocks->macro_blocks[i].macroblock_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) + == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD) + ) + { + struct VL_MOTION_VECTOR motion_vector[2] = + { + { + { + macroblocks->macro_blocks[i].PMV[0][0][0], + macroblocks->macro_blocks[i].PMV[0][0][1], + }, + { + macroblocks->macro_blocks[i].PMV[1][0][0], + macroblocks->macro_blocks[i].PMV[1][0][1], + } + }, + { + { + macroblocks->macro_blocks[i].PMV[0][1][0], + macroblocks->macro_blocks[i].PMV[0][1][1], + }, + { + macroblocks->macro_blocks[i].PMV[1][1][0], + macroblocks->macro_blocks[i].PMV[1][1][1], + } + } + }; + + vlRenderBMacroBlock + ( + PictureToVL(picture_structure), + flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, + macroblocks->macro_blocks[i].x, + macroblocks->macro_blocks[i].y, + macroblocks->macro_blocks[i].motion_type == XVMC_PREDICTION_FRAME ? VL_FIELD_MC : VL_FRAME_MC, + motion_vector, + macroblocks->macro_blocks[i].coded_block_pattern, + macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, + blocks->blocks + (macroblocks->macro_blocks[i].index * 64), + past_vl_surface, + future_vl_surface, + target_vl_surface + ); + } + else + fprintf(stderr, "Unrecognized macroblock\n"); + + return Success; +} + +Status XvMCFlushSurface(Display *display, XvMCSurface *surface) +{ + /* TODO: Check display & surface match */ + return BadImplementation; +} + +Status XvMCSyncSurface(Display *display, XvMCSurface *surface) +{ + /* TODO: Check display & surface match */ + return BadImplementation; +} + +Status XvMCPutSurface +( + Display *display, + XvMCSurface *surface, + Drawable drawable, + short srcx, + short srcy, + unsigned short srcw, + unsigned short srch, + short destx, + short desty, + unsigned short destw, + unsigned short desth, + int flags +) +{ + Window root; + int x, y; + unsigned int width, height; + unsigned int border_width; + unsigned int depth; + struct VL_SURFACE *vl_sfc; + + assert(display); + + if (!surface) + return XvMCBadSurface; + + if (XGetGeometry(display, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable) + return BadDrawable; + + assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE); + + /* TODO: Correct for negative srcx,srcy & destx,desty by clipping */ + + assert(srcx + srcw - 1 < surface->width); + assert(srcy + srch - 1 < surface->height); + assert(destx + destw - 1 < width); + assert(desty + desth - 1 < height); + + vl_sfc = surface->privData; + + vlPutSurface(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, PictureToVL(flags)); + + return Success; +} + +Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status) +{ + struct VL_CONTEXT *vl_ctx; + struct VL_SURFACE *vl_sfc; + + assert(display); + + if (!surface) + return XvMCBadSurface; + + assert(status); + + vl_sfc = surface->privData; + vl_ctx = vl_sfc->context; + + assert(display == vl_ctx->display); + + /* TODO */ + *status = 0; + + return BadImplementation; +} + +Status XvMCDestroySurface(Display *display, XvMCSurface *surface) +{ + struct VL_CONTEXT *vl_ctx; + struct VL_SURFACE *vl_sfc; + + assert(display); + + if (!surface) + return XvMCBadSurface; + + vl_sfc = surface->privData; + vl_ctx = vl_sfc->context; + + assert(display == vl_ctx->display); + + vlDestroySurface(vl_sfc); + + return Success; +} + +Status XvMCHideSurface(Display *display, XvMCSurface *surface) +{ + struct VL_CONTEXT *vl_ctx; + struct VL_SURFACE *vl_sfc; + + assert(display); + + if (!surface) + return XvMCBadSurface; + + vl_sfc = surface->privData; + vl_ctx = vl_sfc->context; + + assert(display == vl_ctx->display); + + /* No op, only for overlaid rendering */ + + return Success; +} + diff --git a/src/libXvMC/tests/.gitignore b/src/libXvMC/tests/.gitignore new file mode 100644 index 00000000000..eb1ef8a0761 --- /dev/null +++ b/src/libXvMC/tests/.gitignore @@ -0,0 +1,5 @@ +test_context +test_surface +test_blocks +test_rendering + diff --git a/src/libXvMC/tests/Makefile b/src/libXvMC/tests/Makefile new file mode 100644 index 00000000000..2cbc97e88ba --- /dev/null +++ b/src/libXvMC/tests/Makefile @@ -0,0 +1,25 @@ +CFLAGS += -g -Wall -Werror +LDFLAGS += +LIBS += -lXvMCW -lXvMC -lXv + +############################################# + +.PHONY = all clean + +all: test_context test_surface test_blocks test_rendering + +test_context: test_context.o testlib.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_surface: test_surface.o testlib.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_blocks: test_blocks.o testlib.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +test_rendering: test_rendering.o testlib.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + +clean: + rm -rf *.o test_context test_surface test_blocks test_rendering + diff --git a/src/libXvMC/tests/test_blocks.c b/src/libXvMC/tests/test_blocks.c new file mode 100644 index 00000000000..0b895ee7737 --- /dev/null +++ b/src/libXvMC/tests/test_blocks.c @@ -0,0 +1,95 @@ +#include +#include +#include "testlib.h" + +int main(int argc, char **argv) +{ + const unsigned int width = 16, height = 16; + const unsigned int min_required_blocks = 1, min_required_macroblocks = 1; + const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; + + Display *display; + XvPortID port_num; + int surface_type_id; + unsigned int is_overlay, intra_unsigned; + int colorkey; + XvMCContext context; + XvMCSurface surface; + XvMCBlockArray blocks = {0}; + XvMCMacroBlockArray macroblocks = {0}; + + display = XOpenDisplay(NULL); + + if (!GetPort + ( + display, + width, + height, + XVMC_CHROMA_FORMAT_420, + mc_types, + 2, + &port_num, + &surface_type_id, + &is_overlay, + &intra_unsigned + )) + { + XCloseDisplay(display); + error(1, 0, "Error, unable to find a good port.\n"); + } + + if (is_overlay) + { + Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); + XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); + } + + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); + assert(XvMCCreateSurface(display, &context, &surface) == Success); + + /* Test NULL context */ + assert(XvMCCreateBlocks(display, NULL, 1, &blocks) == XvMCBadContext); + /* Test 0 blocks */ + assert(XvMCCreateBlocks(display, &context, 0, &blocks) == BadValue); + /* Test too many blocks */ + /*assert(XvMCCreateBlocks(display, &context, 16384, &blocks) == BadAlloc);*/ + + /* Note: No XvMCBadBlock(s) error in spec */ + + /* Test valid params */ + assert(XvMCCreateBlocks(display, &context, min_required_blocks, &blocks) == Success); + /* Test context id assigned and correct */ + assert(blocks.context_id == context.context_id); + /* Test number of blocks assigned and correct */ + assert(blocks.num_blocks == min_required_blocks); + /* Test block pointer valid */ + assert(blocks.blocks != NULL); + /* Test NULL context */ + assert(XvMCCreateMacroBlocks(display, NULL, 1, ¯oblocks) == XvMCBadContext); + /* Test 0 macroblocks */ + assert(XvMCCreateMacroBlocks(display, &context, 0, ¯oblocks) == BadValue); + /* Test too many macroblocks */ + /*assert(XvMCCreateMacroBlocks(display, &context, 16384, ¯oblocks) == BadAlloc);*/ + + /* Note: No XvMCBadMacroBlock(s) error in spec */ + + /* Test valid params */ + assert(XvMCCreateMacroBlocks(display, &context, min_required_macroblocks, ¯oblocks) == Success); + /* Test context id assigned and correct */ + assert(macroblocks.context_id == context.context_id); + /* Test macroblock pointer valid */ + assert(macroblocks.macro_blocks != NULL); + /* Test valid params */ + assert(XvMCDestroyMacroBlocks(display, ¯oblocks) == Success); + /* Test valid params */ + assert(XvMCDestroyBlocks(display, &blocks) == Success); + + assert(XvMCDestroySurface(display, &surface) == Success); + assert(XvMCDestroyContext(display, &context) == Success); + + XvUngrabPort(display, port_num, CurrentTime); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/libXvMC/tests/test_context.c b/src/libXvMC/tests/test_context.c new file mode 100644 index 00000000000..22afb7ada4e --- /dev/null +++ b/src/libXvMC/tests/test_context.c @@ -0,0 +1,94 @@ +#include +#include +#include "testlib.h" + +int main(int argc, char **argv) +{ + const unsigned int width = 16, height = 16; + const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; + + Display *display; + XvPortID port_num; + int surface_type_id; + unsigned int is_overlay, intra_unsigned; + int colorkey; + XvMCContext context = {0}; + + display = XOpenDisplay(NULL); + + if (!GetPort + ( + display, + width, + height, + XVMC_CHROMA_FORMAT_420, + mc_types, + 2, + &port_num, + &surface_type_id, + &is_overlay, + &intra_unsigned + )) + { + XCloseDisplay(display); + error(1, 0, "Error, unable to find a good port.\n"); + } + + if (is_overlay) + { + Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); + XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); + } + + /* Note: XvMCBadContext not a valid return for XvMCCreateContext in the XvMC API, but openChrome driver returns it */ + /* Note: Nvidia binary driver segfaults on NULL context, halts with debug output on bad port */ + + /* Test NULL context */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, NULL) == XvMCBadContext); + /* Test invalid port */ + assert(XvMCCreateContext(display, port_num + 1, surface_type_id, width, height, XVMC_DIRECT, &context) == XvBadPort); + /* Test invalid surface */ + assert(XvMCCreateContext(display, port_num, surface_type_id + 1, width, height, XVMC_DIRECT, &context) == BadMatch); + /* Test invalid flags */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, -1, &context) == BadValue); + /* Test huge width */ + assert(XvMCCreateContext(display, port_num, surface_type_id, 16384, height, XVMC_DIRECT, &context) == BadValue); + /* Test huge height */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width, 16384, XVMC_DIRECT, &context) == BadValue); + /* Test huge width & height */ + assert(XvMCCreateContext(display, port_num, surface_type_id, 16384, 16384, XVMC_DIRECT, &context) == BadValue); + /* Test valid params */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); + /* Test context id assigned */ + assert(context.context_id != 0); + /* Test surface type id assigned and correct */ + assert(context.surface_type_id == surface_type_id); + /* Test width & height assigned and correct */ + assert(context.width == width && context.height == height); + /* Test port assigned and correct */ + assert(context.port == port_num); + /* Test flags assigned and correct */ + assert(context.flags == XVMC_DIRECT); + /* Test NULL context */ + assert(XvMCDestroyContext(display, NULL) == XvMCBadContext); + /* Test valid params */ + assert(XvMCDestroyContext(display, &context) == Success); + /* Test awkward but valid width */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width + 1, height, XVMC_DIRECT, &context) == Success); + assert(context.width >= width + 1); + assert(XvMCDestroyContext(display, &context) == Success); + /* Test awkward but valid height */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height + 1, XVMC_DIRECT, &context) == Success); + assert(context.height >= height + 1); + assert(XvMCDestroyContext(display, &context) == Success); + /* Test awkward but valid width & height */ + assert(XvMCCreateContext(display, port_num, surface_type_id, width + 1, height + 1, XVMC_DIRECT, &context) == Success); + assert(context.width >= width + 1 && context.height >= height + 1); + assert(XvMCDestroyContext(display, &context) == Success); + + XvUngrabPort(display, port_num, CurrentTime); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/libXvMC/tests/test_rendering.c b/src/libXvMC/tests/test_rendering.c new file mode 100644 index 00000000000..1914b1fb628 --- /dev/null +++ b/src/libXvMC/tests/test_rendering.c @@ -0,0 +1,153 @@ +#include +#include +#include +#include "testlib.h" + +int main(int argc, char **argv) +{ + const unsigned int width = 32, height = 32; + const unsigned int mwidth = width / 16, mheight = height / 16; + const unsigned int num_macroblocks = mwidth * mheight; + const unsigned int num_blocks = num_macroblocks * 6; + const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; + + int quit = 0; + Display *display; + Window root, window; + Pixmap framebuffer; + XEvent event; + XvPortID port_num; + int surface_type_id; + unsigned int is_overlay, intra_unsigned; + int colorkey; + XvMCContext context; + XvMCSurface surface; + XvMCBlockArray blocks; + XvMCMacroBlockArray macroblocks; + unsigned int b, x, y; + + display = XOpenDisplay(NULL); + + if (!GetPort + ( + display, + width, + height, + XVMC_CHROMA_FORMAT_420, + mc_types, + 2, + &port_num, + &surface_type_id, + &is_overlay, + &intra_unsigned + )) + { + XCloseDisplay(display); + error(1, 0, "Error, unable to find a good port.\n"); + } + + if (is_overlay) + { + Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); + XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); + } + + root = XDefaultRootWindow(display); + window = XCreateSimpleWindow(display, root, 0, 0, width, height, 0, 0, colorkey); + framebuffer = XCreatePixmap(display, root, width, height, 24); + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); + assert(XvMCCreateSurface(display, &context, &surface) == Success); + assert(XvMCCreateBlocks(display, &context, num_blocks, &blocks) == Success); + assert(XvMCCreateMacroBlocks(display, &context, num_macroblocks, ¯oblocks) == Success); + + for (b = 0; b < 6; ++b) + { + for (y = 0; y < 8; ++y) + { + for (x = 0; x < 8; ++x) + { + blocks.blocks[b * 64 + y * 8 + x] = 0xFFFF; + } + } + } + + for (y = 0; y < mheight; ++y) + { + for (x = 0; x < mwidth; ++x) + { + macroblocks.macro_blocks[y * mwidth + x].x = x; + macroblocks.macro_blocks[y * mwidth + x].y = y; + macroblocks.macro_blocks[y * mwidth + x].index = (y * mwidth + x) * 6; + macroblocks.macro_blocks[y * mwidth + x].macroblock_type = XVMC_MB_TYPE_INTRA; + macroblocks.macro_blocks[y * mwidth + x].coded_block_pattern = 0x3F; + macroblocks.macro_blocks[y * mwidth + x].dct_type = XVMC_DCT_TYPE_FRAME; + } + } + + /* Test NULL context */ + assert(XvMCRenderSurface(display, NULL, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, 1, 0, ¯oblocks, &blocks) == XvMCBadContext); + /* Test NULL surface */ + assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, NULL, NULL, NULL, 0, 1, 0, ¯oblocks, &blocks) == XvMCBadSurface); + /* Test bad picture structure */ + assert(XvMCRenderSurface(display, &context, 0, &surface, NULL, NULL, 0, 1, 0, ¯oblocks, &blocks) == BadValue); + /* Test valid params */ + assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, num_macroblocks, 0, ¯oblocks, &blocks) == Success); + + /* Test NULL surface */ + assert(XvMCPutSurface(display, NULL, window, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == XvMCBadSurface); + /* Test bad window */ + /* X halts with a bad drawable for some reason, doesn't return BadDrawable as expected */ + /*assert(XvMCPutSurface(display, &surface, 0, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == BadDrawable);*/ + /* Test valid params */ + assert(XvMCPutSurface(display, &surface, framebuffer, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == Success); + + puts("Press any key to continue..."); + + while (!quit) + { + XNextEvent(display, &event); + switch (event.type) + { + case Expose: + { + XCopyArea + ( + display, + framebuffer, + window, + XDefaultGC(display, XDefaultScreen(display)), + 0, + 0, + width, + height, + 0, + 0 + ); + break; + } + case KeyPress: + { + quit = 1; + break; + } + } + } + + assert(XvMCDestroyBlocks(display, &blocks) == Success); + assert(XvMCDestroyMacroBlocks(display, ¯oblocks) == Success); + assert(XvMCDestroySurface(display, &surface) == Success); + assert(XvMCDestroyContext(display, &context) == Success); + + XFreePixmap(display, framebuffer); + XvUngrabPort(display, port_num, CurrentTime); + XDestroyWindow(display, window); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/libXvMC/tests/test_surface.c b/src/libXvMC/tests/test_surface.c new file mode 100644 index 00000000000..25ebdcc4fcf --- /dev/null +++ b/src/libXvMC/tests/test_surface.c @@ -0,0 +1,72 @@ +#include +#include +#include "testlib.h" + +int main(int argc, char **argv) +{ + const unsigned int width = 16, height = 16; + const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; + + Display *display; + XvPortID port_num; + int surface_type_id; + unsigned int is_overlay, intra_unsigned; + int colorkey; + XvMCContext context; + XvMCSurface surface = {0}; + + display = XOpenDisplay(NULL); + + if (!GetPort + ( + display, + width, + height, + XVMC_CHROMA_FORMAT_420, + mc_types, + 2, + &port_num, + &surface_type_id, + &is_overlay, + &intra_unsigned + )) + { + XCloseDisplay(display); + error(1, 0, "Error, unable to find a good port.\n"); + } + + if (is_overlay) + { + Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); + XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); + } + + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); + + /* Test NULL context */ + assert(XvMCCreateSurface(display, NULL, &surface) == XvMCBadContext); + /* Test NULL surface */ + assert(XvMCCreateSurface(display, &context, NULL) == XvMCBadSurface); + /* Test valid params */ + assert(XvMCCreateSurface(display, &context, &surface) == Success); + /* Test surface id assigned */ + assert(surface.surface_id != 0); + /* Test context id assigned and correct */ + assert(surface.context_id == context.context_id); + /* Test surface type id assigned and correct */ + assert(surface.surface_type_id == surface_type_id); + /* Test width & height assigned and correct */ + assert(surface.width == width && surface.height == height); + /* Test valid params */ + assert(XvMCDestroySurface(display, &surface) == Success); + /* Test NULL surface */ + assert(XvMCDestroySurface(display, NULL) == XvMCBadSurface); + + assert(XvMCDestroyContext(display, &context) == Success); + + XvUngrabPort(display, port_num, CurrentTime); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/libXvMC/tests/testlib.c b/src/libXvMC/tests/testlib.c new file mode 100644 index 00000000000..8672aa99998 --- /dev/null +++ b/src/libXvMC/tests/testlib.c @@ -0,0 +1,87 @@ +#include "testlib.h" +#include + +/* +void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line) +{ + fputs(doc_string, stderr); + if (!pred) + fprintf(stderr, " FAIL!\n\t\"%s\" at %s:%u\n", pred_string, file, line); + else + fputs(" PASS!\n", stderr); +} +*/ + +int GetPort +( + Display *display, + unsigned int width, + unsigned int height, + unsigned int chroma_format, + const unsigned int *mc_types, + unsigned int num_mc_types, + XvPortID *port_id, + int *surface_type_id, + unsigned int *is_overlay, + unsigned int *intra_unsigned +) +{ + unsigned int found_port = 0; + XvAdaptorInfo *adaptor_info; + unsigned int num_adaptors; + int num_types; + int ev_base, err_base; + unsigned int i, j, k, l; + + if (!XvMCQueryExtension(display, &ev_base, &err_base)) + return 0; + if (XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info) != Success) + return 0; + + for (i = 0; i < num_adaptors && !found_port; ++i) + { + if (adaptor_info[i].type & XvImageMask) + { + XvMCSurfaceInfo *surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types); + + if (surface_info) + { + for (j = 0; j < num_types && !found_port; ++j) + { + if + ( + surface_info[j].chroma_format == chroma_format && + surface_info[j].max_width >= width && + surface_info[j].max_height >= height + ) + { + for (k = 0; k < num_mc_types && !found_port; ++k) + { + if (surface_info[j].mc_type == mc_types[k]) + { + for (l = 0; l < adaptor_info[i].num_ports && !found_port; ++l) + { + if (XvGrabPort(display, adaptor_info[i].base_id + l, CurrentTime) == Success) + { + *port_id = adaptor_info[i].base_id + l; + *surface_type_id = surface_info[j].surface_type_id; + *is_overlay = surface_info[j].flags & XVMC_OVERLAID_SURFACE; + *intra_unsigned = surface_info[j].flags & XVMC_INTRA_UNSIGNED; + found_port = 1; + } + } + } + } + } + } + + XFree(surface_info); + } + } + } + + XvFreeAdaptorInfo(adaptor_info); + + return found_port; +} + diff --git a/src/libXvMC/tests/testlib.h b/src/libXvMC/tests/testlib.h new file mode 100644 index 00000000000..c73845807ec --- /dev/null +++ b/src/libXvMC/tests/testlib.h @@ -0,0 +1,38 @@ +#ifndef testlib_h +#define testlib_h + +/* +#define TEST(pred, doc) test(pred, #pred, doc, __FILE__, __LINE__) + +void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line); +*/ + +#include +#include + +/* + * display: IN A valid X display + * width, height: IN Surface size that the port must display + * chroma_format: IN Chroma format that the port must display + * mc_types, num_mc_types: IN List of MC types that the port must support, first port that matches the first mc_type will be returned + * port_id: OUT Your port's ID + * surface_type_id: OUT Your port's surface ID + * is_overlay: OUT If 1, port uses overlay surfaces, you need to set a colorkey + * intra_unsigned: OUT If 1, port uses unsigned values for intra-coded blocks + */ +int GetPort +( + Display *display, + unsigned int width, + unsigned int height, + unsigned int chroma_format, + const unsigned int *mc_types, + unsigned int num_mc_types, + XvPortID *port_id, + int *surface_type_id, + unsigned int *is_overlay, + unsigned int *intra_unsigned +); + +#endif + -- cgit v1.2.3 From f5a3768c4e7733a11ad0421e3e4b84e6994af0e0 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 8 Jun 2008 14:34:41 -0400 Subject: g3dvl: Reduce number of input vert streams by copying, reusing in shaders. Reduce number of input vertex streams by using same texcoord stream for chroma textures, reusing pos stream when calculating texcoords for P, B macroblocks. --- src/gallium/state_trackers/g3dvl/vl_context.c | 175 ++++++++++---------------- src/gallium/state_trackers/g3dvl/vl_context.h | 4 +- src/gallium/state_trackers/g3dvl/vl_data.c | 8 -- src/gallium/state_trackers/g3dvl/vl_data.h | 1 - src/gallium/state_trackers/g3dvl/vl_surface.c | 6 - 5 files changed, 66 insertions(+), 128 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 7193f7cceaf..2d1d5434955 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -32,15 +32,14 @@ static int vlDestroyIDCT(struct VL_CONTEXT *context) static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 50; - const unsigned int num_attribs = 4; - const unsigned int semantic_names[4] = + const unsigned int num_attribs = 3; + const unsigned int semantic_names[3] = { TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC }; - const unsigned int semantic_indexes[4] = {0, 1, 2, 3}; + const unsigned int semantic_indexes[3] = {0, 1, 2}; const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; @@ -173,7 +172,6 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) /* mov o1, i1 ; Move texcoords to output mov o2, i2 - mov o3, i3 */ for (i = 1; i < num_attribs; ++i) { @@ -251,7 +249,7 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) ti = 3; /* Declare inputs (texcoords) */ - for (i = 0; i < 3; ++i) + for (i = 0; i < 2; ++i) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; @@ -306,7 +304,7 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) /* tex2d o0.x, i0, s0 ; Read texel from luma texture into .x channel tex2d o0.y, i1, s1 ; Read texel from chroma Cb texture into .y channel - tex2d o0.z, i2, s2 ; Read texel from chroma Cr texture into .z channel + tex2d o0.z, i1, s2 ; Read texel from chroma Cr texture into .z channel */ for (i = 0; i < 3; ++i) { @@ -319,7 +317,7 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) inst.Instruction.NumSrcRegs = 2; inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; inst.FullSrcRegisters[1].SrcRegister.Index = i; ti += tgsi_build_full_instruction @@ -354,16 +352,23 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int num_attribs = 5; - const unsigned int semantic_names[5] = + const unsigned int num_input_attribs = 3; + const unsigned int num_output_attribs = 4; + const unsigned int input_semantic_names[3] = { TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC + }; + const unsigned int output_semantic_names[4] = + { + TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC }; - const unsigned int semantic_indexes[5] = {0, 1, 2, 3, 4}; + const unsigned int input_semantic_indexes[3] = {0, 1, 2}; + const unsigned int output_semantic_indexes[4] = {0, 1, 2, 3}; const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; @@ -398,14 +403,14 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) ti = 3; /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + for (i = 0; i < num_input_attribs; i++) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.Semantic.SemanticName = input_semantic_names[i]; + decl.Semantic.SemanticIndex = input_semantic_indexes[i]; decl.u.DeclarationRange.First = i; decl.u.DeclarationRange.Last = i; @@ -438,13 +443,13 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) ); /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + for (i = 0; i < num_output_attribs; i++) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_OUTPUT; decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.Semantic.SemanticName = output_semantic_names[i]; + decl.Semantic.SemanticIndex = output_semantic_indexes[i]; decl.u.DeclarationRange.First = i; decl.u.DeclarationRange.Last = i; ti += tgsi_build_full_declaration @@ -497,9 +502,8 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) /* mov o1, i1 ; Move luma & chroma texcoords to output mov o2, i2 - mov o3, i3 */ - for (i = 1; i < num_attribs - 1; ++i) + for (i = 1; i < num_output_attribs - 1; ++i) { inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_MOV; @@ -517,32 +521,13 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) max_tokens - ti ); } - - /* mul t0, i4, c0 ; Scale normalized coords to window coords */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); - /* add o4, t0, c2 ; Translate texcoords into position */ + /* add o3, t0, c2 ; Translate texcoords into position */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_ADD; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 4; + inst.FullDstRegisters[0].DstRegister.Index = 3; inst.Instruction.NumSrcRegs = 2; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; inst.FullSrcRegisters[0].SrcRegister.Index = 0; @@ -613,7 +598,7 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) ti = 3; /* Declare inputs (texcoords) */ - for (i = 0; i < 4; ++i) + for (i = 0; i < 3; ++i) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; @@ -688,7 +673,7 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) mov t1.x, t0.w ; Move high part from .w channel to .x tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels mov t1.y, t0.w ; Move high part from .w channel to .y - tex2d t0.zw, i2, s2 ; Read texel from chroma Cr texture into .z and .w channels + tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels mov t1.z, t0.w ; Move high part from .w channel to .z */ for (i = 0; i < 3; ++i) @@ -702,7 +687,7 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) inst.Instruction.NumSrcRegs = 2; inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; inst.FullSrcRegisters[1].SrcRegister.Index = i; ti += tgsi_build_full_instruction @@ -792,7 +777,7 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) max_tokens - ti ); - /* tex2d t1, i3, s3 ; Read texel from ref macroblock */ + /* tex2d t1, i2, s3 ; Read texel from ref macroblock */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_TEX; inst.Instruction.NumDstRegs = 1; @@ -801,7 +786,7 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) inst.Instruction.NumSrcRegs = 2; inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[0].SrcRegister.Index = 2; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; inst.FullSrcRegisters[1].SrcRegister.Index = 3; ti += tgsi_build_full_instruction @@ -854,17 +839,24 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int num_attribs = 6; - const unsigned int semantic_names[6] = + const unsigned int num_input_attribs = 3; + const unsigned int num_output_attribs = 5; + const unsigned int input_semantic_names[3] = { TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC + }; + const unsigned int output_semantic_names[5] = + { + TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC }; - const unsigned int semantic_indexes[6] = {0, 1, 2, 3, 4, 5}; + const unsigned int input_semantic_indexes[3] = {0, 1, 2}; + const unsigned int output_semantic_indexes[5] = {0, 1, 2, 3, 4}; const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; @@ -899,14 +891,14 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) ti = 3; /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + for (i = 0; i < num_input_attribs; i++) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.Semantic.SemanticName = input_semantic_names[i]; + decl.Semantic.SemanticIndex = input_semantic_indexes[i]; decl.u.DeclarationRange.First = i; decl.u.DeclarationRange.Last = i; @@ -940,13 +932,13 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) ); /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + for (i = 0; i < num_output_attribs; i++) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_OUTPUT; decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; + decl.Semantic.SemanticName = output_semantic_names[i]; + decl.Semantic.SemanticIndex = output_semantic_indexes[i]; decl.u.DeclarationRange.First = i; decl.u.DeclarationRange.Last = i; ti += tgsi_build_full_declaration @@ -999,9 +991,8 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) /* mov o1, i1 ; Move luma & chroma texcoords to output mov o2, i2 - mov o3, i3 */ - for (i = 1; i < num_attribs - 1; ++i) + for (i = 1; i < num_output_attribs - 2; ++i) { inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_MOV; @@ -1020,38 +1011,18 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) ); } - /* mul t0, i4, c0 ; Scale normalized coords to window coords - add o4, t0, c2 ; Translate texcoords into position - mul t1, i5, c0 ; Repeat for the future surface - add o5, t1, c3 */ + /* add o3, t0, c2 ; Translate past surface texcoords into position + add o4, t0, c3 ; Repeat for future surface texcoords */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i + 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); - inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_ADD; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i + 4; + inst.FullDstRegisters[0].DstRegister.Index = i + 3; inst.Instruction.NumSrcRegs = 2; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; inst.FullSrcRegisters[1].SrcRegister.Index = i + 2; ti += tgsi_build_full_instruction @@ -1120,7 +1091,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) ti = 3; /* Declare inputs (texcoords) */ - for (i = 0; i < 5; ++i) + for (i = 0; i < 4; ++i) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; @@ -1195,7 +1166,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) mov t1.x, t0.w ; Move high part from .w channel to .x tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels mov t1.y, t0.w ; Move high part from .w channel to .y - tex2d t0.zw, i2, s2 ; Read texel from chroma Cr texture into .z and .w channels + tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels mov t1.z, t0.w ; Move high part from .w channel to .z */ for (i = 0; i < 3; ++i) @@ -1209,7 +1180,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) inst.Instruction.NumSrcRegs = 2; inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; + inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; inst.FullSrcRegisters[1].SrcRegister.Index = i; ti += tgsi_build_full_instruction @@ -1299,8 +1270,8 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) max_tokens - ti ); - /* tex2d t1, i3, s3 ; Read texel from past macroblock - tex2d t2, i4, s4 ; Read texel from future macroblock */ + /* tex2d t1, i2, s3 ; Read texel from past macroblock + tex2d t2, i3, s4 ; Read texel from future macroblock */ for (i = 0; i < 2; ++i) { inst = tgsi_default_full_instruction(); @@ -1311,7 +1282,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) inst.Instruction.NumSrcRegs = 2; inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i + 3; + inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; ti += tgsi_build_full_instruction @@ -1409,7 +1380,7 @@ int vlCreateDataBufsMC(struct VL_CONTEXT *context) /* Create our texcoord buffers and texcoord buffer elements */ /* TODO: Should be able to use 1 texcoord buf for chroma textures, 1 buf for ref surfaces */ - for (i = 1; i < 6; ++i) + for (i = 1; i < 3; ++i) { context->states.mc.vertex_bufs[i].pitch = sizeof(struct VL_TEXCOORD2F); context->states.mc.vertex_bufs[i].max_index = 23; @@ -1448,26 +1419,8 @@ int vlCreateDataBufsMC(struct VL_CONTEXT *context) vl_chroma_420_texcoords, sizeof(struct VL_TEXCOORD2F) * 24 ); - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[3].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_chroma_420_texcoords, - sizeof(struct VL_TEXCOORD2F) * 24 - ); - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[4].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_ref_surface_texcoords, - sizeof(struct VL_TEXCOORD2F) * 24 - ); - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[5].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_ref_surface_texcoords, - sizeof(struct VL_TEXCOORD2F) * 24 - ); - for (i = 0; i < 6; ++i) + for (i = 0; i < 3; ++i) pipe->winsys->buffer_unmap(pipe->winsys, context->states.mc.vertex_bufs[i].buffer); /* Create our constant buffer */ @@ -1599,12 +1552,10 @@ static int vlDestroyMC(struct VL_CONTEXT *context) assert(context); for (i = 0; i < 5; ++i) - { context->pipe->delete_sampler_state(context->pipe, context->states.mc.samplers[i]); - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vertex_bufs[i].buffer); - } - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vertex_bufs[5].buffer); + for (i = 0; i < 3; ++i) + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vertex_bufs[i].buffer); /* Textures 3 & 4 are not created directly, no need to release them here */ for (i = 0; i < 3; ++i) @@ -2262,7 +2213,9 @@ int vlBeginRender(struct VL_CONTEXT *context) pipe = context->pipe; /* Frame buffer set in vlRender*Macroblock() */ - /* Shaders, samplers, textures, VBs, VB elements set in vlRender*Macroblock() */ + /* Shaders, samplers, textures set in vlRender*Macroblock() */ + pipe->set_vertex_buffers(pipe, 3, context->states.mc.vertex_bufs); + pipe->set_vertex_elements(pipe, 3, context->states.mc.vertex_buf_elems); pipe->set_viewport_state(pipe, &context->states.mc.viewport); pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &context->states.mc.vs_const_buf); pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &context->states.mc.fs_const_buf); diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h index 0aeba184ccb..f26a4c5b6a7 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.h +++ b/src/gallium/state_trackers/g3dvl/vl_context.h @@ -36,8 +36,8 @@ struct VL_CONTEXT struct pipe_texture *textures[5]; struct pipe_shader_state *i_vs, *p_vs, *b_vs; struct pipe_shader_state *i_fs, *p_fs, *b_fs; - struct pipe_vertex_buffer vertex_bufs[6]; - struct pipe_vertex_element vertex_buf_elems[6]; + struct pipe_vertex_buffer vertex_bufs[3]; + struct pipe_vertex_element vertex_buf_elems[3]; struct pipe_constant_buffer vs_const_buf, fs_const_buf; } mc; diff --git a/src/gallium/state_trackers/g3dvl/vl_data.c b/src/gallium/state_trackers/g3dvl/vl_data.c index c04163276d8..27893aee95e 100644 --- a/src/gallium/state_trackers/g3dvl/vl_data.c +++ b/src/gallium/state_trackers/g3dvl/vl_data.c @@ -61,14 +61,6 @@ const struct VL_TEXCOORD2F *vl_chroma_422_texcoords = (const struct VL_TEXCOORD2 */ const struct VL_TEXCOORD2F *vl_chroma_444_texcoords = vl_luma_texcoords; -/* - * Represents texcoords for the above for rendering a predicted macroblock. - * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. - * Texcoords need to be translated to cover source macroblock on the - * past/future surface. - */ - const struct VL_TEXCOORD2F *vl_ref_surface_texcoords = (const struct VL_TEXCOORD2F*)vl_mb_vertex_positions; - /* * Represents 2 triangles in a strip in normalized coords. * Used to render the surface onto the frame buffer. diff --git a/src/gallium/state_trackers/g3dvl/vl_data.h b/src/gallium/state_trackers/g3dvl/vl_data.h index 67a0a749909..8f347273ad7 100644 --- a/src/gallium/state_trackers/g3dvl/vl_data.h +++ b/src/gallium/state_trackers/g3dvl/vl_data.h @@ -8,7 +8,6 @@ extern const struct VL_TEXCOORD2F vl_luma_texcoords[24]; extern const struct VL_TEXCOORD2F *vl_chroma_420_texcoords; extern const struct VL_TEXCOORD2F *vl_chroma_422_texcoords; extern const struct VL_TEXCOORD2F *vl_chroma_444_texcoords; -extern const struct VL_TEXCOORD2F *vl_ref_surface_texcoords; extern const struct VL_VERTEX2F vl_surface_vertex_positions[4]; extern const struct VL_TEXCOORD2F *vl_surface_texcoords; diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index e58e434dab1..0e1adea4729 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -269,8 +269,6 @@ int vlRenderIMacroBlock pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); pipe->set_sampler_textures(pipe, 3, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 3, (void**)surface->context->states.mc.samplers); - pipe->set_vertex_buffers(pipe, 4, surface->context->states.mc.vertex_bufs); - pipe->set_vertex_elements(pipe, 4, surface->context->states.mc.vertex_buf_elems); pipe->bind_vs_state(pipe, surface->context->states.mc.i_vs); pipe->bind_fs_state(pipe, surface->context->states.mc.i_fs); @@ -354,8 +352,6 @@ int vlRenderPMacroBlock surface->context->states.mc.textures[3] = ref_surface->texture; pipe->set_sampler_textures(pipe, 4, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 4, (void**)surface->context->states.mc.samplers); - pipe->set_vertex_buffers(pipe, 5, surface->context->states.mc.vertex_bufs); - pipe->set_vertex_elements(pipe, 5, surface->context->states.mc.vertex_buf_elems); pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs); pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs); @@ -445,8 +441,6 @@ int vlRenderBMacroBlock surface->context->states.mc.textures[4] = future_surface->texture; pipe->set_sampler_textures(pipe, 5, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 5, (void**)surface->context->states.mc.samplers); - pipe->set_vertex_buffers(pipe, 6, surface->context->states.mc.vertex_bufs); - pipe->set_vertex_elements(pipe, 6, surface->context->states.mc.vertex_buf_elems); pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs); pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs); -- cgit v1.2.3 From 5e1d657d50c247d903b865572bd3e74048e8a8f1 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 20 Jun 2008 22:19:22 +0200 Subject: nv30: Add separate nv30 state stuff for fb, based on nv40 one, need to use it now --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 18 ++++ src/gallium/drivers/nv30/nv30_state_fb.c | 151 +++++++++++++++++++++++++++++++ 3 files changed, 170 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_fb.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index 3f80fb87c9b..ec11de76446 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -14,6 +14,7 @@ DRIVER_SOURCES = \ nv30_screen.c \ nv30_state.c \ nv30_state_emit.c \ + nv30_state_fb.c \ nv30_surface.c \ nv30_vbo.c \ nv30_vertprog.c diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 333bd4875c6..9fbe66dc74e 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -76,6 +76,10 @@ enum nv30_state_index { #define NV30_NEW_ARRAYS (1 << 11) #define NV30_NEW_UCP (1 << 12) +struct nv30_state { + struct nouveau_stateobj *hw[NV30_STATE_MAX]; +}; + struct nv30_context { struct pipe_context pipe; @@ -85,6 +89,9 @@ struct nv30_context { struct draw_context *draw; + /* HW state derived from pipe states */ + struct nv30_state state; + uint32_t dirty; struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; @@ -93,6 +100,9 @@ struct nv30_context { unsigned fp_samplers; unsigned vp_samplers; + /* Context state */ + struct pipe_framebuffer_state framebuffer; + uint32_t rt_enable; struct pipe_buffer *rt[2]; struct pipe_buffer *zeta; @@ -132,6 +142,14 @@ nv30_context(struct pipe_context *pipe) return (struct nv30_context *)pipe; } +struct nv30_state_entry { + boolean (*validate)(struct nv30_context *nv30); + struct { + unsigned pipe; + unsigned hw; + } dirty; +}; + extern void nv30_init_state_functions(struct nv30_context *nv30); extern void nv30_init_surface_functions(struct nv30_context *nv30); extern void nv30_init_query_functions(struct nv30_context *nv30); diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c new file mode 100644 index 00000000000..73c97e298ad --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_fb.c @@ -0,0 +1,151 @@ +#include "nv30_context.h" + +static boolean +nv30_state_framebuffer_validate(struct nv30_context *nv30) +{ + struct pipe_framebuffer_state *fb = &nv30->framebuffer; + struct pipe_surface *rt[4], *zeta = NULL; + uint32_t rt_enable, rt_format; + int i, colour_format = 0, zeta_format = 0; + struct nouveau_stateobj *so = so_new(64, 10); + unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM; + unsigned w = fb->width; + unsigned h = fb->height; + + rt_enable = 0; + for (i = 0; i < fb->num_cbufs; i++) { + if (colour_format) { + assert(colour_format == fb->cbufs[i]->format); + } else { + colour_format = fb->cbufs[i]->format; + rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i); + rt[i] = fb->cbufs[i]; + } + } + + if (rt_enable & (NV34TCL_RT_ENABLE_COLOR1 | NV34TCL_RT_ENABLE_COLOR2 | + NV34TCL_RT_ENABLE_COLOR3)) + rt_enable |= NV34TCL_RT_ENABLE_MRT; + + if (fb->zsbuf) { + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV34TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + switch (zeta_format) { + case PIPE_FORMAT_Z16_UNORM: + rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16; + break; + case PIPE_FORMAT_Z24S8_UNORM: + case 0: + rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8; + break; + default: + assert(0); + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { + uint32_t pitch = rt[0]->pitch * rt[0]->cpp; + if (zeta) { + pitch |= (zeta->pitch * zeta->cpp)<<16; + } else { + pitch |= pitch<<16; + } + + so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR0, 1); + so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv30->nvws->channel->vram->handle, + nv30->nvws->channel->gart->handle); + so_method(so, nv30->screen->rankine, NV34TCL_COLOR0_PITCH, 2); + so_data (so, pitch); + so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { + so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR1, 1); + so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv30->nvws->channel->vram->handle, + nv30->nvws->channel->gart->handle); + so_method(so, nv30->screen->rankine, NV34TCL_COLOR1_OFFSET, 2); + so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, rt[1]->pitch * rt[1]->cpp); + } +/* + if (rt_enable & NV34TCL_RT_ENABLE_COLOR2) { + so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR2, 1); + so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv30->nvws->channel->vram->handle, + nv30->nvws->channel->gart->handle); + so_method(so, nv30->screen->rankine, NV34TCL_COLOR2_OFFSET, 1); + so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv30->screen->rankine, NV34TCL_COLOR2_PITCH, 1); + so_data (so, rt[2]->pitch * rt[2]->cpp); + } + + if (rt_enable & NV34TCL_RT_ENABLE_COLOR3) { + so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR3, 1); + so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv30->nvws->channel->vram->handle, + nv30->nvws->channel->gart->handle); + so_method(so, nv30->screen->rankine, NV34TCL_COLOR3_OFFSET, 1); + so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, nv30->screen->rankine, NV34TCL_COLOR3_PITCH, 1); + so_data (so, rt[3]->pitch * rt[3]->cpp); + } +*/ + if (zeta_format) { + so_method(so, nv30->screen->rankine, NV34TCL_DMA_ZETA, 1); + so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, + nv30->nvws->channel->vram->handle, + nv30->nvws->channel->gart->handle); + so_method(so, nv30->screen->rankine, NV34TCL_ZETA_OFFSET, 1); + so_reloc (so, zeta->buffer, zeta->offset, rt_flags | + NOUVEAU_BO_LOW, 0, 0); + /*so_method(so, nv30->screen->rankine, NV34TCL_ZETA_PITCH, 1); + so_data (so, zeta->pitch * zeta->cpp);*/ + } + + so_method(so, nv30->screen->rankine, NV34TCL_RT_ENABLE, 1); + so_data (so, rt_enable); + so_method(so, nv30->screen->rankine, NV34TCL_RT_HORIZ, 3); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_data (so, rt_format); + so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_HORIZ, 2); + so_data (so, (w << 16) | 0); + so_data (so, (h << 16) | 0); + so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); + so_data (so, ((w - 1) << 16) | 0); + so_data (so, ((h - 1) << 16) | 0); + so_method(so, nv30->screen->rankine, 0x1d88, 1); + so_data (so, (1 << 12) | h); + + so_ref(so, &nv30->state.hw[NV30_STATE_FB]); + return TRUE; +} + +struct nv30_state_entry nv30_state_framebuffer = { + .validate = nv30_state_framebuffer_validate, + .dirty = { + .pipe = NV30_NEW_FB, + .hw = NV30_STATE_FB + } +}; -- cgit v1.2.3 From 582b39ebb9f67e3b67a776be0961fe2e51ee46f7 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 21 Jun 2008 10:11:37 +0200 Subject: nv30: Update nouveau_class.h to get its proper defines instead of using nv40 ones --- src/gallium/drivers/nouveau/nouveau_class.h | 77 ++++++++++++++++++----------- src/gallium/drivers/nv30/nv30_state.c | 4 +- src/gallium/drivers/nv30/nv30_vbo.c | 34 ++++++------- 3 files changed, 66 insertions(+), 49 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 880afe6ce08..58c80ddcd24 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -4,7 +4,7 @@ ************************************************************************** - Copyright (C) 2006-2007 : + Copyright (C) 2006-2008 : Dmitry Baryshkov, Laurent Carlier, Matthieu Castet, @@ -1740,6 +1740,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK 0x00f00000 +#define NV10_DX5_TEXTURED_TRIANGLE 0x00000094 + + + #define NV10TCL 0x00000056 #define NV10TCL_NOP 0x00000100 @@ -3871,13 +3875,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR 0x80020000 #define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 #define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 -#define NV34TCL_BLEND_FUNC_COLOR 0x0000031c -#define NV34TCL_BLEND_FUNC_EQUATION 0x00000320 -#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_ADD 0x00008006 -#define NV34TCL_BLEND_FUNC_EQUATION_MIN 0x00008007 -#define NV34TCL_BLEND_FUNC_EQUATION_MAX 0x00008008 -#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_SUBTRACT 0x0000800a -#define NV34TCL_BLEND_FUNC_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b +#define NV34TCL_BLEND_COLOR 0x0000031c +#define NV34TCL_BLEND_EQUATION 0x00000320 +#define NV34TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 +#define NV34TCL_BLEND_EQUATION_MIN 0x00008007 +#define NV34TCL_BLEND_EQUATION_MAX 0x00008008 +#define NV34TCL_BLEND_EQUATION_FUNC_SUBTRACT 0x0000800a +#define NV34TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT 0x0000800b #define NV34TCL_COLOR_MASK 0x00000324 #define NV34TCL_COLOR_MASK_B_SHIFT 0 #define NV34TCL_COLOR_MASK_B_MASK 0x000000ff @@ -4217,6 +4221,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE 0x00000008 #define NV34TCL_FRONT_MATERIAL_SHININESS(x) (0x00001400+((x)*4)) #define NV34TCL_FRONT_MATERIAL_SHININESS__SIZE 0x00000006 +#define NV34TCL_ENABLED_LIGHTS 0x00001420 #define NV34TCL_FP_REG_CONTROL 0x00001450 #define NV34TCL_FP_REG_CONTROL_UNK1_SHIFT 16 #define NV34TCL_FP_REG_CONTROL_UNK1_MASK 0xffff0000 @@ -4234,12 +4239,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_POLYGON_STIPPLE_ENABLE 0x0000147c #define NV34TCL_POLYGON_STIPPLE_PATTERN(x) (0x00001480+((x)*4)) #define NV34TCL_POLYGON_STIPPLE_PATTERN__SIZE 0x00000020 -#define NV34TCL_VERTEX_ATTR_3F_X(x) (0x00001500+((x)*16)) -#define NV34TCL_VERTEX_ATTR_3F_X__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_3F_Y(x) (0x00001504+((x)*16)) -#define NV34TCL_VERTEX_ATTR_3F_Y__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_3F_Z(x) (0x00001508+((x)*16)) -#define NV34TCL_VERTEX_ATTR_3F_Z__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_3F_X(x) (0x00001500+((x)*16)) +#define NV34TCL_VTX_ATTR_3F_X__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_3F_Y(x) (0x00001504+((x)*16)) +#define NV34TCL_VTX_ATTR_3F_Y__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_3F_Z(x) (0x00001508+((x)*16)) +#define NV34TCL_VTX_ATTR_3F_Z__SIZE 0x00000010 #define NV34TCL_VP_CLIP_PLANE_A(x) (0x00001600+((x)*16)) #define NV34TCL_VP_CLIP_PLANE_A__SIZE 0x00000006 #define NV34TCL_VP_CLIP_PLANE_B(x) (0x00001604+((x)*16)) @@ -4248,22 +4253,22 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VP_CLIP_PLANE_C__SIZE 0x00000006 #define NV34TCL_VP_CLIP_PLANE_D(x) (0x0000160c+((x)*16)) #define NV34TCL_VP_CLIP_PLANE_D__SIZE 0x00000006 -#define NV34TCL_VERTEX_BUFFER_ADDRESS(x) (0x00001680+((x)*4)) -#define NV34TCL_VERTEX_BUFFER_ADDRESS__SIZE 0x00000010 -#define NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1 (1 << 31) -#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_SHIFT 0 -#define NV34TCL_VERTEX_BUFFER_ADDRESS_OFFSET_MASK 0x0fffffff -#define NV34TCL_VERTEX_ARRAY_FORMAT(x) (0x00001740+((x)*4)) -#define NV34TCL_VERTEX_ARRAY_FORMAT__SIZE 0x00000010 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_SHIFT 0 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_MASK 0x0000000f -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT 0x00000002 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE 0x00000004 -#define NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_USHORT 0x00000005 -#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT 4 -#define NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_MASK 0x000000f0 -#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT 8 -#define NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_MASK 0x0000ff00 +#define NV34TCL_VTXBUF_ADDRESS(x) (0x00001680+((x)*4)) +#define NV34TCL_VTXBUF_ADDRESS__SIZE 0x00000010 +#define NV34TCL_VTXBUF_ADDRESS_DMA1 (1 << 31) +#define NV34TCL_VTXBUF_ADDRESS_OFFSET_SHIFT 0 +#define NV34TCL_VTXBUF_ADDRESS_OFFSET_MASK 0x0fffffff +#define NV34TCL_VTXFMT(x) (0x00001740+((x)*4)) +#define NV34TCL_VTXFMT__SIZE 0x00000010 +#define NV34TCL_VTXFMT_TYPE_SHIFT 0 +#define NV34TCL_VTXFMT_TYPE_MASK 0x0000000f +#define NV34TCL_VTXFMT_TYPE_FLOAT 0x00000002 +#define NV34TCL_VTXFMT_TYPE_UBYTE 0x00000004 +#define NV34TCL_VTXFMT_TYPE_USHORT 0x00000005 +#define NV34TCL_VTXFMT_SIZE_SHIFT 4 +#define NV34TCL_VTXFMT_SIZE_MASK 0x000000f0 +#define NV34TCL_VTXFMT_STRIDE_SHIFT 8 +#define NV34TCL_VTXFMT_STRIDE_MASK 0x0000ff00 #define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R 0x000017a0 #define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G 0x000017a4 #define NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B 0x000017a8 @@ -4302,6 +4307,18 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VB_VERTEX_BATCH_COUNT_SHIFT 24 #define NV34TCL_VB_VERTEX_BATCH_COUNT_MASK 0xff000000 #define NV34TCL_VERTEX_DATA 0x00001818 +#define NV34TCL_IDXBUF_ADDRESS 0x0000181c +#define NV34TCL_IDXBUF_FORMAT 0x00001820 +#define NV34TCL_IDXBUF_FORMAT_TYPE_SHIFT 4 +#define NV34TCL_IDXBUF_FORMAT_TYPE_MASK 0x000000f0 +#define NV34TCL_IDXBUF_FORMAT_TYPE_U32 0x00000000 +#define NV34TCL_IDXBUF_FORMAT_TYPE_U16 0x00000010 +#define NV34TCL_IDXBUF_FORMAT_DMA1 (1 << 0) +#define NV34TCL_VB_INDEX_BATCH 0x00001824 +#define NV34TCL_VB_INDEX_BATCH_COUNT_SHIFT 24 +#define NV34TCL_VB_INDEX_BATCH_COUNT_MASK 0xff000000 +#define NV34TCL_VB_INDEX_BATCH_START_SHIFT 0 +#define NV34TCL_VB_INDEX_BATCH_START_MASK 0x00ffffff #define NV34TCL_POLYGON_MODE_FRONT 0x00001828 #define NV34TCL_POLYGON_MODE_FRONT_POINT 0x00001b00 #define NV34TCL_POLYGON_MODE_FRONT_LINE 0x00001b01 diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 620038fa647..84f016eeadc 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -51,7 +51,7 @@ nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso) OUT_RING (cb->b_enable); OUT_RING (cb->b_srcfunc); OUT_RING (cb->b_dstfunc); - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_EQUATION, 1); + BEGIN_RING(rankine, NV34TCL_BLEND_EQUATION, 1); OUT_RING (cb->b_eqn); BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); @@ -531,7 +531,7 @@ nv30_set_blend_color(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_COLOR, 1); + BEGIN_RING(rankine, NV34TCL_BLEND_COLOR, 1); OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | (float_to_ubyte(bcol->color[0]) << 16) | (float_to_ubyte(bcol->color[1]) << 8) | diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index b18a407ec5a..ff0ce6ac810 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -26,12 +26,12 @@ nv30_vbo_type(uint format) { switch (pf_type(format)) { case PIPE_FORMAT_TYPE_FLOAT: - return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + return NV34TCL_VTXFMT_TYPE_FLOAT; case PIPE_FORMAT_TYPE_UNORM: - return NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_UBYTE; + return NV34TCL_VTXFMT_TYPE_UBYTE; default: NOUVEAU_ERR("Unknown format 0x%08x\n", format); - return NV40TCL_VTXFMT_TYPE_FLOAT; + return NV34TCL_VTXFMT_TYPE_FLOAT; } } @@ -51,7 +51,7 @@ nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, map += vb->buffer_offset + ve->src_offset; switch (type) { - case NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT: + case NV34TCL_VTXFMT_TYPE_FLOAT: { float *v = map; @@ -121,7 +121,7 @@ nv30_vbo_arrays_update(struct nv30_context *nv30) struct pipe_vertex_buffer *vb; if (!(inputs & (1 << hw))) { - vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + vtxfmt[hw] = NV34TCL_VTXFMT_TYPE_FLOAT; continue; } @@ -129,7 +129,7 @@ nv30_vbo_arrays_update(struct nv30_context *nv30) vb = &nv30->vtxbuf[ve->vertex_buffer_index]; if (vb->pitch == 0) { - vtxfmt[hw] = NV34TCL_VERTEX_ARRAY_FORMAT_TYPE_FLOAT; + vtxfmt[hw] = NV34TCL_VTXFMT_TYPE_FLOAT; if (nv30_vbo_static_attrib(nv30, hw, ve, vb) == TRUE) continue; } @@ -138,13 +138,13 @@ nv30_vbo_arrays_update(struct nv30_context *nv30) nv30->vb[hw].delta = vb->buffer_offset + ve->src_offset; nv30->vb[hw].buffer = vb->buffer; - vtxfmt[hw] = ((vb->pitch << NV34TCL_VERTEX_ARRAY_FORMAT_STRIDE_SHIFT) | + vtxfmt[hw] = ((vb->pitch << NV34TCL_VTXFMT_STRIDE_SHIFT) | (nv30_vbo_ncomp(ve->src_format) << - NV34TCL_VERTEX_ARRAY_FORMAT_SIZE_SHIFT) | + NV34TCL_VTXFMT_SIZE_SHIFT) | nv30_vbo_type(ve->src_format)); } - BEGIN_RING(rankine, NV34TCL_VERTEX_ARRAY_FORMAT(0), num_hw); + BEGIN_RING(rankine, NV34TCL_VTXFMT(0), num_hw); OUT_RINGp (vtxfmt, num_hw); } @@ -167,20 +167,20 @@ nv30_vbo_validate_state(struct nv30_context *nv30, inputs &= ~(1 << a); - BEGIN_RING(rankine, NV34TCL_VERTEX_BUFFER_ADDRESS(a), 1); + BEGIN_RING(rankine, NV34TCL_VTXBUF_ADDRESS(a), 1); OUT_RELOC (nv30->vb[a].buffer, nv30->vb[a].delta, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV34TCL_VERTEX_BUFFER_ADDRESS_DMA1); + NV34TCL_VTXBUF_ADDRESS_DMA1); } if (ib) { - BEGIN_RING(rankine, NV40TCL_IDXBUF_ADDRESS, 2); + BEGIN_RING(rankine, NV34TCL_IDXBUF_ADDRESS, 2); OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_OR, - 0, NV40TCL_IDXBUF_FORMAT_DMA1); + 0, NV34TCL_IDXBUF_FORMAT_DMA1); } BEGIN_RING(rankine, 0x1710, 1); @@ -360,10 +360,10 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, switch (ib_size) { case 2: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U16; + type = NV34TCL_IDXBUF_FORMAT_TYPE_U16; break; case 4: - type = NV40TCL_IDXBUF_FORMAT_TYPE_U32; + type = NV34TCL_IDXBUF_FORMAT_TYPE_U32; break; default: NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); @@ -381,7 +381,7 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, nr = (count & 0xff); if (nr) { - BEGIN_RING(rankine, NV40TCL_VB_INDEX_BATCH, 1); + BEGIN_RING(rankine, NV34TCL_VB_INDEX_BATCH, 1); OUT_RING (((nr - 1) << 24) | start); start += nr; } @@ -392,7 +392,7 @@ nv30_draw_elements_vbo(struct pipe_context *pipe, nr -= push; - BEGIN_RING_NI(rankine, NV40TCL_VB_INDEX_BATCH, push); + BEGIN_RING_NI(rankine, NV34TCL_VB_INDEX_BATCH, push); while (push--) { OUT_RING(((0x100 - 1) << 24) | start); start += 0x100; -- cgit v1.2.3 From 5fea663b5f7abcdca00c5ff5d1b77f200b0d06ec Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 21 Jun 2008 12:03:05 +0200 Subject: nv30: Add state for blend --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 9 +++++++ src/gallium/drivers/nv30/nv30_state_blend.c | 40 +++++++++++++++++++++++++++++ 3 files changed, 50 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_blend.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index ec11de76446..04f53ee4565 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -13,6 +13,7 @@ DRIVER_SOURCES = \ nv30_query.c \ nv30_screen.c \ nv30_state.c \ + nv30_state_blend.c \ nv30_state_emit.c \ nv30_state_fb.c \ nv30_surface.c \ diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 9fbe66dc74e..1695ba5a017 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -76,6 +76,13 @@ enum nv30_state_index { #define NV30_NEW_ARRAYS (1 << 11) #define NV30_NEW_UCP (1 << 12) +/* TODO: rename when removing the old state emitter */ +struct nv30_blend_state_new { + struct pipe_blend_state pipe; + struct nouveau_stateobj *so; +}; + + struct nv30_state { struct nouveau_stateobj *hw[NV30_STATE_MAX]; }; @@ -101,6 +108,8 @@ struct nv30_context { unsigned vp_samplers; /* Context state */ + struct nv30_blend_state_new *blend; + struct pipe_blend_color blend_colour; struct pipe_framebuffer_state framebuffer; uint32_t rt_enable; diff --git a/src/gallium/drivers/nv30/nv30_state_blend.c b/src/gallium/drivers/nv30/nv30_state_blend.c new file mode 100644 index 00000000000..a1b0100472d --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_blend.c @@ -0,0 +1,40 @@ +#include "nv30_context.h" + +static boolean +nv30_state_blend_validate(struct nv30_context *nv30) +{ + so_ref(nv30->blend->so, &nv30->state.hw[NV30_STATE_BLEND]); + return TRUE; +} + +struct nv30_state_entry nv30_state_blend_new = { + .validate = nv30_state_blend_validate, + .dirty = { + .pipe = NV30_NEW_BLEND, + .hw = NV30_STATE_BLEND + } +}; + +static boolean +nv30_state_blend_colour_validate(struct nv30_context *nv30) +{ + struct nouveau_stateobj *so = so_new(2, 0); + struct pipe_blend_color *bcol = &nv30->blend_colour; + + so_method(so, nv30->screen->rankine, NV34TCL_BLEND_COLOR, 1); + so_data (so, ((float_to_ubyte(bcol->color[3]) << 24) | + (float_to_ubyte(bcol->color[0]) << 16) | + (float_to_ubyte(bcol->color[1]) << 8) | + (float_to_ubyte(bcol->color[2]) << 0))); + + so_ref(so, &nv30->state.hw[NV30_STATE_BCOL]); + return TRUE; +} + +struct nv30_state_entry nv30_state_blend_colour = { + .validate = nv30_state_blend_colour_validate, + .dirty = { + .pipe = NV30_NEW_BCOL, + .hw = NV30_STATE_BCOL + } +}; -- cgit v1.2.3 From 8c26a521ee80f5d8a1d0aabd0910233aad400322 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 21 Jun 2008 22:59:05 +0200 Subject: Update for extra vertex attributes --- src/gallium/drivers/nouveau/nouveau_class.h | 163 +++++++++++++++------------- src/gallium/drivers/nv30/nv30_vbo.c | 2 +- src/gallium/drivers/nv40/nv40_draw.c | 4 +- src/gallium/drivers/nv40/nv40_vbo.c | 2 +- 4 files changed, 93 insertions(+), 78 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 58c80ddcd24..3c29fa0d1be 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -3763,7 +3763,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_RT_ENABLE_COLOR2 (1 << 2) #define NV34TCL_RT_ENABLE_COLOR1 (1 << 1) #define NV34TCL_RT_ENABLE_COLOR0 (1 << 0) -#define NV34TCL_ZETA_PITCH 0x0000022c +#define NV34TCL_LMA_DEPTH_PITCH 0x0000022c #define NV34TCL_LMA_DEPTH_OFFSET 0x00000230 #define NV34TCL_TX_UNITS_ENABLE 0x0000023c #define NV34TCL_TX_UNITS_ENABLE_TX0 (1 << 0) @@ -4145,14 +4145,16 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_DEPTH_TEST_ENABLE 0x00000a74 #define NV34TCL_POLYGON_OFFSET_FACTOR 0x00000a78 #define NV34TCL_POLYGON_OFFSET_UNITS 0x00000a7c -#define NV34TCL_VERTEX_NOR_3I_XY 0x00000a90 -#define NV34TCL_VERTEX_NOR_3I_XY_X_SHIFT 0 -#define NV34TCL_VERTEX_NOR_3I_XY_X_MASK 0x0000ffff -#define NV34TCL_VERTEX_NOR_3I_XY_Y_SHIFT 16 -#define NV34TCL_VERTEX_NOR_3I_XY_Y_MASK 0xffff0000 -#define NV34TCL_VERTEX_NOR_3I_Z 0x00000a94 -#define NV34TCL_VERTEX_NOR_3I_Z_Z_SHIFT 0 -#define NV34TCL_VERTEX_NOR_3I_Z_Z_MASK 0x0000ffff +#define NV34TCL_VTX_ATTR_3I_XY(x) (0x00000a80+((x)*8)) +#define NV34TCL_VTX_ATTR_3I_XY__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_3I_XY_X_SHIFT 0 +#define NV34TCL_VTX_ATTR_3I_XY_X_MASK 0x0000ffff +#define NV34TCL_VTX_ATTR_3I_XY_Y_SHIFT 16 +#define NV34TCL_VTX_ATTR_3I_XY_Y_MASK 0xffff0000 +#define NV34TCL_VTX_ATTR_3I_Z(x) (0x00000a84+((x)*8)) +#define NV34TCL_VTX_ATTR_3I_Z__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_3I_Z_Z_SHIFT 0 +#define NV34TCL_VTX_ATTR_3I_Z_Z_MASK 0x0000ffff #define NV34TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) #define NV34TCL_VP_UPLOAD_INST__SIZE 0x00000004 #define NV34TCL_CLIP_PLANE_A(x) (0x00000e00+((x)*16)) @@ -4336,48 +4338,38 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_FRONT_FACE_CCW 0x00000901 #define NV34TCL_POLYGON_SMOOTH_ENABLE 0x00001838 #define NV34TCL_CULL_FACE_ENABLE 0x0000183c -#define NV34TCL_VERTEX_ATTR_2F_X(x) (0x00001880+((x)*8)) -#define NV34TCL_VERTEX_ATTR_2F_X__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_2F_Y(x) (0x00001884+((x)*8)) -#define NV34TCL_VERTEX_ATTR_2F_Y__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_2I(x) (0x00001900+((x)*4)) -#define NV34TCL_VERTEX_ATTR_2I__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_2I_Y_SHIFT 16 -#define NV34TCL_VERTEX_ATTR_2I_Y_MASK 0xffff0000 -#define NV34TCL_VERTEX_ATTR_2I_X_SHIFT 0 -#define NV34TCL_VERTEX_ATTR_2I_X_MASK 0x0000ffff -#define NV34TCL_VERTEX_COL_4I(x) (0x0000194c+((x)*4)) -#define NV34TCL_VERTEX_COL_4I__SIZE 0x00000002 -#define NV34TCL_VERTEX_COL_4I_R_SHIFT 0 -#define NV34TCL_VERTEX_COL_4I_R_MASK 0x000000ff -#define NV34TCL_VERTEX_COL_4I_G_SHIFT 8 -#define NV34TCL_VERTEX_COL_4I_G_MASK 0x0000ff00 -#define NV34TCL_VERTEX_COL_4I_B_SHIFT 16 -#define NV34TCL_VERTEX_COL_4I_B_MASK 0x00ff0000 -#define NV34TCL_VERTEX_COL_4I_A_SHIFT 24 -#define NV34TCL_VERTEX_COL_4I_A_MASK 0xff000000 -#define NV34TCL_VERTEX_POS_4I_XY 0x00001980 -#define NV34TCL_VERTEX_POS_4I_XY_X_SHIFT 0 -#define NV34TCL_VERTEX_POS_4I_XY_X_MASK 0x0000ffff -#define NV34TCL_VERTEX_POS_4I_XY_Y_SHIFT 16 -#define NV34TCL_VERTEX_POS_4I_XY_Y_MASK 0xffff0000 -#define NV34TCL_VERTEX_POS_4I_ZW 0x00001984 -#define NV34TCL_VERTEX_POS_4I_ZW_Z_SHIFT 0 -#define NV34TCL_VERTEX_POS_4I_ZW_Z_MASK 0x0000ffff -#define NV34TCL_VERTEX_POS_4I_ZW_W_SHIFT 16 -#define NV34TCL_VERTEX_POS_4I_ZW_W_MASK 0xffff0000 -#define NV34TCL_VERTEX_TX_4I_ST(x) (0x000019c0+((x)*8)) -#define NV34TCL_VERTEX_TX_4I_ST__SIZE 0x00000004 -#define NV34TCL_VERTEX_TX_4I_ST_S_SHIFT 0 -#define NV34TCL_VERTEX_TX_4I_ST_S_MASK 0x0000ffff -#define NV34TCL_VERTEX_TX_4I_ST_T_SHIFT 16 -#define NV34TCL_VERTEX_TX_4I_ST_T_MASK 0xffff0000 -#define NV34TCL_VERTEX_TX_4I_RQ(x) (0x000019c4+((x)*8)) -#define NV34TCL_VERTEX_TX_4I_RQ__SIZE 0x00000004 -#define NV34TCL_VERTEX_TX_4I_RQ_R_SHIFT 0 -#define NV34TCL_VERTEX_TX_4I_RQ_R_MASK 0x0000ffff -#define NV34TCL_VERTEX_TX_4I_RQ_Q_SHIFT 16 -#define NV34TCL_VERTEX_TX_4I_RQ_Q_MASK 0xffff0000 +#define NV34TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) +#define NV34TCL_VTX_ATTR_2F_X__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) +#define NV34TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) +#define NV34TCL_VTX_ATTR_2I__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_2I_X_SHIFT 0 +#define NV34TCL_VTX_ATTR_2I_X_MASK 0x0000ffff +#define NV34TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV34TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV34TCL_VTX_ATTR_4UB(x) (0x00001940+((x)*4)) +#define NV34TCL_VTX_ATTR_4UB__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4UB_X_SHIFT 0 +#define NV34TCL_VTX_ATTR_4UB_X_MASK 0x000000ff +#define NV34TCL_VTX_ATTR_4UB_Y_SHIFT 8 +#define NV34TCL_VTX_ATTR_4UB_Y_MASK 0x0000ff00 +#define NV34TCL_VTX_ATTR_4UB_Z_SHIFT 16 +#define NV34TCL_VTX_ATTR_4UB_Z_MASK 0x00ff0000 +#define NV34TCL_VTX_ATTR_4UB_W_SHIFT 24 +#define NV34TCL_VTX_ATTR_4UB_W_MASK 0xff000000 +#define NV34TCL_VTX_ATTR_4I_XY(x) (0x00001980+((x)*8)) +#define NV34TCL_VTX_ATTR_4I_XY__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4I_XY_X_SHIFT 0 +#define NV34TCL_VTX_ATTR_4I_XY_X_MASK 0x0000ffff +#define NV34TCL_VTX_ATTR_4I_XY_Y_SHIFT 16 +#define NV34TCL_VTX_ATTR_4I_XY_Y_MASK 0xffff0000 +#define NV34TCL_VTX_ATTR_4I_ZW(x) (0x00001984+((x)*8)) +#define NV34TCL_VTX_ATTR_4I_ZW__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4I_ZW_Z_SHIFT 0 +#define NV34TCL_VTX_ATTR_4I_ZW_Z_MASK 0x0000ffff +#define NV34TCL_VTX_ATTR_4I_ZW_W_SHIFT 16 +#define NV34TCL_VTX_ATTR_4I_ZW_W_MASK 0xffff0000 #define NV34TCL_TX_OFFSET(x) (0x00001a00+((x)*32)) #define NV34TCL_TX_OFFSET__SIZE 0x00000004 #define NV34TCL_TX_FORMAT(x) (0x00001a04+((x)*32)) @@ -4534,14 +4526,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_BORDER_COLOR_R_MASK 0x00ff0000 #define NV34TCL_TX_BORDER_COLOR_A_SHIFT 24 #define NV34TCL_TX_BORDER_COLOR_A_MASK 0xff000000 -#define NV34TCL_VERTEX_ATTR_4F_X(x) (0x00001c00+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_X__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_Y__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_Z__SIZE 0x00000010 -#define NV34TCL_VERTEX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) -#define NV34TCL_VERTEX_ATTR_4F_W__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4F_X(x) (0x00001c00+((x)*16)) +#define NV34TCL_VTX_ATTR_4F_X__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4F_Y(x) (0x00001c04+((x)*16)) +#define NV34TCL_VTX_ATTR_4F_Y__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4F_Z(x) (0x00001c08+((x)*16)) +#define NV34TCL_VTX_ATTR_4F_Z__SIZE 0x00000010 +#define NV34TCL_VTX_ATTR_4F_W(x) (0x00001c0c+((x)*16)) +#define NV34TCL_VTX_ATTR_4F_W__SIZE 0x00000010 #define NV34TCL_FP_CONTROL 0x00001d60 #define NV34TCL_FP_CONTROL_USES_KIL (1 << 7) #define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_SHIFT 0 @@ -4573,7 +4565,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 #define NV34TCL_BACK_MATERIAL_SHININESS(x) (0x00001e20+((x)*4)) #define NV34TCL_BACK_MATERIAL_SHININESS__SIZE 0x00000006 -#define NV34TCL_VERTEX_FOG_1F 0x00001e54 +#define NV34TCL_VTX_ATTR_1F(x) (0x00001e40+((x)*4)) +#define NV34TCL_VTX_ATTR_1F__SIZE 0x00000010 #define NV34TCL_VP_UPLOAD_FROM_ID 0x00001e9c #define NV34TCL_VP_START_FROM_ID 0x00001ea0 #define NV34TCL_POINT_PARAMETERS(x) (0x00001ec0+((x)*4)) @@ -4986,6 +4979,16 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_DEPTH_TEST_ENABLE 0x00000a74 #define NV40TCL_POLYGON_OFFSET_FACTOR 0x00000a78 #define NV40TCL_POLYGON_OFFSET_UNITS 0x00000a7c +#define NV40TCL_VTX_ATTR_3I_XY(x) (0x00000a80+((x)*8)) +#define NV40TCL_VTX_ATTR_3I_XY__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3I_XY_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_3I_XY_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_3I_XY_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_3I_XY_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_3I_Z(x) (0x00000a84+((x)*8)) +#define NV40TCL_VTX_ATTR_3I_Z__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_3I_Z_Z_SHIFT 0 +#define NV40TCL_VTX_ATTR_3I_Z_Z_MASK 0x0000ffff #define NV40TCL_UNK0B40(x) (0x00000b40+((x)*4)) #define NV40TCL_UNK0B40__SIZE 0x00000008 #define NV40TCL_VP_UPLOAD_INST(x) (0x00000b80+((x)*4)) @@ -5095,22 +5098,32 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_VTX_ATTR_2F_Y__SIZE 0x00000010 #define NV40TCL_VTX_ATTR_2I(x) (0x00001900+((x)*4)) #define NV40TCL_VTX_ATTR_2I__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 -#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 #define NV40TCL_VTX_ATTR_2I_X_SHIFT 0 #define NV40TCL_VTX_ATTR_2I_X_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_4I_0(x) (0x00001900+((x)*8)) -#define NV40TCL_VTX_ATTR_4I_0__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4I_0_Y_SHIFT 16 -#define NV40TCL_VTX_ATTR_4I_0_Y_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_4I_0_X_SHIFT 0 -#define NV40TCL_VTX_ATTR_4I_0_X_MASK 0x0000ffff -#define NV40TCL_VTX_ATTR_4I_1(x) (0x00001904+((x)*8)) -#define NV40TCL_VTX_ATTR_4I_1__SIZE 0x00000010 -#define NV40TCL_VTX_ATTR_4I_1_W_SHIFT 16 -#define NV40TCL_VTX_ATTR_4I_1_W_MASK 0xffff0000 -#define NV40TCL_VTX_ATTR_4I_1_Z_SHIFT 0 -#define NV40TCL_VTX_ATTR_4I_1_Z_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_2I_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_2I_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4UB(x) (0x00001940+((x)*4)) +#define NV40TCL_VTX_ATTR_4UB__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4UB_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_4UB_X_MASK 0x000000ff +#define NV40TCL_VTX_ATTR_4UB_Y_SHIFT 8 +#define NV40TCL_VTX_ATTR_4UB_Y_MASK 0x0000ff00 +#define NV40TCL_VTX_ATTR_4UB_Z_SHIFT 16 +#define NV40TCL_VTX_ATTR_4UB_Z_MASK 0x00ff0000 +#define NV40TCL_VTX_ATTR_4UB_W_SHIFT 24 +#define NV40TCL_VTX_ATTR_4UB_W_MASK 0xff000000 +#define NV40TCL_VTX_ATTR_4I_XY(x) (0x00001980+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_XY__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_XY_X_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_XY_X_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_XY_Y_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_XY_Y_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_4I_ZW(x) (0x00001984+((x)*8)) +#define NV40TCL_VTX_ATTR_4I_ZW__SIZE 0x00000010 +#define NV40TCL_VTX_ATTR_4I_ZW_Z_SHIFT 0 +#define NV40TCL_VTX_ATTR_4I_ZW_Z_MASK 0x0000ffff +#define NV40TCL_VTX_ATTR_4I_ZW_W_SHIFT 16 +#define NV40TCL_VTX_ATTR_4I_ZW_W_MASK 0xffff0000 #define NV40TCL_TEX_OFFSET(x) (0x00001a00+((x)*32)) #define NV40TCL_TEX_OFFSET__SIZE 0x00000010 #define NV40TCL_TEX_FORMAT(x) (0x00001a04+((x)*32)) @@ -5307,6 +5320,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK 0x0000ffff #define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT 16 #define NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK 0xffff0000 +#define NV40TCL_VTX_ATTR_1F(x) (0x00001e40+((x)*4)) +#define NV40TCL_VTX_ATTR_1F__SIZE 0x00000010 #define NV40TCL_VP_UPLOAD_FROM_ID 0x00001e9c #define NV40TCL_VP_START_FROM_ID 0x00001ea0 #define NV40TCL_POINT_SIZE 0x00001ee0 diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index ff0ce6ac810..359e443bb1c 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -55,7 +55,7 @@ nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, { float *v = map; - BEGIN_RING(rankine, NV34TCL_VERTEX_ATTR_4F_X(attrib), 4); + BEGIN_RING(rankine, NV34TCL_VTX_ATTR_4F_X(attrib), 4); switch (ncomp) { case 4: OUT_RINGf(v[0]); diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index a9a939af0c0..1d78324dda8 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -39,7 +39,7 @@ nv40_render_vertex(struct nv40_context *nv40, const struct vertex_header *v) case EMIT_OMIT: break; case EMIT_1F: - BEGIN_RING(curie, 0x1e40 + (hw * 4), 1); + BEGIN_RING(curie, NV40TCL_VTX_ATTR_1F(hw), 1); OUT_RING (fui(v->data[idx][0])); break; case EMIT_2F: @@ -61,7 +61,7 @@ nv40_render_vertex(struct nv40_context *nv40, const struct vertex_header *v) OUT_RING (fui(v->data[idx][3])); break; case EMIT_4UB: - BEGIN_RING(curie, 0x1940 + (hw * 4), 1); + BEGIN_RING(curie, NV40TCL_VTX_ATTR_4UB(hw), 1); OUT_RING (pack_ub4(float_to_ubyte(v->data[idx][0]), float_to_ubyte(v->data[idx][1]), float_to_ubyte(v->data[idx][2]), diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index e5f9bd5668a..93669e6192f 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -149,7 +149,7 @@ nv40_vbo_static_attrib(struct nv40_context *nv40, struct nouveau_stateobj *so, so_data (so, fui(v[1])); break; case 1: - so_method(so, curie, 0x1e40 + (attrib * 4), 1); + so_method(so, curie, NV40TCL_VTX_ATTR_1F(attrib), 1); so_data (so, fui(v[0])); break; default: -- cgit v1.2.3 From 5a01060eb95cb2cb168cb7224ecc805020584c91 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 23 Jun 2008 00:14:21 +1000 Subject: nouveau: update for interface changes + hack around gallium x86_64 bustage --- src/gallium/auxiliary/draw/draw_context.c | 2 ++ src/gallium/drivers/nv10/nv10_context.h | 1 + src/gallium/drivers/nv10/nv10_state.c | 2 ++ src/gallium/drivers/nv10/nv10_vbo.c | 4 +++- src/gallium/drivers/nv30/nv30_fragprog.c | 6 +++--- src/gallium/drivers/nv30/nv30_vertprog.c | 2 +- src/gallium/drivers/nv40/nv40_context.h | 1 + src/gallium/drivers/nv40/nv40_draw.c | 4 +++- src/gallium/drivers/nv40/nv40_fragprog.c | 8 ++++---- src/gallium/drivers/nv40/nv40_state.c | 5 +++-- src/gallium/drivers/nv40/nv40_vertprog.c | 10 +++++----- 11 files changed, 28 insertions(+), 17 deletions(-) diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c index 2f263cf06a9..0e6f55a928b 100644 --- a/src/gallium/auxiliary/draw/draw_context.c +++ b/src/gallium/auxiliary/draw/draw_context.c @@ -63,8 +63,10 @@ struct draw_context *draw_create( void ) if (!draw_pt_init( draw )) goto fail; +#ifndef PIPE_ARCH_X86 if (!draw_vs_init( draw )) goto fail; +#endif return draw; diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 5636dfc9d22..2bdba53db8a 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -66,6 +66,7 @@ struct nv10_context { //struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; float *constbuf[PIPE_SHADER_TYPES][32][4]; + unsigned constbuf_nr[PIPE_SHADER_TYPES]; struct vertex_info vertex_info; diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 4dcb9a31ab1..11664fae2af 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -469,6 +469,8 @@ nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ))) { memcpy(nv10->constbuf[shader], mapped, buf->size); + nv10->constbuf_nr[shader] = + buf->size / (4 * sizeof(float)); ws->buffer_unmap(ws, buf->buffer); } } diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c index 2a334e137d9..f024f534209 100644 --- a/src/gallium/drivers/nv10/nv10_vbo.c +++ b/src/gallium/drivers/nv10/nv10_vbo.c @@ -44,7 +44,9 @@ boolean nv10_draw_elements( struct pipe_context *pipe, draw_set_mapped_element_buffer(draw, 0, NULL); } - draw_set_mapped_constant_buffer(draw, nv10->constbuf[PIPE_SHADER_VERTEX]); + draw_set_mapped_constant_buffer(draw, + nv10->constbuf[PIPE_SHADER_VERTEX], + nv10->constbuf_nr[PIPE_SHADER_VERTEX]); /* draw! */ draw_arrays(nv10->draw, prim, start, count); diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 51000bd6fcd..54d6bea55f7 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -645,7 +645,7 @@ nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc, return FALSE; } - fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; + fpc->attrib_map[fdec->DeclarationRange.First] = hw; return TRUE; } @@ -655,10 +655,10 @@ nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc, { switch (fdec->Semantic.SemanticName) { case TGSI_SEMANTIC_POSITION: - fpc->depth_id = fdec->u.DeclarationRange.First; + fpc->depth_id = fdec->DeclarationRange.First; break; case TGSI_SEMANTIC_COLOR: - fpc->colour_id = fdec->u.DeclarationRange.First; + fpc->colour_id = fdec->DeclarationRange.First; break; default: NOUVEAU_ERR("bad output semantic\n"); diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 71aea3a59cb..1b34d2e4b29 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -530,7 +530,7 @@ nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, return FALSE; } - vpc->output_map[fdec->u.DeclarationRange.First] = hw; + vpc->output_map[fdec->DeclarationRange.First] = hw; return TRUE; } diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index d8d1891dff0..8e60a81e68e 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -136,6 +136,7 @@ struct nv40_context { struct nv40_vertex_program *vertprog; struct nv40_fragment_program *fragprog; struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + unsigned constbuf_nr[PIPE_SHADER_TYPES]; struct nv40_rasterizer_state *rasterizer; struct nv40_zsa_state *zsa; struct nv40_blend_state *blend; diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index 1d78324dda8..2cf58e2950a 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -253,9 +253,11 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe, } if (nv40->constbuf[PIPE_SHADER_VERTEX]) { + const unsigned nr = nv40->constbuf_nr[PIPE_SHADER_VERTEX]; + map = ws->buffer_map(ws, nv40->constbuf[PIPE_SHADER_VERTEX], PIPE_BUFFER_USAGE_CPU_READ); - draw_set_mapped_constant_buffer(nv40->draw, map); + draw_set_mapped_constant_buffer(nv40->draw, map, nr); } draw_arrays(nv40->draw, mode, start, count); diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 4b7667e0381..428348c3388 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -708,7 +708,7 @@ nv40_fragprog_parse_decl_attrib(struct nv40_fpc *fpc, return FALSE; } - fpc->attrib_map[fdec->u.DeclarationRange.First] = hw; + fpc->attrib_map[fdec->DeclarationRange.First] = hw; return TRUE; } @@ -716,7 +716,7 @@ static boolean nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc, const struct tgsi_full_declaration *fdec) { - unsigned idx = fdec->u.DeclarationRange.First; + unsigned idx = fdec->DeclarationRange.First; unsigned hw; switch (fdec->Semantic.SemanticName) { @@ -770,9 +770,9 @@ nv40_fragprog_prepare(struct nv40_fpc *fpc) goto out_err; break; case TGSI_FILE_TEMPORARY: - if (fdec->u.DeclarationRange.Last > high_temp) { + if (fdec->DeclarationRange.Last > high_temp) { high_temp = - fdec->u.DeclarationRange.Last; + fdec->DeclarationRange.Last; } break; default: diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 2d921d2b8a4..afcf336a65b 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -603,12 +603,13 @@ nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, { struct nv40_context *nv40 = nv40_context(pipe); + nv40->constbuf[shader] = buf->buffer; + nv40->constbuf_nr[shader] = buf->size / (4 * sizeof(float)); + if (shader == PIPE_SHADER_VERTEX) { - nv40->constbuf[PIPE_SHADER_VERTEX] = buf->buffer; nv40->dirty |= NV40_NEW_VERTPROG; } else if (shader == PIPE_SHADER_FRAGMENT) { - nv40->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; nv40->dirty |= NV40_NEW_FRAGPROG; } } diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index eb14869bfe0..1e486a66ef0 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -603,7 +603,7 @@ static boolean nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc, const struct tgsi_full_declaration *fdec) { - unsigned idx = fdec->u.DeclarationRange.First; + unsigned idx = fdec->DeclarationRange.First; int hw; switch (fdec->Semantic.SemanticName) { @@ -678,16 +678,16 @@ nv40_vertprog_prepare(struct nv40_vpc *vpc) fdec = &p.FullToken.FullDeclaration; switch (fdec->Declaration.File) { case TGSI_FILE_TEMPORARY: - if (fdec->u.DeclarationRange.Last > high_temp) { + if (fdec->DeclarationRange.Last > high_temp) { high_temp = - fdec->u.DeclarationRange.Last; + fdec->DeclarationRange.Last; } break; #if 0 /* this would be nice.. except gallium doesn't track it */ case TGSI_FILE_ADDRESS: - if (fdec->u.DeclarationRange.Last > high_addr) { + if (fdec->DeclarationRange.Last > high_addr) { high_addr = - fdec->u.DeclarationRange.Last; + fdec->DeclarationRange.Last; } break; #endif -- cgit v1.2.3 From 8b72737a0ba33343673111261265c59546b408c6 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 23 Jun 2008 20:37:27 +0200 Subject: nv30: add state rasterizer, based on nv40 one --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 7 +++++++ src/gallium/drivers/nv30/nv30_state_rasterizer.c | 17 +++++++++++++++++ 3 files changed, 25 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_rasterizer.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index 04f53ee4565..c5208fabf5e 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -16,6 +16,7 @@ DRIVER_SOURCES = \ nv30_state_blend.c \ nv30_state_emit.c \ nv30_state_fb.c \ + nv30_state_rasterizer.c \ nv30_surface.c \ nv30_vbo.c \ nv30_vertprog.c diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 1695ba5a017..43ee8d13a0c 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -76,6 +76,12 @@ enum nv30_state_index { #define NV30_NEW_ARRAYS (1 << 11) #define NV30_NEW_UCP (1 << 12) +/* TODO: rename when removing the old state emitter */ +struct nv30_rasterizer_state_new { + struct pipe_rasterizer_state pipe; + struct nouveau_stateobj *so; +}; + /* TODO: rename when removing the old state emitter */ struct nv30_blend_state_new { struct pipe_blend_state pipe; @@ -111,6 +117,7 @@ struct nv30_context { struct nv30_blend_state_new *blend; struct pipe_blend_color blend_colour; struct pipe_framebuffer_state framebuffer; + struct nv30_rasterizer_state_new *rasterizer; uint32_t rt_enable; struct pipe_buffer *rt[2]; diff --git a/src/gallium/drivers/nv30/nv30_state_rasterizer.c b/src/gallium/drivers/nv30/nv30_state_rasterizer.c new file mode 100644 index 00000000000..6d1b60e043d --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_rasterizer.c @@ -0,0 +1,17 @@ +#include "nv30_context.h" + +static boolean +nv30_state_rasterizer_validate(struct nv30_context *nv30) +{ + so_ref(nv30->rasterizer->so, + &nv30->state.hw[NV30_STATE_RAST]); + return TRUE; +} + +struct nv30_state_entry nv30_state_rasterizer = { + .validate = nv30_state_rasterizer_validate, + .dirty = { + .pipe = NV30_NEW_RAST, + .hw = NV30_STATE_RAST + } +}; -- cgit v1.2.3 From e5bbb18441f34824dc4f9f857b30c71c4ff6466f Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 23 Jun 2008 20:43:22 +0200 Subject: nv30: add state scissor, based on nv40 one --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 3 +++ src/gallium/drivers/nv30/nv30_state_scissor.c | 35 +++++++++++++++++++++++++++ 3 files changed, 39 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_scissor.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index c5208fabf5e..134130c0e00 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -17,6 +17,7 @@ DRIVER_SOURCES = \ nv30_state_emit.c \ nv30_state_fb.c \ nv30_state_rasterizer.c \ + nv30_state_scissor.c \ nv30_surface.c \ nv30_vbo.c \ nv30_vertprog.c diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 43ee8d13a0c..428f17ae0d6 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -90,6 +90,8 @@ struct nv30_blend_state_new { struct nv30_state { + unsigned scissor_enabled; + struct nouveau_stateobj *hw[NV30_STATE_MAX]; }; @@ -104,6 +106,7 @@ struct nv30_context { /* HW state derived from pipe states */ struct nv30_state state; + struct pipe_scissor_state scissor; uint32_t dirty; diff --git a/src/gallium/drivers/nv30/nv30_state_scissor.c b/src/gallium/drivers/nv30/nv30_state_scissor.c new file mode 100644 index 00000000000..1db9bc17955 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_scissor.c @@ -0,0 +1,35 @@ +#include "nv30_context.h" + +static boolean +nv30_state_scissor_validate(struct nv30_context *nv30) +{ + struct pipe_rasterizer_state *rast = &nv30->rasterizer->pipe; + struct pipe_scissor_state *s = &nv30->scissor; + struct nouveau_stateobj *so; + + if (nv30->state.hw[NV30_STATE_SCISSOR] && + (rast->scissor == 0 && nv30->state.scissor_enabled == 0)) + return FALSE; + nv30->state.scissor_enabled = rast->scissor; + + so = so_new(3, 0); + so_method(so, nv30->screen->rankine, NV34TCL_SCISSOR_HORIZ, 2); + if (nv30->state.scissor_enabled) { + so_data (so, ((s->maxx - s->minx) << 16) | s->minx); + so_data (so, ((s->maxy - s->miny) << 16) | s->miny); + } else { + so_data (so, 4096 << 16); + so_data (so, 4096 << 16); + } + + so_ref(so, &nv30->state.hw[NV30_STATE_SCISSOR]); + return TRUE; +} + +struct nv30_state_entry nv30_state_scissor = { + .validate = nv30_state_scissor_validate, + .dirty = { + .pipe = NV30_NEW_SCISSOR | NV30_NEW_RAST, + .hw = NV30_STATE_SCISSOR + } +}; -- cgit v1.2.3 From d4bc56ca49ef39f9f083a2e5adeb3e89ca3bf538 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 23 Jun 2008 21:13:27 +0200 Subject: nv30: add state stipple, based on nv40 one --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 2 ++ src/gallium/drivers/nv30/nv30_state_stipple.c | 39 +++++++++++++++++++++++++++ 3 files changed, 42 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_stipple.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index 134130c0e00..91601087a34 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -18,6 +18,7 @@ DRIVER_SOURCES = \ nv30_state_fb.c \ nv30_state_rasterizer.c \ nv30_state_scissor.c \ + nv30_state_stipple.c \ nv30_surface.c \ nv30_vbo.c \ nv30_vertprog.c diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 428f17ae0d6..446e43a726d 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -91,6 +91,7 @@ struct nv30_blend_state_new { struct nv30_state { unsigned scissor_enabled; + unsigned stipple_enabled; struct nouveau_stateobj *hw[NV30_STATE_MAX]; }; @@ -121,6 +122,7 @@ struct nv30_context { struct pipe_blend_color blend_colour; struct pipe_framebuffer_state framebuffer; struct nv30_rasterizer_state_new *rasterizer; + unsigned stipple[32]; uint32_t rt_enable; struct pipe_buffer *rt[2]; diff --git a/src/gallium/drivers/nv30/nv30_state_stipple.c b/src/gallium/drivers/nv30/nv30_state_stipple.c new file mode 100644 index 00000000000..41b42813b49 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_stipple.c @@ -0,0 +1,39 @@ +#include "nv30_context.h" + +static boolean +nv30_state_stipple_validate(struct nv30_context *nv30) +{ + struct pipe_rasterizer_state *rast = &nv30->rasterizer->pipe; + struct nouveau_grobj *rankine = nv30->screen->rankine; + struct nouveau_stateobj *so; + + if (nv30->state.hw[NV30_STATE_STIPPLE] && + (rast->poly_stipple_enable == 0 && nv30->state.stipple_enabled == 0)) + return FALSE; + + if (rast->poly_stipple_enable) { + unsigned i; + + so = so_new(35, 0); + so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, 1); + so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32); + for (i = 0; i < 32; i++) + so_data(so, nv30->stipple[i]); + } else { + so = so_new(2, 0); + so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, 0); + } + + so_ref(so, &nv30->state.hw[NV30_STATE_STIPPLE]); + return TRUE; +} + +struct nv30_state_entry nv30_state_stipple = { + .validate = nv30_state_stipple_validate, + .dirty = { + .pipe = NV30_NEW_STIPPLE | NV30_NEW_RAST, + .hw = NV30_STATE_STIPPLE, + } +}; -- cgit v1.2.3 From 0e1b36b0b257bfba4427a1e6e12c918e744b9977 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 23 Jun 2008 21:37:41 +0200 Subject: nv30: add state viewport, based on nv40 one --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 2 + src/gallium/drivers/nv30/nv30_state_viewport.c | 70 ++++++++++++++++++++++++++ 3 files changed, 73 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_viewport.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index 91601087a34..a6a8cf1df9c 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -19,6 +19,7 @@ DRIVER_SOURCES = \ nv30_state_rasterizer.c \ nv30_state_scissor.c \ nv30_state_stipple.c \ + nv30_state_viewport.c \ nv30_surface.c \ nv30_vbo.c \ nv30_vertprog.c diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 446e43a726d..eb7e5a8b124 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -92,6 +92,7 @@ struct nv30_blend_state_new { struct nv30_state { unsigned scissor_enabled; unsigned stipple_enabled; + unsigned viewport_bypass; struct nouveau_stateobj *hw[NV30_STATE_MAX]; }; @@ -120,6 +121,7 @@ struct nv30_context { /* Context state */ struct nv30_blend_state_new *blend; struct pipe_blend_color blend_colour; + struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; struct nv30_rasterizer_state_new *rasterizer; unsigned stipple[32]; diff --git a/src/gallium/drivers/nv30/nv30_state_viewport.c b/src/gallium/drivers/nv30/nv30_state_viewport.c new file mode 100644 index 00000000000..951d40ebfdd --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_viewport.c @@ -0,0 +1,70 @@ +#include "nv30_context.h" + +static boolean +nv30_state_viewport_validate(struct nv30_context *nv30) +{ + struct pipe_viewport_state *vpt = &nv30->viewport; + struct nouveau_stateobj *so; + unsigned bypass; + + if (/*nv30->render_mode == HW &&*/ !nv30->rasterizer->pipe.bypass_clipping) + bypass = 0; + else + bypass = 1; + + if (nv30->state.hw[NV30_STATE_VIEWPORT] && + (bypass || !(nv30->dirty & NV30_NEW_VIEWPORT)) && + nv30->state.viewport_bypass == bypass) + return FALSE; + nv30->state.viewport_bypass = bypass; + + so = so_new(11, 0); + if (!bypass) { + so_method(so, nv30->screen->rankine, + NV34TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(vpt->translate[0])); + so_data (so, fui(vpt->translate[1])); + so_data (so, fui(vpt->translate[2])); + so_data (so, fui(vpt->translate[3])); + so_data (so, fui(vpt->scale[0])); + so_data (so, fui(vpt->scale[1])); + so_data (so, fui(vpt->scale[2])); + so_data (so, fui(vpt->scale[3])); +/* so_method(so, nv30->screen->rankine, 0x1d78, 1); + so_data (so, 1); +*/ } else { + so_method(so, nv30->screen->rankine, + NV34TCL_VIEWPORT_TRANSLATE_X, 8); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(0.0)); + so_data (so, fui(1.0)); + so_data (so, fui(1.0)); + so_data (so, fui(1.0)); + so_data (so, fui(0.0)); + /* Not entirely certain what this is yet. The DDX uses this + * value also as it fixes rendering when you pass + * pre-transformed vertices to the GPU. My best gusss is that + * this bypasses some culling/clipping stage. Might be worth + * noting that points/lines are uneffected by whatever this + * value fixes, only filled polygons are effected. + */ +/* so_method(so, nv30->screen->rankine, 0x1d78, 1); + so_data (so, 0x110); +*/ } + /* TODO/FIXME: never saw value 0x0110 in renouveau dumps, only 0x0001 */ + so_method(so, nv30->screen->rankine, 0x1d78, 1); + so_data (so, 1); + + so_ref(so, &nv30->state.hw[NV30_STATE_VIEWPORT]); + return TRUE; +} + +struct nv30_state_entry nv30_state_viewport = { + .validate = nv30_state_viewport_validate, + .dirty = { + .pipe = NV30_NEW_VIEWPORT | NV30_NEW_RAST, + .hw = NV30_STATE_VIEWPORT + } +}; -- cgit v1.2.3 From 95fe122f67024f55d555e2816a95409a8b53a49e Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 23 Jun 2008 21:46:51 +0200 Subject: nv30: add state zsa, based on nv40 one --- src/gallium/drivers/nv30/Makefile | 1 + src/gallium/drivers/nv30/nv30_context.h | 6 ++++++ src/gallium/drivers/nv30/nv30_state_zsa.c | 17 +++++++++++++++++ 3 files changed, 24 insertions(+) create mode 100644 src/gallium/drivers/nv30/nv30_state_zsa.c diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile index a6a8cf1df9c..69f2790dfe2 100644 --- a/src/gallium/drivers/nv30/Makefile +++ b/src/gallium/drivers/nv30/Makefile @@ -20,6 +20,7 @@ DRIVER_SOURCES = \ nv30_state_scissor.c \ nv30_state_stipple.c \ nv30_state_viewport.c \ + nv30_state_zsa.c \ nv30_surface.c \ nv30_vbo.c \ nv30_vertprog.c diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index eb7e5a8b124..dce077b2b4b 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -82,6 +82,11 @@ struct nv30_rasterizer_state_new { struct nouveau_stateobj *so; }; +struct nv30_zsa_state { + struct pipe_depth_stencil_alpha_state pipe; + struct nouveau_stateobj *so; +}; + /* TODO: rename when removing the old state emitter */ struct nv30_blend_state_new { struct pipe_blend_state pipe; @@ -124,6 +129,7 @@ struct nv30_context { struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; struct nv30_rasterizer_state_new *rasterizer; + struct nv30_zsa_state *zsa; unsigned stipple[32]; uint32_t rt_enable; diff --git a/src/gallium/drivers/nv30/nv30_state_zsa.c b/src/gallium/drivers/nv30/nv30_state_zsa.c new file mode 100644 index 00000000000..0940b7269b2 --- /dev/null +++ b/src/gallium/drivers/nv30/nv30_state_zsa.c @@ -0,0 +1,17 @@ +#include "nv30_context.h" + +static boolean +nv30_state_zsa_validate(struct nv30_context *nv30) +{ + so_ref(nv30->zsa->so, + &nv30->state.hw[NV30_STATE_ZSA]); + return TRUE; +} + +struct nv30_state_entry nv30_state_zsa = { + .validate = nv30_state_zsa_validate, + .dirty = { + .pipe = NV30_NEW_ZSA, + .hw = NV30_STATE_ZSA + } +}; -- cgit v1.2.3 From bb1089192572714a604cbef836c34cb16da933f4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 04:08:17 +1000 Subject: nouveau: major thinko --- src/gallium/auxiliary/draw/draw_context.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c index 0e6f55a928b..678c70e4782 100644 --- a/src/gallium/auxiliary/draw/draw_context.c +++ b/src/gallium/auxiliary/draw/draw_context.c @@ -63,7 +63,7 @@ struct draw_context *draw_create( void ) if (!draw_pt_init( draw )) goto fail; -#ifndef PIPE_ARCH_X86 +#ifdef PIPE_ARCH_X86 if (!draw_vs_init( draw )) goto fail; #endif -- cgit v1.2.3 From b831aa06dc0d099185bcaa180683ad10942feaa0 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Tue, 24 Jun 2008 21:04:37 +0200 Subject: nv30: add context value --- src/gallium/drivers/nv30/nv30_screen.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h index 816ece94c4a..b7ddc2a9594 100644 --- a/src/gallium/drivers/nv30/nv30_screen.h +++ b/src/gallium/drivers/nv30/nv30_screen.h @@ -8,6 +8,8 @@ struct nv30_screen { struct nouveau_winsys *nvws; + unsigned cur_pctx; + /* HW graphics objects */ struct nouveau_grobj *rankine; struct nouveau_notifier *sync; -- cgit v1.2.3 From cc8e628563a1cfb26752bc014a75f3087cd8986d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 29 Jun 2008 13:02:31 +1000 Subject: Revert "mesa: Use the appropriate relocation." This reverts commit cb09d1ef60dae3b3c44f1370fd2f726c7044bc17. --- src/mesa/x86-64/glapi_x86-64.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/x86-64/glapi_x86-64.S b/src/mesa/x86-64/glapi_x86-64.S index f8337ff93ee..cb34061b36c 100644 --- a/src/mesa/x86-64/glapi_x86-64.S +++ b/src/mesa/x86-64/glapi_x86-64.S @@ -73,7 +73,7 @@ _x86_64_get_dispatch: .p2align 4,,15 _x86_64_get_dispatch: - movq _gl_DispatchTSD@GOTPCREL(%rip), %rdi + movq _gl_DispatchTSD(%rip), %rdi jmp pthread_getspecific@PLT #elif defined(THREADS) -- cgit v1.2.3 From b6478021d572d9ec30212d6e6992496ee4cf347d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 29 Jun 2008 13:26:04 +1000 Subject: nouveau: adapt to cpp->pf_block changes --- src/gallium/drivers/nv10/nv10_miptree.c | 12 +++-- src/gallium/drivers/nv10/nv10_state_emit.c | 4 +- src/gallium/drivers/nv30/nv30_miptree.c | 10 ++-- src/gallium/drivers/nv30/nv30_state.c | 10 ++-- src/gallium/drivers/nv30/nv30_state_fb.c | 8 +-- src/gallium/drivers/nv40/nv40_miptree.c | 29 ++++++++--- src/gallium/drivers/nv40/nv40_state_fb.c | 10 ++-- src/gallium/drivers/nv40/nv40_surface.c | 2 +- src/gallium/drivers/nv50/nv50_miptree.c | 6 ++- src/gallium/winsys/dri/nouveau/nouveau_context.c | 9 +++- src/gallium/winsys/dri/nouveau/nouveau_screen.c | 7 ++- src/gallium/winsys/dri/nouveau/nv04_surface.c | 63 +++++++++++++----------- src/gallium/winsys/dri/nouveau/nv50_surface.c | 42 +++++++++------- 13 files changed, 129 insertions(+), 83 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index 1b9947354d6..f1486a35df5 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -25,11 +25,13 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; + pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); + pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); if (swizzled) - nv10mt->level[l].pitch = pt->width[l] * pt->cpp; + nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; else - nv10mt->level[l].pitch = pt->width[0] * pt->cpp; + nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; nv10mt->level[l].image_offset = @@ -117,10 +119,12 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, return NULL; pipe_buffer_reference(ws, &ps->buffer, nv10mt->buffer); ps->format = pt->format; - ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; - ps->pitch = nv10mt->level[level].pitch / ps->cpp; + ps->block = pt->block; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->stride = nv10mt->level[level].pitch; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv10mt->level[level].image_offset[face]; diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index 41422c88825..d21368d33f3 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -143,10 +143,10 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10) if (zeta) { BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING ( (rt->pitch * rt->cpp) | ( (zeta->pitch * zeta->cpp) << 16) ); + OUT_RING (rt->stride | (zeta->stride << 16)); } else { BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING ( (rt->pitch * rt->cpp) | ( (rt->pitch * rt->cpp) << 16) ); + OUT_RING (rt->stride | (rt->stride << 16)); } nv10->rt[0] = rt->buffer; diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 6078b1865e2..ad0b257fe2b 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -28,12 +28,14 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; + pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); + pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); if (swizzled) pitch = pt->width[l]; pitch = (pitch + 63) & ~63; - nv30mt->level[l].pitch = pitch * pt->cpp; + nv30mt->level[l].pitch = pitch * pt->block.size; nv30mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); @@ -114,10 +116,12 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); ps->format = pt->format; - ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; - ps->pitch = nv30mt->level[level].pitch / ps->cpp; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->block = pt->block; + ps->stride = nv30mt->level[level].pitch; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv30mt->level[level].image_offset[face]; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 84f016eeadc..c1618041bb0 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -631,21 +631,21 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, } if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { - uint32_t pitch = rt[0]->pitch * rt[0]->cpp; + uint32_t pitch = rt[0]->stride; if (zeta) { - pitch |= (zeta->pitch * zeta->cpp)<<16; + pitch |= (zeta->stride << 16); } else { - pitch |= pitch<<16; + pitch |= (pitch << 16); } BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); - OUT_RING ( pitch ); + OUT_RING (pitch); nv30->rt[0] = rt[0]->buffer; } if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 1); - OUT_RING (rt[1]->pitch * rt[1]->cpp); + OUT_RING (rt[1]->stride); nv30->rt[1] = rt[1]->buffer; } diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c index 73c97e298ad..a20df9f75d1 100644 --- a/src/gallium/drivers/nv30/nv30_state_fb.c +++ b/src/gallium/drivers/nv30/nv30_state_fb.c @@ -59,11 +59,11 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) } if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { - uint32_t pitch = rt[0]->pitch * rt[0]->cpp; + uint32_t pitch = rt[0]->stride; if (zeta) { - pitch |= (zeta->pitch * zeta->cpp)<<16; + pitch |= (zeta->stride << 16); } else { - pitch |= pitch<<16; + pitch |= (pitch << 16); } so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR0, 1); @@ -84,7 +84,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) so_method(so, nv30->screen->rankine, NV34TCL_COLOR1_OFFSET, 2); so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_data (so, rt[1]->pitch * rt[1]->cpp); + so_data (so, rt[1]->stride); } /* if (rt_enable & NV34TCL_RT_ENABLE_COLOR2) { diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 23da6e36a3e..38e1a5f04ca 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -28,19 +28,20 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) pt->width[l] = width; pt->height[l] = height; pt->depth[l] = depth; + pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); + pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); if (swizzled) - pitch = pt->width[l]; - pitch = (pitch + 63) & ~63; + pitch = pt->nblocksx[l]; + pitch = align_int(pitch, 64); - nv40mt->level[l].pitch = pitch * pt->cpp; + nv40mt->level[l].pitch = pitch * pt->block.size; nv40mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); depth = MAX2(1, depth >> 1); - } for (f = 0; f < nr_faces; f++) { @@ -109,15 +110,20 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; struct pipe_surface *ps; - ps = ws->surface_alloc(ws); + ps = CALLOC_STRUCT(pipe_surface); if (!ps) return NULL; + pipe_texture_reference(&ps->texture, pt); pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); ps->format = pt->format; - ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; - ps->pitch = nv40mt->level[level].pitch / ps->cpp; + ps->block = pt->block; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->stride = nv40mt->level[level].pitch; + ps->usage = flags; + ps->status = PIPE_SURFACE_STATUS_DEFINED; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv40mt->level[level].image_offset[face]; @@ -135,6 +141,15 @@ static void nv40_miptree_surface_del(struct pipe_screen *pscreen, struct pipe_surface **psurface) { + struct pipe_surface *ps = *psurface; + + *psurface = NULL; + if (--ps->refcount > 0) + return; + + pipe_texture_reference(&ps->texture, NULL); + pipe_buffer_reference(pscreen->winsys, &ps->buffer, NULL); + FREE(ps); } void diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c index 88baf61ffb3..0e4e60eaa75 100644 --- a/src/gallium/drivers/nv40/nv40_state_fb.c +++ b/src/gallium/drivers/nv40/nv40_state_fb.c @@ -64,7 +64,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) nv40->nvws->channel->vram->handle, nv40->nvws->channel->gart->handle); so_method(so, nv40->screen->curie, NV40TCL_COLOR0_PITCH, 2); - so_data (so, rt[0]->pitch * rt[0]->cpp); + so_data (so, rt[0]->stride); so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); } @@ -77,7 +77,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) so_method(so, nv40->screen->curie, NV40TCL_COLOR1_OFFSET, 2); so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - so_data (so, rt[1]->pitch * rt[1]->cpp); + so_data (so, rt[1]->stride); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) { @@ -89,7 +89,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); so_method(so, nv40->screen->curie, NV40TCL_COLOR2_PITCH, 1); - so_data (so, rt[2]->pitch * rt[2]->cpp); + so_data (so, rt[2]->stride); } if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) { @@ -101,7 +101,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); so_method(so, nv40->screen->curie, NV40TCL_COLOR3_PITCH, 1); - so_data (so, rt[3]->pitch * rt[3]->cpp); + so_data (so, rt[3]->stride); } if (zeta_format) { @@ -113,7 +113,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40) so_reloc (so, zeta->buffer, zeta->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); so_method(so, nv40->screen->curie, NV40TCL_ZETA_PITCH, 1); - so_data (so, zeta->pitch * zeta->cpp); + so_data (so, zeta->stride); } so_method(so, nv40->screen->curie, NV40TCL_RT_ENABLE, 1); diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c index c0d135eb36d..0916555d564 100644 --- a/src/gallium/drivers/nv40/nv40_surface.c +++ b/src/gallium/drivers/nv40/nv40_surface.c @@ -34,7 +34,7 @@ #include "util/p_tile.h" static void -nv40_surface_copy(struct pipe_context *pipe, unsigned do_flip, +nv40_surface_copy(struct pipe_context *pipe, boolean do_flip, struct pipe_surface *dest, unsigned destx, unsigned desty, struct pipe_surface *src, unsigned srcx, unsigned srcy, unsigned width, unsigned height) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index ccb916d6acc..6c838998fd0 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -72,10 +72,12 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, pipe_buffer_reference(ws, &ps->buffer, mt->buffer); ps->format = pt->format; - ps->cpp = pt->cpp; ps->width = pt->width[level]; ps->height = pt->height[level]; - ps->pitch = ps->width; + ps->block = pt->block; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->stride = ps->width * ps->block.size; ps->offset = 0; return ps; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c index ef5fb7ec0d3..74413c408f3 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_context.c @@ -167,9 +167,14 @@ nouveau_context_create(const __GLcontextModes *glVis, fb_buf->bo = &fb_bo->base; fb_surf = calloc(1, sizeof(struct pipe_surface)); - fb_surf->cpp = nv_screen->front_cpp; - fb_surf->pitch = nv_screen->front_pitch / fb_surf->cpp; + if (nv_screen->front_cpp == 2) + fb_surf->format = PIPE_FORMAT_R5G6B5_UNORM; + else + fb_surf->format = PIPE_FORMAT_A8R8G8B8_UNORM; + pf_get_block(fb_surf->format, &fb_surf->block); + fb_surf->width = nv_screen->front_pitch / nv_screen->front_cpp; fb_surf->height = nv_screen->front_height; + fb_surf->stride = fb_surf->width * fb_surf->block.size; fb_surf->refcount = 1; fb_surf->buffer = &fb_buf->base; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c index 9041275a88f..b15ee7509cc 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.c @@ -209,8 +209,9 @@ nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML }; - u_int8_t depth_bits_array[4] = { 0, 16, 24, 24 }; - u_int8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; + uint8_t depth_bits_array[4] = { 0, 16, 24, 24 }; + uint8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; + uint8_t msaa_samples_array[1] = { 0 }; depth_buffer_factor = 4; back_buffer_factor = (have_back_buffer) ? 3 : 1; @@ -229,6 +230,7 @@ nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, depth_buffer_factor, back_buffer_modes, back_buffer_factor, + msaa_samples_array, 1, GLX_TRUE_COLOR)) { fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", __func__, __LINE__ ); @@ -242,6 +244,7 @@ nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, depth_buffer_factor, back_buffer_modes, back_buffer_factor, + msaa_samples_array, 1, GLX_DIRECT_COLOR)) { fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", __func__, __LINE__ ); diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c index 83c790db17a..0085b1c345d 100644 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -1,26 +1,35 @@ #include "pipe/p_context.h" +#include "pipe/p_format.h" #include "nouveau_context.h" static INLINE int -nv04_surface_format(int cpp) +nv04_surface_format(enum pipe_format format) { - switch (cpp) { - case 1: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; - case 2: return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; - case 4: return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; + switch (format) { + case PIPE_FORMAT_A8_UNORM: + return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; + case PIPE_FORMAT_R5G6B5_UNORM: + return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; default: return -1; } } static INLINE int -nv04_rect_format(int cpp) +nv04_rect_format(enum pipe_format format) { - switch (cpp) { - case 1: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - case 2: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; - case 4: return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + switch (format) { + case PIPE_FORMAT_A8_UNORM: + return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + case PIPE_FORMAT_R5G6B5_UNORM: + return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; default: return -1; } @@ -35,8 +44,8 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, struct pipe_surface *src = nv->surf_src; unsigned dst_offset, src_offset; - dst_offset = dst->offset + (dy * dst->pitch + dx) * dst->cpp; - src_offset = src->offset + (sy * src->pitch + sx) * src->cpp; + dst_offset = dst->offset + (dy * dst->stride) + (dx * dst->block.size); + src_offset = src->offset + (sy * src->stride) + (sx * src->block.size); while (h) { int count = (h > 2047) ? 2047 : h; @@ -47,16 +56,16 @@ nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); - OUT_RING (chan, src->pitch * src->cpp); - OUT_RING (chan, dst->pitch * dst->cpp); - OUT_RING (chan, w * src->cpp); + OUT_RING (chan, src->stride); + OUT_RING (chan, dst->stride); + OUT_RING (chan, w * src->block.size); OUT_RING (chan, count); OUT_RING (chan, 0x0101); OUT_RING (chan, 0); h -= count; - src_offset += src->pitch * src->cpp * count; - dst_offset += dst->pitch * dst->cpp * count; + src_offset += src->stride * count; + dst_offset += dst->stride * count; } } @@ -79,7 +88,7 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, struct nouveau_channel *chan = nv->nvc->channel; int format; - if (src->cpp != dst->cpp) + if (src->format != dst->format) return 1; /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback @@ -100,8 +109,8 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, } - if ((format = nv04_surface_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + if ((format = nv04_surface_format(dst->format)) < 0) { + NOUVEAU_ERR("Bad surface format 0x%x\n", dst->format); return 1; } nv->surface_copy = nv04_surface_copy_blit; @@ -116,8 +125,7 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (chan, format); - OUT_RING (chan, ((dst->pitch * dst->cpp) << 16) | - (src->pitch * src->cpp)); + OUT_RING (chan, (dst->stride << 16) | src->stride); OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, @@ -142,13 +150,13 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, struct nouveau_grobj *rect = nv->nvc->NvGdiRect; int cs2d_format, gdirect_format; - if ((cs2d_format = nv04_surface_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + if ((cs2d_format = nv04_surface_format(dst->format)) < 0) { + NOUVEAU_ERR("Bad format = %d\n", dst->format); return 1; } - if ((gdirect_format = nv04_rect_format(dst->cpp)) < 0) { - NOUVEAU_ERR("Bad cpp = %d\n", dst->cpp); + if ((gdirect_format = nv04_rect_format(dst->format)) < 0) { + NOUVEAU_ERR("Bad format = %d\n", dst->format); return 1; } @@ -159,8 +167,7 @@ nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); OUT_RING (chan, cs2d_format); - OUT_RING (chan, ((dst->pitch * dst->cpp) << 16) | - (dst->pitch * dst->cpp)); + OUT_RING (chan, (dst->stride << 16) | dst->stride); OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c index d6da116b774..cf76d76cf52 100644 --- a/src/gallium/winsys/dri/nouveau/nv50_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv50_surface.c @@ -1,15 +1,21 @@ #include "pipe/p_context.h" +#include "pipe/p_format.h" #include "nouveau_context.h" static INLINE int -nv50_format(int cpp) +nv50_format(enum pipe_format format) { - switch (cpp) { - case 4: return NV50_2D_DST_FORMAT_32BPP; - case 3: return NV50_2D_DST_FORMAT_24BPP; - case 2: return NV50_2D_DST_FORMAT_16BPP; - case 1: return NV50_2D_DST_FORMAT_8BPP; + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return NV50_2D_DST_FORMAT_32BPP; + case PIPE_FORMAT_X8R8G8B8_UNORM: + return NV50_2D_DST_FORMAT_24BPP; + case PIPE_FORMAT_R5G6B5_UNORM: + return NV50_2D_DST_FORMAT_16BPP; + case PIPE_FORMAT_A8_UNORM: + return NV50_2D_DST_FORMAT_8BPP; default: return -1; } @@ -23,9 +29,9 @@ nv50_surface_copy_prep(struct nouveau_context *nv, struct nouveau_grobj *eng2d = nv->nvc->Nv2D; int surf_format; - assert(src->cpp == dst->cpp); + assert(src->format == dst->format); - surf_format = nv50_format(dst->cpp); + surf_format = nv50_format(dst->format); assert(surf_format >= 0); BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY0, 2); @@ -38,8 +44,8 @@ nv50_surface_copy_prep(struct nouveau_context *nv, OUT_RING (chan, surf_format); OUT_RING (chan, 1); BEGIN_RING(chan, eng2d, NV50_2D_DST_PITCH, 5); - OUT_RING (chan, dst->pitch * dst->cpp); - OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->stride); + OUT_RING (chan, dst->width); OUT_RING (chan, dst->height); OUT_RELOCh(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); @@ -48,15 +54,15 @@ nv50_surface_copy_prep(struct nouveau_context *nv, BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); OUT_RING (chan, 0); OUT_RING (chan, 0); - OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->width); OUT_RING (chan, dst->height); BEGIN_RING(chan, eng2d, NV50_2D_SRC_FORMAT, 2); OUT_RING (chan, surf_format); OUT_RING (chan, 1); BEGIN_RING(chan, eng2d, NV50_2D_SRC_PITCH, 5); - OUT_RING (chan, src->pitch * src->cpp); - OUT_RING (chan, src->pitch); + OUT_RING (chan, src->stride); + OUT_RING (chan, src->width); OUT_RING (chan, src->height); OUT_RELOCh(chan, nouveau_buffer(src->buffer)->bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); @@ -105,11 +111,11 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, struct nouveau_grobj *eng2d = nv->nvc->Nv2D; int surf_format, rect_format; - surf_format = nv50_format(dst->cpp); + surf_format = nv50_format(dst->format); if (surf_format < 0) return 1; - rect_format = nv50_format(dst->cpp); + rect_format = nv50_format(dst->format); if (rect_format < 0) return 1; @@ -120,8 +126,8 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, OUT_RING (chan, surf_format); OUT_RING (chan, 1); BEGIN_RING(chan, eng2d, NV50_2D_DST_PITCH, 5); - OUT_RING (chan, dst->pitch * dst->cpp); - OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->stride); + OUT_RING (chan, dst->width); OUT_RING (chan, dst->height); OUT_RELOCh(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); @@ -130,7 +136,7 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); OUT_RING (chan, 0); OUT_RING (chan, 0); - OUT_RING (chan, dst->pitch); + OUT_RING (chan, dst->width); OUT_RING (chan, dst->height); BEGIN_RING(chan, eng2d, 0x0580, 3); -- cgit v1.2.3 From f722fd937db2f3cacf1947d538c66528fd16eb89 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 1 Jun 2008 22:41:40 +1000 Subject: nv50: import current "state of the art" nv50 code --- src/gallium/drivers/nv50/Makefile | 1 + src/gallium/drivers/nv50/nv50_context.h | 18 + src/gallium/drivers/nv50/nv50_program.c | 742 +++++++++++++++++++++++++ src/gallium/drivers/nv50/nv50_screen.c | 68 ++- src/gallium/drivers/nv50/nv50_screen.h | 3 + src/gallium/drivers/nv50/nv50_state.c | 54 +- src/gallium/drivers/nv50/nv50_state.h | 30 + src/gallium/drivers/nv50/nv50_state_validate.c | 29 +- src/gallium/drivers/nv50/nv50_vbo.c | 85 ++- 9 files changed, 1017 insertions(+), 13 deletions(-) create mode 100644 src/gallium/drivers/nv50/nv50_program.c diff --git a/src/gallium/drivers/nv50/Makefile b/src/gallium/drivers/nv50/Makefile index a62a4d961bc..2cc6e9de287 100644 --- a/src/gallium/drivers/nv50/Makefile +++ b/src/gallium/drivers/nv50/Makefile @@ -9,6 +9,7 @@ DRIVER_SOURCES = \ nv50_draw.c \ nv50_miptree.c \ nv50_query.c \ + nv50_program.c \ nv50_screen.c \ nv50_state.c \ nv50_state_validate.c \ diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index e68c702deae..d4d716b94bf 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -9,6 +9,7 @@ #include "nouveau/nouveau_winsys.h" #include "nouveau/nouveau_gldefs.h" +#include "nouveau/nouveau_stateobj.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ struct nv50_screen *ctx = nv50->screen @@ -30,6 +31,9 @@ #define NV50_NEW_VIEWPORT (1 << 5) #define NV50_NEW_RASTERIZER (1 << 6) #define NV50_NEW_FRAMEBUFFER (1 << 7) +#define NV50_NEW_VERTPROG (1 << 8) +#define NV50_NEW_FRAGPROG (1 << 9) +#define NV50_NEW_ARRAYS (1 << 10) struct nv50_blend_stateobj { struct pipe_blend_state pipe; @@ -63,6 +67,13 @@ struct nv50_context { struct pipe_scissor_state scissor; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; + struct nv50_program *vertprog; + struct nv50_program *fragprog; + struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + unsigned vtxbuf_nr; + struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; + unsigned vtxelt_nr; }; static INLINE struct nv50_context * @@ -88,11 +99,18 @@ extern boolean nv50_draw_elements(struct pipe_context *pipe, unsigned indexSize, unsigned mode, unsigned start, unsigned count); +extern void nv50_vbo_validate(struct nv50_context *nv50); /* nv50_clear.c */ extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue); +/* nv50_program.c */ +extern void nv50_vertprog_validate(struct nv50_context *nv50); +extern void nv50_fragprog_validate(struct nv50_context *nv50); +extern void nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p); + +/* nv50_state_validate.c */ extern boolean nv50_state_validate(struct nv50_context *nv50); #endif diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c new file mode 100644 index 00000000000..30953b7d8af --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -0,0 +1,742 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_inlines.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_util.h" + +#include "nv50_context.h" +#include "nv50_state.h" + +#define OP_MOV 0x001 +#define OP_RCP 0x009 +#define OP_ADD 0x00b +#define OP_MUL 0x00c +#define NV50_SU_MAX_TEMP 64 + +struct nv50_reg { + enum { + P_TEMP, + P_ATTR, + P_RESULT, + P_CONST, + P_IMMD + } type; + int index; + + int hw; +}; + +struct nv50_pc { + struct nv50_program *p; + + /* hw resources */ + struct nv50_reg *r_temp[NV50_SU_MAX_TEMP]; + + /* tgsi resources */ + struct nv50_reg *temp; + int temp_nr; + struct nv50_reg *attr; + int attr_nr; + struct nv50_reg *result; + int result_nr; + struct nv50_reg *param; + int param_nr; + struct nv50_reg *immd; + float *immd_buf; + int immd_nr; +}; + +static void +alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg) +{ + int i; + + if (reg->type != P_TEMP || reg->hw >= 0) + return; + + for (i = 0; i < NV50_SU_MAX_TEMP; i++) { + if (!(pc->r_temp[i])) { + pc->r_temp[i] = reg; + reg->hw = i; + if (pc->p->cfg.vp.high_temp < (i + 1)) + pc->p->cfg.vp.high_temp = i + 1; + return; + } + } + + assert(0); +} + +static struct nv50_reg * +alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst) +{ + struct nv50_reg *r; + int i; + + if (dst && dst->type == P_TEMP && dst->hw == -1) + return dst; + + for (i = 0; i < NV50_SU_MAX_TEMP; i++) { + if (!pc->r_temp[i]) { + r = CALLOC_STRUCT(nv50_reg); + r->type = P_TEMP; + r->index = -1; + r->hw = i; + pc->r_temp[i] = r; + return r; + } + } + + assert(0); + return NULL; +} + +static void +free_temp(struct nv50_pc *pc, struct nv50_reg *r) +{ + if (r->index == -1) { + FREE(pc->r_temp[r->hw]); + pc->r_temp[r->hw] = NULL; + } +} + +#if 0 +static struct nv50_reg * +constant(struct nv50_pc *pc, int pipe, int c, float v) +{ + struct nv50_reg *r = CALLOC_STRUCT(nv50_reg); + struct nv50_program *p = pc->p; + struct nv50_program_data *pd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < p->nr_consts; idx++) { + if (p->consts[idx].index == pipe) + return nv40_sr(NV40SR_CONST, idx); + } + } + + idx = p->nr_consts++; + p->consts = realloc(p->consts, sizeof(*pd) * p->nr_consts); + pd = &p->consts[idx]; + + pd->index = pipe; + pd->component = c; + pd->value = v; + return nv40_sr(NV40SR_CONST, idx); +} +#endif + +static void +emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, + struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) +{ + struct nv50_program *p = pc->p; + struct nv50_reg *tmp = NULL, *tmp2 = NULL; + unsigned inst[2] = { 0, 0 }; + + /* Grr.. Fun restrictions on where attribs can be sourced from.. */ + if (src1 && src1->type == P_ATTR) { + tmp = alloc_temp(pc, dst); + emit(pc, 1, tmp, src1, NULL, NULL); + src1 = tmp; + } + + if (src2 && src2->type == P_ATTR) { + tmp2 = alloc_temp(pc, dst); + emit(pc, 1, tmp2, src2, NULL, NULL); + src2 = tmp2; + } + + /* Get this out of the way first. What type of opcode do we + * want/need to build? + */ + if ((op & 0x3f0) || dst->type == P_RESULT || + (src0 && src0->type == P_ATTR) || src1 || src2) + inst[0] |= 0x00000001; + + if (inst[0] & 0x00000001) { + inst[0] |= ((op & 0xf) << 28); + inst[1] |= ((op >> 4) << 26); + + alloc_reg(pc, dst); + if (dst->type == P_RESULT) + inst[1] |= 0x00000008; + inst[0] |= (dst->hw << 2); + + if (src0) { + if (src0->type == P_ATTR) + inst[1] |= 0x00200000; + else + if (src0->type == P_CONST || src0->type == P_IMMD) + assert(0); + alloc_reg(pc, src0); + inst[0] |= (src0->hw << 9); + } + + if (src1) { + if (src1->type == P_CONST || src1->type == P_IMMD) { + inst[0] |= 0x00800000; /* src1 is const */ + /*XXX: does src1 come from "src2" now? */ + alloc_reg(pc, src1); + inst[0] |= (src1->hw << 16); + } else { + alloc_reg(pc, src1); + if (op == 0xc || op == 0xe) + inst[0] |= (src1->hw << 16); + else + inst[1] |= (src1->hw << 14); + } + } else { + inst[1] |= 0x0003c000; /*XXX FIXME */ + } + + if (src2) { + if (src2->type == P_CONST || src2->type == P_IMMD) { + inst[0] |= 0x01000000; /* src2 is const */ + inst[1] |= (src2->hw << 14); + } else { + alloc_reg(pc, src2); + if (inst[0] & 0x00800000 || op ==0xe) + inst[1] |= (src2->hw << 14); + else + inst[0] |= (src2->hw << 16); + } + } + + /*XXX: FIXME */ + if (op == 0xb || op == 0xc || op == 0x9 || op == 0xe) { + /* 0x04000000 negates arg0 */ + /* 0x08000000 negates arg1 */ + /*XXX: true for !0xb also ? */ + inst[1] |= 0x00000780; + } else { + /* 0x04000000 == arg0 32 bit, otherwise 16 bit */ + inst[1] |= 0x04000780; + } + } else { + inst[0] |= ((op & 0xf) << 28); + + alloc_reg(pc, dst); + inst[0] |= (dst->hw << 2); + + if (src0) { + alloc_reg(pc, src0); + inst[0] |= (src0->hw << 9); + } + + /*XXX: NFI if this even works - probably not.. */ + if (src1) { + alloc_reg(pc, src1); + inst[0] |= (src1->hw << 16); + } + } + + if (tmp) free_temp(pc, tmp); + if (tmp2) free_temp(pc, tmp2); + + if (inst[0] & 1) { + p->insns_nr += 2; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); + } else { + p->insns_nr += 1; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); + } +} + +static struct nv50_reg * +tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) +{ + switch (dst->DstRegister.File) { + case TGSI_FILE_TEMPORARY: + return &pc->temp[dst->DstRegister.Index * 4 + c]; + case TGSI_FILE_OUTPUT: + return &pc->result[dst->DstRegister.Index * 4 + c]; + case TGSI_FILE_NULL: + return NULL; + default: + break; + } + + return NULL; +} + +static struct nv50_reg * +tgsi_src(struct nv50_pc *pc, int c, const struct tgsi_full_src_register *src) +{ + /* Handle swizzling */ + switch (c) { + case 0: c = src->SrcRegister.SwizzleX; break; + case 1: c = src->SrcRegister.SwizzleY; break; + case 2: c = src->SrcRegister.SwizzleZ; break; + case 3: c = src->SrcRegister.SwizzleW; break; + default: + assert(0); + } + + switch (src->SrcRegister.File) { + case TGSI_FILE_INPUT: + return &pc->attr[src->SrcRegister.Index * 4 + c]; + case TGSI_FILE_TEMPORARY: + return &pc->temp[src->SrcRegister.Index * 4 + c]; + case TGSI_FILE_CONSTANT: + return &pc->param[src->SrcRegister.Index * 4 + c]; + case TGSI_FILE_IMMEDIATE: + return &pc->immd[src->SrcRegister.Index * 4 + c]; + default: + break; + } + + return NULL; +} + +static boolean +nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) +{ + const struct tgsi_full_instruction *inst = &tok->FullInstruction; + struct nv50_reg *dst[4], *src[3][4], *none = NULL, *tmp; + unsigned mask; + int i, c; + + NOUVEAU_ERR("insn %p\n", tok); + + mask = inst->FullDstRegisters[0].DstRegister.WriteMask; + + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + dst[c] = tgsi_dst(pc, c, &inst->FullDstRegisters[0]); + else + dst[c] = NULL; + } + + for (i = 0; i < inst->Instruction.NumSrcRegs; i++) { + for (c = 0; c < 4; c++) + src[i][c] = tgsi_src(pc, c, &inst->FullSrcRegisters[i]); + } + + switch (inst->Instruction.Opcode) { + case TGSI_OPCODE_ADD: + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) { + emit(pc, 0x0b, dst[c], + src[0][c], src[1][c], none); + } + } + break; + case TGSI_OPCODE_DP3: + tmp = alloc_temp(pc, NULL); + emit(pc, 0x0c, tmp, src[0][0], src[1][0], NULL); + emit(pc, 0x0e, tmp, src[0][1], src[1][1], tmp); + emit(pc, 0x0e, tmp, src[0][2], src[1][2], tmp); + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, 0x01, dst[c], tmp, none, none); + } + free_temp(pc, tmp); + break; + case TGSI_OPCODE_DP4: + tmp = alloc_temp(pc, NULL); + emit(pc, 0x0c, tmp, src[0][0], src[1][0], NULL); + emit(pc, 0x0e, tmp, src[0][1], src[1][1], tmp); + emit(pc, 0x0e, tmp, src[0][2], src[1][2], tmp); + emit(pc, 0x0e, tmp, src[0][3], src[1][3], tmp); + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, 0x01, dst[c], tmp, none, none); + } + free_temp(pc, tmp); + break; + case TGSI_OPCODE_MAD: + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, 0x0e, dst[c], + src[0][c], src[1][c], src[2][c]); + } + break; + case TGSI_OPCODE_MOV: + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, 0x01, dst[c], src[0][c], none, none); + } + break; + case TGSI_OPCODE_MUL: + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, 0x0c, dst[c], + src[0][c], src[1][c], none); + } + break; + case TGSI_OPCODE_RCP: + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, 0x09, dst[c], + src[0][c], none, none); + } + break; + case TGSI_OPCODE_END: + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", inst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv50_program_tx_prep(struct nv50_pc *pc) +{ + struct tgsi_parse_context p; + boolean ret = FALSE; + unsigned i, c; + + tgsi_parse_init(&p, pc->p->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch (tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm = + &p.FullToken.FullImmediate; + + pc->immd_nr++; + pc->immd_buf = realloc(pc->immd_buf, 4 * pc->immd_nr * + sizeof(float)); + pc->immd_buf[4 * (pc->immd_nr - 1) + 0] = + imm->u.ImmediateFloat32[0].Float; + pc->immd_buf[4 * (pc->immd_nr - 1) + 1] = + imm->u.ImmediateFloat32[1].Float; + pc->immd_buf[4 * (pc->immd_nr - 1) + 2] = + imm->u.ImmediateFloat32[2].Float; + pc->immd_buf[4 * (pc->immd_nr - 1) + 3] = + imm->u.ImmediateFloat32[3].Float; + } + break; + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *d; + unsigned last; + + d = &p.FullToken.FullDeclaration; + last = d->u.DeclarationRange.Last; + + switch (d->Declaration.File) { + case TGSI_FILE_TEMPORARY: + if (pc->temp_nr < (last + 1)) + pc->temp_nr = last + 1; + break; + case TGSI_FILE_OUTPUT: + if (pc->result_nr < (last + 1)) + pc->result_nr = last + 1; + break; + case TGSI_FILE_INPUT: + if (pc->attr_nr < (last + 1)) + pc->attr_nr = last + 1; + break; + case TGSI_FILE_CONSTANT: + if (pc->param_nr < (last + 1)) + pc->param_nr = last + 1; + break; + default: + NOUVEAU_ERR("bad decl file %d\n", + d->Declaration.File); + goto out_err; + } + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + break; + default: + break; + } + } + + NOUVEAU_ERR("%d temps\n", pc->temp_nr); + if (pc->temp_nr) { + pc->temp = calloc(pc->temp_nr * 4, sizeof(struct nv50_reg)); + if (!pc->temp) + goto out_err; + + for (i = 0; i < pc->temp_nr; i++) { + for (c = 0; c < 4; c++) { + pc->temp[i*4+c].type = P_TEMP; + pc->temp[i*4+c].hw = -1; + pc->temp[i*4+c].index = i; + } + } + } + + NOUVEAU_ERR("%d attrib regs\n", pc->attr_nr); + if (pc->attr_nr) { + int aid = 0; + + pc->attr = calloc(pc->attr_nr * 4, sizeof(struct nv50_reg)); + if (!pc->attr) + goto out_err; + + for (i = 0; i < pc->attr_nr; i++) { + for (c = 0; c < 4; c++) { + pc->p->cfg.vp.attr[aid/32] |= (1 << (aid % 32)); + pc->attr[i*4+c].type = P_ATTR; + pc->attr[i*4+c].hw = aid++; + pc->attr[i*4+c].index = i; + } + } + } + + NOUVEAU_ERR("%d result regs\n", pc->result_nr); + if (pc->result_nr) { + int rid = 0; + + pc->result = calloc(pc->result_nr * 4, sizeof(struct nv50_reg)); + if (!pc->result) + goto out_err; + + for (i = 0; i < pc->result_nr; i++) { + for (c = 0; c < 4; c++) { + pc->result[i*4+c].type = P_RESULT; + pc->result[i*4+c].hw = rid++; + pc->result[i*4+c].index = i; + } + } + } + + NOUVEAU_ERR("%d param regs\n", pc->param_nr); + if (pc->param_nr) { + int rid = 0; + + pc->param = calloc(pc->param_nr * 4, sizeof(struct nv50_reg)); + if (!pc->param) + goto out_err; + + for (i = 0; i < pc->param_nr; i++) { + for (c = 0; c < 4; c++) { + pc->param[i*4+c].type = P_CONST; + pc->param[i*4+c].hw = rid++; + pc->param[i*4+c].index = i; + } + } + } + + if (pc->immd_nr) { + int rid = pc->param_nr * 4; + + pc->immd = calloc(pc->immd_nr * 4, sizeof(struct nv50_reg)); + if (!pc->immd) + goto out_err; + + for (i = 0; i < pc->immd_nr; i++) { + for (c = 0; c < 4; c++) { + pc->immd[i*4+c].type = P_IMMD; + pc->immd[i*4+c].hw = rid++; + pc->immd[i*4+c].index = i; + } + } + } + + ret = TRUE; +out_err: + tgsi_parse_free(&p); + return ret; +} + +static boolean +nv50_program_tx(struct nv50_program *p) +{ + struct tgsi_parse_context parse; + struct nv50_pc *pc; + boolean ret; + + pc = CALLOC_STRUCT(nv50_pc); + if (!pc) + return FALSE; + pc->p = p; + pc->p->cfg.vp.high_temp = 4; + + ret = nv50_program_tx_prep(pc); + if (ret == FALSE) + goto out_cleanup; + + tgsi_parse_init(&parse, pc->p->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&parse)) { + const union tgsi_full_token *tok = &parse.FullToken; + + tgsi_parse_token(&parse); + + switch (tok->Token.Type) { + case TGSI_TOKEN_TYPE_INSTRUCTION: + ret = nv50_program_tx_insn(pc, tok); + if (ret == FALSE) + goto out_err; + break; + default: + break; + } + } + + p->param_nr = pc->param_nr * 4; + p->immd_nr = pc->immd_nr * 4; + p->immd = pc->immd_buf; + +out_err: + tgsi_parse_free(&parse); + +out_cleanup: + return ret; +} + +static void +nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) +{ + struct tgsi_parse_context pc; + + tgsi_parse_init(&pc, p->pipe.tokens); + + if (pc.FullHeader.Processor.Processor == TGSI_PROCESSOR_FRAGMENT) { + p->insns_nr = 8; + p->insns = malloc(p->insns_nr * sizeof(unsigned)); + p->insns[0] = 0x80000000; + p->insns[1] = 0x9000000c; + p->insns[2] = 0x82010600; + p->insns[3] = 0x82020604; + p->insns[4] = 0x80030609; + p->insns[5] = 0x00020780; + p->insns[6] = 0x8004060d; + p->insns[7] = 0x00020781; + } else + if (pc.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX) { + int i; + + if (nv50_program_tx(p) == FALSE) + assert(0); + p->insns[p->insns_nr - 1] |= 0x00000001; + + for (i = 0; i < p->insns_nr; i++) + NOUVEAU_ERR("%d 0x%08x\n", i, p->insns[i]); + } else { + NOUVEAU_ERR("invalid TGSI processor\n"); + tgsi_parse_free(&pc); + return; + } + + tgsi_parse_free(&pc); + + p->translated = TRUE; +} + +void +nv50_vertprog_validate(struct nv50_context *nv50) +{ + struct pipe_winsys *ws = nv50->pipe.winsys; + struct nouveau_winsys *nvws = nv50->screen->nvws; + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nv50_program *p = nv50->vertprog; + struct nouveau_stateobj *so; + void *map; + int i; + + if (!p->translated) { + nv50_program_validate(nv50, p); + if (!p->translated) + assert(0); + } + + if (!p->buffer) + p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); + map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); + memcpy(map, p->insns, p->insns_nr * 4); + ws->buffer_unmap(ws, p->buffer); + + if (p->param_nr) { + float *cb; + + cb = ws->buffer_map(ws, nv50->constbuf[PIPE_SHADER_VERTEX], + PIPE_BUFFER_USAGE_CPU_READ); + for (i = 0; i < p->param_nr; i++) { + BEGIN_RING(tesla, 0x0f00, 2); + OUT_RING (i << 8); + OUT_RING (fui(cb[i])); + } + ws->buffer_unmap(ws, nv50->constbuf[PIPE_SHADER_VERTEX]); + } + + + for (i = 0; i < p->immd_nr; i++) { + BEGIN_RING(tesla, 0x0f00, 2); + OUT_RING ((p->param_nr + i) << 8); + OUT_RING (fui(p->immd[i])); + } + + so = so_new(11, 2); + so_method(so, tesla, NV50TCL_VP_ADDRESS_HIGH, 2); + so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | + NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | + NOUVEAU_BO_LOW, 0, 0); + so_method(so, tesla, 0x1650, 2); + so_data (so, p->cfg.vp.attr[0]); + so_data (so, p->cfg.vp.attr[1]); + so_method(so, tesla, 0x16ac, 2); + so_data (so, 8); + so_data (so, p->cfg.vp.high_temp); + so_method(so, tesla, 0x140c, 1); + so_data (so, 0); /* program start offset */ + so_emit(nv50->screen->nvws, so); + so_ref(NULL, &so); +} + +void +nv50_fragprog_validate(struct nv50_context *nv50) +{ + struct pipe_winsys *ws = nv50->pipe.winsys; + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nv50_program *p = nv50->fragprog; + struct nouveau_stateobj *so; + void *map; + + if (!p->translated) { + nv50_program_validate(nv50, p); + if (!p->translated) + assert(0); + } + + if (!p->buffer) + p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); + map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); + memcpy(map, p->insns, p->insns_nr * 4); + ws->buffer_unmap(ws, p->buffer); + + so = so_new(3, 2); + so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); + so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | + NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | + NOUVEAU_BO_LOW, 0, 0); + so_emit(nv50->screen->nvws, so); + so_ref(NULL, &so); +} + +void +nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p) +{ + struct pipe_winsys *ws = nv50->pipe.winsys; + + if (p->insns_nr) { + if (p->insns) + FREE(p->insns); + p->insns_nr = 0; + } + + if (p->buffer) + pipe_buffer_reference(ws, &p->buffer, NULL); + + p->translated = 0; +} + diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 29c057a1453..fc3eeed9133 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -203,8 +203,19 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) return NULL; } + /* Static constant buffer */ + screen->constbuf = ws->buffer_create(ws, 0, 0, 256 * 4 * 4); + if (nvws->res_init(&screen->vp_data_heap, 0, 256)) { + NOUVEAU_ERR("Error initialising constant buffer\n"); + nv50_screen_destroy(&screen->pipe); + return NULL; + } + /* Static tesla init */ - so = so_new(128, 0); + so = so_new(256, 20); + + so_method(so, screen->tesla, 0x1558, 1); + so_data (so, 1); so_method(so, screen->tesla, NV50TCL_DMA_NOTIFY, 1); so_data (so, screen->sync->handle); so_method(so, screen->tesla, NV50TCL_DMA_IN_MEMORY0(0), @@ -215,6 +226,61 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) NV50TCL_DMA_IN_MEMORY1__SIZE); for (i = 0; i < NV50TCL_DMA_IN_MEMORY1__SIZE; i++) so_data(so, nvws->channel->vram->handle); + so_method(so, screen->tesla, 0x121c, 1); + so_data (so, 1); + + so_method(so, screen->tesla, 0x13bc, 1); + so_data (so, 0x54); + so_method(so, screen->tesla, 0x13ac, 1); + so_data (so, 1); + so_method(so, screen->tesla, 0x16bc, 2); + so_data (so, 0x03020100); + so_data (so, 0x07060504); + so_method(so, screen->tesla, 0x1988, 2); + so_data (so, 0x08040404); + so_data (so, 0x00000004); + so_method(so, screen->tesla, 0x16ac, 2); + so_data (so, 8); + so_data (so, 4); + so_method(so, screen->tesla, 0x16b8, 1); + so_data (so, 8); + + so_method(so, screen->tesla, 0x1280, 3); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0x00001000); + so_method(so, screen->tesla, 0x1280, 3); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0x00014000); + so_method(so, screen->tesla, 0x1280, 3); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0x00024000); + so_method(so, screen->tesla, 0x1280, 3); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0x00034000); + so_method(so, screen->tesla, 0x1280, 3); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0x00040100); + + for (i = 0; i < 16; i++) { + so_method(so, screen->tesla, 0x1080 + (i * 8), 2); + so_data (so, 0x000000ff); + so_data (so, 0xffffffff); + } so_emit(nvws, so); so_ref(NULL, &so); diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index 5a2732e37f4..d63dd485085 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -12,6 +12,9 @@ struct nv50_screen { struct nouveau_grobj *tesla; struct nouveau_notifier *sync; + + struct pipe_buffer *constbuf; + struct nouveau_resource *vp_data_heap; }; static INLINE struct nv50_screen * diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index f42bae0c28d..fd10a383782 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -248,8 +248,8 @@ nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nouveau_stateobj *so = so_new(64, 0); so_method(so, tesla, NV50TCL_DEPTH_WRITE_ENABLE, 1); - so_data (so, cso->depth.writemask ? 1 : 0); - if (cso->depth.enabled) { + so_data (so, 0); //cso->depth.writemask ? 1 : 0); + if (0 && cso->depth.enabled) { so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1); so_data (so, 1); so_method(so, tesla, NV50TCL_DEPTH_TEST_FUNC, 1); @@ -328,34 +328,58 @@ static void * nv50_vp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { - return NULL; + struct nv50_program *p = CALLOC_STRUCT(nv50_program); + + p->pipe = *cso; + tgsi_scan_shader(p->pipe.tokens, &p->info); + return (void *)p; } static void nv50_vp_state_bind(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->vertprog = hwcso; + nv50->dirty |= NV50_NEW_VERTPROG; } static void nv50_vp_state_delete(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50_program_destroy(nv50, hwcso); + FREE(hwcso); } static void * nv50_fp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { - return NULL; + struct nv50_program *p = CALLOC_STRUCT(nv50_program); + + p->pipe = *cso; + tgsi_scan_shader(p->pipe.tokens, &p->info); + return (void *)p; } static void nv50_fp_state_bind(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50->fragprog = hwcso; + nv50->dirty |= NV50_NEW_FRAGPROG; } static void nv50_fp_state_delete(struct pipe_context *pipe, void *hwcso) { + struct nv50_context *nv50 = nv50_context(pipe); + + nv50_program_destroy(nv50, hwcso); + FREE(hwcso); } static void @@ -378,6 +402,16 @@ static void nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, const struct pipe_constant_buffer *buf ) { + struct nv50_context *nv50 = nv50_context(pipe); + + if (shader == PIPE_SHADER_VERTEX) { + nv50->constbuf[PIPE_SHADER_VERTEX] = buf->buffer; + nv50->dirty |= NV50_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv50->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; + nv50->dirty |= NV50_NEW_FRAGPROG; + } } static void @@ -424,12 +458,24 @@ static void nv50_set_vertex_buffers(struct pipe_context *pipe, unsigned count, const struct pipe_vertex_buffer *vb) { + struct nv50_context *nv50 = nv50_context(pipe); + + memcpy(nv50->vtxbuf, vb, sizeof(*vb) * count); + nv50->vtxbuf_nr = count; + + nv50->dirty |= NV50_NEW_ARRAYS; } static void nv50_set_vertex_elements(struct pipe_context *pipe, unsigned count, const struct pipe_vertex_element *ve) { + struct nv50_context *nv50 = nv50_context(pipe); + + memcpy(nv50->vtxelt, ve, sizeof(*ve) * count); + nv50->vtxelt_nr = count; + + nv50->dirty |= NV50_NEW_ARRAYS; } void diff --git a/src/gallium/drivers/nv50/nv50_state.h b/src/gallium/drivers/nv50/nv50_state.h index a3b781d4c61..be0c75ad6ea 100644 --- a/src/gallium/drivers/nv50/nv50_state.h +++ b/src/gallium/drivers/nv50/nv50_state.h @@ -2,6 +2,36 @@ #define __NV50_STATE_H__ #include "pipe/p_state.h" +#include "tgsi/util/tgsi_scan.h" +struct nv50_program_data { + int index; + int component; + float value; +}; + +struct nv50_program { + struct pipe_shader_state pipe; + struct tgsi_shader_info info; + boolean translated; + + unsigned *insns; + unsigned insns_nr; + + struct pipe_buffer *buffer; + + unsigned param_nr; + float *immd; + unsigned immd_nr; + + struct nouveau_resource *data; + + union { + struct { + unsigned high_temp; + unsigned attr[2]; + } vp; + } cfg; +}; #endif diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 68d419f9fc8..4a548378b71 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -15,8 +15,8 @@ nv50_state_validate_fb(struct nv50_context *nv50) h = fb->cbufs[i]->height; gw = 1; } else { - assert(w != fb->cbufs[i]->width); - assert(h != fb->cbufs[i]->height); + assert(w == fb->cbufs[i]->width); + assert(h == fb->cbufs[i]->height); } so_method(so, tesla, NV50TCL_RT_HORIZ(i), 2); @@ -46,6 +46,9 @@ nv50_state_validate_fb(struct nv50_context *nv50) } so_data(so, 0x00000040); so_data(so, 0x00000000); + + so_method(so, tesla, 0x1224, 1); + so_data (so, 1); } if (fb->zsbuf) { @@ -54,11 +57,11 @@ nv50_state_validate_fb(struct nv50_context *nv50) h = fb->zsbuf->height; gw = 1; } else { - assert(w != fb->zsbuf->width); - assert(h != fb->zsbuf->height); + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); } - so_method(so, tesla, NV50TCL_RT_ADDRESS_HIGH(i), 5); + so_method(so, tesla, NV50TCL_ZETA_ADDRESS_HIGH, 5); so_reloc (so, fb->zsbuf->buffer, fb->zsbuf->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, fb->zsbuf->buffer, fb->zsbuf->offset, @@ -83,9 +86,12 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_method(so, tesla, NV50TCL_VIEWPORT_HORIZ, 2); so_data (so, w << 16); so_data (so, h << 16); - so_method(so, tesla, 0xff0, 2); + so_method(so, tesla, 0xff4, 2); so_data (so, w << 16); so_data (so, h << 16); + so_method(so, tesla, 0xdf8, 2); + so_data (so, 0); + so_data (so, h); so_emit(nv50->screen->nvws, so); so_ref(NULL, &so); @@ -108,6 +114,12 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & NV50_NEW_ZSA) so_emit(nvws, nv50->zsa->so); + if (nv50->dirty & NV50_NEW_VERTPROG) + nv50_vertprog_validate(nv50); + + if (nv50->dirty & NV50_NEW_FRAGPROG) + nv50_fragprog_validate(nv50); + if (nv50->dirty & NV50_NEW_RASTERIZER) so_emit(nvws, nv50->rasterizer->so); @@ -150,12 +162,15 @@ nv50_state_validate(struct nv50_context *nv50) so_data (so, fui(nv50->viewport.translate[2])); so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3); so_data (so, fui(nv50->viewport.scale[0])); - so_data (so, fui(nv50->viewport.scale[1])); + so_data (so, fui(-nv50->viewport.scale[1])); so_data (so, fui(nv50->viewport.scale[2])); so_emit(nvws, so); so_ref(NULL, &so); } + if (nv50->dirty & NV50_NEW_ARRAYS) + nv50_vbo_validate(nv50); + nv50->dirty = 0; return TRUE; } diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index b01ce1d42c5..f086c762582 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -5,6 +5,29 @@ #include "nv50_context.h" #include "nv50_state.h" +static INLINE unsigned +nv50_prim(unsigned mode) +{ + switch (mode) { + case PIPE_PRIM_POINTS: return NV50TCL_VERTEX_BEGIN_POINTS; + case PIPE_PRIM_LINES: return NV50TCL_VERTEX_BEGIN_LINES; + case PIPE_PRIM_LINE_LOOP: return NV50TCL_VERTEX_BEGIN_LINE_LOOP; + case PIPE_PRIM_LINE_STRIP: return NV50TCL_VERTEX_BEGIN_LINE_STRIP; + case PIPE_PRIM_TRIANGLES: return NV50TCL_VERTEX_BEGIN_TRIANGLES; + case PIPE_PRIM_TRIANGLE_STRIP: + return NV50TCL_VERTEX_BEGIN_TRIANGLE_STRIP; + case PIPE_PRIM_TRIANGLE_FAN: return NV50TCL_VERTEX_BEGIN_TRIANGLE_FAN; + case PIPE_PRIM_QUADS: return NV50TCL_VERTEX_BEGIN_QUADS; + case PIPE_PRIM_QUAD_STRIP: return NV50TCL_VERTEX_BEGIN_QUAD_STRIP; + case PIPE_PRIM_POLYGON: return NV50TCL_VERTEX_BEGIN_POLYGON; + default: + break; + } + + NOUVEAU_ERR("invalid primitive type %d\n", mode); + return NV50TCL_VERTEX_BEGIN_POINTS; +} + boolean nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) @@ -12,8 +35,22 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, struct nv50_context *nv50 = nv50_context(pipe); nv50_state_validate(nv50); - NOUVEAU_ERR("unimplemented\n"); + + BEGIN_RING(tesla, 0x142c, 1); + OUT_RING (0); + BEGIN_RING(tesla, 0x142c, 1); + OUT_RING (0); + + BEGIN_RING(tesla, NV50TCL_VERTEX_BEGIN, 1); + OUT_RING (nv50_prim(mode)); + BEGIN_RING(tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2); + OUT_RING (start); + OUT_RING (count); + BEGIN_RING(tesla, NV50TCL_VERTEX_END, 1); + OUT_RING (0); + + pipe->flush(pipe, 0, NULL); return TRUE; } @@ -30,3 +67,49 @@ nv50_draw_elements(struct pipe_context *pipe, return TRUE; } +void +nv50_vbo_validate(struct nv50_context *nv50) +{ + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nouveau_stateobj *vtxbuf, *vtxfmt; + int i, vpi = 0; + + vtxbuf = so_new(nv50->vtxelt_nr * 4, nv50->vtxelt_nr * 2); + vtxfmt = so_new(nv50->vtxelt_nr + 1, 0); + so_method(vtxfmt, tesla, 0x1ac0, nv50->vtxelt_nr); + + for (i = 0; i < nv50->vtxelt_nr; i++) { + struct pipe_vertex_element *ve = &nv50->vtxelt[i]; + struct pipe_vertex_buffer *vb = + &nv50->vtxbuf[ve->vertex_buffer_index]; + + switch (ve->src_format) { + case PIPE_FORMAT_R32G32B32_FLOAT: + so_data(vtxfmt, 0x7e100000 | i); + break; + default: + { + char fmt[128]; + pf_sprint_name(fmt, ve->src_format); + NOUVEAU_ERR("invalid vbo format %s\n", fmt); + assert(0); + return; + } + } + + so_method(vtxbuf, tesla, 0x900 + (i * 16), 3); + so_data (vtxbuf, 0x20000000 | vb->pitch); + so_reloc (vtxbuf, vb->buffer, vb->buffer_offset + + ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | + NOUVEAU_BO_HIGH, 0, 0); + so_reloc (vtxbuf, vb->buffer, vb->buffer_offset + + ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | + NOUVEAU_BO_LOW, 0, 0); + } + + so_emit(nv50->screen->nvws, vtxfmt); + so_ref (NULL, &vtxfmt); + so_emit(nv50->screen->nvws, vtxbuf); + so_ref (NULL, &vtxbuf); +} + -- cgit v1.2.3 From 716c1cd2ecbc1e86c0fd747c9fa9e095ded5fd5d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 1 Jun 2008 23:10:31 +1000 Subject: nv50: use "real" constbufs for shaders + tcb uploads --- src/gallium/drivers/nv50/nv50_context.h | 14 ++++++- src/gallium/drivers/nv50/nv50_program.c | 27 +++++------- src/gallium/drivers/nv50/nv50_screen.c | 57 +++++++++++++++----------- src/gallium/drivers/nv50/nv50_screen.h | 3 ++ src/gallium/drivers/nv50/nv50_state.c | 4 +- src/gallium/drivers/nv50/nv50_state.h | 1 - src/gallium/drivers/nv50/nv50_state_validate.c | 29 +++++++++++++ 7 files changed, 90 insertions(+), 45 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index d4d716b94bf..c4a8a4c064e 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -23,6 +23,14 @@ #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); +/* Constant buffer assignment */ +#define NV50_CB_PMISC 0 +#define NV50_CB_PVP 1 +#define NV50_CB_PFP 2 +#define NV50_CB_PGP 3 +#define NV50_CB_TIC 4 +#define NV50_CB_TSC 5 + #define NV50_NEW_BLEND (1 << 0) #define NV50_NEW_ZSA (1 << 1) #define NV50_NEW_BLEND_COLOUR (1 << 2) @@ -32,8 +40,10 @@ #define NV50_NEW_RASTERIZER (1 << 6) #define NV50_NEW_FRAMEBUFFER (1 << 7) #define NV50_NEW_VERTPROG (1 << 8) -#define NV50_NEW_FRAGPROG (1 << 9) -#define NV50_NEW_ARRAYS (1 << 10) +#define NV50_NEW_VERTPROG_CB (1 << 9) +#define NV50_NEW_FRAGPROG (1 << 10) +#define NV50_NEW_FRAGPROG_CB (1 << 11) +#define NV50_NEW_ARRAYS (1 << 12) struct nv50_blend_stateobj { struct pipe_blend_state pipe; diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 30953b7d8af..0a436469238 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -179,6 +179,10 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, if (src1) { if (src1->type == P_CONST || src1->type == P_IMMD) { + if (src1->type == P_IMMD) + inst[1] |= (NV50_CB_PMISC << 22); + else + inst[1] |= (NV50_CB_PVP << 22); inst[0] |= 0x00800000; /* src1 is const */ /*XXX: does src1 come from "src2" now? */ alloc_reg(pc, src1); @@ -196,6 +200,10 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, if (src2) { if (src2->type == P_CONST || src2->type == P_IMMD) { + if (src2->type == P_IMMD) + inst[1] |= (NV50_CB_PMISC << 22); + else + inst[1] |= (NV50_CB_PVP << 22); inst[0] |= 0x01000000; /* src2 is const */ inst[1] |= (src2->hw << 14); } else { @@ -526,7 +534,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) } if (pc->immd_nr) { - int rid = pc->param_nr * 4; + int rid = 0; pc->immd = calloc(pc->immd_nr * 4, sizeof(struct nv50_reg)); if (!pc->immd) @@ -581,7 +589,6 @@ nv50_program_tx(struct nv50_program *p) } } - p->param_nr = pc->param_nr * 4; p->immd_nr = pc->immd_nr * 4; p->immd = pc->immd_buf; @@ -654,23 +661,9 @@ nv50_vertprog_validate(struct nv50_context *nv50) memcpy(map, p->insns, p->insns_nr * 4); ws->buffer_unmap(ws, p->buffer); - if (p->param_nr) { - float *cb; - - cb = ws->buffer_map(ws, nv50->constbuf[PIPE_SHADER_VERTEX], - PIPE_BUFFER_USAGE_CPU_READ); - for (i = 0; i < p->param_nr; i++) { - BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING (i << 8); - OUT_RING (fui(cb[i])); - } - ws->buffer_unmap(ws, nv50->constbuf[PIPE_SHADER_VERTEX]); - } - - for (i = 0; i < p->immd_nr; i++) { BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING ((p->param_nr + i) << 8); + OUT_RING ((NV50_CB_PMISC << 16) | (i << 8)); OUT_RING (fui(p->immd[i])); } diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index fc3eeed9133..6c0810a9cfe 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -203,14 +203,6 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) return NULL; } - /* Static constant buffer */ - screen->constbuf = ws->buffer_create(ws, 0, 0, 256 * 4 * 4); - if (nvws->res_init(&screen->vp_data_heap, 0, 256)) { - NOUVEAU_ERR("Error initialising constant buffer\n"); - nv50_screen_destroy(&screen->pipe); - return NULL; - } - /* Static tesla init */ so = so_new(256, 20); @@ -245,37 +237,56 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_method(so, screen->tesla, 0x16b8, 1); so_data (so, 8); + /* Shared constant buffer */ + screen->constbuf = ws->buffer_create(ws, 0, 0, 256 * 4 * 4); + if (nvws->res_init(&screen->vp_data_heap, 0, 256)) { + NOUVEAU_ERR("Error initialising constant buffer\n"); + nv50_screen_destroy(&screen->pipe); + return NULL; + } + so_method(so, screen->tesla, 0x1280, 3); so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x00001000); + so_data (so, (NV50_CB_PMISC << 16) | 0x00001000); + + /* Texture sampler/image unit setup - we abuse the constant buffer + * upload mechanism for the moment to upload data to the tex config + * blocks. At some point we *may* want to go the NVIDIA way of doing + * things? + */ + screen->tic = ws->buffer_create(ws, 0, 0, 32 * 8 * 4); so_method(so, screen->tesla, 0x1280, 3); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x00014000); - so_method(so, screen->tesla, 0x1280, 3); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_data (so, (NV50_CB_TIC << 16) | 0x0800); + so_method(so, screen->tesla, 0x1574, 3); + so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x00024000); + so_data (so, 0x00000800); + + screen->tsc = ws->buffer_create(ws, 0, 0, 32 * 8 * 4); so_method(so, screen->tesla, 0x1280, 3); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x00034000); - so_method(so, screen->tesla, 0x1280, 3); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_data (so, (NV50_CB_TSC << 16) | 0x0800); + so_method(so, screen->tesla, 0x155c, 3); + so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); - so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM | + so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x00040100); + so_data (so, 0x00000800); + + /* Vertex array limits - max them out */ for (i = 0; i < 16; i++) { so_method(so, screen->tesla, 0x1080 + (i * 8), 2); so_data (so, 0x000000ff); diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index d63dd485085..5acb5003ba4 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -15,6 +15,9 @@ struct nv50_screen { struct pipe_buffer *constbuf; struct nouveau_resource *vp_data_heap; + + struct pipe_buffer *tic; + struct pipe_buffer *tsc; }; static INLINE struct nv50_screen * diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index fd10a383782..ba3d04cede3 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -406,11 +406,11 @@ nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, if (shader == PIPE_SHADER_VERTEX) { nv50->constbuf[PIPE_SHADER_VERTEX] = buf->buffer; - nv50->dirty |= NV50_NEW_VERTPROG; + nv50->dirty |= NV50_NEW_VERTPROG_CB; } else if (shader == PIPE_SHADER_FRAGMENT) { nv50->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer; - nv50->dirty |= NV50_NEW_FRAGPROG; + nv50->dirty |= NV50_NEW_FRAGPROG_CB; } } diff --git a/src/gallium/drivers/nv50/nv50_state.h b/src/gallium/drivers/nv50/nv50_state.h index be0c75ad6ea..9e3876871bd 100644 --- a/src/gallium/drivers/nv50/nv50_state.h +++ b/src/gallium/drivers/nv50/nv50_state.h @@ -20,7 +20,6 @@ struct nv50_program { struct pipe_buffer *buffer; - unsigned param_nr; float *immd; unsigned immd_nr; diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 4a548378b71..05395c6df7c 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -168,6 +168,35 @@ nv50_state_validate(struct nv50_context *nv50) so_ref(NULL, &so); } + if (nv50->dirty & NV50_NEW_VERTPROG_CB) { + so = so_new(4, 2); + so_method(so, tesla, 0x1280, 3); + so_reloc (so, nv50->constbuf[PIPE_SHADER_VERTEX], 0, + NOUVEAU_BO_HIGH | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, + 0, 0); + so_reloc (so, nv50->constbuf[PIPE_SHADER_VERTEX], 0, + NOUVEAU_BO_LOW | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, + 0, 0); + so_data (so, (NV50_CB_PVP << 16) | 0x1000); + so_emit(nvws, so); + so_ref(NULL, &so); + } + + if (nv50->dirty & NV50_NEW_FRAGPROG_CB) { + so = so_new(4, 2); + so_method(so, tesla, 0x1280, 3); + so_reloc (so, nv50->constbuf[PIPE_SHADER_FRAGMENT], 0, + NOUVEAU_BO_HIGH | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, + 0, 0); + so_reloc (so, nv50->constbuf[PIPE_SHADER_FRAGMENT], 0, + NOUVEAU_BO_LOW | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, + 0, 0); + so_data (so, (NV50_CB_PFP << 16) | 0x1000); + so_emit(nvws, so); + so_ref(NULL, &so); + } + + if (nv50->dirty & NV50_NEW_ARRAYS) nv50_vbo_validate(nv50); -- cgit v1.2.3 From 41cd9bddf77ea60f84a957e83ddf098818c95c41 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 1 Jun 2008 23:16:17 +1000 Subject: nv50: fucking horrible hack, I really hate G8x shaders.. --- src/gallium/drivers/nv50/nv50_program.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 0a436469238..1b192b897c4 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -135,10 +135,18 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) { struct nv50_program *p = pc->p; - struct nv50_reg *tmp = NULL, *tmp2 = NULL; + struct nv50_reg *tmp0 = NULL, *tmp = NULL, *tmp2 = NULL; unsigned inst[2] = { 0, 0 }; /* Grr.. Fun restrictions on where attribs can be sourced from.. */ + if (src0 && (src0->type == P_CONST || src0->type == P_IMMD) && + (op == 0xc || op == 0xe)) { + tmp = src1; + src1 = src0; + src0 = tmp; + tmp = NULL; + } + if (src1 && src1->type == P_ATTR) { tmp = alloc_temp(pc, dst); emit(pc, 1, tmp, src1, NULL, NULL); @@ -243,6 +251,7 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, } } + if (tmp0) free_temp(pc, tmp0); if (tmp) free_temp(pc, tmp); if (tmp2) free_temp(pc, tmp2); -- cgit v1.2.3 From 38ce697e5942550888c28bd4859ca2a92f247bf7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 2 Jun 2008 12:08:06 +1000 Subject: nv50: implement SUB --- src/gallium/drivers/nv50/nv50_program.c | 63 +++++++++++++++++++++++---------- 1 file changed, 44 insertions(+), 19 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 1b192b897c4..41ff05dabe7 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -11,9 +11,11 @@ #include "nv50_state.h" #define OP_MOV 0x001 +#define OP_INTERP 0x008 #define OP_RCP 0x009 #define OP_ADD 0x00b #define OP_MUL 0x00c +#define OP_MAD 0x00e #define NV50_SU_MAX_TEMP 64 struct nv50_reg { @@ -27,6 +29,7 @@ struct nv50_reg { int index; int hw; + int neg; }; struct nv50_pc { @@ -140,7 +143,7 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, /* Grr.. Fun restrictions on where attribs can be sourced from.. */ if (src0 && (src0->type == P_CONST || src0->type == P_IMMD) && - (op == 0xc || op == 0xe)) { + (op == OP_MUL || op == OP_MAD)) { tmp = src1; src1 = src0; src0 = tmp; @@ -149,12 +152,14 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, if (src1 && src1->type == P_ATTR) { tmp = alloc_temp(pc, dst); + tmp->neg = src1->neg; src1->neg = 0; emit(pc, 1, tmp, src1, NULL, NULL); src1 = tmp; } if (src2 && src2->type == P_ATTR) { tmp2 = alloc_temp(pc, dst); + tmp2->neg = src2->neg; src2->neg = 0; emit(pc, 1, tmp2, src2, NULL, NULL); src2 = tmp2; } @@ -197,7 +202,7 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, inst[0] |= (src1->hw << 16); } else { alloc_reg(pc, src1); - if (op == 0xc || op == 0xe) + if (op == OP_MUL || op == OP_MAD) inst[0] |= (src1->hw << 16); else inst[1] |= (src1->hw << 14); @@ -216,7 +221,7 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, inst[1] |= (src2->hw << 14); } else { alloc_reg(pc, src2); - if (inst[0] & 0x00800000 || op ==0xe) + if (inst[0] & 0x00800000 || op == OP_MAD) inst[1] |= (src2->hw << 14); else inst[0] |= (src2->hw << 16); @@ -224,14 +229,24 @@ emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, } /*XXX: FIXME */ - if (op == 0xb || op == 0xc || op == 0x9 || op == 0xe) { + switch (op) { + case OP_ADD: + case OP_MUL: + case OP_RCP: + case OP_MAD: /* 0x04000000 negates arg0 */ /* 0x08000000 negates arg1 */ /*XXX: true for !0xb also ? */ + if (src0 && src0->neg) + inst[1] |= 0x04000000; + if (src1 && src1->neg) + inst[1] |= 0x08000000; inst[1] |= 0x00000780; - } else { + break; + default: /* 0x04000000 == arg0 32 bit, otherwise 16 bit */ inst[1] |= 0x04000780; + break; } } else { inst[0] |= ((op & 0xf) << 28); @@ -340,61 +355,71 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) case TGSI_OPCODE_ADD: for (c = 0; c < 4; c++) { if (mask & (1 << c)) { - emit(pc, 0x0b, dst[c], + emit(pc, OP_ADD, dst[c], src[0][c], src[1][c], none); } } break; case TGSI_OPCODE_DP3: tmp = alloc_temp(pc, NULL); - emit(pc, 0x0c, tmp, src[0][0], src[1][0], NULL); - emit(pc, 0x0e, tmp, src[0][1], src[1][1], tmp); - emit(pc, 0x0e, tmp, src[0][2], src[1][2], tmp); + emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); + emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); + emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); for (c = 0; c < 4; c++) { if (mask & (1 << c)) - emit(pc, 0x01, dst[c], tmp, none, none); + emit(pc, OP_MOV, dst[c], tmp, none, none); } free_temp(pc, tmp); break; case TGSI_OPCODE_DP4: tmp = alloc_temp(pc, NULL); - emit(pc, 0x0c, tmp, src[0][0], src[1][0], NULL); - emit(pc, 0x0e, tmp, src[0][1], src[1][1], tmp); - emit(pc, 0x0e, tmp, src[0][2], src[1][2], tmp); - emit(pc, 0x0e, tmp, src[0][3], src[1][3], tmp); + emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); + emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); + emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); + emit(pc, OP_MAD, tmp, src[0][3], src[1][3], tmp); for (c = 0; c < 4; c++) { if (mask & (1 << c)) - emit(pc, 0x01, dst[c], tmp, none, none); + emit(pc, OP_MOV, dst[c], tmp, none, none); } free_temp(pc, tmp); break; case TGSI_OPCODE_MAD: for (c = 0; c < 4; c++) { if (mask & (1 << c)) - emit(pc, 0x0e, dst[c], + emit(pc, OP_MAD, dst[c], src[0][c], src[1][c], src[2][c]); } break; case TGSI_OPCODE_MOV: for (c = 0; c < 4; c++) { if (mask & (1 << c)) - emit(pc, 0x01, dst[c], src[0][c], none, none); + emit(pc, OP_MOV, dst[c], src[0][c], none, none); } break; case TGSI_OPCODE_MUL: for (c = 0; c < 4; c++) { if (mask & (1 << c)) - emit(pc, 0x0c, dst[c], + emit(pc, OP_MUL, dst[c], src[0][c], src[1][c], none); } break; case TGSI_OPCODE_RCP: for (c = 0; c < 4; c++) { if (mask & (1 << c)) - emit(pc, 0x09, dst[c], + emit(pc, OP_RCP, dst[c], src[0][c], none, none); } break; + case TGSI_OPCODE_SUB: + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) { + src[1][c]->neg = 1; + emit(pc, OP_ADD, dst[c], + src[0][c], src[1][c], none); + src[1][c]->neg = 0; + } + } + break; case TGSI_OPCODE_END: break; default: -- cgit v1.2.3 From 207b7974723c6b88aacfa3703a1e049ff35db6a8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 2 Jun 2008 12:12:16 +1000 Subject: nv50: DPH --- src/gallium/drivers/nv50/nv50_program.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 41ff05dabe7..12bd419f4cd 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -383,6 +383,18 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } free_temp(pc, tmp); break; + case TGSI_OPCODE_DPH: + tmp = alloc_temp(pc, NULL); + emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); + emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); + emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); + emit(pc, OP_ADD, tmp, src[1][3], tmp, NULL); + for (c = 0; c < 4; c++) { + if (mask & (1 << c)) + emit(pc, OP_MOV, dst[c], tmp, none, none); + } + free_temp(pc, tmp); + break; case TGSI_OPCODE_MAD: for (c = 0; c < 4; c++) { if (mask & (1 << c)) -- cgit v1.2.3 From 22e0acc466947b203574c88f4964f61ef46ae3fd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 2 Jun 2008 13:01:09 +1000 Subject: nv50: split code/data upload out, fp will use it later on --- src/gallium/drivers/nv50/nv50_program.c | 42 +++++++++++++++++++++------------ 1 file changed, 27 insertions(+), 15 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 12bd419f4cd..97e3eab906f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -684,16 +684,37 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) p->translated = TRUE; } +static void +nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) +{ + int i; + + for (i = 0; i < p->immd_nr; i++) { + BEGIN_RING(tesla, 0x0f00, 2); + OUT_RING ((NV50_CB_PMISC << 16) | (i << 8)); + OUT_RING (fui(p->immd[i])); + } +} + +static void +nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) +{ + struct pipe_winsys *ws = nv50->pipe.winsys; + void *map; + + if (!p->buffer) + p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); + map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); + memcpy(map, p->insns, p->insns_nr * 4); + ws->buffer_unmap(ws, p->buffer); +} + void nv50_vertprog_validate(struct nv50_context *nv50) { - struct pipe_winsys *ws = nv50->pipe.winsys; - struct nouveau_winsys *nvws = nv50->screen->nvws; struct nouveau_grobj *tesla = nv50->screen->tesla; struct nv50_program *p = nv50->vertprog; struct nouveau_stateobj *so; - void *map; - int i; if (!p->translated) { nv50_program_validate(nv50, p); @@ -701,17 +722,8 @@ nv50_vertprog_validate(struct nv50_context *nv50) assert(0); } - if (!p->buffer) - p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); - map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); - memcpy(map, p->insns, p->insns_nr * 4); - ws->buffer_unmap(ws, p->buffer); - - for (i = 0; i < p->immd_nr; i++) { - BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING ((NV50_CB_PMISC << 16) | (i << 8)); - OUT_RING (fui(p->immd[i])); - } + nv50_program_validate_data(nv50, p); + nv50_program_validate_code(nv50, p); so = so_new(11, 2); so_method(so, tesla, NV50TCL_VP_ADDRESS_HIGH, 2); -- cgit v1.2.3 From 55b2fe1047b37d0d86641a252e1c745111030393 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 2 Jun 2008 23:52:58 +1000 Subject: nv50: drop the majority of the old shader code, reimplement, only MOV so far. --- src/gallium/drivers/nv50/nv50_program.c | 411 ++++++++++++-------------------- 1 file changed, 146 insertions(+), 265 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 97e3eab906f..aa848faa496 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -10,13 +10,8 @@ #include "nv50_context.h" #include "nv50_state.h" -#define OP_MOV 0x001 -#define OP_INTERP 0x008 -#define OP_RCP 0x009 -#define OP_ADD 0x00b -#define OP_MUL 0x00c -#define OP_MAD 0x00e #define NV50_SU_MAX_TEMP 64 +#define TX_FRAGPROG 0 struct nv50_reg { enum { @@ -106,181 +101,6 @@ free_temp(struct nv50_pc *pc, struct nv50_reg *r) } } -#if 0 -static struct nv50_reg * -constant(struct nv50_pc *pc, int pipe, int c, float v) -{ - struct nv50_reg *r = CALLOC_STRUCT(nv50_reg); - struct nv50_program *p = pc->p; - struct nv50_program_data *pd; - int idx; - - if (pipe >= 0) { - for (idx = 0; idx < p->nr_consts; idx++) { - if (p->consts[idx].index == pipe) - return nv40_sr(NV40SR_CONST, idx); - } - } - - idx = p->nr_consts++; - p->consts = realloc(p->consts, sizeof(*pd) * p->nr_consts); - pd = &p->consts[idx]; - - pd->index = pipe; - pd->component = c; - pd->value = v; - return nv40_sr(NV40SR_CONST, idx); -} -#endif - -static void -emit(struct nv50_pc *pc, unsigned op, struct nv50_reg *dst, - struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) -{ - struct nv50_program *p = pc->p; - struct nv50_reg *tmp0 = NULL, *tmp = NULL, *tmp2 = NULL; - unsigned inst[2] = { 0, 0 }; - - /* Grr.. Fun restrictions on where attribs can be sourced from.. */ - if (src0 && (src0->type == P_CONST || src0->type == P_IMMD) && - (op == OP_MUL || op == OP_MAD)) { - tmp = src1; - src1 = src0; - src0 = tmp; - tmp = NULL; - } - - if (src1 && src1->type == P_ATTR) { - tmp = alloc_temp(pc, dst); - tmp->neg = src1->neg; src1->neg = 0; - emit(pc, 1, tmp, src1, NULL, NULL); - src1 = tmp; - } - - if (src2 && src2->type == P_ATTR) { - tmp2 = alloc_temp(pc, dst); - tmp2->neg = src2->neg; src2->neg = 0; - emit(pc, 1, tmp2, src2, NULL, NULL); - src2 = tmp2; - } - - /* Get this out of the way first. What type of opcode do we - * want/need to build? - */ - if ((op & 0x3f0) || dst->type == P_RESULT || - (src0 && src0->type == P_ATTR) || src1 || src2) - inst[0] |= 0x00000001; - - if (inst[0] & 0x00000001) { - inst[0] |= ((op & 0xf) << 28); - inst[1] |= ((op >> 4) << 26); - - alloc_reg(pc, dst); - if (dst->type == P_RESULT) - inst[1] |= 0x00000008; - inst[0] |= (dst->hw << 2); - - if (src0) { - if (src0->type == P_ATTR) - inst[1] |= 0x00200000; - else - if (src0->type == P_CONST || src0->type == P_IMMD) - assert(0); - alloc_reg(pc, src0); - inst[0] |= (src0->hw << 9); - } - - if (src1) { - if (src1->type == P_CONST || src1->type == P_IMMD) { - if (src1->type == P_IMMD) - inst[1] |= (NV50_CB_PMISC << 22); - else - inst[1] |= (NV50_CB_PVP << 22); - inst[0] |= 0x00800000; /* src1 is const */ - /*XXX: does src1 come from "src2" now? */ - alloc_reg(pc, src1); - inst[0] |= (src1->hw << 16); - } else { - alloc_reg(pc, src1); - if (op == OP_MUL || op == OP_MAD) - inst[0] |= (src1->hw << 16); - else - inst[1] |= (src1->hw << 14); - } - } else { - inst[1] |= 0x0003c000; /*XXX FIXME */ - } - - if (src2) { - if (src2->type == P_CONST || src2->type == P_IMMD) { - if (src2->type == P_IMMD) - inst[1] |= (NV50_CB_PMISC << 22); - else - inst[1] |= (NV50_CB_PVP << 22); - inst[0] |= 0x01000000; /* src2 is const */ - inst[1] |= (src2->hw << 14); - } else { - alloc_reg(pc, src2); - if (inst[0] & 0x00800000 || op == OP_MAD) - inst[1] |= (src2->hw << 14); - else - inst[0] |= (src2->hw << 16); - } - } - - /*XXX: FIXME */ - switch (op) { - case OP_ADD: - case OP_MUL: - case OP_RCP: - case OP_MAD: - /* 0x04000000 negates arg0 */ - /* 0x08000000 negates arg1 */ - /*XXX: true for !0xb also ? */ - if (src0 && src0->neg) - inst[1] |= 0x04000000; - if (src1 && src1->neg) - inst[1] |= 0x08000000; - inst[1] |= 0x00000780; - break; - default: - /* 0x04000000 == arg0 32 bit, otherwise 16 bit */ - inst[1] |= 0x04000780; - break; - } - } else { - inst[0] |= ((op & 0xf) << 28); - - alloc_reg(pc, dst); - inst[0] |= (dst->hw << 2); - - if (src0) { - alloc_reg(pc, src0); - inst[0] |= (src0->hw << 9); - } - - /*XXX: NFI if this even works - probably not.. */ - if (src1) { - alloc_reg(pc, src1); - inst[0] |= (src1->hw << 16); - } - } - - if (tmp0) free_temp(pc, tmp0); - if (tmp) free_temp(pc, tmp); - if (tmp2) free_temp(pc, tmp2); - - if (inst[0] & 1) { - p->insns_nr += 2; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); - } else { - p->insns_nr += 1; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); - } -} - static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -327,11 +147,147 @@ tgsi_src(struct nv50_pc *pc, int c, const struct tgsi_full_src_register *src) return NULL; } +static void +emit(struct nv50_pc *pc, unsigned *inst) +{ + struct nv50_program *p = pc->p; + + if (inst[0] & 1) { + p->insns_nr += 2; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); + } else { + p->insns_nr += 1; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); + } +} + +static INLINE void set_long(struct nv50_pc *, unsigned *); + +static boolean +is_long(unsigned *inst) +{ + if (inst[0] & 1) + return TRUE; + return FALSE; +} + +static boolean +is_immd(unsigned *inst) +{ + if (is_long(inst) && (inst[1] & 3) == 3) + return TRUE; + return FALSE; +} + +static INLINE void +set_pred(struct nv50_pc *pc, unsigned pred, unsigned idx, unsigned *inst) +{ + set_long(pc, inst); + inst[1] &= ~((0x1f << 7) | (0x3 << 12)); + inst[1] |= (pred << 7) | (idx << 12); +} + +static INLINE void +set_pred_wr(struct nv50_pc *pc, unsigned on, unsigned idx, unsigned *inst) +{ + set_long(pc, inst); + inst[1] &= ~((0x3 << 4) | (1 << 6)); + inst[1] |= (idx << 4) | (on << 6); +} + +static INLINE void +set_long(struct nv50_pc *pc, unsigned *inst) +{ + if (is_long(inst)) + return; + + inst[0] |= 1; + set_pred(pc, 0xf, 0, inst); + set_pred_wr(pc, 0, 0, inst); +} + +static INLINE void +set_dst(struct nv50_pc *pc, struct nv50_reg *dst, unsigned *inst) +{ + if (dst->type == P_RESULT) { + set_long(pc, inst); + inst[1] |= 0x00000008; + } + + alloc_reg(pc, dst); + inst[0] |= (dst->hw << 2); +} + +static INLINE void +set_immd(struct nv50_pc *pc, struct nv50_reg *imm, unsigned *inst) +{ + unsigned val = fui(pc->immd_buf[imm->hw]); /* XXX */ + + set_long(pc, inst); + /*XXX: can't be predicated - bits overlap.. catch cases where both + * are required and avoid them. */ + set_pred(pc, 0, 0, inst); + set_pred_wr(pc, 0, 0, inst); + + inst[1] |= 0x00000002 | 0x00000001; + inst[0] |= (val & 0x3f) << 16; + inst[1] |= (val >> 6) << 2; +} + +static void +emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + int i; + + inst[0] |= 0x10000000; + + set_dst(pc, dst, inst); + + if (dst->type != P_RESULT && src->type == P_IMMD) { + set_immd(pc, src, inst); + /*XXX: 32-bit, but steals part of "half" reg space - need to + * catch and handle this case if/when we do half-regs + */ + inst[0] |= 0x00008000; + } else + if (src->type == P_IMMD || src->type == P_CONST) { + set_long(pc, inst); + if (src->type == P_IMMD) + inst[1] |= (NV50_CB_PMISC << 22); + else + inst[1] |= (NV50_CB_PVP << 22); + inst[0] |= (src->hw << 9); + inst[1] |= 0x20000000; /* src0 const? */ + } else { + if (src->type == P_ATTR) { + set_long(pc, inst); + inst[1] |= 0x00200000; + } + + alloc_reg(pc, src); + inst[0] |= (src->hw << 9); + } + + /* We really should support "half" instructions here at some point, + * but I don't feel confident enough about them yet. + */ + set_long(pc, inst); + if (is_long(inst) && !is_immd(inst)) { + inst[1] |= 0x04000000; /* 32-bit */ + inst[1] |= 0x0003c000; /* "subsubop" 0xf == mov */ + } + + emit(pc, inst); +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { const struct tgsi_full_instruction *inst = &tok->FullInstruction; - struct nv50_reg *dst[4], *src[3][4], *none = NULL, *tmp; + struct nv50_reg *dst[4], *src[3][4]; unsigned mask; int i, c; @@ -352,85 +308,9 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } switch (inst->Instruction.Opcode) { - case TGSI_OPCODE_ADD: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) { - emit(pc, OP_ADD, dst[c], - src[0][c], src[1][c], none); - } - } - break; - case TGSI_OPCODE_DP3: - tmp = alloc_temp(pc, NULL); - emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); - emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); - emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], tmp, none, none); - } - free_temp(pc, tmp); - break; - case TGSI_OPCODE_DP4: - tmp = alloc_temp(pc, NULL); - emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); - emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); - emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); - emit(pc, OP_MAD, tmp, src[0][3], src[1][3], tmp); - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], tmp, none, none); - } - free_temp(pc, tmp); - break; - case TGSI_OPCODE_DPH: - tmp = alloc_temp(pc, NULL); - emit(pc, OP_MUL, tmp, src[0][0], src[1][0], NULL); - emit(pc, OP_MAD, tmp, src[0][1], src[1][1], tmp); - emit(pc, OP_MAD, tmp, src[0][2], src[1][2], tmp); - emit(pc, OP_ADD, tmp, src[1][3], tmp, NULL); - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], tmp, none, none); - } - free_temp(pc, tmp); - break; - case TGSI_OPCODE_MAD: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MAD, dst[c], - src[0][c], src[1][c], src[2][c]); - } - break; case TGSI_OPCODE_MOV: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MOV, dst[c], src[0][c], none, none); - } - break; - case TGSI_OPCODE_MUL: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_MUL, dst[c], - src[0][c], src[1][c], none); - } - break; - case TGSI_OPCODE_RCP: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) - emit(pc, OP_RCP, dst[c], - src[0][c], none, none); - } - break; - case TGSI_OPCODE_SUB: - for (c = 0; c < 4; c++) { - if (mask & (1 << c)) { - src[1][c]->neg = 1; - emit(pc, OP_ADD, dst[c], - src[0][c], src[1][c], none); - src[1][c]->neg = 0; - } - } + for (c = 0; c < 4; c++) + emit_mov(pc, dst[c], src[0][c]); break; case TGSI_OPCODE_END: break; @@ -648,6 +528,7 @@ out_cleanup: static void nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) { +#if TX_FRAGPROG == 0 struct tgsi_parse_context pc; tgsi_parse_init(&pc, p->pipe.tokens); @@ -665,6 +546,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) p->insns[7] = 0x00020781; } else if (pc.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX) { +#endif int i; if (nv50_program_tx(p) == FALSE) @@ -673,6 +555,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) for (i = 0; i < p->insns_nr; i++) NOUVEAU_ERR("%d 0x%08x\n", i, p->insns[i]); +#if TX_FRAGPROG == 0 } else { NOUVEAU_ERR("invalid TGSI processor\n"); tgsi_parse_free(&pc); @@ -680,6 +563,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) } tgsi_parse_free(&pc); +#endif p->translated = TRUE; } @@ -758,11 +642,8 @@ nv50_fragprog_validate(struct nv50_context *nv50) assert(0); } - if (!p->buffer) - p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); - map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); - memcpy(map, p->insns, p->insns_nr * 4); - ws->buffer_unmap(ws, p->buffer); + nv50_program_validate_data(nv50, p); + nv50_program_validate_code(nv50, p); so = so_new(3, 2); so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); -- cgit v1.2.3 From 8ec6415e9fcf876c67bc1624f3eb7dd7624b7791 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 3 Jun 2008 12:37:29 +1000 Subject: nv50: start using interpreter for fragprog too, not hardcoded passthrough --- src/gallium/drivers/nv50/nv50_program.c | 96 ++++++++++++++++++++++++++++----- src/gallium/drivers/nv50/nv50_state.c | 2 + src/gallium/drivers/nv50/nv50_state.h | 8 ++- 3 files changed, 92 insertions(+), 14 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index aa848faa496..9d89f6d0609 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -11,7 +11,7 @@ #include "nv50_state.h" #define NV50_SU_MAX_TEMP 64 -#define TX_FRAGPROG 0 +#define TX_FRAGPROG 1 struct nv50_reg { enum { @@ -52,15 +52,23 @@ alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg) { int i; - if (reg->type != P_TEMP || reg->hw >= 0) + if (reg->type != P_TEMP) return; + if (reg->hw >= 0) { + /*XXX: do this here too to catch FP temp-as-attr usage.. + * not clean, but works */ + if (pc->p->cfg.high_temp < (reg->hw + 1)) + pc->p->cfg.high_temp = reg->hw + 1; + return; + } + for (i = 0; i < NV50_SU_MAX_TEMP; i++) { if (!(pc->r_temp[i])) { pc->r_temp[i] = reg; reg->hw = i; - if (pc->p->cfg.vp.high_temp < (i + 1)) - pc->p->cfg.vp.high_temp = i + 1; + if (pc->p->cfg.high_temp < (i + 1)) + pc->p->cfg.high_temp = i + 1; return; } } @@ -236,6 +244,24 @@ set_immd(struct nv50_pc *pc, struct nv50_reg *imm, unsigned *inst) inst[1] |= (val >> 6) << 2; } +static void +emit_interp(struct nv50_pc *pc, struct nv50_reg *dst, + struct nv50_reg *src, struct nv50_reg *iv, boolean noperspective) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0x80000000; + set_dst(pc, dst, inst); + alloc_reg(pc, iv); + inst[0] |= (iv->hw << 9); + alloc_reg(pc, src); + inst[0] |= (src->hw << 16); + if (noperspective) + inst[0] |= (1 << 25); + + emit(pc, inst); +} + static void emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { @@ -409,20 +435,57 @@ nv50_program_tx_prep(struct nv50_pc *pc) NOUVEAU_ERR("%d attrib regs\n", pc->attr_nr); if (pc->attr_nr) { + struct nv50_reg *iv = NULL, *tmp = NULL; int aid = 0; pc->attr = calloc(pc->attr_nr * 4, sizeof(struct nv50_reg)); if (!pc->attr) goto out_err; + if (pc->p->type == NV50_PROG_FRAGMENT) { + iv = alloc_temp(pc, NULL); + aid++; + } + for (i = 0; i < pc->attr_nr; i++) { + struct nv50_reg *a = &pc->attr[i*4]; + for (c = 0; c < 4; c++) { - pc->p->cfg.vp.attr[aid/32] |= (1 << (aid % 32)); - pc->attr[i*4+c].type = P_ATTR; - pc->attr[i*4+c].hw = aid++; - pc->attr[i*4+c].index = i; + if (pc->p->type == NV50_PROG_FRAGMENT) { + struct nv50_reg *at = + alloc_temp(pc, NULL); + pc->attr[i*4+c].type = at->type; + pc->attr[i*4+c].hw = at->hw; + pc->attr[i*4+c].index = at->index; + } else { + pc->p->cfg.vp.attr[aid/32] |= + (1 << (aid % 32)); + pc->attr[i*4+c].type = P_ATTR; + pc->attr[i*4+c].hw = aid++; + pc->attr[i*4+c].index = i; + } } + + if (pc->p->type != NV50_PROG_FRAGMENT) + continue; + + emit_interp(pc, iv, iv, iv, FALSE); + tmp = alloc_temp(pc, NULL); + { + unsigned inst[2] = { 0, 0 }; + inst[0] = 0x90000000; + inst[0] |= (tmp->hw << 2); + emit(pc, inst); + } + emit_interp(pc, &a[0], &a[0], tmp, TRUE); + emit_interp(pc, &a[1], &a[1], tmp, TRUE); + emit_interp(pc, &a[2], &a[2], tmp, TRUE); + emit_interp(pc, &a[3], &a[3], tmp, TRUE); + free_temp(pc, tmp); } + + if (iv) + free_temp(pc, iv); } NOUVEAU_ERR("%d result regs\n", pc->result_nr); @@ -435,7 +498,10 @@ nv50_program_tx_prep(struct nv50_pc *pc) for (i = 0; i < pc->result_nr; i++) { for (c = 0; c < 4; c++) { - pc->result[i*4+c].type = P_RESULT; + if (pc->p->type == NV50_PROG_FRAGMENT) + pc->result[i*4+c].type = P_TEMP; + else + pc->result[i*4+c].type = P_RESULT; pc->result[i*4+c].hw = rid++; pc->result[i*4+c].index = i; } @@ -492,7 +558,7 @@ nv50_program_tx(struct nv50_program *p) if (!pc) return FALSE; pc->p = p; - pc->p->cfg.vp.high_temp = 4; + pc->p->cfg.high_temp = 4; ret = nv50_program_tx_prep(pc); if (ret == FALSE) @@ -551,6 +617,8 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) if (nv50_program_tx(p) == FALSE) assert(0); + /* *not* sufficient, it's fine if last inst is long and + * NOT immd - otherwise it's fucked fucked fucked */ p->insns[p->insns_nr - 1] |= 0x00000001; for (i = 0; i < p->insns_nr; i++) @@ -620,7 +688,7 @@ nv50_vertprog_validate(struct nv50_context *nv50) so_data (so, p->cfg.vp.attr[1]); so_method(so, tesla, 0x16ac, 2); so_data (so, 8); - so_data (so, p->cfg.vp.high_temp); + so_data (so, p->cfg.high_temp); so_method(so, tesla, 0x140c, 1); so_data (so, 0); /* program start offset */ so_emit(nv50->screen->nvws, so); @@ -645,12 +713,16 @@ nv50_fragprog_validate(struct nv50_context *nv50) nv50_program_validate_data(nv50, p); nv50_program_validate_code(nv50, p); - so = so_new(3, 2); + so = so_new(7, 2); so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); + so_method(so, tesla, 0x198c, 1); + so_data (so, p->cfg.high_temp); + so_method(so, tesla, 0x1414, 1); + so_data (so, 0); /* program start offset */ so_emit(nv50->screen->nvws, so); so_ref(NULL, &so); } diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index ba3d04cede3..48d09c296bb 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -331,6 +331,7 @@ nv50_vp_state_create(struct pipe_context *pipe, struct nv50_program *p = CALLOC_STRUCT(nv50_program); p->pipe = *cso; + p->type = NV50_PROG_VERTEX; tgsi_scan_shader(p->pipe.tokens, &p->info); return (void *)p; } @@ -360,6 +361,7 @@ nv50_fp_state_create(struct pipe_context *pipe, struct nv50_program *p = CALLOC_STRUCT(nv50_program); p->pipe = *cso; + p->type = NV50_PROG_FRAGMENT; tgsi_scan_shader(p->pipe.tokens, &p->info); return (void *)p; } diff --git a/src/gallium/drivers/nv50/nv50_state.h b/src/gallium/drivers/nv50/nv50_state.h index 9e3876871bd..dd352b65856 100644 --- a/src/gallium/drivers/nv50/nv50_state.h +++ b/src/gallium/drivers/nv50/nv50_state.h @@ -15,6 +15,10 @@ struct nv50_program { struct tgsi_shader_info info; boolean translated; + enum { + NV50_PROG_VERTEX, + NV50_PROG_FRAGMENT + } type; unsigned *insns; unsigned insns_nr; @@ -25,9 +29,9 @@ struct nv50_program { struct nouveau_resource *data; - union { + struct { + unsigned high_temp; struct { - unsigned high_temp; unsigned attr[2]; } vp; } cfg; -- cgit v1.2.3 From e55964099b0d47dea80920765daac09b9e2a61a7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 3 Jun 2008 12:38:12 +1000 Subject: nv50: remove hardcoded fp stuff --- src/gallium/drivers/nv50/nv50_program.c | 45 ++++++--------------------------- 1 file changed, 8 insertions(+), 37 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 9d89f6d0609..df29069e628 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -11,7 +11,6 @@ #include "nv50_state.h" #define NV50_SU_MAX_TEMP 64 -#define TX_FRAGPROG 1 struct nv50_reg { enum { @@ -594,44 +593,16 @@ out_cleanup: static void nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) { -#if TX_FRAGPROG == 0 - struct tgsi_parse_context pc; - - tgsi_parse_init(&pc, p->pipe.tokens); - - if (pc.FullHeader.Processor.Processor == TGSI_PROCESSOR_FRAGMENT) { - p->insns_nr = 8; - p->insns = malloc(p->insns_nr * sizeof(unsigned)); - p->insns[0] = 0x80000000; - p->insns[1] = 0x9000000c; - p->insns[2] = 0x82010600; - p->insns[3] = 0x82020604; - p->insns[4] = 0x80030609; - p->insns[5] = 0x00020780; - p->insns[6] = 0x8004060d; - p->insns[7] = 0x00020781; - } else - if (pc.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX) { -#endif - int i; - - if (nv50_program_tx(p) == FALSE) - assert(0); - /* *not* sufficient, it's fine if last inst is long and - * NOT immd - otherwise it's fucked fucked fucked */ - p->insns[p->insns_nr - 1] |= 0x00000001; + int i; - for (i = 0; i < p->insns_nr; i++) - NOUVEAU_ERR("%d 0x%08x\n", i, p->insns[i]); -#if TX_FRAGPROG == 0 - } else { - NOUVEAU_ERR("invalid TGSI processor\n"); - tgsi_parse_free(&pc); - return; - } + if (nv50_program_tx(p) == FALSE) + assert(0); + /* *not* sufficient, it's fine if last inst is long and + * NOT immd - otherwise it's fucked fucked fucked */ + p->insns[p->insns_nr - 1] |= 0x00000001; - tgsi_parse_free(&pc); -#endif + for (i = 0; i < p->insns_nr; i++) + NOUVEAU_ERR("%d 0x%08x\n", i, p->insns[i]); p->translated = TRUE; } -- cgit v1.2.3 From 2a1fb44d75364f2492a1ae5d232218a92b8ca807 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 4 Jun 2008 21:23:14 +1000 Subject: nv50: checkpoint: shader code now exceeds caps of "old" code --- src/gallium/drivers/nv50/nv50_program.c | 279 +++++++++++++++++++++++++++++++- 1 file changed, 272 insertions(+), 7 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index df29069e628..bc0b7b2827d 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -44,6 +44,9 @@ struct nv50_pc { struct nv50_reg *immd; float *immd_buf; int immd_nr; + + struct nv50_reg *temp_temp[4]; + unsigned temp_temp_nr; }; static void @@ -108,6 +111,26 @@ free_temp(struct nv50_pc *pc, struct nv50_reg *r) } } +static struct nv50_reg * +temp_temp(struct nv50_pc *pc) +{ + if (pc->temp_temp_nr >= 4) + assert(0); + + pc->temp_temp[pc->temp_temp_nr] = alloc_temp(pc, NULL); + return pc->temp_temp[pc->temp_temp_nr++]; +} + +static void +kill_temp_temp(struct nv50_pc *pc) +{ + int i; + + for (i = 0; i < pc->temp_temp_nr; i++) + free_temp(pc, pc->temp_temp[i]); + pc->temp_temp_nr = 0; +} + static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -168,6 +191,8 @@ emit(struct nv50_pc *pc, unsigned *inst) p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); } + + kill_temp_temp(pc); } static INLINE void set_long(struct nv50_pc *, unsigned *); @@ -261,11 +286,24 @@ emit_interp(struct nv50_pc *pc, struct nv50_reg *dst, emit(pc, inst); } +static void +set_cseg(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +{ + set_long(pc, inst); + if (src->type == P_IMMD) { + inst[1] |= (NV50_CB_PMISC << 22); + } else { + if (pc->p->type == NV50_PROG_VERTEX) + inst[1] |= (NV50_CB_PVP << 22); + else + inst[1] |= (NV50_CB_PFP << 22); + } +} + static void emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { unsigned inst[2] = { 0, 0 }; - int i; inst[0] |= 0x10000000; @@ -280,10 +318,7 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) } else if (src->type == P_IMMD || src->type == P_CONST) { set_long(pc, inst); - if (src->type == P_IMMD) - inst[1] |= (NV50_CB_PMISC << 22); - else - inst[1] |= (NV50_CB_PVP << 22); + set_cseg(pc, src, inst); inst[0] |= (src->hw << 9); inst[1] |= 0x20000000; /* src0 const? */ } else { @@ -308,11 +343,177 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } +static boolean +check_swap_src_0_1(struct nv50_pc *pc, + struct nv50_reg **s0, struct nv50_reg **s1) +{ + struct nv50_reg *src0 = *s0, *src1 = *s1; + + if (src0->type == P_CONST) { + if (src1->type != P_CONST) { + *s0 = src1; + *s1 = src0; + return TRUE; + } + } else + if (src1->type == P_ATTR) { + if (src0->type != P_ATTR) { + *s0 = src1; + *s1 = src0; + return TRUE; + } + } + + return FALSE; +} + +static void +set_src_0(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +{ + if (src->type == P_ATTR) { + set_long(pc, inst); + inst[1] |= 0x00200000; + } else + if (src->type == P_CONST || src->type == P_IMMD) { + struct nv50_reg *temp = temp_temp(pc); + + emit_mov(pc, temp, src); + src = temp; + } + + alloc_reg(pc, src); + inst[0] |= (src->hw << 9); +} + +static void +set_src_1(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +{ + if (src->type == P_ATTR) { + struct nv50_reg *temp = temp_temp(pc); + + emit_mov(pc, temp, src); + src = temp; + } else + if (src->type == P_CONST || src->type == P_IMMD) { + set_cseg(pc, src, inst); + inst[0] |= 0x00800000; + } + + alloc_reg(pc, src); + inst[0] |= (src->hw << 16); +} + +static void +set_src_2(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +{ + set_long(pc, inst); + + if (src->type == P_ATTR) { + struct nv50_reg *temp = temp_temp(pc); + + emit_mov(pc, temp, src); + src = temp; + } else + if (src->type == P_CONST || src->type == P_IMMD) { + set_cseg(pc, src, inst); + inst[0] |= 0x01000000; + } + + alloc_reg(pc, src); + inst[1] |= (src->hw << 14); +} + +static void +emit_mul(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, + struct nv50_reg *src1) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xc0000000; + + check_swap_src_0_1(pc, &src0, &src1); + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_1(pc, src1, inst); + + emit(pc, inst); +} + +static void +emit_add(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, + struct nv50_reg *src1) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xb0000000; + + check_swap_src_0_1(pc, &src0, &src1); + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_2(pc, src1, inst); + + emit(pc, inst); +} + +static void +emit_sub(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, + struct nv50_reg *src1) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xb0000000; + + set_long(pc, inst); + if (check_swap_src_0_1(pc, &src0, &src1)) + inst[1] |= 0x04000000; + else + inst[1] |= 0x08000000; + + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_2(pc, src1, inst); + + emit(pc, inst); +} + +static void +emit_mad(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, + struct nv50_reg *src1, struct nv50_reg *src2) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xe0000000; + + check_swap_src_0_1(pc, &src0, &src1); + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_1(pc, src1, inst); + set_src_2(pc, src2, inst); + + emit(pc, inst); +} + +static void +emit_flop(struct nv50_pc *pc, unsigned sub, + struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + + set_long(pc, inst); + inst[0] |= 0x90000000; + inst[1] |= (sub << 29); + + set_dst(pc, dst, inst); + set_src_0(pc, src, inst); + + emit(pc, inst); +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { const struct tgsi_full_instruction *inst = &tok->FullInstruction; - struct nv50_reg *dst[4], *src[3][4]; + struct nv50_reg *dst[4], *src[3][4], *temp; unsigned mask; int i, c; @@ -333,10 +534,69 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } switch (inst->Instruction.Opcode) { + case TGSI_OPCODE_ADD: + for (c = 0; c < 4; c++) + emit_add(pc, dst[c], src[0][c], src[1][c]); + break; + case TGSI_OPCODE_COS: + for (c = 0; c < 4; c++) + emit_flop(pc, 5, dst[c], src[0][c]); + break; + case TGSI_OPCODE_DP3: + temp = alloc_temp(pc, NULL); + emit_mul(pc, temp, src[0][0], src[1][0]); + emit_mad(pc, temp, src[0][1], src[1][1], temp); + emit_mad(pc, temp, src[0][2], src[1][2], temp); + for (c = 0; c < 4; c++) + emit_mov(pc, dst[c], temp); + free_temp(pc, temp); + break; + case TGSI_OPCODE_DP4: + temp = alloc_temp(pc, NULL); + emit_mul(pc, temp, src[0][0], src[1][0]); + emit_mad(pc, temp, src[0][1], src[1][1], temp); + emit_mad(pc, temp, src[0][2], src[1][2], temp); + emit_mad(pc, temp, src[0][3], src[1][3], temp); + for (c = 0; c < 4; c++) + emit_mov(pc, dst[c], temp); + free_temp(pc, temp); + break; + case TGSI_OPCODE_EX2: + for (c = 0; c < 4; c++) + emit_flop(pc, 6, dst[c], src[0][c]); + break; + case TGSI_OPCODE_LG2: + for (c = 0; c < 4; c++) + emit_flop(pc, 3, dst[c], src[0][c]); + break; + case TGSI_OPCODE_MAD: + for (c = 0; c < 4; c++) + emit_mad(pc, dst[c], src[0][c], src[1][c], src[2][c]); + break; case TGSI_OPCODE_MOV: for (c = 0; c < 4; c++) emit_mov(pc, dst[c], src[0][c]); break; + case TGSI_OPCODE_MUL: + for (c = 0; c < 4; c++) + emit_mul(pc, dst[c], src[0][c], src[1][c]); + break; + case TGSI_OPCODE_RCP: + for (c = 0; c < 4; c++) + emit_flop(pc, 0, dst[c], src[0][c]); + break; + case TGSI_OPCODE_RSQ: + for (c = 0; c < 4; c++) + emit_flop(pc, 2, dst[c], src[0][c]); + break; + case TGSI_OPCODE_SIN: + for (c = 0; c < 4; c++) + emit_flop(pc, 4, dst[c], src[0][c]); + break; + case TGSI_OPCODE_SUB: + for (c = 0; c < 4; c++) + emit_sub(pc, dst[c], src[0][c], src[1][c]); + break; case TGSI_OPCODE_END: break; default: @@ -601,8 +861,13 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) * NOT immd - otherwise it's fucked fucked fucked */ p->insns[p->insns_nr - 1] |= 0x00000001; + if (p->type == NV50_PROG_VERTEX) { + for (i = 0; i < p->insns_nr; i++) + NOUVEAU_ERR("VP0x%08x\n", p->insns[i]); + } else { for (i = 0; i < p->insns_nr; i++) - NOUVEAU_ERR("%d 0x%08x\n", i, p->insns[i]); + NOUVEAU_ERR("FP0x%08x\n", p->insns[i]); + } p->translated = TRUE; } -- cgit v1.2.3 From 52a69196c1680ff16d1ad1fc88e5869bc6055d00 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 4 Jun 2008 21:45:32 +1000 Subject: nv50: some fixes + MIN/MAX --- src/gallium/drivers/nv50/nv50_program.c | 48 ++++++++++++++++++++++++++------- 1 file changed, 39 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index bc0b7b2827d..afc911cd265 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -45,7 +45,7 @@ struct nv50_pc { float *immd_buf; int immd_nr; - struct nv50_reg *temp_temp[4]; + struct nv50_reg *temp_temp[8]; unsigned temp_temp_nr; }; @@ -114,7 +114,7 @@ free_temp(struct nv50_pc *pc, struct nv50_reg *r) static struct nv50_reg * temp_temp(struct nv50_pc *pc) { - if (pc->temp_temp_nr >= 4) + if (pc->temp_temp_nr >= 8) assert(0); pc->temp_temp[pc->temp_temp_nr] = alloc_temp(pc, NULL); @@ -191,8 +191,6 @@ emit(struct nv50_pc *pc, unsigned *inst) p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); } - - kill_temp_temp(pc); } static INLINE void set_long(struct nv50_pc *, unsigned *); @@ -440,8 +438,8 @@ emit_mul(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, } static void -emit_add(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, - struct nv50_reg *src1) +emit_add(struct nv50_pc *pc, struct nv50_reg *dst, + struct nv50_reg *src0, struct nv50_reg *src1) { unsigned inst[2] = { 0, 0 }; @@ -450,7 +448,28 @@ emit_add(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, check_swap_src_0_1(pc, &src0, &src1); set_dst(pc, dst, inst); set_src_0(pc, src0, inst); - set_src_2(pc, src1, inst); + if (is_long(inst)) + set_src_2(pc, src1, inst); + else + set_src_1(pc, src1, inst); + + emit(pc, inst); +} + +static void +emit_minmax(struct nv50_pc *pc, unsigned sub, struct nv50_reg *dst, + struct nv50_reg *src0, struct nv50_reg *src1) +{ + unsigned inst[2] = { 0, 0 }; + + set_long(pc, inst); + inst[0] |= 0xb0000000; + inst[1] |= (sub << 29); + + check_swap_src_0_1(pc, &src0, &src1); + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_1(pc, src1, inst); emit(pc, inst); } @@ -499,9 +518,11 @@ emit_flop(struct nv50_pc *pc, unsigned sub, { unsigned inst[2] = { 0, 0 }; - set_long(pc, inst); inst[0] |= 0x90000000; - inst[1] |= (sub << 29); + if (sub) { + set_long(pc, inst); + inst[1] |= (sub << 29); + } set_dst(pc, dst, inst); set_src_0(pc, src, inst); @@ -573,6 +594,14 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) for (c = 0; c < 4; c++) emit_mad(pc, dst[c], src[0][c], src[1][c], src[2][c]); break; + case TGSI_OPCODE_MAX: + for (c = 0; c < 4; c++) + emit_minmax(pc, 4, dst[c], src[0][c], src[1][c]); + break; + case TGSI_OPCODE_MIN: + for (c = 0; c < 4; c++) + emit_minmax(pc, 5, dst[c], src[0][c], src[1][c]); + break; case TGSI_OPCODE_MOV: for (c = 0; c < 4; c++) emit_mov(pc, dst[c], src[0][c]); @@ -604,6 +633,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) return FALSE; } + kill_temp_temp(pc); return TRUE; } -- cgit v1.2.3 From 7df7f7bb99441ed8e2fba2840e0459e72691f272 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 5 Jun 2008 18:39:00 +1000 Subject: nv50: big fuckup, fix it --- src/gallium/drivers/nv50/nv50_program.c | 75 ++++++++++++++++++++++++++------- 1 file changed, 60 insertions(+), 15 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index afc911cd265..a293db6f723 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -556,20 +556,29 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) switch (inst->Instruction.Opcode) { case TGSI_OPCODE_ADD: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_add(pc, dst[c], src[0][c], src[1][c]); + } break; case TGSI_OPCODE_COS: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_flop(pc, 5, dst[c], src[0][c]); + } break; case TGSI_OPCODE_DP3: temp = alloc_temp(pc, NULL); emit_mul(pc, temp, src[0][0], src[1][0]); emit_mad(pc, temp, src[0][1], src[1][1], temp); emit_mad(pc, temp, src[0][2], src[1][2], temp); - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_mov(pc, dst[c], temp); + } free_temp(pc, temp); break; case TGSI_OPCODE_DP4: @@ -578,53 +587,89 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_mad(pc, temp, src[0][1], src[1][1], temp); emit_mad(pc, temp, src[0][2], src[1][2], temp); emit_mad(pc, temp, src[0][3], src[1][3], temp); - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_mov(pc, dst[c], temp); + } free_temp(pc, temp); break; case TGSI_OPCODE_EX2: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_flop(pc, 6, dst[c], src[0][c]); + } break; case TGSI_OPCODE_LG2: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_flop(pc, 3, dst[c], src[0][c]); + } break; case TGSI_OPCODE_MAD: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_mad(pc, dst[c], src[0][c], src[1][c], src[2][c]); + } break; case TGSI_OPCODE_MAX: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_minmax(pc, 4, dst[c], src[0][c], src[1][c]); + } break; case TGSI_OPCODE_MIN: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_minmax(pc, 5, dst[c], src[0][c], src[1][c]); + } break; case TGSI_OPCODE_MOV: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_mov(pc, dst[c], src[0][c]); + } break; case TGSI_OPCODE_MUL: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_mul(pc, dst[c], src[0][c], src[1][c]); + } break; case TGSI_OPCODE_RCP: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_flop(pc, 0, dst[c], src[0][c]); + } break; case TGSI_OPCODE_RSQ: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_flop(pc, 2, dst[c], src[0][c]); + } break; case TGSI_OPCODE_SIN: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_flop(pc, 4, dst[c], src[0][c]); + } break; case TGSI_OPCODE_SUB: - for (c = 0; c < 4; c++) + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; emit_sub(pc, dst[c], src[0][c], src[1][c]); + } break; case TGSI_OPCODE_END: break; -- cgit v1.2.3 From fbf4027dd9b279ec159906dcad134f71e34aaec8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 5 Jun 2008 18:52:16 +1000 Subject: nv50: fix EX2.. somehow --- src/gallium/drivers/nv50/nv50_program.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index a293db6f723..6d3d8ff302b 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -595,11 +595,22 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) free_temp(pc, temp); break; case TGSI_OPCODE_EX2: + temp = alloc_temp(pc, NULL); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_flop(pc, 6, dst[c], src[0][c]); + { + unsigned inst[2] = { 0, 0 }; + inst[0] |= 0xb0000000; + set_dst(pc, temp, inst); + set_src_0(pc, src[0][c], inst); + set_long(pc, inst); + inst[1] |= (6 << 29) | 0x00004000; + emit(pc, inst); + } + emit_flop(pc, 6, dst[c], temp); } + free_temp(pc, temp); break; case TGSI_OPCODE_LG2: for (c = 0; c < 4; c++) { -- cgit v1.2.3 From 21e18e2b74d71c93af06ef4c603ca371c4614237 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 5 Jun 2008 23:48:23 +1000 Subject: nv50: DPH, XPD, some TODOs --- src/gallium/drivers/nv50/nv50_program.c | 61 +++++++++++++++++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 6d3d8ff302b..91f24369f14 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -12,6 +12,25 @@ #define NV50_SU_MAX_TEMP 64 +/* ABS + * ARL + * DST - const(1.0) + * FLR + * FRC + * LIT + * POW + * SGE - cvt + * SLT - cvt + * SWZ + * + * MSB - Like MAD, but MUL+SUB + * - Fuck it off, introduce a way to negate args for ops that + * support it. + * + * Need ability to specifiy driver IMMD values, like nv40 constant() + * + * Look into inlining IMMD for ops other than MOV + */ struct nv50_reg { enum { P_TEMP, @@ -512,6 +531,25 @@ emit_mad(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, emit(pc, inst); } +static void +emit_msb(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, + struct nv50_reg *src1, struct nv50_reg *src2) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xe0000000; + set_long(pc, inst); + inst[1] |= 0x08000000; /* src0 * src1 - src2 */ + + check_swap_src_0_1(pc, &src0, &src1); + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_1(pc, src1, inst); + set_src_2(pc, src2, inst); + + emit(pc, inst); +} + static void emit_flop(struct nv50_pc *pc, unsigned sub, struct nv50_reg *dst, struct nv50_reg *src) @@ -594,6 +632,19 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } free_temp(pc, temp); break; + case TGSI_OPCODE_DPH: + temp = alloc_temp(pc, NULL); + emit_mul(pc, temp, src[0][0], src[1][0]); + emit_mad(pc, temp, src[0][1], src[1][1], temp); + emit_mad(pc, temp, src[0][2], src[1][2], temp); + emit_add(pc, temp, src[1][3], temp); + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_mov(pc, dst[c], temp); + } + free_temp(pc, temp); + break; case TGSI_OPCODE_EX2: temp = alloc_temp(pc, NULL); for (c = 0; c < 4; c++) { @@ -682,6 +733,16 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_sub(pc, dst[c], src[0][c], src[1][c]); } break; + case TGSI_OPCODE_XPD: + temp = alloc_temp(pc, NULL); + emit_mul(pc, temp, src[0][2], src[1][1]); + emit_msb(pc, dst[0], src[0][1], src[1][2], temp); + emit_mul(pc, temp, src[0][0], src[1][2]); + emit_msb(pc, dst[1], src[0][2], src[1][0], temp); + emit_mul(pc, temp, src[0][1], src[1][0]); + emit_msb(pc, dst[2], src[0][0], src[1][1], temp); + free_temp(pc, temp); + break; case TGSI_OPCODE_END: break; default: -- cgit v1.2.3 From d69f33423087fc054181c60724f4bcbe29195e08 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 6 Jun 2008 13:57:59 +1000 Subject: nv50: small cleanup --- src/gallium/drivers/nv50/nv50_program.c | 27 ++++++++++++++++----------- 1 file changed, 16 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 91f24369f14..7295c2314cf 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -568,6 +568,21 @@ emit_flop(struct nv50_pc *pc, unsigned sub, emit(pc, inst); } +static void +emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xb0000000; + + set_dst(pc, dst, inst); + set_src_0(pc, src, inst); + set_long(pc, inst); + inst[1] |= (6 << 29) | 0x00004000; + + emit(pc, inst); +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { @@ -650,15 +665,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - { - unsigned inst[2] = { 0, 0 }; - inst[0] |= 0xb0000000; - set_dst(pc, temp, inst); - set_src_0(pc, src[0][c], inst); - set_long(pc, inst); - inst[1] |= (6 << 29) | 0x00004000; - emit(pc, inst); - } + emit_preex2(pc, temp, src[0][c]); emit_flop(pc, 6, dst[c], temp); } free_temp(pc, temp); @@ -1081,11 +1088,9 @@ nv50_vertprog_validate(struct nv50_context *nv50) void nv50_fragprog_validate(struct nv50_context *nv50) { - struct pipe_winsys *ws = nv50->pipe.winsys; struct nouveau_grobj *tesla = nv50->screen->tesla; struct nv50_program *p = nv50->fragprog; struct nouveau_stateobj *so; - void *map; if (!p->translated) { nv50_program_validate(nv50, p); -- cgit v1.2.3 From bdd31c20abb27665ca701a5a46e29d4cfa71f679 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 11:41:47 +1000 Subject: nv50: SGE/SLT --- src/gallium/drivers/nv50/nv50_program.c | 82 ++++++++++++++++++++++++++++----- 1 file changed, 71 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 7295c2314cf..fa16c7522e1 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -19,8 +19,6 @@ * FRC * LIT * POW - * SGE - cvt - * SLT - cvt * SWZ * * MSB - Like MAD, but MUL+SUB @@ -201,15 +199,15 @@ emit(struct nv50_pc *pc, unsigned *inst) { struct nv50_program *p = pc->p; - if (inst[0] & 1) { - p->insns_nr += 2; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); - } else { - p->insns_nr += 1; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); - } + if (inst[0] & 1) { + p->insns_nr += 2; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); + } else { + p->insns_nr += 1; + p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); + memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); + } } static INLINE void set_long(struct nv50_pc *, unsigned *); @@ -582,6 +580,54 @@ emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } +/*XXX: inaccurate results.. why? */ +#define ALLOW_SET_SWAP 0 + +static void +emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, + struct nv50_reg *src0, struct nv50_reg *src1) +{ + unsigned inst[2] = { 0, 0 }; +#if ALLOW_SET_SWAP + unsigned inv_cop[8] = { 0, 6, 2, 4, 3, 5, 1, 7 }; +#endif + struct nv50_reg *rdst; + +#if ALLOW_SET_SWAP + assert(c_op <= 7); + if (check_swap_src_0_1(pc, &src0, &src1)) + c_op = inv_cop[c_op]; +#endif + + rdst = dst; + if (dst->type != P_TEMP) + dst = alloc_temp(pc, NULL); + + /* set.u32 */ + set_long(pc, inst); + inst[0] |= 0xb0000000; + inst[1] |= (3 << 29); + inst[1] |= (c_op << 14); + /*XXX: breaks things, .u32 by default? + * decuda will disasm as .u16 and use .lo/.hi regs, but this + * doesn't seem to match what the hw actually does. + inst[1] |= 0x04000000; << breaks things.. .u32 by default? + */ + set_dst(pc, dst, inst); + set_src_0(pc, src0, inst); + set_src_1(pc, src1, inst); + emit(pc, inst); + + /* cvt.f32.u32 */ + inst[0] = 0xa0000001; + inst[1] = 0x64014780; + set_dst(pc, rdst, inst); + set_src_0(pc, dst, inst); + emit(pc, inst); + + if (dst != rdst) + free_temp(pc, dst); +} static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) @@ -726,6 +772,13 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_flop(pc, 2, dst[c], src[0][c]); } break; + case TGSI_OPCODE_SGE: + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_set(pc, 6, dst[c], src[0][c], src[1][c]); + } + break; case TGSI_OPCODE_SIN: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) @@ -733,6 +786,13 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_flop(pc, 4, dst[c], src[0][c]); } break; + case TGSI_OPCODE_SLT: + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_set(pc, 1, dst[c], src[0][c], src[1][c]); + } + break; case TGSI_OPCODE_SUB: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) -- cgit v1.2.3 From b4f7463585071236d633e4c857dbbdf67b03dc94 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 12:55:06 +1000 Subject: nv50: FLR/FRC --- src/gallium/drivers/nv50/nv50_program.c | 37 +++++++++++++++++++++++++++++++-- 1 file changed, 35 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index fa16c7522e1..448062e767f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -15,8 +15,6 @@ /* ABS * ARL * DST - const(1.0) - * FLR - * FRC * LIT * POW * SWZ @@ -629,6 +627,24 @@ emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, free_temp(pc, dst); } +static void +emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + + set_long(pc, inst); + inst[0] = 0xa0000000; /* cvt */ + inst[1] |= (6 << 29); /* cvt */ + inst[1] |= 0x08000000; /* integer mode */ + inst[1] |= 0x04000000; /* 32 bit */ + inst[1] |= ((0x1 << 3)) << 14; /* .rn */ + inst[1] |= (1 << 14); /* src .f32 */ + set_dst(pc, dst, inst); + set_src_0(pc, src, inst); + + emit(pc, inst); +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { @@ -716,6 +732,23 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } free_temp(pc, temp); break; + case TGSI_OPCODE_FLR: + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_flr(pc, dst[c], src[0][c]); + } + break; + case TGSI_OPCODE_FRC: + temp = alloc_temp(pc, NULL); + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_flr(pc, temp, src[0][c]); + emit_sub(pc, dst[c], src[0][c], temp); + } + free_temp(pc, temp); + break; case TGSI_OPCODE_LG2: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) -- cgit v1.2.3 From 68091b0c89310c309b668c9d6d80640dc6040ab7 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 13:01:58 +1000 Subject: nv50: ABS --- src/gallium/drivers/nv50/nv50_program.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 448062e767f..7c2177d42db 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -12,8 +12,7 @@ #define NV50_SU_MAX_TEMP 64 -/* ABS - * ARL +/* ARL * DST - const(1.0) * LIT * POW @@ -670,6 +669,21 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } switch (inst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + for (c = 0; c < 4; c++) { + unsigned inst[2] = { 0, 0 }; + + set_long(pc, inst); + inst[0] = 0xa0000000; /* cvt */ + inst[1] |= (6 << 29); /* cvt */ + inst[1] |= 0x04000000; /* 32 bit */ + inst[1] |= (1 << 14); /* src .f32 */ + inst[1] |= ((1 << 6) << 14); /* .abs */ + set_dst(pc, dst[c], inst); + set_src_0(pc, src[0][c], inst); + emit(pc, inst); + } + break; case TGSI_OPCODE_ADD: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) -- cgit v1.2.3 From 33e4d30d50344be26398a51365bea1be37487403 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 14:10:48 +1000 Subject: nv50: DST --- src/gallium/drivers/nv50/nv50_program.c | 61 ++++++++++++++++++++++++--------- 1 file changed, 45 insertions(+), 16 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 7c2177d42db..49a731842f7 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -13,18 +13,18 @@ #define NV50_SU_MAX_TEMP 64 /* ARL - * DST - const(1.0) - * LIT + * LIT - other buggery * POW - * SWZ + * SWZ - negation ARGH + * SAT * * MSB - Like MAD, but MUL+SUB * - Fuck it off, introduce a way to negate args for ops that * support it. * - * Need ability to specifiy driver IMMD values, like nv40 constant() - * * Look into inlining IMMD for ops other than MOV + * - Maybe even relax restrictions a bit, can't do P_RESULT + P_IMMD, + * but can emit to P_TEMP first - then MOV later. NVIDIA does this */ struct nv50_reg { enum { @@ -145,6 +145,32 @@ kill_temp_temp(struct nv50_pc *pc) pc->temp_temp_nr = 0; } +static int +ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w) +{ + pc->immd_buf = realloc(pc->immd_buf, (pc->immd_nr + 1) * 4 * + sizeof(float)); + pc->immd_buf[(pc->immd_nr * 4) + 0] = x; + pc->immd_buf[(pc->immd_nr * 4) + 1] = x; + pc->immd_buf[(pc->immd_nr * 4) + 2] = x; + pc->immd_buf[(pc->immd_nr * 4) + 3] = x; + + return pc->immd_nr++; +} + +static struct nv50_reg * +alloc_immd(struct nv50_pc *pc, float f) +{ + struct nv50_reg *r = CALLOC_STRUCT(nv50_reg); + unsigned hw; + + hw = ctor_immd(pc, f, 0, 0, 0); + r->type = P_IMMD; + r->hw = hw; + r->index = -1; + return r; +} + static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -736,6 +762,16 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } free_temp(pc, temp); break; + case TGSI_OPCODE_DST: + { + struct nv50_reg *one = alloc_immd(pc, 1.0); + emit_mov(pc, dst[0], one); + emit_mul(pc, dst[1], src[0][1], src[1][1]); + emit_mov(pc, dst[2], src[0][2]); + emit_mov(pc, dst[3], src[1][3]); + FREE(one); + } + break; case TGSI_OPCODE_EX2: temp = alloc_temp(pc, NULL); for (c = 0; c < 4; c++) { @@ -886,17 +922,10 @@ nv50_program_tx_prep(struct nv50_pc *pc) const struct tgsi_full_immediate *imm = &p.FullToken.FullImmediate; - pc->immd_nr++; - pc->immd_buf = realloc(pc->immd_buf, 4 * pc->immd_nr * - sizeof(float)); - pc->immd_buf[4 * (pc->immd_nr - 1) + 0] = - imm->u.ImmediateFloat32[0].Float; - pc->immd_buf[4 * (pc->immd_nr - 1) + 1] = - imm->u.ImmediateFloat32[1].Float; - pc->immd_buf[4 * (pc->immd_nr - 1) + 2] = - imm->u.ImmediateFloat32[2].Float; - pc->immd_buf[4 * (pc->immd_nr - 1) + 3] = - imm->u.ImmediateFloat32[3].Float; + ctor_immd(pc, imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); } break; case TGSI_TOKEN_TYPE_DECLARATION: -- cgit v1.2.3 From 34abb858e2aaef2c1a066a7cdb3e0376d6c9f6bd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 15:21:05 +1000 Subject: nv50: handle 0/1 SWZ --- src/gallium/drivers/nv50/nv50_program.c | 56 +++++++++++++++++++++++---------- 1 file changed, 39 insertions(+), 17 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 49a731842f7..cad1d9d6795 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -15,7 +15,6 @@ /* ARL * LIT - other buggery * POW - * SWZ - negation ARGH * SAT * * MSB - Like MAD, but MUL+SUB @@ -189,32 +188,55 @@ tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) } static struct nv50_reg * -tgsi_src(struct nv50_pc *pc, int c, const struct tgsi_full_src_register *src) +tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src) { - /* Handle swizzling */ + struct nv50_reg *r = NULL; + unsigned c; + + c = tgsi_util_get_full_src_register_extswizzle(src, chan); switch (c) { - case 0: c = src->SrcRegister.SwizzleX; break; - case 1: c = src->SrcRegister.SwizzleY; break; - case 2: c = src->SrcRegister.SwizzleZ; break; - case 3: c = src->SrcRegister.SwizzleW; break; + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + switch (src->SrcRegister.File) { + case TGSI_FILE_INPUT: + r = &pc->attr[src->SrcRegister.Index * 4 + c]; + break; + case TGSI_FILE_TEMPORARY: + r = &pc->temp[src->SrcRegister.Index * 4 + c]; + break; + case TGSI_FILE_CONSTANT: + r = &pc->param[src->SrcRegister.Index * 4 + c]; + break; + case TGSI_FILE_IMMEDIATE: + r = &pc->immd[src->SrcRegister.Index * 4 + c]; + break; + default: + assert(0); + break; + } + break; + case TGSI_EXTSWIZZLE_ZERO: + r = alloc_immd(pc, 0.0); + break; + case TGSI_EXTSWIZZLE_ONE: + r = alloc_immd(pc, 1.0); + break; default: assert(0); + break; } - switch (src->SrcRegister.File) { - case TGSI_FILE_INPUT: - return &pc->attr[src->SrcRegister.Index * 4 + c]; - case TGSI_FILE_TEMPORARY: - return &pc->temp[src->SrcRegister.Index * 4 + c]; - case TGSI_FILE_CONSTANT: - return &pc->param[src->SrcRegister.Index * 4 + c]; - case TGSI_FILE_IMMEDIATE: - return &pc->immd[src->SrcRegister.Index * 4 + c]; + switch (tgsi_util_get_full_src_register_sign_mode(src, chan)) { + case TGSI_UTIL_SIGN_KEEP: + break; default: + assert(0); break; } - return NULL; + return r; } static void -- cgit v1.2.3 From 688064236ba8b5997014493eb6c6e3fe0739813e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 16:01:06 +1000 Subject: nv50: fixes + untested _SAT modifier --- src/gallium/drivers/nv50/nv50_program.c | 36 ++++++++++++++++++++++++++++----- 1 file changed, 31 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index cad1d9d6795..9fcd7c36c61 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -15,7 +15,6 @@ /* ARL * LIT - other buggery * POW - * SAT * * MSB - Like MAD, but MUL+SUB * - Fuck it off, introduce a way to negate args for ops that @@ -679,8 +678,8 @@ emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { unsigned inst[2] = { 0, 0 }; - set_long(pc, inst); inst[0] = 0xa0000000; /* cvt */ + set_long(pc, inst); inst[1] |= (6 << 29); /* cvt */ inst[1] |= 0x08000000; /* integer mode */ inst[1] |= 0x04000000; /* 32 bit */ @@ -696,13 +695,14 @@ static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { const struct tgsi_full_instruction *inst = &tok->FullInstruction; - struct nv50_reg *dst[4], *src[3][4], *temp; - unsigned mask; + struct nv50_reg *rdst[4], *dst[4], *src[3][4], *temp; + unsigned mask, sat; int i, c; NOUVEAU_ERR("insn %p\n", tok); mask = inst->FullDstRegisters[0].DstRegister.WriteMask; + sat = inst->Instruction.Saturate == TGSI_SAT_ZERO_ONE; for (c = 0; c < 4; c++) { if (mask & (1 << c)) @@ -716,13 +716,20 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) src[i][c] = tgsi_src(pc, c, &inst->FullSrcRegisters[i]); } + if (sat) { + for (c = 0; c < 4; c++) { + rdst[c] = dst[c]; + dst[c] = temp_temp(pc); + } + } + switch (inst->Instruction.Opcode) { case TGSI_OPCODE_ABS: for (c = 0; c < 4; c++) { unsigned inst[2] = { 0, 0 }; - set_long(pc, inst); inst[0] = 0xa0000000; /* cvt */ + set_long(pc, inst); inst[1] |= (6 << 29); /* cvt */ inst[1] |= 0x04000000; /* 32 bit */ inst[1] |= (1 << 14); /* src .f32 */ @@ -922,6 +929,25 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) return FALSE; } + if (sat) { + for (c = 0; c < 4; c++) { + unsigned inst[2] = { 0, 0 }; + + if (!(mask & (1 << c))) + continue; + + inst[0] = 0xa0000000; /* cvt */ + set_long(pc, inst); + inst[1] |= (6 << 29); /* cvt */ + inst[1] |= 0x04000000; /* 32 bit */ + inst[1] |= (1 << 14); /* src .f32 */ + inst[1] |= ((1 << 5) << 14); /* .sat */ + set_dst(pc, rdst[c], inst); + set_src_0(pc, dst[c], inst); + emit(pc, inst); + } + } + kill_temp_temp(pc); return TRUE; } -- cgit v1.2.3 From ea4b09cbcbd9db82648ab30f18c0f46a66ab9f69 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 16:08:36 +1000 Subject: nv50: POW! --- src/gallium/drivers/nv50/nv50_program.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 9fcd7c36c61..79893a9ad4f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -14,15 +14,17 @@ /* ARL * LIT - other buggery - * POW * * MSB - Like MAD, but MUL+SUB * - Fuck it off, introduce a way to negate args for ops that * support it. * - * Look into inlining IMMD for ops other than MOV + * Look into inlining IMMD for ops other than MOV (make it general?) * - Maybe even relax restrictions a bit, can't do P_RESULT + P_IMMD, * but can emit to P_TEMP first - then MOV later. NVIDIA does this + * + * Verify half-insns work where expected - and force disable them where they + * don't work - MUL has it forcibly disabled atm as it fixes POW.. */ struct nv50_reg { enum { @@ -489,6 +491,7 @@ emit_mul(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, unsigned inst[2] = { 0, 0 }; inst[0] |= 0xc0000000; + set_long(pc, inst); check_swap_src_0_1(pc, &src0, &src1); set_dst(pc, dst, inst); @@ -870,6 +873,19 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_mul(pc, dst[c], src[0][c], src[1][c]); } break; + case TGSI_OPCODE_POW: + temp = alloc_temp(pc, NULL); + emit_flop(pc, 3, temp, src[0][0]); + emit_mul(pc, temp, temp, src[1][0]); + emit_preex2(pc, temp, temp); + emit_flop(pc, 6, temp, temp); + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + emit_mov(pc, dst[c], temp); + } + free_temp(pc, temp); + break; case TGSI_OPCODE_RCP: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) -- cgit v1.2.3 From 686bc00c05094e8678747c111a6a70ad4b7063e3 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 16:51:07 +1000 Subject: nv50: oops, copy+pasto --- src/gallium/drivers/nv50/nv50_program.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 79893a9ad4f..b013435f990 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -151,9 +151,9 @@ ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w) pc->immd_buf = realloc(pc->immd_buf, (pc->immd_nr + 1) * 4 * sizeof(float)); pc->immd_buf[(pc->immd_nr * 4) + 0] = x; - pc->immd_buf[(pc->immd_nr * 4) + 1] = x; - pc->immd_buf[(pc->immd_nr * 4) + 2] = x; - pc->immd_buf[(pc->immd_nr * 4) + 3] = x; + pc->immd_buf[(pc->immd_nr * 4) + 1] = y; + pc->immd_buf[(pc->immd_nr * 4) + 2] = z; + pc->immd_buf[(pc->immd_nr * 4) + 3] = w; return pc->immd_nr++; } -- cgit v1.2.3 From faa1c02546db00f69c66db18076b5b0ac86d7138 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 17:36:10 +1000 Subject: nv50: create emit_pow() - emit_lit() will need to use it --- src/gallium/drivers/nv50/nv50_program.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index b013435f990..bc0a834aee6 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -627,6 +627,7 @@ emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } + /*XXX: inaccurate results.. why? */ #define ALLOW_SET_SWAP 0 @@ -694,6 +695,20 @@ emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } +static void +emit_pow(struct nv50_pc *pc, struct nv50_reg *dst, + struct nv50_reg *v, struct nv50_reg *e) +{ + struct nv50_reg *temp = alloc_temp(pc, NULL); + + emit_flop(pc, 3, temp, v); + emit_mul(pc, temp, temp, e); + emit_preex2(pc, temp, temp); + emit_flop(pc, 6, dst, temp); + + free_temp(pc, temp); +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { @@ -875,10 +890,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) break; case TGSI_OPCODE_POW: temp = alloc_temp(pc, NULL); - emit_flop(pc, 3, temp, src[0][0]); - emit_mul(pc, temp, temp, src[1][0]); - emit_preex2(pc, temp, temp); - emit_flop(pc, 6, temp, temp); + emit_pow(pc, temp, src[0][0], src[1][0]); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; -- cgit v1.2.3 From fe90cc509f75772ce202930c934bade1d4b116c8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 17:42:00 +1000 Subject: nv50: obey per-source abs (TGSI_UTIL_SIGN_CLEAR) --- src/gallium/drivers/nv50/nv50_program.c | 175 +++++++++++++++++--------------- 1 file changed, 95 insertions(+), 80 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index bc0a834aee6..fabd0164914 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -171,75 +171,6 @@ alloc_immd(struct nv50_pc *pc, float f) return r; } -static struct nv50_reg * -tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) -{ - switch (dst->DstRegister.File) { - case TGSI_FILE_TEMPORARY: - return &pc->temp[dst->DstRegister.Index * 4 + c]; - case TGSI_FILE_OUTPUT: - return &pc->result[dst->DstRegister.Index * 4 + c]; - case TGSI_FILE_NULL: - return NULL; - default: - break; - } - - return NULL; -} - -static struct nv50_reg * -tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src) -{ - struct nv50_reg *r = NULL; - unsigned c; - - c = tgsi_util_get_full_src_register_extswizzle(src, chan); - switch (c) { - case TGSI_EXTSWIZZLE_X: - case TGSI_EXTSWIZZLE_Y: - case TGSI_EXTSWIZZLE_Z: - case TGSI_EXTSWIZZLE_W: - switch (src->SrcRegister.File) { - case TGSI_FILE_INPUT: - r = &pc->attr[src->SrcRegister.Index * 4 + c]; - break; - case TGSI_FILE_TEMPORARY: - r = &pc->temp[src->SrcRegister.Index * 4 + c]; - break; - case TGSI_FILE_CONSTANT: - r = &pc->param[src->SrcRegister.Index * 4 + c]; - break; - case TGSI_FILE_IMMEDIATE: - r = &pc->immd[src->SrcRegister.Index * 4 + c]; - break; - default: - assert(0); - break; - } - break; - case TGSI_EXTSWIZZLE_ZERO: - r = alloc_immd(pc, 0.0); - break; - case TGSI_EXTSWIZZLE_ONE: - r = alloc_immd(pc, 1.0); - break; - default: - assert(0); - break; - } - - switch (tgsi_util_get_full_src_register_sign_mode(src, chan)) { - case TGSI_UTIL_SIGN_KEEP: - break; - default: - assert(0); - break; - } - - return r; -} - static void emit(struct nv50_pc *pc, unsigned *inst) { @@ -709,6 +640,98 @@ emit_pow(struct nv50_pc *pc, struct nv50_reg *dst, free_temp(pc, temp); } +static void +emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] = 0xa0000000; /* cvt */ + set_long(pc, inst); + inst[1] |= (6 << 29); /* cvt */ + inst[1] |= 0x04000000; /* 32 bit */ + inst[1] |= (1 << 14); /* src .f32 */ + inst[1] |= ((1 << 6) << 14); /* .abs */ + set_dst(pc, dst, inst); + set_src_0(pc, src, inst); + + emit(pc, inst); +} + +static struct nv50_reg * +tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) +{ + switch (dst->DstRegister.File) { + case TGSI_FILE_TEMPORARY: + return &pc->temp[dst->DstRegister.Index * 4 + c]; + case TGSI_FILE_OUTPUT: + return &pc->result[dst->DstRegister.Index * 4 + c]; + case TGSI_FILE_NULL: + return NULL; + default: + break; + } + + return NULL; +} + +static struct nv50_reg * +tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src) +{ + struct nv50_reg *r = NULL; + struct nv50_reg *temp; + unsigned c; + + c = tgsi_util_get_full_src_register_extswizzle(src, chan); + switch (c) { + case TGSI_EXTSWIZZLE_X: + case TGSI_EXTSWIZZLE_Y: + case TGSI_EXTSWIZZLE_Z: + case TGSI_EXTSWIZZLE_W: + switch (src->SrcRegister.File) { + case TGSI_FILE_INPUT: + r = &pc->attr[src->SrcRegister.Index * 4 + c]; + break; + case TGSI_FILE_TEMPORARY: + r = &pc->temp[src->SrcRegister.Index * 4 + c]; + break; + case TGSI_FILE_CONSTANT: + r = &pc->param[src->SrcRegister.Index * 4 + c]; + break; + case TGSI_FILE_IMMEDIATE: + r = &pc->immd[src->SrcRegister.Index * 4 + c]; + break; + default: + assert(0); + break; + } + break; + case TGSI_EXTSWIZZLE_ZERO: + r = alloc_immd(pc, 0.0); + break; + case TGSI_EXTSWIZZLE_ONE: + r = alloc_immd(pc, 1.0); + break; + default: + assert(0); + break; + } + + switch (tgsi_util_get_full_src_register_sign_mode(src, chan)) { + case TGSI_UTIL_SIGN_KEEP: + break; + case TGSI_UTIL_SIGN_CLEAR: + temp = temp_temp(pc); + emit_abs(pc, temp, r); + r = temp; + break; + default: + assert(0); + break; + } + + return r; +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) { @@ -744,17 +767,9 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) switch (inst->Instruction.Opcode) { case TGSI_OPCODE_ABS: for (c = 0; c < 4; c++) { - unsigned inst[2] = { 0, 0 }; - - inst[0] = 0xa0000000; /* cvt */ - set_long(pc, inst); - inst[1] |= (6 << 29); /* cvt */ - inst[1] |= 0x04000000; /* 32 bit */ - inst[1] |= (1 << 14); /* src .f32 */ - inst[1] |= ((1 << 6) << 14); /* .abs */ - set_dst(pc, dst[c], inst); - set_src_0(pc, src[0][c], inst); - emit(pc, inst); + if (!(mask & (1 << c))) + continue; + emit_abs(pc, dst[c], src[0][c]); } break; case TGSI_OPCODE_ADD: -- cgit v1.2.3 From 01e36eb531dfb4b1b3fd38d3fc00c6770833b5ea Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sat, 7 Jun 2008 19:54:04 +1000 Subject: nv50: LIT - sort-of *somehow* we have the exact same bug here as on nv4x, the difference being on nv4x the hw actually has a LIT opcode.. NVIDIA doesn't have the bug on either arch FWIW. --- src/gallium/drivers/nv50/nv50_program.c | 36 +++++++++++++++++++++++++++++++-- src/gallium/drivers/nv50/nv50_vbo.c | 1 - 2 files changed, 34 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index fabd0164914..186eea5e13b 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -13,7 +13,6 @@ #define NV50_SU_MAX_TEMP 64 /* ARL - * LIT - other buggery * * MSB - Like MAD, but MUL+SUB * - Fuck it off, introduce a way to negate args for ops that @@ -23,6 +22,8 @@ * - Maybe even relax restrictions a bit, can't do P_RESULT + P_IMMD, * but can emit to P_TEMP first - then MOV later. NVIDIA does this * + * Hmmm.. what happens if we have src1+src2 both consts.. ouch ! + * * Verify half-insns work where expected - and force disable them where they * don't work - MUL has it forcibly disabled atm as it fixes POW.. */ @@ -164,7 +165,7 @@ alloc_immd(struct nv50_pc *pc, float f) struct nv50_reg *r = CALLOC_STRUCT(nv50_reg); unsigned hw; - hw = ctor_immd(pc, f, 0, 0, 0); + hw = ctor_immd(pc, f, 0, 0, 0) * 4; r->type = P_IMMD; r->hw = hw; r->index = -1; @@ -657,6 +658,34 @@ emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } +static void +emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, struct nv50_reg **src) +{ + struct nv50_reg *one = alloc_immd(pc, 1.0); + struct nv50_reg *zero = alloc_immd(pc, 0.0); + struct nv50_reg *neg128 = alloc_immd(pc, -127.999999); + struct nv50_reg *pos128 = alloc_immd(pc, 127.999999); + struct nv50_reg *tmp[4]; + + emit_mov(pc, dst[0], one); + emit_mov(pc, dst[3], one); + + tmp[0] = temp_temp(pc); + emit_minmax(pc, 4, dst[1], src[0], zero); + set_pred_wr(pc, 1, 0, &pc->p->insns[pc->p->insns_nr - 2]); + + tmp[1] = temp_temp(pc); + emit_minmax(pc, 4, tmp[1], src[1], zero); + + tmp[3] = temp_temp(pc); + emit_minmax(pc, 4, tmp[3], src[3], neg128); + emit_minmax(pc, 5, tmp[3], tmp[3], pos128); + + emit_pow(pc, dst[2], tmp[1], tmp[3]); + emit_mov(pc, dst[2], zero); + set_pred(pc, 3, 0, &pc->p->insns[pc->p->insns_nr - 2]); +} + static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -861,6 +890,9 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } free_temp(pc, temp); break; + case TGSI_OPCODE_LIT: + emit_lit(pc, &dst[0], &src[0][0]); + break; case TGSI_OPCODE_LG2: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index f086c762582..badbef53df5 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -35,7 +35,6 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, struct nv50_context *nv50 = nv50_context(pipe); nv50_state_validate(nv50); - NOUVEAU_ERR("unimplemented\n"); BEGIN_RING(tesla, 0x142c, 1); OUT_RING (0); -- cgit v1.2.3 From 9a37a56c8ab8c64bdadb1e1e807f885d6a5e3121 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 8 Jun 2008 11:23:06 +1000 Subject: nv50: obey writemask in a couple of places --- src/gallium/drivers/nv50/nv50_program.c | 33 ++++++++++++++++++++++----------- 1 file changed, 22 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 186eea5e13b..7206e5d280a 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -12,7 +12,7 @@ #define NV50_SU_MAX_TEMP 64 -/* ARL +/* ARL - gallium craps itself on progs/vp/arl.txt * * MSB - Like MAD, but MUL+SUB * - Fuck it off, introduce a way to negate args for ops that @@ -856,10 +856,14 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) case TGSI_OPCODE_DST: { struct nv50_reg *one = alloc_immd(pc, 1.0); - emit_mov(pc, dst[0], one); - emit_mul(pc, dst[1], src[0][1], src[1][1]); - emit_mov(pc, dst[2], src[0][2]); - emit_mov(pc, dst[3], src[1][3]); + if (mask & (1 << 0)) + emit_mov(pc, dst[0], one); + if (mask & (1 << 1)) + emit_mul(pc, dst[1], src[0][1], src[1][1]); + if (mask & (1 << 2)) + emit_mov(pc, dst[2], src[0][2]); + if (mask & (1 << 3)) + emit_mov(pc, dst[3], src[1][3]); FREE(one); } break; @@ -891,6 +895,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) free_temp(pc, temp); break; case TGSI_OPCODE_LIT: + /*XXX: writemask */ emit_lit(pc, &dst[0], &src[0][0]); break; case TGSI_OPCODE_LG2: @@ -989,12 +994,18 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) break; case TGSI_OPCODE_XPD: temp = alloc_temp(pc, NULL); - emit_mul(pc, temp, src[0][2], src[1][1]); - emit_msb(pc, dst[0], src[0][1], src[1][2], temp); - emit_mul(pc, temp, src[0][0], src[1][2]); - emit_msb(pc, dst[1], src[0][2], src[1][0], temp); - emit_mul(pc, temp, src[0][1], src[1][0]); - emit_msb(pc, dst[2], src[0][0], src[1][1], temp); + if (mask & (1 << 0)) { + emit_mul(pc, temp, src[0][2], src[1][1]); + emit_msb(pc, dst[0], src[0][1], src[1][2], temp); + } + if (mask & (1 << 1)) { + emit_mul(pc, temp, src[0][0], src[1][2]); + emit_msb(pc, dst[1], src[0][2], src[1][0], temp); + } + if (mask & (1 << 2)) { + emit_mul(pc, temp, src[0][1], src[1][0]); + emit_msb(pc, dst[2], src[0][0], src[1][1], temp); + } free_temp(pc, temp); break; case TGSI_OPCODE_END: -- cgit v1.2.3 From 6ca31aa55eada38ad8b8a249b9e79a83b9784c04 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 8 Jun 2008 12:27:36 +1000 Subject: nv50: hehe, damage from tex-surfaces.. surface_fill() reenabled now :) --- src/gallium/drivers/nv50/nv50_miptree.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 6c838998fd0..be85c3fd5ce 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -21,7 +21,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { struct pipe_winsys *ws = pscreen->winsys; struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree); - + NOUVEAU_ERR("unimplemented\n"); mt->base = *pt; @@ -29,7 +29,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) mt->base.screen = pscreen; mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - 512*32*4); + pt->width[0] * pt->cpp * pt->height[0]); if (!mt->buffer) { FREE(mt); return NULL; -- cgit v1.2.3 From afcaeaa0e4dc3ced40621c76304a2c0c5a3ab403 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 8 Jun 2008 14:12:01 +1000 Subject: nv50: note a critical bug --- src/gallium/drivers/nv50/nv50_program.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 7206e5d280a..b151b540bf4 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -26,6 +26,9 @@ * * Verify half-insns work where expected - and force disable them where they * don't work - MUL has it forcibly disabled atm as it fixes POW.. + * + * FUCK! watch dst==src vectors, can overwrite components that are needed. + * ie. SUB R0, R0.yzxw, R0 */ struct nv50_reg { enum { -- cgit v1.2.3 From 34a039ae7b158cacb5b20d91067e9d6458d30a56 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 8 Jun 2008 15:51:54 +1000 Subject: nv50: fix src1 & src2 == const --- src/gallium/drivers/nv50/nv50_program.c | 26 ++++++++++++++++++++------ 1 file changed, 20 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index b151b540bf4..936a8ef9b35 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -22,8 +22,6 @@ * - Maybe even relax restrictions a bit, can't do P_RESULT + P_IMMD, * but can emit to P_TEMP first - then MOV later. NVIDIA does this * - * Hmmm.. what happens if we have src1+src2 both consts.. ouch ! - * * Verify half-insns work where expected - and force disable them where they * don't work - MUL has it forcibly disabled atm as it fixes POW.. * @@ -391,8 +389,16 @@ set_src_1(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) src = temp; } else if (src->type == P_CONST || src->type == P_IMMD) { - set_cseg(pc, src, inst); - inst[0] |= 0x00800000; + assert(!(inst[0] & 0x00800000)); + if (inst[0] & 0x01000000) { + struct nv50_reg *temp = temp_temp(pc); + + emit_mov(pc, temp, src); + src = temp; + } else { + set_cseg(pc, src, inst); + inst[0] |= 0x00800000; + } } alloc_reg(pc, src); @@ -411,8 +417,16 @@ set_src_2(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) src = temp; } else if (src->type == P_CONST || src->type == P_IMMD) { - set_cseg(pc, src, inst); - inst[0] |= 0x01000000; + assert(!(inst[0] & 0x01000000)); + if (inst[0] & 0x00800000) { + struct nv50_reg *temp = temp_temp(pc); + + emit_mov(pc, temp, src); + src = temp; + } else { + set_cseg(pc, src, inst); + inst[0] |= 0x01000000; + } } alloc_reg(pc, src); -- cgit v1.2.3 From 776e9581d16fc0fd28058fbcd879756fd5d40b96 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 13:07:38 +1000 Subject: nv50: delayed write of fragprog result regs until end of program --- src/gallium/drivers/nv50/nv50_program.c | 18 +++++++++++++++--- 1 file changed, 15 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 936a8ef9b35..1733cf97b76 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -27,6 +27,8 @@ * * FUCK! watch dst==src vectors, can overwrite components that are needed. * ie. SUB R0, R0.yzxw, R0 + * + * NV50_PROG* -> PIPE_SHADER* */ struct nv50_reg { enum { @@ -1198,11 +1200,13 @@ nv50_program_tx_prep(struct nv50_pc *pc) for (i = 0; i < pc->result_nr; i++) { for (c = 0; c < 4; c++) { - if (pc->p->type == NV50_PROG_FRAGMENT) + if (pc->p->type == NV50_PROG_FRAGMENT) { pc->result[i*4+c].type = P_TEMP; - else + pc->result[i*4+c].hw = -1; + } else { pc->result[i*4+c].type = P_RESULT; - pc->result[i*4+c].hw = rid++; + pc->result[i*4+c].hw = rid++; + } pc->result[i*4+c].index = i; } } @@ -1281,6 +1285,14 @@ nv50_program_tx(struct nv50_program *p) } } + if (p->type == NV50_PROG_FRAGMENT) { + struct nv50_reg out; + + out.type = P_TEMP; + for (out.hw = 0; out.hw < pc->result_nr * 4; out.hw++) + emit_mov(pc, &out, &pc->result[out.hw]); + } + p->immd_nr = pc->immd_nr * 4; p->immd = pc->immd_buf; -- cgit v1.2.3 From b5bbf09c42a9d563984fad875ced5c4814033a3d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 13:09:55 +1000 Subject: nv50: remove NV50_PROG_{VERTEX,FRAGMENT} --- src/gallium/drivers/nv50/nv50_program.c | 19 +++++++++++-------- src/gallium/drivers/nv50/nv50_state.c | 4 ++-- src/gallium/drivers/nv50/nv50_state.h | 5 +---- 3 files changed, 14 insertions(+), 14 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 1733cf97b76..6b8ba1bdfe5 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -28,7 +28,10 @@ * FUCK! watch dst==src vectors, can overwrite components that are needed. * ie. SUB R0, R0.yzxw, R0 * - * NV50_PROG* -> PIPE_SHADER* + * Things to check with renouveau: + * SGE/SLT with needed src0/1 swap + * FP attr/result assignment - how? + * FP/VP constbuf usage */ struct nv50_reg { enum { @@ -289,7 +292,7 @@ set_cseg(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) if (src->type == P_IMMD) { inst[1] |= (NV50_CB_PMISC << 22); } else { - if (pc->p->type == NV50_PROG_VERTEX) + if (pc->p->type == PIPE_SHADER_VERTEX) inst[1] |= (NV50_CB_PVP << 22); else inst[1] |= (NV50_CB_PFP << 22); @@ -1144,7 +1147,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (!pc->attr) goto out_err; - if (pc->p->type == NV50_PROG_FRAGMENT) { + if (pc->p->type == PIPE_SHADER_FRAGMENT) { iv = alloc_temp(pc, NULL); aid++; } @@ -1153,7 +1156,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) struct nv50_reg *a = &pc->attr[i*4]; for (c = 0; c < 4; c++) { - if (pc->p->type == NV50_PROG_FRAGMENT) { + if (pc->p->type == PIPE_SHADER_FRAGMENT) { struct nv50_reg *at = alloc_temp(pc, NULL); pc->attr[i*4+c].type = at->type; @@ -1168,7 +1171,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) } } - if (pc->p->type != NV50_PROG_FRAGMENT) + if (pc->p->type != PIPE_SHADER_FRAGMENT) continue; emit_interp(pc, iv, iv, iv, FALSE); @@ -1200,7 +1203,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) for (i = 0; i < pc->result_nr; i++) { for (c = 0; c < 4; c++) { - if (pc->p->type == NV50_PROG_FRAGMENT) { + if (pc->p->type == PIPE_SHADER_FRAGMENT) { pc->result[i*4+c].type = P_TEMP; pc->result[i*4+c].hw = -1; } else { @@ -1285,7 +1288,7 @@ nv50_program_tx(struct nv50_program *p) } } - if (p->type == NV50_PROG_FRAGMENT) { + if (p->type == PIPE_SHADER_FRAGMENT) { struct nv50_reg out; out.type = P_TEMP; @@ -1314,7 +1317,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) * NOT immd - otherwise it's fucked fucked fucked */ p->insns[p->insns_nr - 1] |= 0x00000001; - if (p->type == NV50_PROG_VERTEX) { + if (p->type == PIPE_SHADER_VERTEX) { for (i = 0; i < p->insns_nr; i++) NOUVEAU_ERR("VP0x%08x\n", p->insns[i]); } else { diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 48d09c296bb..774117326ad 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -331,7 +331,7 @@ nv50_vp_state_create(struct pipe_context *pipe, struct nv50_program *p = CALLOC_STRUCT(nv50_program); p->pipe = *cso; - p->type = NV50_PROG_VERTEX; + p->type = PIPE_SHADER_VERTEX; tgsi_scan_shader(p->pipe.tokens, &p->info); return (void *)p; } @@ -361,7 +361,7 @@ nv50_fp_state_create(struct pipe_context *pipe, struct nv50_program *p = CALLOC_STRUCT(nv50_program); p->pipe = *cso; - p->type = NV50_PROG_FRAGMENT; + p->type = PIPE_SHADER_FRAGMENT; tgsi_scan_shader(p->pipe.tokens, &p->info); return (void *)p; } diff --git a/src/gallium/drivers/nv50/nv50_state.h b/src/gallium/drivers/nv50/nv50_state.h index dd352b65856..d568bf610d5 100644 --- a/src/gallium/drivers/nv50/nv50_state.h +++ b/src/gallium/drivers/nv50/nv50_state.h @@ -15,10 +15,7 @@ struct nv50_program { struct tgsi_shader_info info; boolean translated; - enum { - NV50_PROG_VERTEX, - NV50_PROG_FRAGMENT - } type; + unsigned type; unsigned *insns; unsigned insns_nr; -- cgit v1.2.3 From 713ef6ccd2590bd866598bb6d4f646e9ec29ba78 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 13:31:42 +1000 Subject: nv50: use emit_flop() instead of building RCP manually on interp --- src/gallium/drivers/nv50/nv50_program.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 6b8ba1bdfe5..592435585d9 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1176,12 +1176,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) emit_interp(pc, iv, iv, iv, FALSE); tmp = alloc_temp(pc, NULL); - { - unsigned inst[2] = { 0, 0 }; - inst[0] = 0x90000000; - inst[0] |= (tmp->hw << 2); - emit(pc, inst); - } + emit_flop(pc, 0, tmp, iv); emit_interp(pc, &a[0], &a[0], tmp, TRUE); emit_interp(pc, &a[1], &a[1], tmp, TRUE); emit_interp(pc, &a[2], &a[2], tmp, TRUE); -- cgit v1.2.3 From 6d0f7ea95475009ee17862786469f7b9a34a797f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 16:26:51 +1000 Subject: nv50: note some things discovered during renouveau session --- src/gallium/drivers/nv50/nv50_program.c | 23 +++++++++++++++++++++-- 1 file changed, 21 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 592435585d9..0274545131e 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -28,10 +28,29 @@ * FUCK! watch dst==src vectors, can overwrite components that are needed. * ie. SUB R0, R0.yzxw, R0 * + * MOV dst, -src + * "delta" tmp, -src (0xa0000204,0xe4004780 - delta r0, -r0) + * mov dst, tmp + * * Things to check with renouveau: - * SGE/SLT with needed src0/1 swap * FP attr/result assignment - how? - * FP/VP constbuf usage + * attrib + * - 0x16bc maps vp output onto fp hpos + * - 0x16c0 maps vp output onto fp col0 + * result + * - colr always 0-3 + * - depr always 4 + * 0x16bc->0x16e8 --> some binding between vp/fp regs + * 0x16b8 --> VP output count + * + * 0x1298 --> "MOV rcol.x, fcol.y" "MOV depr, fcol.y" = 0x00000005 + * "MOV rcol.x, fcol.y" = 0x00000004 + * 0x19a8 --> as above but 0x00000100 and 0x00000000 + * - 0x00100000 used when KIL used + * 0x196c --> as above but 0x00000011 and 0x00000000 + * + * 0x1988 --> 0xXXNNNNNN + * - XX == FP high something */ struct nv50_reg { enum { -- cgit v1.2.3 From 31f6a24b59b0ac18e04336d2e3cbaa643358c88a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 16:35:07 +1000 Subject: nv50: support the other TGSI_UTIL_SIGN modes --- src/gallium/drivers/nv50/nv50_program.c | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 0274545131e..44914f41b66 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -727,6 +727,22 @@ emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, struct nv50_reg **src) set_pred(pc, 3, 0, &pc->p->insns[pc->p->insns_nr - 2]); } +static void +emit_neg(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + + set_long(pc, inst); + inst[0] |= 0xa0000000; /* delta */ + inst[1] |= (7 << 29); /* delta */ + inst[1] |= 0x04000000; /* negate arg0? probably not */ + inst[1] |= (1 << 14); /* src .f32 */ + set_dst(pc, dst, inst); + set_src_0(pc, src, inst); + + emit(pc, inst); +} + static struct nv50_reg * tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst) { @@ -794,6 +810,17 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src) emit_abs(pc, temp, r); r = temp; break; + case TGSI_UTIL_SIGN_TOGGLE: + temp = temp_temp(pc); + emit_neg(pc, temp, r); + r = temp; + break; + case TGSI_UTIL_SIGN_SET: + temp = temp_temp(pc); + emit_abs(pc, temp, r); + emit_neg(pc, temp, r); + r = temp; + break; default: assert(0); break; -- cgit v1.2.3 From 51ea3aae03154046316b814053f7493bdb10c853 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 16:41:08 +1000 Subject: nv50: fix SGE/SLT when sources need swapping --- src/gallium/drivers/nv50/nv50_program.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 44914f41b66..21341619de4 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -601,24 +601,17 @@ emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } -/*XXX: inaccurate results.. why? */ -#define ALLOW_SET_SWAP 0 - static void emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { unsigned inst[2] = { 0, 0 }; -#if ALLOW_SET_SWAP - unsigned inv_cop[8] = { 0, 6, 2, 4, 3, 5, 1, 7 }; -#endif + unsigned inv_cop[8] = { 0, 4, 2, 6, 1, 5, 3, 7 }; struct nv50_reg *rdst; -#if ALLOW_SET_SWAP assert(c_op <= 7); if (check_swap_src_0_1(pc, &src0, &src1)) c_op = inv_cop[c_op]; -#endif rdst = dst; if (dst->type != P_TEMP) -- cgit v1.2.3 From 454394e749feca5ac00e7a270e6ca5529581d228 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 17:36:22 +1000 Subject: nv50: quick hack so progs/fp/fp-tri works for the moment --- src/gallium/drivers/nv50/nv50_context.h | 4 ++-- src/gallium/drivers/nv50/nv50_program.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index c4a8a4c064e..69c9dba675b 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -24,8 +24,8 @@ fprintf(stderr, "nouveau: "fmt, ##args); /* Constant buffer assignment */ -#define NV50_CB_PMISC 0 -#define NV50_CB_PVP 1 +#define NV50_CB_PMISC 1 +#define NV50_CB_PVP 0 #define NV50_CB_PFP 2 #define NV50_CB_PGP 3 #define NV50_CB_TIC 4 diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 21341619de4..843e0362514 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1369,7 +1369,7 @@ nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) for (i = 0; i < p->immd_nr; i++) { BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING ((NV50_CB_PMISC << 16) | (i << 8)); + OUT_RING ((NV50_CB_PMISC << 0) | (i << 8)); OUT_RING (fui(p->immd[i])); } } -- cgit v1.2.3 From 21e688e0a3faeef18b07c4d860bd71cc6e3ddf4a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 18:01:03 +1000 Subject: nv50: LRP --- src/gallium/drivers/nv50/nv50_program.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 843e0362514..33822a1a788 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -85,7 +85,7 @@ struct nv50_pc { float *immd_buf; int immd_nr; - struct nv50_reg *temp_temp[8]; + struct nv50_reg *temp_temp[16]; unsigned temp_temp_nr; }; @@ -154,7 +154,7 @@ free_temp(struct nv50_pc *pc, struct nv50_reg *r) static struct nv50_reg * temp_temp(struct nv50_pc *pc) { - if (pc->temp_temp_nr >= 8) + if (pc->temp_temp_nr >= 16) assert(0); pc->temp_temp[pc->temp_temp_nr] = alloc_temp(pc, NULL); @@ -966,6 +966,18 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_flop(pc, 3, dst[c], src[0][c]); } break; + case TGSI_OPCODE_LRP: + for (c = 0; c < 4; c++) { + if (!(mask & (1 << c))) + continue; + /*XXX: we can do better than this */ + temp = alloc_temp(pc, NULL); + emit_neg(pc, temp, src[0][c]); + emit_mad(pc, temp, temp, src[2][c], src[2][c]); + emit_mad(pc, dst[c], src[0][c], src[1][c], temp); + free_temp(pc, temp); + } + break; case TGSI_OPCODE_MAD: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) -- cgit v1.2.3 From 7b7df34781844c39998d60bbb60880d960da3fb1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 9 Jun 2008 19:50:17 +1000 Subject: nv50: various fixes + SCS --- src/gallium/drivers/nv50/nv50_program.c | 89 ++++++++++++++++++++++++--------- 1 file changed, 66 insertions(+), 23 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 33822a1a788..d69eb4b86be 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -601,6 +601,21 @@ emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) emit(pc, inst); } +static void +emit_precossin(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) +{ + unsigned inst[2] = { 0, 0 }; + + inst[0] |= 0xb0000000; + + set_dst(pc, dst, inst); + set_src_0(pc, src, inst); + set_long(pc, inst); + inst[1] |= (6 << 29); + + emit(pc, inst); +} + static void emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) @@ -693,7 +708,8 @@ emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) } static void -emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, struct nv50_reg **src) +emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, + struct nv50_reg **src) { struct nv50_reg *one = alloc_immd(pc, 1.0); struct nv50_reg *zero = alloc_immd(pc, 0.0); @@ -701,23 +717,34 @@ emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, struct nv50_reg **src) struct nv50_reg *pos128 = alloc_immd(pc, 127.999999); struct nv50_reg *tmp[4]; - emit_mov(pc, dst[0], one); - emit_mov(pc, dst[3], one); + if (mask & (1 << 0)) + emit_mov(pc, dst[0], one); + + if (mask & (1 << 3)) + emit_mov(pc, dst[3], one); + + if (mask & (3 << 1)) { + if (mask & (1 << 1)) + tmp[0] = dst[1]; + else + tmp[0] = temp_temp(pc); + emit_minmax(pc, 4, tmp[0], src[0], zero); + } - tmp[0] = temp_temp(pc); - emit_minmax(pc, 4, dst[1], src[0], zero); - set_pred_wr(pc, 1, 0, &pc->p->insns[pc->p->insns_nr - 2]); + if (mask & (1 << 2)) { + set_pred_wr(pc, 1, 0, &pc->p->insns[pc->p->insns_nr - 2]); - tmp[1] = temp_temp(pc); - emit_minmax(pc, 4, tmp[1], src[1], zero); + tmp[1] = temp_temp(pc); + emit_minmax(pc, 4, tmp[1], src[1], zero); - tmp[3] = temp_temp(pc); - emit_minmax(pc, 4, tmp[3], src[3], neg128); - emit_minmax(pc, 5, tmp[3], tmp[3], pos128); + tmp[3] = temp_temp(pc); + emit_minmax(pc, 4, tmp[3], src[3], neg128); + emit_minmax(pc, 5, tmp[3], tmp[3], pos128); - emit_pow(pc, dst[2], tmp[1], tmp[3]); - emit_mov(pc, dst[2], zero); - set_pred(pc, 3, 0, &pc->p->insns[pc->p->insns_nr - 2]); + emit_pow(pc, dst[2], tmp[1], tmp[3]); + emit_mov(pc, dst[2], zero); + set_pred(pc, 3, 0, &pc->p->insns[pc->p->insns_nr - 2]); + } } static void @@ -870,10 +897,13 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } break; case TGSI_OPCODE_COS: + temp = alloc_temp(pc, NULL); + emit_precossin(pc, temp, src[0][0]); + emit_flop(pc, 5, temp, temp); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_flop(pc, 5, dst[c], src[0][c]); + emit_mov(pc, dst[c], temp); } break; case TGSI_OPCODE_DP3: @@ -930,11 +960,12 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) break; case TGSI_OPCODE_EX2: temp = alloc_temp(pc, NULL); + emit_preex2(pc, temp, src[0][0]); + emit_flop(pc, 6, temp, temp); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_preex2(pc, temp, src[0][c]); - emit_flop(pc, 6, dst[c], temp); + emit_mov(pc, dst[c], temp); } free_temp(pc, temp); break; @@ -956,14 +987,15 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) free_temp(pc, temp); break; case TGSI_OPCODE_LIT: - /*XXX: writemask */ - emit_lit(pc, &dst[0], &src[0][0]); + emit_lit(pc, &dst[0], mask, &src[0][0]); break; case TGSI_OPCODE_LG2: + temp = alloc_temp(pc, NULL); + emit_flop(pc, 3, temp, src[0][0]); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_flop(pc, 3, dst[c], src[0][c]); + emit_mov(pc, dst[c], temp); } break; case TGSI_OPCODE_LRP: @@ -1027,16 +1059,24 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_flop(pc, 0, dst[c], src[0][c]); + emit_flop(pc, 0, dst[c], src[0][0]); } break; case TGSI_OPCODE_RSQ: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_flop(pc, 2, dst[c], src[0][c]); + emit_flop(pc, 2, dst[c], src[0][0]); } break; + case TGSI_OPCODE_SCS: + temp = alloc_temp(pc, NULL); + emit_precossin(pc, temp, src[0][0]); + if (mask & (1 << 0)) + emit_flop(pc, 5, dst[0], temp); + if (mask & (1 << 1)) + emit_flop(pc, 4, dst[1], temp); + break; case TGSI_OPCODE_SGE: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) @@ -1045,10 +1085,13 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } break; case TGSI_OPCODE_SIN: + temp = alloc_temp(pc, NULL); + emit_precossin(pc, temp, src[0][0]); + emit_flop(pc, 4, temp, temp); for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; - emit_flop(pc, 4, dst[c], src[0][c]); + emit_mov(pc, dst[c], temp); } break; case TGSI_OPCODE_SLT: -- cgit v1.2.3 From 7e9f6e290da23d7963857241b541e00c1fcf20dc Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 11 Jun 2008 13:00:20 +1000 Subject: nv50: move some magics --- src/gallium/drivers/nv50/nv50_program.c | 16 ++++++++++++++-- src/gallium/drivers/nv50/nv50_screen.c | 9 --------- 2 files changed, 14 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index d69eb4b86be..934b67fb711 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1492,14 +1492,26 @@ nv50_fragprog_validate(struct nv50_context *nv50) nv50_program_validate_data(nv50, p); nv50_program_validate_code(nv50, p); - so = so_new(7, 2); + so = so_new(64, 2); so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_method(so, tesla, 0x198c, 1); + so_method(so, tesla, 0x1904, 4); + so_data (so, 0x01040404); /* p: 0x01000404 */ + so_data (so, 0x00000004); + so_data (so, 0x00000000); + so_data (so, 0x00000000); + so_method(so, tesla, 0x16bc, 2); + so_data (so, 0x03020100); + so_data (so, 0x07060504); + so_method(so, tesla, 0x1988, 2); + so_data (so, 0x08040404); /* p: 0x0f000401 */ so_data (so, p->cfg.high_temp); + so_method(so, tesla, 0x16ac, 2); + so_data (so, 0x00000008); /* p: 0x00000004 */ + so_data (so, 0x00000004); so_method(so, tesla, 0x1414, 1); so_data (so, 0); /* program start offset */ so_emit(nv50->screen->nvws, so); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 6c0810a9cfe..2417bf80013 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -225,15 +225,6 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 0x54); so_method(so, screen->tesla, 0x13ac, 1); so_data (so, 1); - so_method(so, screen->tesla, 0x16bc, 2); - so_data (so, 0x03020100); - so_data (so, 0x07060504); - so_method(so, screen->tesla, 0x1988, 2); - so_data (so, 0x08040404); - so_data (so, 0x00000004); - so_method(so, screen->tesla, 0x16ac, 2); - so_data (so, 8); - so_data (so, 4); so_method(so, screen->tesla, 0x16b8, 1); so_data (so, 8); -- cgit v1.2.3 From bce558b37cde4be5c70117f49a2570f2988e5849 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 11 Jun 2008 13:15:23 +1000 Subject: Revert "nv50: move some magics" This reverts commit 0a38de30429d3075fc6dfc9ff3729c5ca11f0c2f. --- src/gallium/drivers/nv50/nv50_program.c | 16 ++-------------- src/gallium/drivers/nv50/nv50_screen.c | 9 +++++++++ 2 files changed, 11 insertions(+), 14 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 934b67fb711..d69eb4b86be 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1492,26 +1492,14 @@ nv50_fragprog_validate(struct nv50_context *nv50) nv50_program_validate_data(nv50, p); nv50_program_validate_code(nv50, p); - so = so_new(64, 2); + so = so_new(7, 2); so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_method(so, tesla, 0x1904, 4); - so_data (so, 0x01040404); /* p: 0x01000404 */ - so_data (so, 0x00000004); - so_data (so, 0x00000000); - so_data (so, 0x00000000); - so_method(so, tesla, 0x16bc, 2); - so_data (so, 0x03020100); - so_data (so, 0x07060504); - so_method(so, tesla, 0x1988, 2); - so_data (so, 0x08040404); /* p: 0x0f000401 */ + so_method(so, tesla, 0x198c, 1); so_data (so, p->cfg.high_temp); - so_method(so, tesla, 0x16ac, 2); - so_data (so, 0x00000008); /* p: 0x00000004 */ - so_data (so, 0x00000004); so_method(so, tesla, 0x1414, 1); so_data (so, 0); /* program start offset */ so_emit(nv50->screen->nvws, so); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 2417bf80013..6c0810a9cfe 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -225,6 +225,15 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 0x54); so_method(so, screen->tesla, 0x13ac, 1); so_data (so, 1); + so_method(so, screen->tesla, 0x16bc, 2); + so_data (so, 0x03020100); + so_data (so, 0x07060504); + so_method(so, screen->tesla, 0x1988, 2); + so_data (so, 0x08040404); + so_data (so, 0x00000004); + so_method(so, screen->tesla, 0x16ac, 2); + so_data (so, 8); + so_data (so, 4); so_method(so, screen->tesla, 0x16b8, 1); so_data (so, 8); -- cgit v1.2.3 From 585ae74d87f3d04a4b5b7c068b865292afd1a16b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 11 Jun 2008 13:24:32 +1000 Subject: nv50: move magics take 2 --- src/gallium/drivers/nv50/nv50_program.c | 13 +++++++++++-- src/gallium/drivers/nv50/nv50_screen.c | 9 --------- 2 files changed, 11 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index d69eb4b86be..b73003123d8 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1492,13 +1492,22 @@ nv50_fragprog_validate(struct nv50_context *nv50) nv50_program_validate_data(nv50, p); nv50_program_validate_code(nv50, p); - so = so_new(7, 2); + so = so_new(64, 2); so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_method(so, tesla, 0x198c, 1); + so_method(so, tesla, 0x1904, 4); + so_data (so, 0x01040404); /* p: 0x01000404 */ + so_data (so, 0x00000004); + so_data (so, 0x00000000); + so_data (so, 0x00000000); + so_method(so, tesla, 0x16bc, 2); /*XXX: fixme */ + so_data (so, 0x03020100); + so_data (so, 0x07060504); + so_method(so, tesla, 0x1988, 2); + so_data (so, 0x08040404); /* p: 0x0f000401 */ so_data (so, p->cfg.high_temp); so_method(so, tesla, 0x1414, 1); so_data (so, 0); /* program start offset */ diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 6c0810a9cfe..2417bf80013 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -225,15 +225,6 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 0x54); so_method(so, screen->tesla, 0x13ac, 1); so_data (so, 1); - so_method(so, screen->tesla, 0x16bc, 2); - so_data (so, 0x03020100); - so_data (so, 0x07060504); - so_method(so, screen->tesla, 0x1988, 2); - so_data (so, 0x08040404); - so_data (so, 0x00000004); - so_method(so, screen->tesla, 0x16ac, 2); - so_data (so, 8); - so_data (so, 4); so_method(so, screen->tesla, 0x16b8, 1); so_data (so, 8); -- cgit v1.2.3 From 19a1e9015e4ae429ab26e56848104fa209590338 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 11 Jun 2008 14:59:19 +1000 Subject: nv50: rename nv50_state.h to nv50_program.h --- src/gallium/drivers/nv50/nv50_context.h | 2 +- src/gallium/drivers/nv50/nv50_program.c | 1 - src/gallium/drivers/nv50/nv50_program.h | 37 +++++++++++++++++++++++++++++++++ src/gallium/drivers/nv50/nv50_state.c | 1 - src/gallium/drivers/nv50/nv50_state.h | 37 --------------------------------- src/gallium/drivers/nv50/nv50_vbo.c | 1 - 6 files changed, 38 insertions(+), 41 deletions(-) create mode 100644 src/gallium/drivers/nv50/nv50_program.h delete mode 100644 src/gallium/drivers/nv50/nv50_state.h diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 69c9dba675b..72d859c4685 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -15,8 +15,8 @@ struct nv50_screen *ctx = nv50->screen #include "nouveau/nouveau_push.h" -#include "nv50_state.h" #include "nv50_screen.h" +#include "nv50_program.h" #define NOUVEAU_ERR(fmt, args...) \ fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index b73003123d8..2118d8ef85f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -8,7 +8,6 @@ #include "tgsi/util/tgsi_util.h" #include "nv50_context.h" -#include "nv50_state.h" #define NV50_SU_MAX_TEMP 64 diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h new file mode 100644 index 00000000000..dd5aed799aa --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -0,0 +1,37 @@ +#ifndef __NV50_PROGRAM_H__ +#define __NV50_PROGRAM_H__ + +#include "pipe/p_state.h" +#include "tgsi/util/tgsi_scan.h" + +struct nv50_program_data { + int index; + int component; + float value; +}; + +struct nv50_program { + struct pipe_shader_state pipe; + struct tgsi_shader_info info; + boolean translated; + + unsigned type; + unsigned *insns; + unsigned insns_nr; + + struct pipe_buffer *buffer; + + float *immd; + unsigned immd_nr; + + struct nouveau_resource *data; + + struct { + unsigned high_temp; + struct { + unsigned attr[2]; + } vp; + } cfg; +}; + +#endif diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 774117326ad..ed6cca9a0dd 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -3,7 +3,6 @@ #include "pipe/p_util.h" #include "nv50_context.h" -#include "nv50_state.h" #include "nouveau/nouveau_stateobj.h" diff --git a/src/gallium/drivers/nv50/nv50_state.h b/src/gallium/drivers/nv50/nv50_state.h deleted file mode 100644 index d568bf610d5..00000000000 --- a/src/gallium/drivers/nv50/nv50_state.h +++ /dev/null @@ -1,37 +0,0 @@ -#ifndef __NV50_STATE_H__ -#define __NV50_STATE_H__ - -#include "pipe/p_state.h" -#include "tgsi/util/tgsi_scan.h" - -struct nv50_program_data { - int index; - int component; - float value; -}; - -struct nv50_program { - struct pipe_shader_state pipe; - struct tgsi_shader_info info; - boolean translated; - - unsigned type; - unsigned *insns; - unsigned insns_nr; - - struct pipe_buffer *buffer; - - float *immd; - unsigned immd_nr; - - struct nouveau_resource *data; - - struct { - unsigned high_temp; - struct { - unsigned attr[2]; - } vp; - } cfg; -}; - -#endif diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index badbef53df5..140d60cc9a6 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -3,7 +3,6 @@ #include "pipe/p_util.h" #include "nv50_context.h" -#include "nv50_state.h" static INLINE unsigned nv50_prim(unsigned mode) -- cgit v1.2.3 From f50e78e83cf7bda3537ac82de863096d829f13ee Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 11 Jun 2008 15:28:41 +1000 Subject: nv50: get vp working again, fp is broken regardless somehow.. g8x sucks :) --- src/gallium/drivers/nv50/nv50_context.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 72d859c4685..19ec492eb3f 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -24,8 +24,8 @@ fprintf(stderr, "nouveau: "fmt, ##args); /* Constant buffer assignment */ -#define NV50_CB_PMISC 1 -#define NV50_CB_PVP 0 +#define NV50_CB_PMISC 0 +#define NV50_CB_PVP 1 #define NV50_CB_PFP 2 #define NV50_CB_PGP 3 #define NV50_CB_TIC 4 -- cgit v1.2.3 From 40137ea2631a0c8158f99ae30ca90ed038b72076 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 12:16:43 +1000 Subject: nv50: carry instructions around in nv50_program_exec, not a flat array --- src/gallium/drivers/nv50/nv50_program.c | 482 +++++++++++++++++--------------- src/gallium/drivers/nv50/nv50_program.h | 22 +- 2 files changed, 268 insertions(+), 236 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 2118d8ef85f..3df38487617 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -196,168 +196,175 @@ alloc_immd(struct nv50_pc *pc, float f) return r; } +static struct nv50_program_exec * +exec(struct nv50_pc *pc) +{ + struct nv50_program_exec *e = CALLOC_STRUCT(nv50_program_exec); + + return e; +} + static void -emit(struct nv50_pc *pc, unsigned *inst) +emit(struct nv50_pc *pc, struct nv50_program_exec *e) { struct nv50_program *p = pc->p; - if (inst[0] & 1) { - p->insns_nr += 2; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 2), inst, sizeof(unsigned)*2); - } else { - p->insns_nr += 1; - p->insns = realloc(p->insns, sizeof(unsigned) * p->insns_nr); - memcpy(p->insns + (p->insns_nr - 1), inst, sizeof(unsigned)); - } + if (p->exec_tail) + p->exec_tail->next = e; + if (!p->exec_head) + p->exec_head = e; + p->exec_tail = e; + p->exec_size += (e->inst[0] & 1) ? 2 : 1; } -static INLINE void set_long(struct nv50_pc *, unsigned *); +static INLINE void set_long(struct nv50_pc *, struct nv50_program_exec *); static boolean -is_long(unsigned *inst) +is_long(struct nv50_program_exec *e) { - if (inst[0] & 1) + if (e->inst[0] & 1) return TRUE; return FALSE; } static boolean -is_immd(unsigned *inst) +is_immd(struct nv50_program_exec *e) { - if (is_long(inst) && (inst[1] & 3) == 3) + if (is_long(e) && (e->inst[1] & 3) == 3) return TRUE; return FALSE; } static INLINE void -set_pred(struct nv50_pc *pc, unsigned pred, unsigned idx, unsigned *inst) +set_pred(struct nv50_pc *pc, unsigned pred, unsigned idx, + struct nv50_program_exec *e) { - set_long(pc, inst); - inst[1] &= ~((0x1f << 7) | (0x3 << 12)); - inst[1] |= (pred << 7) | (idx << 12); + set_long(pc, e); + e->inst[1] &= ~((0x1f << 7) | (0x3 << 12)); + e->inst[1] |= (pred << 7) | (idx << 12); } static INLINE void -set_pred_wr(struct nv50_pc *pc, unsigned on, unsigned idx, unsigned *inst) +set_pred_wr(struct nv50_pc *pc, unsigned on, unsigned idx, + struct nv50_program_exec *e) { - set_long(pc, inst); - inst[1] &= ~((0x3 << 4) | (1 << 6)); - inst[1] |= (idx << 4) | (on << 6); + set_long(pc, e); + e->inst[1] &= ~((0x3 << 4) | (1 << 6)); + e->inst[1] |= (idx << 4) | (on << 6); } static INLINE void -set_long(struct nv50_pc *pc, unsigned *inst) +set_long(struct nv50_pc *pc, struct nv50_program_exec *e) { - if (is_long(inst)) + if (is_long(e)) return; - inst[0] |= 1; - set_pred(pc, 0xf, 0, inst); - set_pred_wr(pc, 0, 0, inst); + e->inst[0] |= 1; + set_pred(pc, 0xf, 0, e); + set_pred_wr(pc, 0, 0, e); } static INLINE void -set_dst(struct nv50_pc *pc, struct nv50_reg *dst, unsigned *inst) +set_dst(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_program_exec *e) { if (dst->type == P_RESULT) { - set_long(pc, inst); - inst[1] |= 0x00000008; + set_long(pc, e); + e->inst[1] |= 0x00000008; } alloc_reg(pc, dst); - inst[0] |= (dst->hw << 2); + e->inst[0] |= (dst->hw << 2); } static INLINE void -set_immd(struct nv50_pc *pc, struct nv50_reg *imm, unsigned *inst) +set_immd(struct nv50_pc *pc, struct nv50_reg *imm, struct nv50_program_exec *e) { unsigned val = fui(pc->immd_buf[imm->hw]); /* XXX */ - set_long(pc, inst); + set_long(pc, e); /*XXX: can't be predicated - bits overlap.. catch cases where both * are required and avoid them. */ - set_pred(pc, 0, 0, inst); - set_pred_wr(pc, 0, 0, inst); + set_pred(pc, 0, 0, e); + set_pred_wr(pc, 0, 0, e); - inst[1] |= 0x00000002 | 0x00000001; - inst[0] |= (val & 0x3f) << 16; - inst[1] |= (val >> 6) << 2; + e->inst[1] |= 0x00000002 | 0x00000001; + e->inst[0] |= (val & 0x3f) << 16; + e->inst[1] |= (val >> 6) << 2; } static void emit_interp(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src, struct nv50_reg *iv, boolean noperspective) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0x80000000; - set_dst(pc, dst, inst); + e->inst[0] |= 0x80000000; + set_dst(pc, dst, e); alloc_reg(pc, iv); - inst[0] |= (iv->hw << 9); + e->inst[0] |= (iv->hw << 9); alloc_reg(pc, src); - inst[0] |= (src->hw << 16); + e->inst[0] |= (src->hw << 16); if (noperspective) - inst[0] |= (1 << 25); + e->inst[0] |= (1 << 25); - emit(pc, inst); + emit(pc, e); } static void -set_cseg(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +set_cseg(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) { - set_long(pc, inst); + set_long(pc, e); if (src->type == P_IMMD) { - inst[1] |= (NV50_CB_PMISC << 22); + e->inst[1] |= (NV50_CB_PMISC << 22); } else { if (pc->p->type == PIPE_SHADER_VERTEX) - inst[1] |= (NV50_CB_PVP << 22); + e->inst[1] |= (NV50_CB_PVP << 22); else - inst[1] |= (NV50_CB_PFP << 22); + e->inst[1] |= (NV50_CB_PFP << 22); } } static void emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0x10000000; + e->inst[0] |= 0x10000000; - set_dst(pc, dst, inst); + set_dst(pc, dst, e); if (dst->type != P_RESULT && src->type == P_IMMD) { - set_immd(pc, src, inst); + set_immd(pc, src, e); /*XXX: 32-bit, but steals part of "half" reg space - need to * catch and handle this case if/when we do half-regs */ - inst[0] |= 0x00008000; + e->inst[0] |= 0x00008000; } else if (src->type == P_IMMD || src->type == P_CONST) { - set_long(pc, inst); - set_cseg(pc, src, inst); - inst[0] |= (src->hw << 9); - inst[1] |= 0x20000000; /* src0 const? */ + set_long(pc, e); + set_cseg(pc, src, e); + e->inst[0] |= (src->hw << 9); + e->inst[1] |= 0x20000000; /* src0 const? */ } else { if (src->type == P_ATTR) { - set_long(pc, inst); - inst[1] |= 0x00200000; + set_long(pc, e); + e->inst[1] |= 0x00200000; } alloc_reg(pc, src); - inst[0] |= (src->hw << 9); + e->inst[0] |= (src->hw << 9); } /* We really should support "half" instructions here at some point, * but I don't feel confident enough about them yet. */ - set_long(pc, inst); - if (is_long(inst) && !is_immd(inst)) { - inst[1] |= 0x04000000; /* 32-bit */ - inst[1] |= 0x0003c000; /* "subsubop" 0xf == mov */ + set_long(pc, e); + if (is_long(e) && !is_immd(e)) { + e->inst[1] |= 0x04000000; /* 32-bit */ + e->inst[1] |= 0x0003c000; /* "subsubop" 0xf == mov */ } - emit(pc, inst); + emit(pc, e); } static boolean @@ -385,11 +392,11 @@ check_swap_src_0_1(struct nv50_pc *pc, } static void -set_src_0(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +set_src_0(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) { if (src->type == P_ATTR) { - set_long(pc, inst); - inst[1] |= 0x00200000; + set_long(pc, e); + e->inst[1] |= 0x00200000; } else if (src->type == P_CONST || src->type == P_IMMD) { struct nv50_reg *temp = temp_temp(pc); @@ -399,11 +406,11 @@ set_src_0(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) } alloc_reg(pc, src); - inst[0] |= (src->hw << 9); + e->inst[0] |= (src->hw << 9); } static void -set_src_1(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +set_src_1(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) { if (src->type == P_ATTR) { struct nv50_reg *temp = temp_temp(pc); @@ -412,26 +419,26 @@ set_src_1(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) src = temp; } else if (src->type == P_CONST || src->type == P_IMMD) { - assert(!(inst[0] & 0x00800000)); - if (inst[0] & 0x01000000) { + assert(!(e->inst[0] & 0x00800000)); + if (e->inst[0] & 0x01000000) { struct nv50_reg *temp = temp_temp(pc); emit_mov(pc, temp, src); src = temp; } else { - set_cseg(pc, src, inst); - inst[0] |= 0x00800000; + set_cseg(pc, src, e); + e->inst[0] |= 0x00800000; } } alloc_reg(pc, src); - inst[0] |= (src->hw << 16); + e->inst[0] |= (src->hw << 16); } static void -set_src_2(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) +set_src_2(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) { - set_long(pc, inst); + set_long(pc, e); if (src->type == P_ATTR) { struct nv50_reg *temp = temp_temp(pc); @@ -440,186 +447,186 @@ set_src_2(struct nv50_pc *pc, struct nv50_reg *src, unsigned *inst) src = temp; } else if (src->type == P_CONST || src->type == P_IMMD) { - assert(!(inst[0] & 0x01000000)); - if (inst[0] & 0x00800000) { + assert(!(e->inst[0] & 0x01000000)); + if (e->inst[0] & 0x00800000) { struct nv50_reg *temp = temp_temp(pc); emit_mov(pc, temp, src); src = temp; } else { - set_cseg(pc, src, inst); - inst[0] |= 0x01000000; + set_cseg(pc, src, e); + e->inst[0] |= 0x01000000; } } alloc_reg(pc, src); - inst[1] |= (src->hw << 14); + e->inst[1] |= (src->hw << 14); } static void emit_mul(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xc0000000; - set_long(pc, inst); + e->inst[0] |= 0xc0000000; + set_long(pc, e); check_swap_src_0_1(pc, &src0, &src1); - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - set_src_1(pc, src1, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + set_src_1(pc, src1, e); - emit(pc, inst); + emit(pc, e); } static void emit_add(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xb0000000; + e->inst[0] |= 0xb0000000; check_swap_src_0_1(pc, &src0, &src1); - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - if (is_long(inst)) - set_src_2(pc, src1, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + if (is_long(e)) + set_src_2(pc, src1, e); else - set_src_1(pc, src1, inst); + set_src_1(pc, src1, e); - emit(pc, inst); + emit(pc, e); } static void emit_minmax(struct nv50_pc *pc, unsigned sub, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - set_long(pc, inst); - inst[0] |= 0xb0000000; - inst[1] |= (sub << 29); + set_long(pc, e); + e->inst[0] |= 0xb0000000; + e->inst[1] |= (sub << 29); check_swap_src_0_1(pc, &src0, &src1); - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - set_src_1(pc, src1, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + set_src_1(pc, src1, e); - emit(pc, inst); + emit(pc, e); } static void emit_sub(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xb0000000; + e->inst[0] |= 0xb0000000; - set_long(pc, inst); + set_long(pc, e); if (check_swap_src_0_1(pc, &src0, &src1)) - inst[1] |= 0x04000000; + e->inst[1] |= 0x04000000; else - inst[1] |= 0x08000000; + e->inst[1] |= 0x08000000; - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - set_src_2(pc, src1, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + set_src_2(pc, src1, e); - emit(pc, inst); + emit(pc, e); } static void emit_mad(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xe0000000; + e->inst[0] |= 0xe0000000; check_swap_src_0_1(pc, &src0, &src1); - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - set_src_1(pc, src1, inst); - set_src_2(pc, src2, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + set_src_1(pc, src1, e); + set_src_2(pc, src2, e); - emit(pc, inst); + emit(pc, e); } static void emit_msb(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xe0000000; - set_long(pc, inst); - inst[1] |= 0x08000000; /* src0 * src1 - src2 */ + e->inst[0] |= 0xe0000000; + set_long(pc, e); + e->inst[1] |= 0x08000000; /* src0 * src1 - src2 */ check_swap_src_0_1(pc, &src0, &src1); - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - set_src_1(pc, src1, inst); - set_src_2(pc, src2, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + set_src_1(pc, src1, e); + set_src_2(pc, src2, e); - emit(pc, inst); + emit(pc, e); } static void emit_flop(struct nv50_pc *pc, unsigned sub, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0x90000000; + e->inst[0] |= 0x90000000; if (sub) { - set_long(pc, inst); - inst[1] |= (sub << 29); + set_long(pc, e); + e->inst[1] |= (sub << 29); } - set_dst(pc, dst, inst); - set_src_0(pc, src, inst); + set_dst(pc, dst, e); + set_src_0(pc, src, e); - emit(pc, inst); + emit(pc, e); } static void emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xb0000000; + e->inst[0] |= 0xb0000000; - set_dst(pc, dst, inst); - set_src_0(pc, src, inst); - set_long(pc, inst); - inst[1] |= (6 << 29) | 0x00004000; + set_dst(pc, dst, e); + set_src_0(pc, src, e); + set_long(pc, e); + e->inst[1] |= (6 << 29) | 0x00004000; - emit(pc, inst); + emit(pc, e); } static void emit_precossin(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - inst[0] |= 0xb0000000; + e->inst[0] |= 0xb0000000; - set_dst(pc, dst, inst); - set_src_0(pc, src, inst); - set_long(pc, inst); - inst[1] |= (6 << 29); + set_dst(pc, dst, e); + set_src_0(pc, src, e); + set_long(pc, e); + e->inst[1] |= (6 << 29); - emit(pc, inst); + emit(pc, e); } static void emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); unsigned inv_cop[8] = { 0, 4, 2, 6, 1, 5, 3, 7 }; struct nv50_reg *rdst; @@ -632,26 +639,27 @@ emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, dst = alloc_temp(pc, NULL); /* set.u32 */ - set_long(pc, inst); - inst[0] |= 0xb0000000; - inst[1] |= (3 << 29); - inst[1] |= (c_op << 14); + set_long(pc, e); + e->inst[0] |= 0xb0000000; + e->inst[1] |= (3 << 29); + e->inst[1] |= (c_op << 14); /*XXX: breaks things, .u32 by default? * decuda will disasm as .u16 and use .lo/.hi regs, but this * doesn't seem to match what the hw actually does. inst[1] |= 0x04000000; << breaks things.. .u32 by default? */ - set_dst(pc, dst, inst); - set_src_0(pc, src0, inst); - set_src_1(pc, src1, inst); - emit(pc, inst); + set_dst(pc, dst, e); + set_src_0(pc, src0, e); + set_src_1(pc, src1, e); + emit(pc, e); /* cvt.f32.u32 */ - inst[0] = 0xa0000001; - inst[1] = 0x64014780; - set_dst(pc, rdst, inst); - set_src_0(pc, dst, inst); - emit(pc, inst); + e = exec(pc); + e->inst[0] = 0xa0000001; + e->inst[1] = 0x64014780; + set_dst(pc, rdst, e); + set_src_0(pc, dst, e); + emit(pc, e); if (dst != rdst) free_temp(pc, dst); @@ -660,19 +668,19 @@ emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst, static void emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; - - inst[0] = 0xa0000000; /* cvt */ - set_long(pc, inst); - inst[1] |= (6 << 29); /* cvt */ - inst[1] |= 0x08000000; /* integer mode */ - inst[1] |= 0x04000000; /* 32 bit */ - inst[1] |= ((0x1 << 3)) << 14; /* .rn */ - inst[1] |= (1 << 14); /* src .f32 */ - set_dst(pc, dst, inst); - set_src_0(pc, src, inst); - - emit(pc, inst); + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0xa0000000; /* cvt */ + set_long(pc, e); + e->inst[1] |= (6 << 29); /* cvt */ + e->inst[1] |= 0x08000000; /* integer mode */ + e->inst[1] |= 0x04000000; /* 32 bit */ + e->inst[1] |= ((0x1 << 3)) << 14; /* .rn */ + e->inst[1] |= (1 << 14); /* src .f32 */ + set_dst(pc, dst, e); + set_src_0(pc, src, e); + + emit(pc, e); } static void @@ -692,18 +700,18 @@ emit_pow(struct nv50_pc *pc, struct nv50_reg *dst, static void emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; - - inst[0] = 0xa0000000; /* cvt */ - set_long(pc, inst); - inst[1] |= (6 << 29); /* cvt */ - inst[1] |= 0x04000000; /* 32 bit */ - inst[1] |= (1 << 14); /* src .f32 */ - inst[1] |= ((1 << 6) << 14); /* .abs */ - set_dst(pc, dst, inst); - set_src_0(pc, src, inst); - - emit(pc, inst); + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0xa0000000; /* cvt */ + set_long(pc, e); + e->inst[1] |= (6 << 29); /* cvt */ + e->inst[1] |= 0x04000000; /* 32 bit */ + e->inst[1] |= (1 << 14); /* src .f32 */ + e->inst[1] |= ((1 << 6) << 14); /* .abs */ + set_dst(pc, dst, e); + set_src_0(pc, src, e); + + emit(pc, e); } static void @@ -731,7 +739,7 @@ emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, } if (mask & (1 << 2)) { - set_pred_wr(pc, 1, 0, &pc->p->insns[pc->p->insns_nr - 2]); + set_pred_wr(pc, 1, 0, pc->p->exec_tail); tmp[1] = temp_temp(pc); emit_minmax(pc, 4, tmp[1], src[1], zero); @@ -742,24 +750,24 @@ emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, emit_pow(pc, dst[2], tmp[1], tmp[3]); emit_mov(pc, dst[2], zero); - set_pred(pc, 3, 0, &pc->p->insns[pc->p->insns_nr - 2]); + set_pred(pc, 3, 0, pc->p->exec_tail); } } static void emit_neg(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e = exec(pc); - set_long(pc, inst); - inst[0] |= 0xa0000000; /* delta */ - inst[1] |= (7 << 29); /* delta */ - inst[1] |= 0x04000000; /* negate arg0? probably not */ - inst[1] |= (1 << 14); /* src .f32 */ - set_dst(pc, dst, inst); - set_src_0(pc, src, inst); + set_long(pc, e); + e->inst[0] |= 0xa0000000; /* delta */ + e->inst[1] |= (7 << 29); /* delta */ + e->inst[1] |= 0x04000000; /* negate arg0? probably not */ + e->inst[1] |= (1 << 14); /* src .f32 */ + set_dst(pc, dst, e); + set_src_0(pc, src, e); - emit(pc, inst); + emit(pc, e); } static struct nv50_reg * @@ -1132,20 +1140,21 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) if (sat) { for (c = 0; c < 4; c++) { - unsigned inst[2] = { 0, 0 }; + struct nv50_program_exec *e; if (!(mask & (1 << c))) continue; - - inst[0] = 0xa0000000; /* cvt */ - set_long(pc, inst); - inst[1] |= (6 << 29); /* cvt */ - inst[1] |= 0x04000000; /* 32 bit */ - inst[1] |= (1 << 14); /* src .f32 */ - inst[1] |= ((1 << 5) << 14); /* .sat */ - set_dst(pc, rdst[c], inst); - set_src_0(pc, dst[c], inst); - emit(pc, inst); + e = exec(pc); + + e->inst[0] = 0xa0000000; /* cvt */ + set_long(pc, e); + e->inst[1] |= (6 << 29); /* cvt */ + e->inst[1] |= 0x04000000; /* 32 bit */ + e->inst[1] |= (1 << 14); /* src .f32 */ + e->inst[1] |= ((1 << 5) << 14); /* .sat */ + set_dst(pc, rdst[c], e); + set_src_0(pc, dst[c], e); + emit(pc, e); } } @@ -1384,6 +1393,9 @@ nv50_program_tx(struct nv50_program *p) emit_mov(pc, &out, &pc->result[out.hw]); } + assert(is_long(pc->p->exec_tail) && !is_immd(pc->p->exec_head)); + pc->p->exec_tail->inst[1] |= 0x00000001; + p->immd_nr = pc->immd_nr * 4; p->immd = pc->immd_buf; @@ -1397,20 +1409,17 @@ out_cleanup: static void nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) { - int i; + struct nv50_program_exec *e; if (nv50_program_tx(p) == FALSE) assert(0); - /* *not* sufficient, it's fine if last inst is long and - * NOT immd - otherwise it's fucked fucked fucked */ - p->insns[p->insns_nr - 1] |= 0x00000001; - if (p->type == PIPE_SHADER_VERTEX) { - for (i = 0; i < p->insns_nr; i++) - NOUVEAU_ERR("VP0x%08x\n", p->insns[i]); - } else { - for (i = 0; i < p->insns_nr; i++) - NOUVEAU_ERR("FP0x%08x\n", p->insns[i]); + e = p->exec_head; + while (e) { + NOUVEAU_ERR("0x%08x\n", e->inst[0]); + if (is_long(e)) + NOUVEAU_ERR("0x%08x\n", e->inst[1]); + e = e->next; } p->translated = TRUE; @@ -1432,12 +1441,18 @@ static void nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) { struct pipe_winsys *ws = nv50->pipe.winsys; - void *map; + struct nv50_program_exec *e; + unsigned *map; if (!p->buffer) - p->buffer = ws->buffer_create(ws, 0x100, 0, p->insns_nr * 4); + p->buffer = ws->buffer_create(ws, 0x100, 0, p->exec_size * 4); + map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); - memcpy(map, p->insns, p->insns_nr * 4); + for (e = p->exec_head; e; e = e->next) { + *(map++) = e->inst[0]; + if (is_long(e)) + *(map++) = e->inst[1]; + } ws->buffer_unmap(ws, p->buffer); } @@ -1519,11 +1534,14 @@ nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p) { struct pipe_winsys *ws = nv50->pipe.winsys; - if (p->insns_nr) { - if (p->insns) - FREE(p->insns); - p->insns_nr = 0; + while (p->exec_head) { + struct nv50_program_exec *e = p->exec_head; + + p->exec_head = e->next; + FREE(e); } + p->exec_tail = NULL; + p->exec_size = 0; if (p->buffer) pipe_buffer_reference(ws, &p->buffer, NULL); diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h index dd5aed799aa..6dafe56fbb6 100644 --- a/src/gallium/drivers/nv50/nv50_program.h +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -10,22 +10,36 @@ struct nv50_program_data { float value; }; +struct nv50_program_exec { + struct nv50_program_exec *next; + + unsigned inst[2]; + struct { + int index; + unsigned mask; + unsigned shift; + } param; +}; + struct nv50_program { struct pipe_shader_state pipe; struct tgsi_shader_info info; boolean translated; unsigned type; - unsigned *insns; - unsigned insns_nr; + struct nv50_program_exec *exec_head; + struct nv50_program_exec *exec_tail; + unsigned exec_size; + struct nv50_program_data *data; + struct nouveau_resource *data_res; + unsigned data_nr; + unsigned data_start; struct pipe_buffer *buffer; float *immd; unsigned immd_nr; - struct nouveau_resource *data; - struct { unsigned high_temp; struct { -- cgit v1.2.3 From 1c7489bd7e5391136d0f2e68b467de89eb2d2bfc Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 12:26:43 +1000 Subject: nv50: allow relocating a shader's constants at upload time --- src/gallium/drivers/nv50/nv50_program.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 3df38487617..6c5aeb237db 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -201,6 +201,7 @@ exec(struct nv50_pc *pc) { struct nv50_program_exec *e = CALLOC_STRUCT(nv50_program_exec); + e->param.index = -1; return e; } @@ -311,7 +312,8 @@ emit_interp(struct nv50_pc *pc, struct nv50_reg *dst, } static void -set_cseg(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) +set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s, + struct nv50_program_exec *e) { set_long(pc, e); if (src->type == P_IMMD) { @@ -322,6 +324,10 @@ set_cseg(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) else e->inst[1] |= (NV50_CB_PFP << 22); } + + e->param.index = src->hw; + e->param.shift = s; + e->param.mask = m << (s % 32); } static void @@ -342,8 +348,7 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) } else if (src->type == P_IMMD || src->type == P_CONST) { set_long(pc, e); - set_cseg(pc, src, e); - e->inst[0] |= (src->hw << 9); + set_data(pc, src, 0x7f, 9, e); e->inst[1] |= 0x20000000; /* src0 const? */ } else { if (src->type == P_ATTR) { @@ -426,7 +431,7 @@ set_src_1(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) emit_mov(pc, temp, src); src = temp; } else { - set_cseg(pc, src, e); + set_data(pc, src, 0x7f, 16, e); e->inst[0] |= 0x00800000; } } @@ -454,7 +459,7 @@ set_src_2(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) emit_mov(pc, temp, src); src = temp; } else { - set_cseg(pc, src, e); + set_data(pc, src, 0x7f, 32+14, e); e->inst[0] |= 0x01000000; } } @@ -1449,6 +1454,12 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); for (e = p->exec_head; e; e = e->next) { + if (e->param.index >= 0) { + e->inst[e->param.shift / 32] &= ~e->param.mask; + e->inst[e->param.shift / 32] |= (e->param.index << + e->param.shift); + } + *(map++) = e->inst[0]; if (is_long(e)) *(map++) = e->inst[1]; -- cgit v1.2.3 From aea1669ff221f97682f0be6a60632e40c2739d09 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 12:39:35 +1000 Subject: nv50: use constbuf segment 0 for everything - I can't make the others work.. --- src/gallium/drivers/nv50/nv50_program.c | 70 ++++++++++++++++++++++++++------- src/gallium/drivers/nv50/nv50_program.h | 11 +----- 2 files changed, 57 insertions(+), 24 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 6c5aeb237db..cbbecafb028 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -10,6 +10,7 @@ #include "nv50_context.h" #define NV50_SU_MAX_TEMP 64 +#define NV50_PROGRAM_DUMP /* ARL - gallium craps itself on progs/vp/arl.txt * @@ -316,6 +317,9 @@ set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s, struct nv50_program_exec *e) { set_long(pc, e); +#if 1 + e->inst[1] |= (1 << 22); +#else if (src->type == P_IMMD) { e->inst[1] |= (NV50_CB_PMISC << 22); } else { @@ -324,6 +328,7 @@ set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s, else e->inst[1] |= (NV50_CB_PFP << 22); } +#endif e->param.index = src->hw; e->param.shift = s; @@ -1335,7 +1340,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) } if (pc->immd_nr) { - int rid = 0; + int rid = pc->param_nr * 4; pc->immd = calloc(pc->immd_nr * 4, sizeof(struct nv50_reg)); if (!pc->immd) @@ -1401,6 +1406,7 @@ nv50_program_tx(struct nv50_program *p) assert(is_long(pc->p->exec_tail) && !is_immd(pc->p->exec_head)); pc->p->exec_tail->inst[1] |= 0x00000001; + p->param_nr = pc->param_nr * 4; p->immd_nr = pc->immd_nr * 4; p->immd = pc->immd_buf; @@ -1418,26 +1424,47 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) if (nv50_program_tx(p) == FALSE) assert(0); - - e = p->exec_head; - while (e) { - NOUVEAU_ERR("0x%08x\n", e->inst[0]); - if (is_long(e)) - NOUVEAU_ERR("0x%08x\n", e->inst[1]); - e = e->next; - } - p->translated = TRUE; } static void nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) { + struct nouveau_winsys *nvws = nv50->screen->nvws; + struct pipe_winsys *ws = nv50->pipe.winsys; + unsigned nr = p->param_nr + p->immd_nr; int i; + if (!p->data && nr) { + struct nouveau_resource *heap = nv50->screen->vp_data_heap; + + if (nvws->res_alloc(heap, nr, p, &p->data)) { + while (heap->next && heap->size < nr) { + struct nv50_program *evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, nr, p, &p->data)) + assert(0); + } + } + + if (p->param_nr) { + float *map = ws->buffer_map(ws, nv50->constbuf[p->type], + PIPE_BUFFER_USAGE_CPU_READ); + for (i = 0; i < p->param_nr; i++) { + BEGIN_RING(tesla, 0x0f00, 2); + OUT_RING ((NV50_CB_PMISC << 0) | + ((p->data->start + i) << 8)); + OUT_RING (fui(map[i])); + } + ws->buffer_unmap(ws, nv50->constbuf[p->type]); + } + for (i = 0; i < p->immd_nr; i++) { BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING ((NV50_CB_PMISC << 0) | (i << 8)); + OUT_RING ((NV50_CB_PMISC << 0) | + ((p->data_start + p->param_nr + i) << 8)); OUT_RING (fui(p->immd[i])); } } @@ -1452,17 +1479,30 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) if (!p->buffer) p->buffer = ws->buffer_create(ws, 0x100, 0, p->exec_size * 4); - map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); - for (e = p->exec_head; e; e = e->next) { - if (e->param.index >= 0) { + if (p->data && p->data->start != p->data_start) { + for (e = p->exec_head; e; e = e->next) { + if (e->param.index < 0) + continue; e->inst[e->param.shift / 32] &= ~e->param.mask; e->inst[e->param.shift / 32] |= (e->param.index << e->param.shift); } + p->data_start = p->data->start; + } + + map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); + for (e = p->exec_head; e; e = e->next) { +#ifdef NV50_PROGRAM_DUMP + NOUVEAU_ERR("0x%08x\n", e->inst[0]); +#endif *(map++) = e->inst[0]; - if (is_long(e)) + if (is_long(e)) { +#ifdef NV50_PROGRAM_DUMP + NOUVEAU_ERR("0x%08x\n", e->inst[1]); +#endif *(map++) = e->inst[1]; + } } ws->buffer_unmap(ws, p->buffer); } diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h index 6dafe56fbb6..d143ae97979 100644 --- a/src/gallium/drivers/nv50/nv50_program.h +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -4,12 +4,6 @@ #include "pipe/p_state.h" #include "tgsi/util/tgsi_scan.h" -struct nv50_program_data { - int index; - int component; - float value; -}; - struct nv50_program_exec { struct nv50_program_exec *next; @@ -30,15 +24,14 @@ struct nv50_program { struct nv50_program_exec *exec_head; struct nv50_program_exec *exec_tail; unsigned exec_size; - struct nv50_program_data *data; - struct nouveau_resource *data_res; - unsigned data_nr; + struct nouveau_resource *data; unsigned data_start; struct pipe_buffer *buffer; float *immd; unsigned immd_nr; + unsigned param_nr; struct { unsigned high_temp; -- cgit v1.2.3 From ab3d55e2e3578db8deba84dcf47a024071486bd8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 13:11:41 +1000 Subject: nv50: more efficient const upload + fixes (fp/* works now!) --- src/gallium/drivers/nv50/nv50_program.c | 48 +++++++++++++++++++++------------ src/gallium/drivers/nv50/nv50_screen.c | 4 +-- 2 files changed, 33 insertions(+), 19 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index cbbecafb028..4c41e5a7c24 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1420,20 +1420,35 @@ out_cleanup: static void nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) { - struct nv50_program_exec *e; - if (nv50_program_tx(p) == FALSE) assert(0); p->translated = TRUE; } +static void +nv50_program_upload_data(struct nv50_context *nv50, float *map, + unsigned start, unsigned count) +{ + while (count) { + unsigned nr = count > 2047 ? 2047 : count; + + BEGIN_RING(tesla, 0x00000f00, 1); + OUT_RING ((NV50_CB_PMISC << 0) | (start << 8)); + BEGIN_RING(tesla, 0x40000f04, nr); + OUT_RINGp (map, nr); + + map += nr; + start += nr; + count -= nr; + } +} + static void nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) { struct nouveau_winsys *nvws = nv50->screen->nvws; struct pipe_winsys *ws = nv50->pipe.winsys; unsigned nr = p->param_nr + p->immd_nr; - int i; if (!p->data && nr) { struct nouveau_resource *heap = nv50->screen->vp_data_heap; @@ -1452,20 +1467,15 @@ nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) if (p->param_nr) { float *map = ws->buffer_map(ws, nv50->constbuf[p->type], PIPE_BUFFER_USAGE_CPU_READ); - for (i = 0; i < p->param_nr; i++) { - BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING ((NV50_CB_PMISC << 0) | - ((p->data->start + i) << 8)); - OUT_RING (fui(map[i])); - } + nv50_program_upload_data(nv50, map, p->data->start, + p->param_nr); ws->buffer_unmap(ws, nv50->constbuf[p->type]); } - for (i = 0; i < p->immd_nr; i++) { - BEGIN_RING(tesla, 0x0f00, 2); - OUT_RING ((NV50_CB_PMISC << 0) | - ((p->data_start + p->param_nr + i) << 8)); - OUT_RING (fui(p->immd[i])); + if (p->immd_nr) { + nv50_program_upload_data(nv50, p->immd, + p->data->start + p->param_nr, + p->immd_nr); } } @@ -1481,11 +1491,15 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) if (p->data && p->data->start != p->data_start) { for (e = p->exec_head; e; e = e->next) { + unsigned ei, ci; + if (e->param.index < 0) continue; - e->inst[e->param.shift / 32] &= ~e->param.mask; - e->inst[e->param.shift / 32] |= (e->param.index << - e->param.shift); + ei = e->param.shift >> 5; + ci = e->param.index + p->data->start; + + e->inst[ei] &= ~e->param.mask; + e->inst[ei] |= (ci << e->param.shift); } p->data_start = p->data->start; diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 2417bf80013..5f10fe2e440 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -229,8 +229,8 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 8); /* Shared constant buffer */ - screen->constbuf = ws->buffer_create(ws, 0, 0, 256 * 4 * 4); - if (nvws->res_init(&screen->vp_data_heap, 0, 256)) { + screen->constbuf = ws->buffer_create(ws, 0, 0, 128 * 4 * 4); + if (nvws->res_init(&screen->vp_data_heap, 0, 128)) { NOUVEAU_ERR("Error initialising constant buffer\n"); nv50_screen_destroy(&screen->pipe); return NULL; -- cgit v1.2.3 From f700d6be6335a4d4394296891f783687b6f2d4f2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 13:50:56 +1000 Subject: nv50: remove some cruft, don't upload program unless really needed --- src/gallium/drivers/nv50/nv50_program.c | 9 ++++++- src/gallium/drivers/nv50/nv50_state_validate.c | 33 ++------------------------ 2 files changed, 10 insertions(+), 32 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 4c41e5a7c24..f71aa19312f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1484,10 +1484,13 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) { struct pipe_winsys *ws = nv50->pipe.winsys; struct nv50_program_exec *e; + boolean upload = FALSE; unsigned *map; - if (!p->buffer) + if (!p->buffer) { p->buffer = ws->buffer_create(ws, 0x100, 0, p->exec_size * 4); + upload = TRUE; + } if (p->data && p->data->start != p->data_start) { for (e = p->exec_head; e; e = e->next) { @@ -1503,8 +1506,12 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) } p->data_start = p->data->start; + upload = TRUE; } + if (!upload) + return FALSE; + map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); for (e = p->exec_head; e; e = e->next) { #ifdef NV50_PROGRAM_DUMP diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 05395c6df7c..5f2244e3e80 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -114,10 +114,10 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & NV50_NEW_ZSA) so_emit(nvws, nv50->zsa->so); - if (nv50->dirty & NV50_NEW_VERTPROG) + if (nv50->dirty & (NV50_NEW_VERTPROG | NV50_NEW_VERTPROG_CB)) nv50_vertprog_validate(nv50); - if (nv50->dirty & NV50_NEW_FRAGPROG) + if (nv50->dirty & (NV50_NEW_FRAGPROG | NV50_NEW_FRAGPROG_CB)) nv50_fragprog_validate(nv50); if (nv50->dirty & NV50_NEW_RASTERIZER) @@ -168,35 +168,6 @@ nv50_state_validate(struct nv50_context *nv50) so_ref(NULL, &so); } - if (nv50->dirty & NV50_NEW_VERTPROG_CB) { - so = so_new(4, 2); - so_method(so, tesla, 0x1280, 3); - so_reloc (so, nv50->constbuf[PIPE_SHADER_VERTEX], 0, - NOUVEAU_BO_HIGH | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, - 0, 0); - so_reloc (so, nv50->constbuf[PIPE_SHADER_VERTEX], 0, - NOUVEAU_BO_LOW | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, - 0, 0); - so_data (so, (NV50_CB_PVP << 16) | 0x1000); - so_emit(nvws, so); - so_ref(NULL, &so); - } - - if (nv50->dirty & NV50_NEW_FRAGPROG_CB) { - so = so_new(4, 2); - so_method(so, tesla, 0x1280, 3); - so_reloc (so, nv50->constbuf[PIPE_SHADER_FRAGMENT], 0, - NOUVEAU_BO_HIGH | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, - 0, 0); - so_reloc (so, nv50->constbuf[PIPE_SHADER_FRAGMENT], 0, - NOUVEAU_BO_LOW | NOUVEAU_BO_RD | NOUVEAU_BO_VRAM, - 0, 0); - so_data (so, (NV50_CB_PFP << 16) | 0x1000); - so_emit(nvws, so); - so_ref(NULL, &so); - } - - if (nv50->dirty & NV50_NEW_ARRAYS) nv50_vbo_validate(nv50); -- cgit v1.2.3 From 163d9aa1fe33612e7806549e47b257b61ca5045e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 13:59:51 +1000 Subject: nv50: support a couple more common VBO formats --- src/gallium/drivers/nv50/nv50_vbo.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 140d60cc9a6..74519489f7d 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -82,9 +82,15 @@ nv50_vbo_validate(struct nv50_context *nv50) &nv50->vtxbuf[ve->vertex_buffer_index]; switch (ve->src_format) { + case PIPE_FORMAT_R32G32B32A32_FLOAT: + so_data(vtxfmt, 0x7e080000 | i); + break; case PIPE_FORMAT_R32G32B32_FLOAT: so_data(vtxfmt, 0x7e100000 | i); break; + case PIPE_FORMAT_R8G8B8A8_UNORM: + so_data(vtxfmt, 0x24500000 | i); + break; default: { char fmt[128]; -- cgit v1.2.3 From 027ed25c12f69b39e205d3bbd26b68e9a02bea81 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 14:15:38 +1000 Subject: nv50: draw_elements() - inline only for the moment --- src/gallium/drivers/nv50/nv50_vbo.c | 96 ++++++++++++++++++++++++++++++++++++- 1 file changed, 95 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 74519489f7d..586a6c49de0 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -52,16 +52,110 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, return TRUE; } +static INLINE void +nv50_draw_elements_inline_u08(struct nv50_context *nv50, uint8_t *map, + unsigned start, unsigned count) +{ + map += start; + + if (count & 1) { + BEGIN_RING(tesla, 0x15e8, 1); + OUT_RING (map[0]); + map++; + count--; + } + + while (count) { + unsigned nr = count > 2046 ? 2046 : count; + int i; + + BEGIN_RING(tesla, 0x400015f0, nr >> 1); + for (i = 0; i < nr; i += 2) + OUT_RING ((map[1] << 16) | map[0]); + + count -= nr; + map += nr; + } +} + +static INLINE void +nv50_draw_elements_inline_u16(struct nv50_context *nv50, uint16_t *map, + unsigned start, unsigned count) +{ + map += start; + + if (count & 1) { + BEGIN_RING(tesla, 0x15e8, 1); + OUT_RING (map[0]); + map++; + count--; + } + + while (count) { + unsigned nr = count > 2046 ? 2046 : count; + int i; + + BEGIN_RING(tesla, 0x400015f0, nr >> 1); + for (i = 0; i < nr; i += 2) + OUT_RING ((map[1] << 16) | map[0]); + + count -= nr; + map += nr; + } +} + +static INLINE void +nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint8_t *map, + unsigned start, unsigned count) +{ + map += start; + + while (count) { + unsigned nr = count > 2047 ? 2047 : count; + + BEGIN_RING(tesla, 0x400015e8, nr); + OUT_RINGp (map, nr); + + count -= nr; + map += nr; + } +} + boolean nv50_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { struct nv50_context *nv50 = nv50_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + void *map = ws->buffer_map(ws, indexBuffer, PIPE_BUFFER_USAGE_CPU_READ); nv50_state_validate(nv50); - NOUVEAU_ERR("unimplemented\n"); + BEGIN_RING(tesla, 0x142c, 1); + OUT_RING (0); + BEGIN_RING(tesla, 0x142c, 1); + OUT_RING (0); + + BEGIN_RING(tesla, NV50TCL_VERTEX_BEGIN, 1); + OUT_RING (nv50_prim(mode)); + switch (indexSize) { + case 1: + nv50_draw_elements_inline_u08(nv50, map, start, count); + break; + case 2: + nv50_draw_elements_inline_u16(nv50, map, start, count); + break; + case 4: + nv50_draw_elements_inline_u32(nv50, map, start, count); + break; + default: + assert(0); + } + BEGIN_RING(tesla, NV50TCL_VERTEX_END, 1); + OUT_RING (0); + + pipe->flush(pipe, 0, NULL); return TRUE; } -- cgit v1.2.3 From 619549a6377a58d54c9cf55f8863beed56b09566 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 14:21:28 +1000 Subject: nv50: valgrind complaint --- src/gallium/drivers/nv50/nv50_program.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index f71aa19312f..af5ca84c8e7 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -146,8 +146,10 @@ static void free_temp(struct nv50_pc *pc, struct nv50_reg *r) { if (r->index == -1) { - FREE(pc->r_temp[r->hw]); - pc->r_temp[r->hw] = NULL; + unsigned hw = r->hw; + + FREE(pc->r_temp[hw]); + pc->r_temp[hw] = NULL; } } -- cgit v1.2.3 From 4bde3a72ab0b4246cd779a6d1e2a72943f25c0f6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 14:40:20 +1000 Subject: nv50: fix blend cso --- src/gallium/drivers/nv50/nv50_state.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index ed6cca9a0dd..b56a3992bc4 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -29,12 +29,12 @@ nv50_blend_state_create(struct pipe_context *pipe, so_data(so, 1); so_method(so, tesla, NV50TCL_BLEND_EQUATION_RGB, 5); so_data (so, nvgl_blend_eqn(cso->rgb_func)); - so_data (so, nvgl_blend_func(cso->rgb_src_factor)); - so_data (so, nvgl_blend_func(cso->rgb_dst_factor)); + so_data (so, 0x4000 | nvgl_blend_func(cso->rgb_src_factor)); + so_data (so, 0x4000 | nvgl_blend_func(cso->rgb_dst_factor)); so_data (so, nvgl_blend_eqn(cso->alpha_func)); - so_data (so, nvgl_blend_func(cso->alpha_src_factor)); + so_data (so, 0x4000 | nvgl_blend_func(cso->alpha_src_factor)); so_method(so, tesla, NV50TCL_BLEND_FUNC_DST_ALPHA, 1); - so_data (so, nvgl_blend_func(cso->alpha_dst_factor)); + so_data (so, 0x4000 | nvgl_blend_func(cso->alpha_dst_factor)); } if (cso->logicop_enable == 0 ) { -- cgit v1.2.3 From 4d520e0b76cf54ae8eb5464afc126c6cc5c6bfdc Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 14:47:17 +1000 Subject: nv50: another vbo format --- src/gallium/drivers/nv50/nv50_vbo.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 586a6c49de0..7a1c2f24ef9 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -182,6 +182,9 @@ nv50_vbo_validate(struct nv50_context *nv50) case PIPE_FORMAT_R32G32B32_FLOAT: so_data(vtxfmt, 0x7e100000 | i); break; + case PIPE_FORMAT_R32G32_FLOAT: + so_data(vtxfmt, 0x7e200000 | i); + break; case PIPE_FORMAT_R8G8B8A8_UNORM: so_data(vtxfmt, 0x24500000 | i); break; -- cgit v1.2.3 From 2fdeb4d5a5cc8b93bf885ba646e3a29a68c755ed Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 12 Jun 2008 23:49:26 +1000 Subject: nv50: comment on a so-far unseen bug --- src/gallium/drivers/nv50/nv50_program.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index af5ca84c8e7..39597c54814 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -22,6 +22,9 @@ * - Maybe even relax restrictions a bit, can't do P_RESULT + P_IMMD, * but can emit to P_TEMP first - then MOV later. NVIDIA does this * + * In ops such as ADD it's possible to construct a bad opcode in the !is_long() + * case, if the emit_src() causes the inst to suddenly become long. + * * Verify half-insns work where expected - and force disable them where they * don't work - MUL has it forcibly disabled atm as it fixes POW.. * -- cgit v1.2.3 From 101305f37f7268354a50b825bcb66894e4a0b777 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 13 Jun 2008 10:58:27 +1000 Subject: nv50: flag to indicate to winsys we want a surface for use as a zeta buffer NVIDIA love to make life difficult.. we need different flags in PTEs for zeta.. yay.. not. --- src/gallium/drivers/nouveau/nouveau_winsys.h | 1 + src/gallium/drivers/nv50/nv50_miptree.c | 13 ++++++++++++- src/gallium/drivers/nv50/nv50_screen.c | 4 ++++ src/gallium/drivers/nv50/nv50_state.c | 4 ++-- src/gallium/drivers/nv50/nv50_state_validate.c | 10 ++++++++++ 5 files changed, 29 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 07a41dcf4a1..13810460bfe 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -18,6 +18,7 @@ #define NOUVEAU_CAP_HW_IDXBUF (0xbeef0001) #define NOUVEAU_BUFFER_USAGE_TEXTURE (1 << 16) +#define NOUVEAU_BUFFER_USAGE_ZETA (1 << 17) struct nouveau_winsys { struct nouveau_context *nv; diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index be85c3fd5ce..459e0ff9dd6 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -21,6 +21,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { struct pipe_winsys *ws = pscreen->winsys; struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree); + unsigned usage; NOUVEAU_ERR("unimplemented\n"); @@ -28,7 +29,17 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) mt->base.refcount = 1; mt->base.screen = pscreen; - mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + usage = PIPE_BUFFER_USAGE_PIXEL; + switch (pt->format) { + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + usage |= NOUVEAU_BUFFER_USAGE_ZETA; + break; + default: + break; + } + + mt->buffer = ws->buffer_create(ws, 256, usage, pt->width[0] * pt->cpp * pt->height[0]); if (!mt->buffer) { FREE(mt); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 5f10fe2e440..d76cce9f791 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -284,6 +284,10 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 0xffffffff); } + so_method(so, screen->tesla, NV50TCL_DEPTH_RANGE_NEAR, 2); + so_data (so, fui(0.0)); + so_data (so, fui(1.0)); + so_emit(nvws, so); so_ref(NULL, &so); nvws->push_flush(nvws, 0, NULL); diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index b56a3992bc4..176ebff5da4 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -247,8 +247,8 @@ nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nouveau_stateobj *so = so_new(64, 0); so_method(so, tesla, NV50TCL_DEPTH_WRITE_ENABLE, 1); - so_data (so, 0); //cso->depth.writemask ? 1 : 0); - if (0 && cso->depth.enabled) { + so_data (so, cso->depth.writemask ? 1 : 0); + if (cso->depth.enabled) { so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1); so_data (so, 1); so_method(so, tesla, NV50TCL_DEPTH_TEST_FUNC, 1); diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 5f2244e3e80..a3f399a139f 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -70,6 +70,9 @@ nv50_state_validate_fb(struct nv50_context *nv50) case PIPE_FORMAT_Z24S8_UNORM: so_data(so, 0x16); break; + case PIPE_FORMAT_Z16_UNORM: + so_data(so, 0x15); + break; default: { char fmt[128]; @@ -81,6 +84,13 @@ nv50_state_validate_fb(struct nv50_context *nv50) } so_data(so, 0x00000040); so_data(so, 0x00000000); + + so_method(so, tesla, 0x1538, 1); + so_data (so, 1); + so_method(so, tesla, 0x1228, 3); + so_data (so, fb->zsbuf->width); + so_data (so, fb->zsbuf->height); + so_data (so, 0x00010001); } so_method(so, tesla, NV50TCL_VIEWPORT_HORIZ, 2); -- cgit v1.2.3 From 0d7f25c890e1f1505625542c256d4512c065449a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 13 Jun 2008 11:50:13 +1000 Subject: nv50: disable ztest for now - it doesn't work still --- src/gallium/drivers/nv50/nv50_state.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 176ebff5da4..0129b0fb144 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -247,10 +247,10 @@ nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nouveau_stateobj *so = so_new(64, 0); so_method(so, tesla, NV50TCL_DEPTH_WRITE_ENABLE, 1); - so_data (so, cso->depth.writemask ? 1 : 0); + so_data (so, 0); //cso->depth.writemask ? 1 : 0); if (cso->depth.enabled) { so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1); - so_data (so, 1); + so_data (so, 0); //1); so_method(so, tesla, NV50TCL_DEPTH_TEST_FUNC, 1); so_data (so, nvgl_comparison_op(cso->depth.func)); } else { -- cgit v1.2.3 From c0ed6a871cd3513e17a1fab960f5626485ffed13 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 13 Jun 2008 12:09:46 +1000 Subject: nv50: do tsc/tic upload + stub out shader TEX stuff --- src/gallium/drivers/nv50/nv50_context.h | 17 +++++++++++++ src/gallium/drivers/nv50/nv50_miptree.c | 11 --------- src/gallium/drivers/nv50/nv50_program.c | 6 +++++ src/gallium/drivers/nv50/nv50_screen.c | 1 + src/gallium/drivers/nv50/nv50_state.c | 27 ++++++++++++++++++++- src/gallium/drivers/nv50/nv50_state_validate.c | 33 ++++++++++++++++++++++++++ 6 files changed, 83 insertions(+), 12 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 19ec492eb3f..5214ce1a69c 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -44,6 +44,8 @@ #define NV50_NEW_FRAGPROG (1 << 10) #define NV50_NEW_FRAGPROG_CB (1 << 11) #define NV50_NEW_ARRAYS (1 << 12) +#define NV50_NEW_SAMPLER (1 << 13) +#define NV50_NEW_TEXTURE (1 << 14) struct nv50_blend_stateobj { struct pipe_blend_state pipe; @@ -60,6 +62,17 @@ struct nv50_rasterizer_stateobj { struct nouveau_stateobj *so; }; +struct nv50_miptree { + struct pipe_texture base; + struct pipe_buffer *buffer; +}; + +static INLINE struct nv50_miptree * +nv50_miptree(struct pipe_texture *pt) +{ + return (struct nv50_miptree *)pt; +} + struct nv50_context { struct pipe_context pipe; @@ -84,6 +97,10 @@ struct nv50_context { unsigned vtxbuf_nr; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; unsigned vtxelt_nr; + unsigned *sampler[PIPE_MAX_SAMPLERS]; + unsigned sampler_nr; + struct nv50_miptree *miptree[PIPE_MAX_SAMPLERS]; + unsigned miptree_nr; }; static INLINE struct nv50_context * diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 459e0ff9dd6..cdd98844f94 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -5,17 +5,6 @@ #include "nv50_context.h" -struct nv50_miptree { - struct pipe_texture base; - struct pipe_buffer *buffer; -}; - -static INLINE struct nv50_miptree * -nv50_miptree(struct pipe_texture *pt) -{ - return (struct nv50_miptree *)pt; -} - static struct pipe_texture * nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 39597c54814..5eec68e5e19 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -828,6 +828,8 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src) case TGSI_FILE_IMMEDIATE: r = &pc->immd[src->SrcRegister.Index * 4 + c]; break; + case TGSI_FILE_SAMPLER: + break; default: assert(0); break; @@ -1130,6 +1132,8 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_sub(pc, dst[c], src[0][c], src[1][c]); } break; + case TGSI_OPCODE_TEX: + break; case TGSI_OPCODE_XPD: temp = alloc_temp(pc, NULL); if (mask & (1 << 0)) { @@ -1226,6 +1230,8 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->param_nr < (last + 1)) pc->param_nr = last + 1; break; + case TGSI_FILE_SAMPLER: + break; default: NOUVEAU_ERR("bad decl file %d\n", d->Declaration.File); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index d76cce9f791..b9cecb77f1f 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -28,6 +28,7 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen, break; case PIPE_TEXTURE: switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_I8_UNORM: return TRUE; default: diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 0129b0fb144..f36299db4d2 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -1,6 +1,7 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" +#include "pipe/p_inlines.h" #include "nv50_context.h" @@ -85,23 +86,47 @@ static void * nv50_sampler_state_create(struct pipe_context *pipe, const struct pipe_sampler_state *cso) { - return NULL; + unsigned *tsc = CALLOC(8, sizeof(unsigned)); + + tsc[0] = 0x00024080; + tsc[1] = 0x00000062; + + return (void *)tsc; } static void nv50_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) { + struct nv50_context *nv50 = nv50_context(pipe); + int i; + + nv50->sampler_nr = nr; + for (i = 0; i < nv50->sampler_nr; i++) + nv50->sampler[i] = sampler[i]; + + nv50->dirty |= NV50_NEW_SAMPLER; } static void nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { + FREE(hwcso); } static void nv50_set_sampler_texture(struct pipe_context *pipe, unsigned nr, struct pipe_texture **pt) { + struct nv50_context *nv50 = nv50_context(pipe); + int i; + + for (i = 0; i < nr; i++) + pipe_texture_reference(&nv50->miptree[i], pt[i]); + for (i = nr; i < nv50->miptree_nr; i++) + pipe_texture_reference(&nv50->miptree[i], NULL); + + nv50->miptree_nr = nr; + nv50->dirty |= NV50_NEW_TEXTURE; } static void * diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index a3f399a139f..63c1756bcbf 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -178,6 +178,39 @@ nv50_state_validate(struct nv50_context *nv50) so_ref(NULL, &so); } + if (nv50->dirty & NV50_NEW_SAMPLER) { + int i; + + BEGIN_RING(tesla, 0x0f00, 1); + OUT_RING ((NV50_CB_TSC << 0) | (0 << 8)); + BEGIN_RING(tesla, 0x40000f04, nv50->sampler_nr * 8); + for (i = 0; i < nv50->sampler_nr; i++) + OUT_RINGp(nv50->sampler[i], 8); + } + + if (nv50->dirty & NV50_NEW_TEXTURE) { + int i; + + BEGIN_RING(tesla, 0x0f00, 1); + OUT_RING ((NV50_CB_TIC << 0) | (0 << 8)); + BEGIN_RING(tesla, 0x40000f04, nv50->miptree_nr * 8); + for (i = 0; i < nv50->sampler_nr; i++) { + struct nv50_miptree *mt = nv50->miptree[i]; + + OUT_RING (0x2a712488); + OUT_RELOCl(mt->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW); + OUT_RING (0xd0c05000); + OUT_RING (0x00300000); + OUT_RING (mt->base.width[0]); + OUT_RING ((mt->base.depth[0] << 16) | + mt->base.height[0]); + OUT_RING (0x03000000); + OUT_RELOCh(mt->buffer, 0, + NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH); + } + } + if (nv50->dirty & NV50_NEW_ARRAYS) nv50_vbo_validate(nv50); -- cgit v1.2.3 From fa5cd63f96d2b69ded48d40b9cb7e57c147f7332 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 15 Jun 2008 15:49:25 +1000 Subject: nv50: simplify interp crap a bit... hopefully there wasn't a good reason I went the route I did.. can't recall.. --- src/gallium/drivers/nv50/nv50_program.c | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 5eec68e5e19..5800a347bf6 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1263,7 +1263,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) NOUVEAU_ERR("%d attrib regs\n", pc->attr_nr); if (pc->attr_nr) { - struct nv50_reg *iv = NULL, *tmp = NULL; + struct nv50_reg *iv = NULL; int aid = 0; pc->attr = calloc(pc->attr_nr * 4, sizeof(struct nv50_reg)); @@ -1272,6 +1272,8 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->p->type == PIPE_SHADER_FRAGMENT) { iv = alloc_temp(pc, NULL); + emit_interp(pc, iv, iv, iv, FALSE); + emit_flop(pc, 0, iv, iv); aid++; } @@ -1297,14 +1299,10 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->p->type != PIPE_SHADER_FRAGMENT) continue; - emit_interp(pc, iv, iv, iv, FALSE); - tmp = alloc_temp(pc, NULL); - emit_flop(pc, 0, tmp, iv); - emit_interp(pc, &a[0], &a[0], tmp, TRUE); - emit_interp(pc, &a[1], &a[1], tmp, TRUE); - emit_interp(pc, &a[2], &a[2], tmp, TRUE); - emit_interp(pc, &a[3], &a[3], tmp, TRUE); - free_temp(pc, tmp); + emit_interp(pc, &a[0], &a[0], iv, TRUE); + emit_interp(pc, &a[1], &a[1], iv, TRUE); + emit_interp(pc, &a[2], &a[2], iv, TRUE); + emit_interp(pc, &a[3], &a[3], iv, TRUE); } if (iv) -- cgit v1.2.3 From da66b8a2f4c3c052ad71b2b6d5a845c2fd267c6e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 15 Jun 2008 15:53:22 +1000 Subject: nv50: disable inline IMMD for now, IMMD+pred == BANG! fixes progs/fp/lit.txt --- src/gallium/drivers/nv50/nv50_program.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 5800a347bf6..845fc611905 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -31,10 +31,6 @@ * FUCK! watch dst==src vectors, can overwrite components that are needed. * ie. SUB R0, R0.yzxw, R0 * - * MOV dst, -src - * "delta" tmp, -src (0xa0000204,0xe4004780 - delta r0, -r0) - * mov dst, tmp - * * Things to check with renouveau: * FP attr/result assignment - how? * attrib @@ -349,7 +345,7 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) set_dst(pc, dst, e); - if (dst->type != P_RESULT && src->type == P_IMMD) { + if (0 && dst->type != P_RESULT && src->type == P_IMMD) { set_immd(pc, src, e); /*XXX: 32-bit, but steals part of "half" reg space - need to * catch and handle this case if/when we do half-regs -- cgit v1.2.3 From fea0b1651677444fc6c135e1a4b8ab6463a9fdf9 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 12:55:53 +1000 Subject: nv50: a couple more bits'n'pieces --- src/gallium/drivers/nv50/nv50_program.c | 10 +++++++++- src/gallium/drivers/nv50/nv50_program.h | 1 + 2 files changed, 10 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 845fc611905..2fe60ad51ff 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -93,6 +93,11 @@ alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg) { int i; + if (reg->type == P_RESULT) { + if (pc->p->cfg.high_result < (reg->hw + 1)) + pc->p->cfg.high_result = reg->hw + 1; + } + if (reg->type != P_TEMP) return; @@ -1558,6 +1563,8 @@ nv50_vertprog_validate(struct nv50_context *nv50) so_method(so, tesla, 0x1650, 2); so_data (so, p->cfg.vp.attr[0]); so_data (so, p->cfg.vp.attr[1]); + so_method(so, tesla, 0x16b8, 1); + so_data (so, p->cfg.high_result); so_method(so, tesla, 0x16ac, 2); so_data (so, 8); so_data (so, p->cfg.high_temp); @@ -1594,9 +1601,10 @@ nv50_fragprog_validate(struct nv50_context *nv50) so_data (so, 0x00000004); so_data (so, 0x00000000); so_data (so, 0x00000000); - so_method(so, tesla, 0x16bc, 2); /*XXX: fixme */ + so_method(so, tesla, 0x16bc, 3); /*XXX: fixme */ so_data (so, 0x03020100); so_data (so, 0x07060504); + so_data (so, 0x0b0a0908); so_method(so, tesla, 0x1988, 2); so_data (so, 0x08040404); /* p: 0x0f000401 */ so_data (so, p->cfg.high_temp); diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h index d143ae97979..d643e8db218 100644 --- a/src/gallium/drivers/nv50/nv50_program.h +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -35,6 +35,7 @@ struct nv50_program { struct { unsigned high_temp; + unsigned high_result; struct { unsigned attr[2]; } vp; -- cgit v1.2.3 From cae38d0fcc6c936d3a4dc25ca2dbef3d106d05a5 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 16:29:40 +1000 Subject: nv50: abuse constbuf upload for program upload --- src/gallium/drivers/nv50/nv50_context.h | 1 + src/gallium/drivers/nv50/nv50_program.c | 50 +++++++++++++++++++++++++-------- 2 files changed, 39 insertions(+), 12 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 5214ce1a69c..cbb473410a5 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -30,6 +30,7 @@ #define NV50_CB_PGP 3 #define NV50_CB_TIC 4 #define NV50_CB_TSC 5 +#define NV50_CB_PUPLOAD 6 #define NV50_NEW_BLEND (1 << 0) #define NV50_NEW_ZSA (1 << 1) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 2fe60ad51ff..2cf6275730c 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1494,8 +1494,10 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) { struct pipe_winsys *ws = nv50->pipe.winsys; struct nv50_program_exec *e; + struct nouveau_stateobj *so; + const unsigned flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_WR; + unsigned start, count, *up, *ptr; boolean upload = FALSE; - unsigned *map; if (!p->buffer) { p->buffer = ws->buffer_create(ws, 0x100, 0, p->exec_size * 4); @@ -1522,20 +1524,44 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) if (!upload) return FALSE; - map = ws->buffer_map(ws, p->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); + up = ptr = MALLOC(p->exec_size * 4); for (e = p->exec_head; e; e = e->next) { -#ifdef NV50_PROGRAM_DUMP - NOUVEAU_ERR("0x%08x\n", e->inst[0]); -#endif - *(map++) = e->inst[0]; - if (is_long(e)) { -#ifdef NV50_PROGRAM_DUMP - NOUVEAU_ERR("0x%08x\n", e->inst[1]); -#endif - *(map++) = e->inst[1]; + *(ptr++) = e->inst[0]; + if (is_long(e)) + *(ptr++) = e->inst[1]; + } + + so = so_new(3,2); + so_method(so, nv50->screen->tesla, 0x1280, 3); + so_reloc (so, p->buffer, 0, flags | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, p->buffer, 0, flags | NOUVEAU_BO_LOW, 0, 0); + so_data (so, (NV50_CB_PUPLOAD << 16) | 0x0800); //(p->exec_size * 4)); + + start = 0; count = p->exec_size; + while (count) { + struct nouveau_winsys *nvws = nv50->screen->nvws; + unsigned nr; + + so_emit(nvws, so); + + nr = MIN2(count, 2047); + nr = MIN2(nvws->channel->pushbuf->remaining, nr); + if (nvws->channel->pushbuf->remaining < (nr + 3)) { + FIRE_RING(NULL); + continue; } + + BEGIN_RING(tesla, 0x0f00, 1); + OUT_RING ((start << 8) | NV50_CB_PUPLOAD); + BEGIN_RING(tesla, 0x40000f04, nr); + OUT_RINGp (up + start, nr); + + start += nr; + count -= nr; } - ws->buffer_unmap(ws, p->buffer); + + FREE(up); + so_ref(NULL, &so); } void -- cgit v1.2.3 From bcbe6baac37915563bc120ad558cd930bc1ddec1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 18:03:29 +1000 Subject: nv50: hacks for stuff I don't really get yet --- src/gallium/drivers/nv50/nv50_program.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 2cf6275730c..fa5e24d3e90 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1592,7 +1592,7 @@ nv50_vertprog_validate(struct nv50_context *nv50) so_method(so, tesla, 0x16b8, 1); so_data (so, p->cfg.high_result); so_method(so, tesla, 0x16ac, 2); - so_data (so, 8); + so_data (so, p->cfg.high_result); //8); so_data (so, p->cfg.high_temp); so_method(so, tesla, 0x140c, 1); so_data (so, 0); /* program start offset */ @@ -1632,7 +1632,7 @@ nv50_fragprog_validate(struct nv50_context *nv50) so_data (so, 0x07060504); so_data (so, 0x0b0a0908); so_method(so, tesla, 0x1988, 2); - so_data (so, 0x08040404); /* p: 0x0f000401 */ + so_data (so, 0x08080408); //0x08040404); /* p: 0x0f000401 */ so_data (so, p->cfg.high_temp); so_method(so, tesla, 0x1414, 1); so_data (so, 0); /* program start offset */ -- cgit v1.2.3 From 431504b99cd55948522e86a249e656e78598ddbd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 18:56:39 +1000 Subject: nv50: hack of a TEX opcode --- src/gallium/drivers/nv50/nv50_program.c | 19 +++++++++++++++++++ src/gallium/drivers/nv50/nv50_screen.c | 5 +++++ 2 files changed, 24 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index fa5e24d3e90..21945410b17 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1134,6 +1134,25 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) } break; case TGSI_OPCODE_TEX: + { + struct nv50_reg *t0, *t1; + struct nv50_program_exec *e; + + t0 = alloc_temp(pc, NULL); + t0 = alloc_temp(pc, NULL); + t1 = alloc_temp(pc, NULL); + emit_mov(pc, t0, src[0][0]); + emit_mov(pc, t1, src[0][1]); + + e = exec(pc); + e->inst[0] = 0xf0400000; + set_long(pc, e); + e->inst[1] |= 0x0000c004; + set_dst(pc, t0, e); + emit(pc, e); + free_temp(pc, t0); + free_temp(pc, t1); + } break; case TGSI_OPCODE_XPD: temp = alloc_temp(pc, NULL); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index b9cecb77f1f..8affb0f073f 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -289,6 +289,11 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, fui(0.0)); so_data (so, fui(1.0)); + so_method(so, screen->tesla, 0x1234, 1); + so_data (so, 1); + so_method(so, screen->tesla, 0x1458, 1); + so_data (so, 1); + so_emit(nvws, so); so_ref(NULL, &so); nvws->push_flush(nvws, 0, NULL); -- cgit v1.2.3 From 5d3070149267251bafc1ff982b77e7f422554f50 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 22:02:02 +1000 Subject: nv50: use stateobjs for sampler/image_control uploads --- src/gallium/drivers/nouveau/nouveau_stateobj.h | 18 +++++++++++ src/gallium/drivers/nv50/nv50_state_validate.c | 45 +++++++++++++++----------- 2 files changed, 44 insertions(+), 19 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index d501b76b51e..78b27380702 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -2,6 +2,7 @@ #define __NOUVEAU_STATEOBJ_H__ #include "pipe/p_util.h" +#include "pipe/p_debug.h" struct nouveau_stateobj_reloc { struct pipe_buffer *bo; @@ -67,6 +68,14 @@ so_data(struct nouveau_stateobj *so, unsigned data) so->cur_packet += 4; } +static INLINE void +so_datap(struct nouveau_stateobj *so, unsigned *data, unsigned size) +{ + so->cur_packet += (4 * size); + while (size--) + (*so->cur++) = (*data++); +} + static INLINE void so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr, unsigned mthd, unsigned size) @@ -91,6 +100,15 @@ so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo, so_data(so, data); } +static INLINE void +so_dump(struct nouveau_stateobj *so) +{ + unsigned i, nr = so->cur - so->push; + + for (i = 0; i < nr; i++) + debug_printf("+0x%04x: 0x%08x\n", i, so->push[i]); +} + static INLINE void so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) { diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 63c1756bcbf..b5a9195231b 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -181,34 +181,41 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & NV50_NEW_SAMPLER) { int i; - BEGIN_RING(tesla, 0x0f00, 1); - OUT_RING ((NV50_CB_TSC << 0) | (0 << 8)); - BEGIN_RING(tesla, 0x40000f04, nv50->sampler_nr * 8); + so = so_new(nv50->sampler_nr * 8 + 3, 0); + so_method(so, tesla, 0x0f00, 1); + so_data (so, NV50_CB_TSC); + so_method(so, tesla, 0x40000f04, nv50->sampler_nr * 8); for (i = 0; i < nv50->sampler_nr; i++) - OUT_RINGp(nv50->sampler[i], 8); + so_datap (so, nv50->sampler[i], 8); + so_emit(nvws, so); + so_ref(NULL, &so); } if (nv50->dirty & NV50_NEW_TEXTURE) { int i; - BEGIN_RING(tesla, 0x0f00, 1); - OUT_RING ((NV50_CB_TIC << 0) | (0 << 8)); - BEGIN_RING(tesla, 0x40000f04, nv50->miptree_nr * 8); - for (i = 0; i < nv50->sampler_nr; i++) { + so = so_new(nv50->miptree_nr * 8 + 3, nv50->miptree_nr * 2); + so_method(so, tesla, 0x0f00, 1); + so_data (so, NV50_CB_TIC); + so_method(so, tesla, 0x40000f04, nv50->miptree_nr * 8); + for (i = 0; i < nv50->miptree_nr; i++) { struct nv50_miptree *mt = nv50->miptree[i]; - OUT_RING (0x2a712488); - OUT_RELOCl(mt->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW); - OUT_RING (0xd0c05000); - OUT_RING (0x00300000); - OUT_RING (mt->base.width[0]); - OUT_RING ((mt->base.depth[0] << 16) | - mt->base.height[0]); - OUT_RING (0x03000000); - OUT_RELOCh(mt->buffer, 0, - NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH); + so_data (so, 0x2a712488); + so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0xd0c05000); + so_data (so, 0x00300000); + so_data (so, mt->base.width[0]); + so_data (so, (mt->base.depth[0] << 16) | + mt->base.height[0]); + so_data (so, 0x03000000); + so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_HIGH, 0, 0); } + + so_emit(nvws, so); + so_ref(NULL, &so); } if (nv50->dirty & NV50_NEW_ARRAYS) -- cgit v1.2.3 From 94999d39d43d24a702f4cb55b515906d03a57277 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 22:06:54 +1000 Subject: nv50: fix blend colour --- src/gallium/drivers/nv50/nv50_state_validate.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index b5a9195231b..9d93c04da7d 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -135,11 +135,11 @@ nv50_state_validate(struct nv50_context *nv50) if (nv50->dirty & NV50_NEW_BLEND_COLOUR) { so = so_new(5, 0); - so_method(so, tesla, NV50TCL_BLEND_COLOR(0), 8); - so_data (so, fui(nv50->blend_colour.color[3])); + so_method(so, tesla, NV50TCL_BLEND_COLOR(0), 4); so_data (so, fui(nv50->blend_colour.color[0])); so_data (so, fui(nv50->blend_colour.color[1])); so_data (so, fui(nv50->blend_colour.color[2])); + so_data (so, fui(nv50->blend_colour.color[3])); so_emit(nvws, so); so_ref(NULL, &so); } -- cgit v1.2.3 From bb9efb5534a652878161e28bd73039eff5b11014 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 22:24:16 +1000 Subject: nv50: separate state validation and upload, similar to nv40 --- src/gallium/drivers/nv50/nv50_context.h | 21 +++++++ src/gallium/drivers/nv50/nv50_program.c | 6 +- src/gallium/drivers/nv50/nv50_state_validate.c | 80 ++++++++++++++++++++------ src/gallium/drivers/nv50/nv50_vbo.c | 6 +- 4 files changed, 86 insertions(+), 27 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index cbb473410a5..3e0e523ee3e 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -74,6 +74,25 @@ nv50_miptree(struct pipe_texture *pt) return (struct nv50_miptree *)pt; } +struct nv50_state { + unsigned dirty; + + struct nouveau_stateobj *fb; + struct nouveau_stateobj *blend; + struct nouveau_stateobj *blend_colour; + struct nouveau_stateobj *zsa; + struct nouveau_stateobj *rast; + struct nouveau_stateobj *stipple; + struct nouveau_stateobj *scissor; + struct nouveau_stateobj *viewport; + struct nouveau_stateobj *tsc_upload; + struct nouveau_stateobj *tic_upload; + struct nouveau_stateobj *vertprog; + struct nouveau_stateobj *fragprog; + struct nouveau_stateobj *vtxfmt; + struct nouveau_stateobj *vtxbuf; +}; + struct nv50_context { struct pipe_context pipe; @@ -82,6 +101,8 @@ struct nv50_context { struct draw_context *draw; + struct nv50_state state; + unsigned dirty; struct nv50_blend_stateobj *blend; struct nv50_zsa_stateobj *zsa; diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 21945410b17..7bb2f454bb9 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1615,8 +1615,7 @@ nv50_vertprog_validate(struct nv50_context *nv50) so_data (so, p->cfg.high_temp); so_method(so, tesla, 0x140c, 1); so_data (so, 0); /* program start offset */ - so_emit(nv50->screen->nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.vertprog); } void @@ -1655,8 +1654,7 @@ nv50_fragprog_validate(struct nv50_context *nv50) so_data (so, p->cfg.high_temp); so_method(so, tesla, 0x1414, 1); so_data (so, 0); /* program start offset */ - so_emit(nv50->screen->nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.fragprog); } void diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 9d93c04da7d..566f95f682c 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -103,14 +103,59 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_data (so, 0); so_data (so, h); - so_emit(nv50->screen->nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.fb); +} + +static void +nv50_state_emit(struct nv50_context *nv50) +{ + struct nv50_screen *screen = nv50->screen; + struct nouveau_winsys *nvws = screen->nvws; + + if (nv50->pctx_id != screen->cur_pctx) { + nv50->state.dirty |= 0xffffffff; + screen->cur_pctx = nv50->pctx_id; + } + + if (nv50->state.dirty & NV50_NEW_FRAMEBUFFER) + so_emit(nvws, nv50->state.fb); + if (nv50->state.dirty & NV50_NEW_BLEND) + so_emit(nvws, nv50->state.blend); + if (nv50->state.dirty & NV50_NEW_ZSA) + so_emit(nvws, nv50->state.zsa); + if (nv50->state.dirty & NV50_NEW_VERTPROG) + so_emit(nvws, nv50->state.vertprog); + if (nv50->state.dirty & NV50_NEW_FRAGPROG) + so_emit(nvws, nv50->state.fragprog); + if (nv50->state.dirty & NV50_NEW_RASTERIZER) + so_emit(nvws, nv50->state.rast); + if (nv50->state.dirty & NV50_NEW_BLEND_COLOUR) + so_emit(nvws, nv50->state.blend_colour); + if (nv50->state.dirty & NV50_NEW_STIPPLE) + so_emit(nvws, nv50->state.stipple); + if (nv50->state.dirty & NV50_NEW_SCISSOR) + so_emit(nvws, nv50->state.scissor); + if (nv50->state.dirty & NV50_NEW_VIEWPORT) + so_emit(nvws, nv50->state.viewport); + if (nv50->state.dirty & NV50_NEW_SAMPLER) + so_emit(nvws, nv50->state.tsc_upload); + if (nv50->state.dirty & NV50_NEW_TEXTURE) + so_emit(nvws, nv50->state.tic_upload); + if (nv50->state.dirty & NV50_NEW_ARRAYS) { + so_emit(nvws, nv50->state.vtxfmt); + so_emit(nvws, nv50->state.vtxbuf); + } + nv50->state.dirty = 0; + + so_emit_reloc_markers(nvws, nv50->state.fb); + so_emit_reloc_markers(nvws, nv50->state.vertprog); + so_emit_reloc_markers(nvws, nv50->state.fragprog); + so_emit_reloc_markers(nvws, nv50->state.vtxbuf); } boolean nv50_state_validate(struct nv50_context *nv50) { - struct nouveau_winsys *nvws = nv50->screen->nvws; struct nouveau_grobj *tesla = nv50->screen->tesla; struct nouveau_stateobj *so; unsigned i; @@ -119,10 +164,10 @@ nv50_state_validate(struct nv50_context *nv50) nv50_state_validate_fb(nv50); if (nv50->dirty & NV50_NEW_BLEND) - so_emit(nvws, nv50->blend->so); + so_ref(nv50->blend->so, &nv50->state.blend); if (nv50->dirty & NV50_NEW_ZSA) - so_emit(nvws, nv50->zsa->so); + so_ref(nv50->zsa->so, &nv50->state.zsa); if (nv50->dirty & (NV50_NEW_VERTPROG | NV50_NEW_VERTPROG_CB)) nv50_vertprog_validate(nv50); @@ -131,7 +176,7 @@ nv50_state_validate(struct nv50_context *nv50) nv50_fragprog_validate(nv50); if (nv50->dirty & NV50_NEW_RASTERIZER) - so_emit(nvws, nv50->rasterizer->so); + so_ref(nv50->rasterizer->so, &nv50->state.rast); if (nv50->dirty & NV50_NEW_BLEND_COLOUR) { so = so_new(5, 0); @@ -140,8 +185,7 @@ nv50_state_validate(struct nv50_context *nv50) so_data (so, fui(nv50->blend_colour.color[1])); so_data (so, fui(nv50->blend_colour.color[2])); so_data (so, fui(nv50->blend_colour.color[3])); - so_emit(nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.blend_colour); } if (nv50->dirty & NV50_NEW_STIPPLE) { @@ -149,8 +193,7 @@ nv50_state_validate(struct nv50_context *nv50) so_method(so, tesla, NV50TCL_POLYGON_STIPPLE_PATTERN(0), 32); for (i = 0; i < 32; i++) so_data(so, nv50->stipple.stipple[i]); - so_emit(nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.stipple); } if (nv50->dirty & NV50_NEW_SCISSOR) { @@ -160,8 +203,7 @@ nv50_state_validate(struct nv50_context *nv50) nv50->scissor.minx); so_data (so, (nv50->scissor.maxy << 16) | nv50->scissor.miny); - so_emit(nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.scissor); } if (nv50->dirty & NV50_NEW_VIEWPORT) { @@ -174,8 +216,7 @@ nv50_state_validate(struct nv50_context *nv50) so_data (so, fui(nv50->viewport.scale[0])); so_data (so, fui(-nv50->viewport.scale[1])); so_data (so, fui(nv50->viewport.scale[2])); - so_emit(nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.viewport); } if (nv50->dirty & NV50_NEW_SAMPLER) { @@ -187,8 +228,7 @@ nv50_state_validate(struct nv50_context *nv50) so_method(so, tesla, 0x40000f04, nv50->sampler_nr * 8); for (i = 0; i < nv50->sampler_nr; i++) so_datap (so, nv50->sampler[i], 8); - so_emit(nvws, so); - so_ref(NULL, &so); + so_ref(so, &nv50->state.tsc_upload); } if (nv50->dirty & NV50_NEW_TEXTURE) { @@ -213,15 +253,17 @@ nv50_state_validate(struct nv50_context *nv50) so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0); } - - so_emit(nvws, so); - so_ref(NULL, &so); + + so_ref(so, &nv50->state.tic_upload); } if (nv50->dirty & NV50_NEW_ARRAYS) nv50_vbo_validate(nv50); + nv50->state.dirty |= nv50->dirty; nv50->dirty = 0; + nv50_state_emit(nv50); + return TRUE; } diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 7a1c2f24ef9..3f0b66ae365 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -208,9 +208,7 @@ nv50_vbo_validate(struct nv50_context *nv50) NOUVEAU_BO_LOW, 0, 0); } - so_emit(nv50->screen->nvws, vtxfmt); - so_ref (NULL, &vtxfmt); - so_emit(nv50->screen->nvws, vtxbuf); - so_ref (NULL, &vtxbuf); + so_ref (vtxfmt, &nv50->state.vtxfmt); + so_ref (vtxbuf, &nv50->state.vtxbuf); } -- cgit v1.2.3 From 035a04d9c11e0e90e2dbcdba25f39c3156a64115 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 16 Jun 2008 22:34:50 +1000 Subject: nv50: get tri-scissor-tri working --- src/gallium/drivers/nv50/nv50_context.h | 1 + src/gallium/drivers/nv50/nv50_state_validate.c | 25 +++++++++++++++++++------ 2 files changed, 20 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 3e0e523ee3e..4ea01009fb6 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -84,6 +84,7 @@ struct nv50_state { struct nouveau_stateobj *rast; struct nouveau_stateobj *stipple; struct nouveau_stateobj *scissor; + unsigned scissor_enabled; struct nouveau_stateobj *viewport; struct nouveau_stateobj *tsc_upload; struct nouveau_stateobj *tic_upload; diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 566f95f682c..d99cdc73cac 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -196,15 +196,28 @@ nv50_state_validate(struct nv50_context *nv50) so_ref(so, &nv50->state.stipple); } - if (nv50->dirty & NV50_NEW_SCISSOR) { + if (nv50->dirty & (NV50_NEW_SCISSOR | NV50_NEW_RASTERIZER)) { + struct pipe_rasterizer_state *rast = &nv50->rasterizer->pipe; + struct pipe_scissor_state *s = &nv50->scissor; + + if (nv50->state.scissor && + (rast->scissor == 0 && nv50->state.scissor_enabled == 0)) + goto scissor_uptodate; + nv50->state.scissor_enabled = rast->scissor; + so = so_new(3, 0); - so_method(so, tesla, NV50TCL_SCISSOR_HORIZ, 2); - so_data (so, (nv50->scissor.maxx << 16) | - nv50->scissor.minx); - so_data (so, (nv50->scissor.maxy << 16) | - nv50->scissor.miny); + so_method(so, tesla, 0xff4, 2); //NV50TCL_SCISSOR_HORIZ, 2); + if (nv50->state.scissor_enabled) { + so_data(so, ((s->maxx - s->minx) << 16) | s->minx); + so_data(so, ((s->maxy - s->miny) << 16) | s->miny); + } else { + so_data(so, (8192 << 16)); + so_data(so, (8192 << 16)); + } so_ref(so, &nv50->state.scissor); + nv50->state.dirty |= NV50_NEW_SCISSOR; } +scissor_uptodate: if (nv50->dirty & NV50_NEW_VIEWPORT) { so = so_new(8, 0); -- cgit v1.2.3 From 598b2a51052913521e3059cdef7cf0c66a5adb90 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 17 Jun 2008 01:36:36 +1000 Subject: nv50: make TEX a halfie --- src/gallium/drivers/nv50/nv50_program.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 7bb2f454bb9..5ef09542397 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1145,9 +1145,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_mov(pc, t1, src[0][1]); e = exec(pc); - e->inst[0] = 0xf0400000; - set_long(pc, e); - e->inst[1] |= 0x0000c004; + e->inst[0] = 0xf0400100; set_dst(pc, t0, e); emit(pc, e); free_temp(pc, t0); -- cgit v1.2.3 From fd7412a7f1beab8b81ce307b1054331eee102e8b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 17 Jun 2008 01:51:13 +1000 Subject: nv50: some people are just born stupid.. really.. --- src/gallium/drivers/nv50/nv50_program.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 5ef09542397..ba60b8c5338 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1135,21 +1135,33 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) break; case TGSI_OPCODE_TEX: { - struct nv50_reg *t0, *t1; + struct nv50_reg *t0, *t1, *t2, *t3; struct nv50_program_exec *e; t0 = alloc_temp(pc, NULL); t0 = alloc_temp(pc, NULL); t1 = alloc_temp(pc, NULL); + t2 = alloc_temp(pc, NULL); + t3 = alloc_temp(pc, NULL); emit_mov(pc, t0, src[0][0]); emit_mov(pc, t1, src[0][1]); e = exec(pc); - e->inst[0] = 0xf0400100; + e->inst[0] = 0xf0400000; + set_long(pc, e); + e->inst[1] |= 0x0000c004; set_dst(pc, t0, e); emit(pc, e); + + if (mask & (1 << 0)) emit_mov(pc, dst[0], t0); + if (mask & (1 << 1)) emit_mov(pc, dst[1], t1); + if (mask & (1 << 2)) emit_mov(pc, dst[2], t2); + if (mask & (1 << 3)) emit_mov(pc, dst[3], t3); + free_temp(pc, t0); free_temp(pc, t1); + free_temp(pc, t2); + free_temp(pc, t3); } break; case TGSI_OPCODE_XPD: -- cgit v1.2.3 From 65ad8176ca91b5ed2a01b1b3ee145cfdce369419 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 17 Jun 2008 23:43:38 +1000 Subject: nv50: move surface_map/unmap into nv50_surface.c --- src/gallium/drivers/nv50/nv50_screen.c | 26 +------------------------ src/gallium/drivers/nv50/nv50_screen.h | 2 ++ src/gallium/drivers/nv50/nv50_surface.c | 34 +++++++++++++++++++++++++++++++-- 3 files changed, 35 insertions(+), 27 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 8affb0f073f..18f22c59609 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -118,28 +118,6 @@ nv50_screen_get_paramf(struct pipe_screen *pscreen, int param) } } -static void * -nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, - unsigned flags ) -{ - struct pipe_winsys *ws = screen->winsys; - void *map; - - map = ws->buffer_map(ws, surface->buffer, flags); - if (!map) - return NULL; - - return map + surface->offset; -} - -static void -nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) -{ - struct pipe_winsys *ws = screen->winsys; - - ws->buffer_unmap(ws, surface->buffer); -} - static void nv50_screen_destroy(struct pipe_screen *pscreen) { @@ -309,10 +287,8 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) screen->pipe.is_format_supported = nv50_screen_is_format_supported; - screen->pipe.surface_map = nv50_surface_map; - screen->pipe.surface_unmap = nv50_surface_unmap; - nv50_screen_init_miptree_functions(&screen->pipe); + nv50_surface_init_screen_functions(&screen->pipe); return &screen->pipe; } diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index 5acb5003ba4..08d1f450891 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -26,4 +26,6 @@ nv50_screen(struct pipe_screen *screen) return (struct nv50_screen *)screen; } +void nv50_surface_init_screen_functions(struct pipe_screen *); + #endif diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index 4e294198b07..acd7b501ef4 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -57,9 +57,39 @@ nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, nvws->surface_fill(nvws, dest, destx, desty, width, height, value); } +static void * +nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + void nv50_init_surface_functions(struct nv50_context *nv50) { - nv50->pipe.surface_copy = nv50_surface_copy; - nv50->pipe.surface_fill = nv50_surface_fill; + nv50->pipe.surface_copy = nv50_surface_copy; + nv50->pipe.surface_fill = nv50_surface_fill; } + +void +nv50_surface_init_screen_functions(struct pipe_screen *pscreen) +{ + pscreen->surface_map = nv50_surface_map; + pscreen->surface_unmap = nv50_surface_unmap; +} + -- cgit v1.2.3 From 3b88c3f4112a8bac52b7f7e613b1c2df8a14b752 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 17 Jun 2008 23:52:41 +1000 Subject: nv50: R32_FLOAT vbo format --- src/gallium/drivers/nv50/nv50_vbo.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 3f0b66ae365..c16bfe3a937 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -185,6 +185,9 @@ nv50_vbo_validate(struct nv50_context *nv50) case PIPE_FORMAT_R32G32_FLOAT: so_data(vtxfmt, 0x7e200000 | i); break; + case PIPE_FORMAT_R32_FLOAT: + so_data(vtxfmt, 0x7e900000 | i); + break; case PIPE_FORMAT_R8G8B8A8_UNORM: so_data(vtxfmt, 0x24500000 | i); break; -- cgit v1.2.3 From 714cb4a86c1f503334b37ca6c24272fa1bdf7899 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 17 Jun 2008 23:55:23 +1000 Subject: nv50: make sure static buffers (constbuf, tex control etc) get on reloc list --- src/gallium/drivers/nv50/nv50_screen.c | 2 +- src/gallium/drivers/nv50/nv50_screen.h | 2 ++ src/gallium/drivers/nv50/nv50_state_validate.c | 1 + 3 files changed, 4 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 18f22c59609..5b4c5f96ac3 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -273,7 +273,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 1); so_emit(nvws, so); - so_ref(NULL, &so); + so_ref(so, &screen->static_init); nvws->push_flush(nvws, 0, NULL); screen->pipe.winsys = ws; diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index 08d1f450891..400ddcef06d 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -18,6 +18,8 @@ struct nv50_screen { struct pipe_buffer *tic; struct pipe_buffer *tsc; + + struct nouveau_stateobj *static_init; }; static INLINE struct nv50_screen * diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index d99cdc73cac..8229bce89ee 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -151,6 +151,7 @@ nv50_state_emit(struct nv50_context *nv50) so_emit_reloc_markers(nvws, nv50->state.vertprog); so_emit_reloc_markers(nvws, nv50->state.fragprog); so_emit_reloc_markers(nvws, nv50->state.vtxbuf); + so_emit_reloc_markers(nvws, nv50->screen->static_init); } boolean -- cgit v1.2.3 From 5a3ea9ee59ac586955f7784eb25e7fd70d0c8882 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 18 Jun 2008 13:23:00 +1000 Subject: nv50: simplify emit_interp a bit --- src/gallium/drivers/nv50/nv50_program.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index ba60b8c5338..3248f2aa3c3 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -302,18 +302,19 @@ set_immd(struct nv50_pc *pc, struct nv50_reg *imm, struct nv50_program_exec *e) static void emit_interp(struct nv50_pc *pc, struct nv50_reg *dst, - struct nv50_reg *src, struct nv50_reg *iv, boolean noperspective) + struct nv50_reg *src, struct nv50_reg *iv) { struct nv50_program_exec *e = exec(pc); e->inst[0] |= 0x80000000; set_dst(pc, dst, e); - alloc_reg(pc, iv); - e->inst[0] |= (iv->hw << 9); alloc_reg(pc, src); e->inst[0] |= (src->hw << 16); - if (noperspective) + if (iv) { e->inst[0] |= (1 << 25); + alloc_reg(pc, iv); + e->inst[0] |= (iv->hw << 9); + } emit(pc, e); } @@ -1147,7 +1148,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) emit_mov(pc, t1, src[0][1]); e = exec(pc); - e->inst[0] = 0xf0400000; + e->inst[0] = 0xf6400000; set_long(pc, e); e->inst[1] |= 0x0000c004; set_dst(pc, t0, e); @@ -1302,7 +1303,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->p->type == PIPE_SHADER_FRAGMENT) { iv = alloc_temp(pc, NULL); - emit_interp(pc, iv, iv, iv, FALSE); + emit_interp(pc, iv, iv, NULL); emit_flop(pc, 0, iv, iv); aid++; } @@ -1329,10 +1330,10 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->p->type != PIPE_SHADER_FRAGMENT) continue; - emit_interp(pc, &a[0], &a[0], iv, TRUE); - emit_interp(pc, &a[1], &a[1], iv, TRUE); - emit_interp(pc, &a[2], &a[2], iv, TRUE); - emit_interp(pc, &a[3], &a[3], iv, TRUE); + emit_interp(pc, &a[0], &a[0], iv); + emit_interp(pc, &a[1], &a[1], iv); + emit_interp(pc, &a[2], &a[2], iv); + emit_interp(pc, &a[3], &a[3], iv); } if (iv) -- cgit v1.2.3 From e90130257527aff43f807ae16d802c5515d29e8e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 23 Jun 2008 23:42:53 +1000 Subject: nv50: remove some debug --- src/gallium/drivers/nv50/nv50_program.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 3248f2aa3c3..148de3b7eaf 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -883,8 +883,6 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok) unsigned mask, sat; int i, c; - NOUVEAU_ERR("insn %p\n", tok); - mask = inst->FullDstRegisters[0].DstRegister.WriteMask; sat = inst->Instruction.Saturate == TGSI_SAT_ZERO_ONE; @@ -1277,7 +1275,6 @@ nv50_program_tx_prep(struct nv50_pc *pc) } } - NOUVEAU_ERR("%d temps\n", pc->temp_nr); if (pc->temp_nr) { pc->temp = calloc(pc->temp_nr * 4, sizeof(struct nv50_reg)); if (!pc->temp) @@ -1292,7 +1289,6 @@ nv50_program_tx_prep(struct nv50_pc *pc) } } - NOUVEAU_ERR("%d attrib regs\n", pc->attr_nr); if (pc->attr_nr) { struct nv50_reg *iv = NULL; int aid = 0; @@ -1340,7 +1336,6 @@ nv50_program_tx_prep(struct nv50_pc *pc) free_temp(pc, iv); } - NOUVEAU_ERR("%d result regs\n", pc->result_nr); if (pc->result_nr) { int rid = 0; @@ -1362,7 +1357,6 @@ nv50_program_tx_prep(struct nv50_pc *pc) } } - NOUVEAU_ERR("%d param regs\n", pc->param_nr); if (pc->param_nr) { int rid = 0; -- cgit v1.2.3 From 47771bcd2fb5bcfecfa076c19360436351c21c95 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 23 Jun 2008 23:43:34 +1000 Subject: nv50: maintain pipe surface status field --- src/gallium/drivers/nv50/nv50_clear.c | 1 + src/gallium/drivers/nv50/nv50_state_validate.c | 7 +++++++ 2 files changed, 8 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_clear.c b/src/gallium/drivers/nv50/nv50_clear.c index 552b92f72e2..f35087b37ef 100644 --- a/src/gallium/drivers/nv50/nv50_clear.c +++ b/src/gallium/drivers/nv50/nv50_clear.c @@ -9,4 +9,5 @@ nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); + ps->status = PIPE_SURFACE_STATUS_CLEAR; } diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 8229bce89ee..f5c734699fc 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -157,10 +157,17 @@ nv50_state_emit(struct nv50_context *nv50) boolean nv50_state_validate(struct nv50_context *nv50) { + const struct pipe_framebuffer_state *fb = &nv50->framebuffer; struct nouveau_grobj *tesla = nv50->screen->tesla; struct nouveau_stateobj *so; unsigned i; + for (i = 0; i < fb->num_cbufs; i++) + fb->cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED; + + if (fb->zsbuf) + fb->zsbuf->status = PIPE_SURFACE_STATUS_DEFINED; + if (nv50->dirty & NV50_NEW_FRAMEBUFFER) nv50_state_validate_fb(nv50); -- cgit v1.2.3 From 5a3362521de5e17e4f340fd9136af1d5e3891e23 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 03:56:57 +1000 Subject: nv50: turn on depth test/write again, not 100% but winsys handles it better --- src/gallium/drivers/nv50/nv50_miptree.c | 7 ++++--- src/gallium/drivers/nv50/nv50_state.c | 4 ++-- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index cdd98844f94..6ee09fc4dd4 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -10,7 +10,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) { struct pipe_winsys *ws = pscreen->winsys; struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree); - unsigned usage; + unsigned usage, pitch; NOUVEAU_ERR("unimplemented\n"); @@ -28,8 +28,9 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) break; } - mt->buffer = ws->buffer_create(ws, 256, usage, - pt->width[0] * pt->cpp * pt->height[0]); + pitch = ((pt->width[0] + 63) & ~63) * pt->cpp; + + mt->buffer = ws->buffer_create(ws, 256, usage, pitch * pt->height[0]); if (!mt->buffer) { FREE(mt); return NULL; diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index f36299db4d2..1b765cb5c7d 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -272,10 +272,10 @@ nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, struct nouveau_stateobj *so = so_new(64, 0); so_method(so, tesla, NV50TCL_DEPTH_WRITE_ENABLE, 1); - so_data (so, 0); //cso->depth.writemask ? 1 : 0); + so_data (so, cso->depth.writemask ? 1 : 0); if (cso->depth.enabled) { so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1); - so_data (so, 0); //1); + so_data (so, 1); so_method(so, tesla, NV50TCL_DEPTH_TEST_FUNC, 1); so_data (so, nvgl_comparison_op(cso->depth.func)); } else { -- cgit v1.2.3 From 2c2cb8646168c8709e51d7ff583a86044e3f2040 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 04:53:34 +1000 Subject: nv50: rework miptree/texture/texsurf code a bit --- src/gallium/drivers/nv50/nv50_miptree.c | 30 +++++++++++++++++++----------- 1 file changed, 19 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 6ee09fc4dd4..ec6e09aea08 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -12,8 +12,6 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree); unsigned usage, pitch; - NOUVEAU_ERR("unimplemented\n"); - mt->base = *pt; mt->base.refcount = 1; mt->base.screen = pscreen; @@ -45,9 +43,8 @@ nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt) struct pipe_winsys *ws = pscreen->winsys; struct pipe_texture *pt = *ppt; - NOUVEAU_ERR("unimplemented\n"); - *ppt = NULL; + if (--pt->refcount <= 0) { struct nv50_miptree *mt = nv50_miptree(pt); @@ -65,13 +62,9 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, struct nv50_miptree *mt = nv50_miptree(pt); struct pipe_surface *ps; - NOUVEAU_ERR("unimplemented\n"); - - ps = ws->surface_alloc(ws); - if (!ps) - return NULL; - - pipe_buffer_reference(ws, &ps->buffer, mt->buffer); + ps = CALLOC_STRUCT(pipe_surface); + ps->refcount = 1; + ps->winsys = ws; ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; @@ -80,6 +73,11 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps->nblocksy = pt->nblocksy[level]; ps->stride = ps->width * ps->block.size; ps->offset = 0; + ps->usage = flags; + ps->status = PIPE_SURFACE_STATUS_DEFINED; + + pipe_texture_reference(&ps->texture, pt); + pipe_buffer_reference(ws, &ps->buffer, mt->buffer); return ps; } @@ -88,6 +86,16 @@ static void nv50_miptree_surface_del(struct pipe_screen *pscreen, struct pipe_surface **psurface) { + struct pipe_winsys *ws = pscreen->winsys; + struct pipe_surface *surf = *psurface; + + *psurface = NULL; + + if (--surf->refcount <= 0) { + pipe_texture_reference(&surf->texture, NULL); + pipe_buffer_reference(ws, &surf->buffer, NULL); + FREE(surf); + } } void -- cgit v1.2.3 From e002ad77398fbe14a0efbd91824c3325ca09b4c1 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 04:57:27 +1000 Subject: nv50: whoops --- src/gallium/drivers/nv50/nv50_program.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 148de3b7eaf..bc45fe0bdfd 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1555,7 +1555,7 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) *(ptr++) = e->inst[1]; } - so = so_new(3,2); + so = so_new(4,2); so_method(so, nv50->screen->tesla, 0x1280, 3); so_reloc (so, p->buffer, 0, flags | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, p->buffer, 0, flags | NOUVEAU_BO_LOW, 0, 0); @@ -1604,7 +1604,7 @@ nv50_vertprog_validate(struct nv50_context *nv50) nv50_program_validate_data(nv50, p); nv50_program_validate_code(nv50, p); - so = so_new(11, 2); + so = so_new(13, 2); so_method(so, tesla, NV50TCL_VP_ADDRESS_HIGH, 2); so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); -- cgit v1.2.3 From 95d64ceb5a2b20032e757d6c1b0b5ef5e2b973e2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 05:11:46 +1000 Subject: nv50: vpt translate/scale backwards --- src/gallium/drivers/nv50/nv50_state_validate.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index f5c734699fc..eba13f8e621 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -229,11 +229,11 @@ scissor_uptodate: if (nv50->dirty & NV50_NEW_VIEWPORT) { so = so_new(8, 0); - so_method(so, tesla, NV50TCL_VIEWPORT_UNK0(0), 3); + so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3); so_data (so, fui(nv50->viewport.translate[0])); so_data (so, fui(nv50->viewport.translate[1])); so_data (so, fui(nv50->viewport.translate[2])); - so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3); + so_method(so, tesla, NV50TCL_VIEWPORT_UNK0(0), 3); so_data (so, fui(nv50->viewport.scale[0])); so_data (so, fui(-nv50->viewport.scale[1])); so_data (so, fui(nv50->viewport.scale[2])); -- cgit v1.2.3 From e05f67cbe6d852d01da3c4e0c4d52b28723f3684 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 05:17:08 +1000 Subject: nv50: maybe some scissor fixes.. --- src/gallium/drivers/nv50/nv50_state_validate.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index eba13f8e621..cf5a071a8d6 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -96,7 +96,7 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_method(so, tesla, NV50TCL_VIEWPORT_HORIZ, 2); so_data (so, w << 16); so_data (so, h << 16); - so_method(so, tesla, 0xff4, 2); + so_method(so, tesla, 0x0e04, 2); so_data (so, w << 16); so_data (so, h << 16); so_method(so, tesla, 0xdf8, 2); @@ -214,7 +214,7 @@ nv50_state_validate(struct nv50_context *nv50) nv50->state.scissor_enabled = rast->scissor; so = so_new(3, 0); - so_method(so, tesla, 0xff4, 2); //NV50TCL_SCISSOR_HORIZ, 2); + so_method(so, tesla, 0x0ff4, 2); if (nv50->state.scissor_enabled) { so_data(so, ((s->maxx - s->minx) << 16) | s->minx); so_data(so, ((s->maxy - s->miny) << 16) | s->miny); -- cgit v1.2.3 From fea9eb284248adda65afdc3833385d4b03bb25aa Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 05:58:35 +1000 Subject: nv50: don't multiply polygon offset units by 2.0 like on nv40 --- src/gallium/drivers/nv50/nv50_state.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 1b765cb5c7d..cffcbc3e7d3 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -237,7 +237,7 @@ nv50_rasterizer_state_create(struct pipe_context *pipe, so_method(so, tesla, NV50TCL_POLYGON_OFFSET_FACTOR, 1); so_data (so, fui(cso->offset_scale)); so_method(so, tesla, NV50TCL_POLYGON_OFFSET_UNITS, 1); - so_data (so, fui(cso->offset_units * 2)); + so_data (so, fui(cso->offset_units)); } rso->pipe = *cso; -- cgit v1.2.3 From e52d37d56c91e152bc149230410ed700ff1cffe2 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 06:01:30 +1000 Subject: nv50: fix line stipple --- src/gallium/drivers/nv50/nv50_state.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index cffcbc3e7d3..d72886ac596 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -157,7 +157,7 @@ nv50_rasterizer_state_create(struct pipe_context *pipe, so_method(so, tesla, NV50TCL_LINE_STIPPLE_ENABLE, 1); so_data (so, 1); so_method(so, tesla, NV50TCL_LINE_STIPPLE_PATTERN, 1); - so_data (so, (cso->line_stipple_pattern << 16) | + so_data (so, (cso->line_stipple_pattern << 8) | cso->line_stipple_factor); } else { so_method(so, tesla, NV50TCL_LINE_STIPPLE_ENABLE, 1); -- cgit v1.2.3 From 70f0f0ebdfa40de0fe03ca94294d372b9fa4642d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 25 Jun 2008 06:07:02 +1000 Subject: nv50: reverse stencil sides, header is wrong --- src/gallium/drivers/nv50/nv50_state.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index d72886ac596..665e3cc80d6 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -283,24 +283,25 @@ nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, 0); } + /*XXX: yes, I know they're backwards.. header needs fixing */ if (cso->stencil[0].enabled) { - so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 5); + so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 5); so_data (so, 1); so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); so_data (so, nvgl_comparison_op(cso->stencil[0].func)); - so_method(so, tesla, NV50TCL_STENCIL_FRONT_FUNC_REF, 3); + so_method(so, tesla, NV50TCL_STENCIL_BACK_FUNC_REF, 3); so_data (so, cso->stencil[0].ref_value); so_data (so, cso->stencil[0].write_mask); so_data (so, cso->stencil[0].value_mask); } else { - so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 1); + so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 1); so_data (so, 0); } if (cso->stencil[1].enabled) { - so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 8); + so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 8); so_data (so, 1); so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); @@ -310,7 +311,7 @@ nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe, so_data (so, cso->stencil[1].write_mask); so_data (so, cso->stencil[1].value_mask); } else { - so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 1); + so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 1); so_data (so, 0); } -- cgit v1.2.3 From bf94027fdde51aed476e9bfdd4326aa9040440b0 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 29 Jun 2008 15:59:24 +1000 Subject: nv50: fixes after rebase + commits note on the code that was just pushed. OK, seems a lot of people have been getting the idea that nouveau is dying lately - I decided to commit some of the work I've been doing lately to prove them wrong :) Progress on my side is slow due to lack of time mainly, but I'm still around. Firstly, don't even bother trying to use gallium on G8x/G9x yet, it won't work. I've deliberately left all the necessary winsys changes out of the commits for a very good reason - I don't know what we're going to need from the DRM exactly yet and don't want to be continually breaking interfaces as I discover additional requirements. On my side, I think I've gone through about 3 different DRM interface changes, and have just discovered that I may need more yet. It'd be very annoying for everyone who uses nouveau to keep things in sync. Once I've got it sorted - I'll commit a lot of cool stuff. Stay tuned. Also, don't look at the shader code.. it's horribly nasty and full of hacks, I used it as an opportunity to learn G8x GPU programs at the same time. New semi-decent code is in works, and will follow at some point. :) --- src/gallium/drivers/nv50/nv50_miptree.c | 2 +- src/gallium/drivers/nv50/nv50_program.c | 4 ++-- src/gallium/drivers/nv50/nv50_state.c | 4 ++-- src/gallium/winsys/dri/nouveau/nouveau_screen.c | 2 +- 4 files changed, 6 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index ec6e09aea08..f936a1f0e20 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -26,7 +26,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) break; } - pitch = ((pt->width[0] + 63) & ~63) * pt->cpp; + pitch = ((pt->width[0] + 63) & ~63) * pt->block.size; mt->buffer = ws->buffer_create(ws, 256, usage, pitch * pt->height[0]); if (!mt->buffer) { diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index bc45fe0bdfd..cd6967b3668 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1240,7 +1240,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) unsigned last; d = &p.FullToken.FullDeclaration; - last = d->u.DeclarationRange.Last; + last = d->DeclarationRange.Last; switch (d->Declaration.File) { case TGSI_FILE_TEMPORARY: @@ -1546,7 +1546,7 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) } if (!upload) - return FALSE; + return; up = ptr = MALLOC(p->exec_size * 4); for (e = p->exec_head; e; e = e->next) { diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 665e3cc80d6..743de089d6f 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -121,9 +121,9 @@ nv50_set_sampler_texture(struct pipe_context *pipe, unsigned nr, int i; for (i = 0; i < nr; i++) - pipe_texture_reference(&nv50->miptree[i], pt[i]); + pipe_texture_reference((void *)&nv50->miptree[i], pt[i]); for (i = nr; i < nv50->miptree_nr; i++) - pipe_texture_reference(&nv50->miptree[i], NULL); + pipe_texture_reference((void *)&nv50->miptree[i], NULL); nv50->miptree_nr = nr; nv50->dirty |= NV50_NEW_TEXTURE; diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c index b15ee7509cc..df1fe7e69b4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.c @@ -13,7 +13,7 @@ #include "nouveau_screen.h" #include "nouveau_swapbuffers.h" -#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10 +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11 #error nouveau_drm.h version does not match expected version #endif -- cgit v1.2.3 From c6ddcc10e31f26d17c1c8181013268766aca9ac6 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Sun, 29 Jun 2008 16:02:01 +1000 Subject: nouveau: oops, bump the drm patchlevel back down again --- src/gallium/winsys/dri/nouveau/nouveau_screen.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c index df1fe7e69b4..b15ee7509cc 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.c @@ -13,7 +13,7 @@ #include "nouveau_screen.h" #include "nouveau_swapbuffers.h" -#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11 +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10 #error nouveau_drm.h version does not match expected version #endif -- cgit v1.2.3 From 14d4f9e44e55e2b427579ed6788e579d70b289e7 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 16 Jun 2008 23:18:20 -0400 Subject: g3dvl: Get rid of some Valgrind errors. Get rid of some Valgrind memory leak and uninitialized var errors. --- src/gallium/state_trackers/g3dvl/vl_context.c | 24 +++++++++++++++++++++++- src/gallium/state_trackers/g3dvl/vl_surface.c | 1 + 2 files changed, 24 insertions(+), 1 deletion(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 2d1d5434955..59a1ccd152b 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -209,6 +209,8 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) context->states.mc.i_vs = pipe->create_vs_state(pipe, &vs); + free(tokens); + return 0; } @@ -346,6 +348,8 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) context->states.mc.i_fs = pipe->create_fs_state(pipe, &fs); + free(tokens); + return 0; } @@ -558,6 +562,8 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) context->states.mc.p_vs = pipe->create_vs_state(pipe, &vs); + free(tokens); + return 0; } @@ -833,6 +839,8 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) context->states.mc.p_fs = pipe->create_fs_state(pipe, &fs); + free(tokens); + return 0; } @@ -1051,6 +1059,8 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) context->states.mc.b_vs = pipe->create_vs_state(pipe, &vs); + free(tokens); + return 0; } @@ -1349,6 +1359,8 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) context->states.mc.b_fs = pipe->create_fs_state(pipe, &fs); + free(tokens); + return 0; } @@ -1502,7 +1514,7 @@ static int vlInitMC(struct VL_CONTEXT *context) /*sampler.prefilter = ;*/ /*sampler.shadow_ambient = ;*/ /*sampler.lod_bias = ;*/ - /*sampler.min_lod = ;*/ + sampler.min_lod = 0; /*sampler.max_lod = ;*/ /*sampler.border_color[i] = ;*/ /*sampler.max_anisotropy = ;*/ @@ -1692,6 +1704,8 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) context->states.csc.vertex_shader = pipe->create_vs_state(pipe, &vs); + free(tokens); + return 0; } @@ -1880,6 +1894,8 @@ static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) context->states.csc.fragment_shader = pipe->create_fs_state(pipe, &fs); + free(tokens); + return 0; } @@ -2052,6 +2068,7 @@ static int vlInitCommon(struct VL_CONTEXT *context) pipe = context->pipe; rast.flatshade = 1; + rast.flatshade_first = 0; rast.light_twoside = 0; rast.front_winding = PIPE_WINDING_CCW; rast.cull_mode = PIPE_WINDING_CW; @@ -2061,6 +2078,7 @@ static int vlInitCommon(struct VL_CONTEXT *context) rast.offset_ccw = 0; rast.scissor = 0; rast.poly_smooth = 0; + rast.poly_stipple_enable = 0; rast.point_sprite = 0; rast.point_size_per_vertex = 0; rast.multisample = 0; @@ -2074,6 +2092,7 @@ static int vlInitCommon(struct VL_CONTEXT *context) rast.bypass_vs = 0; rast.origin_lower_left = 0; rast.line_width = 1; + rast.point_smooth = 0; rast.point_size = 1; rast.offset_units = 1; rast.offset_scale = 1; @@ -2199,6 +2218,9 @@ int vlDestroyContext(struct VL_CONTEXT *context) vlDestroy(context); + context->pipe->screen->destroy(context->pipe->screen); + context->pipe->destroy(context->pipe); + free(context); return 0; diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 0e1adea4729..3f59d0f1551 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -195,6 +195,7 @@ int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface) template.depth[0] = 1; template.compressed = 0; template.cpp = 4; + template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; sfc->texture = pipe->screen->texture_create(pipe->screen, &template); -- cgit v1.2.3 From 0a6aec8c0f2173cfb95ce95d12b66f090ea0ba1f Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 18 Jun 2008 22:21:11 -0400 Subject: g3dvl: Work around SP tex cache bug, specify resource usage flags. --- src/gallium/state_trackers/g3dvl/vl_context.c | 1 - src/gallium/state_trackers/g3dvl/vl_surface.c | 14 +++++++++----- 2 files changed, 9 insertions(+), 6 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 59a1ccd152b..3b9afabbb84 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -2218,7 +2218,6 @@ int vlDestroyContext(struct VL_CONTEXT *context) vlDestroy(context); - context->pipe->screen->destroy(context->pipe->screen); context->pipe->destroy(context->pipe); free(context); diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 3f59d0f1551..6451e54953d 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -31,7 +31,7 @@ static int vlGrabBlocks 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE ); - texels = pipe_surface_map(tex_surface, 0); + texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); for (b = 0; b < 4; ++b) { @@ -131,7 +131,7 @@ static int vlGrabBlocks 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE ); - texels = pipe_surface_map(tex_surface, 0); + texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); if ((coded_block_pattern >> (b + 4)) & 1) { @@ -165,6 +165,9 @@ static int vlGrabBlocks pipe_surface_unmap(tex_surface); } + /* XXX: Texture cache is not invalidated when texture contents change */ + context->pipe->flush(context->pipe, PIPE_FLUSH_TEXTURE_CACHE, NULL); + return 0; } @@ -265,7 +268,7 @@ int vlRenderIMacroBlock ( pipe->screen, surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE ); pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); pipe->set_sampler_textures(pipe, 3, surface->context->states.mc.textures); @@ -346,7 +349,7 @@ int vlRenderPMacroBlock ( pipe->screen, surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE ); pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); @@ -434,7 +437,7 @@ int vlRenderBMacroBlock ( pipe->screen, surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE ); pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); @@ -510,6 +513,7 @@ int vlPutSurface destw, desth, PIPE_FORMAT_A8R8G8B8_UNORM, + /*XXX: SoftPipe doesn't change GPU usage to CPU like it does for textures */ PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE, 0 ); -- cgit v1.2.3 From 3933fec6bd62285506fecdc3a254306648cfefb2 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 28 Jun 2008 20:16:01 -0400 Subject: g3dvl: Support for field and frame based MC for progressive pictures. MC support for frame and field based motion prediction. Also various bug fixes, clean up. --- src/gallium/state_trackers/g3dvl/tests/.gitignore | 2 +- src/gallium/state_trackers/g3dvl/tests/Makefile | 7 +- .../state_trackers/g3dvl/tests/test_pf_rendering.c | 214 +++ src/gallium/state_trackers/g3dvl/vl_context.c | 1688 ++++++++++++++++++-- src/gallium/state_trackers/g3dvl/vl_context.h | 4 +- src/gallium/state_trackers/g3dvl/vl_data.c | 3 +- src/gallium/state_trackers/g3dvl/vl_surface.c | 404 ++--- src/gallium/state_trackers/g3dvl/vl_types.h | 12 +- src/libXvMC/surface.c | 34 +- 9 files changed, 2057 insertions(+), 311 deletions(-) create mode 100644 src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c diff --git a/src/gallium/state_trackers/g3dvl/tests/.gitignore b/src/gallium/state_trackers/g3dvl/tests/.gitignore index 939666da9ab..9b1ec4e2122 100644 --- a/src/gallium/state_trackers/g3dvl/tests/.gitignore +++ b/src/gallium/state_trackers/g3dvl/tests/.gitignore @@ -2,5 +2,5 @@ test_context test_surface test_i_rendering test_p_rendering +test_pf_rendering test_b_rendering - diff --git a/src/gallium/state_trackers/g3dvl/tests/Makefile b/src/gallium/state_trackers/g3dvl/tests/Makefile index 8f983593c3a..45cefa2e570 100644 --- a/src/gallium/state_trackers/g3dvl/tests/Makefile +++ b/src/gallium/state_trackers/g3dvl/tests/Makefile @@ -20,7 +20,7 @@ LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil .PHONY = all clean -all: test_context test_surface test_i_rendering test_p_rendering test_b_rendering +all: test_context test_surface test_i_rendering test_p_rendering test_pf_rendering test_b_rendering test_context: test_context.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} @@ -34,9 +34,12 @@ test_i_rendering: test_i_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o test_p_rendering: test_p_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} +test_pf_rendering: test_pf_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} + test_b_rendering: test_b_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} clean: - rm -rf *.o test_context test_surface test_i_rendering test_p_rendering test_b_rendering + rm -rf *.o test_context test_surface test_i_rendering test_p_rendering test_pf_rendering test_b_rendering diff --git a/src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c new file mode 100644 index 00000000000..43586fc553c --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c @@ -0,0 +1,214 @@ +#include +#include +#include +#include +#include + +static const unsigned short ycbcr16x16_420[8*8*6] = +{ + 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, + 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, + 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, + + 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, + 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, + + 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, + 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E +}; + +static const signed short ycbcr16x16_420_2[8*8*6] = +{ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0 +}; + +int main(int argc, char **argv) +{ + const unsigned int video_width = 32, video_height = 32; + const unsigned int window_width = video_width * 2, window_height = video_height * 2; + int quit = 0; + Display *display; + Window root, window; + Pixmap framebuffer; + XEvent event; + struct pipe_context *pipe; + struct VL_CONTEXT *ctx; + struct VL_SURFACE *sfc, *ref_sfc; + struct VL_MOTION_VECTOR motion_vector = + { + {0, 0}, {32, 32} + }; + + display = XOpenDisplay(NULL); + root = XDefaultRootWindow(display); + window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); + framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + + pipe = create_pipe_context(display); + vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); + vlCreateSurface(ctx, &sfc); + vlCreateSurface(ctx, &ref_sfc); + + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); + vlRenderPMacroBlock + ( + VL_FRAME_PICTURE, + VL_FIELD_FIRST, + 0, + 0, + VL_FIELD_MC, + &motion_vector, + 0x3F, + VL_DCT_FRAME_CODED, + (short*)ycbcr16x16_420_2, + ref_sfc, + sfc + ); + vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); + + puts("Press any key to continue..."); + + while (!quit) + { + XNextEvent(display, &event); + switch (event.type) + { + case Expose: + { + XCopyArea + ( + display, + framebuffer, + window, + XDefaultGC(display, XDefaultScreen(display)), + 0, + 0, + window_width, + window_height, + 0, + 0 + ); + break; + } + case KeyPress: + { + quit = 1; + break; + } + } + } + + vlDestroySurface(sfc); + vlDestroySurface(ref_sfc); + vlDestroyContext(ctx); + + XFreePixmap(display, framebuffer); + XDestroyWindow(display, window); + XCloseDisplay(display); + + return 0; +} + diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 3b9afabbb84..d2b1ad7948e 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -36,8 +36,8 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) const unsigned int semantic_names[3] = { TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ }; const unsigned int semantic_indexes[3] = {0, 1, 2}; const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; @@ -353,7 +353,7 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) return 0; } -static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) +static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; const unsigned int num_input_attribs = 3; @@ -361,15 +361,15 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) const unsigned int input_semantic_names[3] = { TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC /* Chroma texcoords */ }; const unsigned int output_semantic_names[4] = { TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ + TGSI_SEMANTIC_GENERIC /* Ref surface texcoords */ }; const unsigned int input_semantic_indexes[3] = {0, 1, 2}; const unsigned int output_semantic_indexes[4] = {0, 1, 2, 3}; @@ -430,14 +430,15 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) /* Declare constant inputs */ /* C[0] scales the normalized MB to cover 16x16 pixels, C[1] translates the macroblock into position on the surface - C[2] translates the ref surface texcoords to the ref macroblock */ + C[2] unused + C[3] translates the ref surface texcoords to the ref macroblock */ decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_CONSTANT; decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; decl.Semantic.SemanticIndex = 0; decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 2; + decl.u.DeclarationRange.Last = 3; ti += tgsi_build_full_declaration ( &decl, @@ -526,7 +527,7 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) ); } - /* add o3, t0, c2 ; Translate texcoords into position */ + /* add o3, t0, c3 ; Translate texcoords into position */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_ADD; inst.Instruction.NumDstRegs = 1; @@ -536,6 +537,264 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.mc.p_vs[0] = pipe->create_vs_state(pipe, &vs); + + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int num_input_attribs = 3; + const unsigned int num_output_attribs = 6; + const unsigned int input_semantic_names[3] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC /* Chroma texcoords */ + }; + const unsigned int output_semantic_names[6] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ + TGSI_SEMANTIC_GENERIC, /* Top field surface texcoords */ + TGSI_SEMANTIC_GENERIC, /* Bottom field surface texcoords */ + TGSI_SEMANTIC_POSITION /* Pos */ + }; + const unsigned int input_semantic_indexes[3] = {0, 1, 2}; + const unsigned int output_semantic_indexes[6] = {0, 1, 2, 3, 4, 5}; + const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_input_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = input_semantic_names[i]; + decl.Semantic.SemanticIndex = input_semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant inputs */ + /* C[0] scales the normalized MB to cover 16x16 pixels, + C[1] translates the macroblock into position on the surface + C[2] denormalizes pos components + C[3] translates the ref surface top field texcoords to the ref macroblock + C[4] translates the ref surface bottom field texcoords to the ref macroblock */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 4; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_output_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = output_semantic_names[i]; + decl.Semantic.SemanticIndex = output_semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add t1, t0, c1 ; Translate vertex into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* mov o0, t1 ; Move vertex pos to output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* + mov o1, i1 ; Move luma & chroma texcoords to output + mov o2, i2 + */ + for (i = 1; i < num_output_attribs - 1; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* add o3, t0, c3 ; Translate top field texcoords into position + add o4, t0, c4 ; Translate bottom field texcoords into position */ + for (i = 0; i < 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i + 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul o5, t1, c2 ; Denorm pos for fragment shader */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 5; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; inst.FullSrcRegisters[1].SrcRegister.Index = 2; ti += tgsi_build_full_instruction ( @@ -560,14 +819,14 @@ static int vlCreateVertexShaderPMC(struct VL_CONTEXT *context) vs.tokens = tokens; - context->states.mc.p_vs = pipe->create_vs_state(pipe, &vs); + context->states.mc.p_vs[1] = pipe->create_vs_state(pipe, &vs); free(tokens); return 0; } -static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) +static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; @@ -837,42 +1096,928 @@ static int vlCreateFragmentShaderPMC(struct VL_CONTEXT *context) fs.tokens = tokens; - context->states.mc.p_fs = pipe->create_fs_state(pipe, &fs); + context->states.mc.p_fs[0] = pipe->create_fs_state(pipe, &fs); free(tokens); return 0; } -static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) +static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) { - const unsigned int max_tokens = 100; - const unsigned int num_input_attribs = 3; - const unsigned int num_output_attribs = 5; - const unsigned int input_semantic_names[3] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC - }; - const unsigned int output_semantic_names[5] = - { - TGSI_SEMANTIC_POSITION, + const unsigned int max_tokens = 200; + const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (texcoords) + I[0] Luma texcoords + I[1] Chroma texcoords + I[2] Ref top field surface texcoords + I[3] Ref bottom field surface texcoords + I[4] Denormalized texel pos */ + for (i = 0; i < 5; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = i + 1; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant input */ + /* C[0] is a multiplier to use when concatenating differential into a single channel + C[1] is a bias to get differential back to -1,1 range + C[2] is constants 2 and 1/2 for Y%2 field selector */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 2; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare output */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare samplers */ + for (i = 0; i < 4; ++i) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* + tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + mov t1.x, t0.w ; Move high part from .w channel to .x + tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + mov t1.y, t0.w ; Move high part from .w channel to .y + tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels + mov t1.z, t0.w ; Move high part from .w channel to .z + */ + for (i = 0; i < 3; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* tex2d t1, i2, s3 ; Read texel from ref macroblock top field + tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field */ + for (i = 0; i < 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = i + 1; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* XXX: Pos values off by 0.5 for rounding? */ + /* sub t4, i4.y, c2.x ; Sub 0.5 from position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 4; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 4; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* mul t3, t4, c2.x ; Divide pos y coord by 2 (mul by 0.5) */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* floor t3, t3 ; Get rid of fractional part */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_FLOOR; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* mul t3, t3, c2.y ; Multiply by 2 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* sub t3, t4, t3 ; Subtract from y to get y % 2 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* lerp t1, t3, t1, t2 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_LERP; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 3; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[2].SrcRegister.Index = 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + fs.tokens = tokens; + + context->states.mc.p_fs[1] = pipe->create_fs_state(pipe, &fs); + + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int num_input_attribs = 3; + const unsigned int num_output_attribs = 5; + const unsigned int input_semantic_names[3] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, + TGSI_SEMANTIC_GENERIC + }; + const unsigned int output_semantic_names[5] = + { + TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC, TGSI_SEMANTIC_GENERIC }; const unsigned int input_semantic_indexes[3] = {0, 1, 2}; - const unsigned int output_semantic_indexes[5] = {0, 1, 2, 3, 4}; + const unsigned int output_semantic_indexes[5] = {0, 1, 2, 3, 4}; + const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_input_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = input_semantic_names[i]; + decl.Semantic.SemanticIndex = input_semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant inputs */ + /* C[0] scales the normalized MB to cover 16x16 pixels, + C[1] translates the macroblock into position on the surface + C[2] unused + C[3] translates the past surface texcoords to the ref macroblock + C[4] unused + C[5] translates the future surface texcoords to the ref macroblock */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 5; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_output_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = output_semantic_names[i]; + decl.Semantic.SemanticIndex = output_semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, c1 ; Translate vertex into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* + mov o1, i1 ; Move luma & chroma texcoords to output + mov o2, i2 + */ + for (i = 1; i < num_output_attribs - 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* add o3, t0, c3 ; Translate past surface texcoords into position + add o4, t0, c5 ; Repeat for future surface texcoords */ + for (i = 0; i < 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i + 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = i * 2 + 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.mc.b_vs[0] = pipe->create_vs_state(pipe, &vs); + + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int num_input_attribs = 3; + const unsigned int num_output_attribs = 8; + const unsigned int input_semantic_names[3] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC /* Chroma texcoords */ + }; + const unsigned int output_semantic_names[8] = + { + TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ + TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ + TGSI_SEMANTIC_GENERIC, /* Top field past surface texcoords */ + TGSI_SEMANTIC_GENERIC, /* Bottom field past surface texcoords */ + TGSI_SEMANTIC_GENERIC, /* Top field future surface texcoords */ + TGSI_SEMANTIC_GENERIC, /* Bottom field future surface texcoords */ + TGSI_SEMANTIC_POSITION /* Pos */ + }; + const unsigned int input_semantic_indexes[3] = {0, 1, 2}; + const unsigned int output_semantic_indexes[8] = {0, 1, 2, 3, 4, 5, 6, 7}; const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; - struct pipe_shader_state vs; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + struct tgsi_processor *processor; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + + /* Processor */ + processor = (struct tgsi_processor*)&tokens[2]; + *processor = tgsi_build_processor(proc_type, header); + + ti = 3; + + /* Declare inputs (pos, texcoords) */ + for (i = 0; i < num_input_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_INPUT; + + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = input_semantic_names[i]; + decl.Semantic.SemanticIndex = input_semantic_indexes[i]; + + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* Declare constant inputs */ + /* C[0] scales the normalized MB to cover 16x16 pixels, + C[1] translates the macroblock into position on the surface + C[2] denormalizes pos components + C[3] translates the past surface top field texcoords to the ref macroblock + C[4] translates the past surface bottom field texcoords to the ref macroblock + C[5] translates the future surface top field texcoords to the ref macroblock + C[6] translates the future surface bottom field texcoords to the ref macroblock */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 6; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare outputs (pos, texcoords) */ + for (i = 0; i < num_output_attribs; i++) + { + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = output_semantic_names[i]; + decl.Semantic.SemanticIndex = output_semantic_indexes[i]; + decl.u.DeclarationRange.First = i; + decl.u.DeclarationRange.Last = i; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add t1, t0, c1 ; Translate vertex into position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* mov o0, t1 ; Move vertex pos to output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* + mov o1, i1 ; Move luma & chroma texcoords to output + mov o2, i2 + */ + for (i = 1; i < num_output_attribs - 1; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* add o3, t0, c3 ; Translate top field past texcoords into position + add o4, t0, c4 ; Translate bottom field past texcoords into position + add o5, t0, c5 ; Translate top field past texcoords into position + add o6, t0, c6 ; Translate bottom field past texcoords into position */ + for (i = 0; i < 4; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = i + 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul o7, t1, c2 ; Denorm pos for fragment shader */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 7; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* END */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + vs.tokens = tokens; + + context->states.mc.b_vs[1] = pipe->create_vs_state(pipe, &vs); + + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 100; + const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; + + struct pipe_context *pipe; + struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; struct tgsi_processor *processor; - + struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -882,12 +2027,12 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - + /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); @@ -898,16 +2043,16 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_input_attribs; i++) + /* Declare inputs (texcoords) */ + for (i = 0; i < 4; ++i) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = input_semantic_names[i]; - decl.Semantic.SemanticIndex = input_semantic_indexes[i]; - + decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; + decl.Semantic.SemanticIndex = i + 1; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; decl.u.DeclarationRange.First = i; decl.u.DeclarationRange.Last = i; ti += tgsi_build_full_declaration @@ -919,18 +2064,17 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) ); } - /* Declare constant inputs */ - /* C[0] scales the normalized MB to cover 16x16 pixels, - C[1] translates the macroblock into position on the surface - C[2] translates the past surface texcoords to the ref macroblock - C[3] translates the future surface texcoords to the ref macroblock */ + /* Declare constant input */ + /* C[0] is a multiplier to use when concatenating differential into a single channel + C[1] is a bias to get differential back to -1,1 range + C[2] contains 0.5 in channel X for use as a weight to blend past and future samples */ decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_CONSTANT; decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; decl.Semantic.SemanticIndex = 0; decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 3; + decl.u.DeclarationRange.Last = 2; ti += tgsi_build_full_declaration ( &decl, @@ -939,14 +2083,27 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) max_tokens - ti ); - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_output_attribs; i++) + /* Declare output */ + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; + decl.Semantic.SemanticIndex = 0; + decl.u.DeclarationRange.First = 0; + decl.u.DeclarationRange.Last = 0; + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + + /* Declare samplers */ + for (i = 0; i < 5; ++i) { decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = output_semantic_names[i]; - decl.Semantic.SemanticIndex = output_semantic_indexes[i]; + decl.Declaration.File = TGSI_FILE_SAMPLER; decl.u.DeclarationRange.First = i; decl.u.DeclarationRange.Last = i; ti += tgsi_build_full_declaration @@ -958,15 +2115,67 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) ); } - /* mul t0, i0, c0 ; Scale normalized coords to window coords */ + /* + tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + mov t1.x, t0.w ; Move high part from .w channel to .x + tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + mov t1.y, t0.w ; Move high part from .w channel to .y + tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels + mov t1.z, t0.w ; Move high part from .w channel to .z + */ + for (i = 0; i < 3; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MOV; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_MUL; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.FullDstRegisters[0].DstRegister.Index = 1; inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 1; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; inst.FullSrcRegisters[1].SrcRegister.Index = 0; ti += tgsi_build_full_instruction @@ -976,17 +2185,17 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) header, max_tokens - ti ); - - /* add o0, t0, c1 ; Translate vertex into position */ + + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_ADD; inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; inst.FullDstRegisters[0].DstRegister.Index = 0; inst.Instruction.NumSrcRegs = 2; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; inst.FullSrcRegisters[1].SrcRegister.Index = 1; ti += tgsi_build_full_instruction ( @@ -996,43 +2205,40 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) max_tokens - ti ); - /* - mov o1, i1 ; Move luma & chroma texcoords to output - mov o2, i2 - */ - for (i = 1; i < num_output_attribs - 2; ++i) - { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); - } + /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); - /* add o3, t0, c2 ; Translate past surface texcoords into position - add o4, t0, c3 ; Repeat for future surface texcoords */ + /* tex2d t1, i2, s3 ; Read texel from past macroblock + tex2d t2, i3, s4 ; Read texel from future macroblock */ for (i = 0; i < 2; ++i) { inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.Opcode = TGSI_OPCODE_TEX; inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i + 3; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = i + 1; inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = i + 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; ti += tgsi_build_full_instruction ( &inst, @@ -1041,6 +2247,50 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) max_tokens - ti ); } + + /* lerp t1, c2.x, t1, t2 ; Blend past and future texels */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_LERP; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 3; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[0].SrcRegister.Index = 2; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[2].SrcRegister.Index = 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; + inst.FullDstRegisters[0].DstRegister.Index = 0; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); /* END */ inst = tgsi_default_full_instruction(); @@ -1054,19 +2304,19 @@ static int vlCreateVertexShaderBMC(struct VL_CONTEXT *context) header, max_tokens - ti ); + + fs.tokens = tokens; - vs.tokens = tokens; - - context->states.mc.b_vs = pipe->create_vs_state(pipe, &vs); + context->states.mc.b_fs[0] = pipe->create_fs_state(pipe, &fs); free(tokens); return 0; } -static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) +static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) { - const unsigned int max_tokens = 100; + const unsigned int max_tokens = 200; const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; struct pipe_context *pipe; @@ -1100,8 +2350,15 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) ti = 3; - /* Declare inputs (texcoords) */ - for (i = 0; i < 4; ++i) + /* Declare inputs (texcoords) + I[0] Luma texcoords + I[1] Chroma texcoords + I[2] Past top field surface texcoords + I[3] Past bottom field surface texcoords + I[4] Future top field surface texcoords + I[5] Future bottom field surface texcoords + I[6] Denormalized texel pos */ + for (i = 0; i < 7; ++i) { decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_INPUT; @@ -1123,14 +2380,15 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) /* Declare constant input */ /* C[0] is a multiplier to use when concatenating differential into a single channel - C[0] is a bias to get differential back to -1,1 range*/ + C[1] is a bias to get differential back to -1,1 range + C[2] is constants 2 and 1/2 for Y%2 field selector */ decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_CONSTANT; decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; decl.Semantic.SemanticIndex = 0; decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 1; + decl.u.DeclarationRange.Last = 2; ti += tgsi_build_full_declaration ( &decl, @@ -1280,8 +2538,118 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) max_tokens - ti ); - /* tex2d t1, i2, s3 ; Read texel from past macroblock - tex2d t2, i3, s4 ; Read texel from future macroblock */ + /* XXX: Pos values off by 0.5 for rounding? */ + /* sub t4, i6.y, c2.x ; Sub 0.5 from position */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 4; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = 6; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* mul t3, t4, c2.x ; Divide pos y coord by 2 (mul by 0.5) */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* floor t3, t3 ; Get rid of fractional part */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_FLOOR; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* mul t3, t3, c2.y ; Multiply by 2 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_MUL; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* sub t3, t4, t3 ; Subtract from y to get y % 2 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_SUB; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 3; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 3; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* tex2d t1, i2, s3 ; Read texel from past macroblock top field + tex2d t2, i3, s3 ; Read texel from past macroblock bottom field */ for (i = 0; i < 2; ++i) { inst = tgsi_default_full_instruction(); @@ -1294,7 +2662,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; + inst.FullSrcRegisters[1].SrcRegister.Index = 3; ti += tgsi_build_full_instruction ( &inst, @@ -1304,17 +2672,89 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) ); } - /* add t0, t0, t1 ; Add past and differential to form partial output */ + /* lerp t1, t3, t1, t2 */ inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; + inst.Instruction.Opcode = TGSI_OPCODE_LERP; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 3; inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; + inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[2].SrcRegister.Index = 2; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* tex2d t4, i4, s4 ; Read texel from future macroblock top field + tex2d t5, i5, s4 ; Read texel from future macroblock bottom field */ + for (i = 0; i < 2; ++i) + { + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = i + 4; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; + inst.FullSrcRegisters[0].SrcRegister.Index = i + 4; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; + inst.FullSrcRegisters[1].SrcRegister.Index = 4; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* lerp t2, t3, t4, t5 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_LERP; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 2; + inst.Instruction.NumSrcRegs = 3; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[0].SrcRegister.Index = 3; + inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[1].SrcRegister.Index = 4; + inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[2].SrcRegister.Index = 5; + ti += tgsi_build_full_instruction + ( + &inst, + &tokens[ti], + header, + max_tokens - ti + ); + + /* lerp t1, c2.x, t1, t2 */ + inst = tgsi_default_full_instruction(); + inst.Instruction.Opcode = TGSI_OPCODE_LERP; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; + inst.FullDstRegisters[0].DstRegister.Index = 1; + inst.Instruction.NumSrcRegs = 3; + inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_CONSTANT; + inst.FullSrcRegisters[0].SrcRegister.Index = 2; + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; inst.FullSrcRegisters[1].SrcRegister.Index = 1; + inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; + inst.FullSrcRegisters[2].SrcRegister.Index = 2; ti += tgsi_build_full_instruction ( &inst, @@ -1323,7 +2763,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) max_tokens - ti ); - /* add o0, t0, t2 ; Add future and differential to form final output */ + /* add o0, t0, t1 ; Add future and differential to form final output */ inst = tgsi_default_full_instruction(); inst.Instruction.Opcode = TGSI_OPCODE_ADD; inst.Instruction.NumDstRegs = 1; @@ -1333,7 +2773,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; + inst.FullSrcRegisters[1].SrcRegister.Index = 1; ti += tgsi_build_full_instruction ( &inst, @@ -1357,7 +2797,7 @@ static int vlCreateFragmentShaderBMC(struct VL_CONTEXT *context) fs.tokens = tokens; - context->states.mc.b_fs = pipe->create_fs_state(pipe, &fs); + context->states.mc.b_fs[1] = pipe->create_fs_state(pipe, &fs); free(tokens); @@ -1491,7 +2931,6 @@ static int vlInitMC(struct VL_CONTEXT *context) context->states.mc.render_target.height = context->video_height; context->states.mc.render_target.num_cbufs = 1; /* FB for MC stage is a VL_SURFACE, set in vlSetRenderSurface() */ - /*context->states.mc.render_target.cbufs[0] = ;*/ context->states.mc.render_target.zsbuf = NULL; filters[0] = PIPE_TEX_FILTER_NEAREST; @@ -1530,6 +2969,7 @@ static int vlInitMC(struct VL_CONTEXT *context) template.depth[0] = 1; template.compressed = 0; template.cpp = 2; + context->states.mc.textures[0] = pipe->screen->texture_create(pipe->screen, &template); if (context->video_format == VL_FORMAT_YCBCR_420) @@ -1548,10 +2988,14 @@ static int vlInitMC(struct VL_CONTEXT *context) vlCreateVertexShaderIMC(context); vlCreateFragmentShaderIMC(context); - vlCreateVertexShaderPMC(context); - vlCreateFragmentShaderPMC(context); - vlCreateVertexShaderBMC(context); - vlCreateFragmentShaderBMC(context); + vlCreateVertexShaderFramePMC(context); + vlCreateVertexShaderFieldPMC(context); + vlCreateFragmentShaderFramePMC(context); + vlCreateFragmentShaderFieldPMC(context); + vlCreateVertexShaderFrameBMC(context); + vlCreateVertexShaderFieldBMC(context); + vlCreateFragmentShaderFrameBMC(context); + vlCreateFragmentShaderFieldBMC(context); vlCreateDataBufsMC(context); return 0; @@ -1575,10 +3019,14 @@ static int vlDestroyMC(struct VL_CONTEXT *context) context->pipe->delete_vs_state(context->pipe, context->states.mc.i_vs); context->pipe->delete_fs_state(context->pipe, context->states.mc.i_fs); - context->pipe->delete_vs_state(context->pipe, context->states.mc.p_vs); - context->pipe->delete_fs_state(context->pipe, context->states.mc.p_fs); - context->pipe->delete_vs_state(context->pipe, context->states.mc.b_vs); - context->pipe->delete_fs_state(context->pipe, context->states.mc.b_fs); + + for (i = 0; i < 2; ++i) + { + context->pipe->delete_vs_state(context->pipe, context->states.mc.p_vs[i]); + context->pipe->delete_fs_state(context->pipe, context->states.mc.p_fs[i]); + context->pipe->delete_vs_state(context->pipe, context->states.mc.b_vs[i]); + context->pipe->delete_fs_state(context->pipe, context->states.mc.b_fs[i]); + } context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vs_const_buf.buffer); context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.fs_const_buf.buffer); @@ -1982,7 +3430,7 @@ static int vlCreateDataBufsCSC(struct VL_CONTEXT *context) /* TODO: Refactor this into a seperate function, - allow changing the csc matrix at runtime to switch between regular & full versions + allow changing the CSC matrix at runtime to switch between regular & full versions */ memcpy ( @@ -2166,7 +3614,7 @@ static int vlDestroy(struct VL_CONTEXT *context) { assert(context); - /* Must unbind shaders before we can delete them for some reason */ + /* XXX: Must unbind shaders before we can delete them for some reason */ context->pipe->bind_vs_state(context->pipe, NULL); context->pipe->bind_fs_state(context->pipe, NULL); diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h index f26a4c5b6a7..8a123180734 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.h +++ b/src/gallium/state_trackers/g3dvl/vl_context.h @@ -34,8 +34,8 @@ struct VL_CONTEXT struct pipe_framebuffer_state render_target; struct pipe_sampler_state *samplers[5]; struct pipe_texture *textures[5]; - struct pipe_shader_state *i_vs, *p_vs, *b_vs; - struct pipe_shader_state *i_fs, *p_fs, *b_fs; + struct pipe_shader_state *i_vs, *p_vs[2], *b_vs[2]; + struct pipe_shader_state *i_fs, *p_fs[2], *b_fs[2]; struct pipe_vertex_buffer vertex_bufs[3]; struct pipe_vertex_element vertex_buf_elems[3]; struct pipe_constant_buffer vs_const_buf, fs_const_buf; diff --git a/src/gallium/state_trackers/g3dvl/vl_data.c b/src/gallium/state_trackers/g3dvl/vl_data.c index 27893aee95e..7e6ee8ac12f 100644 --- a/src/gallium/state_trackers/g3dvl/vl_data.c +++ b/src/gallium/state_trackers/g3dvl/vl_data.c @@ -87,7 +87,8 @@ const struct VL_TEXCOORD2F *vl_surface_texcoords = (const struct VL_TEXCOORD2F*) const struct VL_MC_FS_CONSTS vl_mc_fs_consts = { {256.0f, 256.0f, 256.0f, 0.0f}, - {256.0f / 255.0f, 256.0f / 255.0f, 256.0f / 255.0f, 0.0f} + {256.0f / 255.0f, 256.0f / 255.0f, 256.0f / 255.0f, 0.0f}, + {0.5f, 2.0f, 0.0f, 0.0f} }; /* diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 6451e54953d..d2220d7abf5 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -8,6 +8,85 @@ #include "vl_context.h" #include "vl_defs.h" +static int vlGrabFrameCodedFullBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + memcpy + ( + dst + y * dst_pitch, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabFrameCodedDiffBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int x, y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + dst[y * dst_pitch + x] = src[y * VL_BLOCK_WIDTH + x] + 0x100; + + return 0; +} + +static int vlGrabFieldCodedFullBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) + memcpy + ( + dst + y * dst_pitch * 2, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + dst += VL_BLOCK_HEIGHT * dst_pitch; + + for (; y < VL_BLOCK_HEIGHT; ++y) + memcpy + ( + dst + y * dst_pitch * 2, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabFieldCodedDiffBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int x, y; + + for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + dst[y * dst_pitch * 2 + x] = src[y * VL_BLOCK_WIDTH + x] + 0x100; + + dst += VL_BLOCK_HEIGHT * dst_pitch; + + for (; y < VL_BLOCK_HEIGHT; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + dst[y * dst_pitch * 2 + x] = src[y * VL_BLOCK_WIDTH + x] + 0x100; + + return 0; +} + +static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) +{ + unsigned int x, y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + dst[y * dst_pitch + x] = 0x100; + + return 0; +} + static int vlGrabBlocks ( struct VL_CONTEXT *context, @@ -19,7 +98,7 @@ static int vlGrabBlocks { struct pipe_surface *tex_surface; short *texels; - unsigned int b, x, y, y2; + unsigned int tb, sb = 0; assert(context); assert(blocks); @@ -33,134 +112,81 @@ static int vlGrabBlocks texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); - for (b = 0; b < 4; ++b) + for (tb = 0; tb < 4; ++tb) { - if ((coded_block_pattern >> b) & 1) + if ((coded_block_pattern >> (5 - tb)) & 1) { if (dct_type == VL_DCT_FRAME_CODED) - { if (sample_type == VL_FULL_SAMPLE) - { - for (y = VL_BLOCK_HEIGHT * b; y < VL_BLOCK_HEIGHT * (b + 1); ++y) - memcpy - ( - texels + y * tex_surface->pitch, - blocks + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - } + vlGrabFrameCodedFullBlock + ( + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels + tb * tex_surface->pitch * VL_BLOCK_HEIGHT, + tex_surface->pitch + ); else - { - for (y = VL_BLOCK_HEIGHT * b; y < VL_BLOCK_HEIGHT * (b + 1); ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - texels[y * tex_surface->pitch + x] = - blocks[y * VL_BLOCK_WIDTH + x] + 0x100; - } - } + vlGrabFrameCodedDiffBlock + ( + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels + tb * tex_surface->pitch * VL_BLOCK_HEIGHT, + tex_surface->pitch + ); else - { if (sample_type == VL_FULL_SAMPLE) - { - for + vlGrabFieldCodedFullBlock ( - y = VL_BLOCK_HEIGHT * (b % 2), y2 = VL_BLOCK_HEIGHT * b; - y < VL_BLOCK_HEIGHT * ((b % 2) + 1); - y += 2, ++y2 - ) - memcpy - ( - texels + y * tex_surface->pitch, - blocks + y2 * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - for - ( - y = VL_BLOCK_HEIGHT * ((b % 2) + 2); - y < VL_BLOCK_HEIGHT * (((b % 2) + 2) + 1); - y += 2, ++y2 - ) - memcpy - ( - texels + y * tex_surface->pitch, - blocks + y2 * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - } + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels + (tb % 2) * tex_surface->pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_surface->pitch, + tex_surface->pitch + ); else - { - for + vlGrabFieldCodedDiffBlock ( - y = VL_BLOCK_HEIGHT * (b % 2), y2 = VL_BLOCK_HEIGHT * b; - y < VL_BLOCK_HEIGHT * ((b % 2) + 1); - y += 2, ++y2 - ) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - texels[y * tex_surface->pitch + x] = - blocks[y2 * VL_BLOCK_WIDTH + x] + 0x100; - for - ( - y = VL_BLOCK_HEIGHT * ((b % 2) + 2); - y < VL_BLOCK_HEIGHT * (((b % 2) + 2) + 1); - y += 2, ++y2 - ) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - texels[y * tex_surface->pitch + x] = - blocks[y2 * VL_BLOCK_WIDTH + x] + 0x100; - } - } + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels + (tb % 2) * tex_surface->pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_surface->pitch, + tex_surface->pitch + ); + ++sb; } else - { - for (y = VL_BLOCK_HEIGHT * b; y < VL_BLOCK_HEIGHT * (b + 1); ++y) - { - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - texels[y * tex_surface->pitch + x] = 0x100; - } - } + vlGrabNoBlock(texels + tb * tex_surface->pitch * VL_BLOCK_HEIGHT, tex_surface->pitch); } pipe_surface_unmap(tex_surface); /* TODO: Implement 422, 444 */ - for (b = 0; b < 2; ++b) + for (tb = 0; tb < 2; ++tb) { tex_surface = context->pipe->screen->get_tex_surface - ( - context->pipe->screen, - context->states.mc.textures[b + 1], - 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE - ); + ( + context->pipe->screen, + context->states.mc.textures[tb + 1], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); - if ((coded_block_pattern >> (b + 4)) & 1) - { + if ((coded_block_pattern >> (1 - tb)) & 1) + { if (sample_type == VL_FULL_SAMPLE) - { - for (y = 0; y < tex_surface->height; ++y) - memcpy - ( - texels + y * tex_surface->pitch, - blocks + VL_BLOCK_SIZE * (b + 4) + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - } + vlGrabFrameCodedFullBlock + ( + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels, + tex_surface->pitch + ); else - { - for (y = 0; y < tex_surface->height; ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - texels[y * tex_surface->pitch + x] = - blocks[VL_BLOCK_SIZE * (b + 4) + y * VL_BLOCK_WIDTH + x] + 0x100; - } + vlGrabFrameCodedDiffBlock + ( + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels, + tex_surface->pitch + ); + + ++sb; } else - { - for (y = 0; y < tex_surface->height; ++y) - { - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - texels[y * tex_surface->pitch + x] = 0x100; - } - } + vlGrabNoBlock(texels, tex_surface->pitch); pipe_surface_unmap(tex_surface); } @@ -229,41 +255,35 @@ int vlRenderIMacroBlock ) { struct pipe_context *pipe; - struct VL_MC_VS_CONSTS *vscbdata; + struct VL_MC_VS_CONSTS *vs_consts; assert(blocks); assert(surface); /* TODO: Implement interlaced rendering */ - /*assert(picture_type == VL_FRAME_PICTURE);*/ if (picture_type != VL_FRAME_PICTURE) - { - /*fprintf(stderr, "field picture (I) unimplemented, ignoring\n");*/ return 0; - } pipe = surface->context->pipe; - vscbdata = pipe->winsys->buffer_map + vs_consts = pipe->winsys->buffer_map ( pipe->winsys, surface->context->states.mc.vs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE ); - vscbdata->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; - vscbdata->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; - vscbdata->scale.z = 1.0f; - vscbdata->scale.w = 1.0f; - vscbdata->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; - vscbdata->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; - vscbdata->mb_pos_trans.z = 0.0f; - vscbdata->mb_pos_trans.w = 0.0f; + vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; + vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; + vs_consts->scale.z = 1.0f; + vs_consts->scale.w = 1.0f; + vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; + vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; + vs_consts->mb_pos_trans.z = 0.0f; + vs_consts->mb_pos_trans.w = 0.0f; pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_FULL_SAMPLE, blocks); - surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface ( pipe->screen, @@ -276,6 +296,8 @@ int vlRenderIMacroBlock pipe->bind_vs_state(pipe, surface->context->states.mc.i_vs); pipe->bind_fs_state(pipe, surface->context->states.mc.i_fs); + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_FULL_SAMPLE, blocks); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); return 0; @@ -297,7 +319,7 @@ int vlRenderPMacroBlock ) { struct pipe_context *pipe; - struct VL_MC_VS_CONSTS *vscbdata; + struct VL_MC_VS_CONSTS *vs_consts; assert(motion_vectors); assert(blocks); @@ -305,46 +327,55 @@ int vlRenderPMacroBlock assert(surface); /* TODO: Implement interlaced rendering */ - /*assert(picture_type == VL_FRAME_PICTURE);*/ if (picture_type != VL_FRAME_PICTURE) - { - /*fprintf(stderr, "field picture (P) unimplemented, ignoring\n");*/ return 0; - } - /* TODO: Implement field based motion compensation */ - /*assert(mc_type == VL_FRAME_MC);*/ - if (mc_type != VL_FRAME_MC) - { - /*fprintf(stderr, "field MC (P) unimplemented, ignoring\n");*/ + /* TODO: Implement other MC types */ + if (mc_type != VL_FRAME_MC && mc_type != VL_FIELD_MC) return 0; - } pipe = surface->context->pipe; - vscbdata = pipe->winsys->buffer_map + vs_consts = pipe->winsys->buffer_map ( pipe->winsys, surface->context->states.mc.vs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE ); - vscbdata->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; - vscbdata->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; - vscbdata->scale.z = 1.0f; - vscbdata->scale.w = 1.0f; - vscbdata->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; - vscbdata->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; - vscbdata->mb_pos_trans.z = 0.0f; - vscbdata->mb_pos_trans.w = 0.0f; - vscbdata->mb_tc_trans[0].x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector->top_field.x * 0.5f) / (float)surface->width; - vscbdata->mb_tc_trans[0].y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector->top_field.y * 0.5f) / (float)surface->height; - vscbdata->mb_tc_trans[0].z = 0.0f; - vscbdata->mb_tc_trans[0].w = 0.0f; + vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; + vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; + vs_consts->scale.z = 1.0f; + vs_consts->scale.w = 1.0f; + vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; + vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; + vs_consts->mb_pos_trans.z = 0.0f; + vs_consts->mb_pos_trans.w = 0.0f; + vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector->top_field.x * 0.5f) / (float)surface->width; + vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector->top_field.y * 0.5f) / (float)surface->height; + vs_consts->mb_tc_trans[0].top_field.z = 0.0f; + vs_consts->mb_tc_trans[0].top_field.w = 0.0f; + + if (mc_type == VL_FIELD_MC) + { + vs_consts->denorm.x = (float)surface->width; + vs_consts->denorm.y = (float)surface->height; + + vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector->bottom_field.x * 0.5f) / (float)surface->width; + vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector->bottom_field.y * 0.5f) / (float)surface->height; + vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f; + vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f; + + pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs[1]); + pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs[1]); + } + else + { + pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs[0]); + pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs[0]); + } pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); - surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface ( pipe->screen, @@ -356,8 +387,8 @@ int vlRenderPMacroBlock surface->context->states.mc.textures[3] = ref_surface->texture; pipe->set_sampler_textures(pipe, 4, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 4, (void**)surface->context->states.mc.samplers); - pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs); - pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs); + + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); @@ -381,7 +412,7 @@ int vlRenderBMacroBlock ) { struct pipe_context *pipe; - struct VL_MC_VS_CONSTS *vscbdata; + struct VL_MC_VS_CONSTS *vs_consts; assert(motion_vectors); assert(blocks); @@ -389,50 +420,63 @@ int vlRenderBMacroBlock assert(surface); /* TODO: Implement interlaced rendering */ - /*assert(picture_type == VL_FRAME_PICTURE);*/ if (picture_type != VL_FRAME_PICTURE) - { - /*fprintf(stderr, "field picture (B) unimplemented, ignoring\n");*/ return 0; - } - /* TODO: Implement field based motion compensation */ - /*assert(mc_type == VL_FRAME_MC);*/ - if (mc_type != VL_FRAME_MC) - { - /*fprintf(stderr, "field MC (B) unimplemented, ignoring\n");*/ + /* TODO: Implement other MC types */ + if (mc_type != VL_FRAME_MC && mc_type != VL_FIELD_MC) return 0; - } pipe = surface->context->pipe; - vscbdata = pipe->winsys->buffer_map + vs_consts = pipe->winsys->buffer_map ( pipe->winsys, surface->context->states.mc.vs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE ); - vscbdata->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; - vscbdata->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; - vscbdata->scale.z = 1.0f; - vscbdata->scale.w = 1.0f; - vscbdata->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; - vscbdata->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; - vscbdata->mb_pos_trans.z = 0.0f; - vscbdata->mb_pos_trans.w = 0.0f; - vscbdata->mb_tc_trans[0].x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[0].top_field.x * 0.5f) / (float)surface->width; - vscbdata->mb_tc_trans[0].y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[0].top_field.y * 0.5f) / (float)surface->height; - vscbdata->mb_tc_trans[0].z = 0.0f; - vscbdata->mb_tc_trans[0].w = 0.0f; - vscbdata->mb_tc_trans[1].x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[1].top_field.x * 0.5f) / (float)surface->width; - vscbdata->mb_tc_trans[1].y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[1].top_field.y * 0.5f) / (float)surface->height; - vscbdata->mb_tc_trans[1].z = 0.0f; - vscbdata->mb_tc_trans[1].w = 0.0f; + vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; + vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; + vs_consts->scale.z = 1.0f; + vs_consts->scale.w = 1.0f; + vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; + vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; + vs_consts->mb_pos_trans.z = 0.0f; + vs_consts->mb_pos_trans.w = 0.0f; + vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[0].top_field.x * 0.5f) / (float)surface->width; + vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[0].top_field.y * 0.5f) / (float)surface->height; + vs_consts->mb_tc_trans[0].top_field.z = 0.0f; + vs_consts->mb_tc_trans[0].top_field.w = 0.0f; + vs_consts->mb_tc_trans[1].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[1].top_field.x * 0.5f) / (float)surface->width; + vs_consts->mb_tc_trans[1].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[1].top_field.y * 0.5f) / (float)surface->height; + vs_consts->mb_tc_trans[1].top_field.z = 0.0f; + vs_consts->mb_tc_trans[1].top_field.w = 0.0f; + + if (mc_type == VL_FIELD_MC) + { + vs_consts->denorm.x = (float)surface->width; + vs_consts->denorm.y = (float)surface->height; + + vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[0].bottom_field.x * 0.5f) / (float)surface->width; + vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[0].bottom_field.y * 0.5f) / (float)surface->height; + vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f; + vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f; + vs_consts->mb_tc_trans[1].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[1].bottom_field.x * 0.5f) / (float)surface->width; + vs_consts->mb_tc_trans[1].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[1].bottom_field.y * 0.5f) / (float)surface->height; + vs_consts->mb_tc_trans[1].bottom_field.z = 0.0f; + vs_consts->mb_tc_trans[1].bottom_field.w = 0.0f; + + pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs[1]); + pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs[1]); + } + else + { + pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs[0]); + pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs[0]); + } pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); - surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface ( pipe->screen, @@ -445,8 +489,8 @@ int vlRenderBMacroBlock surface->context->states.mc.textures[4] = future_surface->texture; pipe->set_sampler_textures(pipe, 5, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 5, (void**)surface->context->states.mc.samplers); - pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs); - pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs); + + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); @@ -513,7 +557,7 @@ int vlPutSurface destw, desth, PIPE_FORMAT_A8R8G8B8_UNORM, - /*XXX: SoftPipe doesn't change GPU usage to CPU like it does for textures */ + /* XXX: SoftPipe doesn't change GPU usage to CPU like it does for textures */ PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE, 0 ); diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h index 7040b745036..97753699db6 100644 --- a/src/gallium/state_trackers/g3dvl/vl_types.h +++ b/src/gallium/state_trackers/g3dvl/vl_types.h @@ -36,7 +36,9 @@ enum VL_SAMPLE_TYPE enum VL_MC_TYPE { VL_FIELD_MC, - VL_FRAME_MC + VL_FRAME_MC, + VL_DUAL_PRIME_MC, + VL_16x8_MC = VL_FRAME_MC }; struct VL_VERTEX4F @@ -58,13 +60,19 @@ struct VL_MC_VS_CONSTS { struct VL_VERTEX4F scale; struct VL_VERTEX4F mb_pos_trans; - struct VL_VERTEX4F mb_tc_trans[2]; + struct VL_VERTEX4F denorm; + struct + { + struct VL_VERTEX4F top_field; + struct VL_VERTEX4F bottom_field; + } mb_tc_trans[2]; }; struct VL_MC_FS_CONSTS { struct VL_VERTEX4F multiplier; struct VL_VERTEX4F bias; + struct VL_VERTEX4F y_divider; }; struct VL_CSC_FS_CONSTS diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index e4602d82042..5656895650e 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -32,6 +32,34 @@ static enum VL_PICTURE PictureToVL(int xvmc_pic) return vl_pic; } +static enum VL_MC_TYPE MotionToVL(int xvmc_motion_type) +{ + enum VL_MC_TYPE vl_mc_type; + + switch (xvmc_motion_type) + { + case XVMC_PREDICTION_FRAME: + { + vl_mc_type = VL_FRAME_MC; + break; + } + case XVMC_PREDICTION_FIELD: + { + vl_mc_type = VL_FIELD_MC; + break; + } + case XVMC_PREDICTION_DUAL_PRIME: + { + vl_mc_type = VL_DUAL_PRIME_MC; + break; + } + default: + assert(0); + } + + return vl_mc_type; +} + Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *surface) { struct VL_CONTEXT *vl_ctx; @@ -155,7 +183,7 @@ Status XvMCRenderSurface flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, macroblocks->macro_blocks[i].x, macroblocks->macro_blocks[i].y, - macroblocks->macro_blocks[i].motion_type == XVMC_PREDICTION_FRAME ? VL_FIELD_MC : VL_FRAME_MC, + MotionToVL(macroblocks->macro_blocks[i].motion_type), &motion_vector, macroblocks->macro_blocks[i].coded_block_pattern, macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, @@ -188,7 +216,7 @@ Status XvMCRenderSurface flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, macroblocks->macro_blocks[i].x, macroblocks->macro_blocks[i].y, - macroblocks->macro_blocks[i].motion_type == XVMC_PREDICTION_FRAME ? VL_FIELD_MC : VL_FRAME_MC, + MotionToVL(macroblocks->macro_blocks[i].motion_type), &motion_vector, macroblocks->macro_blocks[i].coded_block_pattern, macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, @@ -233,7 +261,7 @@ Status XvMCRenderSurface flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, macroblocks->macro_blocks[i].x, macroblocks->macro_blocks[i].y, - macroblocks->macro_blocks[i].motion_type == XVMC_PREDICTION_FRAME ? VL_FIELD_MC : VL_FRAME_MC, + MotionToVL(macroblocks->macro_blocks[i].motion_type), motion_vector, macroblocks->macro_blocks[i].coded_block_pattern, macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, -- cgit v1.2.3 From 1c893fd513f5335a81dd72db70d64763634ea856 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 29 Jun 2008 20:52:58 -0400 Subject: g3dvl: Simplify shader code. --- src/gallium/state_trackers/g3dvl/Makefile | 2 +- src/gallium/state_trackers/g3dvl/vl_context.c | 2968 ++++---------------- src/gallium/state_trackers/g3dvl/vl_shader_build.c | 205 ++ src/gallium/state_trackers/g3dvl/vl_shader_build.h | 61 + 4 files changed, 890 insertions(+), 2346 deletions(-) create mode 100644 src/gallium/state_trackers/g3dvl/vl_shader_build.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_shader_build.h diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile index a0d85fbcc8f..bd46d004e23 100644 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -1,5 +1,5 @@ TARGET = libg3dvl.a -OBJECTS = vl_context.o vl_data.o vl_surface.o +OBJECTS = vl_context.o vl_data.o vl_surface.o vl_shader_build.o GALLIUMDIR = ../.. CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index d2b1ad7948e..88da47c06a6 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -9,6 +9,7 @@ #include #include #include +#include "vl_shader_build.h" #include "vl_data.h" static int vlInitIDCT(struct VL_CONTEXT *context) @@ -32,21 +33,11 @@ static int vlDestroyIDCT(struct VL_CONTEXT *context) static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 50; - const unsigned int num_attribs = 3; - const unsigned int semantic_names[3] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ - }; - const unsigned int semantic_indexes[3] = {0, 1, 2}; - const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; struct pipe_shader_state vs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -57,158 +48,71 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; - - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant inputs */ - /* C[0] scales the normalized MB to cover 16x16 pixels, - C[1] translates the macroblock into position on the surface */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 1; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* mul t0, i0, c0 ; Scale normalized coords to window coords */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add o0, t0, c1 ; Translate vertex into position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o0, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - mov o1, i1 ; Move texcoords to output - mov o2, i2 - */ - for (i = 1; i < num_attribs; ++i) + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); vs.tokens = tokens; - context->states.mc.i_vs = pipe->create_vs_state(pipe, &vs); - free(tokens); return 0; @@ -217,13 +121,11 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 50; - const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; - + struct pipe_context *pipe; struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -234,120 +136,61 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); ti = 3; - /* Declare inputs (texcoords) */ + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + */ for (i = 0; i < 2; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = i + 1; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare output (color) */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare samplers */ + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + */ for (i = 0; i < 3; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti); } /* - tex2d o0.x, i0, s0 ; Read texel from luma texture into .x channel - tex2d o0.y, i1, s1 ; Read texel from chroma Cb texture into .y channel - tex2d o0.z, i1, s2 ; Read texel from chroma Cr texture into .z channel + * tex2d o0.x, i0, s0 ; Read texel from luma texture into .x channel + * tex2d o0.y, i1, s1 ; Read texel from chroma Cb texture into .y channel + * tex2d o0.z, i1, s2 ; Read texel from chroma Cr texture into .z channel */ for (i = 0; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); fs.tokens = tokens; - context->states.mc.i_fs = pipe->create_fs_state(pipe, &fs); - free(tokens); return 0; @@ -356,30 +199,11 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int num_input_attribs = 3; - const unsigned int num_output_attribs = 4; - const unsigned int input_semantic_names[3] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC /* Chroma texcoords */ - }; - const unsigned int output_semantic_names[4] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ - TGSI_SEMANTIC_GENERIC /* Ref surface texcoords */ - }; - const unsigned int input_semantic_indexes[3] = {0, 1, 2}; - const unsigned int output_semantic_indexes[4] = {0, 1, 2, 3}; - const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; struct pipe_shader_state vs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -390,179 +214,78 @@ static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_input_attribs; i++) + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = input_semantic_names[i]; - decl.Semantic.SemanticIndex = input_semantic_indexes[i]; - - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant inputs */ - /* C[0] scales the normalized MB to cover 16x16 pixels, - C[1] translates the macroblock into position on the surface - C[2] unused - C[3] translates the ref surface texcoords to the ref macroblock */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 3; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Unused + * decl c3 ; Translation vector to move ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_output_attribs; i++) + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Ref macroblock texcoords + */ + for (i = 0; i < 4; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = output_semantic_names[i]; - decl.Semantic.SemanticIndex = output_semantic_indexes[i]; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* mul t0, i0, c0 ; Scale normalized coords to window coords */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add o0, t0, c1 ; Translate vertex into position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o0, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - mov o1, i1 ; Move luma & chroma texcoords to output - mov o2, i2 - */ - for (i = 1; i < num_output_attribs - 1; ++i) + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* add o3, t0, c3 ; Translate texcoords into position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o3, t0, c3 ; Translate rect into position on ref macroblock */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); vs.tokens = tokens; - context->states.mc.p_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); return 0; @@ -571,32 +294,11 @@ static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int num_input_attribs = 3; - const unsigned int num_output_attribs = 6; - const unsigned int input_semantic_names[3] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC /* Chroma texcoords */ - }; - const unsigned int output_semantic_names[6] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ - TGSI_SEMANTIC_GENERIC, /* Top field surface texcoords */ - TGSI_SEMANTIC_GENERIC, /* Bottom field surface texcoords */ - TGSI_SEMANTIC_POSITION /* Pos */ - }; - const unsigned int input_semantic_indexes[3] = {0, 1, 2}; - const unsigned int output_semantic_indexes[6] = {0, 1, 2, 3, 4, 5}; - const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; struct pipe_shader_state vs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -607,34 +309,26 @@ static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_input_attribs; i++) + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = input_semantic_names[i]; - decl.Semantic.SemanticIndex = input_semantic_indexes[i]; - - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration ( &decl, @@ -644,183 +338,70 @@ static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) ); } - /* Declare constant inputs */ - /* C[0] scales the normalized MB to cover 16x16 pixels, - C[1] translates the macroblock into position on the surface - C[2] denormalizes pos components - C[3] translates the ref surface top field texcoords to the ref macroblock - C[4] translates the ref surface bottom field texcoords to the ref macroblock */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 4; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Denorm coefficients + * decl c3 ; Translation vector to move top field ref macroblock texcoords into position + * decl c4 ; Translation vector to move bottom field ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_output_attribs; i++) + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Top field ref macroblock texcoords + * decl o4 ; Bottom field ref macroblock texcoords + * decl o5 ; Denormalized vertex pos + */ + for (i = 0; i < 6; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = output_semantic_names[i]; - decl.Semantic.SemanticIndex = output_semantic_indexes[i]; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_output((i == 0 || i == 5) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* mul t0, i0, c0 ; Scale normalized coords to window coords */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add t1, t0, c1 ; Translate vertex into position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add t1, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* mov o0, t1 ; Move vertex pos to output */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - mov o1, i1 ; Move luma & chroma texcoords to output - mov o2, i2 + mov o1, i1 ; Move input luma texcoords to output + mov o2, i2 ; Move input chroma texcoords to output */ - for (i = 1; i < num_output_attribs - 1; ++i) + for (i = 1; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* add o3, t0, c3 ; Translate top field texcoords into position - add o4, t0, c4 ; Translate bottom field texcoords into position */ + /* add o3, t0, c3 ; Translate top field rect into position on ref macroblock + add o4, t0, c4 ; Translate bottom field rect into position on ref macroblock */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i + 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul o5, t1, c2 ; Denorm pos for fragment shader */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 5; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul o5, t1, c2 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 5, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); vs.tokens = tokens; - context->states.mc.p_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); return 0; @@ -829,13 +410,11 @@ static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; struct pipe_context *pipe; struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -846,258 +425,101 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); ti = 3; - /* Declare inputs (texcoords) */ + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + */ for (i = 0; i < 3; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = i + 1; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant input */ - /* C[0] is a multiplier to use when concatenating differential into a single channel - C[0] is a bias to get differential back to -1,1 range*/ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 1; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Multiplier to shift 9th bit of differential into place + * decl c1 ; Bias to get differential back to a signed value + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare output */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare samplers */ + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ for (i = 0; i < 4; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } /* - tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - mov t1.x, t0.w ; Move high part from .w channel to .x - tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - mov t1.y, t0.w ; Move high part from .w channel to .y - tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - mov t1.z, t0.w ; Move high part from .w channel to .z - */ + * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + * mov t1.x, t0.w ; Move 9th bit from .w channel to .x + * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + * mov t1.y, t0.w ; Move 9th bit from .w channel to .y + * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels + * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + */ for (i = 0; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t0, t0, c1 ; Subtract bias to get back signed values */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t1, i2, s3 ; Read texel from ref macroblock */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 2; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* tex2d t1, i2, s3 ; Read texel from ref macroblock */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 2, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); fs.tokens = tokens; - context->states.mc.p_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); return 0; @@ -1106,13 +528,11 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 200; - const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; struct pipe_context *pipe; struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -1123,399 +543,150 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); ti = 3; - /* Declare inputs (texcoords) - I[0] Luma texcoords - I[1] Chroma texcoords - I[2] Ref top field surface texcoords - I[3] Ref bottom field surface texcoords - I[4] Denormalized texel pos */ + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s3 + * decl i4 ; Denormalized vertex pos + */ for (i = 0; i < 5; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = i + 1; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant input */ - /* C[0] is a multiplier to use when concatenating differential into a single channel - C[1] is a bias to get differential back to -1,1 range - C[2] is constants 2 and 1/2 for Y%2 field selector */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 2; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Multiplier to shift 9th bit of differential into place + * decl c1 ; Bias to get differential back to a signed value + * decl c2 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare output */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare samplers */ + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ for (i = 0; i < 4; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } /* - tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - mov t1.x, t0.w ; Move high part from .w channel to .x - tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - mov t1.y, t0.w ; Move high part from .w channel to .y - tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - mov t1.z, t0.w ; Move high part from .w channel to .z - */ + * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + * mov t1.x, t0.w ; Move 9th bit from .w channel to .x + * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + * mov t1.y, t0.w ; Move 9th bit from .w channel to .y + * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels + * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + */ for (i = 0; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t0, t0, c1 ; Subtract bias to get back signed values */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t1, i2, s3 ; Read texel from ref macroblock top field - tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field */ + /* tex2d t1, i2, s3 ; Read texel from ref macroblock top field + tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = i + 1; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* XXX: Pos values off by 0.5 for rounding? */ - /* sub t4, i4.y, c2.x ; Sub 0.5 from position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 4; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 4; + /* XXX: Pos values off by 0.5? */ + /* sub t4, i4.y, c2.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 4, TGSI_FILE_INPUT, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t4, c2.x ; Divide pos y coord by 2 (mul by 0.5) */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; + /* mul t3, t4, c2.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 2); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* floor t3, t3 ; Get rid of fractional part */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_FLOOR; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* floor t3, t3 ; Get rid of fractional part */ + inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t3, c2.y ; Multiply by 2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; + /* mul t3, t3, c2.y ; Multiply by 2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 2); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* sub t3, t4, t3 ; Subtract from y to get y % 2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* lerp t1, t3, t1, t2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_LERP; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 3; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[2].SrcRegister.Index = 2; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); fs.tokens = tokens; - context->states.mc.p_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); return 0; @@ -1524,31 +695,11 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int num_input_attribs = 3; - const unsigned int num_output_attribs = 5; - const unsigned int input_semantic_names[3] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC - }; - const unsigned int output_semantic_names[5] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC, - TGSI_SEMANTIC_GENERIC - }; - const unsigned int input_semantic_indexes[3] = {0, 1, 2}; - const unsigned int output_semantic_indexes[5] = {0, 1, 2, 3, 4}; - const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; struct pipe_shader_state vs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -1559,185 +710,85 @@ static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_input_attribs; i++) + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = input_semantic_names[i]; - decl.Semantic.SemanticIndex = input_semantic_indexes[i]; - - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant inputs */ - /* C[0] scales the normalized MB to cover 16x16 pixels, - C[1] translates the macroblock into position on the surface - C[2] unused - C[3] translates the past surface texcoords to the ref macroblock - C[4] unused - C[5] translates the future surface texcoords to the ref macroblock */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 5; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Unused + * decl c3 ; Translation vector to move past ref macroblock texcoords into position + * decl c4 ; Unused + * decl c5 ; Translation vector to move future ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 5); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_output_attribs; i++) + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Past ref macroblock texcoords + * decl o4 ; Future ref macroblock texcoords + */ + for (i = 0; i < 5; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = output_semantic_names[i]; - decl.Semantic.SemanticIndex = output_semantic_indexes[i]; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* mul t0, i0, c0 ; Scale normalized coords to window coords */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); - - /* add o0, t0, c1 ; Translate vertex into position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - mov o1, i1 ; Move luma & chroma texcoords to output - mov o2, i2 - */ - for (i = 1; i < num_output_attribs - 2; ++i) + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* add o3, t0, c3 ; Translate past surface texcoords into position - add o4, t0, c5 ; Repeat for future surface texcoords */ + /* add o3, t0, c3 ; Translate rect into position on past ref macroblock + add o4, t0, c5 ; Translate rect into position on future ref macroblock */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i + 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = i * 2 + 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i * 2 + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); vs.tokens = tokens; - context->states.mc.b_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); return 0; @@ -1746,34 +797,11 @@ static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int num_input_attribs = 3; - const unsigned int num_output_attribs = 8; - const unsigned int input_semantic_names[3] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC /* Chroma texcoords */ - }; - const unsigned int output_semantic_names[8] = - { - TGSI_SEMANTIC_POSITION, - TGSI_SEMANTIC_GENERIC, /* Luma texcoords */ - TGSI_SEMANTIC_GENERIC, /* Chroma texcoords */ - TGSI_SEMANTIC_GENERIC, /* Top field past surface texcoords */ - TGSI_SEMANTIC_GENERIC, /* Bottom field past surface texcoords */ - TGSI_SEMANTIC_GENERIC, /* Top field future surface texcoords */ - TGSI_SEMANTIC_GENERIC, /* Bottom field future surface texcoords */ - TGSI_SEMANTIC_POSITION /* Pos */ - }; - const unsigned int input_semantic_indexes[3] = {0, 1, 2}; - const unsigned int output_semantic_indexes[8] = {0, 1, 2, 3, 4, 5, 6, 7}; - const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; struct pipe_shader_state vs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -1783,225 +811,102 @@ static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) assert(context); - pipe = context->pipe; - + pipe = context->pipe; tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_input_attribs; i++) + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = input_semantic_names[i]; - decl.Semantic.SemanticIndex = input_semantic_indexes[i]; - - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant inputs */ - /* C[0] scales the normalized MB to cover 16x16 pixels, - C[1] translates the macroblock into position on the surface - C[2] denormalizes pos components - C[3] translates the past surface top field texcoords to the ref macroblock - C[4] translates the past surface bottom field texcoords to the ref macroblock - C[5] translates the future surface top field texcoords to the ref macroblock - C[6] translates the future surface bottom field texcoords to the ref macroblock */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 6; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Denorm coefficients + * decl c3 ; Translation vector to move top field past ref macroblock texcoords into position + * decl c4 ; Translation vector to move bottom field past ref macroblock texcoords into position + * decl c5 ; Translation vector to move top field future ref macroblock texcoords into position + * decl c6 ; Translation vector to move bottom field future ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 6); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_output_attribs; i++) + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Top field past ref macroblock texcoords + * decl o4 ; Bottom field past ref macroblock texcoords + * decl o5 ; Top field future ref macroblock texcoords + * decl o6 ; Bottom field future ref macroblock texcoords + * decl o7 ; Denormalized vertex pos + */ + for (i = 0; i < 8; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = output_semantic_names[i]; - decl.Semantic.SemanticIndex = output_semantic_indexes[i]; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_output((i == 0 || i == 7) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* mul t0, i0, c0 ; Scale normalized coords to window coords */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add t1, t0, c1 ; Translate vertex into position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add t1, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* mov o0, t1 ; Move vertex pos to output */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - mov o1, i1 ; Move luma & chroma texcoords to output - mov o2, i2 - */ - for (i = 1; i < num_output_attribs - 1; ++i) + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* add o3, t0, c3 ; Translate top field past texcoords into position - add o4, t0, c4 ; Translate bottom field past texcoords into position - add o5, t0, c5 ; Translate top field past texcoords into position - add o6, t0, c6 ; Translate bottom field past texcoords into position */ + /* + * add o3, t0, c3 ; Translate top field rect into position on past ref macroblock + * add o4, t0, c4 ; Translate bottom field rect into position on past ref macroblock + * add o5, t0, c5 ; Translate top field rect into position on future ref macroblock + * add o6, t0, c6 ; Translate bottom field rect into position on future ref macroblock + */ for (i = 0; i < 4; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i + 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul o7, t1, c2 ; Denorm pos for fragment shader */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 7; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul o7, t1, c2 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 7, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); vs.tokens = tokens; - context->states.mc.b_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); return 0; @@ -2010,13 +915,11 @@ static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 100; - const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; struct pipe_context *pipe; struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -2027,288 +930,118 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); ti = 3; - /* Declare inputs (texcoords) */ + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s4 + */ for (i = 0; i < 4; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = i + 1; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant input */ - /* C[0] is a multiplier to use when concatenating differential into a single channel - C[1] is a bias to get differential back to -1,1 range - C[2] contains 0.5 in channel X for use as a weight to blend past and future samples */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 2; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Multiplier to shift 9th bit of differential into place + * decl c1 ; Bias to get differential back to a signed value + * decl c2 ; Constant 1/2 in .x channel to use as weight to blend past and future texels + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare output */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare samplers */ + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for past ref surface texture + * decl s4 ; Sampler for future ref surface texture + */ for (i = 0; i < 5; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } /* - tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - mov t1.x, t0.w ; Move high part from .w channel to .x - tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - mov t1.y, t0.w ; Move high part from .w channel to .y - tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - mov t1.z, t0.w ; Move high part from .w channel to .z - */ + * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + * mov t1.x, t0.w ; Move 9th bit from .w channel to .x + * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + * mov t1.y, t0.w ; Move 9th bit from .w channel to .y + * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels + * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + */ for (i = 0; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t0, t0, c1 ; Subtract bias to get back signed values */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t1, i2, s3 ; Read texel from past macroblock - tex2d t2, i3, s4 ; Read texel from future macroblock */ + /* + * tex2d t1, i2, s3 ; Read texel from past ref macroblock + * tex2d t2, i3, s4 ; Read texel from future ref macroblock + */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = i + 1; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i + 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* lerp t1, c2.x, t1, t2 ; Blend past and future texels */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_LERP; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 3; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[0].SrcRegister.Index = 2; + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[2].SrcRegister.Index = 2; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); fs.tokens = tokens; - context->states.mc.b_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); return 0; @@ -2317,13 +1050,11 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 200; - const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; struct pipe_context *pipe; struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -2334,471 +1065,179 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); ti = 3; - /* Declare inputs (texcoords) - I[0] Luma texcoords - I[1] Chroma texcoords - I[2] Past top field surface texcoords - I[3] Past bottom field surface texcoords - I[4] Future top field surface texcoords - I[5] Future bottom field surface texcoords - I[6] Denormalized texel pos */ + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s3 + * decl i4 ; Texcoords for s4 + * decl i5 ; Texcoords for s4 + * decl i6 ; Denormalized vertex pos + */ for (i = 0; i < 7; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = i + 1; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare constant input */ - /* C[0] is a multiplier to use when concatenating differential into a single channel - C[1] is a bias to get differential back to -1,1 range - C[2] is constants 2 and 1/2 for Y%2 field selector */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 2; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); - - /* Declare output */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Multiplier to shift 9th bit of differential into place + * decl c1 ; Bias to get differential back to a signed value + * decl c2 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels + * ; and for Y-mod-2 top/bottom field selection + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare samplers */ + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for past ref surface texture + * decl s4 ; Sampler for future ref surface texture + */ for (i = 0; i < 5; ++i) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } /* - tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - mov t1.x, t0.w ; Move high part from .w channel to .x - tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - mov t1.y, t0.w ; Move high part from .w channel to .y - tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - mov t1.z, t0.w ; Move high part from .w channel to .z - */ + * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels + * mov t1.x, t0.w ; Move 9th bit from .w channel to .x + * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels + * mov t1.y, t0.w ; Move 9th bit from .w channel to .y + * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels + * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + */ for (i = 0; i < 3; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i > 0 ? 1 : 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul t1, t1, c0 ; Muliply high part by multiplier to get back its full value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 1; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get a single value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* sub t0, t0, c1 ; Subtract bias to get back the signed value */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t0, t0, c1 ; Subtract bias to get back signed values */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* XXX: Pos values off by 0.5 for rounding? */ - /* sub t4, i6.y, c2.x ; Sub 0.5 from position */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 4; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 6; + /* XXX: Pos values off by 0.5? */ + /* sub t4, i6.y, c2.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 6, TGSI_FILE_CONSTANT, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t4, c2.x ; Divide pos y coord by 2 (mul by 0.5) */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; + /* mul t3, t4, c2.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 2); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* floor t3, t3 ; Get rid of fractional part */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_FLOOR; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* floor t3, t3 ; Get rid of fractional part */ + inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t3, c2.y ; Multiply by 2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MUL; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 2; + /* mul t3, t3, c2.y ; Multiply by 2 */ + inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 2); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* sub t3, t4, t3 ; Subtract from y to get y % 2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 3; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t1, i2, s3 ; Read texel from past macroblock top field - tex2d t2, i3, s3 ; Read texel from past macroblock bottom field */ + /* + * tex2d t1, i2, s3 ; Read texel from past ref macroblock top field + * tex2d t2, i3, s3 ; Read texel from past ref macroblock bottom field + */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = i + 1; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i + 2; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = 3; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* lerp t1, t3, t1, t2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_LERP; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 3; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[2].SrcRegister.Index = 2; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t4, i4, s4 ; Read texel from future macroblock top field - tex2d t5, i5, s4 ; Read texel from future macroblock bottom field */ + /* + * tex2d t4, i4, s4 ; Read texel from future ref macroblock top field + * tex2d t5, i5, s4 ; Read texel from future ref macroblock bottom field + */ for (i = 0; i < 2; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = i + 4; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i + 4; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = 4; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 4, TGSI_FILE_SAMPLER, 4); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* lerp t2, t3, t4, t5 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_LERP; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 2; - inst.Instruction.NumSrcRegs = 3; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 3; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 4; - inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[2].SrcRegister.Index = 5; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t2, t3, t4, t5 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* lerp t1, c2.x, t1, t2 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_LERP; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 1; - inst.Instruction.NumSrcRegs = 3; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[0].SrcRegister.Index = 2; + /* lerp t1, c2.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - inst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[2].SrcRegister.Index = 2; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* add o0, t0, t1 ; Add future and differential to form final output */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_ADD; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[1].SrcRegister.Index = 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* END */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); fs.tokens = tokens; - context->states.mc.b_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); return 0; @@ -2936,8 +1375,8 @@ static int vlInitMC(struct VL_CONTEXT *context) filters[0] = PIPE_TEX_FILTER_NEAREST; filters[1] = context->video_format == VL_FORMAT_YCBCR_444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; filters[2] = context->video_format == VL_FORMAT_YCBCR_444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; - filters[3] = PIPE_TEX_FILTER_NEAREST; - filters[4] = PIPE_TEX_FILTER_NEAREST; + filters[3] = PIPE_TEX_FILTER_LINEAR; + filters[4] = PIPE_TEX_FILTER_LINEAR; for (i = 0; i < 5; ++i) { @@ -3037,16 +1476,11 @@ static int vlDestroyMC(struct VL_CONTEXT *context) static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 50; - const unsigned int num_attribs = 2; - const unsigned int semantic_names[2] = {TGSI_SEMANTIC_POSITION, TGSI_SEMANTIC_GENERIC}; - const unsigned int semantic_indexes[2] = {0, 1}; - const unsigned int proc_type = TGSI_PROCESSOR_VERTEX; struct pipe_context *pipe; struct pipe_shader_state vs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -3057,101 +1491,54 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); ti = 3; - /* Declare inputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + /* + * decl i0 ; Vertex pos + * decl i1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; - - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* Declare outputs (pos, texcoords) */ - for (i = 0; i < num_attribs; i++) + /* + * decl o0 ; Vertex pos + * decl o1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) { - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = semantic_names[i]; - decl.Semantic.SemanticIndex = semantic_indexes[i]; - decl.u.DeclarationRange.First = i; - decl.u.DeclarationRange.Last = i; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* MOV instructions */ - /* mov o0, i0 - mov o1, i1 */ - for (i = 0; i < num_attribs; i++) + /* + * mov o0, i0 ; Move pos in to pos out + * mov o1, i1 ; Move input texcoords to output + */ + for (i = 0; i < 2; i++) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_MOV; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = i; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = i; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* END instruction */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); vs.tokens = tokens; - context->states.csc.vertex_shader = pipe->create_vs_state(pipe, &vs); - free(tokens); return 0; @@ -3160,13 +1547,11 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) { const unsigned int max_tokens = 50; - const unsigned int proc_type = TGSI_PROCESSOR_FRAGMENT; struct pipe_context *pipe; struct pipe_shader_state fs; struct tgsi_token *tokens; struct tgsi_header *header; - struct tgsi_processor *processor; struct tgsi_full_declaration decl; struct tgsi_full_instruction inst; @@ -3177,171 +1562,64 @@ static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) assert(context); pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); /* Version */ *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ header = (struct tgsi_header*)&tokens[1]; *header = tgsi_build_header(); - /* Processor */ - processor = (struct tgsi_processor*)&tokens[2]; - *processor = tgsi_build_processor(proc_type, header); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); ti = 3; - /* Declare TEX[0] input */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 1; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = TGSI_INTERPOLATE_LINEAR; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl i0 ; Texcoords for s0 */ + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare constant input */ - /* Constants include bias vector, 4x4 csc matrix, total 5 vectors */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 4; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* + * decl c0 ; Bias vector for CSC + * decl c1-c4 ; CSC matrix c1-c4 + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare output */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR; - decl.Semantic.SemanticIndex = 0; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* Declare sampler */ - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = 0; - decl.u.DeclarationRange.Last = 0; - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); + /* decl s0 ; Sampler for tex containing picture to display */ + decl = vl_decl_samplers(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* TEX instruction */ - /* tex2d t0, i0, s0 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* tex2d t0, i0, s0 ; Read src pixel */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* SUB instruction */ - /* sub t0, t0, c0 */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_SUB; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY; - inst.FullDstRegisters[0].DstRegister.Index = 0; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* sub t0, t0, c0 ; Subtract bias vector from pixel */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* DP4 instruction */ - /* dp4 o0.x, t0, c1 - dp4 o0.y, t0, c2 - dp4 o0.z, t0, c3 - dp4 o0.w, t0, c4 */ + /* + * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix + * dp4 o0.y, t0, c2 + * dp4 o0.z, t0, c3 + * dp4 o0.w, t0, c4 ; XXX: Don't need 4th coefficient + */ for (i = 0; i < 4; ++i) { - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_DP4; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT; - inst.FullDstRegisters[0].DstRegister.Index = 0; + inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY; - inst.FullSrcRegisters[0].SrcRegister.Index = 0; - inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_CONSTANT; - inst.FullSrcRegisters[1].SrcRegister.Index = i + 1; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* END instruction */ - inst = tgsi_default_full_instruction(); - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - ti += tgsi_build_full_instruction - ( - &inst, - &tokens[ti], - header, - max_tokens - ti - ); + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); fs.tokens = tokens; - context->states.csc.fragment_shader = pipe->create_fs_state(pipe, &fs); - free(tokens); return 0; diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c new file mode 100644 index 00000000000..fd9de7dab8c --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.c @@ -0,0 +1,205 @@ +#include "vl_shader_build.h" +#include +#include +#include + +struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.u.DeclarationRange.First = first; + decl.u.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_interpolated_input +( + unsigned int name, + unsigned int index, + unsigned int first, + unsigned int last, + int interpolation +) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + assert + ( + interpolation == TGSI_INTERPOLATE_CONSTANT || + interpolation == TGSI_INTERPOLATE_LINEAR || + interpolation == TGSI_INTERPOLATE_PERSPECTIVE + ); + + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.Declaration.Interpolate = 1; + decl.Interpolation.Interpolate = interpolation; + decl.u.DeclarationRange.First = first; + decl.u.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.u.DeclarationRange.First = first; + decl.u.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.u.DeclarationRange.First = first; + decl.u.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.u.DeclarationRange.First = first; + decl.u.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_instruction vl_inst2 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src_file, + unsigned int src_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = opcode; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = src_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src_index; + + return inst; +} + +struct tgsi_full_instruction vl_inst3 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = opcode; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = src1_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; + inst.FullSrcRegisters[1].SrcRegister.File = src2_file; + inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; + + return inst; +} + +struct tgsi_full_instruction vl_tex +( + int tex, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = tex; + inst.FullSrcRegisters[0].SrcRegister.File = src1_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; + inst.FullSrcRegisters[1].SrcRegister.File = src2_file; + inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; + + return inst; +} + +struct tgsi_full_instruction vl_inst4 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index, + enum tgsi_file_type src3_file, + unsigned int src3_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = opcode; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 3; + inst.FullSrcRegisters[0].SrcRegister.File = src1_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; + inst.FullSrcRegisters[1].SrcRegister.File = src2_file; + inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; + inst.FullSrcRegisters[2].SrcRegister.File = src3_file; + inst.FullSrcRegisters[2].SrcRegister.Index = src3_index; + + return inst; +} + +struct tgsi_full_instruction vl_end(void) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + + return inst; +} + diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.h b/src/gallium/state_trackers/g3dvl/vl_shader_build.h new file mode 100644 index 00000000000..9e64bbeeae0 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.h @@ -0,0 +1,61 @@ +#ifndef vl_shader_build_h +#define vl_shader_build_h + +#include + +struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_interpolated_input +( + unsigned int name, + unsigned int index, + unsigned int first, + unsigned int last, + int interpolation +); +struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last); +struct tgsi_full_instruction vl_inst2 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src_file, + unsigned int src_index +); +struct tgsi_full_instruction vl_inst3 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +); +struct tgsi_full_instruction vl_tex +( + int tex, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +); +struct tgsi_full_instruction vl_inst4 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index, + enum tgsi_file_type src3_file, + unsigned int src3_index +); +struct tgsi_full_instruction vl_end(void); + +#endif + -- cgit v1.2.3 From 9833aec6cbd113d24277aa5da8625c1427d831ca Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 30 Jun 2008 19:26:22 -0400 Subject: g3dvl: Use block and stride instead of cpp and pitch. --- src/gallium/state_trackers/g3dvl/vl_context.c | 2 +- src/gallium/state_trackers/g3dvl/vl_shader_build.c | 23 ++++++++--------- src/gallium/state_trackers/g3dvl/vl_surface.c | 29 ++++++++++++---------- src/gallium/winsys/g3dvl/xsp_winsys.c | 15 ++++++++--- src/libXvMC/Makefile | 2 +- 5 files changed, 41 insertions(+), 30 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 88da47c06a6..bd8743dc9a5 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -1407,7 +1407,7 @@ static int vlInitMC(struct VL_CONTEXT *context) template.height[0] = 8 * 4; template.depth[0] = 1; template.compressed = 0; - template.cpp = 2; + pf_get_block(template.format, &template.block); context->states.mc.textures[0] = pipe->screen->texture_create(pipe->screen, &template); diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c index fd9de7dab8c..365ad697257 100644 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.c +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.c @@ -11,8 +11,8 @@ struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = name; decl.Semantic.SemanticIndex = index; - decl.u.DeclarationRange.First = first; - decl.u.DeclarationRange.Last = last; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; return decl; } @@ -39,10 +39,9 @@ struct tgsi_full_declaration vl_decl_interpolated_input decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = name; decl.Semantic.SemanticIndex = index; - decl.Declaration.Interpolate = 1; - decl.Interpolation.Interpolate = interpolation; - decl.u.DeclarationRange.First = first; - decl.u.DeclarationRange.Last = last; + decl.Declaration.Interpolate = interpolation;; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; return decl; } @@ -55,8 +54,8 @@ struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int i decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = name; decl.Semantic.SemanticIndex = index; - decl.u.DeclarationRange.First = first; - decl.u.DeclarationRange.Last = last; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; return decl; } @@ -69,8 +68,8 @@ struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int inde decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = name; decl.Semantic.SemanticIndex = index; - decl.u.DeclarationRange.First = first; - decl.u.DeclarationRange.Last = last; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; return decl; } @@ -81,8 +80,8 @@ struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int l decl = tgsi_default_full_declaration(); decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.u.DeclarationRange.First = first; - decl.u.DeclarationRange.Last = last; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; return decl; } diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index d2220d7abf5..68313cc7505 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -98,6 +98,7 @@ static int vlGrabBlocks { struct pipe_surface *tex_surface; short *texels; + unsigned int tex_pitch; unsigned int tb, sb = 0; assert(context); @@ -111,6 +112,7 @@ static int vlGrabBlocks ); texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); + tex_pitch = tex_surface->stride / tex_surface->block.size; for (tb = 0; tb < 4; ++tb) { @@ -121,35 +123,35 @@ static int vlGrabBlocks vlGrabFrameCodedFullBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + tb * tex_surface->pitch * VL_BLOCK_HEIGHT, - tex_surface->pitch + texels + tb * tex_pitch * VL_BLOCK_HEIGHT, + tex_pitch ); else vlGrabFrameCodedDiffBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + tb * tex_surface->pitch * VL_BLOCK_HEIGHT, - tex_surface->pitch + texels + tb * tex_pitch * VL_BLOCK_HEIGHT, + tex_pitch ); else if (sample_type == VL_FULL_SAMPLE) vlGrabFieldCodedFullBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + (tb % 2) * tex_surface->pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_surface->pitch, - tex_surface->pitch + texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, + tex_pitch ); else vlGrabFieldCodedDiffBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + (tb % 2) * tex_surface->pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_surface->pitch, - tex_surface->pitch + texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, + tex_pitch ); ++sb; } else - vlGrabNoBlock(texels + tb * tex_surface->pitch * VL_BLOCK_HEIGHT, tex_surface->pitch); + vlGrabNoBlock(texels + tb * tex_pitch * VL_BLOCK_HEIGHT, tex_pitch); } pipe_surface_unmap(tex_surface); @@ -165,6 +167,7 @@ static int vlGrabBlocks ); texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); + tex_pitch = tex_surface->stride / tex_surface->block.size; if ((coded_block_pattern >> (1 - tb)) & 1) { @@ -173,20 +176,20 @@ static int vlGrabBlocks ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, texels, - tex_surface->pitch + tex_pitch ); else vlGrabFrameCodedDiffBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, texels, - tex_surface->pitch + tex_pitch ); ++sb; } else - vlGrabNoBlock(texels, tex_surface->pitch); + vlGrabNoBlock(texels, tex_pitch); pipe_surface_unmap(tex_surface); } @@ -223,7 +226,7 @@ int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface) template.height[0] = sfc->height; template.depth[0] = 1; template.compressed = 0; - template.cpp = 4; + pf_get_block(template.format, &template.block); template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; sfc->texture = pipe->screen->texture_create(pipe->screen, &template); diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index eb4c9085d6d..d04c4be580f 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -101,6 +101,12 @@ static struct pipe_surface* xsp_surface_alloc(struct pipe_winsys *pws) return surface; } +/* Borrowed from Mesa's xm_winsys */ +static unsigned int round_up(unsigned n, unsigned multiple) +{ + return (n + multiple - 1) & ~(multiple - 1); +} + static int xsp_surface_alloc_storage ( struct pipe_winsys *pws, @@ -120,10 +126,13 @@ static int xsp_surface_alloc_storage surface->width = width; surface->height = height; surface->format = format; - surface->cpp = pf_get_size(format); - surface->pitch = width; + pf_get_block(format, &surface->block); + surface->nblocksx = pf_get_nblocksx(&surface->block, width); + surface->nblocksy = pf_get_nblocksy(&surface->block, height); + surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT); surface->usage = flags; - surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->pitch * surface->cpp * height); + /* XXX: Need to consider block dims? See xm_winsys.c */ + surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * height); return 0; } diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile index b2f41e176d4..4985ecd3e9d 100644 --- a/src/libXvMC/Makefile +++ b/src/libXvMC/Makefile @@ -17,7 +17,7 @@ LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ -L${GALLIUMDIR}/auxiliary/cso_cache \ -L${GALLIUMDIR}/auxiliary/util \ -L${GALLIUMDIR}/auxiliary/rtasm -LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil +LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lm ############################################# -- cgit v1.2.3 From 36488ed052a18f7eafef1d1c5c18b20ad508b2b7 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 20:58:31 +0200 Subject: nv30: Emit framebuffer state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 3 + src/gallium/drivers/nv30/nv30_state.c | 105 +-------------------------- src/gallium/drivers/nv30/nv30_state_emit.c | 111 ++++++++++++++++++++--------- src/gallium/drivers/nv30/nv30_state_fb.c | 35 ++------- src/gallium/drivers/nv30/nv30_vbo.c | 2 + 5 files changed, 92 insertions(+), 164 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index dce077b2b4b..4a6c1d2f47b 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -99,6 +99,7 @@ struct nv30_state { unsigned stipple_enabled; unsigned viewport_bypass; + uint64_t dirty; struct nouveau_stateobj *hw[NV30_STATE_MAX]; }; @@ -208,8 +209,10 @@ extern void nv30_fragprog_destroy(struct nv30_context *, extern void nv30_fragtex_bind(struct nv30_context *); /* nv30_state.c and friends */ +extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_emit_hw_state(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); +extern struct nv30_state_entry nv30_state_framebuffer; /* nv30_vbo.c */ extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index c1618041bb0..3edfb0874da 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -565,110 +565,9 @@ nv30_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct nv30_context *nv30 = nv30_context(pipe); - struct pipe_surface *rt[2], *zeta = NULL; - uint32_t rt_enable, rt_format, w = 0, h = 0; - int i, colour_format = 0, zeta_format = 0; - - rt_enable = 0; - for (i = 0; i < 2; i++) { - if (!fb->cbufs[i]) - continue; - - if (colour_format) { - assert(w == fb->cbufs[i]->width); - assert(h == fb->cbufs[i]->height); - assert(colour_format == fb->cbufs[i]->format); - } else { - w = fb->cbufs[i]->width; - h = fb->cbufs[i]->height; - colour_format = fb->cbufs[i]->format; - rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i); - rt[i] = fb->cbufs[i]; - } - } - - if (rt_enable & (NV34TCL_RT_ENABLE_COLOR1 | NV34TCL_RT_ENABLE_COLOR2 | - NV34TCL_RT_ENABLE_COLOR3)) - rt_enable |= NV34TCL_RT_ENABLE_MRT; - - if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - - zeta_format = fb->zsbuf->format; - zeta = fb->zsbuf; - } - - rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR; - - switch (colour_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case 0: - rt_format |= NV34TCL_RT_FORMAT_COLOR_A8R8G8B8; - break; - case PIPE_FORMAT_R5G6B5_UNORM: - rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5; - break; - default: - assert(0); - } - - switch (zeta_format) { - case PIPE_FORMAT_Z16_UNORM: - rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16; - break; - case PIPE_FORMAT_Z24S8_UNORM: - case 0: - rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8; - break; - default: - assert(0); - } - - if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) { - uint32_t pitch = rt[0]->stride; - if (zeta) { - pitch |= (zeta->stride << 16); - } else { - pitch |= (pitch << 16); - } - - BEGIN_RING(rankine, NV34TCL_COLOR0_PITCH, 1); - OUT_RING (pitch); - nv30->rt[0] = rt[0]->buffer; - } - - if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(rankine, NV34TCL_COLOR1_PITCH, 1); - OUT_RING (rt[1]->stride); - nv30->rt[1] = rt[1]->buffer; - } - - if (zeta_format) - { - nv30->zeta = zeta->buffer; - } - nv30->rt_enable = rt_enable; - BEGIN_RING(rankine, NV34TCL_RT_ENABLE, 1); - OUT_RING (rt_enable); - BEGIN_RING(rankine, NV34TCL_RT_HORIZ, 3); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - OUT_RING (rt_format); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_HORIZ, 2); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0); - OUT_RING (((h - 1) << 16) | 0); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_TX_ORIGIN, 1); - OUT_RING (0); + nv30->framebuffer = *fb; + nv30->dirty |= NV30_NEW_FB; } static void diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index eda2fd45f56..9a3954594a7 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -1,10 +1,55 @@ #include "nv30_context.h" #include "nv30_state.h" +static struct nv30_state_entry *render_states[] = { + &nv30_state_framebuffer, + NULL +}; + +static void +nv30_state_do_validate(struct nv30_context *nv30, + struct nv30_state_entry **states) +{ + const struct pipe_framebuffer_state *fb = &nv30->framebuffer; + unsigned i; + + for (i = 0; i < fb->num_cbufs; i++) + fb->cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED; + if (fb->zsbuf) + fb->zsbuf->status = PIPE_SURFACE_STATUS_DEFINED; + + while (*states) { + struct nv30_state_entry *e = *states; + + if (nv30->dirty & e->dirty.pipe) { + if (e->validate(nv30)) { + nv30->state.dirty |= (1ULL << e->dirty.hw); + } + } + + states++; + } + +/* TODO: uncomment when finished converting + nv30->dirty = 0; +*/ +} + void nv30_emit_hw_state(struct nv30_context *nv30) { - int i; + struct nv30_state *state = &nv30->state; + unsigned i; + uint64 states; + + for (i = 0, states = state->dirty; states; i++) { + if (!(states & (1ULL << i))) + continue; + so_ref (state->hw[i], &nv30->screen->state[i]); + if (state->hw[i]) + so_emit(nv30->nvws, nv30->screen->state[i]); + states &= ~(1ULL << i); + } if (nv30->dirty & NV30_NEW_FRAGPROG) { nv30_fragprog_bind(nv30, nv30->fragprog.current); @@ -28,37 +73,7 @@ nv30_emit_hw_state(struct nv30_context *nv30) nv30->dirty_samplers = 0; - /* Emit relocs for every referenced buffer. - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. This isn't very efficient, but we don't - * seem to take a significant performance hit. Will be improved - * at some point. Vertex arrays are emitted by nv30_vbo.c - */ - - /* Render targets */ - if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR0) { - BEGIN_RING(rankine, NV34TCL_DMA_COLOR0, 1); - OUT_RELOCo(nv30->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(rankine, NV34TCL_COLOR0_OFFSET, 1); - OUT_RELOCl(nv30->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - } - - if (nv30->rt_enable & NV34TCL_RT_ENABLE_COLOR1) { - BEGIN_RING(rankine, NV34TCL_DMA_COLOR1, 1); - OUT_RELOCo(nv30->rt[1], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(rankine, NV34TCL_COLOR1_OFFSET, 1); - OUT_RELOCl(nv30->rt[1], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - } - - if (nv30->zeta) { - BEGIN_RING(rankine, NV34TCL_DMA_ZETA, 1); - OUT_RELOCo(nv30->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(rankine, NV34TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv30->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - /* XXX allocate LMA */ -/* BEGIN_RING(rankine, NV34TCL_LMA_DEPTH_OFFSET, 1); - OUT_RING(0);*/ - } + so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); /* Texture images, emitted in nv30_fragtex_build */ #if 0 @@ -83,3 +98,35 @@ nv30_emit_hw_state(struct nv30_context *nv30) NV34TCL_FP_ACTIVE_PROGRAM_DMA1); } +boolean +nv30_state_validate(struct nv30_context *nv30) +{ +#if 0 + boolean was_sw = nv30->fallback_swtnl ? TRUE : FALSE; + + if (nv30->render_mode != HW) { + /* Don't even bother trying to go back to hw if none + * of the states that caused swtnl previously have changed. + */ + if ((nv30->fallback_swtnl & nv30->dirty) + != nv30->fallback_swtnl) + return FALSE; + + /* Attempt to go to hwtnl again */ + nv30->pipe.flush(&nv30->pipe, 0, NULL); + nv30->dirty |= (NV30_NEW_VIEWPORT | + NV30_NEW_VERTPROG | + NV30_NEW_ARRAYS); + nv30->render_mode = HW; + } +#endif + nv30_state_do_validate(nv30, render_states); +#if 0 + if (nv30->fallback_swtnl || nv30->fallback_swrast) + return FALSE; + + if (was_sw) + NOUVEAU_ERR("swtnl->hw\n"); +#endif + return TRUE; +} diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c index a20df9f75d1..d93e2bb5c89 100644 --- a/src/gallium/drivers/nv30/nv30_state_fb.c +++ b/src/gallium/drivers/nv30/nv30_state_fb.c @@ -4,7 +4,7 @@ static boolean nv30_state_framebuffer_validate(struct nv30_context *nv30) { struct pipe_framebuffer_state *fb = &nv30->framebuffer; - struct pipe_surface *rt[4], *zeta = NULL; + struct pipe_surface *rt[2], *zeta = NULL; uint32_t rt_enable, rt_format; int i, colour_format = 0, zeta_format = 0; struct nouveau_stateobj *so = so_new(64, 10); @@ -23,8 +23,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) } } - if (rt_enable & (NV34TCL_RT_ENABLE_COLOR1 | NV34TCL_RT_ENABLE_COLOR2 | - NV34TCL_RT_ENABLE_COLOR3)) + if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) rt_enable |= NV34TCL_RT_ENABLE_MRT; if (fb->zsbuf) { @@ -86,31 +85,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) NOUVEAU_BO_LOW, 0, 0); so_data (so, rt[1]->stride); } -/* - if (rt_enable & NV34TCL_RT_ENABLE_COLOR2) { - so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR2, 1); - so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv30->nvws->channel->vram->handle, - nv30->nvws->channel->gart->handle); - so_method(so, nv30->screen->rankine, NV34TCL_COLOR2_OFFSET, 1); - so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv30->screen->rankine, NV34TCL_COLOR2_PITCH, 1); - so_data (so, rt[2]->pitch * rt[2]->cpp); - } - if (rt_enable & NV34TCL_RT_ENABLE_COLOR3) { - so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR3, 1); - so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR, - nv30->nvws->channel->vram->handle, - nv30->nvws->channel->gart->handle); - so_method(so, nv30->screen->rankine, NV34TCL_COLOR3_OFFSET, 1); - so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags | - NOUVEAU_BO_LOW, 0, 0); - so_method(so, nv30->screen->rankine, NV34TCL_COLOR3_PITCH, 1); - so_data (so, rt[3]->pitch * rt[3]->cpp); - } -*/ if (zeta_format) { so_method(so, nv30->screen->rankine, NV34TCL_DMA_ZETA, 1); so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR, @@ -119,8 +94,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) so_method(so, nv30->screen->rankine, NV34TCL_ZETA_OFFSET, 1); so_reloc (so, zeta->buffer, zeta->offset, rt_flags | NOUVEAU_BO_LOW, 0, 0); - /*so_method(so, nv30->screen->rankine, NV34TCL_ZETA_PITCH, 1); - so_data (so, zeta->pitch * zeta->cpp);*/ + /* TODO: allocate LMA depth buffer */ } so_method(so, nv30->screen->rankine, NV34TCL_RT_ENABLE, 1); @@ -137,6 +111,9 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30) so_data (so, ((h - 1) << 16) | 0); so_method(so, nv30->screen->rankine, 0x1d88, 1); so_data (so, (1 << 12) | h); + /* Wonder why this is needed, context should all be set to zero on init */ + so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_TX_ORIGIN, 1); + so_data (so, 0); so_ref(so, &nv30->state.hw[NV30_STATE_FB]); return TRUE; diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index 359e443bb1c..d23164eb485 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -154,6 +154,8 @@ nv30_vbo_validate_state(struct nv30_context *nv30, { unsigned inputs; + nv30_state_validate(nv30); + nv30_emit_hw_state(nv30); if (nv30->dirty & NV30_NEW_ARRAYS) { -- cgit v1.2.3 From 52cf7a6c1ccc987859834b640a5ec0a62f84134a Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 21:11:07 +0200 Subject: nv30: Emit blend color state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 1 + src/gallium/drivers/nv30/nv30_state.c | 7 ++----- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 3 files changed, 4 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 4a6c1d2f47b..57cc991cd7a 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -212,6 +212,7 @@ extern void nv30_fragtex_bind(struct nv30_context *); extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_emit_hw_state(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); +extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_framebuffer; /* nv30_vbo.c */ diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 3edfb0874da..e67a701809a 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -531,11 +531,8 @@ nv30_set_blend_color(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); - BEGIN_RING(rankine, NV34TCL_BLEND_COLOR, 1); - OUT_RING ((float_to_ubyte(bcol->color[3]) << 24) | - (float_to_ubyte(bcol->color[0]) << 16) | - (float_to_ubyte(bcol->color[1]) << 8) | - (float_to_ubyte(bcol->color[2]) << 0)); + nv30->blend_colour = *bcol; + nv30->dirty |= NV30_NEW_BCOL; } static void diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 9a3954594a7..541ad0da3e8 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -3,6 +3,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_framebuffer, + &nv30_state_blend_colour, NULL }; -- cgit v1.2.3 From c0e9eb3b095c9769d3deacf4ad4470bd155acdcd Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 21:25:47 +0200 Subject: nv30: Emit blend state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 6 +-- src/gallium/drivers/nv30/nv30_state.c | 84 +++++++++++++++-------------- src/gallium/drivers/nv30/nv30_state.h | 14 ----- src/gallium/drivers/nv30/nv30_state_blend.c | 2 +- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 5 files changed, 48 insertions(+), 59 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 57cc991cd7a..eedebc91b23 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -87,8 +87,7 @@ struct nv30_zsa_state { struct nouveau_stateobj *so; }; -/* TODO: rename when removing the old state emitter */ -struct nv30_blend_state_new { +struct nv30_blend_state { struct pipe_blend_state pipe; struct nouveau_stateobj *so; }; @@ -125,7 +124,7 @@ struct nv30_context { unsigned vp_samplers; /* Context state */ - struct nv30_blend_state_new *blend; + struct nv30_blend_state *blend; struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; @@ -212,6 +211,7 @@ extern void nv30_fragtex_bind(struct nv30_context *); extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_emit_hw_state(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); +extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_framebuffer; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index e67a701809a..cf46dcef936 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -9,63 +9,65 @@ static void * nv30_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { - struct nv30_blend_state *cb; - - cb = malloc(sizeof(struct nv30_blend_state)); - - cb->b_enable = cso->blend_enable ? 1 : 0; - cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | - (nvgl_blend_func(cso->rgb_src_factor))); - cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | - (nvgl_blend_func(cso->rgb_dst_factor))); - cb->b_eqn = ((nvgl_blend_eqn(cso->alpha_func) << 16) | - (nvgl_blend_eqn(cso->rgb_func))); + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_grobj *rankine = nv30->screen->rankine; + struct nv30_blend_state *bso = CALLOC(1, sizeof(*bso)); + struct nouveau_stateobj *so = so_new(16, 0); + + if (cso->blend_enable) { + so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 3); + so_data (so, 1); + so_data (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) | + nvgl_blend_func(cso->rgb_src_factor)); + so_data (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 | + nvgl_blend_func(cso->rgb_dst_factor)); + so_method(so, rankine, NV34TCL_BLEND_EQUATION, 1); + so_data (so, nvgl_blend_eqn(cso->alpha_func) << 16 | + nvgl_blend_eqn(cso->rgb_func)); + } else { + so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 1); + so_data (so, 0); + } - cb->l_enable = cso->logicop_enable ? 1 : 0; - cb->l_op = nvgl_logicop_func(cso->logicop_func); + so_method(so, rankine, NV34TCL_COLOR_MASK, 1); + so_data (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01 << 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01 << 0) : 0))); - cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | - ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | - ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | - ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + if (cso->logicop_enable) { + so_method(so, rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); + so_data (so, 1); + so_data (so, nvgl_logicop_func(cso->logicop_func)); + } else { + so_method(so, rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 1); + so_data (so, 0); + } - cb->d_enable = cso->dither ? 1 : 0; + so_method(so, rankine, NV34TCL_DITHER_ENABLE, 1); + so_data (so, cso->dither ? 1 : 0); - return (void *)cb; + so_ref(so, &bso->so); + bso->pipe = *cso; + return (void *)bso; } static void nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_blend_state *cb = hwcso; - - if (!hwcso) { - return; - } - - BEGIN_RING(rankine, NV34TCL_DITHER_ENABLE, 1); - OUT_RING (cb->d_enable); - BEGIN_RING(rankine, NV34TCL_BLEND_FUNC_ENABLE, 3); - OUT_RING (cb->b_enable); - OUT_RING (cb->b_srcfunc); - OUT_RING (cb->b_dstfunc); - BEGIN_RING(rankine, NV34TCL_BLEND_EQUATION, 1); - OUT_RING (cb->b_eqn); - - BEGIN_RING(rankine, NV34TCL_COLOR_MASK, 1); - OUT_RING (cb->c_mask); - - BEGIN_RING(rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2); - OUT_RING (cb->l_enable); - OUT_RING (cb->l_op); + nv30->blend = hwcso; + nv30->dirty |= NV30_NEW_BLEND; } static void nv30_blend_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nv30_blend_state *bso = hwcso; + + so_ref(NULL, &bso->so); + FREE(bso); } diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index 117520dd13c..8489b7b7969 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -3,20 +3,6 @@ #include "pipe/p_state.h" -struct nv30_blend_state { - uint32_t b_enable; - uint32_t b_srcfunc; - uint32_t b_dstfunc; - uint32_t b_eqn; - - uint32_t l_enable; - uint32_t l_op; - - uint32_t c_mask; - - uint32_t d_enable; -}; - struct nv30_sampler_state { uint32_t fmt; uint32_t wrap; diff --git a/src/gallium/drivers/nv30/nv30_state_blend.c b/src/gallium/drivers/nv30/nv30_state_blend.c index a1b0100472d..44d43e132a5 100644 --- a/src/gallium/drivers/nv30/nv30_state_blend.c +++ b/src/gallium/drivers/nv30/nv30_state_blend.c @@ -7,7 +7,7 @@ nv30_state_blend_validate(struct nv30_context *nv30) return TRUE; } -struct nv30_state_entry nv30_state_blend_new = { +struct nv30_state_entry nv30_state_blend = { .validate = nv30_state_blend_validate, .dirty = { .pipe = NV30_NEW_BLEND, diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 541ad0da3e8..492f4110117 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -3,6 +3,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_framebuffer, + &nv30_state_blend, &nv30_state_blend_colour, NULL }; -- cgit v1.2.3 From 360f7a3e239553fc0e1aff3b38c06c2e3d0a698c Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 21:48:18 +0200 Subject: nv30: Emit rasterizer state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 6 +- src/gallium/drivers/nv30/nv30_state.c | 177 +++++++++++++++-------------- src/gallium/drivers/nv30/nv30_state.h | 23 ---- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 4 files changed, 98 insertions(+), 109 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index eedebc91b23..692d8dd5230 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -76,8 +76,7 @@ enum nv30_state_index { #define NV30_NEW_ARRAYS (1 << 11) #define NV30_NEW_UCP (1 << 12) -/* TODO: rename when removing the old state emitter */ -struct nv30_rasterizer_state_new { +struct nv30_rasterizer_state { struct pipe_rasterizer_state pipe; struct nouveau_stateobj *so; }; @@ -124,11 +123,11 @@ struct nv30_context { unsigned vp_samplers; /* Context state */ + struct nv30_rasterizer_state *rasterizer; struct nv30_blend_state *blend; struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; - struct nv30_rasterizer_state_new *rasterizer; struct nv30_zsa_state *zsa; unsigned stipple[32]; @@ -211,6 +210,7 @@ extern void nv30_fragtex_bind(struct nv30_context *); extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_emit_hw_state(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); +extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_framebuffer; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index cf46dcef936..8c7a902e28f 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -275,123 +275,134 @@ static void * nv30_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { - struct nv30_rasterizer_state *rs; - int i; + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_rasterizer_state *rsso = CALLOC(1, sizeof(*rsso)); + struct nouveau_stateobj *so = so_new(32, 0); + struct nouveau_grobj *rankine = nv30->screen->rankine; /*XXX: ignored: * light_twoside - * offset_cw/ccw -nohw - * scissor * point_smooth -nohw * multisample - * offset_units / offset_scale */ - rs = malloc(sizeof(struct nv30_rasterizer_state)); - - rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; - rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; - rs->line_smooth_en = cso->line_smooth ? 1 : 0; - rs->line_stipple_en = cso->line_stipple_enable ? 1 : 0; - rs->line_stipple = (cso->line_stipple_pattern << 16) | - cso->line_stipple_factor; + so_method(so, rankine, NV34TCL_SHADE_MODEL, 1); + so_data (so, cso->flatshade ? NV34TCL_SHADE_MODEL_FLAT : + NV34TCL_SHADE_MODEL_SMOOTH); - rs->point_size = *(uint32_t*)&cso->point_size; + so_method(so, rankine, NV34TCL_LINE_WIDTH, 2); + so_data (so, (unsigned char)(cso->line_width * 8.0) & 0xff); + so_data (so, cso->line_smooth ? 1 : 0); + so_method(so, rankine, NV34TCL_LINE_STIPPLE_ENABLE, 2); + so_data (so, cso->line_stipple_enable ? 1 : 0); + so_data (so, (cso->line_stipple_pattern << 16) | + cso->line_stipple_factor); - rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; - rs->poly_stipple_en = cso->poly_stipple_enable ? 1 : 0; + so_method(so, rankine, NV34TCL_POINT_SIZE, 1); + so_data (so, fui(cso->point_size)); + so_method(so, rankine, NV34TCL_POLYGON_MODE_FRONT, 6); if (cso->front_winding == PIPE_WINDING_CCW) { - rs->front_face = NV34TCL_FRONT_FACE_CCW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV34TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_CW: + so_data(so, NV34TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV34TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, NV34TCL_CULL_FACE_BACK); + break; + } + so_data(so, NV34TCL_FRONT_FACE_CCW); } else { - rs->front_face = NV34TCL_FRONT_FACE_CW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + so_data(so, nvgl_polygon_mode(cso->fill_cw)); + so_data(so, nvgl_polygon_mode(cso->fill_ccw)); + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + so_data(so, NV34TCL_CULL_FACE_BACK); + break; + case PIPE_WINDING_CW: + so_data(so, NV34TCL_CULL_FACE_FRONT); + break; + case PIPE_WINDING_BOTH: + so_data(so, NV34TCL_CULL_FACE_FRONT_AND_BACK); + break; + default: + so_data(so, NV34TCL_CULL_FACE_BACK); + break; + } + so_data(so, NV34TCL_FRONT_FACE_CW); } - - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CCW) - rs->cull_face = NV34TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV34TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_CW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CW) - rs->cull_face = NV34TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV34TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_BOTH: - rs->cull_face_en = 1; - rs->cull_face = NV34TCL_CULL_FACE_FRONT_AND_BACK; - break; - case PIPE_WINDING_NONE: - default: - rs->cull_face_en = 0; - rs->cull_face = 0; - break; + so_data(so, cso->poly_smooth ? 1 : 0); + so_data(so, (cso->cull_mode != PIPE_WINDING_NONE) ? 1 : 0); + + so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); + so_data (so, cso->poly_stipple_enable ? 1 : 0); + + so_method(so, rankine, NV34TCL_POLYGON_OFFSET_POINT_ENABLE, 3); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT)) + so_data(so, 1); + else + so_data(so, 0); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_LINE) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_LINE)) + so_data(so, 1); + else + so_data(so, 0); + if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_FILL) || + (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_FILL)) + so_data(so, 1); + else + so_data(so, 0); + if (cso->offset_cw || cso->offset_ccw) { + so_method(so, rankine, NV34TCL_POLYGON_OFFSET_FACTOR, 2); + so_data (so, fui(cso->offset_scale)); + so_data (so, fui(cso->offset_units * 2)); } + so_method(so, rankine, NV34TCL_POINT_SPRITE, 1); if (cso->point_sprite) { - rs->point_sprite = (1 << 0); + unsigned psctl = (1 << 0), i; + for (i = 0; i < 8; i++) { if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) - rs->point_sprite |= (1 << (8 + i)); + psctl |= (1 << (8 + i)); } + + so_data(so, psctl); } else { - rs->point_sprite = 0; + so_data(so, 0); } - return (void *)rs; + so_ref(so, &rsso->so); + rsso->pipe = *cso; + return (void *)rsso; } static void nv30_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_rasterizer_state *rs = hwcso; - if (!hwcso) { - return; - } - - BEGIN_RING(rankine, NV34TCL_SHADE_MODEL, 1); - OUT_RING (rs->shade_model); - - BEGIN_RING(rankine, NV34TCL_LINE_WIDTH, 2); - OUT_RING (rs->line_width); - OUT_RING (rs->line_smooth_en); - BEGIN_RING(rankine, NV34TCL_LINE_STIPPLE_ENABLE, 2); - OUT_RING (rs->line_stipple_en); - OUT_RING (rs->line_stipple); - - BEGIN_RING(rankine, NV34TCL_POINT_SIZE, 1); - OUT_RING (rs->point_size); - - BEGIN_RING(rankine, NV34TCL_POLYGON_MODE_FRONT, 6); - OUT_RING (rs->poly_mode_front); - OUT_RING (rs->poly_mode_back); - OUT_RING (rs->cull_face); - OUT_RING (rs->front_face); - OUT_RING (rs->poly_smooth_en); - OUT_RING (rs->cull_face_en); - - BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1); - OUT_RING (rs->poly_stipple_en); - - BEGIN_RING(rankine, NV34TCL_POINT_SPRITE, 1); - OUT_RING (rs->point_sprite); + nv30->rasterizer = hwcso; + nv30->dirty |= NV30_NEW_RAST; + /*nv30->draw_dirty |= NV30_NEW_RAST;*/ } static void nv30_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nv30_rasterizer_state *rsso = hwcso; + + so_ref(NULL, &rsso->so); + FREE(rsso); } static void diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index 8489b7b7969..b33075aceb6 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -11,29 +11,6 @@ struct nv30_sampler_state { uint32_t bcol; }; -struct nv30_rasterizer_state { - uint32_t shade_model; - - uint32_t line_width; - uint32_t line_smooth_en; - uint32_t line_stipple_en; - uint32_t line_stipple; - - uint32_t point_size; - - uint32_t poly_smooth_en; - uint32_t poly_stipple_en; - - uint32_t poly_mode_front; - uint32_t poly_mode_back; - - uint32_t front_face; - uint32_t cull_face; - uint32_t cull_face_en; - - uint32_t point_sprite; -}; - struct nv30_vertex_program_exec { uint32_t data[4]; boolean has_branch_offset; diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 492f4110117..7d77adbfdf0 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -3,6 +3,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_framebuffer, + &nv30_state_rasterizer, &nv30_state_blend, &nv30_state_blend_colour, NULL -- cgit v1.2.3 From c66f376e271427799f777c39bc9221df7c961f77 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 22:10:53 +0200 Subject: nv30: Emit depth/stencil/alpha state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 1 + src/gallium/drivers/nv30/nv30_state.c | 91 +++++++++++++++++------------- src/gallium/drivers/nv30/nv30_state.h | 30 ---------- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 4 files changed, 53 insertions(+), 70 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 692d8dd5230..b8a26712a8b 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -213,6 +213,7 @@ extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; +extern struct nv30_state_entry nv30_state_zsa; extern struct nv30_state_entry nv30_state_framebuffer; /* nv30_vbo.c */ diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 8c7a902e28f..8121e62fb92 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -405,65 +405,76 @@ nv30_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) FREE(rsso); } -static void -nv30_translate_stencil(const struct pipe_depth_stencil_alpha_state *cso, - unsigned idx, struct nv30_stencil_push *hw) -{ - hw->enable = cso->stencil[idx].enabled ? 1 : 0; - hw->wmask = cso->stencil[idx].write_mask; - hw->func = nvgl_comparison_op(cso->stencil[idx].func); - hw->ref = cso->stencil[idx].ref_value; - hw->vmask = cso->stencil[idx].value_mask; - hw->fail = nvgl_stencil_op(cso->stencil[idx].fail_op); - hw->zfail = nvgl_stencil_op(cso->stencil[idx].zfail_op); - hw->zpass = nvgl_stencil_op(cso->stencil[idx].zpass_op); -} - static void * nv30_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { - struct nv30_depth_stencil_alpha_state *hw; - - hw = malloc(sizeof(struct nv30_depth_stencil_alpha_state)); - - hw->depth.func = nvgl_comparison_op(cso->depth.func); - hw->depth.write_enable = cso->depth.writemask ? 1 : 0; - hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + struct nv30_context *nv30 = nv30_context(pipe); + struct nv30_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso)); + struct nouveau_stateobj *so = so_new(32, 0); + struct nouveau_grobj *rankine = nv30->screen->rankine; - nv30_translate_stencil(cso, 0, &hw->stencil.front); - nv30_translate_stencil(cso, 1, &hw->stencil.back); + so_method(so, rankine, NV34TCL_DEPTH_FUNC, 3); + so_data (so, nvgl_comparison_op(cso->depth.func)); + so_data (so, cso->depth.writemask ? 1 : 0); + so_data (so, cso->depth.enabled ? 1 : 0); + + so_method(so, rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3); + so_data (so, cso->alpha.enabled ? 1 : 0); + so_data (so, nvgl_comparison_op(cso->alpha.func)); + so_data (so, float_to_ubyte(cso->alpha.ref)); + + if (cso->stencil[0].enabled) { + so_method(so, rankine, NV34TCL_STENCIL_FRONT_ENABLE, 8); + so_data (so, cso->stencil[0].enabled ? 1 : 0); + so_data (so, cso->stencil[0].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[0].func)); + so_data (so, cso->stencil[0].ref_value); + so_data (so, cso->stencil[0].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[0].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[0].zpass_op)); + } else { + so_method(so, rankine, NV34TCL_STENCIL_FRONT_ENABLE, 1); + so_data (so, 0); + } - hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; - hw->alpha.func = nvgl_comparison_op(cso->alpha.func); - hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + if (cso->stencil[1].enabled) { + so_method(so, rankine, NV34TCL_STENCIL_BACK_ENABLE, 8); + so_data (so, cso->stencil[1].enabled ? 1 : 0); + so_data (so, cso->stencil[1].write_mask); + so_data (so, nvgl_comparison_op(cso->stencil[1].func)); + so_data (so, cso->stencil[1].ref_value); + so_data (so, cso->stencil[1].value_mask); + so_data (so, nvgl_stencil_op(cso->stencil[1].fail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zfail_op)); + so_data (so, nvgl_stencil_op(cso->stencil[1].zpass_op)); + } else { + so_method(so, rankine, NV34TCL_STENCIL_BACK_ENABLE, 1); + so_data (so, 0); + } - return (void *)hw; + so_ref(so, &zsaso->so); + zsaso->pipe = *cso; + return (void *)zsaso; } static void nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_depth_stencil_alpha_state *hw = hwcso; - - if (!hwcso) { - return; - } - BEGIN_RING(rankine, NV34TCL_DEPTH_FUNC, 3); - OUT_RINGp ((uint32_t *)&hw->depth, 3); - BEGIN_RING(rankine, NV34TCL_STENCIL_BACK_ENABLE, 16); - OUT_RINGp ((uint32_t *)&hw->stencil.back, 8); - OUT_RINGp ((uint32_t *)&hw->stencil.front, 8); - BEGIN_RING(rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3); - OUT_RINGp ((uint32_t *)&hw->alpha.enabled, 3); + nv30->zsa = hwcso; + nv30->dirty |= NV30_NEW_ZSA; } static void nv30_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + struct nv30_zsa_state *zsaso = hwcso; + + so_ref(NULL, &zsaso->so); + FREE(zsaso); } static void * diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index b33075aceb6..c65a937467e 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -65,36 +65,6 @@ struct nv30_fragment_program { uint32_t fp_reg_control; }; -struct nv30_stencil_push { - uint32_t enable; - uint32_t wmask; - uint32_t func; - uint32_t ref; - uint32_t vmask; - uint32_t fail; - uint32_t zfail; - uint32_t zpass; -}; - -struct nv30_depth_stencil_alpha_state { - struct { - uint32_t func; - uint32_t write_enable; - uint32_t test_enable; - } depth; - - struct { - struct nv30_stencil_push back; - struct nv30_stencil_push front; - } stencil; - - struct { - uint32_t enabled; - uint32_t func; - uint32_t ref; - } alpha; -}; - struct nv30_miptree { struct pipe_texture base; diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 7d77adbfdf0..83bb3408a5a 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -6,6 +6,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_rasterizer, &nv30_state_blend, &nv30_state_blend_colour, + &nv30_state_zsa, NULL }; -- cgit v1.2.3 From f1d24c1d27255e4ff5ba451e6d58558f9ccdc801 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 22:18:38 +0200 Subject: nv30: Emit scissor state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 1 + src/gallium/drivers/nv30/nv30_state.c | 5 ++--- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 3 files changed, 4 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index b8a26712a8b..2a48903e32a 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -211,6 +211,7 @@ extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_emit_hw_state(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; +extern struct nv30_state_entry nv30_state_scissor; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_zsa; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 8121e62fb92..f2413a3372c 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -607,9 +607,8 @@ nv30_set_scissor_state(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); - BEGIN_RING(rankine, NV34TCL_SCISSOR_HORIZ, 2); - OUT_RING (((s->maxx - s->minx) << 16) | s->minx); - OUT_RING (((s->maxy - s->miny) << 16) | s->miny); + nv30->scissor = *s; + nv30->dirty |= NV30_NEW_SCISSOR; } static void diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 83bb3408a5a..f45a340b5d6 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -4,6 +4,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_framebuffer, &nv30_state_rasterizer, + &nv30_state_scissor, &nv30_state_blend, &nv30_state_blend_colour, &nv30_state_zsa, -- cgit v1.2.3 From e7e231a5116aed9f1ca685a297032a3e3e6a2433 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 22:31:23 +0200 Subject: nv30: Emit polygon stipple state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 10 +++++----- src/gallium/drivers/nv30/nv30_state.c | 4 ++-- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 3 files changed, 8 insertions(+), 7 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 2a48903e32a..ed7b0759ad4 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -112,9 +112,6 @@ struct nv30_context { /* HW state derived from pipe states */ struct nv30_state state; - struct pipe_scissor_state scissor; - - uint32_t dirty; struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; @@ -123,13 +120,15 @@ struct nv30_context { unsigned vp_samplers; /* Context state */ + unsigned dirty; + struct pipe_scissor_state scissor; + unsigned stipple[32]; struct nv30_rasterizer_state *rasterizer; + struct nv30_zsa_state *zsa; struct nv30_blend_state *blend; struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; - struct nv30_zsa_state *zsa; - unsigned stipple[32]; uint32_t rt_enable; struct pipe_buffer *rt[2]; @@ -212,6 +211,7 @@ extern void nv30_emit_hw_state(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; +extern struct nv30_state_entry nv30_state_stipple; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_zsa; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index f2413a3372c..7109deae552 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -597,8 +597,8 @@ nv30_set_polygon_stipple(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); - BEGIN_RING(rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32); - OUT_RINGp ((uint32_t *)stipple->stipple, 32); + memcpy(nv30->stipple, stipple->stipple, 4 * 32); + nv30->dirty |= NV30_NEW_STIPPLE; } static void diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index f45a340b5d6..4f9079184d4 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -5,6 +5,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_framebuffer, &nv30_state_rasterizer, &nv30_state_scissor, + &nv30_state_stipple, &nv30_state_blend, &nv30_state_blend_colour, &nv30_state_zsa, -- cgit v1.2.3 From 568b477b9c118e5ace831b8ecf1811da1c0961cd Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 22:42:30 +0200 Subject: nv30: Emit viewport state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 1 + src/gallium/drivers/nv30/nv30_state.c | 12 +++--------- src/gallium/drivers/nv30/nv30_state_emit.c | 1 + 3 files changed, 5 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index ed7b0759ad4..0ad1dc6f912 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -215,6 +215,7 @@ extern struct nv30_state_entry nv30_state_stipple; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_zsa; +extern struct nv30_state_entry nv30_state_viewport; extern struct nv30_state_entry nv30_state_framebuffer; /* nv30_vbo.c */ diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 7109deae552..e65c4b215d4 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -617,15 +617,9 @@ nv30_set_viewport_state(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); - BEGIN_RING(rankine, NV34TCL_VIEWPORT_TRANSLATE_X, 8); - OUT_RINGf (vpt->translate[0]); - OUT_RINGf (vpt->translate[1]); - OUT_RINGf (vpt->translate[2]); - OUT_RINGf (vpt->translate[3]); - OUT_RINGf (vpt->scale[0]); - OUT_RINGf (vpt->scale[1]); - OUT_RINGf (vpt->scale[2]); - OUT_RINGf (vpt->scale[3]); + nv30->viewport = *vpt; + nv30->dirty |= NV30_NEW_VIEWPORT; + /*nv30->draw_dirty |= NV30_NEW_VIEWPORT;*/ } static void diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 4f9079184d4..cb7f1a7c310 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -9,6 +9,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_blend, &nv30_state_blend_colour, &nv30_state_zsa, + &nv30_state_viewport, NULL }; -- cgit v1.2.3 From 6f56b527d866506a323feb19f9d8529d40034af2 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 3 Jul 2008 22:47:15 +0200 Subject: nv30: Reemit state when changing context --- src/gallium/drivers/nv30/nv30_state_emit.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index cb7f1a7c310..eca1f0652cc 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -46,9 +46,19 @@ void nv30_emit_hw_state(struct nv30_context *nv30) { struct nv30_state *state = &nv30->state; + struct nv30_screen *screen = nv30->screen; unsigned i; uint64 states; + if (nv30->pctx_id != screen->cur_pctx) { + for (i = 0; i < NV30_STATE_MAX; i++) { + if (state->hw[i] && screen->state[i] != state->hw[i]) + state->dirty |= (1ULL << i); + } + + screen->cur_pctx = nv30->pctx_id; + } + for (i = 0, states = state->dirty; states; i++) { if (!(states & (1ULL << i))) continue; -- cgit v1.2.3 From f99643ca6ea3aa05a0b16dc5d99e11fa00185684 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 4 Jul 2008 18:53:44 +0200 Subject: nv30: Emit sampler state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 3 +++ src/gallium/drivers/nv30/nv30_state.c | 42 +++++++++++++++++++++++++----- src/gallium/drivers/nv30/nv30_state_emit.c | 9 ++++++- 3 files changed, 46 insertions(+), 8 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 0ad1dc6f912..5404685da25 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -96,6 +96,7 @@ struct nv30_state { unsigned scissor_enabled; unsigned stipple_enabled; unsigned viewport_bypass; + unsigned fp_samplers; uint64_t dirty; struct nouveau_stateobj *hw[NV30_STATE_MAX]; @@ -129,6 +130,8 @@ struct nv30_context { struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; + unsigned nr_samplers; + unsigned nr_textures; uint32_t rt_enable; struct pipe_buffer *rt[2]; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index e65c4b215d4..97becd2be37 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -1,6 +1,7 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" +#include "pipe/p_inlines.h" #include "nv30_context.h" #include "nv30_state.h" @@ -116,7 +117,7 @@ nv30_sampler_state_create(struct pipe_context *pipe, struct nv30_sampler_state *ps; uint32_t filter = 0; - ps = malloc(sizeof(struct nv30_sampler_state)); + ps = MALLOC(sizeof(struct nv30_sampler_state)); ps->fmt = 0; if (!cso->normalized_coords) @@ -197,6 +198,19 @@ nv30_sampler_state_create(struct pipe_context *pipe, ps->filt = filter; + /*{ + float limit; + + limit = CLAMP(cso->lod_bias, -16.0, 15.0); + ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff; + + limit = CLAMP(cso->max_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 7; + + limit = CLAMP(cso->min_lod, 0.0, 15.0); + ps->en |= (int)(limit * 256.0) << 19; + }*/ + /* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { switch (cso->compare_func) { case PIPE_FUNC_NEVER: @@ -242,20 +256,24 @@ nv30_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) struct nv30_context *nv30 = nv30_context(pipe); unsigned unit; - if (!sampler) { - return; - } - for (unit = 0; unit < nr; unit++) { nv30->tex_sampler[unit] = sampler[unit]; nv30->dirty_samplers |= (1 << unit); } + + for (unit = nr; unit < nv30->nr_samplers; unit++) { + nv30->tex_sampler[unit] = NULL; + nv30->dirty_samplers |= (1 << unit); + } + + nv30->nr_samplers = nr; + nv30->dirty |= NV30_NEW_SAMPLER; } static void nv30_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { - free(hwcso); + FREE(hwcso); } static void @@ -266,9 +284,19 @@ nv30_set_sampler_texture(struct pipe_context *pipe, unsigned nr, unsigned unit; for (unit = 0; unit < nr; unit++) { - nv30->tex_miptree[unit] = (struct nv30_miptree *)miptree[unit]; + pipe_texture_reference((struct pipe_texture **) + &nv30->tex_miptree[unit], miptree[unit]); nv30->dirty_samplers |= (1 << unit); } + + for (unit = nr; unit < nv30->nr_textures; unit++) { + pipe_texture_reference((struct pipe_texture **) + &nv30->tex_miptree[unit], NULL); + nv30->dirty_samplers |= (1 << unit); + } + + nv30->nr_textures = nr; + nv30->dirty |= NV30_NEW_SAMPLER; } static void * diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index eca1f0652cc..0ea78571972 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -47,7 +47,7 @@ nv30_emit_hw_state(struct nv30_context *nv30) { struct nv30_state *state = &nv30->state; struct nv30_screen *screen = nv30->screen; - unsigned i; + unsigned i, samplers; uint64 states; if (nv30->pctx_id != screen->cur_pctx) { @@ -91,6 +91,13 @@ nv30_emit_hw_state(struct nv30_context *nv30) nv30->dirty_samplers = 0; so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); + for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { + if (!(samplers & (1 << i))) + continue; + so_emit_reloc_markers(nv30->nvws, + state->hw[NV30_STATE_FRAGTEX0+i]); + samplers &= ~(1ULL << i); + } /* Texture images, emitted in nv30_fragtex_build */ #if 0 -- cgit v1.2.3 From 49937b99855984dd01a431c026f9308b6c0dac4f Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 3 Jul 2008 20:05:32 -0400 Subject: g3dvl: Round surfaces up to POT, use src rect when outputting surfaces. --- src/gallium/state_trackers/g3dvl/Makefile | 2 +- src/gallium/state_trackers/g3dvl/vl_context.c | 48 +++++++++++++++++++++------ src/gallium/state_trackers/g3dvl/vl_context.h | 2 +- src/gallium/state_trackers/g3dvl/vl_surface.c | 29 +++++++++++++--- src/gallium/state_trackers/g3dvl/vl_types.h | 6 ++++ src/gallium/state_trackers/g3dvl/vl_util.c | 17 ++++++++++ src/gallium/state_trackers/g3dvl/vl_util.h | 7 ++++ 7 files changed, 93 insertions(+), 18 deletions(-) create mode 100644 src/gallium/state_trackers/g3dvl/vl_util.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_util.h diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile index bd46d004e23..50e3c843b55 100644 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -1,5 +1,5 @@ TARGET = libg3dvl.a -OBJECTS = vl_context.o vl_data.o vl_surface.o vl_shader_build.o +OBJECTS = vl_context.o vl_data.o vl_surface.o vl_shader_build.o vl_util.o GALLIUMDIR = ../.. CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index bd8743dc9a5..58971bd7c79 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -11,6 +11,7 @@ #include #include "vl_shader_build.h" #include "vl_data.h" +#include "vl_util.h" static int vlInitIDCT(struct VL_CONTEXT *context) { @@ -1357,8 +1358,9 @@ static int vlInitMC(struct VL_CONTEXT *context) pipe = context->pipe; - context->states.mc.viewport.scale[0] = context->video_width; - context->states.mc.viewport.scale[1] = context->video_height; + /* For MC we render to textures, which are rounded up to nearest POT */ + context->states.mc.viewport.scale[0] = vlRoundUpPOT(context->video_width); + context->states.mc.viewport.scale[1] = vlRoundUpPOT(context->video_height); context->states.mc.viewport.scale[2] = 1; context->states.mc.viewport.scale[3] = 1; context->states.mc.viewport.translate[0] = 0; @@ -1512,6 +1514,13 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + + /* + * decl c0 ; Scaling vector to scale texcoord rect to source size + * decl c1 ; Translation vector to move texcoord rect into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* * decl o0 ; Vertex pos @@ -1522,16 +1531,18 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + + /* mov o0, i0 ; Move pos in to pos out */ + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i1, c0 ; Scale unit texcoord rect to source size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 1, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* - * mov o0, i0 ; Move pos in to pos out - * mov o1, i1 ; Move input texcoords to output - */ - for (i = 0; i < 2; i++) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } + /* add o1, t0, c1 ; Translate texcoord rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* end */ inst = vl_end(); @@ -1693,6 +1704,19 @@ static int vlCreateDataBufsCSC(struct VL_CONTEXT *context) context->states.csc.vertex_buf_elems[1].nr_components = 2; context->states.csc.vertex_buf_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; + /* + Create our vertex shader's constant buffer + Const buffer contains scaling and translation vectors + */ + context->states.csc.vs_const_buf.size = sizeof(struct VL_CSC_VS_CONSTS); + context->states.csc.vs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + context->states.csc.vs_const_buf.size + ); + /* Create our fragment shader's constant buffer Const buffer contains the color conversion matrix and bias vectors @@ -1776,6 +1800,7 @@ static int vlDestroyCSC(struct VL_CONTEXT *context) context->pipe->delete_fs_state(context->pipe, context->states.csc.fragment_shader); context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vertex_bufs[0].buffer); context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vertex_bufs[1].buffer); + context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vs_const_buf.buffer); context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.fs_const_buf.buffer); return 0; @@ -1986,6 +2011,7 @@ int vlEndRender(struct VL_CONTEXT *context) pipe->bind_fs_state(pipe, context->states.csc.fragment_shader); pipe->set_vertex_buffers(pipe, 2, context->states.csc.vertex_bufs); pipe->set_vertex_elements(pipe, 2, context->states.csc.vertex_buf_elems); + pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &context->states.csc.vs_const_buf); pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &context->states.csc.fs_const_buf); return 0; diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h index 8a123180734..9ebda21a1cc 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.h +++ b/src/gallium/state_trackers/g3dvl/vl_context.h @@ -49,7 +49,7 @@ struct VL_CONTEXT struct pipe_shader_state *vertex_shader, *fragment_shader; struct pipe_vertex_buffer vertex_bufs[2]; struct pipe_vertex_element vertex_buf_elems[2]; - struct pipe_constant_buffer fs_const_buf; + struct pipe_constant_buffer vs_const_buf, fs_const_buf; } csc; } states; }; diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 68313cc7505..13f7301f07b 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -7,6 +7,7 @@ #include #include "vl_context.h" #include "vl_defs.h" +#include "vl_util.h" static int vlGrabFrameCodedFullBlock(short *src, short *dst, unsigned int dst_pitch) { @@ -194,9 +195,6 @@ static int vlGrabBlocks pipe_surface_unmap(tex_surface); } - /* XXX: Texture cache is not invalidated when texture contents change */ - context->pipe->flush(context->pipe, PIPE_FLUSH_TEXTURE_CACHE, NULL); - return 0; } @@ -214,8 +212,8 @@ int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface) sfc = calloc(1, sizeof(struct VL_SURFACE)); sfc->context = context; - sfc->width = context->video_width; - sfc->height = context->video_height; + sfc->width = vlRoundUpPOT(context->video_width); + sfc->height = vlRoundUpPOT(context->video_height); sfc->format = context->video_format; memset(&template, 0, sizeof(struct pipe_texture)); @@ -227,6 +225,7 @@ int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface) template.depth[0] = 1; template.compressed = 0; pf_get_block(template.format, &template.block); + /* XXX: Needed? */ template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; sfc->texture = pipe->screen->texture_create(pipe->screen, &template); @@ -517,6 +516,7 @@ int vlPutSurface { unsigned int create_fb = 0; struct pipe_context *pipe; + struct VL_CSC_VS_CONSTS *vs_consts; assert(surface); @@ -568,9 +568,28 @@ int vlPutSurface vlEndRender(surface->context); + vs_consts = pipe->winsys->buffer_map + ( + pipe->winsys, + surface->context->states.csc.vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vs_consts->src_scale.x = srcw / (float)surface->width; + vs_consts->src_scale.y = srch / (float)surface->height; + vs_consts->src_scale.z = 1; + vs_consts->src_scale.w = 1; + vs_consts->src_trans.x = srcx / (float)surface->width; + vs_consts->src_trans.y = srcy / (float)surface->height; + vs_consts->src_trans.z = 0; + vs_consts->src_trans.w = 0; + + pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.csc.vs_const_buf.buffer); + pipe->set_sampler_textures(pipe, 1, &surface->texture); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + /* XXX: Need to take destx, desty into consideration */ pipe->winsys->flush_frontbuffer ( pipe->winsys, diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h index 97753699db6..4d210c9e0aa 100644 --- a/src/gallium/state_trackers/g3dvl/vl_types.h +++ b/src/gallium/state_trackers/g3dvl/vl_types.h @@ -75,6 +75,12 @@ struct VL_MC_FS_CONSTS struct VL_VERTEX4F y_divider; }; +struct VL_CSC_VS_CONSTS +{ + struct VL_VERTEX4F src_scale; + struct VL_VERTEX4F src_trans; +}; + struct VL_CSC_FS_CONSTS { struct VL_VERTEX4F bias; diff --git a/src/gallium/state_trackers/g3dvl/vl_util.c b/src/gallium/state_trackers/g3dvl/vl_util.c new file mode 100644 index 00000000000..2421ae22101 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_util.c @@ -0,0 +1,17 @@ +#include "vl_util.h" +#include + +unsigned int vlRoundUpPOT(unsigned int x) +{ + unsigned int i; + + assert(x > 0); + + --x; + + for (i = 1; i < sizeof(unsigned int) * 8; i <<= 1) + x |= x >> i; + + return x + 1; +} + diff --git a/src/gallium/state_trackers/g3dvl/vl_util.h b/src/gallium/state_trackers/g3dvl/vl_util.h new file mode 100644 index 00000000000..e4b72c4f870 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_util.h @@ -0,0 +1,7 @@ +#ifndef vl_util_h +#define vl_util_h + +unsigned int vlRoundUpPOT(unsigned int x); + +#endif + -- cgit v1.2.3 From 6235141fd2c7af21c2b41ca66f06abc3cb0bbc24 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 6 Jul 2008 22:04:29 -0400 Subject: g3dvl: IDCT part 1. Very basic IDCT support is in, performed CPU-side for now. --- src/gallium/state_trackers/g3dvl/vl_context.c | 241 +++++++++++++++++++++++++- src/gallium/state_trackers/g3dvl/vl_context.h | 10 ++ src/gallium/state_trackers/g3dvl/vl_surface.c | 136 ++++++++++++++- src/libXvMC/surface.c | 2 + 4 files changed, 381 insertions(+), 8 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 58971bd7c79..1668ad1651b 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -11,22 +11,262 @@ #include #include "vl_shader_build.h" #include "vl_data.h" +#include "vl_defs.h" #include "vl_util.h" +static int vlCreateVertexShaderFrameIDCT(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 50; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move pos in to pos out + * mov o1, i1 ; Move texcoord in to texcoord out */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + //context->states.idct.frame_vs = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFrameIDCT(struct VL_CONTEXT *context) +{ + const unsigned int max_tokens = 50; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = context->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* decl i0 ; Texcoords for s0 */ + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl s0 ; Sampler for tex containing picture to display */ + decl = vl_decl_samplers(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* tex2d t0, i0, s0 ; Read src pixel */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* sub t0, t0, c0 ; Subtract bias vector from pixel */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix + * dp4 o0.y, t0, c2 + * dp4 o0.z, t0, c3 + * dp4 o0.w, t0, c4 ; XXX: Don't need 4th coefficient + */ + for (i = 0; i < 4; ++i) + { + inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + //context->states.idct.frame_fs = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + static int vlInitIDCT(struct VL_CONTEXT *context) { + struct pipe_context *pipe; + struct pipe_sampler_state sampler; + struct pipe_texture template; + unsigned int i; + assert(context); + pipe = context->pipe; + + context->states.idct.viewport.scale[0] = VL_BLOCK_WIDTH; + context->states.idct.viewport.scale[1] = VL_BLOCK_HEIGHT; + context->states.idct.viewport.scale[2] = 1; + context->states.idct.viewport.scale[3] = 1; + context->states.idct.viewport.translate[0] = 0; + context->states.idct.viewport.translate[1] = 0; + context->states.idct.viewport.translate[2] = 0; + context->states.idct.viewport.translate[3] = 0; + + context->states.idct.render_target.width = VL_BLOCK_WIDTH; + context->states.idct.render_target.height = VL_BLOCK_HEIGHT; + context->states.idct.render_target.num_cbufs = 1; + context->states.idct.render_target.zsbuf = NULL; + + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = PIPE_TEX_FILTER_NEAREST; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = PIPE_TEX_FILTER_NEAREST; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.shadow_ambient = ;*/ + /*sampler.lod_bias = ;*/ + sampler.min_lod = 0; + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + context->states.idct.sampler = pipe->create_sampler_state(pipe, &sampler); + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + template.format = PIPE_FORMAT_A8L8_UNORM; + template.last_level = 0; + template.width[0] = 8; + template.height[0] = 8; + template.depth[0] = 1; + template.compressed = 0; + pf_get_block(template.format, &template.block); + + context->states.idct.texture = pipe->screen->texture_create(pipe->screen, &template); + + template.format = PIPE_FORMAT_A8R8G8B8_UNORM; + template.width[0] = 16; + template.height[0] = 1; + context->states.idct.basis = pipe->screen->texture_create(pipe->screen, &template); + + for (i = 0; i < 2; ++i) + { + context->states.idct.vertex_bufs[i] = &context->states.csc.vertex_bufs[i]; + context->states.idct.vertex_buf_elems[i] = &context->states.csc.vertex_buf_elems[i]; + /* + context->states.idct.vertex_bufs[i].pitch = sizeof(struct VL_VERTEX2F); + context->states.idct.vertex_bufs[i].max_index = 3; + context->states.idct.vertex_bufs[i].buffer_offset = 0; + context->states.idct.vertex_bufs[i].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct VL_VERTEX2F) * 4 + ); + + context->states.idct.vertex_buf_elems[i].src_offset = 0; + context->states.idct.vertex_buf_elems[i].vertex_buffer_index = i; + context->states.idct.vertex_buf_elems[i].nr_components = 2; + context->states.idct.vertex_buf_elems[i].src_format = PIPE_FORMAT_R32G32_FLOAT; + */ + } + + vlCreateVertexShaderFrameIDCT(context); + vlCreateFragmentShaderFrameIDCT(context); return 0; } static int vlDestroyIDCT(struct VL_CONTEXT *context) { + //unsigned int i; + assert(context); + context->pipe->delete_sampler_state(context->pipe, context->states.idct.sampler); + + //for (i = 0; i < 2; ++i) + //context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.idct.vertex_bufs[i].buffer); + + pipe_texture_release(&context->states.idct.texture); + pipe_texture_release(&context->states.idct.basis); + + //context->pipe->delete_vs_state(context->pipe, context->states.idct.frame_vs); + //context->pipe->delete_fs_state(context->pipe, context->states.idct.frame_fs); + //context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.idct.vs_const_buf.buffer); + //context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.idct.fs_const_buf.buffer); return 0; } @@ -1271,7 +1511,6 @@ int vlCreateDataBufsMC(struct VL_CONTEXT *context) context->states.mc.vertex_buf_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; /* Create our texcoord buffers and texcoord buffer elements */ - /* TODO: Should be able to use 1 texcoord buf for chroma textures, 1 buf for ref surfaces */ for (i = 1; i < 3; ++i) { context->states.mc.vertex_bufs[i].pitch = sizeof(struct VL_TEXCOORD2F); diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h index 9ebda21a1cc..bff318854aa 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.h +++ b/src/gallium/state_trackers/g3dvl/vl_context.h @@ -26,6 +26,16 @@ struct VL_CONTEXT struct { + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state render_target; + struct pipe_sampler_state *sampler; + struct pipe_texture *texture; + struct pipe_texture *basis; + struct pipe_shader_state *frame_vs; + struct pipe_shader_state *frame_fs; + struct pipe_vertex_buffer *vertex_bufs[2]; + struct pipe_vertex_element *vertex_buf_elems[2]; + //struct pipe_constant_buffer vs_const_buf, fs_const_buf; } idct; struct diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 13f7301f07b..145ea32892a 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -9,6 +9,59 @@ #include "vl_defs.h" #include "vl_util.h" +static int vlTransformBlock(short *src, short *dst, short bias) +{ + static const float basis[8][8] = + { + {0.3536, 0.4904, 0.4619, 0.4157, 0.3536, 0.2778, 0.1913, 0.0975}, + {0.3536, 0.4157, 0.1913, -0.0975, -0.3536, -0.4904, -0.4619, -0.2778}, + {0.3536, 0.2778, -0.1913, -0.4904, -0.3536, 0.0975, 0.4619, 0.4157}, + {0.3536, 0.0975, -0.4619, -0.2778, 0.3536, 0.4157, -0.1913, -0.4904}, + {0.3536, -0.0975, -0.4619, 0.2778, 0.3536, -0.4157, -0.1913, 0.4904}, + {0.3536, -0.2778, -0.1913, 0.4904, -0.3536, -0.0975, 0.4619, -0.4157}, + {0.3536, -0.4157, 0.1913, 0.0975, -0.3536, 0.4904, -0.4619, 0.2778}, + {0.3536, -0.4904, 0.4619, -0.4157, 0.3536, -0.2778, 0.1913, -0.0975} + }; + + unsigned int x, y; + short tmp[64]; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + tmp[y * VL_BLOCK_WIDTH + x] = (short) + ( + src[y * VL_BLOCK_WIDTH + 0] * basis[x][0] + + src[y * VL_BLOCK_WIDTH + 1] * basis[x][1] + + src[y * VL_BLOCK_WIDTH + 2] * basis[x][2] + + src[y * VL_BLOCK_WIDTH + 3] * basis[x][3] + + src[y * VL_BLOCK_WIDTH + 4] * basis[x][4] + + src[y * VL_BLOCK_WIDTH + 5] * basis[x][5] + + src[y * VL_BLOCK_WIDTH + 6] * basis[x][6] + + src[y * VL_BLOCK_WIDTH + 7] * basis[x][7] + ); + + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + { + dst[y * VL_BLOCK_WIDTH + x] = bias + (short) + ( + tmp[0 * VL_BLOCK_WIDTH + x] * basis[y][0] + + tmp[1 * VL_BLOCK_WIDTH + x] * basis[y][1] + + tmp[2 * VL_BLOCK_WIDTH + x] * basis[y][2] + + tmp[3 * VL_BLOCK_WIDTH + x] * basis[y][3] + + tmp[4 * VL_BLOCK_WIDTH + x] * basis[y][4] + + tmp[5 * VL_BLOCK_WIDTH + x] * basis[y][5] + + tmp[6 * VL_BLOCK_WIDTH + x] * basis[y][6] + + tmp[7 * VL_BLOCK_WIDTH + x] * basis[y][7] + ); + if (dst[y * VL_BLOCK_WIDTH + x] > 255) + dst[y * VL_BLOCK_WIDTH + x] = 255; + else if (bias > 0 && dst[y * VL_BLOCK_WIDTH + x] < 0) + dst[y * VL_BLOCK_WIDTH + x] = 0; + } + return 0; +} + static int vlGrabFrameCodedFullBlock(short *src, short *dst, unsigned int dst_pitch) { unsigned int y; @@ -102,6 +155,9 @@ static int vlGrabBlocks unsigned int tex_pitch; unsigned int tb, sb = 0; + const int do_idct = 1; + short temp_block[64]; + assert(context); assert(blocks); @@ -121,6 +177,17 @@ static int vlGrabBlocks { if (dct_type == VL_DCT_FRAME_CODED) if (sample_type == VL_FULL_SAMPLE) + if (do_idct) + { + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); + vlGrabFrameCodedFullBlock + ( + temp_block, + texels + tb * tex_pitch * VL_BLOCK_HEIGHT, + tex_pitch + ); + } + else vlGrabFrameCodedFullBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, @@ -128,6 +195,17 @@ static int vlGrabBlocks tex_pitch ); else + if (do_idct) + { + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + vlGrabFrameCodedDiffBlock + ( + temp_block, + texels + tb * tex_pitch * VL_BLOCK_HEIGHT, + tex_pitch + ); + } + else vlGrabFrameCodedDiffBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, @@ -136,6 +214,17 @@ static int vlGrabBlocks ); else if (sample_type == VL_FULL_SAMPLE) + if (do_idct) + { + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); + vlGrabFieldCodedFullBlock + ( + temp_block, + texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, + tex_pitch + ); + } + else vlGrabFieldCodedFullBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, @@ -143,6 +232,17 @@ static int vlGrabBlocks tex_pitch ); else + if (do_idct) + { + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + vlGrabFieldCodedDiffBlock + ( + temp_block, + texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, + tex_pitch + ); + } + else vlGrabFieldCodedDiffBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, @@ -173,6 +273,17 @@ static int vlGrabBlocks if ((coded_block_pattern >> (1 - tb)) & 1) { if (sample_type == VL_FULL_SAMPLE) + if (do_idct) + { + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); + vlGrabFrameCodedFullBlock + ( + temp_block, + texels, + tex_pitch + ); + } + else vlGrabFrameCodedFullBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, @@ -180,6 +291,17 @@ static int vlGrabBlocks tex_pitch ); else + if (do_idct) + { + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + vlGrabFrameCodedDiffBlock + ( + temp_block, + texels, + tex_pitch + ); + } + else vlGrabFrameCodedDiffBlock ( blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, @@ -266,6 +388,8 @@ int vlRenderIMacroBlock if (picture_type != VL_FRAME_PICTURE) return 0; + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_FULL_SAMPLE, blocks); + pipe = surface->context->pipe; vs_consts = pipe->winsys->buffer_map @@ -298,8 +422,6 @@ int vlRenderIMacroBlock pipe->bind_vs_state(pipe, surface->context->states.mc.i_vs); pipe->bind_fs_state(pipe, surface->context->states.mc.i_fs); - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_FULL_SAMPLE, blocks); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); return 0; @@ -335,6 +457,8 @@ int vlRenderPMacroBlock if (mc_type != VL_FRAME_MC && mc_type != VL_FIELD_MC) return 0; + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); + pipe = surface->context->pipe; vs_consts = pipe->winsys->buffer_map @@ -390,8 +514,6 @@ int vlRenderPMacroBlock pipe->set_sampler_textures(pipe, 4, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 4, (void**)surface->context->states.mc.samplers); - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); return 0; @@ -428,6 +550,8 @@ int vlRenderBMacroBlock if (mc_type != VL_FRAME_MC && mc_type != VL_FIELD_MC) return 0; + vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); + pipe = surface->context->pipe; vs_consts = pipe->winsys->buffer_map @@ -492,8 +616,6 @@ int vlRenderBMacroBlock pipe->set_sampler_textures(pipe, 5, surface->context->states.mc.textures); pipe->bind_sampler_states(pipe, 5, (void**)surface->context->states.mc.samplers); - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); return 0; @@ -589,7 +711,7 @@ int vlPutSurface pipe->set_sampler_textures(pipe, 1, &surface->texture); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); - /* XXX: Need to take destx, desty into consideration */ + /* TODO: Need to take destx, desty into consideration */ pipe->winsys->flush_frontbuffer ( pipe->winsys, diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index 5656895650e..a550114655c 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -146,6 +146,8 @@ Status XvMCRenderSurface assert(flags == 0 || flags == XVMC_SECOND_FIELD); + /* TODO: Batch macroblocks by type (I,P,B) */ + for (i = first_macroblock; i < first_macroblock + num_macroblocks; ++i) if (macroblocks->macro_blocks[i].macroblock_type & XVMC_MB_TYPE_INTRA) vlRenderIMacroBlock -- cgit v1.2.3 From 52a68dd9eb1d347aa01ce09db9375793d0d0ceaf Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Tue, 8 Jul 2008 01:06:18 +0200 Subject: nv04. --- configs/default | 2 +- src/gallium/drivers/nouveau/nouveau_winsys.h | 6 + src/gallium/drivers/nv04/Makefile | 28 ++ src/gallium/drivers/nv04/nv04_clear.c | 12 + src/gallium/drivers/nv04/nv04_context.c | 107 ++++++ src/gallium/drivers/nv04/nv04_context.h | 142 ++++++++ src/gallium/drivers/nv04/nv04_fragprog.c | 22 ++ src/gallium/drivers/nv04/nv04_fragtex.c | 100 ++++++ src/gallium/drivers/nv04/nv04_miptree.c | 138 ++++++++ src/gallium/drivers/nv04/nv04_prim_vbuf.c | 312 +++++++++++++++++ src/gallium/drivers/nv04/nv04_screen.c | 216 ++++++++++++ src/gallium/drivers/nv04/nv04_screen.h | 25 ++ src/gallium/drivers/nv04/nv04_state.c | 494 +++++++++++++++++++++++++++ src/gallium/drivers/nv04/nv04_state.h | 71 ++++ src/gallium/drivers/nv04/nv04_state_emit.c | 156 +++++++++ src/gallium/drivers/nv04/nv04_surface.c | 65 ++++ src/gallium/drivers/nv04/nv04_vbo.c | 72 ++++ 17 files changed, 1967 insertions(+), 1 deletion(-) create mode 100644 src/gallium/drivers/nv04/Makefile create mode 100644 src/gallium/drivers/nv04/nv04_clear.c create mode 100644 src/gallium/drivers/nv04/nv04_context.c create mode 100644 src/gallium/drivers/nv04/nv04_context.h create mode 100644 src/gallium/drivers/nv04/nv04_fragprog.c create mode 100644 src/gallium/drivers/nv04/nv04_fragtex.c create mode 100644 src/gallium/drivers/nv04/nv04_miptree.c create mode 100644 src/gallium/drivers/nv04/nv04_prim_vbuf.c create mode 100644 src/gallium/drivers/nv04/nv04_screen.c create mode 100644 src/gallium/drivers/nv04/nv04_screen.h create mode 100644 src/gallium/drivers/nv04/nv04_state.c create mode 100644 src/gallium/drivers/nv04/nv04_state.h create mode 100644 src/gallium/drivers/nv04/nv04_state_emit.c create mode 100644 src/gallium/drivers/nv04/nv04_surface.c create mode 100644 src/gallium/drivers/nv04/nv04_vbo.c diff --git a/configs/default b/configs/default index a8d1fe6bec2..c1bc5fb01d0 100644 --- a/configs/default +++ b/configs/default @@ -70,7 +70,7 @@ PROGRAM_DIRS = demos redbook samples glsl xdemos # Gallium directories and GALLIUM_AUXILIARY_DIRS = draw cso_cache pipebuffer tgsi sct translate rtasm util GALLIUM_AUXILIARIES = $(foreach DIR,$(GALLIUM_AUXILIARY_DIRS),$(TOP)/src/gallium/auxiliary/$(DIR)/lib$(DIR).a) -GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv10 nv30 nv40 nv50 failover +GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv04 nv10 nv30 nv40 nv50 failover GALLIUM_WINSYS_COMMON_DIRS = GALLIUM_DRIVERS = $(foreach DIR,$(GALLIUM_DRIVER_DIRS),$(TOP)/src/gallium/drivers/$(DIR)/lib$(DIR).a) GALLIUM_WINSYS_DIRS = xlib egl_xlib diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 13810460bfe..48feeba309d 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -57,6 +57,12 @@ struct nouveau_winsys { unsigned, unsigned, unsigned, unsigned, unsigned); }; +extern struct pipe_screen * +nv04_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); + +extern struct pipe_context * +nv04_create(struct pipe_screen *, unsigned pctx_id); + extern struct pipe_screen * nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); diff --git a/src/gallium/drivers/nv04/Makefile b/src/gallium/drivers/nv04/Makefile new file mode 100644 index 00000000000..5ea51a2f420 --- /dev/null +++ b/src/gallium/drivers/nv04/Makefile @@ -0,0 +1,28 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv04 + +DRIVER_SOURCES = \ + nv04_clear.c \ + nv04_context.c \ + nv04_fragprog.c \ + nv04_fragtex.c \ + nv04_miptree.c \ + nv04_prim_vbuf.c \ + nv04_screen.c \ + nv04_state.c \ + nv04_state_emit.c \ + nv04_surface.c \ + nv04_vbo.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/drivers/nv04/nv04_clear.c b/src/gallium/drivers/nv04/nv04_clear.c new file mode 100644 index 00000000000..01cacd36fe1 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv04_context.h" + +void +nv04_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv04/nv04_context.c b/src/gallium/drivers/nv04/nv04_context.c new file mode 100644 index 00000000000..852a8edf5ff --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_context.c @@ -0,0 +1,107 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_util.h" + +#include "nv04_context.h" +#include "nv04_screen.h" + +static void +nv04_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + draw_flush(nv04->draw); + + FIRE_RING(fence); +} + +static void +nv04_destroy(struct pipe_context *pipe) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + if (nv04->draw) + draw_destroy(nv04->draw); + + FREE(nv04); +} + +static void +nv04_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + +static boolean +nv04_init_hwctx(struct nv04_context *nv04) +{ + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_NOTIFY, 1); + OUT_RING(0); + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_NOP, 1); + OUT_RING(0); + + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1); + OUT_RING(0x40182800); +// OUT_RING(1<<20/*no cull*/); + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_BLEND, 1); +// OUT_RING(0x24|(1<<6)|(1<<8)); + OUT_RING(0x120001a4); + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_FORMAT, 1); + OUT_RING(0x332213a1); + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_FILTER, 1); + OUT_RING(0x11001010); + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_COLORKEY, 1); + OUT_RING(0x0); +// BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_OFFSET, 1); +// OUT_RING(SCREEN_OFFSET); + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR, 1); + OUT_RING(0xff000000); + + + + FIRE_RING (NULL); + return TRUE; +} + +struct pipe_context * +nv04_create(struct pipe_screen *pscreen, unsigned pctx_id) +{ + struct nv04_screen *screen = nv04_screen(pscreen); + struct pipe_winsys *ws = pscreen->winsys; + struct nv04_context *nv04; + struct nouveau_winsys *nvws = screen->nvws; + + nv04 = CALLOC(1, sizeof(struct nv04_context)); + if (!nv04) + return NULL; + nv04->screen = screen; + nv04->pctx_id = pctx_id; + + nv04->nvws = nvws; + + nv04->pipe.winsys = ws; + nv04->pipe.screen = pscreen; + nv04->pipe.destroy = nv04_destroy; + nv04->pipe.set_edgeflags = nv04_set_edgeflags; + nv04->pipe.draw_arrays = nv04_draw_arrays; + nv04->pipe.draw_elements = nv04_draw_elements; + nv04->pipe.clear = nv04_clear; + nv04->pipe.flush = nv04_flush; + + nv04_init_surface_functions(nv04); + nv04_init_state_functions(nv04); + + nv04->draw = draw_create(); + assert(nv04->draw); + draw_wide_point_threshold(nv04->draw, 0.0); + draw_wide_line_threshold(nv04->draw, 0.0); + draw_enable_line_stipple(nv04->draw, FALSE); + draw_enable_point_sprites(nv04->draw, FALSE); + draw_set_rasterize_stage(nv04->draw, nv04_draw_vbuf_stage(nv04)); + + nv04_init_hwctx(nv04); + + return &nv04->pipe; +} + diff --git a/src/gallium/drivers/nv04/nv04_context.h b/src/gallium/drivers/nv04/nv04_context.h new file mode 100644 index 00000000000..5ba1d4ecdc1 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_context.h @@ -0,0 +1,142 @@ +#ifndef __NV04_CONTEXT_H__ +#define __NV04_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv04_screen *ctx = nv04->screen +#include "nouveau/nouveau_push.h" + +#include "nv04_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#include "nv04_screen.h" + +#define NV04_NEW_VERTPROG (1 << 1) +#define NV04_NEW_FRAGPROG (1 << 2) +#define NV04_NEW_BLEND (1 << 3) +#define NV04_NEW_RAST (1 << 4) +#define NV04_NEW_CONTROL (1 << 5) +#define NV04_NEW_VIEWPORT (1 << 6) +#define NV04_NEW_SAMPLER (1 << 7) + +struct nv04_context { + struct pipe_context pipe; + + struct nouveau_winsys *nvws; + struct nv04_screen *screen; + unsigned pctx_id; + + struct draw_context *draw; + + int chipset; + struct nouveau_notifier *sync; + + uint32_t dirty; + + struct nv04_blend_state *blend; + struct nv04_sampler_state *sampler[PIPE_MAX_SAMPLERS]; + struct nv04_fragtex_state fragtex; + struct nv04_rasterizer_state *rast; + struct nv04_depth_stencil_alpha_state *dsa; + + struct nv04_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + uint32_t rt_enable; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; + + struct { + struct pipe_buffer *buffer; + uint32_t format; + } tex[16]; + + unsigned vb_enable; + struct { + struct pipe_buffer *buffer; + unsigned delta; + } vb[16]; + + struct vertex_info vertex_info; + struct { + + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + + struct nv04_vertex_program *active; + + struct nv04_vertex_program *current; + struct pipe_buffer *constant_buf; + } vertprog; + + struct { + struct nv04_fragment_program *active; + + struct nv04_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS]; + unsigned num_vertex_buffers; + unsigned num_vertex_elements; + + struct pipe_viewport_state viewport; +}; + +static INLINE struct nv04_context * +nv04_context(struct pipe_context *pipe) +{ + return (struct nv04_context *)pipe; +} + +extern void nv04_init_state_functions(struct nv04_context *nv04); +extern void nv04_init_surface_functions(struct nv04_context *nv04); +extern void nv04_init_miptree_functions(struct pipe_screen *screen); + +/* nv04_clear.c */ +extern void nv04_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +/* nv04_draw.c */ +extern struct draw_stage *nv04_draw_render_stage(struct nv04_context *nv04); + +/* nv04_fragprog.c */ +extern void nv04_fragprog_bind(struct nv04_context *, + struct nv04_fragment_program *); +extern void nv04_fragprog_destroy(struct nv04_context *, + struct nv04_fragment_program *); + +/* nv04_fragtex.c */ +extern void nv04_fragtex_bind(struct nv04_context *); + +/* nv04_prim_vbuf.c */ +struct draw_stage *nv04_draw_vbuf_stage( struct nv04_context *nv04 ); + +/* nv04_state.c and friends */ +extern void nv04_emit_hw_state(struct nv04_context *nv04); +extern void nv04_state_tex_update(struct nv04_context *nv04); + +/* nv04_vbo.c */ +extern boolean nv04_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv04_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count); + + +#endif diff --git a/src/gallium/drivers/nv04/nv04_fragprog.c b/src/gallium/drivers/nv04/nv04_fragprog.c new file mode 100644 index 00000000000..11f4360ed80 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_fragprog.c @@ -0,0 +1,22 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" +#include "tgsi/util/tgsi_util.h" + +#include "nv04_context.h" + +void +nv04_fragprog_bind(struct nv04_context *nv04, struct nv04_fragment_program *fp) +{ +} + +void +nv04_fragprog_destroy(struct nv04_context *nv04, + struct nv04_fragment_program *fp) +{ +} + diff --git a/src/gallium/drivers/nv04/nv04_fragtex.c b/src/gallium/drivers/nv04/nv04_fragtex.c new file mode 100644 index 00000000000..3db673cd2c1 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_fragtex.c @@ -0,0 +1,100 @@ +#include "nv04_context.h" + +static INLINE int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + +#define _(m,tf) \ +{ \ + PIPE_FORMAT_##m, \ + NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_##tf, \ +} + +struct nv04_texture_format { + uint pipe; + int format; +}; + +static struct nv04_texture_format +nv04_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8), + _(X8R8G8B8_UNORM, X8R8G8B8), + _(A1R5G5B5_UNORM, A1R5G5B5), + _(A4R4G4B4_UNORM, A4R4G4B4), + _(L8_UNORM, Y8 ), + _(A8_UNORM, Y8 ), +}; + +static uint32_t +nv04_fragtex_format(uint pipe_format) +{ + struct nv04_texture_format *tf = nv04_texture_formats; + char fs[128]; + int i; + + for (i=0; i< sizeof(nv04_texture_formats)/sizeof(nv04_texture_formats[0]); i++) { + if (tf->pipe == pipe_format) + return tf->format; + tf++; + } + + pf_sprint_name(fs, pipe_format); + NOUVEAU_ERR("unknown texture format %s\n", fs); + return 0; +} + + +static void +nv04_fragtex_build(struct nv04_context *nv04, int unit) +{ + struct nv04_miptree *nv04mt = nv04->tex_miptree[unit]; + struct pipe_texture *pt = &nv04mt->base; + + switch (pt->target) { + case PIPE_TEXTURE_2D: + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + nv04->fragtex.format = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER + | NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER + | nv04_fragtex_format(pt->format) + | ( (pt->last_level + 1) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT ) + | ( log2i(pt->width[0]) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT ) + | ( log2i(pt->height[0]) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT ) + | NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE + | NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE + ; +} + + +void +nv04_fragtex_bind(struct nv04_context *nv04) +{ + nv04_fragtex_build(nv04, 0); +} + diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c new file mode 100644 index 00000000000..ab2d2386bee --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_miptree.c @@ -0,0 +1,138 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nv04_context.h" +#include "nv04_screen.h" + +static void +nv04_miptree_layout(struct nv04_miptree *nv04mt) +{ + struct pipe_texture *pt = &nv04mt->base; + uint width = pt->width[0], height = pt->height[0]; + uint offset = 0; + int nr_faces, l, f; + + nr_faces = 1; + + for (l = 0; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + + pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); + pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); + + nv04mt->level[l].pitch = pt->width[0] * pt->block.size; + nv04mt->level[l].pitch = (nv04mt->level[l].pitch + 63) & ~63; + + nv04mt->level[l].image_offset = + CALLOC(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = 0; l <= pt->last_level; l++) { + nv04mt->level[l].image_offset[f] = offset; + offset += nv04mt->level[l].pitch * pt->height[l]; + } + } + + nv04mt->total_size = offset; +} + +static struct pipe_texture * +nv04_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv04_miptree *mt; + + mt = MALLOC(sizeof(struct nv04_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + mt->base.refcount = 1; + mt->base.screen = screen; + + nv04_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + free(mt); + return NULL; + } + + return &mt->base; +} + +static void +nv04_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) +{ + struct pipe_winsys *ws = screen->winsys; + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv04_miptree *nv04mt = (struct nv04_miptree *)mt; + int l; + + pipe_buffer_reference(ws, &nv04mt->buffer, NULL); + for (l = 0; l <= mt->last_level; l++) { + if (nv04mt->level[l].image_offset) + free(nv04mt->level[l].image_offset); + } + free(nv04mt); + } +} + +static struct pipe_surface * +nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice, + unsigned flags) +{ + struct pipe_winsys *ws = pscreen->winsys; + struct nv04_miptree *nv04mt = (struct nv04_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(ws, &ps->buffer, nv04mt->buffer); + ps->format = pt->format; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->block = pt->block; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->stride = nv04mt->level[level].pitch; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv04mt->level[level].image_offset[face]; + } else { + ps->offset = nv04mt->level[level].image_offset[0]; + } + + return ps; +} + +static void +nv04_miptree_surface_del(struct pipe_screen *pscreen, + struct pipe_surface **psurface) +{ +} + +void +nv04_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv04_miptree_create; + pscreen->texture_release = nv04_miptree_release; + pscreen->get_tex_surface = nv04_miptree_surface_new; + pscreen->tex_surface_release = nv04_miptree_surface_del; +} + diff --git a/src/gallium/drivers/nv04/nv04_prim_vbuf.c b/src/gallium/drivers/nv04/nv04_prim_vbuf.c new file mode 100644 index 00000000000..daf28e81f56 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_prim_vbuf.c @@ -0,0 +1,312 @@ + +// XXX this has to go somewhere else +#define NONINC_METHOD 0x40000000 + + +#include "draw/draw_vbuf.h" +#include "pipe/p_debug.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" +#include "pipe/p_winsys.h" + +#include "nv04_context.h" +#include "nv04_state.h" + +#define VERTEX_SIZE 40 +#define VERTEX_BUFFER_SIZE (4096*VERTEX_SIZE) // 4096 vertices of 40 bytes each + +/** + * Primitive renderer for nv04. + */ +struct nv04_vbuf_render { + struct vbuf_render base; + + struct nv04_context *nv04; + + /** Vertex buffer */ + unsigned char* buffer; + + /** Vertex size in bytes */ + unsigned vertex_size; + + /** Current primitive */ + unsigned prim; +}; + + +/** + * Basically a cast wrapper. + */ +static INLINE struct nv04_vbuf_render * +nv04_vbuf_render( struct vbuf_render *render ) +{ + assert(render); + return (struct nv04_vbuf_render *)render; +} + + +static const struct vertex_info * +nv04_vbuf_render_get_vertex_info( struct vbuf_render *render ) +{ + struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); + struct nv04_context *nv04 = nv04_render->nv04; + return &nv04->vertex_info; +} + + +static void * +nv04_vbuf_render_allocate_vertices( struct vbuf_render *render, + ushort vertex_size, + ushort nr_vertices ) +{ + struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); + + nv04_render->buffer = (unsigned char*) malloc(VERTEX_BUFFER_SIZE); + assert(!nv04_render->buffer); + + return nv04_render->buffer; +} + + +static boolean +nv04_vbuf_render_set_primitive( struct vbuf_render *render, + unsigned prim ) +{ + struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); + + if (prim <= PIPE_PRIM_LINE_STRIP) + return FALSE; + + nv04_render->prim = prim; + return TRUE; +} + +static INLINE void nv04_2triangles(struct nv04_context* nv04, unsigned char* buffer, ushort v0, ushort v1, ushort v2, ushort v3, ushort v4, ushort v5) +{ + BEGIN_RING(fahrenheit,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xA),49); + OUT_RINGp(buffer + VERTEX_SIZE * v0,8); + OUT_RINGp(buffer + VERTEX_SIZE * v1,8); + OUT_RINGp(buffer + VERTEX_SIZE * v2,8); + OUT_RINGp(buffer + VERTEX_SIZE * v3,8); + OUT_RINGp(buffer + VERTEX_SIZE * v4,8); + OUT_RINGp(buffer + VERTEX_SIZE * v5,8); + OUT_RING(0xFEDCBA); +} + +static INLINE void nv04_1triangle(struct nv04_context* nv04, unsigned char* buffer, ushort v0, ushort v1, ushort v2) +{ + BEGIN_RING(fahrenheit,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xD),25); + OUT_RINGp(buffer + VERTEX_SIZE * v0,8); + OUT_RINGp(buffer + VERTEX_SIZE * v1,8); + OUT_RINGp(buffer + VERTEX_SIZE * v2,8); + OUT_RING(0xFED); +} + +static INLINE void nv04_1quad(struct nv04_context* nv04, unsigned char* buffer, ushort v0, ushort v1, ushort v2, ushort v3) +{ + BEGIN_RING(fahrenheit,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xC),33); + OUT_RINGp(buffer + VERTEX_SIZE * v0,8); + OUT_RINGp(buffer + VERTEX_SIZE * v1,8); + OUT_RINGp(buffer + VERTEX_SIZE * v2,8); + OUT_RINGp(buffer + VERTEX_SIZE * v3,8); + OUT_RING(0xFECEDC); +} + +static void nv04_vbuf_render_triangles_elts(struct nv04_vbuf_render * render, const ushort * indices, uint nr_indices) +{ + unsigned char* buffer = render->buffer; + struct nv04_context* nv04 = render->nv04; + int i; + + for( i=0; i< nr_indices-5; i+=6) + nv04_2triangles(nv04, + buffer, + indices[i+0], + indices[i+1], + indices[i+2], + indices[i+3], + indices[i+4], + indices[i+5] + ); + if (i != nr_indices) + { + nv04_1triangle(nv04, + buffer, + indices[i+0], + indices[i+1], + indices[i+2] + ); + i+=3; + } + if (i != nr_indices) + NOUVEAU_ERR("Houston, we have lost some vertices\n"); +} + +static void nv04_vbuf_render_tri_strip_elts(struct nv04_vbuf_render* render, const ushort* indices, uint nr_indices) +{ + const uint32_t striptbl[]={0x321210,0x543432,0x765654,0x987876,0xBA9A98,0xDCBCBA,0xFEDEDC}; + unsigned char* buffer = render->buffer; + struct nv04_context* nv04 = render->nv04; + int i,j; + + for(i = 0; ibuffer; + struct nv04_context* nv04 = render->nv04; + int i,j; + + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0), 8); + OUT_RINGp(buffer + VERTEX_SIZE * indices[0], 8); + + for(i = 1; ibuffer; + struct nv04_context* nv04 = render->nv04; + int i; + + for(i = 0; i < nr_indices; i += 4) + nv04_1quad(nv04, + buffer, + indices[i+0], + indices[i+1], + indices[i+2], + indices[i+3] + ); +} + + +static void +nv04_vbuf_render_draw( struct vbuf_render *render, + const ushort *indices, + uint nr_indices) +{ + struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); + + // emit the indices + switch( nv04_render->prim ) + { + case PIPE_PRIM_TRIANGLES: + nv04_vbuf_render_triangles_elts(nv04_render, indices, nr_indices); + break; + case PIPE_PRIM_QUAD_STRIP: + case PIPE_PRIM_TRIANGLE_STRIP: + nv04_vbuf_render_tri_strip_elts(nv04_render, indices, nr_indices); + break; + case PIPE_PRIM_TRIANGLE_FAN: + case PIPE_PRIM_POLYGON: + nv04_vbuf_render_tri_fan_elts(nv04_render, indices, nr_indices); + break; + case PIPE_PRIM_QUADS: + nv04_vbuf_render_quads_elts(nv04_render, indices, nr_indices); + break; + default: + NOUVEAU_ERR("You have to implement primitive %d, young padawan\n", nv04_render->prim); + break; + } +} + + +static void +nv04_vbuf_render_release_vertices( struct vbuf_render *render, + void *vertices, + unsigned vertex_size, + unsigned vertices_used ) +{ + struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); + + free(nv04_render->buffer); + nv04_render->buffer = NULL; +} + + +static void +nv04_vbuf_render_destroy( struct vbuf_render *render ) +{ + struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); + FREE(nv04_render); +} + + +/** + * Create a new primitive render. + */ +static struct vbuf_render * +nv04_vbuf_render_create( struct nv04_context *nv04 ) +{ + struct nv04_vbuf_render *nv04_render = CALLOC_STRUCT(nv04_vbuf_render); + + nv04_render->nv04 = nv04; + + nv04_render->base.max_vertex_buffer_bytes = VERTEX_BUFFER_SIZE; + nv04_render->base.max_indices = 65536; + nv04_render->base.get_vertex_info = nv04_vbuf_render_get_vertex_info; + nv04_render->base.allocate_vertices = nv04_vbuf_render_allocate_vertices; + nv04_render->base.set_primitive = nv04_vbuf_render_set_primitive; + nv04_render->base.draw = nv04_vbuf_render_draw; + nv04_render->base.release_vertices = nv04_vbuf_render_release_vertices; + nv04_render->base.destroy = nv04_vbuf_render_destroy; + + return &nv04_render->base; +} + + +/** + * Create a new primitive vbuf/render stage. + */ +struct draw_stage *nv04_draw_vbuf_stage( struct nv04_context *nv04 ) +{ + struct vbuf_render *render; + struct draw_stage *stage; + + render = nv04_vbuf_render_create(nv04); + if(!render) + return NULL; + + stage = draw_vbuf_stage( nv04->draw, render ); + if(!stage) { + render->destroy(render); + return NULL; + } + + return stage; +} diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c new file mode 100644 index 00000000000..9f34117b8c8 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_screen.c @@ -0,0 +1,216 @@ +#include "pipe/p_screen.h" +#include "pipe/p_util.h" + +#include "nv04_context.h" +#include "nv04_screen.h" + +static const char * +nv04_screen_get_name(struct pipe_screen *screen) +{ + struct nv04_screen *nv04screen = nv04_screen(screen); + struct nouveau_device *dev = nv04screen->nvws->channel->device; + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); + return buffer; +} + +static const char * +nv04_screen_get_vendor(struct pipe_screen *screen) +{ + return "nouveau"; +} + +static int +nv04_screen_get_param(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 1; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 0; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 0; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 1; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 10; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 0; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv04_screen_get_paramf(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 0.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 0.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 0.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 0.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static boolean +nv04_screen_is_format_supported(struct pipe_screen *screen, + enum pipe_format format, uint type) +{ + switch (type) { + case PIPE_SURFACE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + break; + case PIPE_TEXTURE: + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_X8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + return TRUE; + default: + break; + } + break; + default: + assert(0); + }; + + return FALSE; +} + +static void * +nv04_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv04_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + +static void +nv04_screen_destroy(struct pipe_screen *pscreen) +{ + struct nv04_screen *screen = nv04_screen(pscreen); + struct nouveau_winsys *nvws = screen->nvws; + + nvws->notifier_free(&screen->sync); + nvws->grobj_free(&screen->fahrenheit); + + FREE(pscreen); +} + +struct pipe_screen * +nv04_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) +{ + struct nv04_screen *screen = CALLOC_STRUCT(nv04_screen); + unsigned fahrenheit_class = 0, sub3d_class = 0; + unsigned chipset = nvws->channel->device->chipset; + int ret; + + if (!screen) + return NULL; + screen->nvws = nvws; + + if (chipset>=0x20) { + fahrenheit_class = 0; + sub3d_class = 0; + } else if (chipset>=0x10) { + fahrenheit_class = NV10_DX5_TEXTURED_TRIANGLE; + sub3d_class = NV10_CONTEXT_SURFACES_3D; + } else { + fahrenheit_class=NV04_DX5_TEXTURED_TRIANGLE; + sub3d_class = NV04_CONTEXT_SURFACES_3D; + } + + if (!fahrenheit_class) { + NOUVEAU_ERR("Unknown nv04 chipset: nv%02x\n", chipset); + return NULL; + } + + /* 3D object */ + ret = nvws->grobj_alloc(nvws, fahrenheit_class, &screen->fahrenheit); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return NULL; + } + + /* 3D surface object */ + ret = nvws->grobj_alloc(nvws, sub3d_class, &screen->context_surfaces_3d); + if (ret) { + NOUVEAU_ERR("Error creating 3D surface object: %d\n", ret); + return NULL; + } + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv04_screen_destroy(&screen->pipe); + return NULL; + } + + screen->pipe.winsys = ws; + screen->pipe.destroy = nv04_screen_destroy; + + screen->pipe.get_name = nv04_screen_get_name; + screen->pipe.get_vendor = nv04_screen_get_vendor; + screen->pipe.get_param = nv04_screen_get_param; + screen->pipe.get_paramf = nv04_screen_get_paramf; + + screen->pipe.is_format_supported = nv04_screen_is_format_supported; + + screen->pipe.surface_map = nv04_surface_map; + screen->pipe.surface_unmap = nv04_surface_unmap; + + nv04_screen_init_miptree_functions(&screen->pipe); + + return &screen->pipe; +} + diff --git a/src/gallium/drivers/nv04/nv04_screen.h b/src/gallium/drivers/nv04/nv04_screen.h new file mode 100644 index 00000000000..99a49cdf7a9 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_screen.h @@ -0,0 +1,25 @@ +#ifndef __NV04_SCREEN_H__ +#define __NV04_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv04_screen { + struct pipe_screen pipe; + + struct nouveau_winsys *nvws; + unsigned chipset; + + /* HW graphics objects */ + struct nouveau_grobj *fahrenheit; + struct nouveau_grobj *context_surfaces_3d; + struct nouveau_notifier *sync; + +}; + +static INLINE struct nv04_screen * +nv04_screen(struct pipe_screen *screen) +{ + return (struct nv04_screen *)screen; +} + +#endif diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c new file mode 100644 index 00000000000..39d7e8b42d0 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_state.c @@ -0,0 +1,494 @@ +#include "draw/draw_context.h" +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_shader_tokens.h" + + +#include "nv04_context.h" +#include "nv04_state.h" + +static void * +nv04_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv04_blend_state *cb; + + cb = malloc(sizeof(struct nv04_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + cb->b_src = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dst = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + + + return (void *)cb; +} + +static void +nv04_blend_state_bind(struct pipe_context *pipe, void *blend) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + nv04->blend = (struct nv04_blend_state*)blend; + + nv04->dirty |= NV04_NEW_BLEND; +} + +static void +nv04_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + case PIPE_TEX_WRAP_MIRROR_CLAMP: + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP; + } + return ret >> NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT; +} + +static void * +nv04_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + + struct nv04_sampler_state *ss; + uint32_t filter = 0; + + ss = malloc(sizeof(struct nv04_sampler_state)); + + ss->format = ((wrap_mode(cso->wrap_s) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT) | + (wrap_mode(cso->wrap_t) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT)); + + if (cso->max_anisotropy > 1.0) { + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE | NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE; + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST; + break; + } + break; + } + + ss->filter = filter; + + return (void *)ss; +} + +static void +nv04_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) +{ + struct nv04_context *nv04 = nv04_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv04->sampler[unit] = sampler[unit]; + nv04->dirty_samplers |= (1 << unit); + } +} + +static void +nv04_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void +nv04_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **miptree) +{ + struct nv04_context *nv04 = nv04_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv04->tex_miptree[unit] = (struct nv04_miptree *)miptree[unit]; + nv04->dirty_samplers |= (1 << unit); + } +} + +static void * +nv04_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv04_rasterizer_state *rs; + + /*XXX: ignored: + * scissor + * points/lines (no hw support, emulated with tris in gallium) + */ + rs = malloc(sizeof(struct nv04_rasterizer_state)); + + rs->blend = cso->flatshade ? NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT : NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD; + + return (void *)rs; +} + +static void +nv04_rasterizer_state_bind(struct pipe_context *pipe, void *rast) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + nv04->rast = (struct nv04_rasterizer_state*)rast; + + draw_set_rasterizer_state(nv04->draw, (nv04->rast ? nv04->rast->templ : NULL)); + + nv04->dirty |= NV04_NEW_RAST | NV04_NEW_BLEND; +} + +static void +nv04_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static INLINE uint32_t nv04_compare_func(uint32_t f) +{ + switch ( f ) { + case PIPE_FUNC_NEVER: return 1; + case PIPE_FUNC_LESS: return 2; + case PIPE_FUNC_EQUAL: return 3; + case PIPE_FUNC_LEQUAL: return 4; + case PIPE_FUNC_GREATER: return 5; + case PIPE_FUNC_NOTEQUAL: return 6; + case PIPE_FUNC_GEQUAL: return 7; + case PIPE_FUNC_ALWAYS: return 8; + } + NOUVEAU_MSG("Unable to find the function\n"); + return 0; +} + +static void * +nv04_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv04_depth_stencil_alpha_state *hw; + + hw = malloc(sizeof(struct nv04_depth_stencil_alpha_state)); + + hw->control = float_to_ubyte(cso->alpha.ref); + hw->control |= ( nv04_compare_func(cso->alpha.func) << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT ); + hw->control |= cso->alpha.enabled ? NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE : 0; + hw->control |= NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN; + hw->control |= cso->depth.enabled ? (1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT) : 0; + hw->control |= ( nv04_compare_func(cso->depth.func)<< NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT ); + hw->control |= 1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT; // no culling, handled by the draw module + hw->control |= NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE; + hw->control |= NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE; + hw->control |= cso->depth.writemask ? (1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT) : 0; + hw->control |= 1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT; // integer zbuffer format + + return (void *)hw; +} + +static void +nv04_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + nv04->dsa = hwcso; + nv04->dirty |= NV04_NEW_CONTROL; +} + +static void +nv04_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + free(hwcso); +} + +static void * +nv04_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *templ) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + return draw_create_vertex_shader(nv04->draw, templ); +} + +static void +nv04_vp_state_bind(struct pipe_context *pipe, void *shader) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + draw_bind_vertex_shader(nv04->draw, (struct draw_vertex_shader *) shader); + + nv04->dirty |= NV04_NEW_VERTPROG; +} + +static void +nv04_vp_state_delete(struct pipe_context *pipe, void *shader) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + draw_delete_vertex_shader(nv04->draw, (struct draw_vertex_shader *) shader); +} + +static void * +nv04_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv04_fragment_program *fp; + + fp = CALLOC(1, sizeof(struct nv04_fragment_program)); + fp->pipe = cso; + + return (void *)fp; +} + +static void +nv04_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv04_context *nv04 = nv04_context(pipe); + struct nv04_fragment_program *fp = hwcso; + + nv04->fragprog.current = fp; + nv04->dirty |= NV04_NEW_FRAGPROG; +} + +static void +nv04_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv04_context *nv04 = nv04_context(pipe); + struct nv04_fragment_program *fp = hwcso; + + nv04_fragprog_destroy(nv04, fp); + free(fp); +} + +static void +nv04_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ +} + +static void +nv04_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ +} + +static void +nv04_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + if (shader == PIPE_SHADER_VERTEX) { + nv04->vertprog.constant_buf = buf->buffer; + nv04->dirty |= NV04_NEW_VERTPROG; + } else + if (shader == PIPE_SHADER_FRAGMENT) { + nv04->fragprog.constant_buf = buf->buffer; + nv04->dirty |= NV04_NEW_FRAGPROG; + } +} + +static void +nv04_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv04_context *nv04 = nv04_context(pipe); + struct pipe_surface *rt, *zeta; + uint32_t rt_format, w, h; + int colour_format = 0, zeta_format = 0; + + w = fb->cbufs[0]->width; + h = fb->cbufs[0]->height; + colour_format = fb->cbufs[0]->format; + rt = fb->cbufs[0]; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format = 0x108; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format = 0x103; + break; + default: + assert(0); + } + + BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_FORMAT, 1); + OUT_RING(rt_format); + + /* FIXME pitches have to be aligned ! */ + BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_PITCH, 2); + OUT_RING(rt->stride|(zeta->stride<<16)); + OUT_RELOCl(rt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + if (fb->zsbuf) { + BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1); + OUT_RELOCl(zeta->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + } +} + +static void +nv04_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + NOUVEAU_ERR("line stipple hahaha\n"); +} + +static void +nv04_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ +/* struct nv04_context *nv04 = nv04_context(pipe); + + // XXX + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ +} + +static void +nv04_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *viewport) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + nv04->viewport = *viewport; + + draw_set_viewport_state(nv04->draw, &nv04->viewport); +} + +static void +nv04_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *buffers) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + draw_flush(nv04->draw); + + memcpy(nv04->vertex_buffer, buffers, count * sizeof(buffers[0])); + nv04->num_vertex_buffers = count; + + draw_set_vertex_buffers(nv04->draw, count, buffers); +} + +static void +nv04_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *elements) +{ + struct nv04_context *nv04 = nv04_context(pipe); + + draw_flush(nv04->draw); + + nv04->num_vertex_elements = count; + draw_set_vertex_elements(nv04->draw, count, elements); +} + +void +nv04_init_state_functions(struct nv04_context *nv04) +{ + nv04->pipe.create_blend_state = nv04_blend_state_create; + nv04->pipe.bind_blend_state = nv04_blend_state_bind; + nv04->pipe.delete_blend_state = nv04_blend_state_delete; + + nv04->pipe.create_sampler_state = nv04_sampler_state_create; + nv04->pipe.bind_sampler_states = nv04_sampler_state_bind; + nv04->pipe.delete_sampler_state = nv04_sampler_state_delete; + nv04->pipe.set_sampler_textures = nv04_set_sampler_texture; + + nv04->pipe.create_rasterizer_state = nv04_rasterizer_state_create; + nv04->pipe.bind_rasterizer_state = nv04_rasterizer_state_bind; + nv04->pipe.delete_rasterizer_state = nv04_rasterizer_state_delete; + + nv04->pipe.create_depth_stencil_alpha_state = nv04_depth_stencil_alpha_state_create; + nv04->pipe.bind_depth_stencil_alpha_state = nv04_depth_stencil_alpha_state_bind; + nv04->pipe.delete_depth_stencil_alpha_state = nv04_depth_stencil_alpha_state_delete; + + nv04->pipe.create_vs_state = nv04_vp_state_create; + nv04->pipe.bind_vs_state = nv04_vp_state_bind; + nv04->pipe.delete_vs_state = nv04_vp_state_delete; + + nv04->pipe.create_fs_state = nv04_fp_state_create; + nv04->pipe.bind_fs_state = nv04_fp_state_bind; + nv04->pipe.delete_fs_state = nv04_fp_state_delete; + + nv04->pipe.set_blend_color = nv04_set_blend_color; + nv04->pipe.set_clip_state = nv04_set_clip_state; + nv04->pipe.set_constant_buffer = nv04_set_constant_buffer; + nv04->pipe.set_framebuffer_state = nv04_set_framebuffer_state; + nv04->pipe.set_polygon_stipple = nv04_set_polygon_stipple; + nv04->pipe.set_scissor_state = nv04_set_scissor_state; + nv04->pipe.set_viewport_state = nv04_set_viewport_state; + + nv04->pipe.set_vertex_buffers = nv04_set_vertex_buffers; + nv04->pipe.set_vertex_elements = nv04_set_vertex_elements; +} + diff --git a/src/gallium/drivers/nv04/nv04_state.h b/src/gallium/drivers/nv04/nv04_state.h new file mode 100644 index 00000000000..7487819c3ab --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_state.h @@ -0,0 +1,71 @@ +#ifndef __NV04_STATE_H__ +#define __NV04_STATE_H__ + +#include "pipe/p_state.h" +#include "tgsi/util/tgsi_scan.h" + +struct nv04_blend_state { + uint32_t b_enable; + uint32_t b_src; + uint32_t b_dst; +}; + +struct nv04_fragtex_state { + uint32_t format; +}; + +struct nv04_sampler_state { + uint32_t filter; + uint32_t format; +}; + +struct nv04_depth_stencil_alpha_state { + uint32_t control; +}; + +struct nv04_rasterizer_state { + uint32_t blend; + + const struct pipe_rasterizer_state *templ; +}; + +struct nv04_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +struct nv04_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv04_fragment_program { + const struct pipe_shader_state *pipe; + struct tgsi_shader_info info; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv04_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + uint32_t fp_reg_control; +}; + + + +#endif diff --git a/src/gallium/drivers/nv04/nv04_state_emit.c b/src/gallium/drivers/nv04/nv04_state_emit.c new file mode 100644 index 00000000000..0ad40a092ea --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_state_emit.c @@ -0,0 +1,156 @@ +#include "nv04_context.h" +#include "nv04_state.h" + +static void nv04_vertex_layout(struct pipe_context* pipe) +{ + struct nv04_context *nv04 = nv04_context(pipe); + struct nv04_fragment_program *fp = nv04->fragprog.current; + uint32_t src = 0; + int i; + struct vertex_info vinfo; + + memset(&vinfo, 0, sizeof(vinfo)); + + for (i = 0; i < fp->info.num_inputs; i++) { + switch (i) { + case TGSI_SEMANTIC_POSITION: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + case TGSI_SEMANTIC_COLOR: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + default: + case TGSI_SEMANTIC_GENERIC: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + case TGSI_SEMANTIC_FOG: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + } + } + draw_compute_vertex_size(&vinfo); +} + +static uint32_t nv04_blend_func(uint32_t f) +{ + switch ( f ) { + case PIPE_BLENDFACTOR_ZERO: return 0x1; + case PIPE_BLENDFACTOR_ONE: return 0x2; + case PIPE_BLENDFACTOR_SRC_COLOR: return 0x3; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: return 0x4; + case PIPE_BLENDFACTOR_SRC_ALPHA: return 0x5; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: return 0x6; + case PIPE_BLENDFACTOR_DST_ALPHA: return 0x7; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: return 0x8; + case PIPE_BLENDFACTOR_DST_COLOR: return 0x9; + case PIPE_BLENDFACTOR_INV_DST_COLOR: return 0xA; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: return 0xB; + } + NOUVEAU_MSG("Unable to find the blend function 0x%x\n",f); + return 0; +} + +static void nv04_emit_control(struct nv04_context* nv04) +{ + uint32_t control = nv04->dsa->control; + + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1); + OUT_RING(control); +} + +static void nv04_emit_blend(struct nv04_context* nv04) +{ + uint32_t blend; + + blend=0x4; // texture MODULATE_ALPHA + blend|=0x20; // alpha is MSB + blend|=(2<<6); // flat shading + blend|=(1<<8); // persp correct + blend|=(0<<16); // no fog + blend|=(nv04->blend->b_enable<<20); + blend|=(nv04_blend_func(nv04->blend->b_src)<<24); + blend|=(nv04_blend_func(nv04->blend->b_dst)<<28); + + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_BLEND, 1); + OUT_RING(blend); +} + +static void nv04_emit_sampler(struct nv04_context *nv04, int unit) +{ + struct nv04_miptree *nv04mt = nv04->tex_miptree[unit]; + struct pipe_texture *pt = &nv04mt->base; + + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_OFFSET, 3); + OUT_RELOCl(nv04mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv04mt->buffer, (nv04->fragtex.format | nv04->sampler[unit]->format), NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RING(nv04->sampler[unit]->filter); +} + +void +nv04_emit_hw_state(struct nv04_context *nv04) +{ + int i; + + if (nv04->dirty & NV04_NEW_VERTPROG) { + //nv04_vertprog_bind(nv04, nv04->vertprog.current); + nv04->dirty &= ~NV04_NEW_VERTPROG; + } + + if (nv04->dirty & NV04_NEW_FRAGPROG) { + nv04_fragprog_bind(nv04, nv04->fragprog.current); + /*XXX: clear NV04_NEW_FRAGPROG if no new program uploaded */ + nv04->dirty_samplers |= (1<<10); + nv04->dirty_samplers = 0; + } + + if (nv04->dirty & NV04_NEW_CONTROL) { + nv04->dirty &= ~NV04_NEW_CONTROL; + + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1); + OUT_RING(nv04->dsa->control); + } + + if (nv04->dirty & NV04_NEW_BLEND) { + nv04->dirty &= ~NV04_NEW_BLEND; + + nv04_emit_blend(nv04); + } + + if (nv04->dirty & NV04_NEW_SAMPLER) { + nv04->dirty &= ~NV04_NEW_SAMPLER; + + nv04_emit_sampler(nv04, 0); + } + + if (nv04->dirty & NV04_NEW_VIEWPORT) { + nv04->dirty &= ~NV04_NEW_VIEWPORT; +// nv04_state_emit_viewport(nv04); + } + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv04_vbo.c + */ + + /* Render target */ +/* BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_PITCH, 2); + OUT_RING(rt->stride|(zeta->stride<<16)); + OUT_RELOCl(rt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + if (fb->zsbuf) { + BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1); + OUT_RELOCl(zeta->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + }*/ + + /* Texture images */ + for (i = 0; i < 1; i++) { + if (!(nv04->fp_samplers & (1 << i))) + continue; + struct nv04_miptree *nv04mt = nv04->tex_miptree[i]; + BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_OFFSET, 2); + OUT_RELOCl(nv04mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv04mt->buffer, (nv04->fragtex.format | nv04->sampler[i]->format), NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + } +} + diff --git a/src/gallium/drivers/nv04/nv04_surface.c b/src/gallium/drivers/nv04/nv04_surface.c new file mode 100644 index 00000000000..b13ebf9f9b5 --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_surface.c @@ -0,0 +1,65 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv04_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/p_tile.h" + +static void +nv04_surface_copy(struct pipe_context *pipe, unsigned do_flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv04_context *nv04 = nv04_context(pipe); + struct nouveau_winsys *nvws = nv04->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv04_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv04_context *nv04 = nv04_context(pipe); + struct nouveau_winsys *nvws = nv04->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv04_init_surface_functions(struct nv04_context *nv04) +{ + nv04->pipe.surface_copy = nv04_surface_copy; + nv04->pipe.surface_fill = nv04_surface_fill; +} diff --git a/src/gallium/drivers/nv04/nv04_vbo.c b/src/gallium/drivers/nv04/nv04_vbo.c new file mode 100644 index 00000000000..fbfe0cf406d --- /dev/null +++ b/src/gallium/drivers/nv04/nv04_vbo.c @@ -0,0 +1,72 @@ +#include "draw/draw_context.h" +#include "pipe/p_context.h" +#include "pipe/p_state.h" +#include "pipe/p_util.h" + +#include "nv04_context.h" +#include "nv04_state.h" + +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_pushbuf.h" + +boolean nv04_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count) +{ + struct nv04_context *nv04 = nv04_context( pipe ); + struct draw_context *draw = nv04->draw; + unsigned i; + + /* + * Map vertex buffers + */ + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (nv04->vertex_buffer[i].buffer) { + void *buf + = pipe->winsys->buffer_map(pipe->winsys, + nv04->vertex_buffer[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(draw, i, buf); + } + } + /* Map index buffer, if present */ + if (indexBuffer) { + void *mapped_indexes + = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes); + } + else { + /* no index/element buffer */ + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + /* draw! */ + draw_arrays(nv04->draw, prim, start, count); + + /* + * unmap vertex/index buffers + */ + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (nv04->vertex_buffer[i].buffer) { + pipe->winsys->buffer_unmap(pipe->winsys, nv04->vertex_buffer[i].buffer); + draw_set_mapped_vertex_buffer(draw, i, NULL); + } + } + if (indexBuffer) { + pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + return TRUE; +} + +boolean nv04_draw_arrays( struct pipe_context *pipe, + unsigned prim, unsigned start, unsigned count) +{ + return nv04_draw_elements(pipe, NULL, 0, prim, start, count); +} + + + -- cgit v1.2.3 From 97646f10459dd743d9bfd091aec27b28b24222f2 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Tue, 8 Jul 2008 01:30:48 +0200 Subject: nv04: use BEGIN_RING_NI --- src/gallium/drivers/nv04/nv04_prim_vbuf.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) diff --git a/src/gallium/drivers/nv04/nv04_prim_vbuf.c b/src/gallium/drivers/nv04/nv04_prim_vbuf.c index daf28e81f56..80a4dbc311c 100644 --- a/src/gallium/drivers/nv04/nv04_prim_vbuf.c +++ b/src/gallium/drivers/nv04/nv04_prim_vbuf.c @@ -1,8 +1,4 @@ -// XXX this has to go somewhere else -#define NONINC_METHOD 0x40000000 - - #include "draw/draw_vbuf.h" #include "pipe/p_debug.h" #include "pipe/p_util.h" @@ -160,7 +156,7 @@ static void nv04_vbuf_render_tri_strip_elts(struct nv04_vbuf_render* render, con for(j = 0; j Date: Tue, 8 Jul 2008 01:32:32 +0200 Subject: nv04: Hook the lib into the build. --- src/gallium/winsys/dri/nouveau/Makefile | 1 + 1 file changed, 1 insertion(+) diff --git a/src/gallium/winsys/dri/nouveau/Makefile b/src/gallium/winsys/dri/nouveau/Makefile index f637b2cebfc..be630ff6d16 100644 --- a/src/gallium/winsys/dri/nouveau/Makefile +++ b/src/gallium/winsys/dri/nouveau/Makefile @@ -8,6 +8,7 @@ MINIGLX_SOURCES = PIPE_DRIVERS = \ $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ + $(TOP)/src/gallium/drivers/nv04/libnv04.a \ $(TOP)/src/gallium/drivers/nv10/libnv10.a \ $(TOP)/src/gallium/drivers/nv30/libnv30.a \ $(TOP)/src/gallium/drivers/nv40/libnv40.a \ -- cgit v1.2.3 From 205101dafb0042ebeeab3dd50f549d8a3d3c08ce Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Tue, 8 Jul 2008 01:34:36 +0200 Subject: nv10: there are no 3D textures. --- src/gallium/drivers/nv10/nv10_miptree.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index f1486a35df5..9a68df29250 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -11,7 +11,7 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) { struct pipe_texture *pt = &nv10mt->base; boolean swizzled = FALSE; - uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; + uint width = pt->width[0], height = pt->height[0]; uint offset = 0; int nr_faces, l, f; @@ -24,7 +24,6 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) for (l = 0; l <= pt->last_level; l++) { pt->width[l] = width; pt->height[l] = height; - pt->depth[l] = depth; pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); @@ -39,7 +38,6 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); - depth = MAX2(1, depth >> 1); } -- cgit v1.2.3 From 06d87b44cffef9e335b9a0e70fd2aec61fcd171b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 8 Jul 2008 12:51:29 +1000 Subject: nv50: make use of nouveau drm 0.0.11 to get 3d going --- src/gallium/drivers/nouveau/nouveau_bo.h | 2 + src/gallium/drivers/nouveau/nouveau_stateobj.h | 3 + src/gallium/drivers/nv50/nv50_state_validate.c | 6 +- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 16 +++ src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 1 + src/gallium/winsys/dri/nouveau/nouveau_screen.c | 2 +- .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 12 ++ src/gallium/winsys/dri/nouveau/nv50_surface.c | 158 +++++++++++---------- 8 files changed, 120 insertions(+), 80 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_bo.h b/src/gallium/drivers/nouveau/nouveau_bo.h index 18020e9c652..65b138283c4 100644 --- a/src/gallium/drivers/nouveau/nouveau_bo.h +++ b/src/gallium/drivers/nouveau/nouveau_bo.h @@ -35,6 +35,8 @@ #define NOUVEAU_BO_HIGH (1 << 7) #define NOUVEAU_BO_OR (1 << 8) #define NOUVEAU_BO_LOCAL (1 << 9) +#define NOUVEAU_BO_TILED (1 << 10) +#define NOUVEAU_BO_ZTILE (1 << 11) #define NOUVEAU_BO_DUMMY (1 << 31) struct nouveau_bo { diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index 78b27380702..998ec2d4ad4 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -136,6 +136,9 @@ so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so) struct nouveau_pushbuf *pb = nvws->channel->pushbuf; unsigned i; + if (!so) + return; + i = so->cur_reloc << 1; if (nvws->channel->pushbuf->remaining < i) nvws->push_flush(nvws, i, NULL); diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index cf5a071a8d6..88ef685bb1f 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -44,7 +44,7 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_data(so, 0xe6); break; } - so_data(so, 0x00000040); + so_data(so, 0x00000000); so_data(so, 0x00000000); so_method(so, tesla, 0x1224, 1); @@ -82,7 +82,7 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_data(so, 0x16); break; } - so_data(so, 0x00000040); + so_data(so, 0x00000000); so_data(so, 0x00000000); so_method(so, tesla, 0x1538, 1); @@ -265,7 +265,7 @@ scissor_uptodate: so_data (so, 0x2a712488); so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0xd0c05000); + so_data (so, 0xd0005000); so_data (so, 0x00300000); so_data (so, mt->base.width[0]); so_data (so, (mt->base.depth[0] << 16) | diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c index 57f5b7d1ae1..b5942994d94 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_bo.c @@ -182,6 +182,13 @@ nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, nvbo->drm.alignment = align; nvbo->refcount = 1; + if (flags & NOUVEAU_BO_TILED) { + nvbo->tiled = 1; + if (flags & NOUVEAU_BO_ZTILE) + nvbo->tiled |= 2; + flags &= ~NOUVEAU_BO_TILED; + } + ret = nouveau_bo_set_status(&nvbo->base, flags); if (ret) { free(nvbo); @@ -332,6 +339,12 @@ nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) else if (flags & NOUVEAU_BO_GART) new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); + + if (nvbo->tiled && flags) { + new_flags |= NOUVEAU_MEM_TILE; + if (nvbo->tiled & 2) + new_flags |= NOUVEAU_MEM_TILE_ZETA; + } if (new_flags) { ret = nouveau_mem_alloc(bo->device, bo->size, @@ -347,9 +360,12 @@ nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) /* Copy old -> new */ /*XXX: use M2MF */ if (nvbo->sysmem || nvbo->map) { + struct nouveau_pushbuf_bo *pbo = nvbo->pending; + nvbo->pending = NULL; nouveau_bo_map(bo, NOUVEAU_BO_RD); memcpy(new_map, bo->map, bo->size); nouveau_bo_unmap(bo); + nvbo->pending = pbo; } /* Free old memory */ diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h index 5829f649cd5..dcd6a5eb0a4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h +++ b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h @@ -259,6 +259,7 @@ struct nouveau_bo_priv { uint64_t offset; uint64_t flags; + int tiled; }; #define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c index b15ee7509cc..df1fe7e69b4 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_screen.c @@ -13,7 +13,7 @@ #include "nouveau_screen.h" #include "nouveau_swapbuffers.h" -#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10 +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11 #error nouveau_drm.h version does not match expected version #endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index ba5dd2eea4b..765ea36f881 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -76,6 +76,18 @@ nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE) flags |= NOUVEAU_BO_GART; flags |= NOUVEAU_BO_VRAM; + + switch (dev->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + flags |= NOUVEAU_BO_TILED; + if (usage & NOUVEAU_BUFFER_USAGE_ZETA) + flags |= NOUVEAU_BO_ZTILE; + break; + default: + break; + } } if (usage & PIPE_BUFFER_USAGE_VERTEX) { diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c index cf76d76cf52..c8ab7f690f3 100644 --- a/src/gallium/winsys/dri/nouveau/nv50_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv50_surface.c @@ -22,52 +22,70 @@ nv50_format(enum pipe_format format) } static int -nv50_surface_copy_prep(struct nouveau_context *nv, - struct pipe_surface *dst, struct pipe_surface *src) +nv50_surface_set(struct nouveau_context *nv, struct pipe_surface *surf, int dst) { struct nouveau_channel *chan = nv->nvc->channel; struct nouveau_grobj *eng2d = nv->nvc->Nv2D; - int surf_format; - - assert(src->format == dst->format); + struct nouveau_bo *bo = nouveau_buffer(surf->buffer)->bo; + int surf_format, mthd = dst ? NV50_2D_DST_FORMAT : NV50_2D_SRC_FORMAT; + int flags = NOUVEAU_BO_VRAM | (dst ? NOUVEAU_BO_WR : NOUVEAU_BO_RD); + + surf_format = nv50_format(surf->format); + if (surf_format < 0) + return 1; + + if (!nouveau_bo(bo)->tiled) { + BEGIN_RING(chan, eng2d, mthd, 2); + OUT_RING (chan, surf_format); + OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, mthd + 0x14, 5); + OUT_RING (chan, surf->stride); + OUT_RING (chan, surf->width); + OUT_RING (chan, surf->height); + OUT_RELOCh(chan, bo, surf->offset, flags); + OUT_RELOCl(chan, bo, surf->offset, flags); + } else { + BEGIN_RING(chan, eng2d, mthd, 5); + OUT_RING (chan, surf_format); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, eng2d, mthd + 0x18, 4); + OUT_RING (chan, surf->width); + OUT_RING (chan, surf->height); + OUT_RELOCh(chan, bo, surf->offset, flags); + OUT_RELOCl(chan, bo, surf->offset, flags); + } + +#if 0 + if (dst) { + BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, surf->width); + OUT_RING (chan, surf->height); + } +#endif + + return 0; +} - surf_format = nv50_format(dst->format); - assert(surf_format >= 0); +static int +nv50_surface_copy_prep(struct nouveau_context *nv, + struct pipe_surface *dst, struct pipe_surface *src) +{ + int ret; - BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + assert(src->format == dst->format); - BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 2); - OUT_RING (chan, surf_format); - OUT_RING (chan, 1); - BEGIN_RING(chan, eng2d, NV50_2D_DST_PITCH, 5); - OUT_RING (chan, dst->stride); - OUT_RING (chan, dst->width); - OUT_RING (chan, dst->height); - OUT_RELOCh(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); - OUT_RING (chan, 0); - OUT_RING (chan, 0); - OUT_RING (chan, dst->width); - OUT_RING (chan, dst->height); + ret = nv50_surface_set(nv, dst, 1); + if (ret) + return ret; - BEGIN_RING(chan, eng2d, NV50_2D_SRC_FORMAT, 2); - OUT_RING (chan, surf_format); - OUT_RING (chan, 1); - BEGIN_RING(chan, eng2d, NV50_2D_SRC_PITCH, 5); - OUT_RING (chan, src->stride); - OUT_RING (chan, src->width); - OUT_RING (chan, src->height); - OUT_RELOCh(chan, nouveau_buffer(src->buffer)->bo, src->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + ret = nv50_surface_set(nv, src, 0); + if (ret) + return ret; return 0; } @@ -79,17 +97,19 @@ nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, struct nouveau_channel *chan = nv->nvc->channel; struct nouveau_grobj *eng2d = nv->nvc->Nv2D; - BEGIN_RING(chan, eng2d, 0x0110, 1); + BEGIN_RING(chan, eng2d, 0x088c, 1); OUT_RING (chan, 0); - BEGIN_RING(chan, eng2d, NV50_2D_BLIT_DST_X, 12); + BEGIN_RING(chan, eng2d, NV50_2D_BLIT_DST_X, 4); OUT_RING (chan, dx); OUT_RING (chan, dy); OUT_RING (chan, w); OUT_RING (chan, h); + BEGIN_RING(chan, eng2d, 0x08c0, 4); OUT_RING (chan, 0); OUT_RING (chan, 1); OUT_RING (chan, 0); OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, 0x08d0, 4); OUT_RING (chan, 0); OUT_RING (chan, sx); OUT_RING (chan, 0); @@ -109,35 +129,15 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, { struct nouveau_channel *chan = nv->nvc->channel; struct nouveau_grobj *eng2d = nv->nvc->Nv2D; - int surf_format, rect_format; - - surf_format = nv50_format(dst->format); - if (surf_format < 0) - return 1; + int rect_format, ret; rect_format = nv50_format(dst->format); if (rect_format < 0) return 1; - BEGIN_RING(chan, eng2d, NV50_2D_DMA_IN_MEMORY1, 1); - OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 2); - OUT_RING (chan, surf_format); - OUT_RING (chan, 1); - BEGIN_RING(chan, eng2d, NV50_2D_DST_PITCH, 5); - OUT_RING (chan, dst->stride); - OUT_RING (chan, dst->width); - OUT_RING (chan, dst->height); - OUT_RELOCh(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); - OUT_RING (chan, 0); - OUT_RING (chan, 0); - OUT_RING (chan, dst->width); - OUT_RING (chan, dst->height); + ret = nv50_surface_set(nv, dst, 1); + if (ret) + return ret; BEGIN_RING(chan, eng2d, 0x0580, 3); OUT_RING (chan, 4); @@ -151,27 +151,33 @@ nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, OUT_RING (chan, dy + h); FIRE_RING(chan); - return 0; } int nouveau_surface_channel_create_nv50(struct nouveau_channel_context *nvc) { + struct nouveau_channel *chan = nvc->channel; + struct nouveau_grobj *eng2d = NULL; int ret; - ret = nouveau_grobj_alloc(nvc->channel, nvc->next_handle++, NV50_2D, - &nvc->Nv2D); + ret = nouveau_grobj_alloc(chan, nvc->next_handle++, NV50_2D, &eng2d); if (ret) return ret; - BIND_RING (nvc->channel, nvc->Nv2D, nvc->next_subchannel++); - BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_DMA_NOTIFY, 1); - OUT_RING (nvc->channel, nvc->sync_notifier->handle); - BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_DMA_IN_MEMORY0, 2); - OUT_RING (nvc->channel, nvc->channel->vram->handle); - OUT_RING (nvc->channel, nvc->channel->vram->handle); - BEGIN_RING(nvc->channel, nvc->Nv2D, NV50_2D_OPERATION, 1); - OUT_RING (nvc->channel, NV50_2D_OPERATION_SRCCOPY); + nvc->Nv2D = eng2d; + + BIND_RING (chan, eng2d, nvc->next_subchannel++); + BEGIN_RING(chan, eng2d, NV50_2D_DMA_NOTIFY, 4); + OUT_RING (chan, nvc->sync_notifier->handle); + OUT_RING (chan, chan->vram->handle); + OUT_RING (chan, chan->vram->handle); + OUT_RING (chan, chan->vram->handle); + BEGIN_RING(chan, eng2d, NV50_2D_OPERATION, 1); + OUT_RING (chan, NV50_2D_OPERATION_SRCCOPY); + BEGIN_RING(chan, eng2d, 0x0290, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, eng2d, 0x0888, 1); + OUT_RING (chan, 1); return 0; } -- cgit v1.2.3 From 28268f7b22905974a2cc3c099b1a842d20a7712a Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Tue, 8 Jul 2008 16:38:27 +0200 Subject: nv30: use native instructions. --- src/gallium/drivers/nv30/nv30_fragprog.c | 26 ++++---------------------- 1 file changed, 4 insertions(+), 22 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 54d6bea55f7..740d3ab430d 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -510,9 +510,7 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, break; // case TGSI_OPCODE_LIT: case TGSI_OPCODE_LRP: - tmp = temp(fpc); - arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]); - arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp); + arith(fpc, sat, LRP, dst, mask, src[0], src[1], src[2]); break; case TGSI_OPCODE_MAD: arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]); @@ -530,13 +528,7 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); break; case TGSI_OPCODE_POW: - tmp = temp(fpc); - arith(fpc, 0, LG2, tmp, MASK_X, - swz(src[0], X, X, X, X), none, none); - arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X), - swz(src[1], X, X, X, X), none); - arith(fpc, sat, EX2, dst, mask, - swz(tmp, X, X, X, X), none, none); + arith(fpc, sat, POW, dst, mask, src[0], src[1], none); break; case TGSI_OPCODE_RCP: arith(fpc, sat, RCP, dst, mask, src[0], none, none); @@ -545,20 +537,10 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, assert(0); break; case TGSI_OPCODE_RFL: - tmp = temp(fpc); - arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none); - arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none); - arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z, - swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none); - arith(fpc, sat, MAD, dst, mask, - swz(tmp, Z, Z, Z, Z), src[0], neg(src[1])); + arith(fpc, 0, RFL, dst, mask, src[0], src[1], none); break; case TGSI_OPCODE_RSQ: - tmp = temp(fpc); - arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X, - abs(swz(src[0], X, X, X, X)), none, none); - arith(fpc, sat, EX2, dst, mask, - neg(swz(tmp, X, X, X, X)), none, none); + arith(fpc, sat, RSQ, dst, mask, abs(swz(src[0], X, X, X, X)), none, none); break; case TGSI_OPCODE_SCS: if (mask & MASK_X) { -- cgit v1.2.3 From 511693d00c5f8641375efd16e08289586d62e562 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Tue, 8 Jul 2008 20:06:57 +0200 Subject: nv04: use FREE/MALLOC like a good gallium boy should. --- src/gallium/drivers/nv04/nv04_miptree.c | 6 +++--- src/gallium/drivers/nv04/nv04_prim_vbuf.c | 2 +- src/gallium/drivers/nv04/nv04_state.c | 8 ++++---- 3 files changed, 8 insertions(+), 8 deletions(-) diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c index ab2d2386bee..97f679731ea 100644 --- a/src/gallium/drivers/nv04/nv04_miptree.c +++ b/src/gallium/drivers/nv04/nv04_miptree.c @@ -62,7 +62,7 @@ nv04_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, mt->total_size); if (!mt->buffer) { - free(mt); + FREE(mt); return NULL; } @@ -83,9 +83,9 @@ nv04_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) pipe_buffer_reference(ws, &nv04mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv04mt->level[l].image_offset) - free(nv04mt->level[l].image_offset); + FREE(nv04mt->level[l].image_offset); } - free(nv04mt); + FREE(nv04mt); } } diff --git a/src/gallium/drivers/nv04/nv04_prim_vbuf.c b/src/gallium/drivers/nv04/nv04_prim_vbuf.c index 80a4dbc311c..d3963d1f59d 100644 --- a/src/gallium/drivers/nv04/nv04_prim_vbuf.c +++ b/src/gallium/drivers/nv04/nv04_prim_vbuf.c @@ -57,7 +57,7 @@ nv04_vbuf_render_allocate_vertices( struct vbuf_render *render, { struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render); - nv04_render->buffer = (unsigned char*) malloc(VERTEX_BUFFER_SIZE); + nv04_render->buffer = (unsigned char*) MALLOC(VERTEX_BUFFER_SIZE); assert(!nv04_render->buffer); return nv04_render->buffer; diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c index 39d7e8b42d0..d618465a201 100644 --- a/src/gallium/drivers/nv04/nv04_state.c +++ b/src/gallium/drivers/nv04/nv04_state.c @@ -14,7 +14,7 @@ nv04_blend_state_create(struct pipe_context *pipe, { struct nv04_blend_state *cb; - cb = malloc(sizeof(struct nv04_blend_state)); + cb = MALLOC(sizeof(struct nv04_blend_state)); cb->b_enable = cso->blend_enable ? 1 : 0; cb->b_src = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | @@ -81,7 +81,7 @@ nv04_sampler_state_create(struct pipe_context *pipe, struct nv04_sampler_state *ss; uint32_t filter = 0; - ss = malloc(sizeof(struct nv04_sampler_state)); + ss = MALLOC(sizeof(struct nv04_sampler_state)); ss->format = ((wrap_mode(cso->wrap_s) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT) | (wrap_mode(cso->wrap_t) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT)); @@ -178,7 +178,7 @@ nv04_rasterizer_state_create(struct pipe_context *pipe, * scissor * points/lines (no hw support, emulated with tris in gallium) */ - rs = malloc(sizeof(struct nv04_rasterizer_state)); + rs = MALLOC(sizeof(struct nv04_rasterizer_state)); rs->blend = cso->flatshade ? NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT : NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD; @@ -225,7 +225,7 @@ nv04_depth_stencil_alpha_state_create(struct pipe_context *pipe, { struct nv04_depth_stencil_alpha_state *hw; - hw = malloc(sizeof(struct nv04_depth_stencil_alpha_state)); + hw = MALLOC(sizeof(struct nv04_depth_stencil_alpha_state)); hw->control = float_to_ubyte(cso->alpha.ref); hw->control |= ( nv04_compare_func(cso->alpha.func) << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT ); -- cgit v1.2.3 From a62a738f3df054ef802f9ea7bc778d468fd0a4ba Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Tue, 8 Jul 2008 20:26:04 +0200 Subject: nv10: use the gallium alloc/free wrappers. --- src/gallium/drivers/nv10/nv10_state.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 11664fae2af..9b8b7801cdf 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -14,7 +14,7 @@ nv10_blend_state_create(struct pipe_context *pipe, { struct nv10_blend_state *cb; - cb = malloc(sizeof(struct nv10_blend_state)); + cb = MALLOC(sizeof(struct nv10_blend_state)); cb->b_enable = cso->blend_enable ? 1 : 0; cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | @@ -88,7 +88,7 @@ nv10_sampler_state_create(struct pipe_context *pipe, struct nv10_sampler_state *ps; uint32_t filter = 0; - ps = malloc(sizeof(struct nv10_sampler_state)); + ps = MALLOC(sizeof(struct nv10_sampler_state)); ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) | (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT)); @@ -249,7 +249,7 @@ nv10_rasterizer_state_create(struct pipe_context *pipe, * multisample * offset_units / offset_scale */ - rs = malloc(sizeof(struct nv10_rasterizer_state)); + rs = MALLOC(sizeof(struct nv10_rasterizer_state)); rs->templ = cso; @@ -335,7 +335,7 @@ nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, { struct nv10_depth_stencil_alpha_state *hw; - hw = malloc(sizeof(struct nv10_depth_stencil_alpha_state)); + hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state)); hw->depth.func = nvgl_comparison_op(cso->depth.func); hw->depth.write_enable = cso->depth.writemask ? 1 : 0; -- cgit v1.2.3 From 456550293a50d72c02ed5ec930c12899e7458a6f Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Wed, 9 Jul 2008 21:51:22 +0200 Subject: nv30: update nouveau_class.h --- src/gallium/drivers/nouveau/nouveau_class.h | 61 ++++++++++++++++++++++++----- 1 file changed, 52 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 3c29fa0d1be..4ec5062709e 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -3759,8 +3759,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_COLOR1_PITCH 0x0000021c #define NV34TCL_RT_ENABLE 0x00000220 #define NV34TCL_RT_ENABLE_MRT (1 << 4) -#define NV34TCL_RT_ENABLE_COLOR3 (1 << 3) -#define NV34TCL_RT_ENABLE_COLOR2 (1 << 2) #define NV34TCL_RT_ENABLE_COLOR1 (1 << 1) #define NV34TCL_RT_ENABLE_COLOR0 (1 << 0) #define NV34TCL_LMA_DEPTH_PITCH 0x0000022c @@ -3876,6 +3874,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA 0x80030000 #define NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA 0x80040000 #define NV34TCL_BLEND_COLOR 0x0000031c +#define NV34TCL_BLEND_COLOR_B_SHIFT 0 +#define NV34TCL_BLEND_COLOR_B_MASK 0x000000ff +#define NV34TCL_BLEND_COLOR_G_SHIFT 8 +#define NV34TCL_BLEND_COLOR_G_MASK 0x0000ff00 +#define NV34TCL_BLEND_COLOR_R_SHIFT 16 +#define NV34TCL_BLEND_COLOR_R_MASK 0x00ff0000 +#define NV34TCL_BLEND_COLOR_A_SHIFT 24 +#define NV34TCL_BLEND_COLOR_A_MASK 0xff000000 #define NV34TCL_BLEND_EQUATION 0x00000320 #define NV34TCL_BLEND_EQUATION_FUNC_ADD 0x00008006 #define NV34TCL_BLEND_EQUATION_MIN 0x00008007 @@ -4338,6 +4344,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_FRONT_FACE_CCW 0x00000901 #define NV34TCL_POLYGON_SMOOTH_ENABLE 0x00001838 #define NV34TCL_CULL_FACE_ENABLE 0x0000183c +#define NV34TCL_TX_PALETTE_OFFSET(x) (0x00001840+((x)*4)) +#define NV34TCL_TX_PALETTE_OFFSET__SIZE 0x00000004 #define NV34TCL_VTX_ATTR_2F_X(x) (0x00001880+((x)*8)) #define NV34TCL_VTX_ATTR_2F_X__SIZE 0x00000010 #define NV34TCL_VTX_ATTR_2F_Y(x) (0x00001884+((x)*8)) @@ -4376,7 +4384,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_FORMAT__SIZE 0x00000004 #define NV34TCL_TX_FORMAT_DMA0 (1 << 0) #define NV34TCL_TX_FORMAT_DMA1 (1 << 1) -#define NV34TCL_TX_FORMAT_CUBE_MAP (1 << 2) +#define NV34TCL_TX_FORMAT_CUBIC (1 << 2) +#define NV34TCL_TX_FORMAT_NO_BORDER (1 << 3) #define NV34TCL_TX_FORMAT_DIMS_SHIFT 4 #define NV34TCL_TX_FORMAT_DIMS_MASK 0x000000f0 #define NV34TCL_TX_FORMAT_DIMS_1D 0x00000010 @@ -4389,6 +4398,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000200 #define NV34TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000300 #define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000400 +#define NV34TCL_TX_FORMAT_FORMAT_R5G6B5 0x00000500 #define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000600 #define NV34TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000700 #define NV34TCL_TX_FORMAT_FORMAT_INDEX8 0x00000b00 @@ -4396,22 +4406,26 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_FORMAT_FORMAT_DXT3 0x00000e00 #define NV34TCL_TX_FORMAT_FORMAT_DXT5 0x00000f00 #define NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00001000 +#define NV34TCL_TX_FORMAT_FORMAT_R5G6B5_RECT 0x00001100 #define NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00001200 #define NV34TCL_TX_FORMAT_FORMAT_L8_RECT 0x00001300 #define NV34TCL_TX_FORMAT_FORMAT_A8L8 0x00001a00 #define NV34TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00001b00 -#define NV34TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 #define NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00001d00 +#define NV34TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00001e00 #define NV34TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00002000 #define NV34TCL_TX_FORMAT_FORMAT_A16 0x00003200 +#define NV34TCL_TX_FORMAT_FORMAT_HILO16 0x00003300 #define NV34TCL_TX_FORMAT_FORMAT_A16_RECT 0x00003500 +#define NV34TCL_TX_FORMAT_FORMAT_HILO16_RECT 0x00003600 +#define NV34TCL_TX_FORMAT_FORMAT_HILO8 0x00004400 +#define NV34TCL_TX_FORMAT_FORMAT_SIGNED_HILO8 0x00004500 +#define NV34TCL_TX_FORMAT_FORMAT_HILO8_RECT 0x00004600 +#define NV34TCL_TX_FORMAT_FORMAT_SIGNED_HILO8_RECT 0x00004700 #define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00004a00 #define NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00004b00 #define NV34TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00004c00 -#define NV34TCL_TX_FORMAT_NPOT (1 << 12) -#define NV34TCL_TX_FORMAT_RECT (1 << 14) -#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT 16 -#define NV34TCL_TX_FORMAT_MIPMAP_LEVELS_MASK 0x000f0000 +#define NV34TCL_TX_FORMAT_MIPMAP (1 << 19) #define NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT 20 #define NV34TCL_TX_FORMAT_BASE_SIZE_U_MASK 0x00f00000 #define NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT 24 @@ -4428,12 +4442,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER 0x00000004 #define NV34TCL_TX_WRAP_S_CLAMP 0x00000005 #define NV34TCL_TX_WRAP_T_SHIFT 8 -#define NV34TCL_TX_WRAP_T_MASK 0x0000ff00 +#define NV34TCL_TX_WRAP_T_MASK 0x00000f00 #define NV34TCL_TX_WRAP_T_REPEAT 0x00000100 #define NV34TCL_TX_WRAP_T_MIRRORED_REPEAT 0x00000200 #define NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE 0x00000300 #define NV34TCL_TX_WRAP_T_CLAMP_TO_BORDER 0x00000400 #define NV34TCL_TX_WRAP_T_CLAMP 0x00000500 +#define NV34TCL_TX_WRAP_EXPAND_NORMAL_SHIFT 12 +#define NV34TCL_TX_WRAP_EXPAND_NORMAL_MASK 0x0000f000 #define NV34TCL_TX_WRAP_R_SHIFT 16 #define NV34TCL_TX_WRAP_R_MASK 0x00ff0000 #define NV34TCL_TX_WRAP_R_REPEAT 0x00010000 @@ -4441,8 +4457,28 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE 0x00030000 #define NV34TCL_TX_WRAP_R_CLAMP_TO_BORDER 0x00040000 #define NV34TCL_TX_WRAP_R_CLAMP 0x00050000 +#define NV34TCL_TX_WRAP_RCOMP_SHIFT 28 +#define NV34TCL_TX_WRAP_RCOMP_MASK 0xf0000000 +#define NV34TCL_TX_WRAP_RCOMP_NEVER 0x00000000 +#define NV34TCL_TX_WRAP_RCOMP_GREATER 0x10000000 +#define NV34TCL_TX_WRAP_RCOMP_EQUAL 0x20000000 +#define NV34TCL_TX_WRAP_RCOMP_GEQUAL 0x30000000 +#define NV34TCL_TX_WRAP_RCOMP_LESS 0x40000000 +#define NV34TCL_TX_WRAP_RCOMP_NOTEQUAL 0x50000000 +#define NV34TCL_TX_WRAP_RCOMP_LEQUAL 0x60000000 +#define NV34TCL_TX_WRAP_RCOMP_ALWAYS 0x70000000 #define NV34TCL_TX_ENABLE(x) (0x00001a0c+((x)*32)) #define NV34TCL_TX_ENABLE__SIZE 0x00000004 +#define NV34TCL_TX_ENABLE_ANISO_SHIFT 4 +#define NV34TCL_TX_ENABLE_ANISO_MASK 0x00000030 +#define NV34TCL_TX_ENABLE_ANISO_NONE 0x00000000 +#define NV34TCL_TX_ENABLE_ANISO_2X 0x00000010 +#define NV34TCL_TX_ENABLE_ANISO_4X 0x00000020 +#define NV34TCL_TX_ENABLE_ANISO_8X 0x00000030 +#define NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT 14 +#define NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_MASK 0x0003c000 +#define NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT 26 +#define NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_MASK 0x3c000000 #define NV34TCL_TX_ENABLE_ENABLE (1 << 30) #define NV34TCL_TX_SWIZZLE(x) (0x00001a10+((x)*32)) #define NV34TCL_TX_SWIZZLE__SIZE 0x00000004 @@ -4494,6 +4530,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_TX_SWIZZLE_RECT_PITCH_MASK 0xffff0000 #define NV34TCL_TX_FILTER(x) (0x00001a14+((x)*32)) #define NV34TCL_TX_FILTER__SIZE 0x00000004 +#define NV34TCL_TX_FILTER_LOD_BIAS_SHIFT 8 +#define NV34TCL_TX_FILTER_LOD_BIAS_MASK 0x00000f00 #define NV34TCL_TX_FILTER_MINIFY_SHIFT 16 #define NV34TCL_TX_FILTER_MINIFY_MASK 0x000f0000 #define NV34TCL_TX_FILTER_MINIFY_NEAREST 0x00010000 @@ -4539,6 +4577,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_SHIFT 0 #define NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_MASK 0x0000000f #define NV34TCL_MULTISAMPLE_CONTROL 0x00001d7c +#define NV34TCL_MULTISAMPLE_CONTROL_ENABLE (1 << 0) +#define NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_ALPHA_TO_COVERAGE (1 << 4) +#define NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_ALPHA_TO_ONE (1 << 8) +#define NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_COVERAGE_SHIFT 16 +#define NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_COVERAGE_MASK 0xffff0000 #define NV34TCL_CLEAR_DEPTH_VALUE 0x00001d8c #define NV34TCL_CLEAR_COLOR_VALUE 0x00001d90 #define NV34TCL_CLEAR_COLOR_VALUE_B_SHIFT 0 -- cgit v1.2.3 From 7a838ef411d2f4716bdcbcad9f593541c43a2ce7 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Wed, 9 Jul 2008 22:03:59 +0200 Subject: nv30: Update defines from nouveau_class.h --- src/gallium/drivers/nv30/nv30_state.c | 46 ++++++++++++++--------------------- 1 file changed, 18 insertions(+), 28 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 97becd2be37..72b9515eb1a 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -120,38 +120,28 @@ nv30_sampler_state_create(struct pipe_context *pipe, ps = MALLOC(sizeof(struct nv30_sampler_state)); ps->fmt = 0; + /* TODO: Not all RECTs formats have this bit set, bits 15-8 of format + are the tx format to use. We should store normalized coord flag + in sampler state structure, and set appropriate format in + nvxx_fragtex_build() + */ if (!cso->normalized_coords) - ps->fmt |= NV34TCL_TX_FORMAT_RECT; + ps->fmt |= (1<<14) /*NV34TCL_TX_FORMAT_RECT*/; ps->wrap = ((wrap_mode(cso->wrap_s) << NV34TCL_TX_WRAP_S_SHIFT) | (wrap_mode(cso->wrap_t) << NV34TCL_TX_WRAP_T_SHIFT) | (wrap_mode(cso->wrap_r) << NV34TCL_TX_WRAP_R_SHIFT)); ps->en = 0; + + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV34TCL_TX_ENABLE_ANISO_4X; + } else if (cso->max_anisotropy >= 2.0) { - /* no idea, binary driver sets it, works without it.. meh.. */ - ps->wrap |= (1 << 5); - -/* if (cso->max_anisotropy >= 16.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_16X; - } else - if (cso->max_anisotropy >= 12.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_12X; - } else - if (cso->max_anisotropy >= 10.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_10X; - } else - if (cso->max_anisotropy >= 8.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_8X; - } else - if (cso->max_anisotropy >= 6.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_6X; - } else - if (cso->max_anisotropy >= 4.0) { - ps->en |= NV34TCL_TX_ENABLE_ANISO_4X; - } else { - ps->en |= NV34TCL_TX_ENABLE_ANISO_2X; - }*/ + ps->en |= NV34TCL_TX_ENABLE_ANISO_2X; } switch (cso->mag_img_filter) { @@ -198,7 +188,7 @@ nv30_sampler_state_create(struct pipe_context *pipe, ps->filt = filter; - /*{ + { float limit; limit = CLAMP(cso->lod_bias, -16.0, 15.0); @@ -209,9 +199,9 @@ nv30_sampler_state_create(struct pipe_context *pipe, limit = CLAMP(cso->min_lod, 0.0, 15.0); ps->en |= (int)(limit * 256.0) << 19; - }*/ + } -/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { switch (cso->compare_func) { case PIPE_FUNC_NEVER: ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NEVER; @@ -240,7 +230,7 @@ nv30_sampler_state_create(struct pipe_context *pipe, default: break; } - }*/ + } ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | (float_to_ubyte(cso->border_color[0]) << 16) | -- cgit v1.2.3 From 2fed964a27f208f69edc0b2d9a2d0b0717184ee5 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Wed, 9 Jul 2008 22:13:01 +0200 Subject: nv30: Update defines from nouveau_class.h --- src/gallium/drivers/nv30/nv30_fragtex.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index 47f3ff047db..92b3f939297 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -87,6 +87,7 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) struct nv30_texture_format *tf; uint32_t txf, txs, txp; int swizzled = 0; /*XXX: implement in region code? */ + unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; tf = nv30_fragtex_format(pt->format); if (!tf || !tf->defined) { @@ -99,20 +100,20 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) txf |= log2i(pt->width[0]) << 20; txf |= log2i(pt->height[0]) << 24; txf |= log2i(pt->depth[0]) << 28; - txf |= 8; + txf |= NV34TCL_TX_FORMAT_NO_BORDER; switch (pt->target) { -/* case PIPE_TEXTURE_CUBE: - txf |= NV34TCL_TEX_FORMAT_CUBIC;*/ + case PIPE_TEXTURE_CUBE: + txf |= NV34TCL_TX_FORMAT_CUBIC; /* fall-through */ case PIPE_TEXTURE_2D: - txf |= (2<<4); + txf |= NV34TCL_TX_FORMAT_DIMS_2D; break; case PIPE_TEXTURE_3D: - txf |= (3<<4); + txf |= NV34TCL_TX_FORMAT_DIMS_3D; break; case PIPE_TEXTURE_1D: - txf |= (1<<4); + txf |= NV34TCL_TX_FORMAT_DIMS_1D; break; default: NOUVEAU_ERR("Unknown target %d\n", pt->target); @@ -122,13 +123,13 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) txs = tf->swizzle; BEGIN_RING(rankine, NV34TCL_TX_OFFSET(unit), 8); - OUT_RELOCl(nv30mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv30mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RELOCl(nv30mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW); + OUT_RELOCd(nv30mt->buffer,txf, tex_flags | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); OUT_RING (ps->wrap); - OUT_RING (0x40000000); /* enable */ + OUT_RING (NV34TCL_TX_ENABLE_ENABLE | ps->en); OUT_RING (txs); OUT_RING (ps->filt | 0x2000 /* magic */); - OUT_RING ((pt->width[0] << 16) | pt->height[0]); + OUT_RING ((pt->width[0] << NV34TCL_TX_NPOT_SIZE_W_SHIFT) | pt->height[0]); OUT_RING (ps->bcol); } -- cgit v1.2.3 From 225863aeb5f2dfe4980ae5887f5623ecb05e9ced Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Wed, 9 Jul 2008 23:23:39 +0200 Subject: nv30: min/max lod are used for mipmap, there is just enable bit in tx_format --- src/gallium/drivers/nv30/nv30_fragtex.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index 92b3f939297..f253087c884 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -96,11 +96,11 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) } txf = tf->format << 8; - txf |= (pt->last_level + 1) << 16; + txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0); txf |= log2i(pt->width[0]) << 20; txf |= log2i(pt->height[0]) << 24; txf |= log2i(pt->depth[0]) << 28; - txf |= NV34TCL_TX_FORMAT_NO_BORDER; + txf |= NV34TCL_TX_FORMAT_NO_BORDER | 0x10000; switch (pt->target) { case PIPE_TEXTURE_CUBE: -- cgit v1.2.3 From 861629d1fd4a1d256c913470c33d9522e83d615d Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 00:15:01 +1000 Subject: nv50: more "abuse" by using libc malloc etc.. --- src/gallium/drivers/nv50/nv50_context.c | 2 +- src/gallium/drivers/nv50/nv50_draw.c | 2 +- src/gallium/drivers/nv50/nv50_program.c | 14 +++++++------- 3 files changed, 9 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index a225c4bf728..2494784a94e 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -21,7 +21,7 @@ nv50_destroy(struct pipe_context *pipe) struct nv50_context *nv50 = (struct nv50_context *)pipe; draw_destroy(nv50->draw); - free(nv50); + FREE(nv50); } diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c index d185d999502..2b0c0384a37 100644 --- a/src/gallium/drivers/nv50/nv50_draw.c +++ b/src/gallium/drivers/nv50/nv50_draw.c @@ -46,7 +46,7 @@ nv50_render_reset_stipple_counter(struct draw_stage *stage) static void nv50_render_destroy(struct draw_stage *stage) { - free(stage); + FREE(stage); } struct draw_stage * diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index cd6967b3668..ff81719f03e 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -180,8 +180,8 @@ kill_temp_temp(struct nv50_pc *pc) static int ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w) { - pc->immd_buf = realloc(pc->immd_buf, (pc->immd_nr + 1) * 4 * - sizeof(float)); + pc->immd_buf = REALLOC(pc->immd_buf, (pc->immd_nr * r * sizeof(float)), + (pc->immd_nr + 1) * 4 * sizeof(float)); pc->immd_buf[(pc->immd_nr * 4) + 0] = x; pc->immd_buf[(pc->immd_nr * 4) + 1] = y; pc->immd_buf[(pc->immd_nr * 4) + 2] = z; @@ -1276,7 +1276,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) } if (pc->temp_nr) { - pc->temp = calloc(pc->temp_nr * 4, sizeof(struct nv50_reg)); + pc->temp = CALLOC(pc->temp_nr * 4, sizeof(struct nv50_reg)); if (!pc->temp) goto out_err; @@ -1293,7 +1293,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) struct nv50_reg *iv = NULL; int aid = 0; - pc->attr = calloc(pc->attr_nr * 4, sizeof(struct nv50_reg)); + pc->attr = CALLOC(pc->attr_nr * 4, sizeof(struct nv50_reg)); if (!pc->attr) goto out_err; @@ -1339,7 +1339,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->result_nr) { int rid = 0; - pc->result = calloc(pc->result_nr * 4, sizeof(struct nv50_reg)); + pc->result = CALLOC(pc->result_nr * 4, sizeof(struct nv50_reg)); if (!pc->result) goto out_err; @@ -1360,7 +1360,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->param_nr) { int rid = 0; - pc->param = calloc(pc->param_nr * 4, sizeof(struct nv50_reg)); + pc->param = CALLOC(pc->param_nr * 4, sizeof(struct nv50_reg)); if (!pc->param) goto out_err; @@ -1376,7 +1376,7 @@ nv50_program_tx_prep(struct nv50_pc *pc) if (pc->immd_nr) { int rid = pc->param_nr * 4; - pc->immd = calloc(pc->immd_nr * 4, sizeof(struct nv50_reg)); + pc->immd = CALLOC(pc->immd_nr * 4, sizeof(struct nv50_reg)); if (!pc->immd) goto out_err; -- cgit v1.2.3 From add89c78455f04654c3706d46e3d3e6b92b73b71 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 00:21:42 +1000 Subject: nouveau: update to latest object header --- src/gallium/drivers/nouveau/nouveau_class.h | 52 +++++++++++++++++++++-------- src/gallium/drivers/nv50/nv50_screen.c | 12 +++---- 2 files changed, 44 insertions(+), 20 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 4ec5062709e..18d8d3677db 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -1798,6 +1798,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5 0x00000100 #define NV10TCL_TX_FORMAT_FORMAT_A8_RECT 0x00000180 #define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4 0x00000200 +#define NV10TCL_TX_FORMAT_FORMAT_R5G6B5 0x00000280 #define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8 0x00000300 #define NV10TCL_TX_FORMAT_FORMAT_X8R8G8B8 0x00000380 #define NV10TCL_TX_FORMAT_FORMAT_INDEX8 0x00000580 @@ -1805,15 +1806,22 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV10TCL_TX_FORMAT_FORMAT_DXT3 0x00000700 #define NV10TCL_TX_FORMAT_FORMAT_DXT5 0x00000780 #define NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT 0x00000800 +#define NV10TCL_TX_FORMAT_FORMAT_R5G6B5_RECT 0x00000880 #define NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT 0x00000900 #define NV10TCL_TX_FORMAT_FORMAT_L8_RECT 0x00000980 #define NV10TCL_TX_FORMAT_FORMAT_A8L8 0x00000d00 #define NV10TCL_TX_FORMAT_FORMAT_A8_RECT2 0x00000d80 -#define NV10TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 #define NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT 0x00000e80 +#define NV10TCL_TX_FORMAT_FORMAT_R8G8B8_RECT 0x00000f00 #define NV10TCL_TX_FORMAT_FORMAT_L8A8_RECT 0x00001000 #define NV10TCL_TX_FORMAT_FORMAT_A16 0x00001900 +#define NV10TCL_TX_FORMAT_FORMAT_HILO16 0x00001980 #define NV10TCL_TX_FORMAT_FORMAT_A16_RECT 0x00001a80 +#define NV10TCL_TX_FORMAT_FORMAT_HILO16_RECT 0x00001b00 +#define NV10TCL_TX_FORMAT_FORMAT_HILO8 0x00002200 +#define NV10TCL_TX_FORMAT_FORMAT_SIGNED_HILO8 0x00002280 +#define NV10TCL_TX_FORMAT_FORMAT_HILO8_RECT 0x00002300 +#define NV10TCL_TX_FORMAT_FORMAT_SIGNED_HILO8_RECT 0x00002380 #define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV 0x00002500 #define NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV 0x00002580 #define NV10TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV 0x00002600 @@ -5486,18 +5494,23 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50TCL_NOP 0x00000100 #define NV50TCL_NOTIFY 0x00000104 #define NV50TCL_DMA_NOTIFY 0x00000180 -#define NV50TCL_DMA_IN_MEMORY0(x) (0x00000184+((x)*4)) -#define NV50TCL_DMA_IN_MEMORY0__SIZE 0x0000000b -#define NV50TCL_DMA_IN_MEMORY1(x) (0x000001c0+((x)*4)) -#define NV50TCL_DMA_IN_MEMORY1__SIZE 0x00000008 +#define NV50TCL_DMA_UNK0(x) (0x00000184+((x)*4)) +#define NV50TCL_DMA_UNK0__SIZE 0x0000000b +#define NV50TCL_DMA_UNK1(x) (0x000001c0+((x)*4)) +#define NV50TCL_DMA_UNK1__SIZE 0x00000008 #define NV50TCL_RT_ADDRESS_HIGH(x) (0x00000200+((x)*32)) #define NV50TCL_RT_ADDRESS_HIGH__SIZE 0x00000008 #define NV50TCL_RT_ADDRESS_LOW(x) (0x00000204+((x)*32)) #define NV50TCL_RT_ADDRESS_LOW__SIZE 0x00000008 #define NV50TCL_RT_FORMAT(x) (0x00000208+((x)*32)) #define NV50TCL_RT_FORMAT__SIZE 0x00000008 -#define NV50TCL_RT_UNK3(x) (0x0000020c+((x)*32)) -#define NV50TCL_RT_UNK3__SIZE 0x00000008 +#define NV50TCL_RT_FORMAT_32BPP 0x000000cf +#define NV50TCL_RT_FORMAT_24BPP 0x000000e6 +#define NV50TCL_RT_FORMAT_16BPP 0x000000e8 +#define NV50TCL_RT_FORMAT_8BPP 0x000000f3 +#define NV50TCL_RT_FORMAT_15BPP 0x000000f8 +#define NV50TCL_RT_TILE_UNK(x) (0x0000020c+((x)*32)) +#define NV50TCL_RT_TILE_UNK__SIZE 0x00000008 #define NV50TCL_RT_UNK4(x) (0x00000210+((x)*32)) #define NV50TCL_RT_UNK4__SIZE 0x00000008 #define NV50TCL_VTX_ATTR_1F(x) (0x00000300+((x)*4)) @@ -5602,9 +5615,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50TCL_SCISSOR_VERT_T_MASK 0x0000ffff #define NV50TCL_SCISSOR_VERT_B_SHIFT 16 #define NV50TCL_SCISSOR_VERT_B_MASK 0xffff0000 -#define NV50TCL_VP_UPLOAD_CONST_ID 0x00000f00 -#define NV50TCL_VP_UPLOAD_CONST(x) (0x00000f04+((x)*4)) -#define NV50TCL_VP_UPLOAD_CONST__SIZE 0x00000010 +#define NV50TCL_CB_ADDR 0x00000f00 +#define NV50TCL_CB_ADDR_ID_SHIFT 8 +#define NV50TCL_CB_ADDR_ID_MASK 0xffffff00 +#define NV50TCL_CB_ADDR_BUFFER_SHIFT 0 +#define NV50TCL_CB_ADDR_BUFFER_MASK 0x000000ff +#define NV50TCL_CB_DATA(x) (0x00000f04+((x)*4)) +#define NV50TCL_CB_DATA__SIZE 0x00000010 #define NV50TCL_STENCIL_FRONT_FUNC_REF 0x00000f54 #define NV50TCL_STENCIL_FRONT_MASK 0x00000f58 #define NV50TCL_STENCIL_FRONT_FUNC_MASK 0x00000f5c @@ -5626,6 +5643,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50TCL_RT_HORIZ__SIZE 0x00000008 #define NV50TCL_RT_VERT(x) (0x00001244+((x)*8)) #define NV50TCL_RT_VERT__SIZE 0x00000008 +#define NV50TCL_CB_DEF_ADDRESS_HIGH 0x00001280 +#define NV50TCL_CB_DEF_ADDRESS_LOW 0x00001284 +#define NV50TCL_CB_DEF_SET 0x00001288 +#define NV50TCL_CB_DEF_SET_SIZE_SHIFT 0 +#define NV50TCL_CB_DEF_SET_SIZE_MASK 0x0000ffff +#define NV50TCL_CB_DEF_SET_BUFFER_SHIFT 16 +#define NV50TCL_CB_DEF_SET_BUFFER_MASK 0xffff0000 #define NV50TCL_DEPTH_TEST_ENABLE 0x000012cc #define NV50TCL_SHADE_MODEL 0x000012d4 #define NV50TCL_SHADE_MODEL_FLAT 0x00001d00 @@ -5779,12 +5803,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV50TCL_GP_START_ID 0x00001410 #define NV50TCL_FP_START_ID 0x00001414 #define NV50TCL_POINT_SIZE 0x00001518 -#define NV50TCL_TEX_CB0_ADDRESS_HIGH 0x0000155c -#define NV50TCL_TEX_CB0_ADDRESS_LOW 0x00001560 +#define NV50TCL_TSC_ADDRESS_HIGH 0x0000155c +#define NV50TCL_TSC_ADDRESS_LOW 0x00001560 #define NV50TCL_POLYGON_OFFSET_FACTOR 0x0000156c #define NV50TCL_LINE_SMOOTH_ENABLE 0x00001570 -#define NV50TCL_TEX_CB1_ADDRESS_HIGH 0x00001574 -#define NV50TCL_TEX_CB1_ADDRESS_LOW 0x00001578 +#define NV50TCL_TIC_ADDRESS_HIGH 0x00001574 +#define NV50TCL_TIC_ADDRESS_LOW 0x00001578 #define NV50TCL_STENCIL_FRONT_ENABLE 0x00001594 #define NV50TCL_STENCIL_FRONT_OP_FAIL 0x00001598 #define NV50TCL_STENCIL_FRONT_OP_FAIL_ZERO 0x00000000 diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 5b4c5f96ac3..e4c09a99d8b 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -189,13 +189,13 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) so_data (so, 1); so_method(so, screen->tesla, NV50TCL_DMA_NOTIFY, 1); so_data (so, screen->sync->handle); - so_method(so, screen->tesla, NV50TCL_DMA_IN_MEMORY0(0), - NV50TCL_DMA_IN_MEMORY0__SIZE); - for (i = 0; i < NV50TCL_DMA_IN_MEMORY0__SIZE; i++) + so_method(so, screen->tesla, NV50TCL_DMA_UNK0(0), + NV50TCL_DMA_UNK0__SIZE); + for (i = 0; i < NV50TCL_DMA_UNK0__SIZE; i++) so_data(so, nvws->channel->vram->handle); - so_method(so, screen->tesla, NV50TCL_DMA_IN_MEMORY1(0), - NV50TCL_DMA_IN_MEMORY1__SIZE); - for (i = 0; i < NV50TCL_DMA_IN_MEMORY1__SIZE; i++) + so_method(so, screen->tesla, NV50TCL_DMA_UNK1(0), + NV50TCL_DMA_UNK1__SIZE); + for (i = 0; i < NV50TCL_DMA_UNK1__SIZE; i++) so_data(so, nvws->channel->vram->handle); so_method(so, screen->tesla, 0x121c, 1); so_data (so, 1); -- cgit v1.2.3 From 7a81ffa1154e6efd09fe91a944514a6f56369b4a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 01:53:28 +1000 Subject: nv50: get clear-scissor working --- src/gallium/drivers/nv50/nv50_context.h | 1 + src/gallium/drivers/nv50/nv50_state_validate.c | 44 ++++++++++++++++++++------ 2 files changed, 36 insertions(+), 9 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 4ea01009fb6..9b77fc1225d 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -86,6 +86,7 @@ struct nv50_state { struct nouveau_stateobj *scissor; unsigned scissor_enabled; struct nouveau_stateobj *viewport; + unsigned viewport_bypass; struct nouveau_stateobj *tsc_upload; struct nouveau_stateobj *tic_upload; struct nouveau_stateobj *vertprog; diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 88ef685bb1f..bd1f048f878 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -228,17 +228,43 @@ nv50_state_validate(struct nv50_context *nv50) scissor_uptodate: if (nv50->dirty & NV50_NEW_VIEWPORT) { - so = so_new(8, 0); - so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3); - so_data (so, fui(nv50->viewport.translate[0])); - so_data (so, fui(nv50->viewport.translate[1])); - so_data (so, fui(nv50->viewport.translate[2])); - so_method(so, tesla, NV50TCL_VIEWPORT_UNK0(0), 3); - so_data (so, fui(nv50->viewport.scale[0])); - so_data (so, fui(-nv50->viewport.scale[1])); - so_data (so, fui(nv50->viewport.scale[2])); + unsigned bypass; + + if (!nv50->rasterizer->pipe.bypass_clipping) + bypass = 0; + else + bypass = 1; + + if (nv50->state.viewport && + (bypass || !(nv50->dirty & NV50_NEW_VIEWPORT)) && + nv50->state.viewport_bypass == bypass) + goto viewport_uptodate; + nv50->state.viewport_bypass = bypass; + + so = so_new(12, 0); + if (!bypass) { + so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3); + so_data (so, fui(nv50->viewport.translate[0])); + so_data (so, fui(nv50->viewport.translate[1])); + so_data (so, fui(nv50->viewport.translate[2])); + so_method(so, tesla, NV50TCL_VIEWPORT_UNK0(0), 3); + so_data (so, fui(nv50->viewport.scale[0])); + so_data (so, fui(-nv50->viewport.scale[1])); + so_data (so, fui(nv50->viewport.scale[2])); + so_method(so, tesla, 0x192c, 1); + so_data (so, 1); + so_method(so, tesla, 0x0f90, 1); + so_data (so, 0); + } else { + so_method(so, tesla, 0x192c, 1); + so_data (so, 0); + so_method(so, tesla, 0x0f90, 1); + so_data (so, 1); + } + so_ref(so, &nv50->state.viewport); } +viewport_uptodate: if (nv50->dirty & NV50_NEW_SAMPLER) { int i; -- cgit v1.2.3 From 17e95de85250540c8e2448f138d81b7b055be511 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 02:01:23 +1000 Subject: nv50: enable GART usage for vertex buffers AKA "I can haz fast b0rk3d glxgears!!!" --- src/gallium/drivers/nv50/nv50_screen.c | 1 + src/gallium/drivers/nv50/nv50_vbo.c | 8 ++++---- 2 files changed, 5 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index e4c09a99d8b..9f1fba6e42d 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -90,6 +90,7 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param) case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; case NOUVEAU_CAP_HW_VTXBUF: + return 1; case NOUVEAU_CAP_HW_IDXBUF: return 0; default: diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index c16bfe3a937..0318b9b5503 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -204,11 +204,11 @@ nv50_vbo_validate(struct nv50_context *nv50) so_method(vtxbuf, tesla, 0x900 + (i * 16), 3); so_data (vtxbuf, 0x20000000 | vb->pitch); so_reloc (vtxbuf, vb->buffer, vb->buffer_offset + - ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | - NOUVEAU_BO_HIGH, 0, 0); + ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (vtxbuf, vb->buffer, vb->buffer_offset + - ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | - NOUVEAU_BO_LOW, 0, 0); + ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); } so_ref (vtxfmt, &nv50->state.vtxfmt); -- cgit v1.2.3 From 857a3294a959015bf893241199f7fd7f7882a6ab Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 20:44:39 +1000 Subject: nv50: add license headers to .c files --- src/gallium/drivers/nv50/nv50_clear.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_context.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_draw.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_miptree.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_program.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_query.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_screen.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_state.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_state_validate.c | 22 ++++++++++++ src/gallium/drivers/nv50/nv50_surface.c | 46 +++++++++++--------------- src/gallium/drivers/nv50/nv50_vbo.c | 22 ++++++++++++ 11 files changed, 240 insertions(+), 26 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_clear.c b/src/gallium/drivers/nv50/nv50_clear.c index f35087b37ef..fbc6cd09cee 100644 --- a/src/gallium/drivers/nv50/nv50_clear.c +++ b/src/gallium/drivers/nv50/nv50_clear.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 2494784a94e..07987c7d02d 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "draw/draw_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c index 2b0c0384a37..4fd81bd27a1 100644 --- a/src/gallium/drivers/nv50/nv50_draw.c +++ b/src/gallium/drivers/nv50/nv50_draw.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "draw/draw_pipe.h" #include "pipe/p_util.h" diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index f936a1f0e20..8b61ca2a1fa 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index ff81719f03e..2668acf2999 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" diff --git a/src/gallium/drivers/nv50/nv50_query.c b/src/gallium/drivers/nv50/nv50_query.c index d8c3491c2c5..26bd90ccc5e 100644 --- a/src/gallium/drivers/nv50/nv50_query.c +++ b/src/gallium/drivers/nv50/nv50_query.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_context.h" #include "nv50_context.h" diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 9f1fba6e42d..ec6d34e6845 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_screen.h" #include "pipe/p_util.h" diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 743de089d6f..5c364c4ddff 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_state.h" #include "pipe/p_defines.h" #include "pipe/p_util.h" diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index bd1f048f878..5c75a242cc6 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "nv50_context.h" #include "nouveau/nouveau_stateobj.h" diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index acd7b501ef4..6229b636260 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -1,30 +1,24 @@ - -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * +/* + * Copyright 2008 Ben Skeggs + * * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ #include "nv50_context.h" #include "pipe/p_defines.h" diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 0318b9b5503..8e420ad172e 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -1,3 +1,25 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + #include "pipe/p_context.h" #include "pipe/p_state.h" #include "pipe/p_util.h" -- cgit v1.2.3 From 62100692b8ad7fa868743c4698dac109beceaf7f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 20:49:06 +1000 Subject: nv50: split tic construction out into own file --- src/gallium/drivers/nv50/Makefile | 1 + src/gallium/drivers/nv50/nv50_context.h | 3 + src/gallium/drivers/nv50/nv50_state_validate.c | 27 +----- src/gallium/drivers/nv50/nv50_tex.c | 57 ++++++++++++ src/gallium/drivers/nv50/nv50_texture.h | 124 +++++++++++++++++++++++++ 5 files changed, 187 insertions(+), 25 deletions(-) create mode 100644 src/gallium/drivers/nv50/nv50_tex.c create mode 100644 src/gallium/drivers/nv50/nv50_texture.h diff --git a/src/gallium/drivers/nv50/Makefile b/src/gallium/drivers/nv50/Makefile index 2cc6e9de287..be30400c039 100644 --- a/src/gallium/drivers/nv50/Makefile +++ b/src/gallium/drivers/nv50/Makefile @@ -14,6 +14,7 @@ DRIVER_SOURCES = \ nv50_state.c \ nv50_state_validate.c \ nv50_surface.c \ + nv50_tex.c \ nv50_vbo.c C_SOURCES = \ diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 9b77fc1225d..cb51fb02ebd 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -164,4 +164,7 @@ extern void nv50_program_destroy(struct nv50_context *nv50, struct nv50_program /* nv50_state_validate.c */ extern boolean nv50_state_validate(struct nv50_context *nv50); +/* nv50_tex.c */ +extern void nv50_tex_validate(struct nv50_context *); + #endif diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 5c75a242cc6..8f9ee05acc8 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -300,31 +300,8 @@ viewport_uptodate: so_ref(so, &nv50->state.tsc_upload); } - if (nv50->dirty & NV50_NEW_TEXTURE) { - int i; - - so = so_new(nv50->miptree_nr * 8 + 3, nv50->miptree_nr * 2); - so_method(so, tesla, 0x0f00, 1); - so_data (so, NV50_CB_TIC); - so_method(so, tesla, 0x40000f04, nv50->miptree_nr * 8); - for (i = 0; i < nv50->miptree_nr; i++) { - struct nv50_miptree *mt = nv50->miptree[i]; - - so_data (so, 0x2a712488); - so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0xd0005000); - so_data (so, 0x00300000); - so_data (so, mt->base.width[0]); - so_data (so, (mt->base.depth[0] << 16) | - mt->base.height[0]); - so_data (so, 0x03000000); - so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_HIGH, 0, 0); - } - - so_ref(so, &nv50->state.tic_upload); - } + if (nv50->dirty & NV50_NEW_TEXTURE) + nv50_tex_validate(nv50); if (nv50->dirty & NV50_NEW_ARRAYS) nv50_vbo_validate(nv50); diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c new file mode 100644 index 00000000000..581a9e2ffb6 --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_tex.c @@ -0,0 +1,57 @@ +/* + * Copyright 2008 Ben Skeggs + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include "nv50_context.h" +#include "nv50_texture.h" + +#include "nouveau/nouveau_stateobj.h" + +void +nv50_tex_validate(struct nv50_context *nv50) +{ + struct nouveau_grobj *tesla = nv50->screen->tesla; + struct nouveau_stateobj *so; + int i; + + so = so_new(nv50->miptree_nr * 8 + 3, nv50->miptree_nr * 2); + so_method(so, tesla, 0x0f00, 1); + so_data (so, NV50_CB_TIC); + so_method(so, tesla, 0x40000f04, nv50->miptree_nr * 8); + for (i = 0; i < nv50->miptree_nr; i++) { + struct nv50_miptree *mt = nv50->miptree[i]; + + so_data (so, 0x2a712488); + so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0xd0005000); + so_data (so, 0x00300000); + so_data (so, mt->base.width[0]); + so_data (so, (mt->base.depth[0] << 16) | + mt->base.height[0]); + so_data (so, 0x03000000); + so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_HIGH, 0, 0); + } + + so_ref(so, &nv50->state.tic_upload); +} + diff --git a/src/gallium/drivers/nv50/nv50_texture.h b/src/gallium/drivers/nv50/nv50_texture.h new file mode 100644 index 00000000000..c8fb282f21f --- /dev/null +++ b/src/gallium/drivers/nv50/nv50_texture.h @@ -0,0 +1,124 @@ +#ifndef __NV50_TEXTURE_H__ +#define __NV50_TEXTURE_H__ + +/* It'd be really nice to have these in nouveau_class.h generated by + * renouveau like the rest of the object header - but not sure it can + * handle non-object stuff nicely - need to look into it. + */ + +/* Texture image control block */ +#define NV50TIC_0_0_MAPA_MASK 0x38000000 +#define NV50TIC_0_0_MAPA_ZERO 0x00000000 +#define NV50TIC_0_0_MAPA_C0 0x10000000 +#define NV50TIC_0_0_MAPA_C1 0x18000000 +#define NV50TIC_0_0_MAPA_C2 0x20000000 +#define NV50TIC_0_0_MAPA_C3 0x28000000 +#define NV50TIC_0_0_MAPA_ONE 0x38000000 +#define NV50TIC_0_0_MAPR_MASK 0x07000000 +#define NV50TIC_0_0_MAPR_ZERO 0x00000000 +#define NV50TIC_0_0_MAPR_C0 0x02000000 +#define NV50TIC_0_0_MAPR_C1 0x03000000 +#define NV50TIC_0_0_MAPR_C2 0x04000000 +#define NV50TIC_0_0_MAPR_C3 0x05000000 +#define NV50TIC_0_0_MAPR_ONE 0x07000000 +#define NV50TIC_0_0_MAPG_MASK 0x00e00000 +#define NV50TIC_0_0_MAPG_ZERO 0x00000000 +#define NV50TIC_0_0_MAPG_C0 0x00400000 +#define NV50TIC_0_0_MAPG_C1 0x00600000 +#define NV50TIC_0_0_MAPG_C2 0x00800000 +#define NV50TIC_0_0_MAPG_C3 0x00a00000 +#define NV50TIC_0_0_MAPG_ONE 0x00e00000 +#define NV50TIC_0_0_MAPB_MASK 0x001c0000 +#define NV50TIC_0_0_MAPB_ZERO 0x00000000 +#define NV50TIC_0_0_MAPB_C0 0x00080000 +#define NV50TIC_0_0_MAPB_C1 0x000c0000 +#define NV50TIC_0_0_MAPB_C2 0x00100000 +#define NV50TIC_0_0_MAPB_C3 0x00140000 +#define NV50TIC_0_0_MAPB_ONE 0x001c0000 +#define NV50TIC_0_0_TYPEA_MASK 0x00038000 +#define NV50TIC_0_0_TYPEA_UNORM 0x00010000 +#define NV50TIC_0_0_TYPER_MASK 0x00007000 +#define NV50TIC_0_0_TYPER_UNORM 0x00002000 +#define NV50TIC_0_0_TYPEG_MASK 0x00000e00 +#define NV50TIC_0_0_TYPEG_UNORM 0x00000400 +#define NV50TIC_0_0_TYPEB_MASK 0x000001c0 +#define NV50TIC_0_0_TYPEB_UNORM 0x00000080 +#define NV50TIC_0_0_FMT_MASK 0x0000003c +#define NV50TIC_0_0_FMT_8_8_8_8 0x00000008 +#define NV50TIC_0_0_FMT_5_6_5 0x00000015 +#define NV50TIC_0_0_FMT_8_8 0x00000018 +#define NV50TIC_0_0_FMT_8 0x0000001d + +#define NV50TIC_0_1_OFFSET_LOW_MASK 0xffffffff +#define NV50TIC_0_1_OFFSET_LOW_SHIFT 0 + +#define NV50TIC_0_2_UNKNOWN_MASK 0xffffffff + +#define NV50TIC_0_3_UNKNOWN_MASK 0xffffffff + +#define NV50TIC_0_4_WIDTH_MASK 0x0000ffff +#define NV50TIC_0_4_WIDTH_SHIFT 0 + +#define NV50TIC_0_5_DEPTH_MASK 0xffff0000 +#define NV50TIC_0_5_DEPTH_SHIFT 16 +#define NV50TIC_0_5_HEIGHT_MASK 0x0000ffff +#define NV50TIC_0_5_HEIGHT_SHIFT 0 + +#define NV50TIC_0_6_UNKNOWN_MASK 0xffffffff + +#define NV50TIC_0_7_OFFSET_HIGH_MASK 0xffffffff +#define NV50TIC_0_7_OFFSET_HIGH_SHIFT 0 + +/* Texture sampler control block */ +#define NV50TSC_1_0_WRAPS_MASK 0x00000007 +#define NV50TSC_1_0_WRAPS_REPEAT 0x00000000 +#define NV50TSC_1_0_WRAPS_MIRROR_REPEAT 0x00000001 +#define NV50TSC_1_0_WRAPS_CLAMP_TO_EDGE 0x00000002 +#define NV50TSC_1_0_WRAPS_CLAMP_TO_BORDER 0x00000003 +#define NV50TSC_1_0_WRAPS_CLAMP 0x00000004 +#define NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_EDGE 0x00000005 +#define NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_BORDER 0x00000006 +#define NV50TSC_1_0_WRAPS_MIRROR_CLAMP 0x00000007 +#define NV50TSC_1_0_WRAPT_MASK 0x00000038 +#define NV50TSC_1_0_WRAPT_REPEAT 0x00000000 +#define NV50TSC_1_0_WRAPT_MIRROR_REPEAT 0x00000008 +#define NV50TSC_1_0_WRAPT_CLAMP_TO_EDGE 0x00000010 +#define NV50TSC_1_0_WRAPT_CLAMP_TO_BORDER 0x00000018 +#define NV50TSC_1_0_WRAPT_CLAMP 0x00000020 +#define NV50TSC_1_0_WRAPT_MIRROR_CLAMP_TO_EDGE 0x00000028 +#define NV50TSC_1_0_WRAPT_MIRROR_CLAMP_TO_BORDER 0x00000030 +#define NV50TSC_1_0_WRAPT_MIRROR_CLAMP 0x00000038 +#define NV50TSC_1_0_WRAPR_MASK 0x000001c0 +#define NV50TSC_1_0_WRAPR_REPEAT 0x00000000 +#define NV50TSC_1_0_WRAPR_MIRROR_REPEAT 0x00000040 +#define NV50TSC_1_0_WRAPR_CLAMP_TO_EDGE 0x00000080 +#define NV50TSC_1_0_WRAPR_CLAMP_TO_BORDER 0x000000c0 +#define NV50TSC_1_0_WRAPR_CLAMP 0x00000100 +#define NV50TSC_1_0_WRAPR_MIRROR_CLAMP_TO_EDGE 0x00000140 +#define NV50TSC_1_0_WRAPR_MIRROR_CLAMP_TO_BORDER 0x00000180 +#define NV50TSC_1_0_WRAPR_MIRROR_CLAMP 0x000001c0 + +#define NV50TSC_1_1_MAGF_MASK 0x00000003 +#define NV50TSC_1_1_MAGF_NEAREST 0x00000001 +#define NV50TSC_1_1_MAGF_LINEAR 0x00000002 +#define NV50TSC_1_1_MINF_MASK 0x00000030 +#define NV50TSC_1_1_MINF_NEAREST 0x00000010 +#define NV50TSC_1_1_MINF_LINEAR 0x00000020 +#define NV50TSC_1_1_MIPF_MASK 0x000000c0 +#define NV50TSC_1_1_MIPF_NONE 0x00000040 +#define NV50TSC_1_1_MIPF_NEAREST 0x00000080 +#define NV50TSC_1_1_MIPF_LINEAR 0x000000c0 + +#define NV50TSC_1_2_UNKNOWN_MASK 0xffffffff + +#define NV50TSC_1_3_UNKNOWN_MASK 0xffffffff + +#define NV50TSC_1_4_UNKNOWN_MASK 0xffffffff + +#define NV50TSC_1_5_UNKNOWN_MASK 0xffffffff + +#define NV50TSC_1_6_UNKNOWN_MASK 0xffffffff + +#define NV50TSC_1_7_UNKNOWN_MASK 0xffffffff + +#endif -- cgit v1.2.3 From 7c745de74997e859d7e2640092bda9ad900e28a9 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 21:19:41 +1000 Subject: nv50: add some texture formats --- src/gallium/drivers/nv50/nv50_program.c | 2 + src/gallium/drivers/nv50/nv50_screen.c | 6 +++ src/gallium/drivers/nv50/nv50_tex.c | 93 ++++++++++++++++++++++++++++----- src/gallium/drivers/nv50/nv50_texture.h | 2 + 4 files changed, 90 insertions(+), 13 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 2668acf2999..0d3ddb8a596 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -1701,6 +1701,8 @@ nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p) if (p->buffer) pipe_buffer_reference(ws, &p->buffer, NULL); + nv50->screen->nvws->res_free(&p->data); + p->translated = 0; } diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index ec6d34e6845..bba5ca4d4e5 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -51,7 +51,13 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen, case PIPE_TEXTURE: switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: case PIPE_FORMAT_I8_UNORM: + case PIPE_FORMAT_A8L8_UNORM: return TRUE; default: break; diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c index 581a9e2ffb6..fde3c97c059 100644 --- a/src/gallium/drivers/nv50/nv50_tex.c +++ b/src/gallium/drivers/nv50/nv50_tex.c @@ -25,6 +25,81 @@ #include "nouveau/nouveau_stateobj.h" +static int +nv50_tex_construct(struct nouveau_stateobj *so, struct nv50_miptree *mt) +{ + switch (mt->base.format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_8_8_8_8); + break; + case PIPE_FORMAT_A1R5G5B5_UNORM: + so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_1_5_5_5); + break; + case PIPE_FORMAT_A4R4G4B4_UNORM: + so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_4_4_4_4); + break; + case PIPE_FORMAT_R5G6B5_UNORM: + so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_5_6_5); + break; + case PIPE_FORMAT_L8_UNORM: + so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_8); + break; + case PIPE_FORMAT_A8_UNORM: + so_data(so, NV50TIC_0_0_MAPA_C0 | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_ZERO | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_ZERO | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_ZERO | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_8); + break; + case PIPE_FORMAT_I8_UNORM: + so_data(so, NV50TIC_0_0_MAPA_C0 | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_8); + break; + case PIPE_FORMAT_A8L8_UNORM: + so_data(so, NV50TIC_0_0_MAPA_C1 | NV50TIC_0_0_TYPEA_UNORM | + NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM | + NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM | + NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM | + NV50TIC_0_0_FMT_8_8); + break; + default: + return 1; + } + + so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0); + so_data (so, 0xd0005000); + so_data (so, 0x00300000); + so_data (so, mt->base.width[0]); + so_data (so, (mt->base.depth[0] << 16) | mt->base.height[0]); + so_data (so, 0x03000000); + so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0); + + return 0; +} + void nv50_tex_validate(struct nv50_context *nv50) { @@ -37,19 +112,11 @@ nv50_tex_validate(struct nv50_context *nv50) so_data (so, NV50_CB_TIC); so_method(so, tesla, 0x40000f04, nv50->miptree_nr * 8); for (i = 0; i < nv50->miptree_nr; i++) { - struct nv50_miptree *mt = nv50->miptree[i]; - - so_data (so, 0x2a712488); - so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0xd0005000); - so_data (so, 0x00300000); - so_data (so, mt->base.width[0]); - so_data (so, (mt->base.depth[0] << 16) | - mt->base.height[0]); - so_data (so, 0x03000000); - so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_HIGH, 0, 0); + if (nv50_tex_construct(so, nv50->miptree[i])) { + NOUVEAU_ERR("failed tex validate\n"); + so_ref(NULL, &so); + return; + } } so_ref(so, &nv50->state.tic_upload); diff --git a/src/gallium/drivers/nv50/nv50_texture.h b/src/gallium/drivers/nv50/nv50_texture.h index c8fb282f21f..6861d67b4d5 100644 --- a/src/gallium/drivers/nv50/nv50_texture.h +++ b/src/gallium/drivers/nv50/nv50_texture.h @@ -45,6 +45,8 @@ #define NV50TIC_0_0_TYPEB_UNORM 0x00000080 #define NV50TIC_0_0_FMT_MASK 0x0000003c #define NV50TIC_0_0_FMT_8_8_8_8 0x00000008 +#define NV50TIC_0_0_FMT_4_4_4_4 0x00000012 +#define NV50TIC_0_0_FMT_1_5_5_5 0x00000013 #define NV50TIC_0_0_FMT_5_6_5 0x00000015 #define NV50TIC_0_0_FMT_8_8 0x00000018 #define NV50TIC_0_0_FMT_8 0x0000001d -- cgit v1.2.3 From 866a68dbd1ec9c3a6100de6d50890ac5068247dc Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 22:01:29 +1000 Subject: nv50: demagic tex filter / wrap mode --- src/gallium/drivers/nv50/nv50_state.c | 65 +++++++++++++++++++++++++++++++++-- 1 file changed, 63 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 5c364c4ddff..c552a0b0aa0 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -26,6 +26,7 @@ #include "pipe/p_inlines.h" #include "nv50_context.h" +#include "nv50_texture.h" #include "nouveau/nouveau_stateobj.h" @@ -104,14 +105,74 @@ nv50_blend_state_delete(struct pipe_context *pipe, void *hwcso) FREE(bso); } +static INLINE unsigned +wrap_mode(unsigned wrap) +{ + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + return NV50TSC_1_0_WRAPS_REPEAT; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + return NV50TSC_1_0_WRAPS_MIRROR_REPEAT; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + return NV50TSC_1_0_WRAPS_CLAMP_TO_EDGE; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return NV50TSC_1_0_WRAPS_CLAMP_TO_BORDER; + case PIPE_TEX_WRAP_CLAMP: + return NV50TSC_1_0_WRAPS_CLAMP; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + return NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_EDGE; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + return NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_BORDER; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + return NV50TSC_1_0_WRAPS_MIRROR_CLAMP; + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + return NV50TSC_1_0_WRAPS_REPEAT; + } +} static void * nv50_sampler_state_create(struct pipe_context *pipe, const struct pipe_sampler_state *cso) { unsigned *tsc = CALLOC(8, sizeof(unsigned)); - tsc[0] = 0x00024080; - tsc[1] = 0x00000062; + tsc[0] = (0x00024000 | + (wrap_mode(cso->wrap_s) << 0) | + (wrap_mode(cso->wrap_t) << 3) | + (wrap_mode(cso->wrap_r) << 6)); + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + tsc[1] |= NV50TSC_1_1_MAGF_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + tsc[1] |= NV50TSC_1_1_MAGF_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + tsc[1] |= NV50TSC_1_1_MINF_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + tsc[1] |= NV50TSC_1_1_MINF_NEAREST; + break; + } + + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_LINEAR: + tsc[1] |= NV50TSC_1_1_MIPF_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NEAREST: + tsc[1] |= NV50TSC_1_1_MIPF_NEAREST; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + tsc[1] |= NV50TSC_1_1_MIPF_NONE; + break; + } return (void *)tsc; } -- cgit v1.2.3 From 5180a668a7da627a40e024a2ed765458d5bac43e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 22:24:24 +1000 Subject: nouveau: winsys surface funcs deprecated, remove them --- .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 30 ---------------------- 1 file changed, 30 deletions(-) diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c index 765ea36f881..8a2870a2ff2 100644 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c @@ -25,33 +25,6 @@ nouveau_get_name(struct pipe_winsys *pws) return "Nouveau/DRI"; } -static struct pipe_surface * -nouveau_surface_alloc(struct pipe_winsys *ws) -{ - struct pipe_surface *surf; - - surf = CALLOC_STRUCT(pipe_surface); - if (!surf) - return NULL; - - surf->refcount = 1; - surf->winsys = ws; - return surf; -} - -static void -nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) -{ - struct pipe_surface *surf = *s; - - *s = NULL; - if (--surf->refcount <= 0) { - if (surf->buffer) - pipe_buffer_reference(ws, &surf->buffer, NULL); - free(surf); - } -} - static struct pipe_buffer * nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size) @@ -215,9 +188,6 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) pws->flush_frontbuffer = nouveau_flush_frontbuffer; - pws->surface_alloc = nouveau_surface_alloc; - pws->surface_release = nouveau_surface_release; - pws->buffer_create = nouveau_pipe_bo_create; pws->buffer_destroy = nouveau_pipe_bo_del; pws->user_buffer_create = nouveau_pipe_bo_user_create; -- cgit v1.2.3 From 9b0add0be4a3ba7fc72779daf8361d8cd98d9f64 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 22:45:11 +1000 Subject: nv50: quick hack to get textures untiled on map, and tiled on unmap progs/fp/tri-tex is all good now rather than all scrambled :) --- src/gallium/drivers/nv50/nv50_context.h | 11 ++++++++++ src/gallium/drivers/nv50/nv50_miptree.c | 18 +++++++++++------ src/gallium/drivers/nv50/nv50_surface.c | 36 ++++++++++++++++++++++++++++----- 3 files changed, 54 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index cb51fb02ebd..1c069f1625f 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -74,6 +74,17 @@ nv50_miptree(struct pipe_texture *pt) return (struct nv50_miptree *)pt; } +struct nv50_surface { + struct pipe_surface base; + struct pipe_buffer *untiled; +}; + +static INLINE struct nv50_surface * +nv50_surface(struct pipe_surface *pt) +{ + return (struct nv50_surface *)pt; +} + struct nv50_state { unsigned dirty; diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 8b61ca2a1fa..a02ad41885b 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -82,9 +82,14 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, { struct pipe_winsys *ws = pscreen->winsys; struct nv50_miptree *mt = nv50_miptree(pt); + struct nv50_surface *s; struct pipe_surface *ps; - ps = CALLOC_STRUCT(pipe_surface); + s = CALLOC_STRUCT(nv50_surface); + if (!s) + return NULL; + ps = &s->base; + ps->refcount = 1; ps->winsys = ws; ps->format = pt->format; @@ -109,14 +114,15 @@ nv50_miptree_surface_del(struct pipe_screen *pscreen, struct pipe_surface **psurface) { struct pipe_winsys *ws = pscreen->winsys; - struct pipe_surface *surf = *psurface; + struct pipe_surface *ps = *psurface; + struct nv50_surface *s = nv50_surface(ps); *psurface = NULL; - if (--surf->refcount <= 0) { - pipe_texture_reference(&surf->texture, NULL); - pipe_buffer_reference(ws, &surf->buffer, NULL); - FREE(surf); + if (--ps->refcount <= 0) { + pipe_texture_reference(&ps->texture, NULL); + pipe_buffer_reference(ws, &ps->buffer, NULL); + FREE(s); } } diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index 6229b636260..8d3f1edcfe9 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -52,25 +52,51 @@ nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, } static void * -nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, +nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *ps, unsigned flags ) { + struct nouveau_winsys *nvws = nv50_screen(screen)->nvws; struct pipe_winsys *ws = screen->winsys; + struct nv50_surface *s = nv50_surface(ps); + struct nv50_surface m = *s; void *map; - map = ws->buffer_map(ws, surface->buffer, flags); + if (!s->untiled) { + s->untiled = ws->buffer_create(ws, 0, 0, ps->buffer->size); + + m.base.buffer = s->untiled; + nvws->surface_copy(nvws, &m.base, 0, 0, &s->base, 0, 0, + ps->width, ps->height); + } + + /* Map original tiled surface to disallow it being validated while + * untiled mirror is mapped. + */ + ws->buffer_map(ws, ps->buffer, flags); + + map = ws->buffer_map(ws, s->untiled, flags); if (!map) return NULL; - return map + surface->offset; + return map; } static void -nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *ps) { + struct nouveau_winsys *nvws = nv50_screen(screen)->nvws; struct pipe_winsys *ws = screen->winsys; + struct nv50_surface *s = nv50_surface(ps); + struct nv50_surface m = *s; + + ws->buffer_unmap(ws, s->untiled); + ws->buffer_unmap(ws, ps->buffer); + + m.base.buffer = s->untiled; + nvws->surface_copy(nvws, &s->base, 0, 0, &m.base, 0, 0, + ps->width, ps->height); - ws->buffer_unmap(ws, surface->buffer); + pipe_buffer_reference(ws, &s->untiled, NULL); } void -- cgit v1.2.3 From 7c949fb2a2cd4e9b05efd4133e5ae8ea938934d8 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 11 Jul 2008 22:48:11 +1000 Subject: nv50: obey do_flip in surface_copy() --- src/gallium/drivers/nv50/nv50_surface.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index 8d3f1edcfe9..a9daeee3694 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -28,7 +28,7 @@ #include "util/p_tile.h" static void -nv50_surface_copy(struct pipe_context *pipe, unsigned flip, +nv50_surface_copy(struct pipe_context *pipe, boolean flip, struct pipe_surface *dest, unsigned destx, unsigned desty, struct pipe_surface *src, unsigned srcx, unsigned srcy, unsigned width, unsigned height) @@ -36,8 +36,16 @@ nv50_surface_copy(struct pipe_context *pipe, unsigned flip, struct nv50_context *nv50 = (struct nv50_context *)pipe; struct nouveau_winsys *nvws = nv50->screen->nvws; - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); + if (flip) { + desty += height; + while (height--) { + nvws->surface_copy(nvws, dest, destx, desty--, src, + srcx, srcy++, width, 1); + } + } else { + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); + } } static void -- cgit v1.2.3 From 36ac2ade17a1b8f2c47d41945430983834b9ef01 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Jul 2008 21:14:32 +0200 Subject: nv30: Update miptree to match latest changes --- src/gallium/drivers/nv30/nv30_miptree.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index ad0b257fe2b..a0e488c09b1 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -32,8 +32,8 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); if (swizzled) - pitch = pt->width[l]; - pitch = (pitch + 63) & ~63; + pitch = pt->nblocksx[l]; + pitch = align_int(pitch, 64); nv30mt->level[l].pitch = pitch * pt->block.size; nv30mt->level[l].image_offset = @@ -42,7 +42,6 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) width = MAX2(1, width >> 1); height = MAX2(1, height >> 1); depth = MAX2(1, depth >> 1); - } for (f = 0; f < nr_faces; f++) { @@ -111,17 +110,20 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; struct pipe_surface *ps; - ps = ws->surface_alloc(ws); + ps = CALLOC_STRUCT(pipe_surface); if (!ps) return NULL; + pipe_texture_reference(&ps->texture, pt); pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; + ps->block = pt->block; ps->nblocksx = pt->nblocksx[level]; ps->nblocksy = pt->nblocksy[level]; - ps->block = pt->block; ps->stride = nv30mt->level[level].pitch; + ps->usage = flags; + ps->status = PIPE_SURFACE_STATUS_DEFINED; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv30mt->level[level].image_offset[face]; @@ -139,6 +141,15 @@ static void nv30_miptree_surface_del(struct pipe_screen *pscreen, struct pipe_surface **psurface) { + struct pipe_surface *ps = *psurface; + + *psurface = NULL; + if (--ps->refcount > 0) + return; + + pipe_texture_reference(&ps->texture, NULL); + pipe_buffer_reference(pscreen->winsys, &ps->buffer, NULL); + FREE(ps); } void -- cgit v1.2.3 From 37a418b3b05db51aea0a1010bad8118b92e0e9d6 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Jul 2008 21:31:24 +0200 Subject: nv30: split fragprog_prepare from fragprog_translate --- src/gallium/drivers/nv30/nv30_fragprog.c | 91 ++++++++++++++++++++++++-------- 1 file changed, 68 insertions(+), 23 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 740d3ab430d..7278bf58869 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -1,7 +1,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi/util/tgsi_parse.h" @@ -650,30 +649,22 @@ nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc, return TRUE; } -void -nv30_fragprog_translate(struct nv30_context *nv30, - struct nv30_fragment_program *fp) +static boolean +nv30_fragprog_prepare(struct nv30_fpc *fpc) { - struct tgsi_parse_context parse; - struct nv30_fpc *fpc = NULL; - - fpc = CALLOC(1, sizeof(struct nv30_fpc)); - if (!fpc) - return; - fpc->fp = fp; - fpc->high_temp = -1; - fpc->num_regs = 2; + struct tgsi_parse_context p; + /*int high_temp = -1, i;*/ - tgsi_parse_init(&parse, fp->pipe.tokens); + tgsi_parse_init(&p, fpc->fp->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { + tgsi_parse_token(&p); + switch(tok->Token.Type) { case TGSI_TOKEN_TYPE_DECLARATION: { const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; + fdec = &p.FullToken.FullDeclaration; switch (fdec->Declaration.File) { case TGSI_FILE_INPUT: if (!nv30_fragprog_parse_decl_attrib(fpc, fdec)) @@ -683,6 +674,12 @@ nv30_fragprog_translate(struct nv30_context *nv30, if (!nv30_fragprog_parse_decl_output(fpc, fdec)) goto out_err; break; + /*case TGSI_FILE_TEMPORARY: + if (fdec->DeclarationRange.Last > high_temp) { + high_temp = + fdec->DeclarationRange.Last; + } + break;*/ default: break; } @@ -692,17 +689,65 @@ nv30_fragprog_translate(struct nv30_context *nv30, { struct tgsi_full_immediate *imm; float vals[4]; - int i; - imm = &parse.FullToken.FullImmediate; + imm = &p.FullToken.FullImmediate; assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); assert(fpc->nr_imm < MAX_IMM); - for (i = 0; i < 4; i++) - vals[i] = imm->u.ImmediateFloat32[i].Float; + vals[0] = imm->u.ImmediateFloat32[0].Float; + vals[1] = imm->u.ImmediateFloat32[1].Float; + vals[2] = imm->u.ImmediateFloat32[2].Float; + vals[3] = imm->u.ImmediateFloat32[3].Float; fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals); } break; + default: + break; + } + } + tgsi_parse_free(&p); + + /*if (++high_temp) { + fpc->r_temp = CALLOC(high_temp, sizeof(struct nv30_sreg)); + for (i = 0; i < high_temp; i++) + fpc->r_temp[i] = temp(fpc); + fpc->r_temps_discard = 0; + }*/ + + return TRUE; + +out_err: + /*if (fpc->r_temp) + FREE(fpc->r_temp);*/ + tgsi_parse_free(&p); + return FALSE; +} + +void +nv30_fragprog_translate(struct nv30_context *nv30, + struct nv30_fragment_program *fp) +{ + struct tgsi_parse_context parse; + struct nv30_fpc *fpc = NULL; + + fpc = CALLOC(1, sizeof(struct nv30_fpc)); + if (!fpc) + return; + fpc->fp = fp; + fpc->high_temp = -1; + fpc->num_regs = 2; + + if (!nv30_fragprog_prepare(fpc)) { + FREE(fpc); + return; + } + + tgsi_parse_init(&parse, fp->pipe.tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { case TGSI_TOKEN_TYPE_INSTRUCTION: { const struct tgsi_full_instruction *finst; -- cgit v1.2.3 From b327b064058aaab1debb1af08f4b9abe1385adc6 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Jul 2008 22:06:12 +0200 Subject: nv30: split fragprog_upload from fragprog_bind --- src/gallium/drivers/nv30/nv30_fragprog.c | 57 ++++++++++++++++++++------------ 1 file changed, 35 insertions(+), 22 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 7278bf58869..6af86e4f4f9 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -783,6 +783,40 @@ out_err: FREE(fpc); } +static void +nv30_fragprog_upload(struct nv30_context *nv30, + struct nv30_fragment_program *fp) +{ + struct pipe_winsys *ws = nv30->pipe.winsys; + const uint32_t le = 1; + uint32_t *map; + int i; + + map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); + +#if 0 + for (i = 0; i < fp->insn_len; i++) { + fflush(stdout); fflush(stderr); + NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); + fflush(stdout); fflush(stderr); + } +#endif + + if ((*(const uint8_t *)&le)) { + for (i = 0; i < fp->insn_len; i++) { + map[i] = fp->insn[i]; + } + } else { + /* Weird swapping for big-endian chips */ + for (i = 0; i < fp->insn_len; i++) { + map[i] = ((fp->insn[i] & 0xffff) << 16) | + ((fp->insn[i] >> 16) & 0xffff); + } + } + + ws->buffer_unmap(ws, fp->buffer); +} + void nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) { @@ -818,28 +852,7 @@ nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) if (!fp->buffer) fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); - map = ws->buffer_map(ws, fp->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); - -#if 0 - for (i = 0; i < fp->insn_len; i++) { - NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]); - } -#endif - - if ((*(const uint8_t *)&le)) { - for (i = 0; i < fp->insn_len; i++) { - map[i] = fp->insn[i]; - } - } else { - /* Weird swapping for big-endian chips */ - for (i = 0; i < fp->insn_len; i++) { - map[i] = ((fp->insn[i] & 0xffff) << 16) | - ((fp->insn[i] >> 16) & 0xffff); - } - } - - ws->buffer_unmap(ws, fp->buffer); + nv30_fragprog_upload(nv30, fp); fp->on_hw = TRUE; } -- cgit v1.2.3 From 5acbd0b0961089f9553adbe9b3d1341997ccb220 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Jul 2008 22:42:42 +0200 Subject: nv30: Emit fragtex state using state objects --- src/gallium/drivers/nv30/nv30_context.h | 1 + src/gallium/drivers/nv30/nv30_fragtex.c | 82 +++++++++++++++++++----------- src/gallium/drivers/nv30/nv30_state_emit.c | 38 +++----------- 3 files changed, 59 insertions(+), 62 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 5404685da25..72f803c80ad 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -220,6 +220,7 @@ extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_zsa; extern struct nv30_state_entry nv30_state_viewport; extern struct nv30_state_entry nv30_state_framebuffer; +extern struct nv30_state_entry nv30_state_fragtex; /* nv30_vbo.c */ extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index f253087c884..abe77b51df2 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -49,17 +49,17 @@ nv30_texture_formats[] = { _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), -// _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), + _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X), _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X), _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y), // _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), // _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), -// _(RGB_DXT1 , 0x86, S1, S1, S1, ONE, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT1 , 0x86, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT3 , 0x87, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), -// _(RGBA_DXT5 , 0x88, S1, S1, S1, S1, X, Y, Z, W, 0x00, 0x00), + _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W), + _(DXT1_RGBA , DXT1 , S1, S1, S1, S1, X, Y, Z, W), + _(DXT3_RGBA , DXT3 , S1, S1, S1, S1, X, Y, Z, W), + _(DXT5_RGBA , DXT5 , S1, S1, S1, S1, X, Y, Z, W), {}, }; @@ -67,6 +67,7 @@ static struct nv30_texture_format * nv30_fragtex_format(uint pipe_format) { struct nv30_texture_format *tf = nv30_texture_formats; + char fs[128]; while (tf->defined) { if (tf->pipe == pipe_format) @@ -74,26 +75,27 @@ nv30_fragtex_format(uint pipe_format) tf++; } + pf_sprint_name(fs, pipe_format); + NOUVEAU_ERR("unknown texture format %s\n", fs); return NULL; } -static void +static struct nouveau_stateobj * nv30_fragtex_build(struct nv30_context *nv30, int unit) { struct nv30_sampler_state *ps = nv30->tex_sampler[unit]; struct nv30_miptree *nv30mt = nv30->tex_miptree[unit]; struct pipe_texture *pt = &nv30mt->base; struct nv30_texture_format *tf; - uint32_t txf, txs, txp; - int swizzled = 0; /*XXX: implement in region code? */ + struct nouveau_stateobj *so; + uint32_t txf, txs /*, txp*/; + /*int swizzled = 0;*/ /*XXX: implement in region code? */ unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; tf = nv30_fragtex_format(pt->format); - if (!tf || !tf->defined) { - NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); - return; - } + if (!tf) + assert(0); txf = tf->format << 8; txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0); @@ -117,35 +119,45 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) break; default: NOUVEAU_ERR("Unknown target %d\n", pt->target); - return; + return NULL; } txs = tf->swizzle; - BEGIN_RING(rankine, NV34TCL_TX_OFFSET(unit), 8); - OUT_RELOCl(nv30mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW); - OUT_RELOCd(nv30mt->buffer,txf, tex_flags | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); - OUT_RING (ps->wrap); - OUT_RING (NV34TCL_TX_ENABLE_ENABLE | ps->en); - OUT_RING (txs); - OUT_RING (ps->filt | 0x2000 /* magic */); - OUT_RING ((pt->width[0] << NV34TCL_TX_NPOT_SIZE_W_SHIFT) | pt->height[0]); - OUT_RING (ps->bcol); + so = so_new(16, 2); + so_method(so, nv30->screen->rankine, NV34TCL_TX_OFFSET(unit), 8); + so_reloc (so, nv30mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (so, nv30mt->buffer, txf, tex_flags | NOUVEAU_BO_OR, + NV34TCL_TX_FORMAT_DMA0, NV34TCL_TX_FORMAT_DMA1); + so_data (so, ps->wrap); + so_data (so, NV34TCL_TX_ENABLE_ENABLE | ps->en); + so_data (so, txs); + so_data (so, ps->filt | 0x2000 /*voodoo*/); + so_data (so, (pt->width[0] << NV34TCL_TX_NPOT_SIZE_W_SHIFT) | + pt->height[0]); + so_data (so, ps->bcol); + + return so; } -void -nv30_fragtex_bind(struct nv30_context *nv30) +static boolean +nv30_fragtex_validate(struct nv30_context *nv30) { - struct nv30_fragment_program *fp = nv30->fragprog.active; + struct nv30_fragment_program *fp = nv30->fragprog.current; + struct nv30_state *state = &nv30->state; + struct nouveau_stateobj *so; unsigned samplers, unit; - samplers = nv30->fp_samplers & ~fp->samplers; + samplers = state->fp_samplers & ~fp->samplers; while (samplers) { unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - BEGIN_RING(rankine, NV34TCL_TX_ENABLE(unit), 1); - OUT_RING (0); + so = so_new(2, 0); + so_method(so, nv30->screen->rankine, NV34TCL_TX_ENABLE(unit), 1); + so_data (so, 0); + so_ref(so, &nv30->state.hw[NV30_STATE_FRAGTEX0 + unit]); + state->dirty |= (1ULL << (NV30_STATE_FRAGTEX0 + unit)); } samplers = nv30->dirty_samplers & fp->samplers; @@ -153,9 +165,19 @@ nv30_fragtex_bind(struct nv30_context *nv30) unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - nv30_fragtex_build(nv30, unit); + so = nv30_fragtex_build(nv30, unit); + so_ref(so, &nv30->state.hw[NV30_STATE_FRAGTEX0 + unit]); + state->dirty |= (1ULL << (NV30_STATE_FRAGTEX0 + unit)); } - nv30->fp_samplers = fp->samplers; + nv30->state.fp_samplers = fp->samplers; + return FALSE; } +struct nv30_state_entry nv30_state_fragtex = { + .validate = nv30_fragtex_validate, + .dirty = { + .pipe = NV30_NEW_SAMPLER | NV30_NEW_FRAGPROG, + .hw = 0 + } +}; diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 0ea78571972..cf10ddd4b63 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -6,6 +6,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_rasterizer, &nv30_state_scissor, &nv30_state_stipple, + &nv30_state_fragtex, &nv30_state_blend, &nv30_state_blend_colour, &nv30_state_zsa, @@ -59,6 +60,11 @@ nv30_emit_hw_state(struct nv30_context *nv30) screen->cur_pctx = nv30->pctx_id; } + if (nv30->dirty & NV30_NEW_FRAGPROG) { + nv30_fragprog_bind(nv30, nv30->fragprog.current); + /*XXX: clear NV30_NEW_FRAGPROG if no new program uploaded */ + } + for (i = 0, states = state->dirty; states; i++) { if (!(states & (1ULL << i))) continue; @@ -68,28 +74,11 @@ nv30_emit_hw_state(struct nv30_context *nv30) states &= ~(1ULL << i); } - if (nv30->dirty & NV30_NEW_FRAGPROG) { - nv30_fragprog_bind(nv30, nv30->fragprog.current); - /*XXX: clear NV30_NEW_FRAGPROG if no new program uploaded */ - } - - if (nv30->dirty_samplers || (nv30->dirty & NV30_NEW_FRAGPROG)) { - nv30_fragtex_bind(nv30); -/* - BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); - OUT_RING (2); - BEGIN_RING(rankine, NV34TCL_TX_CACHE_CTL, 1); - OUT_RING (1);*/ - nv30->dirty &= ~NV30_NEW_FRAGPROG; - } - if (nv30->dirty & NV30_NEW_VERTPROG) { nv30_vertprog_bind(nv30, nv30->vertprog.current); nv30->dirty &= ~NV30_NEW_VERTPROG; } - nv30->dirty_samplers = 0; - so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { if (!(samplers & (1 << i))) @@ -99,21 +88,6 @@ nv30_emit_hw_state(struct nv30_context *nv30) samplers &= ~(1ULL << i); } - /* Texture images, emitted in nv30_fragtex_build */ -#if 0 - for (i = 0; i < 16; i++) { - if (!(nv30->fp_samplers & (1 << i))) - continue; - BEGIN_RING(rankine, NV34TCL_TX_OFFSET(i), 2); - OUT_RELOCl(nv30->tex[i].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv30->tex[i].buffer, nv30->tex[i].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV34TCL_TX_FORMAT_DMA0, - NV34TCL_TX_FORMAT_DMA1); - } -#endif - /* Fragment program */ BEGIN_RING(rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); OUT_RELOC (nv30->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | -- cgit v1.2.3 From c7086277546d065eb94ba8dbeca1620605f167ea Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Fri, 11 Jul 2008 23:17:47 +0200 Subject: nv30: Move constant buffers out of vert/frag prog structures --- src/gallium/drivers/nv30/nv30_context.h | 6 ++---- src/gallium/drivers/nv30/nv30_fragprog.c | 12 ++++++------ src/gallium/drivers/nv30/nv30_state.c | 5 +++-- src/gallium/drivers/nv30/nv30_vertprog.c | 18 ++++++++---------- 4 files changed, 19 insertions(+), 22 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 72f803c80ad..c3c8b733090 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -124,6 +124,8 @@ struct nv30_context { unsigned dirty; struct pipe_scissor_state scissor; unsigned stipple[32]; + struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + unsigned constbuf_nr[PIPE_SHADER_TYPES]; struct nv30_rasterizer_state *rasterizer; struct nv30_zsa_state *zsa; struct nv30_blend_state *blend; @@ -150,16 +152,12 @@ struct nv30_context { struct { struct nv30_vertex_program *active; - struct nv30_vertex_program *current; - struct pipe_buffer *constant_buf; } vertprog; struct { struct nv30_fragment_program *active; - struct nv30_fragment_program *current; - struct pipe_buffer *constant_buf; } fragprog; struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 6af86e4f4f9..b560455a02f 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -820,6 +820,8 @@ nv30_fragprog_upload(struct nv30_context *nv30, void nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) { + struct pipe_buffer *constbuf = + nv30->constbuf[PIPE_SHADER_FRAGMENT]; struct pipe_winsys *ws = nv30->pipe.winsys; int i; @@ -830,8 +832,9 @@ nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) } if (fp->nr_consts) { - float *map = ws->buffer_map(ws, nv30->fragprog.constant_buf, - PIPE_BUFFER_USAGE_CPU_READ); + float *map; + + map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ); for (i = 0; i < fp->nr_consts; i++) { struct nv30_fragment_program_data *fpd = &fp->consts[i]; uint32_t *p = &fp->insn[fpd->offset]; @@ -842,13 +845,10 @@ nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) memcpy(p, cb, 4 * sizeof(float)); fp->on_hw = 0; } - ws->buffer_unmap(ws, nv30->fragprog.constant_buf); + ws->buffer_unmap(ws, constbuf); } if (!fp->on_hw) { - const uint32_t le = 1; - uint32_t *map; - if (!fp->buffer) fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 72b9515eb1a..aec6fa2b155 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -589,12 +589,13 @@ nv30_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, { struct nv30_context *nv30 = nv30_context(pipe); + nv30->constbuf[shader] = buf->buffer; + nv30->constbuf_nr[shader] = buf->size / (4 * sizeof(float)); + if (shader == PIPE_SHADER_VERTEX) { - nv30->vertprog.constant_buf = buf->buffer; nv30->dirty |= NV30_NEW_VERTPROG; } else if (shader == PIPE_SHADER_FRAGMENT) { - nv30->fragprog.constant_buf = buf->buffer; nv30->dirty |= NV30_NEW_FRAGPROG; } } diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 1b34d2e4b29..90756febd28 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -135,7 +135,7 @@ emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) /* * |VVV| - * d°.°b + * d�.�b * \u/ * */ @@ -641,9 +641,12 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) { struct nouveau_winsys *nvws = nv30->nvws; struct pipe_winsys *ws = nv30->pipe.winsys; + struct pipe_buffer *constbuf; boolean upload_code = FALSE, upload_data = FALSE; int i; + constbuf = nv30->constbuf[PIPE_SHADER_VERTEX]; + /* Translate TGSI shader into hw bytecode */ if (!vp->translated) { nv30_vertprog_translate(nv30, vp); @@ -730,8 +733,8 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) if (vp->nr_consts) { float *map = NULL; - if (nv30->vertprog.constant_buf) { - map = ws->buffer_map(ws, nv30->vertprog.constant_buf, + if (constbuf) { + map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ); } @@ -750,15 +753,10 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); OUT_RING (i + vp->data->start); OUT_RINGp ((uint32_t *)vpd->value, 4); -#if 0 - NOUVEAU_MSG("VP const %d: %f %f %f %f\n", - i, vpd->value[0], vpd->value[1], - vpd->value[2], vpd->value[3]); -#endif } - if (map) { - ws->buffer_unmap(ws, nv30->vertprog.constant_buf); + if (constbuf) { + ws->buffer_unmap(ws, constbuf); } } -- cgit v1.2.3 From 2419a5fe3601851989506a11b0bd4e3cfb071035 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 00:19:15 +0200 Subject: nv30: Emit vertex program using state objects --- src/gallium/drivers/nv30/nv30_context.h | 11 ++------ src/gallium/drivers/nv30/nv30_state.c | 11 ++++---- src/gallium/drivers/nv30/nv30_state.h | 2 ++ src/gallium/drivers/nv30/nv30_state_emit.c | 6 +---- src/gallium/drivers/nv30/nv30_vbo.c | 2 +- src/gallium/drivers/nv30/nv30_vertprog.c | 40 +++++++++++++++++++++++------- 6 files changed, 42 insertions(+), 30 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index c3c8b733090..5d7080a555d 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -124,6 +124,7 @@ struct nv30_context { unsigned dirty; struct pipe_scissor_state scissor; unsigned stipple[32]; + struct nv30_vertex_program *vertprog; struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; unsigned constbuf_nr[PIPE_SHADER_TYPES]; struct nv30_rasterizer_state *rasterizer; @@ -150,11 +151,6 @@ struct nv30_context { unsigned delta; } vb[16]; - struct { - struct nv30_vertex_program *active; - struct nv30_vertex_program *current; - } vertprog; - struct { struct nv30_fragment_program *active; struct nv30_fragment_program *current; @@ -188,10 +184,6 @@ extern void nv30_screen_init_miptree_functions(struct pipe_screen *pscreen); extern struct draw_stage *nv30_draw_render_stage(struct nv30_context *nv30); /* nv30_vertprog.c */ -extern void nv30_vertprog_translate(struct nv30_context *, - struct nv30_vertex_program *); -extern void nv30_vertprog_bind(struct nv30_context *, - struct nv30_vertex_program *); extern void nv30_vertprog_destroy(struct nv30_context *, struct nv30_vertex_program *); @@ -213,6 +205,7 @@ extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; extern struct nv30_state_entry nv30_state_stipple; +extern struct nv30_state_entry nv30_state_vertprog; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; extern struct nv30_state_entry nv30_state_zsa; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index aec6fa2b155..92695ce23c0 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -499,10 +499,12 @@ static void * nv30_vp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { + /*struct nv30_context *nv30 = nv30_context(pipe);*/ struct nv30_vertex_program *vp; vp = CALLOC(1, sizeof(struct nv30_vertex_program)); vp->pipe = *cso; + /*vp->draw = draw_create_vertex_shader(nv30->draw, &vp->pipe);*/ return (void *)vp; } @@ -511,14 +513,10 @@ static void nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_vertex_program *vp = hwcso; - - if (!hwcso) { - return; - } - nv30->vertprog.current = vp; + nv30->vertprog = hwcso; nv30->dirty |= NV30_NEW_VERTPROG; + /*nv30->draw_dirty |= NV30_NEW_VERTPROG;*/ } static void @@ -527,6 +525,7 @@ nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv30_context *nv30 = nv30_context(pipe); struct nv30_vertex_program *vp = hwcso; + /*draw_delete_vertex_shader(nv30->draw, vp->draw);*/ nv30_vertprog_destroy(nv30, vp); FREE(vp); } diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index c65a937467e..a897bc50687 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -26,6 +26,7 @@ struct nv30_vertex_program { struct pipe_shader_state pipe; boolean translated; + struct nv30_vertex_program_exec *insns; unsigned nr_insns; struct nv30_vertex_program_data *consts; @@ -39,6 +40,7 @@ struct nv30_vertex_program { uint32_t ir; uint32_t or; + struct nouveau_stateobj *so; }; struct nv30_fragment_program_data { diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index cf10ddd4b63..4ab62ddc016 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -7,6 +7,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_scissor, &nv30_state_stipple, &nv30_state_fragtex, + &nv30_state_vertprog, &nv30_state_blend, &nv30_state_blend_colour, &nv30_state_zsa, @@ -74,11 +75,6 @@ nv30_emit_hw_state(struct nv30_context *nv30) states &= ~(1ULL << i); } - if (nv30->dirty & NV30_NEW_VERTPROG) { - nv30_vertprog_bind(nv30, nv30->vertprog.current); - nv30->dirty &= ~NV30_NEW_VERTPROG; - } - so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { if (!(samplers & (1 << i))) diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index d23164eb485..8e4ee7a8742 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -100,7 +100,7 @@ nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, static void nv30_vbo_arrays_update(struct nv30_context *nv30) { - struct nv30_vertex_program *vp = nv30->vertprog.active; + struct nv30_vertex_program *vp = nv30->vertprog; uint32_t inputs, vtxfmt[16]; int hw, num_hw = 0; diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 90756febd28..65e00132055 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -563,7 +563,7 @@ nv30_vertprog_prepare(struct nv30_vpc *vpc) return TRUE; } -void +static void nv30_vertprog_translate(struct nv30_context *nv30, struct nv30_vertex_program *vp) { @@ -636,27 +636,31 @@ out_err: FREE(vpc); } -void -nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) +static boolean +nv30_vertprog_validate(struct nv30_context *nv30) { struct nouveau_winsys *nvws = nv30->nvws; struct pipe_winsys *ws = nv30->pipe.winsys; + struct nouveau_grobj *rankine = nv30->screen->rankine; + struct nv30_vertex_program *vp; struct pipe_buffer *constbuf; boolean upload_code = FALSE, upload_data = FALSE; int i; + vp = nv30->vertprog; constbuf = nv30->constbuf[PIPE_SHADER_VERTEX]; /* Translate TGSI shader into hw bytecode */ if (!vp->translated) { nv30_vertprog_translate(nv30, vp); if (!vp->translated) - assert(0); + return FALSE; } /* Allocate hw vtxprog exec slots */ if (!vp->exec) { struct nouveau_resource *heap = nv30->screen->vp_exec_heap; + struct nouveau_stateobj *so; uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { @@ -671,6 +675,15 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) assert(0); } + so = so_new(2, 0); + so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); + so_data (so, vp->exec->start); + /* Add these, and you'll go 1/3 speed */ + /*so_method(so, rankine, NV34TCL_VP_ATTRIB_EN, 2); + so_data (so, vp->ir); + so_data (so, vp->or);*/ + so_ref(so, &vp->so); + upload_code = TRUE; } @@ -777,10 +790,12 @@ nv30_vertprog_bind(struct nv30_context *nv30, struct nv30_vertex_program *vp) } } - BEGIN_RING(rankine, NV34TCL_VP_START_FROM_ID, 1); - OUT_RING (vp->exec->start); + if (vp->so != nv30->state.hw[NV30_STATE_VERTPROG]) { + so_ref(vp->so, &nv30->state.hw[NV30_STATE_VERTPROG]); + return TRUE; + } - nv30->vertprog.active = vp; + return FALSE; } void @@ -808,7 +823,14 @@ nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) vp->data_start = 0; vp->data_start_min = 0; - /* vp->ir = vp->or = vp->clip_ctrl = 0; - so_ref(NULL, &vp->so); */ + vp->ir = vp->or = 0; + so_ref(NULL, &vp->so); } +struct nv30_state_entry nv30_state_vertprog = { + .validate = nv30_vertprog_validate, + .dirty = { + .pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/, + .hw = NV30_STATE_VERTPROG, + } +}; -- cgit v1.2.3 From 58737dc87575625438d288fe2f816b6a9d2086f1 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 00:48:26 +0200 Subject: nv30: Emit fragment program using state objects --- src/gallium/drivers/nv30/nv30_context.h | 11 +---- src/gallium/drivers/nv30/nv30_fragprog.c | 65 +++++++++++++++++++++--------- src/gallium/drivers/nv30/nv30_fragtex.c | 2 +- src/gallium/drivers/nv30/nv30_state.c | 9 ++--- src/gallium/drivers/nv30/nv30_state.h | 3 ++ src/gallium/drivers/nv30/nv30_state_emit.c | 14 +------ 6 files changed, 56 insertions(+), 48 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 5d7080a555d..0c900c5598f 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -125,6 +125,7 @@ struct nv30_context { struct pipe_scissor_state scissor; unsigned stipple[32]; struct nv30_vertex_program *vertprog; + struct nv30_fragment_program *fragprog; struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; unsigned constbuf_nr[PIPE_SHADER_TYPES]; struct nv30_rasterizer_state *rasterizer; @@ -151,11 +152,6 @@ struct nv30_context { unsigned delta; } vb[16]; - struct { - struct nv30_fragment_program *active; - struct nv30_fragment_program *current; - } fragprog; - struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; }; @@ -188,10 +184,6 @@ extern void nv30_vertprog_destroy(struct nv30_context *, struct nv30_vertex_program *); /* nv30_fragprog.c */ -extern void nv30_fragprog_translate(struct nv30_context *, - struct nv30_fragment_program *); -extern void nv30_fragprog_bind(struct nv30_context *, - struct nv30_fragment_program *); extern void nv30_fragprog_destroy(struct nv30_context *, struct nv30_fragment_program *); @@ -205,6 +197,7 @@ extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; extern struct nv30_state_entry nv30_state_stipple; +extern struct nv30_state_entry nv30_state_fragprog; extern struct nv30_state_entry nv30_state_vertprog; extern struct nv30_state_entry nv30_state_blend; extern struct nv30_state_entry nv30_state_blend_colour; diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index b560455a02f..06ed04cbfeb 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -723,7 +723,7 @@ out_err: return FALSE; } -void +static void nv30_fragprog_translate(struct nv30_context *nv30, struct nv30_fragment_program *fp) { @@ -817,23 +817,46 @@ nv30_fragprog_upload(struct nv30_context *nv30, ws->buffer_unmap(ws, fp->buffer); } -void -nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) +static boolean +nv30_fragprog_validate(struct nv30_context *nv30) { + struct nv30_fragment_program *fp = nv30->fragprog; struct pipe_buffer *constbuf = nv30->constbuf[PIPE_SHADER_FRAGMENT]; struct pipe_winsys *ws = nv30->pipe.winsys; + struct nouveau_stateobj *so; + boolean new_consts = FALSE; int i; + if (fp->translated) + goto update_constants; + + /*nv30->fallback_swrast &= ~NV30_NEW_FRAGPROG;*/ + nv30_fragprog_translate(nv30, fp); if (!fp->translated) { - nv30_fragprog_translate(nv30, fp); - if (!fp->translated) - assert(0); + /*nv30->fallback_swrast |= NV30_NEW_FRAGPROG;*/ + return FALSE; } + fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); + nv30_fragprog_upload(nv30, fp); + + so = so_new(4, 1); + so_method(so, nv30->screen->rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); + so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | + NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + NV34TCL_FP_ACTIVE_PROGRAM_DMA0, NV34TCL_FP_ACTIVE_PROGRAM_DMA1); + so_method(so, nv30->screen->rankine, NV34TCL_FP_CONTROL, 1); + so_data (so, fp->fp_control); + /* FIXME: Add these, and you'll have big slowdown */ + /*so_method(so, nv30->screen->rankine, NV34TCL_FP_REG_CONTROL, 1); + so_data (so, fp->fp_control);*/ + so_ref(so, &fp->so); + +update_constants: if (fp->nr_consts) { float *map; - + map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ); for (i = 0; i < fp->nr_consts; i++) { struct nv30_fragment_program_data *fpd = &fp->consts[i]; @@ -843,25 +866,20 @@ nv30_fragprog_bind(struct nv30_context *nv30, struct nv30_fragment_program *fp) if (!memcmp(p, cb, 4 * sizeof(float))) continue; memcpy(p, cb, 4 * sizeof(float)); - fp->on_hw = 0; + new_consts = TRUE; } ws->buffer_unmap(ws, constbuf); - } - if (!fp->on_hw) { - if (!fp->buffer) - fp->buffer = ws->buffer_create(ws, 0x100, 0, - fp->insn_len * 4); - nv30_fragprog_upload(nv30, fp); - fp->on_hw = TRUE; + if (new_consts) + nv30_fragprog_upload(nv30, fp); } - BEGIN_RING(rankine, NV34TCL_FP_CONTROL, 1); - OUT_RING (fp->fp_control); - BEGIN_RING(rankine, NV34TCL_FP_REG_CONTROL, 1); - OUT_RING (fp->fp_reg_control); + if (new_consts || fp->so != nv30->state.hw[NV30_STATE_FRAGPROG]) { + so_ref(fp->so, &nv30->state.hw[NV30_STATE_FRAGPROG]); + return TRUE; + } - nv30->fragprog.active = fp; + return FALSE; } void @@ -872,3 +890,10 @@ nv30_fragprog_destroy(struct nv30_context *nv30, FREE(fp->insn); } +struct nv30_state_entry nv30_state_fragprog = { + .validate = nv30_fragprog_validate, + .dirty = { + .pipe = NV30_NEW_FRAGPROG, + .hw = NV30_STATE_FRAGPROG + } +}; diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index abe77b51df2..8c5e88ea1da 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -143,7 +143,7 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) static boolean nv30_fragtex_validate(struct nv30_context *nv30) { - struct nv30_fragment_program *fp = nv30->fragprog.current; + struct nv30_fragment_program *fp = nv30->fragprog; struct nv30_state *state = &nv30->state; struct nouveau_stateobj *so; unsigned samplers, unit; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 92695ce23c0..8dc16d361d8 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -539,6 +539,8 @@ nv30_fp_state_create(struct pipe_context *pipe, fp = CALLOC(1, sizeof(struct nv30_fragment_program)); fp->pipe = *cso; + tgsi_scan_shader(fp->pipe.tokens, &fp->info); + return (void *)fp; } @@ -546,13 +548,8 @@ static void nv30_fp_state_bind(struct pipe_context *pipe, void *hwcso) { struct nv30_context *nv30 = nv30_context(pipe); - struct nv30_fragment_program *fp = hwcso; - - if (!hwcso) { - return; - } - nv30->fragprog.current = fp; + nv30->fragprog = hwcso; nv30->dirty |= NV30_NEW_FRAGPROG; } diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index a897bc50687..20c51761607 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -2,6 +2,7 @@ #define __NV30_STATE_H__ #include "pipe/p_state.h" +#include "tgsi/util/tgsi_scan.h" struct nv30_sampler_state { uint32_t fmt; @@ -50,6 +51,7 @@ struct nv30_fragment_program_data { struct nv30_fragment_program { struct pipe_shader_state pipe; + struct tgsi_shader_info info; boolean translated; boolean on_hw; @@ -65,6 +67,7 @@ struct nv30_fragment_program { uint32_t fp_control; uint32_t fp_reg_control; + struct nouveau_stateobj *so; }; struct nv30_miptree { diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 4ab62ddc016..c4ccc9422be 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -6,6 +6,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_rasterizer, &nv30_state_scissor, &nv30_state_stipple, + &nv30_state_fragprog, &nv30_state_fragtex, &nv30_state_vertprog, &nv30_state_blend, @@ -61,11 +62,6 @@ nv30_emit_hw_state(struct nv30_context *nv30) screen->cur_pctx = nv30->pctx_id; } - if (nv30->dirty & NV30_NEW_FRAGPROG) { - nv30_fragprog_bind(nv30, nv30->fragprog.current); - /*XXX: clear NV30_NEW_FRAGPROG if no new program uploaded */ - } - for (i = 0, states = state->dirty; states; i++) { if (!(states & (1ULL << i))) continue; @@ -83,13 +79,7 @@ nv30_emit_hw_state(struct nv30_context *nv30) state->hw[NV30_STATE_FRAGTEX0+i]); samplers &= ~(1ULL << i); } - - /* Fragment program */ - BEGIN_RING(rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); - OUT_RELOC (nv30->fragprog.active->buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR, NV34TCL_FP_ACTIVE_PROGRAM_DMA0, - NV34TCL_FP_ACTIVE_PROGRAM_DMA1); + so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FRAGPROG]); } boolean -- cgit v1.2.3 From 740c93a08ce7f5fb43a0f4cd5a50d95459c8aa8e Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 00:48:44 +0200 Subject: nv30: Change comment about slowdown --- src/gallium/drivers/nv30/nv30_vertprog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 65e00132055..be6a327dad5 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -678,7 +678,7 @@ nv30_vertprog_validate(struct nv30_context *nv30) so = so_new(2, 0); so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); so_data (so, vp->exec->start); - /* Add these, and you'll go 1/3 speed */ + /* FIXME: Add these, and you'll have big slowdown */ /*so_method(so, rankine, NV34TCL_VP_ATTRIB_EN, 2); so_data (so, vp->ir); so_data (so, vp->or);*/ -- cgit v1.2.3 From b23e20a386729e75492069445bb924412dc29a0c Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 10:10:16 +0200 Subject: nv30: does not have vp_attrib/result --- src/gallium/drivers/nouveau/nouveau_class.h | 2 -- src/gallium/drivers/nv30/nv30_vertprog.c | 4 ---- 2 files changed, 6 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 18d8d3677db..749fbf041e7 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -4643,8 +4643,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV34TCL_VP_UPLOAD_CONST_W__SIZE 0x00000004 #define NV34TCL_UNK1f80(x) (0x00001f80+((x)*4)) #define NV34TCL_UNK1f80__SIZE 0x00000010 -#define NV34TCL_VP_ATTRIB_EN 0x00001ff0 -#define NV34TCL_VP_RESULT_EN 0x00001ff4 #define NV40_CONTEXT_SURFACES_2D 0x00003062 diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index be6a327dad5..39852ce9834 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -678,10 +678,6 @@ nv30_vertprog_validate(struct nv30_context *nv30) so = so_new(2, 0); so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); so_data (so, vp->exec->start); - /* FIXME: Add these, and you'll have big slowdown */ - /*so_method(so, rankine, NV34TCL_VP_ATTRIB_EN, 2); - so_data (so, vp->ir); - so_data (so, vp->or);*/ so_ref(so, &vp->so); upload_code = TRUE; -- cgit v1.2.3 From 9489de99802e635271c1ae84630fc02892af1699 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 10:13:58 +0200 Subject: nv30: was setting wrong register --- src/gallium/drivers/nv30/nv30_fragprog.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 06ed04cbfeb..59e79d66f2a 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -841,16 +841,15 @@ nv30_fragprog_validate(struct nv30_context *nv30) fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); nv30_fragprog_upload(nv30, fp); - so = so_new(4, 1); + so = so_new(6, 1); so_method(so, nv30->screen->rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, NV34TCL_FP_ACTIVE_PROGRAM_DMA0, NV34TCL_FP_ACTIVE_PROGRAM_DMA1); so_method(so, nv30->screen->rankine, NV34TCL_FP_CONTROL, 1); so_data (so, fp->fp_control); - /* FIXME: Add these, and you'll have big slowdown */ - /*so_method(so, nv30->screen->rankine, NV34TCL_FP_REG_CONTROL, 1); - so_data (so, fp->fp_control);*/ + so_method(so, nv30->screen->rankine, NV34TCL_FP_REG_CONTROL, 1); + so_data (so, fp->fp_reg_control); so_ref(so, &fp->so); update_constants: -- cgit v1.2.3 From 4ca346a8c0d0d4ea38705f8d3a3e5e690aa77daf Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 11:45:51 +0200 Subject: nv30: Rename state emission func --- src/gallium/drivers/nv30/nv30_context.h | 2 +- src/gallium/drivers/nv30/nv30_state_emit.c | 2 +- src/gallium/drivers/nv30/nv30_vbo.c | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 0c900c5598f..c41cb292017 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -192,7 +192,7 @@ extern void nv30_fragtex_bind(struct nv30_context *); /* nv30_state.c and friends */ extern boolean nv30_state_validate(struct nv30_context *nv30); -extern void nv30_emit_hw_state(struct nv30_context *nv30); +extern void nv30_state_emit(struct nv30_context *nv30); extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index c4ccc9422be..1e69a6a6e26 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -46,7 +46,7 @@ nv30_state_do_validate(struct nv30_context *nv30, } void -nv30_emit_hw_state(struct nv30_context *nv30) +nv30_state_emit(struct nv30_context *nv30) { struct nv30_state *state = &nv30->state; struct nv30_screen *screen = nv30->screen; diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index 8e4ee7a8742..0f82adfee04 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -156,7 +156,7 @@ nv30_vbo_validate_state(struct nv30_context *nv30, nv30_state_validate(nv30); - nv30_emit_hw_state(nv30); + nv30_state_emit(nv30); if (nv30->dirty & NV30_NEW_ARRAYS) { nv30_vbo_arrays_update(nv30); -- cgit v1.2.3 From 3d0e18ff5fba368a66bf34d18d219bf9a2dfba90 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 11:53:10 +0200 Subject: nv30: Reorder, remove useless stuff --- src/gallium/drivers/nv30/nv30_context.h | 19 +++---------------- 1 file changed, 3 insertions(+), 16 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index c41cb292017..ef8c16bdbb5 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -114,12 +114,6 @@ struct nv30_context { /* HW state derived from pipe states */ struct nv30_state state; - struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; - unsigned dirty_samplers; - unsigned fp_samplers; - unsigned vp_samplers; - /* Context state */ unsigned dirty; struct pipe_scissor_state scissor; @@ -134,17 +128,11 @@ struct nv30_context { struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; + struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; unsigned nr_samplers; unsigned nr_textures; - - uint32_t rt_enable; - struct pipe_buffer *rt[2]; - struct pipe_buffer *zeta; - - /*struct { - struct pipe_buffer *buffer; - uint32_t format; - } tex[16];*/ + unsigned dirty_samplers; unsigned vb_enable; struct { @@ -193,7 +181,6 @@ extern void nv30_fragtex_bind(struct nv30_context *); /* nv30_state.c and friends */ extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_state_emit(struct nv30_context *nv30); -extern void nv30_state_tex_update(struct nv30_context *nv30); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; extern struct nv30_state_entry nv30_state_stipple; -- cgit v1.2.3 From ac44f334e3492ab68eb310cfe43ed22206a042d8 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 12:24:37 +0200 Subject: nv30: Move edgeflag stuff --- src/gallium/drivers/nv30/nv30_context.c | 7 ------- src/gallium/drivers/nv30/nv30_context.h | 1 + src/gallium/drivers/nv30/nv30_state.c | 11 +++++++++++ 3 files changed, 12 insertions(+), 7 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index b2d9d3f1814..eefc614e5b5 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -32,12 +32,6 @@ nv30_destroy(struct pipe_context *pipe) FREE(nv30); } - -static void -nv30_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) -{ -} - struct pipe_context * nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) { @@ -57,7 +51,6 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30->pipe.winsys = ws; nv30->pipe.screen = pscreen; nv30->pipe.destroy = nv30_destroy; - nv30->pipe.set_edgeflags = nv30_set_edgeflags; nv30->pipe.draw_arrays = nv30_draw_arrays; nv30->pipe.draw_elements = nv30_draw_elements; nv30->pipe.clear = nv30_clear; diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index ef8c16bdbb5..2e7b62a69a3 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -142,6 +142,7 @@ struct nv30_context { struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; + const unsigned *edgeflags; }; static INLINE struct nv30_context * diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 8dc16d361d8..4fe1def74d6 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -657,6 +657,16 @@ nv30_set_vertex_elements(struct pipe_context *pipe, unsigned count, nv30->dirty |= NV30_NEW_ARRAYS; } +static void +nv30_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ + struct nv30_context *nv30 = nv30_context(pipe); + + nv30->edgeflags = bitfield; + nv30->dirty |= NV30_NEW_ARRAYS; + /*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/ +} + void nv30_init_state_functions(struct nv30_context *nv30) { @@ -696,6 +706,7 @@ nv30_init_state_functions(struct nv30_context *nv30) nv30->pipe.set_scissor_state = nv30_set_scissor_state; nv30->pipe.set_viewport_state = nv30_set_viewport_state; + nv30->pipe.set_edgeflags = nv30_set_edgeflags; nv30->pipe.set_vertex_buffers = nv30_set_vertex_buffers; nv30->pipe.set_vertex_elements = nv30_set_vertex_elements; } -- cgit v1.2.3 From 12118fcd123992f48ce78629e79e9949b96cd525 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 13:16:47 +0200 Subject: nv30: Emit vertex buffer objects using state objects --- src/gallium/drivers/nv30/nv30_context.h | 14 +- src/gallium/drivers/nv30/nv30_screen.c | 2 +- src/gallium/drivers/nv30/nv30_state.c | 6 + src/gallium/drivers/nv30/nv30_state_emit.c | 8 +- src/gallium/drivers/nv30/nv30_vbo.c | 628 +++++++++++++++++------------ 5 files changed, 399 insertions(+), 259 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 2e7b62a69a3..823b34a7c3b 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -128,20 +128,17 @@ struct nv30_context { struct pipe_blend_color blend_colour; struct pipe_viewport_state viewport; struct pipe_framebuffer_state framebuffer; + struct pipe_buffer *idxbuf; + unsigned idxbuf_format; struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; unsigned nr_samplers; unsigned nr_textures; unsigned dirty_samplers; - - unsigned vb_enable; - struct { - struct pipe_buffer *buffer; - unsigned delta; - } vb[16]; - - struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + unsigned vtxbuf_nr; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; + unsigned vtxelt_nr; const unsigned *edgeflags; }; @@ -193,6 +190,7 @@ extern struct nv30_state_entry nv30_state_zsa; extern struct nv30_state_entry nv30_state_viewport; extern struct nv30_state_entry nv30_state_framebuffer; extern struct nv30_state_entry nv30_state_fragtex; +extern struct nv30_state_entry nv30_state_vbo; /* nv30_vbo.c */ extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode, diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 9c576369891..de5590af146 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -57,7 +57,7 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param) return 13; case NOUVEAU_CAP_HW_VTXBUF: case NOUVEAU_CAP_HW_IDXBUF: - return 0; + return 1; default: NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); return 0; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 4fe1def74d6..4d6303ebc28 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -644,7 +644,10 @@ nv30_set_vertex_buffers(struct pipe_context *pipe, unsigned count, struct nv30_context *nv30 = nv30_context(pipe); memcpy(nv30->vtxbuf, vb, sizeof(*vb) * count); + nv30->vtxbuf_nr = count; + nv30->dirty |= NV30_NEW_ARRAYS; + /*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/ } static void @@ -654,7 +657,10 @@ nv30_set_vertex_elements(struct pipe_context *pipe, unsigned count, struct nv30_context *nv30 = nv30_context(pipe); memcpy(nv30->vtxelt, ve, sizeof(*ve) * count); + nv30->vtxelt_nr = count; + nv30->dirty |= NV30_NEW_ARRAYS; + /*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/ } static void diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 1e69a6a6e26..40fed621b24 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -13,6 +13,7 @@ static struct nv30_state_entry *render_states[] = { &nv30_state_blend_colour, &nv30_state_zsa, &nv30_state_viewport, + &nv30_state_vbo, NULL }; @@ -39,10 +40,7 @@ nv30_state_do_validate(struct nv30_context *nv30, states++; } - -/* TODO: uncomment when finished converting nv30->dirty = 0; -*/ } void @@ -71,6 +69,8 @@ nv30_state_emit(struct nv30_context *nv30) states &= ~(1ULL << i); } + state->dirty = 0; + so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { if (!(samplers & (1 << i))) @@ -80,6 +80,8 @@ nv30_state_emit(struct nv30_context *nv30) samplers &= ~(1ULL << i); } so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FRAGPROG]); + if (state->hw[NV30_STATE_VTXBUF] /*&& nv30->render_mode == HW*/) + so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_VTXBUF]); } boolean diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index 0f82adfee04..d930557f6bb 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -7,45 +7,119 @@ #include "nouveau/nouveau_channel.h" #include "nouveau/nouveau_pushbuf.h" +#include "nouveau/nouveau_util.h" + +#define FORCE_SWTNL 0 static INLINE int -nv30_vbo_ncomp(uint format) +nv30_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) { - int ncomp = 0; + char fs[128]; + + switch (pipe) { + case PIPE_FORMAT_R32_FLOAT: + case PIPE_FORMAT_R32G32_FLOAT: + case PIPE_FORMAT_R32G32B32_FLOAT: + case PIPE_FORMAT_R32G32B32A32_FLOAT: + *fmt = NV34TCL_VTXFMT_TYPE_FLOAT; + break; + case PIPE_FORMAT_R8_UNORM: + case PIPE_FORMAT_R8G8_UNORM: + case PIPE_FORMAT_R8G8B8_UNORM: + case PIPE_FORMAT_R8G8B8A8_UNORM: + *fmt = NV34TCL_VTXFMT_TYPE_UBYTE; + break; + case PIPE_FORMAT_R16_SSCALED: + case PIPE_FORMAT_R16G16_SSCALED: + case PIPE_FORMAT_R16G16B16_SSCALED: + case PIPE_FORMAT_R16G16B16A16_SSCALED: + *fmt = NV34TCL_VTXFMT_TYPE_USHORT; + break; + default: + pf_sprint_name(fs, pipe); + NOUVEAU_ERR("Unknown format %s\n", fs); + return 1; + } - if (pf_size_x(format)) ncomp++; - if (pf_size_y(format)) ncomp++; - if (pf_size_z(format)) ncomp++; - if (pf_size_w(format)) ncomp++; + switch (pipe) { + case PIPE_FORMAT_R8_UNORM: + case PIPE_FORMAT_R32_FLOAT: + case PIPE_FORMAT_R16_SSCALED: + *ncomp = 1; + break; + case PIPE_FORMAT_R8G8_UNORM: + case PIPE_FORMAT_R32G32_FLOAT: + case PIPE_FORMAT_R16G16_SSCALED: + *ncomp = 2; + break; + case PIPE_FORMAT_R8G8B8_UNORM: + case PIPE_FORMAT_R32G32B32_FLOAT: + case PIPE_FORMAT_R16G16B16_SSCALED: + *ncomp = 3; + break; + case PIPE_FORMAT_R8G8B8A8_UNORM: + case PIPE_FORMAT_R32G32B32A32_FLOAT: + case PIPE_FORMAT_R16G16B16A16_SSCALED: + *ncomp = 4; + break; + default: + pf_sprint_name(fs, pipe); + NOUVEAU_ERR("Unknown format %s\n", fs); + return 1; + } - return ncomp; + return 0; } -static INLINE int -nv30_vbo_type(uint format) +static boolean +nv30_vbo_set_idxbuf(struct nv30_context *nv30, struct pipe_buffer *ib, + unsigned ib_size) { - switch (pf_type(format)) { - case PIPE_FORMAT_TYPE_FLOAT: - return NV34TCL_VTXFMT_TYPE_FLOAT; - case PIPE_FORMAT_TYPE_UNORM: - return NV34TCL_VTXFMT_TYPE_UBYTE; + struct pipe_screen *pscreen = &nv30->screen->pipe; + unsigned type; + + if (!ib) { + nv30->idxbuf = NULL; + nv30->idxbuf_format = 0xdeadbeef; + return FALSE; + } + + if (!pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF) || ib_size == 1) + return FALSE; + + switch (ib_size) { + case 2: + type = NV34TCL_IDXBUF_FORMAT_TYPE_U16; + break; + case 4: + type = NV34TCL_IDXBUF_FORMAT_TYPE_U32; + break; default: - NOUVEAU_ERR("Unknown format 0x%08x\n", format); - return NV34TCL_VTXFMT_TYPE_FLOAT; + return FALSE; } + + if (ib != nv30->idxbuf || + type != nv30->idxbuf_format) { + nv30->dirty |= NV30_NEW_ARRAYS; + nv30->idxbuf = ib; + nv30->idxbuf_format = type; + } + + return TRUE; } static boolean -nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, - struct pipe_vertex_element *ve, +nv30_vbo_static_attrib(struct nv30_context *nv30, struct nouveau_stateobj *so, + int attrib, struct pipe_vertex_element *ve, struct pipe_vertex_buffer *vb) { struct pipe_winsys *ws = nv30->pipe.winsys; - int type, ncomp; + struct nouveau_grobj *rankine = nv30->screen->rankine; + unsigned type, ncomp; void *map; - type = nv30_vbo_type(ve->src_format); - ncomp = nv30_vbo_ncomp(ve->src_format); + if (nv30_vbo_format_to_hw(ve->src_format, &type, &ncomp)) + return FALSE; map = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ); map += vb->buffer_offset + ve->src_offset; @@ -55,31 +129,28 @@ nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, { float *v = map; - BEGIN_RING(rankine, NV34TCL_VTX_ATTR_4F_X(attrib), 4); switch (ncomp) { case 4: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(v[3]); + so_method(so, rankine, NV34TCL_VTX_ATTR_4F_X(attrib), 4); + so_data (so, fui(v[0])); + so_data (so, fui(v[1])); + so_data (so, fui(v[2])); + so_data (so, fui(v[3])); break; case 3: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(v[2]); - OUT_RINGf(1.0); + so_method(so, rankine, NV34TCL_VTX_ATTR_3F_X(attrib), 3); + so_data (so, fui(v[0])); + so_data (so, fui(v[1])); + so_data (so, fui(v[2])); break; case 2: - OUT_RINGf(v[0]); - OUT_RINGf(v[1]); - OUT_RINGf(0.0); - OUT_RINGf(1.0); + so_method(so, rankine, NV34TCL_VTX_ATTR_2F_X(attrib), 2); + so_data (so, fui(v[0])); + so_data (so, fui(v[1])); break; case 1: - OUT_RINGf(v[0]); - OUT_RINGf(0.0); - OUT_RINGf(0.0); - OUT_RINGf(1.0); + so_method(so, rankine, NV34TCL_VTX_ATTR_1F(attrib), 1); + so_data (so, fui(v[0])); break; default: ws->buffer_unmap(ws, vb->buffer); @@ -97,209 +168,202 @@ nv30_vbo_static_attrib(struct nv30_context *nv30, int attrib, return TRUE; } -static void -nv30_vbo_arrays_update(struct nv30_context *nv30) +boolean +nv30_draw_arrays(struct pipe_context *pipe, + unsigned mode, unsigned start, unsigned count) { - struct nv30_vertex_program *vp = nv30->vertprog; - uint32_t inputs, vtxfmt[16]; - int hw, num_hw = 0; - - nv30->vb_enable = 0; + struct nv30_context *nv30 = nv30_context(pipe); + struct nouveau_channel *chan = nv30->nvws->channel; + unsigned restart = 0; - inputs = vp->ir; - for (hw = 0; hw < 16 && inputs; hw++) { - if (inputs & (1 << hw)) { - num_hw = hw; - inputs &= ~(1 << hw); - } + nv30_vbo_set_idxbuf(nv30, NULL, 0); + if (FORCE_SWTNL || !nv30_state_validate(nv30)) { + /*return nv30_draw_elements_swtnl(pipe, NULL, 0, + mode, start, count);*/ + return FALSE; } - num_hw++; - inputs = vp->ir; - for (hw = 0; hw < num_hw; hw++) { - struct pipe_vertex_element *ve; - struct pipe_vertex_buffer *vb; + while (count) { + unsigned vc, nr; - if (!(inputs & (1 << hw))) { - vtxfmt[hw] = NV34TCL_VTXFMT_TYPE_FLOAT; + nv30_state_emit(nv30); + + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256, + mode, start, count, &restart); + if (!vc) { + FIRE_RING(NULL); continue; } - ve = &nv30->vtxelt[hw]; - vb = &nv30->vtxbuf[ve->vertex_buffer_index]; + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); - if (vb->pitch == 0) { - vtxfmt[hw] = NV34TCL_VTXFMT_TYPE_FLOAT; - if (nv30_vbo_static_attrib(nv30, hw, ve, vb) == TRUE) - continue; + nr = (vc & 0xff); + if (nr) { + BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; } - nv30->vb_enable |= (1 << hw); - nv30->vb[hw].delta = vb->buffer_offset + ve->src_offset; - nv30->vb[hw].buffer = vb->buffer; + nr = vc >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; + + nr -= push; - vtxfmt[hw] = ((vb->pitch << NV34TCL_VTXFMT_STRIDE_SHIFT) | - (nv30_vbo_ncomp(ve->src_format) << - NV34TCL_VTXFMT_SIZE_SHIFT) | - nv30_vbo_type(ve->src_format)); + BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + count -= vc; + start = restart; } - BEGIN_RING(rankine, NV34TCL_VTXFMT(0), num_hw); - OUT_RINGp (vtxfmt, num_hw); + pipe->flush(pipe, 0, NULL); + return TRUE; } -static boolean -nv30_vbo_validate_state(struct nv30_context *nv30, - struct pipe_buffer *ib, unsigned ib_format) +static INLINE void +nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, + unsigned mode, unsigned start, unsigned count) { - unsigned inputs; + struct nouveau_channel *chan = nv30->nvws->channel; - nv30_state_validate(nv30); + while (count) { + uint8_t *elts = (uint8_t *)ib + start; + unsigned vc, push, restart = 0; - nv30_state_emit(nv30); + nv30_state_emit(nv30); - if (nv30->dirty & NV30_NEW_ARRAYS) { - nv30_vbo_arrays_update(nv30); - nv30->dirty &= ~NV30_NEW_ARRAYS; - } + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2, + mode, start, count, &restart); + if (vc == 0) { + FIRE_RING(NULL); + continue; + } + count -= vc; - inputs = nv30->vb_enable; - while (inputs) { - unsigned a = ffs(inputs) - 1; + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); - inputs &= ~(1 << a); + if (vc & 1) { + BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; vc--; + } - BEGIN_RING(rankine, NV34TCL_VTXBUF_ADDRESS(a), 1); - OUT_RELOC (nv30->vb[a].buffer, nv30->vb[a].delta, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_LOW | - NOUVEAU_BO_OR | NOUVEAU_BO_RD, 0, - NV34TCL_VTXBUF_ADDRESS_DMA1); - } + while (vc) { + unsigned i; - if (ib) { - BEGIN_RING(rankine, NV34TCL_IDXBUF_ADDRESS, 2); - OUT_RELOCl(ib, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD); - OUT_RELOCd(ib, ib_format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | - NOUVEAU_BO_RD | NOUVEAU_BO_OR, - 0, NV34TCL_IDXBUF_FORMAT_DMA1); - } + push = MIN2(vc, 2047 * 2); - BEGIN_RING(rankine, 0x1710, 1); - OUT_RING (0); /* vtx cache flush */ + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); - return TRUE; -} + vc -= push; + elts += push; + } -boolean -nv30_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, - unsigned count) -{ - struct nv30_context *nv30 = nv30_context(pipe); - unsigned nr; - boolean ret; + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); - ret = nv30_vbo_validate_state(nv30, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; + start = restart; } +} - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - - nr = (count & 0xff); - if (nr) { - BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } +static INLINE void +nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, + unsigned mode, unsigned start, unsigned count) +{ + struct nouveau_channel *chan = nv30->nvws->channel; - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; + while (count) { + uint16_t *elts = (uint16_t *)ib + start; + unsigned vc, push, restart = 0; - nr -= push; + nv30_state_emit(nv30); - BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2, + mode, start, count, &restart); + if (vc == 0) { + FIRE_RING(NULL); + continue; } - } + count -= vc; - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); - pipe->flush(pipe, 0, NULL); - return TRUE; -} + if (vc & 1) { + BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); + OUT_RING (elts[0]); + elts++; vc--; + } -static INLINE void -nv30_draw_elements_u08(struct nv30_context *nv30, void *ib, - unsigned start, unsigned count) -{ - uint8_t *elts = (uint8_t *)ib + start; - int push, i; + while (vc) { + unsigned i; - if (count & 1) { - BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } + push = MIN2(vc, 2047 * 2); - while (count) { - push = MIN2(count, 2047 * 2); + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((elts[i+1] << 16) | elts[i]); - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); + vc -= push; + elts += push; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); - count -= push; - elts += push; + start = restart; } } static INLINE void -nv30_draw_elements_u16(struct nv30_context *nv30, void *ib, - unsigned start, unsigned count) +nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, + unsigned mode, unsigned start, unsigned count) { - uint16_t *elts = (uint16_t *)ib + start; - int push, i; - - if (count & 1) { - BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1); - OUT_RING (elts[0]); - elts++; count--; - } + struct nouveau_channel *chan = nv30->nvws->channel; while (count) { - push = MIN2(count, 2047 * 2); + uint32_t *elts = (uint32_t *)ib + start; + unsigned vc, push, restart = 0; - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((elts[i+1] << 16) | elts[i]); + nv30_state_emit(nv30); - count -= push; - elts += push; - } -} + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 5, 1, + mode, start, count, &restart); + if (vc == 0) { + FIRE_RING(NULL); + continue; + } + count -= vc; -static INLINE void -nv30_draw_elements_u32(struct nv30_context *nv30, void *ib, - unsigned start, unsigned count) -{ - uint32_t *elts = (uint32_t *)ib + start; - int push; + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); - while (count) { - push = MIN2(count, 2047); + while (vc) { + push = MIN2(vc, 2047); - BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push); - OUT_RINGp (elts, push); + BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push); + OUT_RINGp (elts, push); - count -= push; - elts += push; + vc -= push; + elts += push; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + start = restart; } } @@ -310,99 +374,82 @@ nv30_draw_elements_inline(struct pipe_context *pipe, { struct nv30_context *nv30 = nv30_context(pipe); struct pipe_winsys *ws = pipe->winsys; - boolean ret; void *map; - ret = nv30_vbo_validate_state(nv30, NULL, 0); - if (!ret) { - NOUVEAU_ERR("state validate failed\n"); - return FALSE; - } - map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ); if (!ib) { NOUVEAU_ERR("failed mapping ib\n"); return FALSE; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); - switch (ib_size) { case 1: - nv30_draw_elements_u08(nv30, map, start, count); + nv30_draw_elements_u08(nv30, map, mode, start, count); break; case 2: - nv30_draw_elements_u16(nv30, map, start, count); + nv30_draw_elements_u16(nv30, map, mode, start, count); break; case 4: - nv30_draw_elements_u32(nv30, map, start, count); + nv30_draw_elements_u32(nv30, map, mode, start, count); break; default: NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); break; } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); - ws->buffer_unmap(ws, ib); - return TRUE; } static boolean nv30_draw_elements_vbo(struct pipe_context *pipe, - struct pipe_buffer *ib, unsigned ib_size, unsigned mode, unsigned start, unsigned count) { struct nv30_context *nv30 = nv30_context(pipe); - unsigned nr, type; - boolean ret; - - switch (ib_size) { - case 2: - type = NV34TCL_IDXBUF_FORMAT_TYPE_U16; - break; - case 4: - type = NV34TCL_IDXBUF_FORMAT_TYPE_U32; - break; - default: - NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size); - return FALSE; - } + struct nouveau_channel *chan = nv30->nvws->channel; + unsigned restart = 0; - ret = nv30_vbo_validate_state(nv30, ib, type); - if (!ret) { - NOUVEAU_ERR("failed state validation\n"); - return FALSE; - } + while (count) { + unsigned nr, vc; - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (nvgl_primitive(mode)); + nv30_state_emit(nv30); - nr = (count & 0xff); - if (nr) { - BEGIN_RING(rankine, NV34TCL_VB_INDEX_BATCH, 1); - OUT_RING (((nr - 1) << 24) | start); - start += nr; - } + vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256, + mode, start, count, &restart); + if (!vc) { + FIRE_RING(NULL); + continue; + } + + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (nvgl_primitive(mode)); + + nr = (vc & 0xff); + if (nr) { + BEGIN_RING(rankine, NV34TCL_VB_INDEX_BATCH, 1); + OUT_RING (((nr - 1) << 24) | start); + start += nr; + } - nr = count >> 8; - while (nr) { - unsigned push = nr > 2047 ? 2047 : nr; + nr = vc >> 8; + while (nr) { + unsigned push = nr > 2047 ? 2047 : nr; - nr -= push; + nr -= push; - BEGIN_RING_NI(rankine, NV34TCL_VB_INDEX_BATCH, push); - while (push--) { - OUT_RING(((0x100 - 1) << 24) | start); - start += 0x100; + BEGIN_RING_NI(rankine, NV34TCL_VB_INDEX_BATCH, push); + while (push--) { + OUT_RING(((0x100 - 1) << 24) | start); + start += 0x100; + } } - } - BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); - OUT_RING (0); + BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1); + OUT_RING (0); + + count -= vc; + start = restart; + } return TRUE; } @@ -412,10 +459,19 @@ nv30_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { -/* if (indexSize != 1) { - nv30_draw_elements_vbo(pipe, indexBuffer, indexSize, - mode, start, count); - } else */{ + struct nv30_context *nv30 = nv30_context(pipe); + boolean idxbuf; + + idxbuf = nv30_vbo_set_idxbuf(nv30, indexBuffer, indexSize); + if (FORCE_SWTNL || !nv30_state_validate(nv30)) { + /*return nv30_draw_elements_swtnl(pipe, NULL, 0, + mode, start, count);*/ + return FALSE; + } + + if (idxbuf) { + nv30_draw_elements_vbo(pipe, mode, start, count); + } else { nv30_draw_elements_inline(pipe, indexBuffer, indexSize, mode, start, count); } @@ -424,4 +480,82 @@ nv30_draw_elements(struct pipe_context *pipe, return TRUE; } +static boolean +nv30_vbo_validate(struct nv30_context *nv30) +{ + struct nouveau_stateobj *vtxbuf, *vtxfmt, *sattr = NULL; + struct nouveau_grobj *rankine = nv30->screen->rankine; + struct pipe_buffer *ib = nv30->idxbuf; + unsigned ib_format = nv30->idxbuf_format; + unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD; + int hw; + + if (nv30->edgeflags) { + /*nv30->fallback_swtnl |= NV30_NEW_ARRAYS;*/ + return FALSE; + } + + vtxbuf = so_new(20, 18); + so_method(vtxbuf, rankine, NV34TCL_VTXBUF_ADDRESS(0), nv30->vtxelt_nr); + vtxfmt = so_new(17, 0); + so_method(vtxfmt, rankine, NV34TCL_VTXFMT(0), nv30->vtxelt_nr); + for (hw = 0; hw < nv30->vtxelt_nr; hw++) { + struct pipe_vertex_element *ve; + struct pipe_vertex_buffer *vb; + unsigned type, ncomp; + + ve = &nv30->vtxelt[hw]; + vb = &nv30->vtxbuf[ve->vertex_buffer_index]; + + if (!vb->pitch) { + if (!sattr) + sattr = so_new(16 * 5, 0); + + if (nv30_vbo_static_attrib(nv30, sattr, hw, ve, vb)) { + so_data(vtxbuf, 0); + so_data(vtxfmt, NV34TCL_VTXFMT_TYPE_FLOAT); + continue; + } + } + + if (nv30_vbo_format_to_hw(ve->src_format, &type, &ncomp)) { + /*nv30->fallback_swtnl |= NV30_NEW_ARRAYS;*/ + so_ref(NULL, &vtxbuf); + so_ref(NULL, &vtxfmt); + return FALSE; + } + + so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset, + vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, + 0, NV34TCL_VTXBUF_ADDRESS_DMA1); + so_data (vtxfmt, ((vb->pitch << NV34TCL_VTXFMT_STRIDE_SHIFT) | + (ncomp << NV34TCL_VTXFMT_SIZE_SHIFT) | type)); + } + + if (ib) { + so_method(vtxbuf, rankine, NV34TCL_IDXBUF_ADDRESS, 2); + so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0); + so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR, + 0, NV34TCL_IDXBUF_FORMAT_DMA1); + } + + so_method(vtxbuf, rankine, 0x1710, 1); + so_data (vtxbuf, 0); + + so_ref(vtxbuf, &nv30->state.hw[NV30_STATE_VTXBUF]); + nv30->state.dirty |= (1ULL << NV30_STATE_VTXBUF); + so_ref(vtxfmt, &nv30->state.hw[NV30_STATE_VTXFMT]); + nv30->state.dirty |= (1ULL << NV30_STATE_VTXFMT); + so_ref(sattr, &nv30->state.hw[NV30_STATE_VTXATTR]); + nv30->state.dirty |= (1ULL << NV30_STATE_VTXATTR); + return FALSE; +} + +struct nv30_state_entry nv30_state_vbo = { + .validate = nv30_vbo_validate, + .dirty = { + .pipe = NV30_NEW_ARRAYS, + .hw = 0, + } +}; -- cgit v1.2.3 From a9863caefc9ed1ae098ea9033a82dbdc556edd30 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 13:59:01 +0200 Subject: nv30: do not shift texture format --- src/gallium/drivers/nv30/nv30_fragtex.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index 8c5e88ea1da..4242f86f76f 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -97,7 +97,7 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) if (!tf) assert(0); - txf = tf->format << 8; + txf = tf->format; txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0); txf |= log2i(pt->width[0]) << 20; txf |= log2i(pt->height[0]) << 24; -- cgit v1.2.3 From fa167eedeba601f3281655f779331ea9f4eaa5cf Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 13:59:24 +0200 Subject: nv30: update caps --- src/gallium/drivers/nv30/nv30_screen.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index de5590af146..54fc245bc63 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -32,7 +32,7 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param) case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: return 16; case PIPE_CAP_NPOT_TEXTURES: - return 0; + return 1; case PIPE_CAP_TWO_SIDED_STENCIL: return 1; case PIPE_CAP_GLSL: @@ -75,7 +75,7 @@ nv30_screen_get_paramf(struct pipe_screen *pscreen, int param) case PIPE_CAP_MAX_POINT_WIDTH_AA: return 64.0; case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 16.0; + return 8.0; case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: return 4.0; default: -- cgit v1.2.3 From 1d50e26f4afc0c7cdcd843a1336a90cdfc76765b Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 12 Jul 2008 15:07:02 +0200 Subject: nv30: no npot textures --- src/gallium/drivers/nv30/nv30_screen.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 54fc245bc63..b216a703187 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -32,7 +32,7 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param) case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: return 16; case PIPE_CAP_NPOT_TEXTURES: - return 1; + return 0; case PIPE_CAP_TWO_SIDED_STENCIL: return 1; case PIPE_CAP_GLSL: -- cgit v1.2.3 From 8878d8d4b35c9550632f5684bdc97f3958bf43c5 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 10 Jul 2008 00:32:18 -0400 Subject: g3dvl: Minor SP winsys cleanup. --- src/gallium/winsys/g3dvl/xsp_winsys.c | 46 ++++++++++++++++++++++------------- src/libXvMC/block.c | 3 +-- 2 files changed, 30 insertions(+), 19 deletions(-) diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index d04c4be580f..7c5c592fc77 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -62,6 +62,7 @@ static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, assert(buffer); xsp_buf->mapped_data = xsp_buf->data; + return xsp_buf->mapped_data; } @@ -131,8 +132,7 @@ static int xsp_surface_alloc_storage surface->nblocksy = pf_get_nblocksy(&surface->block, height); surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT); surface->usage = flags; - /* XXX: Need to consider block dims? See xm_winsys.c */ - surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * height); + surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * surface->nblocksy); return 0; } @@ -158,6 +158,29 @@ static void xsp_surface_release(struct pipe_winsys *pws, struct pipe_surface **s *surface = NULL; } +static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence) +{ + assert(pws); + assert(ptr); + assert(fence); +} + +static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag) +{ + assert(pws); + assert(fence); + + return 0; +} + +static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag) +{ + assert(pws); + assert(fence); + + return 0; +} + static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private) { struct xsp_pipe_winsys *xsp_winsys; @@ -196,22 +219,11 @@ static const char* xsp_get_name(struct pipe_winsys *pws) return "X11 SoftPipe"; } -/* softpipe_winsys implementation */ - -static boolean xsp_is_format_supported(struct softpipe_winsys *spws, enum pipe_format format) -{ - assert(spws); - - /* TODO: Test that 'format' is equal to our output window's format */ - return TRUE; -} - /* Show starts here */ struct pipe_context* create_pipe_context(Display *display) { struct xsp_pipe_winsys *xsp_winsys; - struct softpipe_winsys *sp_winsys; struct pipe_screen *p_screen; struct pipe_context *p_context; @@ -226,6 +238,9 @@ struct pipe_context* create_pipe_context(Display *display) xsp_winsys->base.surface_alloc = xsp_surface_alloc; xsp_winsys->base.surface_alloc_storage = xsp_surface_alloc_storage; xsp_winsys->base.surface_release = xsp_surface_release; + xsp_winsys->base.fence_reference = xsp_fence_reference; + xsp_winsys->base.fence_signalled = xsp_fence_signalled; + xsp_winsys->base.fence_finish = xsp_fence_finish; xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer; xsp_winsys->base.get_name = xsp_get_name; xsp_winsys->display = display; @@ -253,12 +268,9 @@ struct pipe_context* create_pipe_context(Display *display) XDestroyImage(template); } - - sp_winsys = calloc(1, sizeof(struct softpipe_winsys)); - sp_winsys->is_format_supported = xsp_is_format_supported; p_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); - p_context = softpipe_create(p_screen, (struct pipe_winsys*)xsp_winsys, sp_winsys); + p_context = softpipe_create(p_screen, (struct pipe_winsys*)xsp_winsys, NULL); return p_context; } diff --git a/src/libXvMC/block.c b/src/libXvMC/block.c index b56348d4645..deca305bdcf 100644 --- a/src/libXvMC/block.c +++ b/src/libXvMC/block.c @@ -8,8 +8,7 @@ * XvMC defines 64 element blocks (8x8 elements). * Elements are 8 bits when they represent color values, * 9 bits when they reprecent DCT coefficients, we - * store them in 2 bytes in either case. DCT coefficients - * can be signed or unsigned, at our option. + * store them in 2 bytes in either case. */ #define BLOCK_SIZE (64 * 2) -- cgit v1.2.3 From c243573fafe8e83d4964535b201c499164d7c172 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 19 Jul 2008 16:22:56 -0400 Subject: g3dvl: Fix some memory leaks in the winsys. --- src/gallium/state_trackers/g3dvl/vl_context.c | 2 -- src/gallium/winsys/g3dvl/xsp_winsys.c | 26 +++++++++++++++++++++----- src/gallium/winsys/g3dvl/xsp_winsys.h | 1 + src/libXvMC/context.c | 5 ++++- 4 files changed, 26 insertions(+), 8 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 1668ad1651b..3d4ca7cf4e2 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -2208,8 +2208,6 @@ int vlDestroyContext(struct VL_CONTEXT *context) vlDestroy(context); - context->pipe->destroy(context->pipe); - free(context); return 0; diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index 7c5c592fc77..c660c6e0186 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -224,8 +224,8 @@ static const char* xsp_get_name(struct pipe_winsys *pws) struct pipe_context* create_pipe_context(Display *display) { struct xsp_pipe_winsys *xsp_winsys; - struct pipe_screen *p_screen; - struct pipe_context *p_context; + struct pipe_screen *screen; + struct pipe_context *pipe; assert(display); @@ -269,9 +269,25 @@ struct pipe_context* create_pipe_context(Display *display) XDestroyImage(template); } - p_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); - p_context = softpipe_create(p_screen, (struct pipe_winsys*)xsp_winsys, NULL); + screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); + pipe = softpipe_create(screen, (struct pipe_winsys*)xsp_winsys, NULL); - return p_context; + return pipe; +} + +int destroy_pipe_context(struct pipe_context *pipe) +{ + struct pipe_screen *screen; + struct pipe_winsys *winsys; + + assert(pipe); + + screen = pipe->screen; + winsys = pipe->winsys; + pipe->destroy(pipe); + screen->destroy(screen); + free(winsys); + + return 0; } diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.h b/src/gallium/winsys/g3dvl/xsp_winsys.h index 47a6ac6e251..cb163dc24d2 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.h +++ b/src/gallium/winsys/g3dvl/xsp_winsys.h @@ -6,6 +6,7 @@ struct pipe_context; struct pipe_context* create_pipe_context(Display *display); +int destroy_pipe_context(struct pipe_context *pipe); #endif diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c index c835a6acf7a..e5590025777 100644 --- a/src/libXvMC/context.c +++ b/src/libXvMC/context.c @@ -136,7 +136,8 @@ Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, i Status XvMCDestroyContext(Display *display, XvMCContext *context) { - struct VL_CONTEXT *vl_ctx; + struct VL_CONTEXT *vl_ctx; + struct pipe_context *pipe; assert(display); @@ -147,7 +148,9 @@ Status XvMCDestroyContext(Display *display, XvMCContext *context) assert(display == vl_ctx->display); + pipe = vl_ctx->pipe; vlDestroyContext(vl_ctx); + destroy_pipe_context(pipe); return Success; } -- cgit v1.2.3 From a3616067fee97f06fb52131ac13c39aeca6dc06a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 21 Jul 2008 19:05:55 +1000 Subject: nv50: add NV86 and NV94 to list of "supported" chips --- src/gallium/drivers/nv50/nv50_screen.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index bba5ca4d4e5..2f3cf041d1c 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -29,8 +29,8 @@ #include "nouveau/nouveau_stateobj.h" #define NV5X_GRCLASS5097_CHIPSETS 0x00000001 -#define NV8X_GRCLASS8297_CHIPSETS 0x00000010 -#define NV9X_GRCLASS8297_CHIPSETS 0x00000004 +#define NV8X_GRCLASS8297_CHIPSETS 0x00000050 +#define NV9X_GRCLASS8297_CHIPSETS 0x00000014 static boolean nv50_screen_is_format_supported(struct pipe_screen *pscreen, -- cgit v1.2.3 From 90bd0e338d315c426c2d0255331610055023739e Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 20 Jul 2008 21:30:14 -0400 Subject: g3dvl: Add stubs for some XvMC functions. Some of these may need to be implemented on the DDX side. --- src/libXvMC/context.c | 43 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c index e5590025777..22eadc1889b 100644 --- a/src/libXvMC/context.c +++ b/src/libXvMC/context.c @@ -155,3 +155,46 @@ Status XvMCDestroyContext(Display *display, XvMCContext *context) return Success; } +/* XXX: The following are here temporarily, need to be implemented in the DDX driver */ +/* TODO: Figure out which of these need to be in DDX, which are better off in DDX, which can stay */ + +Bool XvMCQueryExtension(Display *display, int *event_base, int *err_base) +{ + *event_base = 0; + *err_base = 0; + + return True; +} + +Status XvMCQueryVersion(Display *display, int *major, int *minor) +{ + *major = 1; + *minor = 0; + + return Success; +} + +XvMCSurfaceInfo* XvMCListSurfaceTypes(Display *display, XvPortID port, int *num) +{ + XvMCSurfaceInfo *surface_info = calloc(1, sizeof(XvMCSurfaceInfo)); + + *num = 1; + + surface_info->chroma_format = XVMC_CHROMA_FORMAT_420; + surface_info->max_width = 2048; + surface_info->max_height = 2048; + surface_info->mc_type = XVMC_IDCT | XVMC_MPEG_2; + surface_info->surface_type_id = 123; /* FIXME: XAllocID(display)*/; + surface_info->flags |= XVMC_INTRA_UNSIGNED; + + return surface_info; +} + +XvImageFormatValues* XvMCListSubpictureTypes(Display* display, XvPortID port, int surface_type_id, int *count_return) +{ + /* TODO */ + *count_return = 0; + + return NULL; +} + -- cgit v1.2.3 From 0c25ac52425e6d6eb037b99ab90f41b47e3f4491 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Tue, 22 Jul 2008 22:26:26 -0400 Subject: g3dvl: Add Nouveau winsys, libdriclient. Nouveau winsys is based on Mesa's Nouveau winsys and soft-links to most of it. The 'nouveau_context' and 'nouveau_screen' code contains most of the changes, 'nouveau_winsys_pipe', 'nouveau_swapbuffers' and 'nouveau_lock' contain some minor changes. The driclient library contains the DRI userland stuff, most of which was based on Mesa's DRI code. --- src/driclient/include/driclient.h | 98 ++++ src/driclient/include/xf86dri.h | 119 ++++ src/driclient/src/Makefile | 22 + src/driclient/src/XF86dri.c | 619 +++++++++++++++++++++ src/driclient/src/driclient.c | 292 ++++++++++ src/driclient/src/test | Bin 0 -> 68389 bytes src/driclient/src/test.c | 41 ++ src/driclient/src/xf86dri.h | 119 ++++ src/driclient/src/xf86dristr.h | 342 ++++++++++++ src/gallium/state_trackers/g3dvl/Makefile | 2 +- src/gallium/state_trackers/g3dvl/vl_context.c | 68 ++- src/gallium/state_trackers/g3dvl/vl_shader_build.c | 12 + src/gallium/state_trackers/g3dvl/vl_shader_build.h | 1 + src/gallium/state_trackers/g3dvl/vl_surface.c | 4 +- src/gallium/winsys/g3dvl/nouveau/Makefile | 46 ++ src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_context.c | 371 ++++++++++++ src/gallium/winsys/g3dvl/nouveau/nouveau_context.h | 105 ++++ src/gallium/winsys/g3dvl/nouveau/nouveau_device.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_local.h | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c | 92 +++ .../winsys/g3dvl/nouveau/nouveau_notifier.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c | 1 + .../winsys/g3dvl/nouveau/nouveau_resource.c | 1 + src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c | 88 +++ src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h | 24 + .../winsys/g3dvl/nouveau/nouveau_swapbuffers.c | 64 +++ .../winsys/g3dvl/nouveau/nouveau_swapbuffers.h | 10 + src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c | 1 + .../winsys/g3dvl/nouveau/nouveau_winsys_pipe.c | 261 +++++++++ .../winsys/g3dvl/nouveau/nouveau_winsys_pipe.h | 1 + .../winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c | 1 + src/gallium/winsys/g3dvl/nouveau/nv04_surface.c | 1 + src/gallium/winsys/g3dvl/nouveau/nv50_surface.c | 1 + src/gallium/winsys/g3dvl/vl_winsys.h | 14 + src/gallium/winsys/g3dvl/xsp_winsys.c | 72 ++- src/gallium/winsys/g3dvl/xsp_winsys.h | 12 - src/libXvMC/Makefile | 19 +- src/libXvMC/context.c | 4 +- 46 files changed, 2896 insertions(+), 43 deletions(-) create mode 100644 src/driclient/include/driclient.h create mode 100644 src/driclient/include/xf86dri.h create mode 100644 src/driclient/src/Makefile create mode 100644 src/driclient/src/XF86dri.c create mode 100644 src/driclient/src/driclient.c create mode 100755 src/driclient/src/test create mode 100644 src/driclient/src/test.c create mode 100644 src/driclient/src/xf86dri.h create mode 100644 src/driclient/src/xf86dristr.h create mode 100644 src/gallium/winsys/g3dvl/nouveau/Makefile create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_context.c create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_context.h create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_device.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_local.h create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c create mode 100644 src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h create mode 120000 src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nv04_surface.c create mode 120000 src/gallium/winsys/g3dvl/nouveau/nv50_surface.c create mode 100644 src/gallium/winsys/g3dvl/vl_winsys.h delete mode 100644 src/gallium/winsys/g3dvl/xsp_winsys.h diff --git a/src/driclient/include/driclient.h b/src/driclient/include/driclient.h new file mode 100644 index 00000000000..36438a9f790 --- /dev/null +++ b/src/driclient/include/driclient.h @@ -0,0 +1,98 @@ +#ifndef driclient_h +#define driclient_h + +#include +#include +#include +#include "xf86dri.h" + +/* TODO: Bring in DRI XML options */ + +typedef struct dri_version +{ + int major; + int minor; + int patch; +} dri_version_t; + +typedef struct dri_screen +{ + Display *display; + unsigned int num; + dri_version_t ddx, dri, drm; + int draw_lock_id; + int fd; + drm_sarea_t *sarea; + void *drawable_hash; + void *private; +} dri_screen_t; + +struct dri_context; + +typedef struct dri_drawable +{ + drm_drawable_t drm_drawable; + Drawable x_drawable; + unsigned int sarea_index; + unsigned int *sarea_stamp; + unsigned int last_sarea_stamp; + int x, y, w, h; + int back_x, back_y; + int num_cliprects, num_back_cliprects; + drm_clip_rect_t *cliprects, *back_cliprects; + dri_screen_t *dri_screen; + unsigned int refcount; + void *private; +} dri_drawable_t; + +typedef struct dri_context +{ + XID id; + drm_context_t drm_context; + dri_screen_t *dri_screen; + void *private; +} dri_context_t; + +typedef struct dri_framebuffer +{ + drm_handle_t drm_handle; + int base, size, stride; + int private_size; + void *private; +} dri_framebuffer_t; + +int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri_framebuffer_t *dri_framebuf); +int driDestroyScreen(dri_screen_t *dri_screen); +int driCreateDrawable(dri_screen_t *dri_screen, Drawable drawable, dri_drawable_t **dri_drawable); +int driUpdateDrawableInfo(dri_drawable_t *dri_drawable); +int driDestroyDrawable(dri_drawable_t *dri_drawable); +int driCreateContext(dri_screen_t *dri_screen, Visual *visual, dri_context_t **dri_context); +int driDestroyContext(dri_context_t *dri_context); +int driCompareVersions(const dri_version_t *v1, const dri_version_t *v2); + +#define DRI_VALIDATE_DRAWABLE_INFO_ONCE(dri_drawable) \ +do \ +{ \ + if (*(dri_drawable->sarea_stamp) != dri_drawable->last_sarea_stamp) \ + driUpdateDrawableInfo(dri_drawable); \ +} while (0) + +#define DRI_VALIDATE_DRAWABLE_INFO(dri_screen, dri_drawable) \ +do \ +{ \ + while (*(dri_drawable->sarea_stamp) != dri_drawable->last_sarea_stamp) \ + { \ + register unsigned int hwContext = dri_screen->sarea->lock.lock & \ + ~(DRM_LOCK_HELD | DRM_LOCK_CONT); \ + DRM_UNLOCK(dri_screen->fd, &dri_screen->sarea->lock, hwContext); \ + \ + DRM_SPINLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id); \ + DRI_VALIDATE_DRAWABLE_INFO_ONCE(dri_drawable); \ + DRM_SPINUNLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id); \ + \ + DRM_LIGHT_LOCK(dri_screen->fd, &dri_screen->sarea->lock, hwContext); \ + } \ +} while (0) + +#endif + diff --git a/src/driclient/include/xf86dri.h b/src/driclient/include/xf86dri.h new file mode 100644 index 00000000000..baf80a7a9dd --- /dev/null +++ b/src/driclient/include/xf86dri.h @@ -0,0 +1,119 @@ +/************************************************************************** + +Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas. +Copyright 2000 VA Linux Systems, Inc. +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial portions +of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS +OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. +IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR +ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/** + * \file xf86dri.h + * Protocol numbers and function prototypes for DRI X protocol. + * + * \author Kevin E. Martin + * \author Jens Owen + * \author Rickard E. (Rik) Faith + */ + +#ifndef _XF86DRI_H_ +#define _XF86DRI_H_ + +#include +#include + +#define X_XF86DRIQueryVersion 0 +#define X_XF86DRIQueryDirectRenderingCapable 1 +#define X_XF86DRIOpenConnection 2 +#define X_XF86DRICloseConnection 3 +#define X_XF86DRIGetClientDriverName 4 +#define X_XF86DRICreateContext 5 +#define X_XF86DRIDestroyContext 6 +#define X_XF86DRICreateDrawable 7 +#define X_XF86DRIDestroyDrawable 8 +#define X_XF86DRIGetDrawableInfo 9 +#define X_XF86DRIGetDeviceInfo 10 +#define X_XF86DRIAuthConnection 11 +#define X_XF86DRIOpenFullScreen 12 /* Deprecated */ +#define X_XF86DRICloseFullScreen 13 /* Deprecated */ + +#define XF86DRINumberEvents 0 + +#define XF86DRIClientNotLocal 0 +#define XF86DRIOperationNotSupported 1 +#define XF86DRINumberErrors (XF86DRIOperationNotSupported + 1) + +#ifndef _XF86DRI_SERVER_ + +_XFUNCPROTOBEGIN + +Bool XF86DRIQueryExtension( Display *dpy, int *event_base, int *error_base ); + +Bool XF86DRIQueryVersion( Display *dpy, int *majorVersion, int *minorVersion, + int *patchVersion ); + +Bool XF86DRIQueryDirectRenderingCapable( Display *dpy, int screen, + Bool *isCapable ); + +Bool XF86DRIOpenConnection( Display *dpy, int screen, drm_handle_t *hSAREA, + char **busIDString ); + +Bool XF86DRIAuthConnection( Display *dpy, int screen, drm_magic_t magic ); + +Bool XF86DRICloseConnection( Display *dpy, int screen ); + +Bool XF86DRIGetClientDriverName( Display *dpy, int screen, + int *ddxDriverMajorVersion, int *ddxDriverMinorVersion, + int *ddxDriverPatchVersion, char **clientDriverName ); + +Bool XF86DRICreateContext( Display *dpy, int screen, Visual *visual, + XID *ptr_to_returned_context_id, drm_context_t *hHWContext ); + +Bool XF86DRICreateContextWithConfig( Display *dpy, int screen, int configID, + XID *ptr_to_returned_context_id, drm_context_t *hHWContext ); + +Bool XF86DRIDestroyContext( Display *dpy, int screen, + XID context_id ); + +Bool XF86DRICreateDrawable( Display *dpy, int screen, + Drawable drawable, drm_drawable_t *hHWDrawable ); + +Bool XF86DRIDestroyDrawable( Display *dpy, int screen, + Drawable drawable); + +Bool XF86DRIGetDrawableInfo( Display *dpy, int screen, Drawable drawable, + unsigned int *index, unsigned int *stamp, + int *X, int *Y, int *W, int *H, + int *numClipRects, drm_clip_rect_t ** pClipRects, + int *backX, int *backY, + int *numBackClipRects, drm_clip_rect_t **pBackClipRects ); + +Bool XF86DRIGetDeviceInfo( Display *dpy, int screen, + drm_handle_t *hFrameBuffer, int *fbOrigin, int *fbSize, + int *fbStride, int *devPrivateSize, void **pDevPrivate ); + +_XFUNCPROTOEND + +#endif /* _XF86DRI_SERVER_ */ + +#endif /* _XF86DRI_H_ */ + diff --git a/src/driclient/src/Makefile b/src/driclient/src/Makefile new file mode 100644 index 00000000000..0fac552de5b --- /dev/null +++ b/src/driclient/src/Makefile @@ -0,0 +1,22 @@ +TARGET = libdriclient.a +OBJECTS = driclient.o XF86dri.o +DRMDIR ?= /usr + +CFLAGS += -g -Wall -fPIC -Werror -I../include -I${DRMDIR}/include -I${DRMDIR}/include/drm + +############################################# + +.PHONY = all clean + +all: ${TARGET} test + +${TARGET}: ${OBJECTS} + ar rcs $@ $^ + cp ${TARGET} ../lib + +test: test.o + $(CC) -L../lib -L${DRMDIR}/lib ${LDFLAGS} -o $@ $^ -ldriclient -lX11 -lXext -ldrm + +clean: + rm -rf ${OBJECTS} ${TARGET} test test.o + diff --git a/src/driclient/src/XF86dri.c b/src/driclient/src/XF86dri.c new file mode 100644 index 00000000000..9e359a92384 --- /dev/null +++ b/src/driclient/src/XF86dri.c @@ -0,0 +1,619 @@ +/************************************************************************** + +Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas. +Copyright 2000 VA Linux Systems, Inc. +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial portions +of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS +OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. +IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR +ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/* + * Authors: + * Kevin E. Martin + * Jens Owen + * Rickard E. (Rik) Faith + * + */ + +/* THIS IS NOT AN X CONSORTIUM STANDARD */ + +#define NEED_REPLIES +#include +#include +#include +#include "xf86dristr.h" + +static XExtensionInfo _xf86dri_info_data; +static XExtensionInfo *xf86dri_info = &_xf86dri_info_data; +static char xf86dri_extension_name[] = XF86DRINAME; + +#define XF86DRICheckExtension(dpy,i,val) \ + XextCheckExtension (dpy, i, xf86dri_extension_name, val) + +/***************************************************************************** + * * + * private utility routines * + * * + *****************************************************************************/ + +static int close_display(Display *dpy, XExtCodes *extCodes); +static /* const */ XExtensionHooks xf86dri_extension_hooks = { + NULL, /* create_gc */ + NULL, /* copy_gc */ + NULL, /* flush_gc */ + NULL, /* free_gc */ + NULL, /* create_font */ + NULL, /* free_font */ + close_display, /* close_display */ + NULL, /* wire_to_event */ + NULL, /* event_to_wire */ + NULL, /* error */ + NULL, /* error_string */ +}; + +static XEXT_GENERATE_FIND_DISPLAY (find_display, xf86dri_info, + xf86dri_extension_name, + &xf86dri_extension_hooks, + 0, NULL) + +static XEXT_GENERATE_CLOSE_DISPLAY (close_display, xf86dri_info) + + +/***************************************************************************** + * * + * public XFree86-DRI Extension routines * + * * + *****************************************************************************/ + +#if 0 +#include +#define TRACE(msg) fprintf(stderr,"XF86DRI%s\n", msg); +#else +#define TRACE(msg) +#endif + +#define PUBLIC + +PUBLIC Bool XF86DRIQueryExtension (dpy, event_basep, error_basep) + Display *dpy; + int *event_basep, *error_basep; +{ + XExtDisplayInfo *info = find_display (dpy); + + TRACE("QueryExtension..."); + if (XextHasExtension(info)) { + *event_basep = info->codes->first_event; + *error_basep = info->codes->first_error; + TRACE("QueryExtension... return True"); + return True; + } else { + TRACE("QueryExtension... return False"); + return False; + } +} + +PUBLIC Bool XF86DRIQueryVersion(dpy, majorVersion, minorVersion, patchVersion) + Display* dpy; + int* majorVersion; + int* minorVersion; + int* patchVersion; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIQueryVersionReply rep; + xXF86DRIQueryVersionReq *req; + + TRACE("QueryVersion..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIQueryVersion, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIQueryVersion; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("QueryVersion... return False"); + return False; + } + *majorVersion = rep.majorVersion; + *minorVersion = rep.minorVersion; + *patchVersion = rep.patchVersion; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("QueryVersion... return True"); + return True; +} + +PUBLIC Bool XF86DRIQueryDirectRenderingCapable(dpy, screen, isCapable) + Display* dpy; + int screen; + Bool* isCapable; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIQueryDirectRenderingCapableReply rep; + xXF86DRIQueryDirectRenderingCapableReq *req; + + TRACE("QueryDirectRenderingCapable..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIQueryDirectRenderingCapable, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIQueryDirectRenderingCapable; + req->screen = screen; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("QueryDirectRenderingCapable... return False"); + return False; + } + *isCapable = rep.isCapable; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("QueryDirectRenderingCapable... return True"); + return True; +} + +PUBLIC Bool XF86DRIOpenConnection(dpy, screen, hSAREA, busIdString) + Display* dpy; + int screen; + drm_handle_t * hSAREA; + char **busIdString; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIOpenConnectionReply rep; + xXF86DRIOpenConnectionReq *req; + + TRACE("OpenConnection..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIOpenConnection, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIOpenConnection; + req->screen = screen; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("OpenConnection... return False"); + return False; + } + + *hSAREA = rep.hSAREALow; + if (sizeof(drm_handle_t) == 8) { + int shift = 32; /* var to prevent warning on next line */ + *hSAREA |= ((drm_handle_t) rep.hSAREAHigh) << shift; + } + + if (rep.length) { + if (!(*busIdString = (char *)Xcalloc(rep.busIdStringLength + 1, 1))) { + _XEatData(dpy, ((rep.busIdStringLength+3) & ~3)); + UnlockDisplay(dpy); + SyncHandle(); + TRACE("OpenConnection... return False"); + return False; + } + _XReadPad(dpy, *busIdString, rep.busIdStringLength); + } else { + *busIdString = NULL; + } + UnlockDisplay(dpy); + SyncHandle(); + TRACE("OpenConnection... return True"); + return True; +} + +PUBLIC Bool XF86DRIAuthConnection(dpy, screen, magic) + Display* dpy; + int screen; + drm_magic_t magic; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIAuthConnectionReq *req; + xXF86DRIAuthConnectionReply rep; + + TRACE("AuthConnection..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIAuthConnection, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIAuthConnection; + req->screen = screen; + req->magic = magic; + rep.authenticated = 0; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse) || !rep.authenticated) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("AuthConnection... return False"); + return False; + } + UnlockDisplay(dpy); + SyncHandle(); + TRACE("AuthConnection... return True"); + return True; +} + +PUBLIC Bool XF86DRICloseConnection(dpy, screen) + Display* dpy; + int screen; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRICloseConnectionReq *req; + + TRACE("CloseConnection..."); + + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRICloseConnection, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRICloseConnection; + req->screen = screen; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("CloseConnection... return True"); + return True; +} + +PUBLIC Bool XF86DRIGetClientDriverName(dpy, screen, ddxDriverMajorVersion, + ddxDriverMinorVersion, ddxDriverPatchVersion, clientDriverName) + Display* dpy; + int screen; + int* ddxDriverMajorVersion; + int* ddxDriverMinorVersion; + int* ddxDriverPatchVersion; + char** clientDriverName; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIGetClientDriverNameReply rep; + xXF86DRIGetClientDriverNameReq *req; + + TRACE("GetClientDriverName..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIGetClientDriverName, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIGetClientDriverName; + req->screen = screen; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetClientDriverName... return False"); + return False; + } + + *ddxDriverMajorVersion = rep.ddxDriverMajorVersion; + *ddxDriverMinorVersion = rep.ddxDriverMinorVersion; + *ddxDriverPatchVersion = rep.ddxDriverPatchVersion; + + if (rep.length) { + if (!(*clientDriverName = (char *)Xcalloc(rep.clientDriverNameLength + 1, 1))) { + _XEatData(dpy, ((rep.clientDriverNameLength+3) & ~3)); + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetClientDriverName... return False"); + return False; + } + _XReadPad(dpy, *clientDriverName, rep.clientDriverNameLength); + } else { + *clientDriverName = NULL; + } + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetClientDriverName... return True"); + return True; +} + +PUBLIC Bool XF86DRICreateContextWithConfig(dpy, screen, configID, context, + hHWContext) + Display* dpy; + int screen; + int configID; + XID* context; + drm_context_t * hHWContext; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRICreateContextReply rep; + xXF86DRICreateContextReq *req; + + TRACE("CreateContext..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRICreateContext, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRICreateContext; + req->visual = configID; + req->screen = screen; + *context = XAllocID(dpy); + req->context = *context; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("CreateContext... return False"); + return False; + } + *hHWContext = rep.hHWContext; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("CreateContext... return True"); + return True; +} + +PUBLIC Bool XF86DRICreateContext(dpy, screen, visual, context, hHWContext) + Display* dpy; + int screen; + Visual* visual; + XID* context; + drm_context_t * hHWContext; +{ + return XF86DRICreateContextWithConfig( dpy, screen, visual->visualid, + context, hHWContext ); +} + +PUBLIC Bool XF86DRIDestroyContext( Display * ndpy, int screen, + XID context ) +{ + Display * const dpy = (Display *) ndpy; + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIDestroyContextReq *req; + + TRACE("DestroyContext..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIDestroyContext, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIDestroyContext; + req->screen = screen; + req->context = context; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("DestroyContext... return True"); + return True; +} + +PUBLIC Bool XF86DRICreateDrawable( Display * ndpy, int screen, + Drawable drawable, drm_drawable_t * hHWDrawable ) +{ + Display * const dpy = (Display *) ndpy; + XExtDisplayInfo *info = find_display (dpy); + xXF86DRICreateDrawableReply rep; + xXF86DRICreateDrawableReq *req; + + TRACE("CreateDrawable..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRICreateDrawable, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRICreateDrawable; + req->screen = screen; + req->drawable = drawable; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("CreateDrawable... return False"); + return False; + } + *hHWDrawable = rep.hHWDrawable; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("CreateDrawable... return True"); + return True; +} + +PUBLIC Bool XF86DRIDestroyDrawable( Display * ndpy, int screen, + Drawable drawable ) +{ + Display * const dpy = (Display *) ndpy; + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIDestroyDrawableReq *req; + + TRACE("DestroyDrawable..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIDestroyDrawable, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIDestroyDrawable; + req->screen = screen; + req->drawable = drawable; + UnlockDisplay(dpy); + SyncHandle(); + TRACE("DestroyDrawable... return True"); + return True; +} + +PUBLIC Bool XF86DRIGetDrawableInfo(Display* dpy, int screen, Drawable drawable, + unsigned int* index, unsigned int* stamp, + int* X, int* Y, int* W, int* H, + int* numClipRects, drm_clip_rect_t ** pClipRects, + int* backX, int* backY, + int* numBackClipRects, drm_clip_rect_t ** pBackClipRects ) +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIGetDrawableInfoReply rep; + xXF86DRIGetDrawableInfoReq *req; + int total_rects; + + TRACE("GetDrawableInfo..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIGetDrawableInfo, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIGetDrawableInfo; + req->screen = screen; + req->drawable = drawable; + + if (!_XReply(dpy, (xReply *)&rep, 1, xFalse)) + { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetDrawableInfo... return False"); + return False; + } + *index = rep.drawableTableIndex; + *stamp = rep.drawableTableStamp; + *X = (int)rep.drawableX; + *Y = (int)rep.drawableY; + *W = (int)rep.drawableWidth; + *H = (int)rep.drawableHeight; + *numClipRects = rep.numClipRects; + total_rects = *numClipRects; + + *backX = rep.backX; + *backY = rep.backY; + *numBackClipRects = rep.numBackClipRects; + total_rects += *numBackClipRects; + +#if 0 + /* Because of the fix in Xserver/GL/dri/xf86dri.c, this check breaks + * backwards compatibility (Because of the >> 2 shift) but the fix + * enables multi-threaded apps to work. + */ + if (rep.length != ((((SIZEOF(xXF86DRIGetDrawableInfoReply) - + SIZEOF(xGenericReply) + + total_rects * sizeof(drm_clip_rect_t)) + 3) & ~3) >> 2)) { + _XEatData(dpy, rep.length); + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetDrawableInfo... return False"); + return False; + } +#endif + + if (*numClipRects) { + int len = sizeof(drm_clip_rect_t) * (*numClipRects); + + *pClipRects = (drm_clip_rect_t *)Xcalloc(len, 1); + if (*pClipRects) + _XRead(dpy, (char*)*pClipRects, len); + } else { + *pClipRects = NULL; + } + + if (*numBackClipRects) { + int len = sizeof(drm_clip_rect_t) * (*numBackClipRects); + + *pBackClipRects = (drm_clip_rect_t *)Xcalloc(len, 1); + if (*pBackClipRects) + _XRead(dpy, (char*)*pBackClipRects, len); + } else { + *pBackClipRects = NULL; + } + + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetDrawableInfo... return True"); + return True; +} + +PUBLIC Bool XF86DRIGetDeviceInfo(dpy, screen, hFrameBuffer, + fbOrigin, fbSize, fbStride, devPrivateSize, pDevPrivate) + Display* dpy; + int screen; + drm_handle_t * hFrameBuffer; + int* fbOrigin; + int* fbSize; + int* fbStride; + int* devPrivateSize; + void** pDevPrivate; +{ + XExtDisplayInfo *info = find_display (dpy); + xXF86DRIGetDeviceInfoReply rep; + xXF86DRIGetDeviceInfoReq *req; + + TRACE("GetDeviceInfo..."); + XF86DRICheckExtension (dpy, info, False); + + LockDisplay(dpy); + GetReq(XF86DRIGetDeviceInfo, req); + req->reqType = info->codes->major_opcode; + req->driReqType = X_XF86DRIGetDeviceInfo; + req->screen = screen; + if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) { + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetDeviceInfo... return False"); + return False; + } + + *hFrameBuffer = rep.hFrameBufferLow; + if (sizeof(drm_handle_t) == 8) { + int shift = 32; /* var to prevent warning on next line */ + *hFrameBuffer |= ((drm_handle_t) rep.hFrameBufferHigh) << shift; + } + + *fbOrigin = rep.framebufferOrigin; + *fbSize = rep.framebufferSize; + *fbStride = rep.framebufferStride; + *devPrivateSize = rep.devPrivateSize; + + if (rep.length) { + if (!(*pDevPrivate = (void *)Xcalloc(rep.devPrivateSize, 1))) { + _XEatData(dpy, ((rep.devPrivateSize+3) & ~3)); + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetDeviceInfo... return False"); + return False; + } + _XRead(dpy, (char*)*pDevPrivate, rep.devPrivateSize); + } else { + *pDevPrivate = NULL; + } + + UnlockDisplay(dpy); + SyncHandle(); + TRACE("GetDeviceInfo... return True"); + return True; +} + +PUBLIC Bool XF86DRIOpenFullScreen(dpy, screen, drawable) + Display* dpy; + int screen; + Drawable drawable; +{ + /* This function and the underlying X protocol are deprecated. + */ + (void) dpy; + (void) screen; + (void) drawable; + return False; +} + +PUBLIC Bool XF86DRICloseFullScreen(dpy, screen, drawable) + Display* dpy; + int screen; + Drawable drawable; +{ + /* This function and the underlying X protocol are deprecated. + */ + (void) dpy; + (void) screen; + (void) drawable; + return True; +} + diff --git a/src/driclient/src/driclient.c b/src/driclient/src/driclient.c new file mode 100644 index 00000000000..7a7ca95702a --- /dev/null +++ b/src/driclient/src/driclient.c @@ -0,0 +1,292 @@ +#include "driclient.h" +#include +#include + +int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri_framebuffer_t *dri_framebuf) +{ + int evbase, errbase; + char *driver_name; + int newly_opened; + drm_magic_t magic; + drmVersionPtr drm_version; + drm_handle_t sarea_handle; + char *bus_id; + dri_screen_t *dri_scrn; + + assert(display); + assert(dri_screen); + + if (!XF86DRIQueryExtension(display, &evbase, &errbase)) + return 1; + + dri_scrn = calloc(1, sizeof(dri_screen_t)); + + if (!dri_scrn) + return 1; + + if (!XF86DRIQueryVersion(display, &dri_scrn->dri.major, &dri_scrn->dri.minor, &dri_scrn->dri.patch)) + goto free_screen; + + dri_scrn->display = display; + dri_scrn->num = screen; + dri_scrn->draw_lock_id = 1; + + if (!XF86DRIOpenConnection(display, screen, &sarea_handle, &bus_id)) + goto free_screen; + + dri_scrn->fd = -1; + dri_scrn->fd = drmOpenOnce(NULL, bus_id, &newly_opened); + XFree(bus_id); + + if (dri_scrn->fd < 0) + goto close_connection; + + if (drmGetMagic(dri_scrn->fd, &magic)) + goto close_drm; + + drm_version = drmGetVersion(dri_scrn->fd); + + if (!drm_version) + goto close_drm; + + dri_scrn->drm.major = drm_version->version_major; + dri_scrn->drm.minor = drm_version->version_minor; + dri_scrn->drm.patch = drm_version->version_patchlevel; + drmFreeVersion(drm_version); + + if (!XF86DRIAuthConnection(display, screen, magic)) + goto close_drm; + + if (!XF86DRIGetClientDriverName + ( + display, + screen, + &dri_scrn->ddx.major, + &dri_scrn->ddx.minor, + &dri_scrn->ddx.patch, + &driver_name + )) + goto close_drm; + + if (drmMap(dri_scrn->fd, sarea_handle, SAREA_MAX, (drmAddress)&dri_scrn->sarea)) + goto close_drm; + + dri_scrn->drawable_hash = drmHashCreate(); + + if (!dri_scrn->drawable_hash) + goto unmap_sarea; + + if (dri_framebuf) + { + if (!XF86DRIGetDeviceInfo + ( + display, + screen, &dri_framebuf->drm_handle, + &dri_framebuf->base, + &dri_framebuf->size, + &dri_framebuf->stride, + &dri_framebuf->private_size, + &dri_framebuf->private + )) + goto destroy_hash; + } + + *dri_screen = dri_scrn; + + return 0; + +destroy_hash: + drmHashDestroy(dri_scrn->drawable_hash); +unmap_sarea: + drmUnmap(dri_scrn->sarea, SAREA_MAX); +close_drm: + drmCloseOnce(dri_scrn->fd); +close_connection: + XF86DRICloseConnection(display, screen); +free_screen: + free(dri_scrn); + + return 1; +} + +int driDestroyScreen(dri_screen_t *dri_screen) +{ + assert(dri_screen); + + drmHashDestroy(dri_screen->drawable_hash); + drmUnmap(dri_screen->sarea, SAREA_MAX); + drmCloseOnce(dri_screen->fd); + XF86DRICloseConnection(dri_screen->display, dri_screen->num); + free(dri_screen); + + return 0; +} + +int driCreateDrawable(dri_screen_t *dri_screen, Drawable drawable, dri_drawable_t **dri_drawable) +{ + int evbase, errbase; + dri_drawable_t *dri_draw; + + assert(dri_screen); + assert(dri_drawable); + + if (!XF86DRIQueryExtension(dri_screen->display, &evbase, &errbase)) + return 1; + + if (!drmHashLookup(dri_screen->drawable_hash, drawable, (void**)dri_drawable)) + { + /* Found */ + (*dri_drawable)->refcount++; + return 0; + } + + dri_draw = calloc(1, sizeof(dri_drawable_t)); + + if (!dri_draw) + return 1; + + if (!XF86DRICreateDrawable(dri_screen->display, 0, drawable, &dri_draw->drm_drawable)) + { + free(dri_draw); + return 1; + } + + dri_draw->x_drawable = drawable; + dri_draw->sarea_index = 0; + dri_draw->sarea_stamp = NULL; + dri_draw->last_sarea_stamp = 0; + dri_draw->dri_screen = dri_screen; + dri_draw->refcount = 1; + + if (drmHashInsert(dri_screen->drawable_hash, drawable, dri_draw)) + { + XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable); + free(dri_draw); + return 1; + } + + /* + * XXX: Need this to initialize sarea pointer and other stuff in dri_drawable_t + * to be able to use the DRI_VALIDATE_DRAWABLE_INFO macro, but is it safe to + * call without any sync? + */ + if (driUpdateDrawableInfo(dri_draw)) + { + XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable); + free(dri_draw); + return 1; + } + + *dri_drawable = dri_draw; + + return 0; +} + +int driUpdateDrawableInfo(dri_drawable_t *dri_drawable) +{ + assert(dri_drawable); + + if (dri_drawable->cliprects) + XFree(dri_drawable->cliprects); + if (dri_drawable->back_cliprects) + XFree(dri_drawable->back_cliprects); + + DRM_SPINUNLOCK(&dri_drawable->dri_screen->sarea->drawable_lock, dri_drawable->dri_screen->draw_lock_id); + + if (!XF86DRIGetDrawableInfo + ( + dri_drawable->dri_screen->display, + dri_drawable->dri_screen->num, + dri_drawable->x_drawable, + &dri_drawable->sarea_index, + &dri_drawable->last_sarea_stamp, + &dri_drawable->x, + &dri_drawable->y, + &dri_drawable->w, + &dri_drawable->h, + &dri_drawable->num_cliprects, + &dri_drawable->cliprects, + &dri_drawable->back_x, + &dri_drawable->back_y, + &dri_drawable->num_back_cliprects, + &dri_drawable->back_cliprects + )) + { + dri_drawable->sarea_stamp = &dri_drawable->last_sarea_stamp; + dri_drawable->num_cliprects = 0; + dri_drawable->cliprects = NULL; + dri_drawable->num_back_cliprects = 0; + dri_drawable->back_cliprects = 0; + + return 1; + } + else + dri_drawable->sarea_stamp = &dri_drawable->dri_screen->sarea->drawableTable[dri_drawable->sarea_index].stamp; + + DRM_SPINLOCK(&dri_drawable->dri_screen->sarea->drawable_lock, dri_drawable->dri_screen->draw_lock_id); + + return 0; +} + +int driDestroyDrawable(dri_drawable_t *dri_drawable) +{ + assert(dri_drawable); + + if (--dri_drawable->refcount == 0) + { + if (dri_drawable->cliprects) + XFree(dri_drawable->cliprects); + if (dri_drawable->back_cliprects) + XFree(dri_drawable->back_cliprects); + drmHashDelete(dri_drawable->dri_screen->drawable_hash, dri_drawable->x_drawable); + XF86DRIDestroyDrawable(dri_drawable->dri_screen->display, dri_drawable->dri_screen->num, dri_drawable->x_drawable); + free(dri_drawable); + } + + return 0; +} + +int driCreateContext(dri_screen_t *dri_screen, Visual *visual, dri_context_t **dri_context) +{ + int evbase, errbase; + dri_context_t *dri_ctx; + + assert(dri_screen); + assert(visual); + assert(dri_context); + + if (!XF86DRIQueryExtension(dri_screen->display, &evbase, &errbase)) + return 1; + + dri_ctx = calloc(1, sizeof(dri_context_t)); + + if (!dri_ctx) + return 1; + + if (!XF86DRICreateContext(dri_screen->display, dri_screen->num, visual, &dri_ctx->id, &dri_ctx->drm_context)) + { + free(dri_ctx); + return 1; + } + + dri_ctx->dri_screen = dri_screen; + *dri_context = dri_ctx; + + return 0; +} + +int driDestroyContext(dri_context_t *dri_context) +{ + assert(dri_context); + + XF86DRIDestroyContext(dri_context->dri_screen->display, dri_context->dri_screen->num, dri_context->id); + free(dri_context); + + return 0; +} + +int driCompareVersions(const dri_version_t *v1, const dri_version_t *v2) +{ + return (v1->major == v2->major) && (v1->minor == v2->minor) && (v1->patch == v2->patch); +} + diff --git a/src/driclient/src/test b/src/driclient/src/test new file mode 100755 index 00000000000..57cddf8e00a Binary files /dev/null and b/src/driclient/src/test differ diff --git a/src/driclient/src/test.c b/src/driclient/src/test.c new file mode 100644 index 00000000000..15f75d928b6 --- /dev/null +++ b/src/driclient/src/test.c @@ -0,0 +1,41 @@ +#include +#include +#include "driclient.h" + +int main(int argc, char **argv) +{ + Display *dpy; + Window root, window; + + dri_screen_t *screen; + dri_drawable_t *dri_drawable; + dri_context_t *context; + + dpy = XOpenDisplay(NULL); + root = XDefaultRootWindow(dpy); + window = XCreateSimpleWindow(dpy, root, 0, 0, 100, 100, 0, 0, 0); + + XSelectInput(dpy, window, 0); + XMapWindow(dpy, window); + XSync(dpy, 0); + + assert(driCreateScreen(dpy, 0, &screen, NULL) == 0); + assert(driCreateDrawable(screen, window, &dri_drawable) == 0); + assert(driCreateContext(screen, XDefaultVisual(dpy, 0), &context) == 0); + assert(driUpdateDrawableInfo(dri_drawable) == 0); + + DRI_VALIDATE_DRAWABLE_INFO(screen, dri_drawable); + + assert(drmGetLock(screen->fd, context->drm_context, 0) == 0); + assert(drmUnlock(screen->fd, context->drm_context) == 0); + + assert(driDestroyContext(context) == 0); + assert(driDestroyDrawable(dri_drawable) == 0); + assert(driDestroyScreen(screen) == 0); + + XDestroyWindow(dpy, window); + XCloseDisplay(dpy); + + return 0; +} + diff --git a/src/driclient/src/xf86dri.h b/src/driclient/src/xf86dri.h new file mode 100644 index 00000000000..baf80a7a9dd --- /dev/null +++ b/src/driclient/src/xf86dri.h @@ -0,0 +1,119 @@ +/************************************************************************** + +Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas. +Copyright 2000 VA Linux Systems, Inc. +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial portions +of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS +OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. +IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR +ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/** + * \file xf86dri.h + * Protocol numbers and function prototypes for DRI X protocol. + * + * \author Kevin E. Martin + * \author Jens Owen + * \author Rickard E. (Rik) Faith + */ + +#ifndef _XF86DRI_H_ +#define _XF86DRI_H_ + +#include +#include + +#define X_XF86DRIQueryVersion 0 +#define X_XF86DRIQueryDirectRenderingCapable 1 +#define X_XF86DRIOpenConnection 2 +#define X_XF86DRICloseConnection 3 +#define X_XF86DRIGetClientDriverName 4 +#define X_XF86DRICreateContext 5 +#define X_XF86DRIDestroyContext 6 +#define X_XF86DRICreateDrawable 7 +#define X_XF86DRIDestroyDrawable 8 +#define X_XF86DRIGetDrawableInfo 9 +#define X_XF86DRIGetDeviceInfo 10 +#define X_XF86DRIAuthConnection 11 +#define X_XF86DRIOpenFullScreen 12 /* Deprecated */ +#define X_XF86DRICloseFullScreen 13 /* Deprecated */ + +#define XF86DRINumberEvents 0 + +#define XF86DRIClientNotLocal 0 +#define XF86DRIOperationNotSupported 1 +#define XF86DRINumberErrors (XF86DRIOperationNotSupported + 1) + +#ifndef _XF86DRI_SERVER_ + +_XFUNCPROTOBEGIN + +Bool XF86DRIQueryExtension( Display *dpy, int *event_base, int *error_base ); + +Bool XF86DRIQueryVersion( Display *dpy, int *majorVersion, int *minorVersion, + int *patchVersion ); + +Bool XF86DRIQueryDirectRenderingCapable( Display *dpy, int screen, + Bool *isCapable ); + +Bool XF86DRIOpenConnection( Display *dpy, int screen, drm_handle_t *hSAREA, + char **busIDString ); + +Bool XF86DRIAuthConnection( Display *dpy, int screen, drm_magic_t magic ); + +Bool XF86DRICloseConnection( Display *dpy, int screen ); + +Bool XF86DRIGetClientDriverName( Display *dpy, int screen, + int *ddxDriverMajorVersion, int *ddxDriverMinorVersion, + int *ddxDriverPatchVersion, char **clientDriverName ); + +Bool XF86DRICreateContext( Display *dpy, int screen, Visual *visual, + XID *ptr_to_returned_context_id, drm_context_t *hHWContext ); + +Bool XF86DRICreateContextWithConfig( Display *dpy, int screen, int configID, + XID *ptr_to_returned_context_id, drm_context_t *hHWContext ); + +Bool XF86DRIDestroyContext( Display *dpy, int screen, + XID context_id ); + +Bool XF86DRICreateDrawable( Display *dpy, int screen, + Drawable drawable, drm_drawable_t *hHWDrawable ); + +Bool XF86DRIDestroyDrawable( Display *dpy, int screen, + Drawable drawable); + +Bool XF86DRIGetDrawableInfo( Display *dpy, int screen, Drawable drawable, + unsigned int *index, unsigned int *stamp, + int *X, int *Y, int *W, int *H, + int *numClipRects, drm_clip_rect_t ** pClipRects, + int *backX, int *backY, + int *numBackClipRects, drm_clip_rect_t **pBackClipRects ); + +Bool XF86DRIGetDeviceInfo( Display *dpy, int screen, + drm_handle_t *hFrameBuffer, int *fbOrigin, int *fbSize, + int *fbStride, int *devPrivateSize, void **pDevPrivate ); + +_XFUNCPROTOEND + +#endif /* _XF86DRI_SERVER_ */ + +#endif /* _XF86DRI_H_ */ + diff --git a/src/driclient/src/xf86dristr.h b/src/driclient/src/xf86dristr.h new file mode 100644 index 00000000000..b834bd1a1a0 --- /dev/null +++ b/src/driclient/src/xf86dristr.h @@ -0,0 +1,342 @@ +/************************************************************************** + +Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas. +Copyright 2000 VA Linux Systems, Inc. +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial portions +of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS +OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. +IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR +ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/* + * Authors: + * Kevin E. Martin + * Jens Owen + * Rickard E. (Rik) Fiath + * + */ + +#ifndef _XF86DRISTR_H_ +#define _XF86DRISTR_H_ + +#include "xf86dri.h" + +#define XF86DRINAME "XFree86-DRI" + +/* The DRI version number. This was originally set to be the same of the + * XFree86 version number. However, this version is really indepedent of + * the XFree86 version. + * + * Version History: + * 4.0.0: Original + * 4.0.1: Patch to bump clipstamp when windows are destroyed, 28 May 02 + * 4.1.0: Add transition from single to multi in DRMInfo rec, 24 Jun 02 + */ +#define XF86DRI_MAJOR_VERSION 4 +#define XF86DRI_MINOR_VERSION 1 +#define XF86DRI_PATCH_VERSION 0 + +typedef struct _XF86DRIQueryVersion { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIQueryVersion */ + CARD16 length B16; +} xXF86DRIQueryVersionReq; +#define sz_xXF86DRIQueryVersionReq 4 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD16 majorVersion B16; /* major version of DRI protocol */ + CARD16 minorVersion B16; /* minor version of DRI protocol */ + CARD32 patchVersion B32; /* patch version of DRI protocol */ + CARD32 pad3 B32; + CARD32 pad4 B32; + CARD32 pad5 B32; + CARD32 pad6 B32; +} xXF86DRIQueryVersionReply; +#define sz_xXF86DRIQueryVersionReply 32 + +typedef struct _XF86DRIQueryDirectRenderingCapable { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* X_DRIQueryDirectRenderingCapable */ + CARD16 length B16; + CARD32 screen B32; +} xXF86DRIQueryDirectRenderingCapableReq; +#define sz_xXF86DRIQueryDirectRenderingCapableReq 8 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + BOOL isCapable; + BOOL pad2; + BOOL pad3; + BOOL pad4; + CARD32 pad5 B32; + CARD32 pad6 B32; + CARD32 pad7 B32; + CARD32 pad8 B32; + CARD32 pad9 B32; +} xXF86DRIQueryDirectRenderingCapableReply; +#define sz_xXF86DRIQueryDirectRenderingCapableReply 32 + +typedef struct _XF86DRIOpenConnection { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIOpenConnection */ + CARD16 length B16; + CARD32 screen B32; +} xXF86DRIOpenConnectionReq; +#define sz_xXF86DRIOpenConnectionReq 8 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 hSAREALow B32; + CARD32 hSAREAHigh B32; + CARD32 busIdStringLength B32; + CARD32 pad6 B32; + CARD32 pad7 B32; + CARD32 pad8 B32; +} xXF86DRIOpenConnectionReply; +#define sz_xXF86DRIOpenConnectionReply 32 + +typedef struct _XF86DRIAuthConnection { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRICloseConnection */ + CARD16 length B16; + CARD32 screen B32; + CARD32 magic B32; +} xXF86DRIAuthConnectionReq; +#define sz_xXF86DRIAuthConnectionReq 12 + +typedef struct { + BYTE type; + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 authenticated B32; + CARD32 pad2 B32; + CARD32 pad3 B32; + CARD32 pad4 B32; + CARD32 pad5 B32; + CARD32 pad6 B32; +} xXF86DRIAuthConnectionReply; +#define zx_xXF86DRIAuthConnectionReply 32 + +typedef struct _XF86DRICloseConnection { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRICloseConnection */ + CARD16 length B16; + CARD32 screen B32; +} xXF86DRICloseConnectionReq; +#define sz_xXF86DRICloseConnectionReq 8 + +typedef struct _XF86DRIGetClientDriverName { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIGetClientDriverName */ + CARD16 length B16; + CARD32 screen B32; +} xXF86DRIGetClientDriverNameReq; +#define sz_xXF86DRIGetClientDriverNameReq 8 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 ddxDriverMajorVersion B32; + CARD32 ddxDriverMinorVersion B32; + CARD32 ddxDriverPatchVersion B32; + CARD32 clientDriverNameLength B32; + CARD32 pad5 B32; + CARD32 pad6 B32; +} xXF86DRIGetClientDriverNameReply; +#define sz_xXF86DRIGetClientDriverNameReply 32 + +typedef struct _XF86DRICreateContext { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRICreateContext */ + CARD16 length B16; + CARD32 screen B32; + CARD32 visual B32; + CARD32 context B32; +} xXF86DRICreateContextReq; +#define sz_xXF86DRICreateContextReq 16 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 hHWContext B32; + CARD32 pad2 B32; + CARD32 pad3 B32; + CARD32 pad4 B32; + CARD32 pad5 B32; + CARD32 pad6 B32; +} xXF86DRICreateContextReply; +#define sz_xXF86DRICreateContextReply 32 + +typedef struct _XF86DRIDestroyContext { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIDestroyContext */ + CARD16 length B16; + CARD32 screen B32; + CARD32 context B32; +} xXF86DRIDestroyContextReq; +#define sz_xXF86DRIDestroyContextReq 12 + +typedef struct _XF86DRICreateDrawable { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRICreateDrawable */ + CARD16 length B16; + CARD32 screen B32; + CARD32 drawable B32; +} xXF86DRICreateDrawableReq; +#define sz_xXF86DRICreateDrawableReq 12 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 hHWDrawable B32; + CARD32 pad2 B32; + CARD32 pad3 B32; + CARD32 pad4 B32; + CARD32 pad5 B32; + CARD32 pad6 B32; +} xXF86DRICreateDrawableReply; +#define sz_xXF86DRICreateDrawableReply 32 + +typedef struct _XF86DRIDestroyDrawable { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIDestroyDrawable */ + CARD16 length B16; + CARD32 screen B32; + CARD32 drawable B32; +} xXF86DRIDestroyDrawableReq; +#define sz_xXF86DRIDestroyDrawableReq 12 + +typedef struct _XF86DRIGetDrawableInfo { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIGetDrawableInfo */ + CARD16 length B16; + CARD32 screen B32; + CARD32 drawable B32; +} xXF86DRIGetDrawableInfoReq; +#define sz_xXF86DRIGetDrawableInfoReq 12 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 drawableTableIndex B32; + CARD32 drawableTableStamp B32; + INT16 drawableX B16; + INT16 drawableY B16; + INT16 drawableWidth B16; + INT16 drawableHeight B16; + CARD32 numClipRects B32; + INT16 backX B16; + INT16 backY B16; + CARD32 numBackClipRects B32; +} xXF86DRIGetDrawableInfoReply; + +#define sz_xXF86DRIGetDrawableInfoReply 36 + + +typedef struct _XF86DRIGetDeviceInfo { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIGetDeviceInfo */ + CARD16 length B16; + CARD32 screen B32; +} xXF86DRIGetDeviceInfoReq; +#define sz_xXF86DRIGetDeviceInfoReq 8 + +typedef struct { + BYTE type; /* X_Reply */ + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 hFrameBufferLow B32; + CARD32 hFrameBufferHigh B32; + CARD32 framebufferOrigin B32; + CARD32 framebufferSize B32; + CARD32 framebufferStride B32; + CARD32 devPrivateSize B32; +} xXF86DRIGetDeviceInfoReply; +#define sz_xXF86DRIGetDeviceInfoReply 32 + +typedef struct _XF86DRIOpenFullScreen { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRIOpenFullScreen */ + CARD16 length B16; + CARD32 screen B32; + CARD32 drawable B32; +} xXF86DRIOpenFullScreenReq; +#define sz_xXF86DRIOpenFullScreenReq 12 + +typedef struct { + BYTE type; + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 isFullScreen B32; + CARD32 pad2 B32; + CARD32 pad3 B32; + CARD32 pad4 B32; + CARD32 pad5 B32; + CARD32 pad6 B32; +} xXF86DRIOpenFullScreenReply; +#define sz_xXF86DRIOpenFullScreenReply 32 + +typedef struct _XF86DRICloseFullScreen { + CARD8 reqType; /* always DRIReqCode */ + CARD8 driReqType; /* always X_DRICloseFullScreen */ + CARD16 length B16; + CARD32 screen B32; + CARD32 drawable B32; +} xXF86DRICloseFullScreenReq; +#define sz_xXF86DRICloseFullScreenReq 12 + +typedef struct { + BYTE type; + BOOL pad1; + CARD16 sequenceNumber B16; + CARD32 length B32; + CARD32 pad2 B32; + CARD32 pad3 B32; + CARD32 pad4 B32; + CARD32 pad5 B32; + CARD32 pad6 B32; + CARD32 pad7 B32; +} xXF86DRICloseFullScreenReply; +#define sz_xXF86DRICloseFullScreenReply 32 + + +#endif /* _XF86DRISTR_H_ */ diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile index 50e3c843b55..c6a22cad4e2 100644 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -2,7 +2,7 @@ TARGET = libg3dvl.a OBJECTS = vl_context.o vl_data.o vl_surface.o vl_shader_build.o vl_util.o GALLIUMDIR = ../.. -CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary +CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary -I${GALLIUMDIR}/winsys/g3dvl ############################################# diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 3d4ca7cf4e2..850a769376e 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -330,6 +330,10 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -354,7 +358,7 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.i_vs = pipe->create_vs_state(pipe, &vs); - free(tokens); + //free(tokens); return 0; } @@ -432,7 +436,7 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.i_fs = pipe->create_fs_state(pipe, &fs); - free(tokens); + //free(tokens); return 0; } @@ -499,6 +503,10 @@ static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -527,7 +535,7 @@ static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.p_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); + //free(tokens); return 0; } @@ -603,6 +611,10 @@ static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -643,7 +655,7 @@ static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.p_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); + //free(tokens); return 0; } @@ -700,6 +712,10 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture @@ -761,7 +777,7 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.p_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); + //free(tokens); return 0; } @@ -821,6 +837,10 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl t0-t4 */ + decl = vl_decl_temps(0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture @@ -928,7 +948,7 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.p_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); + //free(tokens); return 0; } @@ -998,6 +1018,10 @@ static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -1030,7 +1054,7 @@ static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.b_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); + //free(tokens); return 0; } @@ -1104,6 +1128,10 @@ static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -1148,7 +1176,7 @@ static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.b_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); + //free(tokens); return 0; } @@ -1207,6 +1235,10 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl t0-t2 */ + decl = vl_decl_temps(0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture @@ -1283,7 +1315,7 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.b_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); + //free(tokens); return 0; } @@ -1346,6 +1378,10 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl t0-t5 */ + decl = vl_decl_temps(0, 5); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture @@ -1479,7 +1515,7 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.b_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); + //free(tokens); return 0; } @@ -1771,6 +1807,10 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* mov o0, i0 ; Move pos in to pos out */ inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -1789,7 +1829,7 @@ static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.csc.vertex_shader = pipe->create_vs_state(pipe, &vs); - free(tokens); + //free(tokens); return 0; } @@ -1839,6 +1879,10 @@ static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl s0 ; Sampler for tex containing picture to display */ decl = vl_decl_samplers(0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1870,7 +1914,7 @@ static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.csc.fragment_shader = pipe->create_fs_state(pipe, &fs); - free(tokens); + //free(tokens); return 0; } diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c index 365ad697257..1dc5be6fdbb 100644 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.c +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.c @@ -74,6 +74,18 @@ struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int inde return decl; } +struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_TEMPORARY; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last) { struct tgsi_full_declaration decl = tgsi_default_full_declaration(); diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.h b/src/gallium/state_trackers/g3dvl/vl_shader_build.h index 9e64bbeeae0..878d7e2c457 100644 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.h +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.h @@ -14,6 +14,7 @@ struct tgsi_full_declaration vl_decl_interpolated_input ); struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last); struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last); struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last); struct tgsi_full_instruction vl_inst2 ( diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 145ea32892a..9b91ab4e223 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -5,6 +5,7 @@ #include #include #include +#include #include "vl_context.h" #include "vl_defs.h" #include "vl_util.h" @@ -711,12 +712,13 @@ int vlPutSurface pipe->set_sampler_textures(pipe, 1, &surface->texture); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + bind_pipe_drawable(pipe, drawable); /* TODO: Need to take destx, desty into consideration */ pipe->winsys->flush_frontbuffer ( pipe->winsys, surface->context->states.csc.framebuffer.cbufs[0], - &drawable + pipe->priv ); vlBeginRender(surface->context); diff --git a/src/gallium/winsys/g3dvl/nouveau/Makefile b/src/gallium/winsys/g3dvl/nouveau/Makefile new file mode 100644 index 00000000000..644ac083e65 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/Makefile @@ -0,0 +1,46 @@ +TARGET = libnouveau_dri.so +GALLIUMDIR = ../../.. +DRMDIR ?= /usr +DRIDIR = ../../../../driclient + +OBJECTS = nouveau_bo.o nouveau_fence.o nouveau_swapbuffers.o nouveau_channel.o \ + nouveau_grobj.o nouveau_context.o nouveau_winsys.o nouveau_lock.o \ + nouveau_winsys_pipe.o nouveau_device.o nouveau_notifier.o nouveau_dma.o \ + nouveau_pushbuf.o nouveau_resource.o nouveau_screen.o nv04_surface.o \ + nv50_surface.o #nouveau_winsys_softpipe.o + +CFLAGS += -g -Wall -Werror -fPIC \ + -I${GALLIUMDIR}/include \ + -I${GALLIUMDIR}/winsys/g3dvl \ + -I${DRMDIR}/include \ + -I${DRMDIR}/include/drm \ + -I${GALLIUMDIR}/drivers \ + -I${GALLIUMDIR}/auxiliary \ + -I${DRIDIR}/include + +LDFLAGS += -L${DRMDIR}/lib \ + -L${DRIDIR}/lib \ + -L${GALLIUMDIR}/auxiliary/draw \ + -L${GALLIUMDIR}/auxiliary/tgsi \ + -L${GALLIUMDIR}/auxiliary/translate \ + -L${GALLIUMDIR}/auxiliary/rtasm \ + -L${GALLIUMDIR}/auxiliary/cso_cache \ + -L${GALLIUMDIR}/drivers/nv10 \ + -L${GALLIUMDIR}/drivers/nv30 \ + -L${GALLIUMDIR}/drivers/nv40 \ + -L${GALLIUMDIR}/drivers/nv50 + +LIBS += -ldriclient -ldrm -lnv10 -lnv30 -lnv40 -lnv50 -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lm + +############################################# + +.PHONY = all clean + +all: ${TARGET} + +${TARGET}: ${OBJECTS} + $(CC) ${LDFLAGS} -shared -o $@ $^ ${LIBS} + +clean: + rm -rf ${OBJECTS} ${TARGET} + diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c new file mode 120000 index 00000000000..73ac4a4171a --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_bo.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c new file mode 120000 index 00000000000..6c9b2c48d84 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_channel.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c new file mode 100644 index 00000000000..5e173c76728 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c @@ -0,0 +1,371 @@ +#include "pipe/p_defines.h" +#include "pipe/p_context.h" +#include "pipe/p_screen.h" +#include "pipe/p_util.h" + +#include "nouveau_context.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_winsys_pipe.h" + +/* +#ifdef DEBUG +static const struct dri_debug_control debug_control[] = { + { "bo", DEBUG_BO }, + { NULL, 0 } +}; +int __nouveau_debug = 0; +#endif +*/ + +/* + * TODO: Re-examine dri_screen, dri_context, nouveau_screen, nouveau_context + * relationships, seems like there is a lot of room for simplification there. + */ + +static void +nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) +{ + nouveau_grobj_free(&nvc->NvCtxSurf2D); + nouveau_grobj_free(&nvc->NvImageBlit); + nouveau_grobj_free(&nvc->NvGdiRect); + nouveau_grobj_free(&nvc->NvM2MF); + nouveau_grobj_free(&nvc->Nv2D); + nouveau_grobj_free(&nvc->NvSwzSurf); + nouveau_grobj_free(&nvc->NvSIFM); + + nouveau_notifier_free(&nvc->sync_notifier); + + nouveau_channel_free(&nvc->channel); + + FREE(nvc); +} + +static struct nouveau_channel_context * +nouveau_channel_context_create(struct nouveau_device *dev) +{ + struct nouveau_channel_context *nvc; + int ret; + + nvc = CALLOC_STRUCT(nouveau_channel_context); + if (!nvc) + return NULL; + + if ((ret = nouveau_channel_alloc(dev, 0x8003d001, 0x8003d002, + &nvc->channel))) { + NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + nvc->next_handle = 0x80000000; + + if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1, + &nvc->sync_notifier))) { + NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + switch (dev->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + ret = nouveau_surface_channel_create_nv50(nvc); + break; + default: + ret = nouveau_surface_channel_create_nv04(nvc); + break; + } + + if (ret) { + NOUVEAU_ERR("Error initialising surface objects: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + return nvc; +} + +int +nouveau_context_create(dri_context_t *dri_context) +{ + dri_screen_t *dri_screen = dri_context->dri_screen; + struct nouveau_screen *nv_screen = dri_screen->private; + struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); + struct pipe_context *pipe = NULL; + struct nouveau_channel_context *nvc = NULL; + struct nouveau_device *dev = nv_screen->device; + int i; + + switch (dev->chipset & 0xf0) { + case 0x10: + case 0x20: + /* NV10 */ + case 0x30: + /* NV30 */ + case 0x40: + case 0x60: + /* NV40 */ + case 0x50: + case 0x80: + case 0x90: + /* G80 */ + break; + default: + NOUVEAU_ERR("Unsupported chipset: NV%02x\n", dev->chipset); + return 1; + } + + dri_context->private = (void*)nv; + nv->dri_context = dri_context; + nv->nv_screen = nv_screen; + + { + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + nvdev->ctx = dri_context->drm_context; + nvdev->lock = (drmLock*)&dri_screen->sarea->lock; + } + + /* + driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, + nv->dri_screen->myNum, "nouveau"); +#ifdef DEBUG + __nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"), + debug_control); +#endif + */ + + /*XXX: Hack up a fake region and buffer object for front buffer. + * This will go away with TTM, replaced with a simple reference + * of the front buffer handle passed to us by the DDX. + */ + { + struct pipe_surface *fb_surf; + struct nouveau_pipe_buffer *fb_buf; + struct nouveau_bo_priv *fb_bo; + + fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); + fb_bo->drm.offset = nv_screen->front_offset; + fb_bo->drm.flags = NOUVEAU_MEM_FB; + fb_bo->drm.size = nv_screen->front_pitch * + nv_screen->front_height; + fb_bo->refcount = 1; + fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; + fb_bo->base.offset = fb_bo->drm.offset; + fb_bo->base.handle = (unsigned long)fb_bo; + fb_bo->base.size = fb_bo->drm.size; + fb_bo->base.device = nv_screen->device; + + fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer)); + fb_buf->bo = &fb_bo->base; + + fb_surf = calloc(1, sizeof(struct pipe_surface)); + if (nv_screen->front_cpp == 2) + fb_surf->format = PIPE_FORMAT_R5G6B5_UNORM; + else + fb_surf->format = PIPE_FORMAT_A8R8G8B8_UNORM; + pf_get_block(fb_surf->format, &fb_surf->block); + fb_surf->width = nv_screen->front_pitch / nv_screen->front_cpp; + fb_surf->height = nv_screen->front_height; + fb_surf->stride = fb_surf->width * fb_surf->block.size; + fb_surf->refcount = 1; + fb_surf->buffer = &fb_buf->base; + + nv->frontbuffer = fb_surf; + } + + nvc = nv_screen->nvc; + + if (!nvc) { + nvc = nouveau_channel_context_create(dev); + if (!nvc) { + NOUVEAU_ERR("Failed initialising GPU context\n"); + return 1; + } + nv_screen->nvc = nvc; + } + + nvc->refcount++; + nv->nvc = nvc; + + /* Find a free slot for a pipe context, allocate a new one if needed */ + nv->pctx_id = -1; + for (i = 0; i < nvc->nr_pctx; i++) { + if (nvc->pctx[i] == NULL) { + nv->pctx_id = i; + break; + } + } + + if (nv->pctx_id < 0) { + nv->pctx_id = nvc->nr_pctx++; + nvc->pctx = + realloc(nvc->pctx, + sizeof(struct pipe_context *) * nvc->nr_pctx); + } + + /* Create pipe */ + switch (dev->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + if (nouveau_surface_init_nv50(nv)) + return 1; + break; + default: + if (nouveau_surface_init_nv04(nv)) + return 1; + break; + } + + if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { + struct pipe_screen *pscreen; + + pipe = nouveau_pipe_create(nv); + if (!pipe) + NOUVEAU_ERR("Couldn't create hw pipe\n"); + pscreen = nvc->pscreen; + + nv->cap.hw_vertex_buffer = + pscreen->get_param(pscreen, NOUVEAU_CAP_HW_VTXBUF); + nv->cap.hw_index_buffer = + pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF); + } + + /* XXX: nouveau_winsys_softpipe needs a mesa header removed before we can compile it. */ + /* + if (!pipe) { + NOUVEAU_MSG("Using softpipe\n"); + pipe = nouveau_create_softpipe(nv); + if (!pipe) { + NOUVEAU_ERR("Error creating pipe, bailing\n"); + return 1; + } + } + */ + if (!pipe) { + NOUVEAU_ERR("Error creating pipe, bailing\n"); + return 1; + } + + pipe->priv = nv; + + return 0; +} + +void +nouveau_context_destroy(dri_context_t *dri_context) +{ + struct nouveau_context *nv = dri_context->private; + struct nouveau_channel_context *nvc = nv->nvc; + + assert(nv); + + if (nv->pctx_id >= 0) { + nvc->pctx[nv->pctx_id] = NULL; + if (--nvc->refcount <= 0) { + nouveau_channel_context_destroy(nvc); + nv->nv_screen->nvc = NULL; + } + } + + free(nv); +} + +int +nouveau_context_bind(struct nouveau_context *nv, dri_drawable_t *dri_drawable) +{ + assert(nv); + assert(dri_drawable); + + if (nv->dri_drawable != dri_drawable) + { + nv->dri_drawable = dri_drawable; + dri_drawable->private = nv; + } + + return 0; +} + +int +nouveau_context_unbind(struct nouveau_context *nv) +{ + assert(nv); + + nv->dri_drawable = NULL; + + return 0; +} + +/* Show starts here */ + +int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable) +{ + struct nouveau_context *nv; + dri_drawable_t *dri_drawable; + + nv = pipe->priv; + + driCreateDrawable(nv->nv_screen->dri_screen, drawable, &dri_drawable); + + nouveau_context_bind(nv, dri_drawable); + + return 0; +} + +int unbind_pipe_drawable(struct pipe_context *pipe) +{ + nouveau_context_unbind(pipe->priv); + + return 0; +} + +struct pipe_context* create_pipe_context(Display *display, int screen) +{ + dri_screen_t *dri_screen; + dri_framebuffer_t dri_framebuf; + dri_context_t *dri_context; + struct nouveau_context *nv; + + driCreateScreen(display, screen, &dri_screen, &dri_framebuf); + driCreateContext(dri_screen, XDefaultVisual(display, screen), &dri_context); + + nouveau_screen_create(dri_screen, &dri_framebuf); + nouveau_context_create(dri_context); + + nv = dri_context->private; + + return nv->nvc->pctx[nv->pctx_id]; +} + +int destroy_pipe_context(struct pipe_context *pipe) +{ + struct pipe_screen *screen; + struct pipe_winsys *winsys; + struct nouveau_context *nv; + dri_screen_t *dri_screen; + dri_context_t *dri_context; + + assert(pipe); + + screen = pipe->screen; + winsys = pipe->winsys; + nv = pipe->priv; + dri_context = nv->dri_context; + dri_screen = dri_context->dri_screen; + + pipe->destroy(pipe); + screen->destroy(screen); + free(winsys); + + nouveau_context_destroy(dri_context); + nouveau_screen_destroy(dri_screen); + driDestroyContext(dri_context); + driDestroyScreen(dri_screen); + + return 0; +} + diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_context.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.h new file mode 100644 index 00000000000..395a3ab7903 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.h @@ -0,0 +1,105 @@ +#ifndef __NOUVEAU_CONTEXT_H__ +#define __NOUVEAU_CONTEXT_H__ + +/*#include "xmlconfig.h"*/ + +#include +#include "nouveau/nouveau_winsys.h" +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +struct nouveau_channel_context { + struct pipe_screen *pscreen; + int refcount; + + unsigned cur_pctx; + unsigned nr_pctx; + struct pipe_context **pctx; + + struct nouveau_channel *channel; + + struct nouveau_notifier *sync_notifier; + + /* Common */ + struct nouveau_grobj *NvM2MF; + /* NV04-NV40 */ + struct nouveau_grobj *NvCtxSurf2D; + struct nouveau_grobj *NvSwzSurf; + struct nouveau_grobj *NvImageBlit; + struct nouveau_grobj *NvGdiRect; + struct nouveau_grobj *NvSIFM; + /* G80 */ + struct nouveau_grobj *Nv2D; + + uint32_t next_handle; + uint32_t next_subchannel; + uint32_t next_sequence; +}; + +struct nouveau_context { + /* DRI stuff */ + dri_context_t *dri_context; + dri_drawable_t *dri_drawable; + unsigned int last_stamp; + /*driOptionCache dri_option_cache;*/ + drm_context_t drm_context; + drmLock drm_lock; + int locked; + struct nouveau_screen *nv_screen; + struct pipe_surface *frontbuffer; + + struct { + int hw_vertex_buffer; + int hw_index_buffer; + } cap; + + /* Hardware context */ + struct nouveau_channel_context *nvc; + int pctx_id; + + /* pipe_surface accel */ + struct pipe_surface *surf_src, *surf_dst; + unsigned surf_src_offset, surf_dst_offset; + + int (*surface_copy_prep)(struct nouveau_context *, + struct pipe_surface *dst, + struct pipe_surface *src); + void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h); + void (*surface_copy_done)(struct nouveau_context *); + int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned, unsigned); +}; + +extern int nouveau_context_create(dri_context_t *); +extern void nouveau_context_destroy(dri_context_t *); +extern int nouveau_context_bind(struct nouveau_context *, dri_drawable_t *); +extern int nouveau_context_unbind(struct nouveau_context *); + +#ifdef DEBUG +extern int __nouveau_debug; + +#define DEBUG_BO (1 << 0) + +#define DBG(flag, ...) do { \ + if (__nouveau_debug & (DEBUG_##flag)) \ + NOUVEAU_ERR(__VA_ARGS__); \ +} while(0) +#else +#define DBG(flag, ...) +#endif + +extern void LOCK_HARDWARE(struct nouveau_context *); +extern void UNLOCK_HARDWARE(struct nouveau_context *); + +extern int +nouveau_surface_channel_create_nv04(struct nouveau_channel_context *); +extern int +nouveau_surface_channel_create_nv50(struct nouveau_channel_context *); +extern int nouveau_surface_init_nv04(struct nouveau_context *); +extern int nouveau_surface_init_nv50(struct nouveau_context *); + +extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int); +extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *); + +#endif diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c new file mode 120000 index 00000000000..47d52dafa89 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_device.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c new file mode 120000 index 00000000000..45078c964f4 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_dma.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h new file mode 120000 index 00000000000..6b9ec777417 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_dma.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h new file mode 120000 index 00000000000..0e6c9fce355 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_dri.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h new file mode 120000 index 00000000000..473b7d4812d --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_drmif.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c new file mode 120000 index 00000000000..ef1f0c6afe1 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_fence.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c new file mode 120000 index 00000000000..428186544c6 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_grobj.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h new file mode 120000 index 00000000000..6b878aad224 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_local.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c new file mode 100644 index 00000000000..375634bd059 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c @@ -0,0 +1,92 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include +#include +#include "nouveau_context.h" +#include "nouveau_screen.h" + +static pthread_mutex_t lockMutex = PTHREAD_MUTEX_INITIALIZER; + +static void +nouveau_contended_lock(struct nouveau_context *nv, unsigned int flags) +{ + dri_drawable_t *dri_drawable = nv->dri_drawable; + dri_screen_t *dri_screen = nv->dri_context->dri_screen; + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + drmGetLock(nvdev->fd, nvdev->ctx, flags); + + /* If the window moved, may need to set a new cliprect now. + * + * NOTE: This releases and regains the hw lock, so all state + * checking must be done *after* this call: + */ + if (dri_drawable) + DRI_VALIDATE_DRAWABLE_INFO(dri_screen, dri_drawable); +} + +/* Lock the hardware and validate our state. + */ +void +LOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + char __ret=0; + + pthread_mutex_lock(&lockMutex); + assert(!nv->locked); + + DRM_CAS(nvdev->lock, nvdev->ctx, + (DRM_LOCK_HELD | nvdev->ctx), __ret); + + if (__ret) + nouveau_contended_lock(nv, 0); + nv->locked = 1; +} + + +/* Unlock the hardware using the global current context + */ +void +UNLOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + assert(nv->locked); + nv->locked = 0; + + DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx); + + pthread_mutex_unlock(&lockMutex); +} diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c new file mode 120000 index 00000000000..3024a612a7a --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_notifier.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c new file mode 120000 index 00000000000..dae31d97998 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_pushbuf.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c new file mode 120000 index 00000000000..e0d71e9d2bf --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_resource.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c new file mode 100644 index 00000000000..0087ce0056b --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c @@ -0,0 +1,88 @@ +#include "pipe/p_context.h" +#include "pipe/p_util.h" +#include "nouveau_context.h" +#include "nouveau_drm.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11 +#error nouveau_drm.h version does not match expected version +#endif + +/* +PUBLIC const char __driConfigOptions[] = +DRI_CONF_BEGIN +DRI_CONF_END; +static const GLuint __driNConfigOptions = 0; +*/ + +int +nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf) +{ + /* XXX: Someone forgot to bump this? */ + static const dri_version_t ddx_expected = {0, 0, 10 /*NOUVEAU_DRM_HEADER_PATCHLEVEL*/}; + static const dri_version_t dri_expected = {4, 1, 0}; + static const dri_version_t drm_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL}; + + struct nouveau_dri *nv_dri = dri_framebuf->private; + struct nouveau_screen *nv_screen; + int ret; + + if (!driCompareVersions(&ddx_expected, &dri_screen->ddx)) + { + NOUVEAU_ERR("Unexpected DDX version.\n"); + return 1; + } + + if (!driCompareVersions(&drm_expected, &dri_screen->drm)) + { + NOUVEAU_ERR("Unexpected DRM version.\n"); + return 1; + } + + if (!driCompareVersions(&dri_expected, &dri_screen->dri)) + { + NOUVEAU_ERR("Unexpected DRI version.\n"); + return 1; + } + + if (dri_framebuf->private_size != sizeof(struct nouveau_dri)) { + NOUVEAU_ERR("DRI struct mismatch between DDX/DRI.\n"); + return 1; + } + + nv_screen = CALLOC_STRUCT(nouveau_screen); + if (!nv_screen) + return 1; + nv_screen->dri_screen = dri_screen; + dri_screen->private = (void*)nv_screen; + + /* + driParseOptionInfo(&nv_screen->option_cache, + __driConfigOptions, __driNConfigOptions); + */ + + if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, + dri_screen->fd, 0))) { + NOUVEAU_ERR("Failed opening nouveau device: %d.\n", ret); + return 1; + } + + nv_screen->front_offset = nv_dri->front_offset; + nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); + nv_screen->front_cpp = nv_dri->bpp / 8; + nv_screen->front_height = nv_dri->height; + + return 0; +} + +void +nouveau_screen_destroy(dri_screen_t *dri_screen) +{ + struct nouveau_screen *nv_screen = dri_screen->private; + + FREE(nv_screen); +} + diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h new file mode 100644 index 00000000000..8a58bb75564 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h @@ -0,0 +1,24 @@ +#ifndef __NOUVEAU_SCREEN_H__ +#define __NOUVEAU_SCREEN_H__ + +/* TODO: Investigate using DRI options for interesting things */ +/*#include "xmlconfig.h"*/ + +struct nouveau_screen { + dri_screen_t *dri_screen; + struct nouveau_device *device; + struct nouveau_channel_context *nvc; + + uint32_t front_offset; + uint32_t front_pitch; + uint32_t front_cpp; + uint32_t front_height; + + /*driOptionCache option_cache;*/ +}; + +int nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf); +void nouveau_screen_destroy(dri_screen_t *dri_screen); + +#endif + diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c new file mode 100644 index 00000000000..7916c806151 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c @@ -0,0 +1,64 @@ +#include "pipe/p_context.h" +#include "nouveau_context.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +void +nouveau_copy_buffer(dri_drawable_t *dri_drawable, struct pipe_surface *surf, + const drm_clip_rect_t *rect) +{ + struct nouveau_context *nv = dri_drawable->private; + drm_clip_rect_t *pbox; + int nbox, i; + + LOCK_HARDWARE(nv); + if (!dri_drawable->num_cliprects) { + UNLOCK_HARDWARE(nv); + return; + } + pbox = dri_drawable->cliprects; + nbox = dri_drawable->num_cliprects; + + nv->surface_copy_prep(nv, nv->frontbuffer, surf); + for (i = 0; i < nbox; i++, pbox++) { + int sx, sy, dx, dy, w, h; + + sx = pbox->x1 - dri_drawable->x; + sy = pbox->y1 - dri_drawable->y; + dx = pbox->x1; + dy = pbox->y1; + w = pbox->x2 - pbox->x1; + h = pbox->y2 - pbox->y1; + + nv->surface_copy(nv, dx, dy, sx, sy, w, h); + } + + FIRE_RING(nv->nvc->channel); + UNLOCK_HARDWARE(nv); + + //if (nv->last_stamp != dri_drawable->last_sarea_stamp) + //nv->last_stamp = dri_drawable->last_sarea_stamp; +} + +void +nouveau_copy_sub_buffer(dri_drawable_t *dri_drawable, struct pipe_surface *surf, int x, int y, int w, int h) +{ + if (surf) { + drm_clip_rect_t rect; + rect.x1 = x; + rect.y1 = y; + rect.x2 = x + w; + rect.y2 = y + h; + + nouveau_copy_buffer(dri_drawable, surf, &rect); + } +} + +void +nouveau_swap_buffers(dri_drawable_t *dri_drawable, struct pipe_surface *surf) +{ + if (surf) + nouveau_copy_buffer(dri_drawable, surf, NULL); +} + diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h new file mode 100644 index 00000000000..35e934adba8 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h @@ -0,0 +1,10 @@ +#ifndef __NOUVEAU_SWAPBUFFERS_H__ +#define __NOUVEAU_SWAPBUFFERS_H__ + +extern void nouveau_copy_buffer(dri_drawable_t *, struct pipe_surface *, + const drm_clip_rect_t *); +extern void nouveau_copy_sub_buffer(dri_drawable_t *, struct pipe_surface *, + int x, int y, int w, int h); +extern void nouveau_swap_buffers(dri_drawable_t *, struct pipe_surface *); + +#endif diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c new file mode 120000 index 00000000000..43de49b98bb --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_winsys.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c new file mode 100644 index 00000000000..b835bd57606 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c @@ -0,0 +1,261 @@ +#include "pipe/p_winsys.h" +#include "pipe/p_defines.h" +#include "pipe/p_util.h" +#include "pipe/p_inlines.h" + +#include "nouveau_context.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" +#include "nouveau_winsys_pipe.h" + +static void +nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, + void *context_private) +{ + struct nouveau_context *nv = context_private; + dri_drawable_t *dri_drawable = nv->dri_drawable; + + nouveau_copy_buffer(dri_drawable, surf, NULL); +} + +static const char * +nouveau_get_name(struct pipe_winsys *pws) +{ + return "Nouveau/DRI"; +} + +static struct pipe_surface * +nouveau_surface_alloc(struct pipe_winsys *ws) +{ + struct pipe_surface *surf; + + surf = CALLOC_STRUCT(pipe_surface); + if (!surf) + return NULL; + + surf->refcount = 1; + surf->winsys = ws; + return surf; +} + +/* Borrowed from Mesa's xm_winsys */ +static unsigned int +round_up(unsigned n, unsigned multiple) +{ + return (n + multiple - 1) & ~(multiple - 1); +} + +static int +nouveau_surface_alloc_storage +( + struct pipe_winsys *pws, + struct pipe_surface *surface, + unsigned width, + unsigned height, + enum pipe_format format, + unsigned flags, + unsigned tex_usage +) +{ + const unsigned int ALIGNMENT = 256; + + assert(pws); + assert(surface); + + surface->width = width; + surface->height = height; + surface->format = format; + pf_get_block(format, &surface->block); + surface->nblocksx = pf_get_nblocksx(&surface->block, width); + surface->nblocksy = pf_get_nblocksy(&surface->block, height); + surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT); + surface->usage = flags; + surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * surface->nblocksy); + + return 0; +} + +static void +nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) +{ + struct pipe_surface *surf = *s; + + *s = NULL; + if (--surf->refcount <= 0) { + if (surf->buffer) + pipe_buffer_reference(ws, &surf->buffer, NULL); + free(surf); + } +} + +static struct pipe_buffer * +nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, + unsigned usage, unsigned size) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_context *nv = nvpws->nv; + struct nouveau_device *dev = nv->nv_screen->device; + struct nouveau_pipe_buffer *nvbuf; + uint32_t flags; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.alignment = alignment; + nvbuf->base.usage = usage; + nvbuf->base.size = size; + + flags = NOUVEAU_BO_LOCAL; + + if (usage & PIPE_BUFFER_USAGE_PIXEL) { + if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE) + flags |= NOUVEAU_BO_GART; + flags |= NOUVEAU_BO_VRAM; + } + + if (usage & PIPE_BUFFER_USAGE_VERTEX) { + if (nv->cap.hw_vertex_buffer) + flags |= NOUVEAU_BO_GART; + } + + if (usage & PIPE_BUFFER_USAGE_INDEX) { + if (nv->cap.hw_index_buffer) + flags |= NOUVEAU_BO_GART; + } + + if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { + free(nvbuf); + return NULL; + } + + return &nvbuf->base; +} + +static struct pipe_buffer * +nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_pipe_buffer *nvbuf; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.size = bytes; + + if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) { + free(nvbuf); + return NULL; + } + + return &nvbuf->base; +} + +static void +nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + + nouveau_bo_del(&nvbuf->bo); + free(nvbuf); +} + +static void * +nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf, + unsigned flags) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + uint32_t map_flags = 0; + + if (flags & PIPE_BUFFER_USAGE_CPU_READ) + map_flags |= NOUVEAU_BO_RD; + if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) + map_flags |= NOUVEAU_BO_WR; + + if (nouveau_bo_map(nvbuf->bo, map_flags)) + return NULL; + return nvbuf->bo->map; +} + +static void +nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + + nouveau_bo_unmap(nvbuf->bo); +} + +static INLINE struct nouveau_fence * +nouveau_pipe_fence(struct pipe_fence_handle *pfence) +{ + return (struct nouveau_fence *)pfence; +} + +static void +nouveau_pipe_fence_reference(struct pipe_winsys *ws, + struct pipe_fence_handle **ptr, + struct pipe_fence_handle *pfence) +{ + nouveau_fence_ref((void *)pfence, (void *)ptr); +} + +static int +nouveau_pipe_fence_signalled(struct pipe_winsys *ws, + struct pipe_fence_handle *pfence, unsigned flag) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)ws; + struct nouveau_fence *fence = nouveau_pipe_fence(pfence); + + if (nouveau_fence(fence)->signalled == 0) + nouveau_fence_flush(nvpws->nv->nvc->channel); + + return !nouveau_fence(fence)->signalled; +} + +static int +nouveau_pipe_fence_finish(struct pipe_winsys *ws, + struct pipe_fence_handle *pfence, unsigned flag) +{ + struct nouveau_fence *fence = nouveau_pipe_fence(pfence); + struct nouveau_fence *ref = NULL; + + nouveau_fence_ref(fence, &ref); + return nouveau_fence_wait(&ref); +} + +struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv) +{ + struct nouveau_pipe_winsys *nvpws; + struct pipe_winsys *pws; + + nvpws = CALLOC_STRUCT(nouveau_pipe_winsys); + if (!nvpws) + return NULL; + nvpws->nv = nv; + pws = &nvpws->pws; + + pws->flush_frontbuffer = nouveau_flush_frontbuffer; + + pws->surface_alloc = nouveau_surface_alloc; + pws->surface_alloc_storage = nouveau_surface_alloc_storage; + pws->surface_release = nouveau_surface_release; + + pws->buffer_create = nouveau_pipe_bo_create; + pws->buffer_destroy = nouveau_pipe_bo_del; + pws->user_buffer_create = nouveau_pipe_bo_user_create; + pws->buffer_map = nouveau_pipe_bo_map; + pws->buffer_unmap = nouveau_pipe_bo_unmap; + + pws->fence_reference = nouveau_pipe_fence_reference; + pws->fence_signalled = nouveau_pipe_fence_signalled; + pws->fence_finish = nouveau_pipe_fence_finish; + + pws->get_name = nouveau_get_name; + + return &nvpws->pws; +} + diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h new file mode 120000 index 00000000000..264716fef04 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_winsys_pipe.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c new file mode 120000 index 00000000000..83faccde969 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c @@ -0,0 +1 @@ +../../dri/nouveau/nouveau_winsys_softpipe.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c b/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c new file mode 120000 index 00000000000..e05f0671d63 --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c @@ -0,0 +1 @@ +../../dri/nouveau/nv04_surface.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c b/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c new file mode 120000 index 00000000000..3850748229d --- /dev/null +++ b/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c @@ -0,0 +1 @@ +../../dri/nouveau/nv50_surface.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/vl_winsys.h b/src/gallium/winsys/g3dvl/vl_winsys.h new file mode 100644 index 00000000000..c83db28dd98 --- /dev/null +++ b/src/gallium/winsys/g3dvl/vl_winsys.h @@ -0,0 +1,14 @@ +#ifndef vl_winsys_h +#define vl_winsys_h + +#include + +struct pipe_context; + +struct pipe_context* create_pipe_context(Display *display, int screen); +int destroy_pipe_context(struct pipe_context *pipe); +int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable); +int unbind_pipe_drawable(struct pipe_context *pipe); + +#endif + diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index c660c6e0186..0100fe37bdb 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -1,4 +1,4 @@ -#include "xsp_winsys.h" +#include "vl_winsys.h" #include #include #include @@ -11,10 +11,17 @@ struct xsp_pipe_winsys { struct pipe_winsys base; - Display *display; XImage fbimage; }; +struct xsp_context +{ + Display *display; + int screen; + Drawable drawable; + int drawable_bound; +}; + struct xsp_buffer { struct pipe_buffer base; @@ -183,13 +190,18 @@ static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *f static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private) { - struct xsp_pipe_winsys *xsp_winsys; + struct xsp_pipe_winsys *xsp_winsys; + struct xsp_context *xsp_context; assert(pws); assert(surface); assert(context_private); xsp_winsys = (struct xsp_pipe_winsys*)pws; + xsp_context = (struct xsp_context*)context_private; + + if (!xsp_context->drawable_bound) + return; xsp_winsys->fbimage.width = surface->width; xsp_winsys->fbimage.height = surface->height; @@ -198,9 +210,9 @@ static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface * XPutImage ( - xsp_winsys->display, - *(Drawable*)context_private, - XDefaultGC(xsp_winsys->display, XDefaultScreen(xsp_winsys->display)), + xsp_context->display, + xsp_context->drawable, + XDefaultGC(xsp_context->display, xsp_context->screen), &xsp_winsys->fbimage, 0, 0, @@ -209,7 +221,7 @@ static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface * surface->width, surface->height ); - XFlush(xsp_winsys->display); + XFlush(xsp_context->display); pipe_surface_unmap(surface); } @@ -221,11 +233,37 @@ static const char* xsp_get_name(struct pipe_winsys *pws) /* Show starts here */ -struct pipe_context* create_pipe_context(Display *display) +int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable) +{ + struct xsp_context *xsp_context; + + assert(pipe); + + xsp_context = pipe->priv; + xsp_context->drawable = drawable; + xsp_context->drawable_bound = 1; + + return 0; +} + +int unbind_pipe_drawable(struct pipe_context *pipe) +{ + struct xsp_context *xsp_context; + + assert(pipe); + + xsp_context = pipe->priv; + xsp_context->drawable_bound = 0; + + return 0; +} + +struct pipe_context* create_pipe_context(Display *display, int screen) { struct xsp_pipe_winsys *xsp_winsys; - struct pipe_screen *screen; - struct pipe_context *pipe; + struct xsp_context *xsp_context; + struct pipe_screen *sp_screen; + struct pipe_context *sp_pipe; assert(display); @@ -243,7 +281,6 @@ struct pipe_context* create_pipe_context(Display *display) xsp_winsys->base.fence_finish = xsp_fence_finish; xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer; xsp_winsys->base.get_name = xsp_get_name; - xsp_winsys->display = display; { /* XXX: Can't use the returned XImage* directly, @@ -269,10 +306,16 @@ struct pipe_context* create_pipe_context(Display *display) XDestroyImage(template); } - screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); - pipe = softpipe_create(screen, (struct pipe_winsys*)xsp_winsys, NULL); + sp_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); + sp_pipe = softpipe_create(sp_screen, (struct pipe_winsys*)xsp_winsys, NULL); + + xsp_context = calloc(1, sizeof(struct xsp_context)); + xsp_context->display = display; + xsp_context->screen = screen; + + sp_pipe->priv = xsp_context; - return pipe; + return sp_pipe; } int destroy_pipe_context(struct pipe_context *pipe) @@ -284,6 +327,7 @@ int destroy_pipe_context(struct pipe_context *pipe) screen = pipe->screen; winsys = pipe->winsys; + free(pipe->priv); pipe->destroy(pipe); screen->destroy(screen); free(winsys); diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.h b/src/gallium/winsys/g3dvl/xsp_winsys.h deleted file mode 100644 index cb163dc24d2..00000000000 --- a/src/gallium/winsys/g3dvl/xsp_winsys.h +++ /dev/null @@ -1,12 +0,0 @@ -#ifndef xsp_winsys_h -#define xsp_winsys_h - -#include - -struct pipe_context; - -struct pipe_context* create_pipe_context(Display *display); -int destroy_pipe_context(struct pipe_context *pipe); - -#endif - diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile index 4985ecd3e9d..83fcfc523c1 100644 --- a/src/libXvMC/Makefile +++ b/src/libXvMC/Makefile @@ -1,7 +1,12 @@ TARGET = libXvMCg3dvl.so SONAME = libXvMCg3dvl.so.1 GALLIUMDIR = ../gallium -OBJECTS = block.o surface.o context.o subpicture.o attributes.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o + +OBJECTS = block.o surface.o context.o subpicture.o attributes.o + +ifeq (${DRIVER}, softpipe) +OBJECTS += ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o +endif CFLAGS += -g -fPIC -Wall -Werror \ -I${GALLIUMDIR}/state_trackers/g3dvl \ @@ -9,6 +14,8 @@ CFLAGS += -g -fPIC -Wall -Werror \ -I${GALLIUMDIR}/include \ -I${GALLIUMDIR}/auxiliary \ -I${GALLIUMDIR}/drivers + +ifeq (${DRIVER}, softpipe) LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ -L${GALLIUMDIR}/drivers/softpipe \ -L${GALLIUMDIR}/auxiliary/tgsi \ @@ -17,7 +24,17 @@ LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ -L${GALLIUMDIR}/auxiliary/cso_cache \ -L${GALLIUMDIR}/auxiliary/util \ -L${GALLIUMDIR}/auxiliary/rtasm +else +LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ + -L${GALLIUMDIR}/winsys/g3dvl/nouveau \ + -L${GALLIUMDIR}/auxiliary/util +endif + +ifeq (${DRIVER}, softpipe) LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lm +else +LIBS += -lg3dvl -lnouveau_dri -lutil +endif ############################################# diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c index 22eadc1889b..59e1cb2b256 100644 --- a/src/libXvMC/context.c +++ b/src/libXvMC/context.c @@ -2,7 +2,7 @@ #include #include #include -#include +#include static Status Validate(Display *display, XvPortID port, int surface_type_id, unsigned int width, unsigned int height, int flags, int *chroma_format) { @@ -117,7 +117,7 @@ Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, i if (ret != Success) return ret; - pipe = create_pipe_context(display); + pipe = create_pipe_context(display, XDefaultScreen(display)); assert(pipe); -- cgit v1.2.3 From 39793a262f055adf49e6bbd7b74728f744074e8e Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 23 Jul 2008 14:14:55 +1000 Subject: nouveau: is_format_supported() interface changes --- src/gallium/drivers/nv04/nv04_screen.c | 15 ++++++--------- src/gallium/drivers/nv10/nv10_screen.c | 15 ++++++--------- src/gallium/drivers/nv30/nv30_screen.c | 15 ++++++--------- src/gallium/drivers/nv40/nv40_screen.c | 15 ++++++--------- src/gallium/drivers/nv50/nv50_screen.c | 13 +++++-------- 5 files changed, 29 insertions(+), 44 deletions(-) diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c index 9f34117b8c8..da09a3a5fe8 100644 --- a/src/gallium/drivers/nv04/nv04_screen.c +++ b/src/gallium/drivers/nv04/nv04_screen.c @@ -79,10 +79,11 @@ nv04_screen_get_paramf(struct pipe_screen *screen, int param) static boolean nv04_screen_is_format_supported(struct pipe_screen *screen, - enum pipe_format format, uint type) + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) { - switch (type) { - case PIPE_SURFACE: + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: @@ -91,8 +92,7 @@ nv04_screen_is_format_supported(struct pipe_screen *screen, default: break; } - break; - case PIPE_TEXTURE: + } else { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_X8R8G8B8_UNORM: @@ -104,10 +104,7 @@ nv04_screen_is_format_supported(struct pipe_screen *screen, default: break; } - break; - default: - assert(0); - }; + } return FALSE; } diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 67787d8e9ce..403f7b98cde 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -82,10 +82,11 @@ nv10_screen_get_paramf(struct pipe_screen *screen, int param) static boolean nv10_screen_is_format_supported(struct pipe_screen *screen, - enum pipe_format format, uint type) + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) { - switch (type) { - case PIPE_SURFACE: + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: @@ -95,8 +96,7 @@ nv10_screen_is_format_supported(struct pipe_screen *screen, default: break; } - break; - case PIPE_TEXTURE: + } else { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_A1R5G5B5_UNORM: @@ -109,10 +109,7 @@ nv10_screen_is_format_supported(struct pipe_screen *screen, default: break; } - break; - default: - assert(0); - }; + } return FALSE; } diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index b216a703187..0ffcd772351 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -86,10 +86,11 @@ nv30_screen_get_paramf(struct pipe_screen *pscreen, int param) static boolean nv30_screen_surface_format_supported(struct pipe_screen *pscreen, - enum pipe_format format, uint type) + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) { - switch (type) { - case PIPE_SURFACE: + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: @@ -99,8 +100,7 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen, default: break; } - break; - case PIPE_TEXTURE: + } else { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_A1R5G5B5_UNORM: @@ -116,10 +116,7 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen, default: break; } - break; - default: - assert(0); - }; + } return FALSE; } diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index ed0215b4862..3872c0a0c9b 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -91,10 +91,11 @@ nv40_screen_get_paramf(struct pipe_screen *pscreen, int param) static boolean nv40_screen_surface_format_supported(struct pipe_screen *pscreen, - enum pipe_format format, uint type) + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) { - switch (type) { - case PIPE_SURFACE: + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: @@ -104,8 +105,7 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, default: break; } - break; - case PIPE_TEXTURE: + } else { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_A1R5G5B5_UNORM: @@ -125,10 +125,7 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, default: break; } - break; - default: - assert(0); - }; + } return FALSE; } diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 2f3cf041d1c..996b0b3e8f2 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -34,10 +34,11 @@ static boolean nv50_screen_is_format_supported(struct pipe_screen *pscreen, - enum pipe_format format, uint type) + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) { - switch (type) { - case PIPE_SURFACE: + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_R5G6B5_UNORM: @@ -47,8 +48,7 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen, default: break; } - break; - case PIPE_TEXTURE: + } else { switch (format) { case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_A1R5G5B5_UNORM: @@ -62,9 +62,6 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen, default: break; } - break; - default: - assert(0); } return FALSE; -- cgit v1.2.3 From b4d198e47704f3b79c5a61877846172ae9d4b4c0 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 23 Jul 2008 02:28:02 -0400 Subject: g3dvl: Minor fixes. --- src/driclient/include/driclient.h | 1 - src/driclient/src/.gitignore | 1 + src/driclient/src/Makefile | 1 + src/driclient/src/driclient.c | 5 --- src/driclient/src/test | Bin 68389 -> 0 bytes src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c | 42 ++++++++++++---------- 6 files changed, 25 insertions(+), 25 deletions(-) create mode 100644 src/driclient/src/.gitignore delete mode 100755 src/driclient/src/test diff --git a/src/driclient/include/driclient.h b/src/driclient/include/driclient.h index 36438a9f790..d3915250392 100644 --- a/src/driclient/include/driclient.h +++ b/src/driclient/include/driclient.h @@ -68,7 +68,6 @@ int driUpdateDrawableInfo(dri_drawable_t *dri_drawable); int driDestroyDrawable(dri_drawable_t *dri_drawable); int driCreateContext(dri_screen_t *dri_screen, Visual *visual, dri_context_t **dri_context); int driDestroyContext(dri_context_t *dri_context); -int driCompareVersions(const dri_version_t *v1, const dri_version_t *v2); #define DRI_VALIDATE_DRAWABLE_INFO_ONCE(dri_drawable) \ do \ diff --git a/src/driclient/src/.gitignore b/src/driclient/src/.gitignore new file mode 100644 index 00000000000..9daeafb9864 --- /dev/null +++ b/src/driclient/src/.gitignore @@ -0,0 +1 @@ +test diff --git a/src/driclient/src/Makefile b/src/driclient/src/Makefile index 0fac552de5b..3c0fc284409 100644 --- a/src/driclient/src/Makefile +++ b/src/driclient/src/Makefile @@ -12,6 +12,7 @@ all: ${TARGET} test ${TARGET}: ${OBJECTS} ar rcs $@ $^ + if ! test -d ../lib; then mkdir ../lib; fi cp ${TARGET} ../lib test: test.o diff --git a/src/driclient/src/driclient.c b/src/driclient/src/driclient.c index 7a7ca95702a..94d01aca4f9 100644 --- a/src/driclient/src/driclient.c +++ b/src/driclient/src/driclient.c @@ -285,8 +285,3 @@ int driDestroyContext(dri_context_t *dri_context) return 0; } -int driCompareVersions(const dri_version_t *v1, const dri_version_t *v2) -{ - return (v1->major == v2->major) && (v1->minor == v2->minor) && (v1->patch == v2->patch); -} - diff --git a/src/driclient/src/test b/src/driclient/src/test deleted file mode 100755 index 57cddf8e00a..00000000000 Binary files a/src/driclient/src/test and /dev/null differ diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c index 0087ce0056b..daea3fff680 100644 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c @@ -1,7 +1,7 @@ #include "pipe/p_context.h" #include "pipe/p_util.h" #include "nouveau_context.h" -#include "nouveau_drm.h" +#include #include "nouveau_dri.h" #include "nouveau_local.h" #include "nouveau_screen.h" @@ -18,40 +18,44 @@ DRI_CONF_END; static const GLuint __driNConfigOptions = 0; */ -int -nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf) +int nouveau_check_dri_drm_ddx(dri_version_t *dri, dri_version_t *drm, dri_version_t *ddx) { - /* XXX: Someone forgot to bump this? */ - static const dri_version_t ddx_expected = {0, 0, 10 /*NOUVEAU_DRM_HEADER_PATCHLEVEL*/}; - static const dri_version_t dri_expected = {4, 1, 0}; + static const dri_version_t ddx_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL}; + static const dri_version_t dri_expected = {4, 0, 0}; static const dri_version_t drm_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL}; - struct nouveau_dri *nv_dri = dri_framebuf->private; - struct nouveau_screen *nv_screen; - int ret; + assert(dri); + assert(drm); + assert(ddx); - if (!driCompareVersions(&ddx_expected, &dri_screen->ddx)) + if (dri->major != dri_expected.major || dri->minor < dri_expected.minor) { - NOUVEAU_ERR("Unexpected DDX version.\n"); + NOUVEAU_ERR("Unexpected DRI version.\n"); return 1; } - - if (!driCompareVersions(&drm_expected, &dri_screen->drm)) + if (drm->major != drm_expected.major || drm->minor < drm_expected.minor) { NOUVEAU_ERR("Unexpected DRM version.\n"); return 1; } - - if (!driCompareVersions(&dri_expected, &dri_screen->dri)) + if (ddx->major != ddx_expected.major || ddx->minor < ddx_expected.minor) { - NOUVEAU_ERR("Unexpected DRI version.\n"); + NOUVEAU_ERR("Unexpected DDX version.\n"); return 1; } + + return 0; +} - if (dri_framebuf->private_size != sizeof(struct nouveau_dri)) { - NOUVEAU_ERR("DRI struct mismatch between DDX/DRI.\n"); +int +nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf) +{ + struct nouveau_dri *nv_dri = dri_framebuf->private; + struct nouveau_screen *nv_screen; + int ret; + + if (nouveau_check_dri_drm_ddx(&dri_screen->dri, &dri_screen->drm, &dri_screen->ddx)) return 1; - } nv_screen = CALLOC_STRUCT(nouveau_screen); if (!nv_screen) -- cgit v1.2.3 From a8da04cb861b8f9caf3acd33f52f64621f0c15e2 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 23 Jul 2008 23:35:23 -0400 Subject: nv all: Copy shader tokens on create, free on delete. Must copy token stream on shader create, client is allowed to free their copy after creating the state object. --- src/gallium/drivers/nv04/nv04_state.c | 4 +++- src/gallium/drivers/nv04/nv04_state.h | 2 +- src/gallium/drivers/nv10/nv10_state.c | 4 +++- src/gallium/drivers/nv10/nv10_state.h | 2 +- src/gallium/drivers/nv30/nv30_state.c | 8 ++++++-- src/gallium/drivers/nv40/nv40_state.c | 8 ++++++-- src/gallium/drivers/nv50/nv50_state.c | 18 ++++++++++++------ src/gallium/state_trackers/g3dvl/vl_context.c | 20 ++++++++++---------- 8 files changed, 42 insertions(+), 24 deletions(-) diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c index d618465a201..7e71dee7b5d 100644 --- a/src/gallium/drivers/nv04/nv04_state.c +++ b/src/gallium/drivers/nv04/nv04_state.c @@ -4,6 +4,7 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" #include "nv04_context.h" #include "nv04_state.h" @@ -291,7 +292,7 @@ nv04_fp_state_create(struct pipe_context *pipe, struct nv04_fragment_program *fp; fp = CALLOC(1, sizeof(struct nv04_fragment_program)); - fp->pipe = cso; + fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); return (void *)fp; } @@ -313,6 +314,7 @@ nv04_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv04_fragment_program *fp = hwcso; nv04_fragprog_destroy(nv04, fp); + free((void*)fp->pipe.tokens); free(fp); } diff --git a/src/gallium/drivers/nv04/nv04_state.h b/src/gallium/drivers/nv04/nv04_state.h index 7487819c3ab..39f7cd17b30 100644 --- a/src/gallium/drivers/nv04/nv04_state.h +++ b/src/gallium/drivers/nv04/nv04_state.h @@ -47,7 +47,7 @@ struct nv04_fragment_program_data { }; struct nv04_fragment_program { - const struct pipe_shader_state *pipe; + struct pipe_shader_state pipe; struct tgsi_shader_info info; boolean translated; diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 9b8b7801cdf..43b9c32f12d 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -4,6 +4,7 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" +#include "tgsi/util/tgsi_parse.h" #include "nv10_context.h" #include "nv10_state.h" @@ -407,7 +408,7 @@ nv10_fp_state_create(struct pipe_context *pipe, struct nv10_fragment_program *fp; fp = CALLOC(1, sizeof(struct nv10_fragment_program)); - fp->pipe = cso; + fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); tgsi_scan_shader(cso->tokens, &fp->info); @@ -431,6 +432,7 @@ nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv10_fragment_program *fp = hwcso; nv10_fragprog_destroy(nv10, fp); + FREE((void*)fp->pipe.tokens); FREE(fp); } diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h index 3ca501d1356..f1f9a12110f 100644 --- a/src/gallium/drivers/nv10/nv10_state.h +++ b/src/gallium/drivers/nv10/nv10_state.h @@ -79,7 +79,7 @@ struct nv10_fragment_program_data { }; struct nv10_fragment_program { - const struct pipe_shader_state *pipe; + struct pipe_shader_state pipe; struct tgsi_shader_info info; boolean translated; diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 4d6303ebc28..ba02413de5c 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -3,6 +3,8 @@ #include "pipe/p_util.h" #include "pipe/p_inlines.h" +#include "tgsi/util/tgsi_parse.h" + #include "nv30_context.h" #include "nv30_state.h" @@ -503,7 +505,7 @@ nv30_vp_state_create(struct pipe_context *pipe, struct nv30_vertex_program *vp; vp = CALLOC(1, sizeof(struct nv30_vertex_program)); - vp->pipe = *cso; + vp->pipe.tokens = tgsi_dup_tokens(cso->tokens); /*vp->draw = draw_create_vertex_shader(nv30->draw, &vp->pipe);*/ return (void *)vp; @@ -527,6 +529,7 @@ nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso) /*draw_delete_vertex_shader(nv30->draw, vp->draw);*/ nv30_vertprog_destroy(nv30, vp); + FREE((void*)vp->pipe.tokens); FREE(vp); } @@ -537,7 +540,7 @@ nv30_fp_state_create(struct pipe_context *pipe, struct nv30_fragment_program *fp; fp = CALLOC(1, sizeof(struct nv30_fragment_program)); - fp->pipe = *cso; + fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); tgsi_scan_shader(fp->pipe.tokens, &fp->info); @@ -560,6 +563,7 @@ nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv30_fragment_program *fp = hwcso; nv30_fragprog_destroy(nv30, fp); + FREE((void*)fp->pipe.tokens); FREE(fp); } diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index afcf336a65b..5d2c3ab8810 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -5,6 +5,8 @@ #include "draw/draw_context.h" +#include "tgsi/util/tgsi_parse.h" + #include "nv40_context.h" #include "nv40_state.h" @@ -516,7 +518,7 @@ nv40_vp_state_create(struct pipe_context *pipe, struct nv40_vertex_program *vp; vp = CALLOC(1, sizeof(struct nv40_vertex_program)); - vp->pipe = *cso; + vp->pipe.tokens = tgsi_dup_tokens(cso->tokens); vp->draw = draw_create_vertex_shader(nv40->draw, &vp->pipe); return (void *)vp; @@ -540,6 +542,7 @@ nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso) draw_delete_vertex_shader(nv40->draw, vp->draw); nv40_vertprog_destroy(nv40, vp); + FREE((void*)vp->pipe.tokens); FREE(vp); } @@ -550,7 +553,7 @@ nv40_fp_state_create(struct pipe_context *pipe, struct nv40_fragment_program *fp; fp = CALLOC(1, sizeof(struct nv40_fragment_program)); - fp->pipe = *cso; + fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); tgsi_scan_shader(fp->pipe.tokens, &fp->info); @@ -573,6 +576,7 @@ nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso) struct nv40_fragment_program *fp = hwcso; nv40_fragprog_destroy(nv40, fp); + FREE((void*)fp->pipe.tokens); FREE(fp); } diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index c552a0b0aa0..731409bed4e 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -25,6 +25,8 @@ #include "pipe/p_util.h" #include "pipe/p_inlines.h" +#include "tgsi/util/tgsi_parse.h" + #include "nv50_context.h" #include "nv50_texture.h" @@ -438,7 +440,7 @@ nv50_vp_state_create(struct pipe_context *pipe, { struct nv50_program *p = CALLOC_STRUCT(nv50_program); - p->pipe = *cso; + p->pipe.tokens = tgsi_dup_tokens(cso->tokens); p->type = PIPE_SHADER_VERTEX; tgsi_scan_shader(p->pipe.tokens, &p->info); return (void *)p; @@ -457,9 +459,11 @@ static void nv50_vp_state_delete(struct pipe_context *pipe, void *hwcso) { struct nv50_context *nv50 = nv50_context(pipe); + struct nv50_program *p = hwcso; - nv50_program_destroy(nv50, hwcso); - FREE(hwcso); + nv50_program_destroy(nv50, p); + FREE((void*)p->pipe.tokens); + FREE(p); } static void * @@ -468,7 +472,7 @@ nv50_fp_state_create(struct pipe_context *pipe, { struct nv50_program *p = CALLOC_STRUCT(nv50_program); - p->pipe = *cso; + p->pipe.tokens = tgsi_dup_tokens(cso->tokens); p->type = PIPE_SHADER_FRAGMENT; tgsi_scan_shader(p->pipe.tokens, &p->info); return (void *)p; @@ -487,9 +491,11 @@ static void nv50_fp_state_delete(struct pipe_context *pipe, void *hwcso) { struct nv50_context *nv50 = nv50_context(pipe); + struct nv50_program *p = hwcso; - nv50_program_destroy(nv50, hwcso); - FREE(hwcso); + nv50_program_destroy(nv50, p); + FREE((void*)p->pipe.tokens); + FREE(p); } static void diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 850a769376e..638900b3f47 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -358,7 +358,7 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.i_vs = pipe->create_vs_state(pipe, &vs); - //free(tokens); + free(tokens); return 0; } @@ -436,7 +436,7 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.i_fs = pipe->create_fs_state(pipe, &fs); - //free(tokens); + free(tokens); return 0; } @@ -535,7 +535,7 @@ static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.p_vs[0] = pipe->create_vs_state(pipe, &vs); - //free(tokens); + free(tokens); return 0; } @@ -655,7 +655,7 @@ static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.p_vs[1] = pipe->create_vs_state(pipe, &vs); - //free(tokens); + free(tokens); return 0; } @@ -777,7 +777,7 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.p_fs[0] = pipe->create_fs_state(pipe, &fs); - //free(tokens); + free(tokens); return 0; } @@ -948,7 +948,7 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.p_fs[1] = pipe->create_fs_state(pipe, &fs); - //free(tokens); + free(tokens); return 0; } @@ -1054,7 +1054,7 @@ static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.b_vs[0] = pipe->create_vs_state(pipe, &vs); - //free(tokens); + free(tokens); return 0; } @@ -1176,7 +1176,7 @@ static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) vs.tokens = tokens; context->states.mc.b_vs[1] = pipe->create_vs_state(pipe, &vs); - //free(tokens); + free(tokens); return 0; } @@ -1315,7 +1315,7 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.b_fs[0] = pipe->create_fs_state(pipe, &fs); - //free(tokens); + free(tokens); return 0; } @@ -1515,7 +1515,7 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) fs.tokens = tokens; context->states.mc.b_fs[1] = pipe->create_fs_state(pipe, &fs); - //free(tokens); + free(tokens); return 0; } -- cgit v1.2.3 From d1dc937c62083c3d8ab9f420df8e54f5c3217468 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 24 Jul 2008 16:50:18 -0400 Subject: g3dvl: Added more XvMC stubs to get mplayer working. --- src/libXvMC/context.c | 28 +++++++++-- src/libXvMC/subpicture.c | 123 +++++++++++++++++++++++++++++++++++++++++++---- src/libXvMC/surface.c | 17 +++++-- 3 files changed, 150 insertions(+), 18 deletions(-) diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c index 59e1cb2b256..9cf654d6bb2 100644 --- a/src/libXvMC/context.c +++ b/src/libXvMC/context.c @@ -183,18 +183,36 @@ XvMCSurfaceInfo* XvMCListSurfaceTypes(Display *display, XvPortID port, int *num) surface_info->chroma_format = XVMC_CHROMA_FORMAT_420; surface_info->max_width = 2048; surface_info->max_height = 2048; + surface_info->subpicture_max_width = 2048; + surface_info->subpicture_max_height = 2048; surface_info->mc_type = XVMC_IDCT | XVMC_MPEG_2; surface_info->surface_type_id = 123; /* FIXME: XAllocID(display)*/; - surface_info->flags |= XVMC_INTRA_UNSIGNED; + surface_info->flags = XVMC_INTRA_UNSIGNED | XVMC_SUBPICTURE_INDEPENDENT_SCALING | XVMC_BACKEND_SUBPICTURE; return surface_info; } XvImageFormatValues* XvMCListSubpictureTypes(Display* display, XvPortID port, int surface_type_id, int *count_return) { - /* TODO */ - *count_return = 0; - - return NULL; + XvImageFormatValues *image_formats = calloc(1, sizeof(XvImageFormatValues)); + + *count_return = 1; + + image_formats[0].id = 123; + image_formats[0].type = XvRGB; + image_formats[0].byte_order = LSBFirst; + image_formats[0].bits_per_pixel = 8; + image_formats[0].format = XvPacked; + image_formats[0].num_planes = 1; + image_formats[0].depth = 8; + image_formats[0].red_mask = 0x0000FF; + image_formats[0].green_mask = 0x00FF00; + image_formats[0].blue_mask = 0xFF0000; + image_formats[0].component_order[0] = 'R'; + image_formats[0].component_order[0] = 'G'; + image_formats[0].component_order[0] = 'B'; + image_formats[0].scanline_order = XvTopToBottom; + + return image_formats; } diff --git a/src/libXvMC/subpicture.c b/src/libXvMC/subpicture.c index 38d93438334..c8f70c90d0c 100644 --- a/src/libXvMC/subpicture.c +++ b/src/libXvMC/subpicture.c @@ -13,7 +13,33 @@ Status XvMCCreateSubpicture int xvimage_id ) { - return BadImplementation; + assert(display); + + if (!context) + return XvMCBadContext; + + assert(subpicture); + + if (width > 2048 || height > 2048) + return BadValue; + + if (xvimage_id != 123) + return BadMatch; + + subpicture->subpicture_id = XAllocID(display); + subpicture->context_id = context->context_id; + subpicture->xvimage_id = xvimage_id; + subpicture->width = width; + subpicture->height = height; + subpicture->num_palette_entries = 0; + subpicture->entry_bytes = 0; + subpicture->component_order[0] = 0; + subpicture->component_order[1] = 0; + subpicture->component_order[2] = 0; + subpicture->component_order[3] = 0; + /* TODO: subpicture->privData = ;*/ + + return Success; } Status XvMCClearSubpicture @@ -27,7 +53,14 @@ Status XvMCClearSubpicture unsigned int color ) { - return BadImplementation; + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + + /* TODO: Assert clear rect is within bounds? Or clip? */ + + return Success; } Status XvMCCompositeSubpicture @@ -43,17 +76,42 @@ Status XvMCCompositeSubpicture short dsty ) { - return BadImplementation; + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + + assert(image); + + if (subpicture->xvimage_id != image->id) + return BadMatch; + + /* TODO: Assert rects are within bounds? Or clip? */ + + return Success; } Status XvMCDestroySubpicture(Display *display, XvMCSubpicture *subpicture) { + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + return BadImplementation; } Status XvMCSetSubpicturePalette(Display *display, XvMCSubpicture *subpicture, unsigned char *palette) { - return BadImplementation; + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + + assert(palette); + + /* We don't support paletted subpictures */ + return BadMatch; } Status XvMCBlendSubpicture @@ -71,7 +129,19 @@ Status XvMCBlendSubpicture unsigned short surfh ) { - return BadImplementation; + assert(display); + + if (!target_surface) + return XvMCBadSurface; + + if (!subpicture) + return XvMCBadSubpicture; + + if (target_surface->context_id != subpicture->context_id) + return BadMatch; + + /* TODO: Assert rects are within bounds? Or clip? */ + return Success; } Status XvMCBlendSubpicture2 @@ -90,21 +160,56 @@ Status XvMCBlendSubpicture2 unsigned short surfh ) { - return BadImplementation; + assert(display); + + if (!source_surface || !target_surface) + return XvMCBadSurface; + + if (!subpicture) + return XvMCBadSubpicture; + + if (source_surface->context_id != subpicture->context_id) + return BadMatch; + + if (source_surface->context_id != subpicture->context_id) + return BadMatch; + + /* TODO: Assert rects are within bounds? Or clip? */ + return Success; } Status XvMCSyncSubpicture(Display *display, XvMCSubpicture *subpicture) { - return BadImplementation; + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + + return Success; } Status XvMCFlushSubpicture(Display *display, XvMCSubpicture *subpicture) { - return BadImplementation; + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + + return Success; } Status XvMCGetSubpictureStatus(Display *display, XvMCSubpicture *subpicture, int *status) { - return BadImplementation; + assert(display); + + if (!subpicture) + return XvMCBadSubpicture; + + assert(status); + + /* TODO */ + *status = 0; + + return Success; } diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index a550114655c..1c07220e844 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -281,14 +281,23 @@ Status XvMCRenderSurface Status XvMCFlushSurface(Display *display, XvMCSurface *surface) { + assert(display); + + if (!surface) + return XvMCBadSurface; + /* TODO: Check display & surface match */ - return BadImplementation; + return Success; } Status XvMCSyncSurface(Display *display, XvMCSurface *surface) { - /* TODO: Check display & surface match */ - return BadImplementation; + assert(display); + + if (!surface) + return XvMCBadSurface; + + return Success; } Status XvMCPutSurface @@ -358,7 +367,7 @@ Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status) /* TODO */ *status = 0; - return BadImplementation; + return Success; } Status XvMCDestroySurface(Display *display, XvMCSurface *surface) -- cgit v1.2.3 From a17e6c046cd990f2ed4d0dfe9ed0a59bccd9aade Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 26 Jul 2008 16:51:09 -0400 Subject: g3dvl: Recursively build dependencies. Run `make` in src/libXvMC and everything should be built for Nouveau. Run `make DRIVER=softpipe` in src/libXvMC for SoftPipe. --- src/driclient/src/Makefile | 2 +- src/gallium/winsys/g3dvl/nouveau/Makefile | 2 ++ src/libXvMC/Makefile | 14 ++++++++++++++ 3 files changed, 17 insertions(+), 1 deletion(-) diff --git a/src/driclient/src/Makefile b/src/driclient/src/Makefile index 3c0fc284409..89ffb84477f 100644 --- a/src/driclient/src/Makefile +++ b/src/driclient/src/Makefile @@ -19,5 +19,5 @@ test: test.o $(CC) -L../lib -L${DRMDIR}/lib ${LDFLAGS} -o $@ $^ -ldriclient -lX11 -lXext -ldrm clean: - rm -rf ${OBJECTS} ${TARGET} test test.o + rm -rf ${OBJECTS} ${TARGET} ../lib/${TARGET} test test.o diff --git a/src/gallium/winsys/g3dvl/nouveau/Makefile b/src/gallium/winsys/g3dvl/nouveau/Makefile index 644ac083e65..2861bd7db4f 100644 --- a/src/gallium/winsys/g3dvl/nouveau/Makefile +++ b/src/gallium/winsys/g3dvl/nouveau/Makefile @@ -39,8 +39,10 @@ LIBS += -ldriclient -ldrm -lnv10 -lnv30 -lnv40 -lnv50 -ldraw -ltgsi -ltranslate all: ${TARGET} ${TARGET}: ${OBJECTS} + cd ${DRIDIR}/src; ${MAKE} $(CC) ${LDFLAGS} -shared -o $@ $^ ${LIBS} clean: + cd ${DRIDIR}/src; ${MAKE} clean rm -rf ${OBJECTS} ${TARGET} diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile index 83fcfc523c1..c154e17dc35 100644 --- a/src/libXvMC/Makefile +++ b/src/libXvMC/Makefile @@ -42,9 +42,23 @@ endif all: ${TARGET} +ifeq (${DRIVER}, softpipe) +${TARGET}: ${OBJECTS} + cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} + $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ $^ ${LIBS} + +clean: + cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean + rm -rf ${OBJECTS} ${TARGET} +else ${TARGET}: ${OBJECTS} + cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} + cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ $^ ${LIBS} clean: + cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean + cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} clean rm -rf ${OBJECTS} ${TARGET} +endif -- cgit v1.2.3 From 1d27b4bc0f291ec955e59b1ead943100d8a15505 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 30 Jul 2008 16:47:10 -0400 Subject: g3dvl: Use R16_SNORM instead of A8L8_UNORM for block rendering. --- src/gallium/state_trackers/g3dvl/vl_context.c | 248 ++++++++++++-------------- src/gallium/state_trackers/g3dvl/vl_data.c | 3 +- src/gallium/state_trackers/g3dvl/vl_surface.c | 178 ++++-------------- 3 files changed, 153 insertions(+), 276 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 638900b3f47..542ba996a7e 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -365,7 +365,7 @@ static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) { - const unsigned int max_tokens = 50; + const unsigned int max_tokens = 100; struct pipe_context *pipe; struct pipe_shader_state fs; @@ -402,11 +402,19 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* decl o0 ; Fragment color */ decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture @@ -419,16 +427,30 @@ static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) } /* - * tex2d o0.x, i0, s0 ; Read texel from luma texture into .x channel - * tex2d o0.y, i1, s1 ; Read texel from chroma Cb texture into .y channel - * tex2d o0.z, i1, s2 ; Read texel from chroma Cr texture into .z channel - */ + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul o0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* end */ inst = vl_end(); @@ -701,10 +723,7 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* - * decl c0 ; Multiplier to shift 9th bit of differential into place - * decl c1 ; Bias to get differential back to a signed value - */ + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -729,38 +748,29 @@ static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) } /* - * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - * mov t1.x, t0.w ; Move 9th bit from .w channel to .x - * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - * mov t1.y, t0.w ; Move 9th bit from .w channel to .y - * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } - /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c1 ; Subtract bias to get back signed values */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* tex2d t1, i2, s3 ; Read texel from ref macroblock */ @@ -826,11 +836,10 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) } /* - * decl c0 ; Multiplier to shift 9th bit of differential into place - * decl c1 ; Bias to get differential back to a signed value - * decl c2 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 2); + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* decl o0 ; Fragment color */ @@ -854,42 +863,35 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) } /* - * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - * mov t1.x, t0.w ; Move 9th bit from .w channel to .x - * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - * mov t1.y, t0.w ; Move 9th bit from .w channel to .y - * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } - /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c1 ; Subtract bias to get back signed values */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t1, i2, s3 ; Read texel from ref macroblock top field - tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field */ + /* + * tex2d t1, i2, s3 ; Read texel from ref macroblock top field + * tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field + */ for (i = 0; i < 2; ++i) { inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); @@ -897,8 +899,8 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) } /* XXX: Pos values off by 0.5? */ - /* sub t4, i4.y, c2.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 4, TGSI_FILE_INPUT, 2); + /* sub t4, i4.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 4, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; @@ -909,8 +911,8 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t4, c2.x ; Multiply pos Y-coord by 1/2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 2); + /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; @@ -921,8 +923,8 @@ static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t3, c2.y ; Multiply by 2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 2); + /* mul t3, t3, c1.y ; Multiply by 2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; @@ -1224,11 +1226,10 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) } /* - * decl c0 ; Multiplier to shift 9th bit of differential into place - * decl c1 ; Bias to get differential back to a signed value - * decl c2 ; Constant 1/2 in .x channel to use as weight to blend past and future texels + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constant 1/2 in .x channel to use as weight to blend past and future texels */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 2); + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* decl o0 ; Fragment color */ @@ -1253,38 +1254,29 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) } /* - * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - * mov t1.x, t0.w ; Move 9th bit from .w channel to .x - * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - * mov t1.y, t0.w ; Move 9th bit from .w channel to .y - * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } - /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c1 ; Subtract bias to get back signed values */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* @@ -1297,8 +1289,8 @@ static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* lerp t1, c2.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; @@ -1366,12 +1358,11 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) } /* - * decl c0 ; Multiplier to shift 9th bit of differential into place - * decl c1 ; Bias to get differential back to a signed value - * decl c2 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels * ; and for Y-mod-2 top/bottom field selection */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 2); + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* decl o0 ; Fragment color */ @@ -1396,43 +1387,34 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) } /* - * tex2d t0.xw, i0, s0 ; Read texel from luma texture into .x and .w channels - * mov t1.x, t0.w ; Move 9th bit from .w channel to .x - * tex2d t0.yw, i1, s1 ; Read texel from chroma Cb texture into .y and .w channels - * mov t1.y, t0.w ; Move 9th bit from .w channel to .y - * tex2d t0.zw, i1, s2 ; Read texel from chroma Cr texture into .z and .w channels - * mov t1.z, t0.w ; Move 9th bit from .w channel to .z + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - inst.FullDstRegisters[0].DstRegister.WriteMask = (TGSI_WRITEMASK_X << i) | TGSI_WRITEMASK_W; + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_W; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } - /* mul t1, t1, c0 ; Muliply 9th bit by multiplier to shift it into place */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add t0, t0, t1 ; Add luma and chroma low and high parts to get normalized unsigned 9-bit values */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c1 ; Subtract bias to get back signed values */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* XXX: Pos values off by 0.5? */ - /* sub t4, i6.y, c2.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 6, TGSI_FILE_CONSTANT, 2); + /* sub t4, i6.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 6, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; @@ -1443,8 +1425,8 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t4, c2.x ; Multiply pos Y-coord by 1/2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 2); + /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; @@ -1455,8 +1437,8 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* mul t3, t3, c2.y ; Multiply by 2 */ - inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 2); + /* mul t3, t3, c1.y ; Multiply by 2 */ + inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; @@ -1497,8 +1479,8 @@ static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* lerp t1, c2.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; @@ -1678,7 +1660,7 @@ static int vlInitMC(struct VL_CONTEXT *context) memset(&template, 0, sizeof(struct pipe_texture)); template.target = PIPE_TEXTURE_2D; - template.format = PIPE_FORMAT_A8L8_UNORM; + template.format = PIPE_FORMAT_R16_SNORM; template.last_level = 0; template.width[0] = 8; template.height[0] = 8 * 4; diff --git a/src/gallium/state_trackers/g3dvl/vl_data.c b/src/gallium/state_trackers/g3dvl/vl_data.c index 7e6ee8ac12f..0e5c8c77f90 100644 --- a/src/gallium/state_trackers/g3dvl/vl_data.c +++ b/src/gallium/state_trackers/g3dvl/vl_data.c @@ -86,8 +86,7 @@ const struct VL_TEXCOORD2F *vl_surface_texcoords = (const struct VL_TEXCOORD2F*) */ const struct VL_MC_FS_CONSTS vl_mc_fs_consts = { - {256.0f, 256.0f, 256.0f, 0.0f}, - {256.0f / 255.0f, 256.0f / 255.0f, 256.0f / 255.0f, 0.0f}, + {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, {0.5f, 2.0f, 0.0f, 0.0f} }; diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 9b91ab4e223..6d4e14b95c3 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -63,7 +63,7 @@ static int vlTransformBlock(short *src, short *dst, short bias) return 0; } -static int vlGrabFrameCodedFullBlock(short *src, short *dst, unsigned int dst_pitch) +static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) { unsigned int y; @@ -78,18 +78,7 @@ static int vlGrabFrameCodedFullBlock(short *src, short *dst, unsigned int dst_pi return 0; } -static int vlGrabFrameCodedDiffBlock(short *src, short *dst, unsigned int dst_pitch) -{ - unsigned int x, y; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - dst[y * dst_pitch + x] = src[y * VL_BLOCK_WIDTH + x] + 0x100; - - return 0; -} - -static int vlGrabFieldCodedFullBlock(short *src, short *dst, unsigned int dst_pitch) +static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) { unsigned int y; @@ -114,30 +103,17 @@ static int vlGrabFieldCodedFullBlock(short *src, short *dst, unsigned int dst_pi return 0; } -static int vlGrabFieldCodedDiffBlock(short *src, short *dst, unsigned int dst_pitch) -{ - unsigned int x, y; - - for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - dst[y * dst_pitch * 2 + x] = src[y * VL_BLOCK_WIDTH + x] + 0x100; - - dst += VL_BLOCK_HEIGHT * dst_pitch; - - for (; y < VL_BLOCK_HEIGHT; ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - dst[y * dst_pitch * 2 + x] = src[y * VL_BLOCK_WIDTH + x] + 0x100; - - return 0; -} - static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) { - unsigned int x, y; + unsigned int y; for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - dst[y * dst_pitch + x] = 0x100; + memset + ( + dst + y * dst_pitch, + 0, + VL_BLOCK_WIDTH * 2 + ); return 0; } @@ -156,7 +132,6 @@ static int vlGrabBlocks unsigned int tex_pitch; unsigned int tb, sb = 0; - const int do_idct = 1; short temp_block[64]; assert(context); @@ -176,80 +151,26 @@ static int vlGrabBlocks { if ((coded_block_pattern >> (5 - tb)) & 1) { + if (sample_type == VL_FULL_SAMPLE) + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); + else + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + if (dct_type == VL_DCT_FRAME_CODED) - if (sample_type == VL_FULL_SAMPLE) - if (do_idct) - { - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); - vlGrabFrameCodedFullBlock - ( - temp_block, - texels + tb * tex_pitch * VL_BLOCK_HEIGHT, - tex_pitch - ); - } - else - vlGrabFrameCodedFullBlock - ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + tb * tex_pitch * VL_BLOCK_HEIGHT, - tex_pitch - ); - else - if (do_idct) - { - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); - vlGrabFrameCodedDiffBlock - ( - temp_block, - texels + tb * tex_pitch * VL_BLOCK_HEIGHT, - tex_pitch - ); - } - else - vlGrabFrameCodedDiffBlock - ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + tb * tex_pitch * VL_BLOCK_HEIGHT, - tex_pitch - ); + vlGrabFrameCodedBlock + ( + temp_block, + texels + tb * tex_pitch * VL_BLOCK_HEIGHT, + tex_pitch + ); else - if (sample_type == VL_FULL_SAMPLE) - if (do_idct) - { - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); - vlGrabFieldCodedFullBlock - ( - temp_block, - texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, - tex_pitch - ); - } - else - vlGrabFieldCodedFullBlock - ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, - tex_pitch - ); - else - if (do_idct) - { - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); - vlGrabFieldCodedDiffBlock - ( - temp_block, - texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, - tex_pitch - ); - } - else - vlGrabFieldCodedDiffBlock - ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, - tex_pitch - ); + vlGrabFieldCodedBlock + ( + blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, + tex_pitch + ); + ++sb; } else @@ -272,43 +193,18 @@ static int vlGrabBlocks tex_pitch = tex_surface->stride / tex_surface->block.size; if ((coded_block_pattern >> (1 - tb)) & 1) - { + { if (sample_type == VL_FULL_SAMPLE) - if (do_idct) - { - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); - vlGrabFrameCodedFullBlock - ( - temp_block, - texels, - tex_pitch - ); - } - else - vlGrabFrameCodedFullBlock - ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels, - tex_pitch - ); + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); else - if (do_idct) - { - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); - vlGrabFrameCodedDiffBlock - ( - temp_block, - texels, - tex_pitch - ); - } - else - vlGrabFrameCodedDiffBlock - ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, - texels, - tex_pitch - ); + vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + + vlGrabFrameCodedBlock + ( + temp_block, + texels, + tex_pitch + ); ++sb; } -- cgit v1.2.3 From a55ced56760f90fe84794b23c557050403d514b2 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 3 Aug 2008 17:20:48 -0400 Subject: nv40: Support for PIPE_FORMAT_R16_SNORM. --- src/gallium/drivers/nouveau/nouveau_class.h | 1 + src/gallium/drivers/nv40/nv40_fragtex.c | 36 ++++++++++++++++------------- src/gallium/drivers/nv40/nv40_screen.c | 3 ++- 3 files changed, 23 insertions(+), 17 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h index 749fbf041e7..c3d8d7539d1 100644 --- a/src/gallium/drivers/nouveau/nouveau_class.h +++ b/src/gallium/drivers/nouveau/nouveau_class.h @@ -5194,6 +5194,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define NV40TCL_TEX_FORMAT_FORMAT_A8L8 0x00000b00 #define NV40TCL_TEX_FORMAT_FORMAT_Z24 0x00001000 #define NV40TCL_TEX_FORMAT_FORMAT_Z16 0x00001200 +#define NV40TCL_TEX_FORMAT_FORMAT_A16 0x00001400 #define NV40TCL_TEX_FORMAT_FORMAT_HILO8 0x00001800 #define NV40TCL_TEX_FORMAT_FORMAT_RGBA16F 0x00001a00 #define NV40TCL_TEX_FORMAT_FORMAT_RGBA32F 0x00001b00 diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index c79ea8becb3..2d45c2545cd 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -1,6 +1,6 @@ #include "nv40_context.h" -#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w,sx,sy,sz,sw) \ { \ TRUE, \ PIPE_FORMAT_##m, \ @@ -9,6 +9,8 @@ NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w | \ NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y | \ NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w), \ + ((NV40TCL_TEX_FILTER_SIGNED_RED*sx) | (NV40TCL_TEX_FILTER_SIGNED_GREEN*sy) | \ + (NV40TCL_TEX_FILTER_SIGNED_BLUE*sz) | (NV40TCL_TEX_FILTER_SIGNED_ALPHA*sw)) \ } struct nv40_texture_format { @@ -16,24 +18,26 @@ struct nv40_texture_format { uint pipe; int format; int swizzle; + int sign; }; static struct nv40_texture_format nv40_texture_formats[] = { - _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), - _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), - _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), - _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X), - _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X), - _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y), - _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), - _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), - _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W), - _(DXT1_RGBA , DXT1 , S1, S1, S1, S1, X, Y, Z, W), - _(DXT3_RGBA , DXT3 , S1, S1, S1, S1, X, Y, Z, W), - _(DXT5_RGBA , DXT5 , S1, S1, S1, S1, X, Y, Z, W), + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W, 0, 0, 0, 0), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W, 0, 0, 0, 0), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W, 0, 0, 0, 0), + _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W, 0, 0, 0, 0), + _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X, 0, 0, 0, 0), + _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X, 0, 0, 0, 0), + _(R16_SNORM , A16 , ZERO, ZERO, S1, ONE, X, X, X, Y, 1, 1, 1, 1), + _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X, 0, 0, 0, 0), + _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y, 0, 0, 0, 0), + _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X, 0, 0, 0, 0), + _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X, 0, 0, 0, 0), + _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W, 0, 0, 0, 0), + _(DXT1_RGBA , DXT1 , S1, S1, S1, S1, X, Y, Z, W, 0, 0, 0, 0), + _(DXT3_RGBA , DXT3 , S1, S1, S1, S1, X, Y, Z, W, 0, 0, 0, 0), + _(DXT5_RGBA , DXT5 , S1, S1, S1, S1, X, Y, Z, W, 0, 0, 0, 0), {}, }; @@ -113,7 +117,7 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit) so_data (so, ps->wrap); so_data (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en); so_data (so, txs); - so_data (so, ps->filt | 0x2000 /*voodoo*/); + so_data (so, ps->filt | tf->sign | 0x2000 /*voodoo*/); so_data (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) | pt->height[0]); so_data (so, ps->bcol); diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 3872c0a0c9b..a63a6136c52 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -110,7 +110,8 @@ nv40_screen_surface_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_A8R8G8B8_UNORM: case PIPE_FORMAT_A1R5G5B5_UNORM: case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_R16_SNORM: case PIPE_FORMAT_L8_UNORM: case PIPE_FORMAT_A8_UNORM: case PIPE_FORMAT_I8_UNORM: -- cgit v1.2.3 From 7fde9febd6f212494730ebef916fe25c95d30be9 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 3 Aug 2008 18:21:13 -0400 Subject: g3dvl: Temporarily disable IDCT. --- src/gallium/state_trackers/g3dvl/vl_context.c | 4 ++-- src/gallium/state_trackers/g3dvl/vl_surface.c | 30 +++++++++++++++------------ 2 files changed, 19 insertions(+), 15 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 542ba996a7e..d446d218a44 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -1625,8 +1625,8 @@ static int vlInitMC(struct VL_CONTEXT *context) context->states.mc.viewport.translate[2] = 0; context->states.mc.viewport.translate[3] = 0; - context->states.mc.render_target.width = context->video_width; - context->states.mc.render_target.height = context->video_height; + context->states.mc.render_target.width = vlRoundUpPOT(context->video_width); + context->states.mc.render_target.height = vlRoundUpPOT(context->video_height); context->states.mc.render_target.num_cbufs = 1; /* FB for MC stage is a VL_SURFACE, set in vlSetRenderSurface() */ context->states.mc.render_target.zsbuf = NULL; diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 6d4e14b95c3..1386b1107cc 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -10,6 +10,9 @@ #include "vl_defs.h" #include "vl_util.h" +/*#define DO_IDCT*/ + +#ifdef DO_IDCT static int vlTransformBlock(short *src, short *dst, short bias) { static const float basis[8][8] = @@ -62,6 +65,7 @@ static int vlTransformBlock(short *src, short *dst, short bias) } return 0; } +#endif static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) { @@ -132,8 +136,6 @@ static int vlGrabBlocks unsigned int tex_pitch; unsigned int tb, sb = 0; - short temp_block[64]; - assert(context); assert(blocks); @@ -151,22 +153,23 @@ static int vlGrabBlocks { if ((coded_block_pattern >> (5 - tb)) & 1) { - if (sample_type == VL_FULL_SAMPLE) - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); - else - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; + +#ifdef DO_IDCT + vlTransformBlock(cur_block, cur_block, sample_type == VL_FULL_SAMPLE ? 128 : 0); +#endif if (dct_type == VL_DCT_FRAME_CODED) vlGrabFrameCodedBlock ( - temp_block, + cur_block, texels + tb * tex_pitch * VL_BLOCK_HEIGHT, tex_pitch ); else vlGrabFieldCodedBlock ( - blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, + cur_block, texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, tex_pitch ); @@ -194,14 +197,15 @@ static int vlGrabBlocks if ((coded_block_pattern >> (1 - tb)) & 1) { - if (sample_type == VL_FULL_SAMPLE) - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 128); - else - vlTransformBlock(blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT, temp_block, 0); + short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; + +#ifdef DO_IDCT + vlTransformBlock(cur_block, cur_block, sample_type == VL_FULL_SAMPLE ? 128 : 0); +#endif vlGrabFrameCodedBlock ( - temp_block, + cur_block, texels, tex_pitch ); -- cgit v1.2.3 From fda01b584715c05696a0e6768fda669ef1eb5f3b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 7 Aug 2008 11:26:17 +1000 Subject: nouveau: fix build --- src/gallium/drivers/nv04/nv04_fragprog.c | 4 ++-- src/gallium/drivers/nv04/nv04_state.c | 2 +- src/gallium/drivers/nv04/nv04_state.h | 2 +- src/gallium/drivers/nv10/nv10_fragprog.c | 4 ++-- src/gallium/drivers/nv10/nv10_state.c | 2 +- src/gallium/drivers/nv10/nv10_state.h | 2 +- src/gallium/drivers/nv30/nv30_fragprog.c | 4 ++-- src/gallium/drivers/nv30/nv30_state.c | 2 +- src/gallium/drivers/nv30/nv30_state.h | 2 +- src/gallium/drivers/nv30/nv30_vertprog.c | 2 +- src/gallium/drivers/nv40/nv40_fragprog.c | 4 ++-- src/gallium/drivers/nv40/nv40_state.c | 2 +- src/gallium/drivers/nv40/nv40_state.h | 2 +- src/gallium/drivers/nv40/nv40_vertprog.c | 4 ++-- src/gallium/drivers/nv50/nv50_program.c | 4 ++-- src/gallium/drivers/nv50/nv50_program.h | 2 +- src/gallium/drivers/nv50/nv50_state.c | 2 +- 17 files changed, 23 insertions(+), 23 deletions(-) diff --git a/src/gallium/drivers/nv04/nv04_fragprog.c b/src/gallium/drivers/nv04/nv04_fragprog.c index 11f4360ed80..215974eec00 100644 --- a/src/gallium/drivers/nv04/nv04_fragprog.c +++ b/src/gallium/drivers/nv04/nv04_fragprog.c @@ -4,8 +4,8 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" -#include "tgsi/util/tgsi_util.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" #include "nv04_context.h" diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c index 7e71dee7b5d..668d875671f 100644 --- a/src/gallium/drivers/nv04/nv04_state.c +++ b/src/gallium/drivers/nv04/nv04_state.c @@ -4,7 +4,7 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" +#include "tgsi/tgsi_parse.h" #include "nv04_context.h" #include "nv04_state.h" diff --git a/src/gallium/drivers/nv04/nv04_state.h b/src/gallium/drivers/nv04/nv04_state.h index 39f7cd17b30..399f750dbe7 100644 --- a/src/gallium/drivers/nv04/nv04_state.h +++ b/src/gallium/drivers/nv04/nv04_state.h @@ -2,7 +2,7 @@ #define __NV04_STATE_H__ #include "pipe/p_state.h" -#include "tgsi/util/tgsi_scan.h" +#include "tgsi/tgsi_scan.h" struct nv04_blend_state { uint32_t b_enable; diff --git a/src/gallium/drivers/nv10/nv10_fragprog.c b/src/gallium/drivers/nv10/nv10_fragprog.c index 2a63c8a704e..137de9d53ef 100644 --- a/src/gallium/drivers/nv10/nv10_fragprog.c +++ b/src/gallium/drivers/nv10/nv10_fragprog.c @@ -4,8 +4,8 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" -#include "tgsi/util/tgsi_util.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" #include "nv10_context.h" diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index 43b9c32f12d..f902fd54b66 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -4,7 +4,7 @@ #include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" +#include "tgsi/tgsi_parse.h" #include "nv10_context.h" #include "nv10_state.h" diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h index f1f9a12110f..3a3fd0d4f4f 100644 --- a/src/gallium/drivers/nv10/nv10_state.h +++ b/src/gallium/drivers/nv10/nv10_state.h @@ -2,7 +2,7 @@ #define __NV10_STATE_H__ #include "pipe/p_state.h" -#include "tgsi/util/tgsi_scan.h" +#include "tgsi/tgsi_scan.h" struct nv10_blend_state { uint32_t b_enable; diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 59e79d66f2a..68058264e39 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -3,8 +3,8 @@ #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" -#include "tgsi/util/tgsi_util.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" #include "nv30_context.h" diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index ba02413de5c..8d88d6c806e 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -3,7 +3,7 @@ #include "pipe/p_util.h" #include "pipe/p_inlines.h" -#include "tgsi/util/tgsi_parse.h" +#include "tgsi/tgsi_parse.h" #include "nv30_context.h" #include "nv30_state.h" diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index 20c51761607..e6f23bf1667 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -2,7 +2,7 @@ #define __NV30_STATE_H__ #include "pipe/p_state.h" -#include "tgsi/util/tgsi_scan.h" +#include "tgsi/tgsi_scan.h" struct nv30_sampler_state { uint32_t fmt; diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 39852ce9834..6c075cdb752 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -3,7 +3,7 @@ #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" +#include "tgsi/tgsi_parse.h" #include "nv30_context.h" #include "nv30_state.h" diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index 428348c3388..a361509e8f1 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -3,8 +3,8 @@ #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" -#include "tgsi/util/tgsi_util.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" #include "nv40_context.h" diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 5d2c3ab8810..63d0ecc9158 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -5,7 +5,7 @@ #include "draw/draw_context.h" -#include "tgsi/util/tgsi_parse.h" +#include "tgsi/tgsi_parse.h" #include "nv40_context.h" #include "nv40_state.h" diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index 2b4225deb2e..8a9d8c8fdf6 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -2,7 +2,7 @@ #define __NV40_STATE_H__ #include "pipe/p_state.h" -#include "tgsi/util/tgsi_scan.h" +#include "tgsi/tgsi_scan.h" struct nv40_sampler_state { uint32_t fmt; diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index 1e486a66ef0..ff988e6a5f4 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -3,8 +3,8 @@ #include "pipe/p_state.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" -#include "tgsi/util/tgsi_util.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" #include "nv40_context.h" #include "nv40_state.h" diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index 0d3ddb8a596..d6fbdd18243 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -26,8 +26,8 @@ #include "pipe/p_inlines.h" #include "pipe/p_shader_tokens.h" -#include "tgsi/util/tgsi_parse.h" -#include "tgsi/util/tgsi_util.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" #include "nv50_context.h" diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h index d643e8db218..78deed6a384 100644 --- a/src/gallium/drivers/nv50/nv50_program.h +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -2,7 +2,7 @@ #define __NV50_PROGRAM_H__ #include "pipe/p_state.h" -#include "tgsi/util/tgsi_scan.h" +#include "tgsi/tgsi_scan.h" struct nv50_program_exec { struct nv50_program_exec *next; diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 731409bed4e..4055527d9f8 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -25,7 +25,7 @@ #include "pipe/p_util.h" #include "pipe/p_inlines.h" -#include "tgsi/util/tgsi_parse.h" +#include "tgsi/tgsi_parse.h" #include "nv50_context.h" #include "nv50_texture.h" -- cgit v1.2.3 From ce8e846ffea8e1a11b8ae4ba05a7386e7c34cc9f Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 7 Aug 2008 11:38:56 +1000 Subject: nv40/nv50: enable mirror wrap modes --- src/gallium/drivers/nv40/nv40_screen.c | 3 +++ src/gallium/drivers/nv50/nv50_screen.c | 3 +++ 2 files changed, 6 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index a63a6136c52..0e1df89ee85 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -57,6 +57,9 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param) return 10; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; + case PIPE_CAP_TEXTURE_MIRROR_CLAMP: + case PIPE_CAP_TEXTURE_MIRROR_REPEAT: + return 1; case NOUVEAU_CAP_HW_VTXBUF: return 1; case NOUVEAU_CAP_HW_IDXBUF: diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 996b0b3e8f2..ec439239298 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -114,6 +114,9 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param) return 10; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; + case PIPE_CAP_TEXTURE_MIRROR_CLAMP: + case PIPE_CAP_TEXTURE_MIRROR_REPEAT: + return 1; case NOUVEAU_CAP_HW_VTXBUF: return 1; case NOUVEAU_CAP_HW_IDXBUF: -- cgit v1.2.3 From df4228deddea36b9d5b41ea395a216137e046205 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 11 Aug 2008 16:14:42 +1000 Subject: nouveau: pf_sprint_name -> pf_name --- src/gallium/drivers/nv04/nv04_fragtex.c | 4 +--- src/gallium/drivers/nv30/nv30_fragtex.c | 3 +-- src/gallium/drivers/nv30/nv30_vbo.c | 8 ++------ src/gallium/drivers/nv40/nv40_fragtex.c | 4 +--- src/gallium/drivers/nv40/nv40_vbo.c | 8 ++------ src/gallium/drivers/nv50/nv50_state_validate.c | 14 ++++---------- src/gallium/drivers/nv50/nv50_vbo.c | 5 ++--- 7 files changed, 13 insertions(+), 33 deletions(-) diff --git a/src/gallium/drivers/nv04/nv04_fragtex.c b/src/gallium/drivers/nv04/nv04_fragtex.c index 3db673cd2c1..1b866aae199 100644 --- a/src/gallium/drivers/nv04/nv04_fragtex.c +++ b/src/gallium/drivers/nv04/nv04_fragtex.c @@ -51,7 +51,6 @@ static uint32_t nv04_fragtex_format(uint pipe_format) { struct nv04_texture_format *tf = nv04_texture_formats; - char fs[128]; int i; for (i=0; i< sizeof(nv04_texture_formats)/sizeof(nv04_texture_formats[0]); i++) { @@ -60,8 +59,7 @@ nv04_fragtex_format(uint pipe_format) tf++; } - pf_sprint_name(fs, pipe_format); - NOUVEAU_ERR("unknown texture format %s\n", fs); + NOUVEAU_ERR("unknown texture format %s\n", pf_name(pipe_format)); return 0; } diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index 4242f86f76f..91246f5ca0a 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -75,8 +75,7 @@ nv30_fragtex_format(uint pipe_format) tf++; } - pf_sprint_name(fs, pipe_format); - NOUVEAU_ERR("unknown texture format %s\n", fs); + NOUVEAU_ERR("unknown texture format %s\n", pf_name(pipe_format)); return NULL; } diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index d930557f6bb..b1c73793bf2 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -14,8 +14,6 @@ static INLINE int nv30_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) { - char fs[128]; - switch (pipe) { case PIPE_FORMAT_R32_FLOAT: case PIPE_FORMAT_R32G32_FLOAT: @@ -36,8 +34,7 @@ nv30_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) *fmt = NV34TCL_VTXFMT_TYPE_USHORT; break; default: - pf_sprint_name(fs, pipe); - NOUVEAU_ERR("Unknown format %s\n", fs); + NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe)); return 1; } @@ -63,8 +60,7 @@ nv30_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) *ncomp = 4; break; default: - pf_sprint_name(fs, pipe); - NOUVEAU_ERR("Unknown format %s\n", fs); + NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe)); return 1; } diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c index 2d45c2545cd..566d5a8d5ba 100644 --- a/src/gallium/drivers/nv40/nv40_fragtex.c +++ b/src/gallium/drivers/nv40/nv40_fragtex.c @@ -45,7 +45,6 @@ static struct nv40_texture_format * nv40_fragtex_format(uint pipe_format) { struct nv40_texture_format *tf = nv40_texture_formats; - char fs[128]; while (tf->defined) { if (tf->pipe == pipe_format) @@ -53,8 +52,7 @@ nv40_fragtex_format(uint pipe_format) tf++; } - pf_sprint_name(fs, pipe_format); - NOUVEAU_ERR("unknown texture format %s\n", fs); + NOUVEAU_ERR("unknown texture format %s\n", pf_name(pipe_format)); return NULL; } diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 93669e6192f..755d5586b7b 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -14,8 +14,6 @@ static INLINE int nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) { - char fs[128]; - switch (pipe) { case PIPE_FORMAT_R32_FLOAT: case PIPE_FORMAT_R32G32_FLOAT: @@ -36,8 +34,7 @@ nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) *fmt = NV40TCL_VTXFMT_TYPE_USHORT; break; default: - pf_sprint_name(fs, pipe); - NOUVEAU_ERR("Unknown format %s\n", fs); + NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe)); return 1; } @@ -63,8 +60,7 @@ nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp) *ncomp = 4; break; default: - pf_sprint_name(fs, pipe); - NOUVEAU_ERR("Unknown format %s\n", fs); + NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe)); return 1; } diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 8f9ee05acc8..198e25f4482 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -58,11 +58,8 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_data(so, 0xe8); break; default: - { - char fmt[128]; - pf_sprint_name(fmt, fb->cbufs[i]->format); - NOUVEAU_ERR("AIIII unknown format %s\n", fmt); - } + NOUVEAU_ERR("AIIII unknown format %s\n", + pf_name(fb->cbufs[i]->format)); so_data(so, 0xe6); break; } @@ -96,11 +93,8 @@ nv50_state_validate_fb(struct nv50_context *nv50) so_data(so, 0x15); break; default: - { - char fmt[128]; - pf_sprint_name(fmt, fb->zsbuf->format); - NOUVEAU_ERR("AIIII unknown format %s\n", fmt); - } + NOUVEAU_ERR("AIIII unknown format %s\n", + pf_name(fb->zsbuf->format)); so_data(so, 0x16); break; } diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index 8e420ad172e..c94531723bc 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -215,9 +215,8 @@ nv50_vbo_validate(struct nv50_context *nv50) break; default: { - char fmt[128]; - pf_sprint_name(fmt, ve->src_format); - NOUVEAU_ERR("invalid vbo format %s\n", fmt); + NOUVEAU_ERR("invalid vbo format %s\n", + pf_name(ve->src_format)); assert(0); return; } -- cgit v1.2.3 From dbec107c2549c00120ca6d67282f2a1a4bd8bdbd Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 14 Aug 2008 00:22:49 +1000 Subject: nv30/nv40: KIL/KILP swapped meanings --- src/gallium/drivers/nv30/nv30_fragprog.c | 4 ++-- src/gallium/drivers/nv40/nv40_fragprog.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 68058264e39..d3693fdf567 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -494,10 +494,10 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, case TGSI_OPCODE_FRC: arith(fpc, sat, FRC, dst, mask, src[0], none, none); break; - case TGSI_OPCODE_KIL: + case TGSI_OPCODE_KILP: arith(fpc, 0, KIL, none, 0, none, none, none); break; - case TGSI_OPCODE_KILP: + case TGSI_OPCODE_KIL: dst = nv30_sr(NV30SR_NONE, 0); dst.cc_update = 1; arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c index a361509e8f1..91dcbebda0d 100644 --- a/src/gallium/drivers/nv40/nv40_fragprog.c +++ b/src/gallium/drivers/nv40/nv40_fragprog.c @@ -533,10 +533,10 @@ nv40_fragprog_parse_instruction(struct nv40_fpc *fpc, case TGSI_OPCODE_FRC: arith(fpc, sat, FRC, dst, mask, src[0], none, none); break; - case TGSI_OPCODE_KIL: + case TGSI_OPCODE_KILP: arith(fpc, 0, KIL, none, 0, none, none, none); break; - case TGSI_OPCODE_KILP: + case TGSI_OPCODE_KIL: dst = nv40_sr(NV40SR_NONE, 0); dst.cc_update = 1; arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none); -- cgit v1.2.3 From a7ea6bae4e618e28636909ac2db7783632bc81b8 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 14 Aug 2008 16:36:52 +0200 Subject: nv30: does not support mirror clamp, only mirror repeat --- src/gallium/drivers/nv30/nv30_screen.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 0ffcd772351..d5514c2aba5 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -55,6 +55,10 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param) return 10; case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; + case PIPE_CAP_TEXTURE_MIRROR_CLAMP: + return 0; + case PIPE_CAP_TEXTURE_MIRROR_REPEAT: + return 1; case NOUVEAU_CAP_HW_VTXBUF: case NOUVEAU_CAP_HW_IDXBUF: return 1; -- cgit v1.2.3 From a145c107c12715105e14bb56b245eeb660cf433a Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 14 Aug 2008 16:52:51 +0200 Subject: nv30: disable setting nv40 RECT bit, this is not the same on nv30, plus gallium does not support rectangle textures currently, only full POT or NPOT --- src/gallium/drivers/nv30/nv30_state.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 8d88d6c806e..77b0c9e08b0 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -127,8 +127,10 @@ nv30_sampler_state_create(struct pipe_context *pipe, in sampler state structure, and set appropriate format in nvxx_fragtex_build() */ - if (!cso->normalized_coords) - ps->fmt |= (1<<14) /*NV34TCL_TX_FORMAT_RECT*/; + /*NV34TCL_TX_FORMAT_RECT*/ + /*if (!cso->normalized_coords) { + ps->fmt |= (1<<14) ; + }*/ ps->wrap = ((wrap_mode(cso->wrap_s) << NV34TCL_TX_WRAP_S_SHIFT) | (wrap_mode(cso->wrap_t) << NV34TCL_TX_WRAP_T_SHIFT) | -- cgit v1.2.3 From d4c199d05691878bbc4c72a06d3042ef00ff38e0 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Thu, 14 Aug 2008 17:24:06 +0200 Subject: nv30: set mipmap min/max lod accordingly --- src/gallium/drivers/nv30/nv30_state.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index 77b0c9e08b0..eceb5353152 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -199,10 +199,10 @@ nv30_sampler_state_create(struct pipe_context *pipe, ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff; limit = CLAMP(cso->max_lod, 0.0, 15.0); - ps->en |= (int)(limit * 256.0) << 7; + ps->en |= (int)(limit) << 14 /*NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT*/; limit = CLAMP(cso->min_lod, 0.0, 15.0); - ps->en |= (int)(limit * 256.0) << 19; + ps->en |= (int)(limit) << 26 /*NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT*/; } if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { -- cgit v1.2.3 From fb2732ff056a83418a379b81c1e5da50675c41ac Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 9 Aug 2008 20:52:36 -0400 Subject: g3dvl: Merged tgsi/util and tgsi/exec moved some headers around. --- src/gallium/state_trackers/g3dvl/vl_context.c | 4 ++-- src/gallium/state_trackers/g3dvl/vl_shader_build.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index d446d218a44..5616de0ba41 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -7,8 +7,8 @@ #include #include #include -#include -#include +#include +#include #include "vl_shader_build.h" #include "vl_data.h" #include "vl_defs.h" diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c index 1dc5be6fdbb..5f30e23ff8e 100644 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.c +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.c @@ -1,7 +1,7 @@ #include "vl_shader_build.h" #include -#include -#include +#include +#include struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last) { -- cgit v1.2.3 From 6858dd50c9b696c1c6044f5a403000f9d20b286b Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 16 Aug 2008 13:04:23 -0400 Subject: g3dvl: Modularized rendering, refactored to accommodate VAAPI, other APIs. --- src/gallium/state_trackers/g3dvl/Makefile | 4 +- src/gallium/state_trackers/g3dvl/vl_basic_csc.c | 694 ++++++ src/gallium/state_trackers/g3dvl/vl_basic_csc.h | 13 + src/gallium/state_trackers/g3dvl/vl_context.c | 2272 +------------------ src/gallium/state_trackers/g3dvl/vl_context.h | 118 +- src/gallium/state_trackers/g3dvl/vl_csc.h | 53 + src/gallium/state_trackers/g3dvl/vl_data.c | 130 +- src/gallium/state_trackers/g3dvl/vl_data.h | 19 +- src/gallium/state_trackers/g3dvl/vl_defs.h | 1 - src/gallium/state_trackers/g3dvl/vl_display.c | 48 + src/gallium/state_trackers/g3dvl/vl_display.h | 29 + src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c | 2315 ++++++++++++++++++++ src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h | 18 + src/gallium/state_trackers/g3dvl/vl_render.h | 33 + src/gallium/state_trackers/g3dvl/vl_screen.c | 115 + src/gallium/state_trackers/g3dvl/vl_screen.h | 63 + src/gallium/state_trackers/g3dvl/vl_shader_build.c | 37 +- src/gallium/state_trackers/g3dvl/vl_shader_build.h | 1 - src/gallium/state_trackers/g3dvl/vl_surface.c | 671 +----- src/gallium/state_trackers/g3dvl/vl_surface.h | 91 +- src/gallium/state_trackers/g3dvl/vl_types.h | 124 +- src/gallium/state_trackers/g3dvl/vl_util.c | 9 +- src/gallium/state_trackers/g3dvl/vl_util.h | 1 - src/libXvMC/block.c | 47 +- src/libXvMC/context.c | 194 +- src/libXvMC/surface.c | 403 ++-- 26 files changed, 4061 insertions(+), 3442 deletions(-) create mode 100644 src/gallium/state_trackers/g3dvl/vl_basic_csc.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_basic_csc.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_csc.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_display.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_display.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_render.h create mode 100644 src/gallium/state_trackers/g3dvl/vl_screen.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_screen.h diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile index c6a22cad4e2..9995c554ab8 100644 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -1,5 +1,6 @@ TARGET = libg3dvl.a -OBJECTS = vl_context.o vl_data.o vl_surface.o vl_shader_build.o vl_util.o +OBJECTS = vl_display.o vl_screen.o vl_context.o vl_surface.o vl_data.o vl_shader_build.o vl_util.o vl_basic_csc.o \ + vl_r16snorm_mc.o GALLIUMDIR = ../.. CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary -I${GALLIUMDIR}/winsys/g3dvl @@ -15,4 +16,3 @@ ${TARGET}: ${OBJECTS} clean: rm -rf ${OBJECTS} ${TARGET} - diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c new file mode 100644 index 00000000000..ea003a31d17 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c @@ -0,0 +1,694 @@ +#define VL_INTERNAL +#include "vl_basic_csc.h" +#include +#include +#include +#include +#include +#include +#include +#include "vl_csc.h" +#include "vl_surface.h" +#include "vl_shader_build.h" +#include "vl_types.h" + +struct vlVertexShaderConsts +{ + struct vlVertex4f src_scale; + struct vlVertex4f src_trans; +}; + +struct vlFragmentShaderConsts +{ + struct vlVertex4f bias; + float matrix[16]; +}; + +struct vlBasicCSC +{ + struct vlCSC base; + + struct pipe_context *pipe; + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state framebuffer; + void *sampler; + void *vertex_shader, *fragment_shader; + struct pipe_vertex_buffer vertex_bufs[2]; + struct pipe_vertex_element vertex_elems[2]; + struct pipe_constant_buffer vs_const_buf, fs_const_buf; +}; + +static int vlResizeFrameBuffer +( + struct vlCSC *csc, + unsigned int width, + unsigned int height +) +{ + struct vlBasicCSC *basic_csc; + struct pipe_context *pipe; + + assert(csc); + + basic_csc = (struct vlBasicCSC*)csc; + pipe = basic_csc->pipe; + + if (basic_csc->framebuffer.width == width && basic_csc->framebuffer.height == height) + return 0; + + if (basic_csc->framebuffer.cbufs[0]) + pipe->winsys->surface_release + ( + pipe->winsys, + &basic_csc->framebuffer.cbufs[0] + ); + + basic_csc->viewport.scale[0] = width; + basic_csc->viewport.scale[1] = height; + basic_csc->viewport.scale[2] = 1; + basic_csc->viewport.scale[3] = 1; + basic_csc->viewport.translate[0] = 0; + basic_csc->viewport.translate[1] = 0; + basic_csc->viewport.translate[2] = 0; + basic_csc->viewport.translate[3] = 0; + + basic_csc->framebuffer.width = width; + basic_csc->framebuffer.height = height; + basic_csc->framebuffer.cbufs[0] = pipe->winsys->surface_alloc(pipe->winsys); + pipe->winsys->surface_alloc_storage + ( + pipe->winsys, + basic_csc->framebuffer.cbufs[0], + width, + height, + PIPE_FORMAT_A8R8G8B8_UNORM, + /* XXX: SoftPipe doesn't change GPU usage to CPU like it does for textures */ + PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE, + 0 + ); + + return 0; +} + +static int vlBegin +( + struct vlCSC *csc +) +{ + struct vlBasicCSC *basic_csc; + struct pipe_context *pipe; + + assert(csc); + + basic_csc = (struct vlBasicCSC*)csc; + pipe = basic_csc->pipe; + + pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer); + pipe->set_viewport_state(pipe, &basic_csc->viewport); + pipe->bind_sampler_states(pipe, 1, (void**)&basic_csc->sampler); + /* Source texture set in vlPutSurface() */ + pipe->bind_vs_state(pipe, basic_csc->vertex_shader); + pipe->bind_fs_state(pipe, basic_csc->fragment_shader); + pipe->set_vertex_buffers(pipe, 2, basic_csc->vertex_bufs); + pipe->set_vertex_elements(pipe, 2, basic_csc->vertex_elems); + pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &basic_csc->vs_const_buf); + pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &basic_csc->fs_const_buf); + + return 0; +} + +static int vlPutPictureCSC +( + struct vlCSC *csc, + struct vlSurface *surface, + int srcx, + int srcy, + int srcw, + int srch, + int destx, + int desty, + int destw, + int desth, + enum vlPictureType picture_type +) +{ + struct vlBasicCSC *basic_csc; + struct pipe_context *pipe; + struct vlVertexShaderConsts *vs_consts; + + assert(csc); + assert(surface); + + basic_csc = (struct vlBasicCSC*)csc; + pipe = basic_csc->pipe; + + vs_consts = pipe->winsys->buffer_map + ( + pipe->winsys, + basic_csc->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vs_consts->src_scale.x = srcw / (float)surface->texture->width[0]; + vs_consts->src_scale.y = srch / (float)surface->texture->height[0]; + vs_consts->src_scale.z = 1; + vs_consts->src_scale.w = 1; + vs_consts->src_trans.x = srcx / (float)surface->texture->width[0]; + vs_consts->src_trans.y = srcy / (float)surface->texture->height[0]; + vs_consts->src_trans.z = 0; + vs_consts->src_trans.w = 0; + + pipe->winsys->buffer_unmap(pipe->winsys, basic_csc->vs_const_buf.buffer); + + pipe->set_sampler_textures(pipe, 1, &surface->texture); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); + + return 0; +} + +static int vlEnd +( + struct vlCSC *csc +) +{ + assert(csc); + + return 0; +} + +static struct pipe_surface* vlGetFrameBuffer +( + struct vlCSC *csc +) +{ + struct vlBasicCSC *basic_csc; + + assert(csc); + + basic_csc = (struct vlBasicCSC*)csc; + + return basic_csc->framebuffer.cbufs[0]; +} + +static int vlDestroy +( + struct vlCSC *csc +) +{ + struct vlBasicCSC *basic_csc; + struct pipe_context *pipe; + unsigned int i; + + assert(csc); + + basic_csc = (struct vlBasicCSC*)csc; + pipe = basic_csc->pipe; + + if (basic_csc->framebuffer.cbufs[0]) + pipe->winsys->surface_release + ( + pipe->winsys, + &basic_csc->framebuffer.cbufs[0] + ); + + pipe->delete_sampler_state(pipe, basic_csc->sampler); + pipe->delete_vs_state(pipe, basic_csc->vertex_shader); + pipe->delete_fs_state(pipe, basic_csc->fragment_shader); + + for (i = 0; i < 2; ++i) + pipe->winsys->buffer_destroy(pipe->winsys, basic_csc->vertex_bufs[i].buffer); + + pipe->winsys->buffer_destroy(pipe->winsys, basic_csc->vs_const_buf.buffer); + pipe->winsys->buffer_destroy(pipe->winsys, basic_csc->fs_const_buf.buffer); + + free(basic_csc); + + return 0; +} + +/* + * Represents 2 triangles in a strip in normalized coords. + * Used to render the surface onto the frame buffer. + */ +static const struct vlVertex2f surface_verts[4] = +{ + {0.0f, 0.0f}, + {0.0f, 1.0f}, + {1.0f, 0.0f}, + {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above. We can use the position values directly. + * TODO: Duplicate these in the shader, no need to create a buffer. + */ +static const struct vlVertex2f *surface_texcoords = surface_verts; + +/* + * Identity color conversion constants, for debugging + */ +static const struct vlFragmentShaderConsts identity = +{ + { + 0.0f, 0.0f, 0.0f, 0.0f + }, + { + 1.0f, 0.0f, 0.0f, 0.0f, + 0.0f, 1.0f, 0.0f, 0.0f, + 0.0f, 0.0f, 1.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [16,235] + */ +static const struct vlFragmentShaderConsts bt_601 = +{ + { + 0.0f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.0f, 0.0f, 1.371f, 0.0f, + 1.0f, -0.336f, -0.698f, 0.0f, + 1.0f, 1.732f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [0,255] + */ +static const struct vlFragmentShaderConsts bt_601_full = +{ + { + 0.062745098f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.164f, 0.0f, 1.596f, 0.0f, + 1.164f, -0.391f, -0.813f, 0.0f, + 1.164f, 2.018f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [16,235] + */ +static const struct vlFragmentShaderConsts bt_709 = +{ + { + 0.0f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.0f, 0.0f, 1.540f, 0.0f, + 1.0f, -0.183f, -0.459f, 0.0f, + 1.0f, 1.816f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +/* + * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [0,255] + */ +const struct vlFragmentShaderConsts bt_709_full = +{ + { + 0.062745098f, 0.501960784f, 0.501960784f, 0.0f + }, + { + 1.164f, 0.0f, 1.793f, 0.0f, + 1.164f, -0.213f, -0.534f, 0.0f, + 1.164f, 2.115f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f + } +}; + +static int vlCreateVertexShader +( + struct vlBasicCSC *csc +) +{ + const unsigned int max_tokens = 50; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = csc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling vector to scale texcoord rect to source size + * decl c1 ; Translation vector to move texcoord rect into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* mov o0, i0 ; Move pos in to pos out */ + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i1, c0 ; Scale unit texcoord rect to source size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 1, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o1, t0, c1 ; Translate texcoord rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + csc->vertex_shader = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShader +( + struct vlBasicCSC *csc +) +{ + const unsigned int max_tokens = 50; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(context); + + pipe = csc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* decl i0 ; Texcoords for s0 */ + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl c0 ; Bias vector for CSC + * decl c1-c4 ; CSC matrix c1-c4 + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl s0 ; Sampler for tex containing picture to display */ + decl = vl_decl_samplers(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* tex2d t0, i0, s0 ; Read src pixel */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* sub t0, t0, c0 ; Subtract bias vector from pixel */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix + * dp4 o0.y, t0, c2 + * dp4 o0.z, t0, c3 + * dp4 o0.w, t0, c4 ; XXX: Don't need 4th coefficient + */ + for (i = 0; i < 4; ++i) + { + inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + csc->fragment_shader = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateDataBufs +( + struct vlBasicCSC *csc +) +{ + struct pipe_context *pipe; + + assert(csc); + + pipe = csc->pipe; + + /* + Create our vertex buffer and vertex buffer element + VB contains 4 vertices that render a quad covering the entire window + to display a rendered surface + Quad is rendered as a tri strip + */ + csc->vertex_bufs[0].pitch = sizeof(struct vlVertex2f); + csc->vertex_bufs[0].max_index = 3; + csc->vertex_bufs[0].buffer_offset = 0; + csc->vertex_bufs[0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 4 + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, csc->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + surface_verts, + sizeof(struct vlVertex2f) * 4 + ); + + pipe->winsys->buffer_unmap(pipe->winsys, csc->vertex_bufs[0].buffer); + + csc->vertex_elems[0].src_offset = 0; + csc->vertex_elems[0].vertex_buffer_index = 0; + csc->vertex_elems[0].nr_components = 2; + csc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* + Create our texcoord buffer and texcoord buffer element + Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices + */ + csc->vertex_bufs[1].pitch = sizeof(struct vlVertex2f); + csc->vertex_bufs[1].max_index = 3; + csc->vertex_bufs[1].buffer_offset = 0; + csc->vertex_bufs[1].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 4 + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, csc->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + surface_texcoords, + sizeof(struct vlVertex2f) * 4 + ); + + pipe->winsys->buffer_unmap(pipe->winsys, csc->vertex_bufs[1].buffer); + + csc->vertex_elems[1].src_offset = 0; + csc->vertex_elems[1].vertex_buffer_index = 1; + csc->vertex_elems[1].nr_components = 2; + csc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* + Create our vertex shader's constant buffer + Const buffer contains scaling and translation vectors + */ + csc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts); + csc->vs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + csc->vs_const_buf.size + ); + + /* + Create our fragment shader's constant buffer + Const buffer contains the color conversion matrix and bias vectors + */ + csc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts); + csc->fs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + csc->fs_const_buf.size + ); + + /* + TODO: Refactor this into a seperate function, + allow changing the CSC matrix at runtime to switch between regular & full versions + */ + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, csc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + &bt_601, + sizeof(struct vlFragmentShaderConsts) + ); + + pipe->winsys->buffer_unmap(pipe->winsys, csc->fs_const_buf.buffer); + + return 0; +} + +static int vlInit +( + struct vlBasicCSC *csc +) +{ + struct pipe_context *pipe; + struct pipe_sampler_state sampler; + + assert(csc); + + pipe = csc->pipe; + + /* Delay creating the FB until vlPutSurface() so we know window size */ + csc->framebuffer.num_cbufs = 1; + csc->framebuffer.cbufs[0] = NULL; + csc->framebuffer.zsbuf = NULL; + + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.shadow_ambient = ;*/ + /*sampler.lod_bias = ;*/ + /*sampler.min_lod = ;*/ + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + csc->sampler = pipe->create_sampler_state(pipe, &sampler); + + vlCreateVertexShader(csc); + vlCreateFragmentShader(csc); + vlCreateDataBufs(csc); + + return 0; +} + +int vlCreateBasicCSC +( + struct pipe_context *pipe, + struct vlCSC **csc +) +{ + struct vlBasicCSC *basic_csc; + + assert(pipe); + assert(csc); + + basic_csc = calloc(1, sizeof(struct vlBasicCSC)); + + if (!basic_csc) + return 1; + + basic_csc->base.vlResizeFrameBuffer = &vlResizeFrameBuffer; + basic_csc->base.vlBegin = &vlBegin; + basic_csc->base.vlPutPicture = &vlPutPictureCSC; + basic_csc->base.vlEnd = &vlEnd; + basic_csc->base.vlGetFrameBuffer = &vlGetFrameBuffer; + basic_csc->base.vlDestroy = &vlDestroy; + basic_csc->pipe = pipe; + + vlInit(basic_csc); + + *csc = &basic_csc->base; + + return 0; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.h b/src/gallium/state_trackers/g3dvl/vl_basic_csc.h new file mode 100644 index 00000000000..2e17f1d814a --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_basic_csc.h @@ -0,0 +1,13 @@ +#ifndef vl_basic_csc_h +#define vl_basic_csc_h + +struct pipe_context; +struct vlCSC; + +int vlCreateBasicCSC +( + struct pipe_context *pipe, + struct vlCSC **csc +); + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 5616de0ba41..56d360c05b9 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -1,2088 +1,26 @@ +#define VL_INTERNAL #include "vl_context.h" #include #include #include -#include -#include #include -#include -#include -#include -#include -#include "vl_shader_build.h" -#include "vl_data.h" -#include "vl_defs.h" -#include "vl_util.h" +#include "vl_render.h" +#include "vl_r16snorm_mc.h" +#include "vl_csc.h" +#include "vl_basic_csc.h" -static int vlCreateVertexShaderFrameIDCT(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Vertex texcoords - */ - for (i = 0; i < 2; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl o0 ; Vertex pos - * decl o1 ; Vertex texcoords - */ - for (i = 0; i < 2; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * mov o0, i0 ; Move pos in to pos out - * mov o1, i1 ; Move texcoord in to texcoord out */ - for (i = 0; i < 2; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - //context->states.idct.frame_vs = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFrameIDCT(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* decl i0 ; Texcoords for s0 */ - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl s0 ; Sampler for tex containing picture to display */ - decl = vl_decl_samplers(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* tex2d t0, i0, s0 ; Read src pixel */ - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c0 ; Subtract bias vector from pixel */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix - * dp4 o0.y, t0, c2 - * dp4 o0.z, t0, c3 - * dp4 o0.w, t0, c4 ; XXX: Don't need 4th coefficient - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - //context->states.idct.frame_fs = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlInitIDCT(struct VL_CONTEXT *context) -{ - struct pipe_context *pipe; - struct pipe_sampler_state sampler; - struct pipe_texture template; - unsigned int i; - - assert(context); - - pipe = context->pipe; - - context->states.idct.viewport.scale[0] = VL_BLOCK_WIDTH; - context->states.idct.viewport.scale[1] = VL_BLOCK_HEIGHT; - context->states.idct.viewport.scale[2] = 1; - context->states.idct.viewport.scale[3] = 1; - context->states.idct.viewport.translate[0] = 0; - context->states.idct.viewport.translate[1] = 0; - context->states.idct.viewport.translate[2] = 0; - context->states.idct.viewport.translate[3] = 0; - - context->states.idct.render_target.width = VL_BLOCK_WIDTH; - context->states.idct.render_target.height = VL_BLOCK_HEIGHT; - context->states.idct.render_target.num_cbufs = 1; - context->states.idct.render_target.zsbuf = NULL; - - sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.min_img_filter = PIPE_TEX_FILTER_NEAREST; - sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; - sampler.mag_img_filter = PIPE_TEX_FILTER_NEAREST; - sampler.compare_mode = PIPE_TEX_COMPARE_NONE; - sampler.compare_func = PIPE_FUNC_ALWAYS; - sampler.normalized_coords = 1; - /*sampler.prefilter = ;*/ - /*sampler.shadow_ambient = ;*/ - /*sampler.lod_bias = ;*/ - sampler.min_lod = 0; - /*sampler.max_lod = ;*/ - /*sampler.border_color[i] = ;*/ - /*sampler.max_anisotropy = ;*/ - context->states.idct.sampler = pipe->create_sampler_state(pipe, &sampler); - - memset(&template, 0, sizeof(struct pipe_texture)); - template.target = PIPE_TEXTURE_2D; - template.format = PIPE_FORMAT_A8L8_UNORM; - template.last_level = 0; - template.width[0] = 8; - template.height[0] = 8; - template.depth[0] = 1; - template.compressed = 0; - pf_get_block(template.format, &template.block); - - context->states.idct.texture = pipe->screen->texture_create(pipe->screen, &template); - - template.format = PIPE_FORMAT_A8R8G8B8_UNORM; - template.width[0] = 16; - template.height[0] = 1; - - context->states.idct.basis = pipe->screen->texture_create(pipe->screen, &template); - - for (i = 0; i < 2; ++i) - { - context->states.idct.vertex_bufs[i] = &context->states.csc.vertex_bufs[i]; - context->states.idct.vertex_buf_elems[i] = &context->states.csc.vertex_buf_elems[i]; - /* - context->states.idct.vertex_bufs[i].pitch = sizeof(struct VL_VERTEX2F); - context->states.idct.vertex_bufs[i].max_index = 3; - context->states.idct.vertex_bufs[i].buffer_offset = 0; - context->states.idct.vertex_bufs[i].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct VL_VERTEX2F) * 4 - ); - - context->states.idct.vertex_buf_elems[i].src_offset = 0; - context->states.idct.vertex_buf_elems[i].vertex_buffer_index = i; - context->states.idct.vertex_buf_elems[i].nr_components = 2; - context->states.idct.vertex_buf_elems[i].src_format = PIPE_FORMAT_R32G32_FLOAT; - */ - } - - vlCreateVertexShaderFrameIDCT(context); - vlCreateFragmentShaderFrameIDCT(context); - - return 0; -} - -static int vlDestroyIDCT(struct VL_CONTEXT *context) -{ - //unsigned int i; - - assert(context); - - context->pipe->delete_sampler_state(context->pipe, context->states.idct.sampler); - - //for (i = 0; i < 2; ++i) - //context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.idct.vertex_bufs[i].buffer); - - pipe_texture_release(&context->states.idct.texture); - pipe_texture_release(&context->states.idct.basis); - - //context->pipe->delete_vs_state(context->pipe, context->states.idct.frame_vs); - //context->pipe->delete_fs_state(context->pipe, context->states.idct.frame_fs); - - //context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.idct.vs_const_buf.buffer); - //context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.idct.fs_const_buf.buffer); - - return 0; -} - -static int vlCreateVertexShaderIMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma texcoords - */ - for (i = 0; i < 3; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling vector to scale unit rect to macroblock size - * decl c1 ; Translation vector to move macroblock into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma texcoords - */ - for (i = 0; i < 3; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0 */ - decl = vl_decl_temps(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, c1 ; Translate rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma texcoords to output - */ - for (i = 1; i < 3; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - context->states.mc.i_vs = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderIMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Texcoords for s0 - * decl i1 ; Texcoords for s1, s2 - */ - for (i = 0; i < 2; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - */ - for (i = 0; i < 3; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i1, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - } - - /* mul o0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - context->states.mc.i_fs = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFramePMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma texcoords - */ - for (i = 0; i < 3; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling vector to scale unit rect to macroblock size - * decl c1 ; Translation vector to move macroblock into position - * decl c2 ; Unused - * decl c3 ; Translation vector to move ref macroblock texcoords into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma texcoords - * decl o3 ; Ref macroblock texcoords - */ - for (i = 0; i < 4; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0 */ - decl = vl_decl_temps(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, c1 ; Translate rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma texcoords to output - */ - for (i = 1; i < 3; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* add o3, t0, c3 ; Translate rect into position on ref macroblock */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - context->states.mc.p_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFieldPMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma texcoords - */ - for (i = 0; i < 3; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration - ( - &decl, - &tokens[ti], - header, - max_tokens - ti - ); - } - - /* - * decl c0 ; Scaling vector to scale unit rect to macroblock size - * decl c1 ; Translation vector to move macroblock into position - * decl c2 ; Denorm coefficients - * decl c3 ; Translation vector to move top field ref macroblock texcoords into position - * decl c4 ; Translation vector to move bottom field ref macroblock texcoords into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma texcoords - * decl o3 ; Top field ref macroblock texcoords - * decl o4 ; Bottom field ref macroblock texcoords - * decl o5 ; Denormalized vertex pos - */ - for (i = 0; i < 6; i++) - { - decl = vl_decl_output((i == 0 || i == 5) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add t1, t0, c1 ; Translate rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mov o0, t1 ; Move vertex pos to output */ - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - mov o1, i1 ; Move input luma texcoords to output - mov o2, i2 ; Move input chroma texcoords to output - */ - for (i = 1; i < 3; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* add o3, t0, c3 ; Translate top field rect into position on ref macroblock - add o4, t0, c4 ; Translate bottom field rect into position on ref macroblock */ - for (i = 0; i < 2; ++i) - { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul o5, t1, c2 ; Denorm vertex pos */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 5, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - context->states.mc.p_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFramePMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Texcoords for s0 - * decl i1 ; Texcoords for s1, s2 - * decl i2 ; Texcoords for s3 - */ - for (i = 0; i < 3; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for ref surface texture - */ - for (i = 0; i < 4; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i1, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* tex2d t1, i2, s3 ; Read texel from ref macroblock */ - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 2, TGSI_FILE_SAMPLER, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - context->states.mc.p_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFieldPMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 200; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Texcoords for s0 - * decl i1 ; Texcoords for s1, s2 - * decl i2 ; Texcoords for s3 - * decl i3 ; Texcoords for s3 - * decl i4 ; Denormalized vertex pos - */ - for (i = 0; i < 5; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm - * decl c1 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0-t4 */ - decl = vl_decl_temps(0, 4); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for ref surface texture - */ - for (i = 0; i < 4; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i1, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t1, i2, s3 ; Read texel from ref macroblock top field - * tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* XXX: Pos values off by 0.5? */ - /* sub t4, i4.y, c1.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 4, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* floor t3, t3 ; Get rid of fractional part */ - inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t3, c1.y ; Multiply by 2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* TODO: Move to conditional tex fetch on t3 instead of lerp */ - /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - context->states.mc.p_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFrameBMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma texcoords - */ - for (i = 0; i < 3; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling vector to scale unit rect to macroblock size - * decl c1 ; Translation vector to move macroblock into position - * decl c2 ; Unused - * decl c3 ; Translation vector to move past ref macroblock texcoords into position - * decl c4 ; Unused - * decl c5 ; Translation vector to move future ref macroblock texcoords into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 5); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma texcoords - * decl o3 ; Past ref macroblock texcoords - * decl o4 ; Future ref macroblock texcoords - */ - for (i = 0; i < 5; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0 */ - decl = vl_decl_temps(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, c1 ; Translate rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma texcoords to output - */ - for (i = 1; i < 3; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* add o3, t0, c3 ; Translate rect into position on past ref macroblock - add o4, t0, c5 ; Translate rect into position on future ref macroblock */ - for (i = 0; i < 2; ++i) - { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i * 2 + 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - context->states.mc.b_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFieldBMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma texcoords - */ - for (i = 0; i < 3; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling vector to scale unit rect to macroblock size - * decl c1 ; Translation vector to move macroblock into position - * decl c2 ; Denorm coefficients - * decl c3 ; Translation vector to move top field past ref macroblock texcoords into position - * decl c4 ; Translation vector to move bottom field past ref macroblock texcoords into position - * decl c5 ; Translation vector to move top field future ref macroblock texcoords into position - * decl c6 ; Translation vector to move bottom field future ref macroblock texcoords into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 6); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma texcoords - * decl o3 ; Top field past ref macroblock texcoords - * decl o4 ; Bottom field past ref macroblock texcoords - * decl o5 ; Top field future ref macroblock texcoords - * decl o6 ; Bottom field future ref macroblock texcoords - * decl o7 ; Denormalized vertex pos - */ - for (i = 0; i < 8; i++) - { - decl = vl_decl_output((i == 0 || i == 7) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add t1, t0, c1 ; Translate rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mov o0, t1 ; Move vertex pos to output */ - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma texcoords to output - */ - for (i = 1; i < 3; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* - * add o3, t0, c3 ; Translate top field rect into position on past ref macroblock - * add o4, t0, c4 ; Translate bottom field rect into position on past ref macroblock - * add o5, t0, c5 ; Translate top field rect into position on future ref macroblock - * add o6, t0, c6 ; Translate bottom field rect into position on future ref macroblock - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul o7, t1, c2 ; Denorm vertex pos */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 7, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - context->states.mc.b_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFrameBMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Texcoords for s0 - * decl i1 ; Texcoords for s1, s2 - * decl i2 ; Texcoords for s3 - * decl i3 ; Texcoords for s4 - */ - for (i = 0; i < 4; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm - * decl c1 ; Constant 1/2 in .x channel to use as weight to blend past and future texels - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0-t2 */ - decl = vl_decl_temps(0, 2); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for past ref surface texture - * decl s4 ; Sampler for future ref surface texture - */ - for (i = 0; i < 5; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i1, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t1, i2, s3 ; Read texel from past ref macroblock - * tex2d t2, i3, s4 ; Read texel from future ref macroblock - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, i + 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - context->states.mc.b_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFieldBMC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 200; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Texcoords for s0 - * decl i1 ; Texcoords for s1, s2 - * decl i2 ; Texcoords for s3 - * decl i3 ; Texcoords for s3 - * decl i4 ; Texcoords for s4 - * decl i5 ; Texcoords for s4 - * decl i6 ; Denormalized vertex pos - */ - for (i = 0; i < 7; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm - * decl c1 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels - * ; and for Y-mod-2 top/bottom field selection - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0-t5 */ - decl = vl_decl_temps(0, 5); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for past ref surface texture - * decl s4 ; Sampler for future ref surface texture - */ - for (i = 0; i < 5; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i1, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* XXX: Pos values off by 0.5? */ - /* sub t4, i6.y, c1.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 6, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* floor t3, t3 ; Get rid of fractional part */ - inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t3, c1.y ; Multiply by 2 */ - inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t1, i2, s3 ; Read texel from past ref macroblock top field - * tex2d t2, i3, s3 ; Read texel from past ref macroblock bottom field - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* TODO: Move to conditional tex fetch on t3 instead of lerp */ - /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t4, i4, s4 ; Read texel from future ref macroblock top field - * tex2d t5, i5, s4 ; Read texel from future ref macroblock bottom field - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 4, TGSI_FILE_SAMPLER, 4); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* TODO: Move to conditional tex fetch on t3 instead of lerp */ - /* lerp t2, t3, t4, t5 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - context->states.mc.b_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -int vlCreateDataBufsMC(struct VL_CONTEXT *context) -{ - struct pipe_context *pipe; - unsigned int i; - - assert(context); - - pipe = context->pipe; - - /* Create our vertex buffer and vertex buffer element */ - context->states.mc.vertex_bufs[0].pitch = sizeof(struct VL_VERTEX2F); - context->states.mc.vertex_bufs[0].max_index = 23; - context->states.mc.vertex_bufs[0].buffer_offset = 0; - context->states.mc.vertex_bufs[0].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct VL_VERTEX2F) * 24 - ); - - context->states.mc.vertex_buf_elems[0].src_offset = 0; - context->states.mc.vertex_buf_elems[0].vertex_buffer_index = 0; - context->states.mc.vertex_buf_elems[0].nr_components = 2; - context->states.mc.vertex_buf_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Create our texcoord buffers and texcoord buffer elements */ - for (i = 1; i < 3; ++i) - { - context->states.mc.vertex_bufs[i].pitch = sizeof(struct VL_TEXCOORD2F); - context->states.mc.vertex_bufs[i].max_index = 23; - context->states.mc.vertex_bufs[i].buffer_offset = 0; - context->states.mc.vertex_bufs[i].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct VL_TEXCOORD2F) * 24 - ); - - context->states.mc.vertex_buf_elems[i].src_offset = 0; - context->states.mc.vertex_buf_elems[i].vertex_buffer_index = i; - context->states.mc.vertex_buf_elems[i].nr_components = 2; - context->states.mc.vertex_buf_elems[i].src_format = PIPE_FORMAT_R32G32_FLOAT; - } - - /* Fill buffers */ - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_chroma_420_texcoords, - sizeof(struct VL_VERTEX2F) * 24 - ); - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_luma_texcoords, - sizeof(struct VL_TEXCOORD2F) * 24 - ); - /* TODO: Accomodate 422, 444 */ - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.vertex_bufs[2].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_chroma_420_texcoords, - sizeof(struct VL_TEXCOORD2F) * 24 - ); - - for (i = 0; i < 3; ++i) - pipe->winsys->buffer_unmap(pipe->winsys, context->states.mc.vertex_bufs[i].buffer); - - /* Create our constant buffer */ - context->states.mc.vs_const_buf.size = sizeof(struct VL_MC_VS_CONSTS); - context->states.mc.vs_const_buf.buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_CONSTANT, - context->states.mc.vs_const_buf.size - ); - - context->states.mc.fs_const_buf.size = sizeof(struct VL_MC_FS_CONSTS); - context->states.mc.fs_const_buf.buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_CONSTANT, - context->states.mc.fs_const_buf.size - ); - - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.mc.fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - &vl_mc_fs_consts, - sizeof(struct VL_MC_FS_CONSTS) - ); - - pipe->winsys->buffer_unmap(pipe->winsys, context->states.mc.fs_const_buf.buffer); - - return 0; -} - -static int vlInitMC(struct VL_CONTEXT *context) -{ - struct pipe_context *pipe; - struct pipe_sampler_state sampler; - struct pipe_texture template; - unsigned int filters[5]; - unsigned int i; - - assert(context); - - pipe = context->pipe; - - /* For MC we render to textures, which are rounded up to nearest POT */ - context->states.mc.viewport.scale[0] = vlRoundUpPOT(context->video_width); - context->states.mc.viewport.scale[1] = vlRoundUpPOT(context->video_height); - context->states.mc.viewport.scale[2] = 1; - context->states.mc.viewport.scale[3] = 1; - context->states.mc.viewport.translate[0] = 0; - context->states.mc.viewport.translate[1] = 0; - context->states.mc.viewport.translate[2] = 0; - context->states.mc.viewport.translate[3] = 0; - - context->states.mc.render_target.width = vlRoundUpPOT(context->video_width); - context->states.mc.render_target.height = vlRoundUpPOT(context->video_height); - context->states.mc.render_target.num_cbufs = 1; - /* FB for MC stage is a VL_SURFACE, set in vlSetRenderSurface() */ - context->states.mc.render_target.zsbuf = NULL; - - filters[0] = PIPE_TEX_FILTER_NEAREST; - filters[1] = context->video_format == VL_FORMAT_YCBCR_444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; - filters[2] = context->video_format == VL_FORMAT_YCBCR_444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; - filters[3] = PIPE_TEX_FILTER_LINEAR; - filters[4] = PIPE_TEX_FILTER_LINEAR; - - for (i = 0; i < 5; ++i) - { - sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.min_img_filter = filters[i]; - sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; - sampler.mag_img_filter = filters[i]; - sampler.compare_mode = PIPE_TEX_COMPARE_NONE; - sampler.compare_func = PIPE_FUNC_ALWAYS; - sampler.normalized_coords = 1; - /*sampler.prefilter = ;*/ - /*sampler.shadow_ambient = ;*/ - /*sampler.lod_bias = ;*/ - sampler.min_lod = 0; - /*sampler.max_lod = ;*/ - /*sampler.border_color[i] = ;*/ - /*sampler.max_anisotropy = ;*/ - context->states.mc.samplers[i] = pipe->create_sampler_state(pipe, &sampler); - } - - memset(&template, 0, sizeof(struct pipe_texture)); - template.target = PIPE_TEXTURE_2D; - template.format = PIPE_FORMAT_R16_SNORM; - template.last_level = 0; - template.width[0] = 8; - template.height[0] = 8 * 4; - template.depth[0] = 1; - template.compressed = 0; - pf_get_block(template.format, &template.block); - - context->states.mc.textures[0] = pipe->screen->texture_create(pipe->screen, &template); - - if (context->video_format == VL_FORMAT_YCBCR_420) - template.height[0] = 8; - else if (context->video_format == VL_FORMAT_YCBCR_422) - template.height[0] = 8 * 2; - else if (context->video_format == VL_FORMAT_YCBCR_444) - template.height[0] = 8 * 4; - else - assert(0); - - context->states.mc.textures[1] = pipe->screen->texture_create(pipe->screen, &template); - context->states.mc.textures[2] = pipe->screen->texture_create(pipe->screen, &template); - - /* textures[3] & textures[4] are assigned from VL_SURFACEs for P and B macroblocks at render time */ - - vlCreateVertexShaderIMC(context); - vlCreateFragmentShaderIMC(context); - vlCreateVertexShaderFramePMC(context); - vlCreateVertexShaderFieldPMC(context); - vlCreateFragmentShaderFramePMC(context); - vlCreateFragmentShaderFieldPMC(context); - vlCreateVertexShaderFrameBMC(context); - vlCreateVertexShaderFieldBMC(context); - vlCreateFragmentShaderFrameBMC(context); - vlCreateFragmentShaderFieldBMC(context); - vlCreateDataBufsMC(context); - - return 0; -} - -static int vlDestroyMC(struct VL_CONTEXT *context) -{ - unsigned int i; - - assert(context); - - for (i = 0; i < 5; ++i) - context->pipe->delete_sampler_state(context->pipe, context->states.mc.samplers[i]); - - for (i = 0; i < 3; ++i) - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vertex_bufs[i].buffer); - - /* Textures 3 & 4 are not created directly, no need to release them here */ - for (i = 0; i < 3; ++i) - pipe_texture_release(&context->states.mc.textures[i]); - - context->pipe->delete_vs_state(context->pipe, context->states.mc.i_vs); - context->pipe->delete_fs_state(context->pipe, context->states.mc.i_fs); - - for (i = 0; i < 2; ++i) - { - context->pipe->delete_vs_state(context->pipe, context->states.mc.p_vs[i]); - context->pipe->delete_fs_state(context->pipe, context->states.mc.p_fs[i]); - context->pipe->delete_vs_state(context->pipe, context->states.mc.b_vs[i]); - context->pipe->delete_fs_state(context->pipe, context->states.mc.b_fs[i]); - } - - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.vs_const_buf.buffer); - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.mc.fs_const_buf.buffer); - - return 0; -} - -static int vlCreateVertexShaderCSC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Vertex texcoords - */ - for (i = 0; i < 2; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling vector to scale texcoord rect to source size - * decl c1 ; Translation vector to move texcoord rect into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Vertex texcoords - */ - for (i = 0; i < 2; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0 */ - decl = vl_decl_temps(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* mov o0, i0 ; Move pos in to pos out */ - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i1, c0 ; Scale unit texcoord rect to source size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 1, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o1, t0, c1 ; Translate texcoord rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - context->states.csc.vertex_shader = pipe->create_vs_state(pipe, &vs); - //free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderCSC(struct VL_CONTEXT *context) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(context); - - pipe = context->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* decl i0 ; Texcoords for s0 */ - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl c0 ; Bias vector for CSC - * decl c1-c4 ; CSC matrix c1-c4 - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0 */ - decl = vl_decl_temps(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl s0 ; Sampler for tex containing picture to display */ - decl = vl_decl_samplers(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* tex2d t0, i0, s0 ; Read src pixel */ - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c0 ; Subtract bias vector from pixel */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix - * dp4 o0.y, t0, c2 - * dp4 o0.z, t0, c3 - * dp4 o0.w, t0, c4 ; XXX: Don't need 4th coefficient - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - context->states.csc.fragment_shader = pipe->create_fs_state(pipe, &fs); - //free(tokens); - - return 0; -} - -static int vlCreateDataBufsCSC(struct VL_CONTEXT *context) -{ - struct pipe_context *pipe; - - assert(context); - - pipe = context->pipe; - - /* - Create our vertex buffer and vertex buffer element - VB contains 4 vertices that render a quad covering the entire window - to display a rendered surface - Quad is rendered as a tri strip - */ - context->states.csc.vertex_bufs[0].pitch = sizeof(struct VL_VERTEX2F); - context->states.csc.vertex_bufs[0].max_index = 3; - context->states.csc.vertex_bufs[0].buffer_offset = 0; - context->states.csc.vertex_bufs[0].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct VL_VERTEX2F) * 4 - ); - - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.csc.vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_surface_vertex_positions, - sizeof(struct VL_VERTEX2F) * 4 - ); - - pipe->winsys->buffer_unmap(pipe->winsys, context->states.csc.vertex_bufs[0].buffer); - - context->states.csc.vertex_buf_elems[0].src_offset = 0; - context->states.csc.vertex_buf_elems[0].vertex_buffer_index = 0; - context->states.csc.vertex_buf_elems[0].nr_components = 2; - context->states.csc.vertex_buf_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* - Create our texcoord buffer and texcoord buffer element - Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices - */ - context->states.csc.vertex_bufs[1].pitch = sizeof(struct VL_TEXCOORD2F); - context->states.csc.vertex_bufs[1].max_index = 3; - context->states.csc.vertex_bufs[1].buffer_offset = 0; - context->states.csc.vertex_bufs[1].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct VL_TEXCOORD2F) * 4 - ); - - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.csc.vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - vl_surface_texcoords, - sizeof(struct VL_TEXCOORD2F) * 4 - ); - - pipe->winsys->buffer_unmap(pipe->winsys, context->states.csc.vertex_bufs[1].buffer); - - context->states.csc.vertex_buf_elems[1].src_offset = 0; - context->states.csc.vertex_buf_elems[1].vertex_buffer_index = 1; - context->states.csc.vertex_buf_elems[1].nr_components = 2; - context->states.csc.vertex_buf_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* - Create our vertex shader's constant buffer - Const buffer contains scaling and translation vectors - */ - context->states.csc.vs_const_buf.size = sizeof(struct VL_CSC_VS_CONSTS); - context->states.csc.vs_const_buf.buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_CONSTANT, - context->states.csc.vs_const_buf.size - ); - - /* - Create our fragment shader's constant buffer - Const buffer contains the color conversion matrix and bias vectors - */ - context->states.csc.fs_const_buf.size = sizeof(struct VL_CSC_FS_CONSTS); - context->states.csc.fs_const_buf.buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_CONSTANT, - context->states.csc.fs_const_buf.size - ); - - /* - TODO: Refactor this into a seperate function, - allow changing the CSC matrix at runtime to switch between regular & full versions - */ - memcpy - ( - pipe->winsys->buffer_map(pipe->winsys, context->states.csc.fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - &vl_csc_fs_consts_601, - sizeof(struct VL_CSC_FS_CONSTS) - ); - - pipe->winsys->buffer_unmap(pipe->winsys, context->states.csc.fs_const_buf.buffer); - - return 0; -} - -static int vlInitCSC(struct VL_CONTEXT *context) -{ - struct pipe_context *pipe; - struct pipe_sampler_state sampler; - - assert(context); - - pipe = context->pipe; - - /* Delay creating the FB until vlPutSurface() so we know window size */ - context->states.csc.framebuffer.num_cbufs = 1; - context->states.csc.framebuffer.cbufs[0] = NULL; - context->states.csc.framebuffer.zsbuf = NULL; - - sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; - sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; - sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; - sampler.compare_mode = PIPE_TEX_COMPARE_NONE; - sampler.compare_func = PIPE_FUNC_ALWAYS; - sampler.normalized_coords = 1; - /*sampler.prefilter = ;*/ - /*sampler.shadow_ambient = ;*/ - /*sampler.lod_bias = ;*/ - /*sampler.min_lod = ;*/ - /*sampler.max_lod = ;*/ - /*sampler.border_color[i] = ;*/ - /*sampler.max_anisotropy = ;*/ - context->states.csc.sampler = pipe->create_sampler_state(pipe, &sampler); - - vlCreateVertexShaderCSC(context); - vlCreateFragmentShaderCSC(context); - vlCreateDataBufsCSC(context); - - return 0; -} - -static int vlDestroyCSC(struct VL_CONTEXT *context) -{ - assert(context); - - /* - Since we create the final FB when we display our first surface, - it may not be created if vlPutSurface() is never called - */ - if (context->states.csc.framebuffer.cbufs[0]) - context->pipe->winsys->surface_release(context->pipe->winsys, &context->states.csc.framebuffer.cbufs[0]); - context->pipe->delete_sampler_state(context->pipe, context->states.csc.sampler); - context->pipe->delete_vs_state(context->pipe, context->states.csc.vertex_shader); - context->pipe->delete_fs_state(context->pipe, context->states.csc.fragment_shader); - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vertex_bufs[0].buffer); - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vertex_bufs[1].buffer); - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.vs_const_buf.buffer); - context->pipe->winsys->buffer_destroy(context->pipe->winsys, context->states.csc.fs_const_buf.buffer); - - return 0; -} - -static int vlInitCommon(struct VL_CONTEXT *context) +static int vlInitCommon(struct vlContext *context) { struct pipe_context *pipe; struct pipe_rasterizer_state rast; struct pipe_blend_state blend; struct pipe_depth_stencil_alpha_state dsa; unsigned int i; - + assert(context); - + pipe = context->pipe; - + rast.flatshade = 1; rast.flatshade_first = 0; rast.light_twoside = 0; @@ -2113,9 +51,9 @@ static int vlInitCommon(struct VL_CONTEXT *context) rast.offset_units = 1; rast.offset_scale = 1; /*rast.sprite_coord_mode[i] = ;*/ - context->states.common.raster = pipe->create_rasterizer_state(pipe, &rast); - pipe->bind_rasterizer_state(pipe, context->states.common.raster); - + context->raster = pipe->create_rasterizer_state(pipe, &rast); + pipe->bind_rasterizer_state(pipe, context->raster); + blend.blend_enable = 0; blend.rgb_func = PIPE_BLEND_ADD; blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE; @@ -2128,9 +66,9 @@ static int vlInitCommon(struct VL_CONTEXT *context) /* Needed to allow color writes to FB, even if blending disabled */ blend.colormask = PIPE_MASK_RGBA; blend.dither = 0; - context->states.common.blend = pipe->create_blend_state(pipe, &blend); - pipe->bind_blend_state(pipe, context->states.common.blend); - + context->blend = pipe->create_blend_state(pipe, &blend); + pipe->bind_blend_state(pipe, context->blend); + dsa.depth.enabled = 0; dsa.depth.writemask = 0; dsa.depth.func = PIPE_FUNC_ALWAYS; @@ -2149,134 +87,122 @@ static int vlInitCommon(struct VL_CONTEXT *context) dsa.alpha.enabled = 0; dsa.alpha.func = PIPE_FUNC_ALWAYS; dsa.alpha.ref = 0; - context->states.common.dsa = pipe->create_depth_stencil_alpha_state(pipe, &dsa); - pipe->bind_depth_stencil_alpha_state(pipe, context->states.common.dsa); - - return 0; -} + context->dsa = pipe->create_depth_stencil_alpha_state(pipe, &dsa); + pipe->bind_depth_stencil_alpha_state(pipe, context->dsa); -static int vlDestroyCommon(struct VL_CONTEXT *context) -{ - assert(context); - - context->pipe->delete_blend_state(context->pipe, context->states.common.blend); - context->pipe->delete_rasterizer_state(context->pipe, context->states.common.raster); - context->pipe->delete_depth_stencil_alpha_state(context->pipe, context->states.common.dsa); - return 0; } -static int vlInit(struct VL_CONTEXT *context) +int vlCreateContext +( + struct vlScreen *screen, + struct pipe_context *pipe, + unsigned int picture_width, + unsigned int picture_height, + enum vlFormat picture_format, + enum vlProfile profile, + enum vlEntryPoint entry_point, + struct vlContext **context +) { + struct vlContext *ctx; + + assert(screen); assert(context); - - vlInitCommon(context); - vlInitCSC(context); - vlInitMC(context); - vlInitIDCT(context); - + assert(pipe); + + ctx = calloc(1, sizeof(struct vlContext)); + + if (!ctx) + return 1; + + ctx->screen = screen; + ctx->pipe = pipe; + ctx->picture_width = picture_width; + ctx->picture_height = picture_height; + ctx->picture_format = picture_format; + ctx->profile = profile; + ctx->entry_point = entry_point; + + vlInitCommon(ctx); + + vlCreateR16SNormMC(pipe, picture_width, picture_height, picture_format, &ctx->render); + vlCreateBasicCSC(pipe, &ctx->csc); + + *context = ctx; + return 0; } -static int vlDestroy(struct VL_CONTEXT *context) +int vlDestroyContext +( + struct vlContext *context +) { assert(context); - + /* XXX: Must unbind shaders before we can delete them for some reason */ context->pipe->bind_vs_state(context->pipe, NULL); context->pipe->bind_fs_state(context->pipe, NULL); - - vlDestroyCommon(context); - vlDestroyCSC(context); - vlDestroyMC(context); - vlDestroyIDCT(context); - + + context->render->vlDestroy(context->render); + context->csc->vlDestroy(context->csc); + + context->pipe->delete_blend_state(context->pipe, context->blend); + context->pipe->delete_rasterizer_state(context->pipe, context->raster); + context->pipe->delete_depth_stencil_alpha_state(context->pipe, context->dsa); + + free(context); + return 0; } -int vlCreateContext +struct vlScreen* vlContextGetScreen ( - Display *display, - struct pipe_context *pipe, - unsigned int video_width, - unsigned int video_height, - enum VL_FORMAT video_format, - struct VL_CONTEXT **context + struct vlContext *context ) { - struct VL_CONTEXT *ctx; - - assert(display); - assert(pipe); assert(context); - - ctx = calloc(1, sizeof(struct VL_CONTEXT)); - - ctx->display = display; - ctx->pipe = pipe; - ctx->video_width = video_width; - ctx->video_height = video_height; - ctx->video_format = video_format; - - vlInit(ctx); - - /* Since we only change states in vlPutSurface() we need to start in render mode */ - vlBeginRender(ctx); - - *context = ctx; - - return 0; + + return context->screen; } -int vlDestroyContext(struct VL_CONTEXT *context) +struct pipe_context* vlGetPipeContext +( + struct vlContext *context +) { assert(context); - - vlDestroy(context); - - free(context); - - return 0; + + return context->pipe; } -int vlBeginRender(struct VL_CONTEXT *context) +unsigned int vlGetPictureWidth +( + struct vlContext *context +) { - struct pipe_context *pipe; - assert(context); - - pipe = context->pipe; - - /* Frame buffer set in vlRender*Macroblock() */ - /* Shaders, samplers, textures set in vlRender*Macroblock() */ - pipe->set_vertex_buffers(pipe, 3, context->states.mc.vertex_bufs); - pipe->set_vertex_elements(pipe, 3, context->states.mc.vertex_buf_elems); - pipe->set_viewport_state(pipe, &context->states.mc.viewport); - pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &context->states.mc.vs_const_buf); - pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &context->states.mc.fs_const_buf); - - return 0; + + return context->picture_width; } -int vlEndRender(struct VL_CONTEXT *context) +unsigned int vlGetPictureHeight +( + struct vlContext *context +) { - struct pipe_context *pipe; - assert(context); - - pipe = context->pipe; - - pipe->set_framebuffer_state(pipe, &context->states.csc.framebuffer); - pipe->set_viewport_state(pipe, &context->states.csc.viewport); - pipe->bind_sampler_states(pipe, 1, (void**)&context->states.csc.sampler); - /* Source texture set in vlPutSurface() */ - pipe->bind_vs_state(pipe, context->states.csc.vertex_shader); - pipe->bind_fs_state(pipe, context->states.csc.fragment_shader); - pipe->set_vertex_buffers(pipe, 2, context->states.csc.vertex_bufs); - pipe->set_vertex_elements(pipe, 2, context->states.csc.vertex_buf_elems); - pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &context->states.csc.vs_const_buf); - pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &context->states.csc.fs_const_buf); - - return 0; + + return context->picture_height; } +enum vlFormat vlGetPictureFormat +( + struct vlContext *context +) +{ + assert(context); + + return context->picture_format; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h index bff318854aa..3d14634c44e 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.h +++ b/src/gallium/state_trackers/g3dvl/vl_context.h @@ -1,83 +1,73 @@ #ifndef vl_context_h #define vl_context_h -#include -#include #include "vl_types.h" struct pipe_context; -struct VL_CONTEXT +#ifdef VL_INTERNAL +struct vlRender; +struct vlCSC; + +struct vlContext { - Display *display; + struct vlScreen *screen; struct pipe_context *pipe; - unsigned int video_width; - unsigned int video_height; - enum VL_FORMAT video_format; - - struct - { - struct - { - struct pipe_rasterizer_state *raster; - struct pipe_depth_stencil_alpha_state *dsa; - struct pipe_blend_state *blend; - } common; - - struct - { - struct pipe_viewport_state viewport; - struct pipe_framebuffer_state render_target; - struct pipe_sampler_state *sampler; - struct pipe_texture *texture; - struct pipe_texture *basis; - struct pipe_shader_state *frame_vs; - struct pipe_shader_state *frame_fs; - struct pipe_vertex_buffer *vertex_bufs[2]; - struct pipe_vertex_element *vertex_buf_elems[2]; - //struct pipe_constant_buffer vs_const_buf, fs_const_buf; - } idct; - - struct - { - struct pipe_viewport_state viewport; - struct pipe_framebuffer_state render_target; - struct pipe_sampler_state *samplers[5]; - struct pipe_texture *textures[5]; - struct pipe_shader_state *i_vs, *p_vs[2], *b_vs[2]; - struct pipe_shader_state *i_fs, *p_fs[2], *b_fs[2]; - struct pipe_vertex_buffer vertex_bufs[3]; - struct pipe_vertex_element vertex_buf_elems[3]; - struct pipe_constant_buffer vs_const_buf, fs_const_buf; - } mc; - - struct - { - struct pipe_viewport_state viewport; - struct pipe_framebuffer_state framebuffer; - struct pipe_sampler_state *sampler; - struct pipe_shader_state *vertex_shader, *fragment_shader; - struct pipe_vertex_buffer vertex_bufs[2]; - struct pipe_vertex_element vertex_buf_elems[2]; - struct pipe_constant_buffer vs_const_buf, fs_const_buf; - } csc; - } states; + unsigned int picture_width; + unsigned int picture_height; + enum vlFormat picture_format; + enum vlProfile profile; + enum vlEntryPoint entry_point; + + void *raster; + void *dsa; + void *blend; + + struct vlRender *render; + struct vlCSC *csc; }; +#endif int vlCreateContext ( - Display *display, + struct vlScreen *screen, struct pipe_context *pipe, - unsigned int video_width, - unsigned int video_height, - enum VL_FORMAT video_format, - struct VL_CONTEXT **context + unsigned int picture_width, + unsigned int picture_height, + enum vlFormat picture_format, + enum vlProfile profile, + enum vlEntryPoint entry_point, + struct vlContext **context ); -int vlDestroyContext(struct VL_CONTEXT *context); +int vlDestroyContext +( + struct vlContext *context +); -int vlBeginRender(struct VL_CONTEXT *context); -int vlEndRender(struct VL_CONTEXT *context); +struct vlScreen* vlContextGetScreen +( + struct vlContext *context +); -#endif +struct pipe_context* vlGetPipeContext +( + struct vlContext *context +); +unsigned int vlGetPictureWidth +( + struct vlContext *context +); + +unsigned int vlGetPictureHeight +( + struct vlContext *context +); + +enum vlFormat vlGetPictureFormat +( + struct vlContext *context +); + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_csc.h b/src/gallium/state_trackers/g3dvl/vl_csc.h new file mode 100644 index 00000000000..36417a27929 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_csc.h @@ -0,0 +1,53 @@ +#ifndef vl_csc_h +#define vl_csc_h + +#include "vl_types.h" + +struct pipe_surface; + +struct vlCSC +{ + int (*vlResizeFrameBuffer) + ( + struct vlCSC *csc, + unsigned int width, + unsigned int height + ); + + int (*vlBegin) + ( + struct vlCSC *csc + ); + + int (*vlPutPicture) + ( + struct vlCSC *csc, + struct vlSurface *surface, + int srcx, + int srcy, + int srcw, + int srch, + int destx, + int desty, + int destw, + int desth, + enum vlPictureType picture_type + ); + + int (*vlEnd) + ( + struct vlCSC *csc + ); + + struct pipe_surface* (*vlGetFrameBuffer) + ( + struct vlCSC *csc + ); + + int (*vlDestroy) + ( + struct vlCSC *csc + ); +}; + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_data.c b/src/gallium/state_trackers/g3dvl/vl_data.c index 0e5c8c77f90..f2476dbf1ee 100644 --- a/src/gallium/state_trackers/g3dvl/vl_data.c +++ b/src/gallium/state_trackers/g3dvl/vl_data.c @@ -6,17 +6,17 @@ * Need to be scaled to cover mbW*mbH macroblock pixels and translated into * position on target surface. */ -const struct VL_VERTEX2F vl_mb_vertex_positions[24] = +const struct vlVertex2f macroblock_verts[24] = { {0.0f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.0f}, {0.5f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.5f}, - + {0.5f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.0f}, {1.0f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.5f}, - + {0.0f, 0.5f}, {0.0f, 1.0f}, {0.5f, 0.5f}, {0.5f, 0.5f}, {0.0f, 1.0f}, {0.5f, 1.0f}, - + {0.5f, 0.5f}, {0.5f, 1.0f}, {1.0f, 0.5f}, {1.0f, 0.5f}, {0.5f, 1.0f}, {1.0f, 1.0f} }; @@ -26,17 +26,17 @@ const struct VL_VERTEX2F vl_mb_vertex_positions[24] = * in a bW*(bH*4) texture. First luma block located at 0,0->bW,bH; second at * 0,bH->bW,2bH; third at 0,2bH->bW,3bH; fourth at 0,3bH->bW,4bH. */ -const struct VL_TEXCOORD2F vl_luma_texcoords[24] = +const struct vlVertex2f macroblock_luma_texcoords[24] = { {0.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.0f}, {1.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.25f}, - + {0.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.25f}, {1.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.5f}, - + {0.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.5f}, {1.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.75f}, - + {0.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 0.75f}, {1.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 1.0f} }; @@ -45,7 +45,7 @@ const struct VL_TEXCOORD2F vl_luma_texcoords[24] = * Represents texcoords for the above for rendering 1 chroma block. * Straight forward 0,0->1,1 mapping so we can reuse the MB pos vectors. */ -const struct VL_TEXCOORD2F *vl_chroma_420_texcoords = (const struct VL_TEXCOORD2F*)vl_mb_vertex_positions; +const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts; /* * Represents texcoords for the above for rendering 2 chroma blocks arranged @@ -53,30 +53,13 @@ const struct VL_TEXCOORD2F *vl_chroma_420_texcoords = (const struct VL_TEXCOORD2 * 0,bH->bW,2bH. We can render this with 0,0->1,1 mapping. * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. */ -const struct VL_TEXCOORD2F *vl_chroma_422_texcoords = (const struct VL_TEXCOORD2F*)vl_mb_vertex_positions; +const struct vlVertex2f *macroblock_chroma_422_texcoords = macroblock_verts; /* * Represents texcoords for the above for rendering 4 chroma blocks. * Same case as 4 luma blocks. */ -const struct VL_TEXCOORD2F *vl_chroma_444_texcoords = vl_luma_texcoords; - -/* - * Represents 2 triangles in a strip in normalized coords. - * Used to render the surface onto the frame buffer. - */ -const struct VL_VERTEX2F vl_surface_vertex_positions[4] = -{ - {0.0f, 0.0f}, - {0.0f, 1.0f}, - {1.0f, 0.0f}, - {1.0f, 1.0f} -}; - -/* - * Represents texcoords for the above. We can use the position values directly. - */ -const struct VL_TEXCOORD2F *vl_surface_texcoords = (const struct VL_TEXCOORD2F*)vl_surface_vertex_positions; +const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texcoords; /* * Used when rendering P and B macroblocks, multiplier is applied to the A channel, @@ -84,97 +67,10 @@ const struct VL_TEXCOORD2F *vl_surface_texcoords = (const struct VL_TEXCOORD2F*) * get back the differential. The differential is then added to the samples from the * reference surface(s). */ +#if 0 const struct VL_MC_FS_CONSTS vl_mc_fs_consts = { {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, {0.5f, 2.0f, 0.0f, 0.0f} }; - -/* - * Identity color conversion constants, for debugging - */ -const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_identity = -{ - { - 0.0f, 0.0f, 0.0f, 0.0f - }, - { - 1.0f, 0.0f, 0.0f, 0.0f, - 0.0f, 1.0f, 0.0f, 0.0f, - 0.0f, 0.0f, 1.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [16,235] - */ -const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601 = -{ - { - 0.0f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.0f, 0.0f, 1.371f, 0.0f, - 1.0f, -0.336f, -0.698f, 0.0f, - 1.0f, 1.732f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [0,255] - */ -const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601_full = -{ - { - 0.062745098f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.164f, 0.0f, 1.596f, 0.0f, - 1.164f, -0.391f, -0.813f, 0.0f, - 1.164f, 2.018f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [16,235] - */ -const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709 = -{ - { - 0.0f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.0f, 0.0f, 1.540f, 0.0f, - 1.0f, -0.183f, -0.459f, 0.0f, - 1.0f, 1.816f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [0,255] - */ -const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709_full = -{ - { - 0.062745098f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.164f, 0.0f, 1.793f, 0.0f, - 1.164f, -0.213f, -0.534f, 0.0f, - 1.164f, 2.115f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_data.h b/src/gallium/state_trackers/g3dvl/vl_data.h index 8f347273ad7..f0de2e976ce 100644 --- a/src/gallium/state_trackers/g3dvl/vl_data.h +++ b/src/gallium/state_trackers/g3dvl/vl_data.h @@ -3,15 +3,18 @@ #include "vl_types.h" -extern const struct VL_VERTEX2F vl_mb_vertex_positions[24]; -extern const struct VL_TEXCOORD2F vl_luma_texcoords[24]; -extern const struct VL_TEXCOORD2F *vl_chroma_420_texcoords; -extern const struct VL_TEXCOORD2F *vl_chroma_422_texcoords; -extern const struct VL_TEXCOORD2F *vl_chroma_444_texcoords; +/* TODO: Needs to be rolled into the appropriate stage */ -extern const struct VL_VERTEX2F vl_surface_vertex_positions[4]; -extern const struct VL_TEXCOORD2F *vl_surface_texcoords; +extern const struct vlVertex2f macroblock_verts[24]; +extern const struct vlVertex2f macroblock_luma_texcoords[24]; +extern const struct vlVertex2f *macroblock_chroma_420_texcoords; +extern const struct vlVertex2f *macroblock_chroma_422_texcoords; +extern const struct vlVertex2f *macroblock_chroma_444_texcoords; +extern const struct vlVertex2f surface_verts[4]; +extern const struct vlVertex2f *surface_texcoords; + +/* extern const struct VL_MC_FS_CONSTS vl_mc_fs_consts; extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_identity; @@ -19,6 +22,6 @@ extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601; extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601_full; extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709; extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709_full; +*/ #endif - diff --git a/src/gallium/state_trackers/g3dvl/vl_defs.h b/src/gallium/state_trackers/g3dvl/vl_defs.h index e668a7a10e0..d612d02502f 100644 --- a/src/gallium/state_trackers/g3dvl/vl_defs.h +++ b/src/gallium/state_trackers/g3dvl/vl_defs.h @@ -9,4 +9,3 @@ #define VL_MACROBLOCK_SIZE (VL_MACROBLOCK_WIDTH * VL_MACROBLOCK_HEIGHT) #endif - diff --git a/src/gallium/state_trackers/g3dvl/vl_display.c b/src/gallium/state_trackers/g3dvl/vl_display.c new file mode 100644 index 00000000000..af80faa7f52 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_display.c @@ -0,0 +1,48 @@ +#define VL_INTERNAL +#include "vl_display.h" +#include +#include + +int vlCreateDisplay +( + vlNativeDisplay native_display, + struct vlDisplay **display +) +{ + struct vlDisplay *dpy; + + assert(native_display); + assert(display); + + dpy = calloc(1, sizeof(struct vlDisplay)); + + if (!dpy) + return 1; + + dpy->native = native_display; + *display = dpy; + + return 0; +} + +int vlDestroyDisplay +( + struct vlDisplay *display +) +{ + assert(display); + + free(display); + + return 0; +} + +vlNativeDisplay vlGetNativeDisplay +( + struct vlDisplay *display +) +{ + assert(display); + + return display->native; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_display.h b/src/gallium/state_trackers/g3dvl/vl_display.h new file mode 100644 index 00000000000..e11fd407993 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_display.h @@ -0,0 +1,29 @@ +#ifndef vl_display_h +#define vl_display_h + +#include "vl_types.h" + +#ifdef VL_INTERNAL +struct vlDisplay +{ + vlNativeDisplay native; +}; +#endif + +int vlCreateDisplay +( + vlNativeDisplay native_display, + struct vlDisplay **display +); + +int vlDestroyDisplay +( + struct vlDisplay *display +); + +vlNativeDisplay vlGetNativeDisplay +( + struct vlDisplay *display +); + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c new file mode 100644 index 00000000000..4fae2244310 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c @@ -0,0 +1,2315 @@ +#define VL_INTERNAL +#include "vl_r16snorm_mc.h" +#include +#include +#include +#include +#include +#include +#include +#include +#include "vl_render.h" +#include "vl_shader_build.h" +#include "vl_surface.h" +#include "vl_util.h" +#include "vl_types.h" +#include "vl_defs.h" + +struct vlVertexShaderConsts +{ + /*struct vlVertex4f scale; + struct vlVertex4f denorm;*/ + struct vlVertex4f scale; + struct vlVertex4f mb_pos_trans; + struct vlVertex4f denorm; + struct + { + struct vlVertex4f top_field; + struct vlVertex4f bottom_field; + } mb_tc_trans[2]; +}; + +struct vlFragmentShaderConsts +{ + struct vlVertex4f multiplier; + struct vlVertex4f div; +}; + +struct vlR16SnormMC +{ + struct vlRender base; + + unsigned int video_width, video_height; + enum vlFormat video_format; + + struct pipe_context *pipe; + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state render_target; + struct pipe_sampler_state *samplers[5]; + struct pipe_texture *textures[5]; + void *i_vs, *p_vs[2], *b_vs[2]; + void *i_fs, *p_fs[2], *b_fs[2]; + struct pipe_vertex_buffer vertex_bufs[3]; + struct pipe_vertex_element vertex_elems[3]; + struct pipe_constant_buffer vs_const_buf, fs_const_buf; +}; + +int vlBegin +( + struct vlRender *render +) +{ + struct vlR16SnormMC *mc; + struct pipe_context *pipe; + + assert(render); + + mc = (struct vlR16SnormMC*)render; + pipe = mc->pipe; + + /* Frame buffer set in vlRender*Macroblock() */ + /* Shaders, samplers, textures set in vlRender*Macroblock() */ + pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs); + pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + pipe->set_viewport_state(pipe, &mc->viewport); + pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &mc->vs_const_buf); + pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf); + + return 0; +} + +/*static int vlGrabMacroBlock +( + struct vlR16SnormMC *mc, + struct vlMpeg2MacroBlock *macroblock +) +{ + assert(mc); + assert(macroblock); + + + + return 0; +}*/ + +/*#define DO_IDCT*/ + +#ifdef DO_IDCT +static int vlTransformBlock(short *src, short *dst, short bias) +{ + static const float basis[8][8] = + { + {0.3536, 0.4904, 0.4619, 0.4157, 0.3536, 0.2778, 0.1913, 0.0975}, + {0.3536, 0.4157, 0.1913, -0.0975, -0.3536, -0.4904, -0.4619, -0.2778}, + {0.3536, 0.2778, -0.1913, -0.4904, -0.3536, 0.0975, 0.4619, 0.4157}, + {0.3536, 0.0975, -0.4619, -0.2778, 0.3536, 0.4157, -0.1913, -0.4904}, + {0.3536, -0.0975, -0.4619, 0.2778, 0.3536, -0.4157, -0.1913, 0.4904}, + {0.3536, -0.2778, -0.1913, 0.4904, -0.3536, -0.0975, 0.4619, -0.4157}, + {0.3536, -0.4157, 0.1913, 0.0975, -0.3536, 0.4904, -0.4619, 0.2778}, + {0.3536, -0.4904, 0.4619, -0.4157, 0.3536, -0.2778, 0.1913, -0.0975} + }; + + unsigned int x, y; + short tmp[64]; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + tmp[y * VL_BLOCK_WIDTH + x] = (short) + ( + src[y * VL_BLOCK_WIDTH + 0] * basis[x][0] + + src[y * VL_BLOCK_WIDTH + 1] * basis[x][1] + + src[y * VL_BLOCK_WIDTH + 2] * basis[x][2] + + src[y * VL_BLOCK_WIDTH + 3] * basis[x][3] + + src[y * VL_BLOCK_WIDTH + 4] * basis[x][4] + + src[y * VL_BLOCK_WIDTH + 5] * basis[x][5] + + src[y * VL_BLOCK_WIDTH + 6] * basis[x][6] + + src[y * VL_BLOCK_WIDTH + 7] * basis[x][7] + ); + + for (x = 0; x < VL_BLOCK_WIDTH; ++x) + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + { + dst[y * VL_BLOCK_WIDTH + x] = bias + (short) + ( + tmp[0 * VL_BLOCK_WIDTH + x] * basis[y][0] + + tmp[1 * VL_BLOCK_WIDTH + x] * basis[y][1] + + tmp[2 * VL_BLOCK_WIDTH + x] * basis[y][2] + + tmp[3 * VL_BLOCK_WIDTH + x] * basis[y][3] + + tmp[4 * VL_BLOCK_WIDTH + x] * basis[y][4] + + tmp[5 * VL_BLOCK_WIDTH + x] * basis[y][5] + + tmp[6 * VL_BLOCK_WIDTH + x] * basis[y][6] + + tmp[7 * VL_BLOCK_WIDTH + x] * basis[y][7] + ); + if (dst[y * VL_BLOCK_WIDTH + x] > 255) + dst[y * VL_BLOCK_WIDTH + x] = 255; + else if (bias > 0 && dst[y * VL_BLOCK_WIDTH + x] < 0) + dst[y * VL_BLOCK_WIDTH + x] = 0; + } + return 0; +} +#endif + +static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + memcpy + ( + dst + y * dst_pitch, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) + memcpy + ( + dst + y * dst_pitch * 2, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + dst += VL_BLOCK_HEIGHT * dst_pitch; + + for (; y < VL_BLOCK_HEIGHT; ++y) + memcpy + ( + dst + y * dst_pitch * 2, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + memset + ( + dst + y * dst_pitch, + 0, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +enum vlSampleType +{ + vlSampleTypeFull, + vlSampleTypeDiff +}; + +static int vlGrabBlocks +( + struct vlR16SnormMC *mc, + unsigned int coded_block_pattern, + enum vlDCTType dct_type, + enum vlSampleType sample_type, + short *blocks +) +{ + struct pipe_surface *tex_surface; + short *texels; + unsigned int tex_pitch; + unsigned int tb, sb = 0; + + assert(mc); + assert(blocks); + + tex_surface = mc->pipe->screen->get_tex_surface + ( + mc->pipe->screen, + mc->textures[0], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); + + texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); + tex_pitch = tex_surface->stride / tex_surface->block.size; + + for (tb = 0; tb < 4; ++tb) + { + if ((coded_block_pattern >> (5 - tb)) & 1) + { + short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; + +#ifdef DO_IDCT + vlTransformBlock(cur_block, cur_block, sample_type == vlSampleTypeFull ? 128 : 0); +#endif + + if (dct_type == vlDCTTypeFrameCoded) + vlGrabFrameCodedBlock + ( + cur_block, + texels + tb * tex_pitch * VL_BLOCK_HEIGHT, + tex_pitch + ); + else + vlGrabFieldCodedBlock + ( + cur_block, + texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, + tex_pitch + ); + + ++sb; + } + else + vlGrabNoBlock(texels + tb * tex_pitch * VL_BLOCK_HEIGHT, tex_pitch); + } + + pipe_surface_unmap(tex_surface); + + /* TODO: Implement 422, 444 */ + for (tb = 0; tb < 2; ++tb) + { + tex_surface = mc->pipe->screen->get_tex_surface + ( + mc->pipe->screen, + mc->textures[tb + 1], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); + + texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); + tex_pitch = tex_surface->stride / tex_surface->block.size; + + if ((coded_block_pattern >> (1 - tb)) & 1) + { + short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; + +#ifdef DO_IDCT + vlTransformBlock(cur_block, cur_block, sample_type == vlSampleTypeFull ? 128 : 0); +#endif + + vlGrabFrameCodedBlock + ( + cur_block, + texels, + tex_pitch + ); + + ++sb; + } + else + vlGrabNoBlock(texels, tex_pitch); + + pipe_surface_unmap(tex_surface); + } + + return 0; +} + +int vlRenderIMacroBlock +( + struct vlR16SnormMC *mc, + enum vlPictureType picture_type, + enum vlFieldOrder field_order, + unsigned int mbx, + unsigned int mby, + unsigned int coded_block_pattern, + enum vlDCTType dct_type, + short *blocks, + struct vlSurface *surface +) +{ + struct pipe_context *pipe; + struct vlVertexShaderConsts *vs_consts; + + assert(blocks); + assert(surface); + + /* TODO: Implement interlaced rendering */ + if (picture_type != vlPictureTypeFrame) + return 0; + + vlGrabBlocks(mc, coded_block_pattern, dct_type, vlSampleTypeFull, blocks); + + pipe = mc->pipe; + + vs_consts = pipe->winsys->buffer_map + ( + pipe->winsys, + mc->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->texture->width[0]; + vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->texture->height[0]; + vs_consts->scale.z = 1.0f; + vs_consts->scale.w = 1.0f; + vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->texture->width[0]; + vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->texture->height[0]; + vs_consts->mb_pos_trans.z = 0.0f; + vs_consts->mb_pos_trans.w = 0.0f; + + pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer); + + mc->render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + ); + pipe->set_framebuffer_state(pipe, &mc->render_target); + pipe->set_sampler_textures(pipe, 3, mc->textures); + pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->i_vs); + pipe->bind_fs_state(pipe, mc->i_fs); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + return 0; +} + +int vlRenderPMacroBlock +( + struct vlR16SnormMC *mc, + enum vlPictureType picture_type, + enum vlFieldOrder field_order, + unsigned int mbx, + unsigned int mby, + enum vlMotionType mc_type, + short top_x, + short top_y, + short bottom_x, + short bottom_y, + unsigned int coded_block_pattern, + enum vlDCTType dct_type, + short *blocks, + struct vlSurface *ref_surface, + struct vlSurface *surface +) +{ + struct pipe_context *pipe; + struct vlVertexShaderConsts *vs_consts; + + assert(motion_vectors); + assert(blocks); + assert(ref_surface); + assert(surface); + + /* TODO: Implement interlaced rendering */ + if (picture_type != vlPictureTypeFrame) + return 0; + /* TODO: Implement other MC types */ + if (mc_type != vlMotionTypeFrame && mc_type != vlMotionTypeField) + return 0; + + vlGrabBlocks(mc, coded_block_pattern, dct_type, vlSampleTypeDiff, blocks); + + pipe = mc->pipe; + + vs_consts = pipe->winsys->buffer_map + ( + pipe->winsys, + mc->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->texture->width[0]; + vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->texture->height[0]; + vs_consts->scale.z = 1.0f; + vs_consts->scale.w = 1.0f; + vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->texture->width[0]; + vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->texture->height[0]; + vs_consts->mb_pos_trans.z = 0.0f; + vs_consts->mb_pos_trans.w = 0.0f; + vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + top_x * 0.5f) / (float)surface->texture->width[0]; + vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + top_y * 0.5f) / (float)surface->texture->height[0]; + vs_consts->mb_tc_trans[0].top_field.z = 0.0f; + vs_consts->mb_tc_trans[0].top_field.w = 0.0f; + + if (mc_type == vlMotionTypeField) + { + vs_consts->denorm.x = (float)surface->texture->width[0]; + vs_consts->denorm.y = (float)surface->texture->height[0]; + + vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + bottom_x * 0.5f) / (float)surface->texture->width[0]; + vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + bottom_y * 0.5f) / (float)surface->texture->height[0]; + vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f; + vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f; + + pipe->bind_vs_state(pipe, mc->p_vs[1]); + pipe->bind_fs_state(pipe, mc->p_fs[1]); + } + else + { + pipe->bind_vs_state(pipe, mc->p_vs[0]); + pipe->bind_fs_state(pipe, mc->p_fs[0]); + } + + pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer); + + mc->render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + ); + pipe->set_framebuffer_state(pipe, &mc->render_target); + + mc->textures[3] = ref_surface->texture; + pipe->set_sampler_textures(pipe, 4, mc->textures); + pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + return 0; +} + +int vlRenderBMacroBlock +( + struct vlR16SnormMC *mc, + enum vlPictureType picture_type, + enum vlFieldOrder field_order, + unsigned int mbx, + unsigned int mby, + enum vlMotionType mc_type, + short top_past_x, + short top_past_y, + short bottom_past_x, + short bottom_past_y, + short top_future_x, + short top_future_y, + short bottom_future_x, + short bottom_future_y, + unsigned int coded_block_pattern, + enum vlDCTType dct_type, + short *blocks, + struct vlSurface *past_surface, + struct vlSurface *future_surface, + struct vlSurface *surface +) +{ + struct pipe_context *pipe; + struct vlVertexShaderConsts *vs_consts; + + assert(motion_vectors); + assert(blocks); + assert(ref_surface); + assert(surface); + + /* TODO: Implement interlaced rendering */ + if (picture_type != vlPictureTypeFrame) + return 0; + /* TODO: Implement other MC types */ + if (mc_type != vlMotionTypeFrame && mc_type != vlMotionTypeField) + return 0; + + vlGrabBlocks(mc, coded_block_pattern, dct_type, vlSampleTypeDiff, blocks); + + pipe = mc->pipe; + + vs_consts = pipe->winsys->buffer_map + ( + pipe->winsys, + mc->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->texture->width[0]; + vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->texture->height[0]; + vs_consts->scale.z = 1.0f; + vs_consts->scale.w = 1.0f; + vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->texture->width[0]; + vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->texture->height[0]; + vs_consts->mb_pos_trans.z = 0.0f; + vs_consts->mb_pos_trans.w = 0.0f; + vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + top_past_x * 0.5f) / (float)surface->texture->width[0]; + vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + top_past_y * 0.5f) / (float)surface->texture->height[0]; + vs_consts->mb_tc_trans[0].top_field.z = 0.0f; + vs_consts->mb_tc_trans[0].top_field.w = 0.0f; + vs_consts->mb_tc_trans[1].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + top_future_x * 0.5f) / (float)surface->texture->width[0]; + vs_consts->mb_tc_trans[1].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + top_future_y * 0.5f) / (float)surface->texture->height[0]; + vs_consts->mb_tc_trans[1].top_field.z = 0.0f; + vs_consts->mb_tc_trans[1].top_field.w = 0.0f; + + if (mc_type == vlMotionTypeField) + { + vs_consts->denorm.x = (float)surface->texture->width[0]; + vs_consts->denorm.y = (float)surface->texture->height[0]; + + vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + bottom_past_x * 0.5f) / (float)surface->texture->width[0]; + vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + bottom_past_y * 0.5f) / (float)surface->texture->height[0]; + vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f; + vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f; + vs_consts->mb_tc_trans[1].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + bottom_future_x * 0.5f) / (float)surface->texture->width[0]; + vs_consts->mb_tc_trans[1].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + bottom_future_y * 0.5f) / (float)surface->texture->height[0]; + vs_consts->mb_tc_trans[1].bottom_field.z = 0.0f; + vs_consts->mb_tc_trans[1].bottom_field.w = 0.0f; + + pipe->bind_vs_state(pipe, mc->b_vs[1]); + pipe->bind_fs_state(pipe, mc->b_fs[1]); + } + else + { + pipe->bind_vs_state(pipe, mc->b_vs[0]); + pipe->bind_fs_state(pipe, mc->b_fs[0]); + } + + pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer); + + mc->render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + ); + pipe->set_framebuffer_state(pipe, &mc->render_target); + + mc->textures[3] = past_surface->texture; + mc->textures[4] = future_surface->texture; + pipe->set_sampler_textures(pipe, 5, mc->textures); + pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + return 0; +} + +int vlRenderMacroBlocksMpeg2R16Snorm +( + struct vlRender *render, + struct vlMpeg2MacroBlockBatch *batch, + struct vlSurface *surface +) +{ + struct vlR16SnormMC *mc; + unsigned int i; + + assert(render); + + mc = (struct vlR16SnormMC*)render; + + /*for (i = 0; i < batch->num_macroblocks; ++i) + vlGrabMacroBlock(batch->macroblocks[i]);*/ + + for (i = 0; i < batch->num_macroblocks; ++i) + { + switch (batch->macroblocks[i].mb_type) + { + case vlMacroBlockTypeIntra: + { + vlRenderIMacroBlock + ( + mc, + batch->picture_type, + batch->field_order, + batch->macroblocks[i].mbx, + batch->macroblocks[i].mby, + batch->macroblocks[i].cbp, + batch->macroblocks[i].dct_type, + batch->macroblocks[i].blocks, + surface + ); + break; + } + case vlMacroBlockTypeFwdPredicted: + { + vlRenderPMacroBlock + ( + mc, + batch->picture_type, + batch->field_order, + batch->macroblocks[i].mbx, + batch->macroblocks[i].mby, + batch->macroblocks[i].mo_type, + batch->macroblocks[i].PMV[0][0][0], + batch->macroblocks[i].PMV[0][0][1], + batch->macroblocks[i].PMV[1][0][0], + batch->macroblocks[i].PMV[1][0][1], + batch->macroblocks[i].cbp, + batch->macroblocks[i].dct_type, + batch->macroblocks[i].blocks, + batch->past_surface, + surface + ); + break; + } + case vlMacroBlockTypeBkwdPredicted: + { + vlRenderPMacroBlock + ( + mc, + batch->picture_type, + batch->field_order, + batch->macroblocks[i].mbx, + batch->macroblocks[i].mby, + batch->macroblocks[i].mo_type, + batch->macroblocks[i].PMV[0][1][0], + batch->macroblocks[i].PMV[0][1][1], + batch->macroblocks[i].PMV[1][1][0], + batch->macroblocks[i].PMV[1][1][1], + batch->macroblocks[i].cbp, + batch->macroblocks[i].dct_type, + batch->macroblocks[i].blocks, + batch->future_surface, + surface + ); + break; + } + case vlMacroBlockTypeBiPredicted: + { + vlRenderBMacroBlock + ( + mc, + batch->picture_type, + batch->field_order, + batch->macroblocks[i].mbx, + batch->macroblocks[i].mby, + batch->macroblocks[i].mo_type, + batch->macroblocks[i].PMV[0][0][0], + batch->macroblocks[i].PMV[0][0][1], + batch->macroblocks[i].PMV[1][0][0], + batch->macroblocks[i].PMV[1][0][1], + batch->macroblocks[i].PMV[0][1][0], + batch->macroblocks[i].PMV[0][1][1], + batch->macroblocks[i].PMV[1][1][0], + batch->macroblocks[i].PMV[1][1][1], + batch->macroblocks[i].cbp, + batch->macroblocks[i].dct_type, + batch->macroblocks[i].blocks, + batch->past_surface, + batch->future_surface, + surface + ); + break; + } + default: + assert(0); + } + } + + return 0; +} + +int vlEnd +( + struct vlRender *render +) +{ + assert(render); + + return 0; +} + +int vlDestroy +( + struct vlRender *render +) +{ + struct vlR16SnormMC *mc; + struct pipe_context *pipe; + unsigned int i; + + assert(render); + + mc = (struct vlR16SnormMC*)render; + pipe = mc->pipe; + + for (i = 0; i < 5; ++i) + pipe->delete_sampler_state(pipe, mc->samplers[i]); + + for (i = 0; i < 3; ++i) + pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[i].buffer); + + /* Textures 3 & 4 are not created directly, no need to release them here */ + for (i = 0; i < 3; ++i) + pipe_texture_release(&mc->textures[i]); + + pipe->delete_vs_state(pipe, mc->i_vs); + pipe->delete_fs_state(pipe, mc->i_fs); + + for (i = 0; i < 2; ++i) + { + pipe->delete_vs_state(pipe, mc->p_vs[i]); + pipe->delete_fs_state(pipe, mc->p_fs[i]); + pipe->delete_vs_state(pipe, mc->b_vs[i]); + pipe->delete_fs_state(pipe, mc->b_fs[i]); + } + + pipe->winsys->buffer_destroy(pipe->winsys, mc->vs_const_buf.buffer); + pipe->winsys->buffer_destroy(pipe->winsys, mc->fs_const_buf.buffer); + + free(mc); + + return 0; +} + +/* + * Represents 8 triangles (4 quads, 1 per block) in noormalized coords + * that render a macroblock. + * Need to be scaled to cover mbW*mbH macroblock pixels and translated into + * position on target surface. + */ +const struct vlVertex2f macroblock_verts[24] = +{ + {0.0f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.0f}, + {0.5f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.5f}, + + {0.5f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.0f}, + {1.0f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.5f}, + + {0.0f, 0.5f}, {0.0f, 1.0f}, {0.5f, 0.5f}, + {0.5f, 0.5f}, {0.0f, 1.0f}, {0.5f, 1.0f}, + + {0.5f, 0.5f}, {0.5f, 1.0f}, {1.0f, 0.5f}, + {1.0f, 0.5f}, {0.5f, 1.0f}, {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above for rendering 4 luma blocks arranged + * in a bW*(bH*4) texture. First luma block located at 0,0->bW,bH; second at + * 0,bH->bW,2bH; third at 0,2bH->bW,3bH; fourth at 0,3bH->bW,4bH. + */ +const struct vlVertex2f macroblock_luma_texcoords[24] = +{ + {0.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.0f}, + {1.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.25f}, + + {0.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.25f}, + {1.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.5f}, + + {0.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.5f}, + {1.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.75f}, + + {0.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 0.75f}, + {1.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above for rendering 1 chroma block. + * Straight forward 0,0->1,1 mapping so we can reuse the MB pos vectors. + */ +const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts; + +/* + * Represents texcoords for the above for rendering 2 chroma blocks arranged + * in a bW*(bH*2) texture. First chroma block located at 0,0->bW,bH; second at + * 0,bH->bW,2bH. We can render this with 0,0->1,1 mapping. + * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. + */ +const struct vlVertex2f *macroblock_chroma_422_texcoords = macroblock_verts; + +/* + * Represents texcoords for the above for rendering 4 chroma blocks. + * Same case as 4 luma blocks. + */ +const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texcoords; + +/* + * Used when rendering P and B macroblocks, multiplier is applied to the A channel, + * which is then added to the L channel, then the bias is subtracted from that to + * get back the differential. The differential is then added to the samples from the + * reference surface(s). + */ +const struct vlFragmentShaderConsts fs_consts = +{ + {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, + {0.5f, 2.0f, 0.0f, 0.0f} +}; + +static int vlCreateVertexShaderIMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 50; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->i_vs = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderIMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + */ + for (i = 0; i < 2; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + */ + for (i = 0; i < 3; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul o0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->i_fs = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFramePMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Unused + * decl c3 ; Translation vector to move ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Ref macroblock texcoords + */ + for (i = 0; i < 4; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* add o3, t0, c3 ; Translate rect into position on ref macroblock */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->p_vs[0] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFieldPMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration + ( + &decl, + &tokens[ti], + header, + max_tokens - ti + ); + } + + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Denorm coefficients + * decl c3 ; Translation vector to move top field ref macroblock texcoords into position + * decl c4 ; Translation vector to move bottom field ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Top field ref macroblock texcoords + * decl o4 ; Bottom field ref macroblock texcoords + * decl o5 ; Denormalized vertex pos + */ + for (i = 0; i < 6; i++) + { + decl = vl_decl_output((i == 0 || i == 5) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add t1, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mov o0, t1 ; Move vertex pos to output */ + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + mov o1, i1 ; Move input luma texcoords to output + mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* add o3, t0, c3 ; Translate top field rect into position on ref macroblock + add o4, t0, c4 ; Translate bottom field rect into position on ref macroblock */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul o5, t1, c2 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 5, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->p_vs[1] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFramePMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + */ + for (i = 0; i < 3; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ + for (i = 0; i < 4; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* tex2d t1, i2, s3 ; Read texel from ref macroblock */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 2, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->p_fs[0] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFieldPMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 200; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s3 + * decl i4 ; Denormalized vertex pos + */ + for (i = 0; i < 5; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t4 */ + decl = vl_decl_temps(0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ + for (i = 0; i < 4; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i2, s3 ; Read texel from ref macroblock top field + * tex2d t2, i3, s3 ; Read texel from ref macroblock bottom field + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* XXX: Pos values off by 0.5? */ + /* sub t4, i4.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 4, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* floor t3, t3 ; Get rid of fractional part */ + inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t3, c1.y ; Multiply by 2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->p_fs[1] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFrameBMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Unused + * decl c3 ; Translation vector to move past ref macroblock texcoords into position + * decl c4 ; Unused + * decl c5 ; Translation vector to move future ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 5); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Past ref macroblock texcoords + * decl o4 ; Future ref macroblock texcoords + */ + for (i = 0; i < 5; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* add o3, t0, c3 ; Translate rect into position on past ref macroblock + add o4, t0, c5 ; Translate rect into position on future ref macroblock */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i * 2 + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->b_vs[0] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFieldBMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling vector to scale unit rect to macroblock size + * decl c1 ; Translation vector to move macroblock into position + * decl c2 ; Denorm coefficients + * decl c3 ; Translation vector to move top field past ref macroblock texcoords into position + * decl c4 ; Translation vector to move bottom field past ref macroblock texcoords into position + * decl c5 ; Translation vector to move top field future ref macroblock texcoords into position + * decl c6 ; Translation vector to move bottom field future ref macroblock texcoords into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 6); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma texcoords + * decl o3 ; Top field past ref macroblock texcoords + * decl o4 ; Bottom field past ref macroblock texcoords + * decl o5 ; Top field future ref macroblock texcoords + * decl o6 ; Bottom field future ref macroblock texcoords + * decl o7 ; Denormalized vertex pos + */ + for (i = 0; i < 8; i++) + { + decl = vl_decl_output((i == 0 || i == 7) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* mul t0, i0, c0 ; Scale unit rect to normalized MB size */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add t1, t0, c1 ; Translate rect into position */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mov o0, t1 ; Move vertex pos to output */ + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma texcoords to output + */ + for (i = 1; i < 3; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* + * add o3, t0, c3 ; Translate top field rect into position on past ref macroblock + * add o4, t0, c4 ; Translate bottom field rect into position on past ref macroblock + * add o5, t0, c5 ; Translate top field rect into position on future ref macroblock + * add o6, t0, c6 ; Translate bottom field rect into position on future ref macroblock + */ + for (i = 0; i < 4; ++i) + { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul o7, t1, c2 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 7, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->b_vs[1] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFrameBMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s4 + */ + for (i = 0; i < 4; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constant 1/2 in .x channel to use as weight to blend past and future texels + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t2 */ + decl = vl_decl_temps(0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for past ref surface texture + * decl s4 ; Sampler for future ref surface texture + */ + for (i = 0; i < 5; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i2, s3 ; Read texel from past ref macroblock + * tex2d t2, i3, s4 ; Read texel from future ref macroblock + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->b_fs[0] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFieldBMB +( + struct vlR16SnormMC *mc +) +{ + const unsigned int max_tokens = 200; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0 + * decl i1 ; Texcoords for s1, s2 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s3 + * decl i4 ; Texcoords for s4 + * decl i5 ; Texcoords for s4 + * decl i6 ; Denormalized vertex pos + */ + for (i = 0; i < 7; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels + * ; and for Y-mod-2 top/bottom field selection + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t5 */ + decl = vl_decl_temps(0, 5); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for past ref surface texture + * decl s4 ; Sampler for future ref surface texture + */ + for (i = 0; i < 5; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i1, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* XXX: Pos values off by 0.5? */ + /* sub t4, i6.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 6, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* floor t3, t3 ; Get rid of fractional part */ + inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t3, c1.y ; Multiply by 2 */ + inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i2, s3 ; Read texel from past ref macroblock top field + * tex2d t2, i3, s3 ; Read texel from past ref macroblock bottom field + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t4, i4, s4 ; Read texel from future ref macroblock top field + * tex2d t5, i5, s4 ; Read texel from future ref macroblock bottom field + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 4, TGSI_FILE_SAMPLER, 4); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t2, t3, t4, t5 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->b_fs[1] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +int vlCreateDataBufs +( + struct vlR16SnormMC *mc +) +{ + struct pipe_context *pipe; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + + /* Create our vertex buffer and vertex buffer element */ + mc->vertex_bufs[0].pitch = sizeof(struct vlVertex2f); + mc->vertex_bufs[0].max_index = 23; + mc->vertex_bufs[0].buffer_offset = 0; + mc->vertex_bufs[0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 24 + ); + + mc->vertex_elems[0].src_offset = 0; + mc->vertex_elems[0].vertex_buffer_index = 0; + mc->vertex_elems[0].nr_components = 2; + mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Create our texcoord buffers and texcoord buffer elements */ + for (i = 1; i < 3; ++i) + { + mc->vertex_bufs[i].pitch = sizeof(struct vlVertex2f); + mc->vertex_bufs[i].max_index = 23; + mc->vertex_bufs[i].buffer_offset = 0; + mc->vertex_bufs[i].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 24 + ); + + mc->vertex_elems[i].src_offset = 0; + mc->vertex_elems[i].vertex_buffer_index = i; + mc->vertex_elems[i].nr_components = 2; + mc->vertex_elems[i].src_format = PIPE_FORMAT_R32G32_FLOAT; + } + + /* Fill buffers */ + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, mc->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + macroblock_verts, + sizeof(struct vlVertex2f) * 24 + ); + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, mc->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + macroblock_luma_texcoords, + sizeof(struct vlVertex2f) * 24 + ); + /* TODO: Accomodate 422, 444 */ + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, mc->vertex_bufs[2].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + macroblock_chroma_420_texcoords, + sizeof(struct vlVertex2f) * 24 + ); + + for (i = 0; i < 3; ++i) + pipe->winsys->buffer_unmap(pipe->winsys, mc->vertex_bufs[i].buffer); + + /* Create our constant buffer */ + mc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts); + mc->vs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + mc->vs_const_buf.size + ); + + mc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts); + mc->fs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + mc->fs_const_buf.size + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, mc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + &fs_consts, + sizeof(struct vlFragmentShaderConsts) + ); + + pipe->winsys->buffer_unmap(pipe->winsys, mc->fs_const_buf.buffer); + + return 0; +} + +static int vlInit +( + struct vlR16SnormMC *mc +) +{ + struct pipe_context *pipe; + struct pipe_sampler_state sampler; + struct pipe_texture template; + unsigned int filters[5]; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + + /* For MC we render to textures, which are rounded up to nearest POT */ + mc->viewport.scale[0] = vlRoundUpPOT(mc->video_width); + mc->viewport.scale[1] = vlRoundUpPOT(mc->video_height); + mc->viewport.scale[2] = 1; + mc->viewport.scale[3] = 1; + mc->viewport.translate[0] = 0; + mc->viewport.translate[1] = 0; + mc->viewport.translate[2] = 0; + mc->viewport.translate[3] = 0; + + mc->render_target.width = vlRoundUpPOT(mc->video_width); + mc->render_target.height = vlRoundUpPOT(mc->video_height); + mc->render_target.num_cbufs = 1; + /* FB for MC stage is a vlSurface, set in vlSetRenderSurface() */ + mc->render_target.zsbuf = NULL; + + filters[0] = PIPE_TEX_FILTER_NEAREST; + filters[1] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + filters[2] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + filters[3] = PIPE_TEX_FILTER_LINEAR; + filters[4] = PIPE_TEX_FILTER_LINEAR; + + for (i = 0; i < 5; ++i) + { + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = filters[i]; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = filters[i]; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.shadow_ambient = ;*/ + /*sampler.lod_bias = ;*/ + sampler.min_lod = 0; + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + mc->samplers[i] = pipe->create_sampler_state(pipe, &sampler); + } + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + template.format = PIPE_FORMAT_R16_SNORM; + template.last_level = 0; + template.width[0] = 8; + template.height[0] = 8 * 4; + template.depth[0] = 1; + template.compressed = 0; + pf_get_block(template.format, &template.block); + + mc->textures[0] = pipe->screen->texture_create(pipe->screen, &template); + + if (mc->video_format == vlFormatYCbCr420) + template.height[0] = 8; + else if (mc->video_format == vlFormatYCbCr422) + template.height[0] = 8 * 2; + else if (mc->video_format == vlFormatYCbCr444) + template.height[0] = 8 * 4; + else + assert(0); + + mc->textures[1] = pipe->screen->texture_create(pipe->screen, &template); + mc->textures[2] = pipe->screen->texture_create(pipe->screen, &template); + + /* textures[3] & textures[4] are assigned from vlSurfaces for P and B macroblocks at render time */ + + vlCreateVertexShaderIMB(mc); + vlCreateFragmentShaderIMB(mc); + vlCreateVertexShaderFramePMB(mc); + vlCreateVertexShaderFieldPMB(mc); + vlCreateFragmentShaderFramePMB(mc); + vlCreateFragmentShaderFieldPMB(mc); + vlCreateVertexShaderFrameBMB(mc); + vlCreateVertexShaderFieldBMB(mc); + vlCreateFragmentShaderFrameBMB(mc); + vlCreateFragmentShaderFieldBMB(mc); + vlCreateDataBufs(mc); + + return 0; +} + +int vlCreateR16SNormMC +( + struct pipe_context *pipe, + unsigned int video_width, + unsigned int video_height, + enum vlFormat video_format, + struct vlRender **render +) +{ + struct vlR16SnormMC *mc; + + assert(pipe); + assert(render); + + mc = calloc(1, sizeof(struct vlR16SnormMC)); + + mc->base.vlBegin = &vlBegin; + mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16Snorm; + mc->base.vlEnd = &vlEnd; + mc->base.vlDestroy = &vlDestroy; + mc->pipe = pipe; + mc->video_width = video_width; + mc->video_height = video_height; + + vlInit(mc); + + *render = &mc->base; + + return 0; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h new file mode 100644 index 00000000000..a6eecf05b6c --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h @@ -0,0 +1,18 @@ +#ifndef vl_mc_h +#define vl_mc_h + +#include "vl_types.h" + +struct pipe_context; +struct vlRender; + +int vlCreateR16SNormMC +( + struct pipe_context *pipe, + unsigned int video_width, + unsigned int video_height, + enum vlFormat video_format, + struct vlRender **render +); + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_render.h b/src/gallium/state_trackers/g3dvl/vl_render.h new file mode 100644 index 00000000000..63016b5cbe0 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_render.h @@ -0,0 +1,33 @@ +#ifndef vl_render_h +#define vl_render_h + +#include "vl_types.h" + +struct pipe_surface; + +struct vlRender +{ + int (*vlBegin) + ( + struct vlRender *render + ); + + int (*vlRenderMacroBlocksMpeg2) + ( + struct vlRender *render, + struct vlMpeg2MacroBlockBatch *batch, + struct vlSurface *surface + ); + + int (*vlEnd) + ( + struct vlRender *render + ); + + int (*vlDestroy) + ( + struct vlRender *render + ); +}; + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.c b/src/gallium/state_trackers/g3dvl/vl_screen.c new file mode 100644 index 00000000000..484f63b0d4d --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_screen.c @@ -0,0 +1,115 @@ +#define VL_INTERNAL +#include "vl_screen.h" +#include +#include + +int vlCreateScreen +( + struct vlDisplay *display, + int screen, + struct pipe_screen *pscreen, + struct vlScreen **vl_screen +) +{ + struct vlScreen *scrn; + + assert(display); + assert(pscreen); + assert(vl_screen); + + scrn = calloc(1, sizeof(struct vlScreen)); + + if (!scrn) + return 1; + + scrn->display = display; + scrn->ordinal = screen; + scrn->pscreen = pscreen; + *vl_screen = scrn; + + return 0; +} + +int vlDestroyScreen +( + struct vlScreen *screen +) +{ + assert(screen); + + free(screen); + + return 0; +} + +struct vlDisplay* vlGetDisplay +( + struct vlScreen *screen +) +{ + assert(screen); + + return screen->display; +} + +struct pipe_screen* vlGetPipeScreen +( + struct vlScreen *screen +) +{ + assert(screen); + + return screen->pscreen; +} + +unsigned int vlGetMaxProfiles +( + struct vlScreen *screen +) +{ + assert(screen); + + return vlProfileCount; +} + +int vlQueryProfiles +( + struct vlScreen *screen, + enum vlProfile *profiles +) +{ + assert(screen); + assert(profiles); + + profiles[0] = vlProfileMpeg2Simple; + profiles[1] = vlProfileMpeg2Main; + + return 0; +} + +unsigned int vlGetMaxEntryPoints +( + struct vlScreen *screen +) +{ + assert(screen); + + return vlEntryPointCount; +} + +int vlQueryEntryPoints +( + struct vlScreen *screen, + enum vlProfile profile, + enum vlEntryPoint *entry_points +) +{ + assert(screen); + assert(entry_points); + + entry_points[0] = vlEntryPointIDCT; + entry_points[1] = vlEntryPointMC; + entry_points[2] = vlEntryPointCSC; + + return 0; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.h b/src/gallium/state_trackers/g3dvl/vl_screen.h new file mode 100644 index 00000000000..98f3d429b61 --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_screen.h @@ -0,0 +1,63 @@ +#ifndef vl_screen_h +#define vl_screen_h + +#include "vl_types.h" + +struct pipe_screen; + +#ifdef VL_INTERNAL +struct vlScreen +{ + struct vlDisplay *display; + unsigned int ordinal; + struct pipe_screen *pscreen; +}; +#endif + +int vlCreateScreen +( + struct vlDisplay *display, + int screen, + struct pipe_screen *pscreen, + struct vlScreen **vl_screen +); + +int vlDestroyScreen +( + struct vlScreen *screen +); + +struct vlDisplay* vlGetDisplay +( + struct vlScreen *screen +); + +struct pipe_screen* vlGetPipeScreen +( + struct vlScreen *screen +); + +unsigned int vlGetMaxProfiles +( + struct vlScreen *screen +); + +int vlQueryProfiles +( + struct vlScreen *screen, + enum vlProfile *profiles +); + +unsigned int vlGetMaxEntryPoints +( + struct vlScreen *screen +); + +int vlQueryEntryPoints +( + struct vlScreen *screen, + enum vlProfile profile, + enum vlEntryPoint *entry_points +); + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c index 5f30e23ff8e..51f1721a332 100644 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.c +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.c @@ -13,7 +13,7 @@ struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index decl.Semantic.SemanticIndex = index; decl.DeclarationRange.First = first; decl.DeclarationRange.Last = last; - + return decl; } @@ -27,7 +27,7 @@ struct tgsi_full_declaration vl_decl_interpolated_input ) { struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - + assert ( interpolation == TGSI_INTERPOLATE_CONSTANT || @@ -42,21 +42,21 @@ struct tgsi_full_declaration vl_decl_interpolated_input decl.Declaration.Interpolate = interpolation;; decl.DeclarationRange.First = first; decl.DeclarationRange.Last = last; - + return decl; } struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last) { struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - + decl.Declaration.File = TGSI_FILE_CONSTANT; decl.Declaration.Semantic = 1; decl.Semantic.SemanticName = name; decl.Semantic.SemanticIndex = index; decl.DeclarationRange.First = first; decl.DeclarationRange.Last = last; - + return decl; } @@ -70,7 +70,7 @@ struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int inde decl.Semantic.SemanticIndex = index; decl.DeclarationRange.First = first; decl.DeclarationRange.Last = last; - + return decl; } @@ -82,7 +82,7 @@ struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last decl.Declaration.File = TGSI_FILE_TEMPORARY; decl.DeclarationRange.First = first; decl.DeclarationRange.Last = last; - + return decl; } @@ -94,7 +94,7 @@ struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int l decl.Declaration.File = TGSI_FILE_SAMPLER; decl.DeclarationRange.First = first; decl.DeclarationRange.Last = last; - + return decl; } @@ -108,7 +108,7 @@ struct tgsi_full_instruction vl_inst2 ) { struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - + inst.Instruction.Opcode = opcode; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = dst_file; @@ -116,7 +116,7 @@ struct tgsi_full_instruction vl_inst2 inst.Instruction.NumSrcRegs = 1; inst.FullSrcRegisters[0].SrcRegister.File = src_file; inst.FullSrcRegisters[0].SrcRegister.Index = src_index; - + return inst; } @@ -132,7 +132,7 @@ struct tgsi_full_instruction vl_inst3 ) { struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - + inst.Instruction.Opcode = opcode; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = dst_file; @@ -142,7 +142,7 @@ struct tgsi_full_instruction vl_inst3 inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; inst.FullSrcRegisters[1].SrcRegister.File = src2_file; inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; - + return inst; } @@ -158,7 +158,7 @@ struct tgsi_full_instruction vl_tex ) { struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - + inst.Instruction.Opcode = TGSI_OPCODE_TEX; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = dst_file; @@ -169,7 +169,7 @@ struct tgsi_full_instruction vl_tex inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; inst.FullSrcRegisters[1].SrcRegister.File = src2_file; inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; - + return inst; } @@ -187,7 +187,7 @@ struct tgsi_full_instruction vl_inst4 ) { struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - + inst.Instruction.Opcode = opcode; inst.Instruction.NumDstRegs = 1; inst.FullDstRegisters[0].DstRegister.File = dst_file; @@ -199,18 +199,17 @@ struct tgsi_full_instruction vl_inst4 inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; inst.FullSrcRegisters[2].SrcRegister.File = src3_file; inst.FullSrcRegisters[2].SrcRegister.Index = src3_index; - + return inst; } struct tgsi_full_instruction vl_end(void) { struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - + inst.Instruction.Opcode = TGSI_OPCODE_END; inst.Instruction.NumDstRegs = 0; inst.Instruction.NumSrcRegs = 0; - + return inst; } - diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.h b/src/gallium/state_trackers/g3dvl/vl_shader_build.h index 878d7e2c457..dc615cb1566 100644 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.h +++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.h @@ -59,4 +59,3 @@ struct tgsi_full_instruction vl_inst4 struct tgsi_full_instruction vl_end(void); #endif - diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 1386b1107cc..ffc81221728 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -1,628 +1,177 @@ +#define VL_INTERNAL #include "vl_surface.h" #include #include -#include +#include +#include #include -#include #include #include +#include "vl_screen.h" #include "vl_context.h" -#include "vl_defs.h" +#include "vl_render.h" +#include "vl_csc.h" #include "vl_util.h" -/*#define DO_IDCT*/ - -#ifdef DO_IDCT -static int vlTransformBlock(short *src, short *dst, short bias) +int vlCreateSurface +( + struct vlScreen *screen, + unsigned int width, + unsigned int height, + enum vlFormat format, + struct vlSurface **surface +) { - static const float basis[8][8] = - { - {0.3536, 0.4904, 0.4619, 0.4157, 0.3536, 0.2778, 0.1913, 0.0975}, - {0.3536, 0.4157, 0.1913, -0.0975, -0.3536, -0.4904, -0.4619, -0.2778}, - {0.3536, 0.2778, -0.1913, -0.4904, -0.3536, 0.0975, 0.4619, 0.4157}, - {0.3536, 0.0975, -0.4619, -0.2778, 0.3536, 0.4157, -0.1913, -0.4904}, - {0.3536, -0.0975, -0.4619, 0.2778, 0.3536, -0.4157, -0.1913, 0.4904}, - {0.3536, -0.2778, -0.1913, 0.4904, -0.3536, -0.0975, 0.4619, -0.4157}, - {0.3536, -0.4157, 0.1913, 0.0975, -0.3536, 0.4904, -0.4619, 0.2778}, - {0.3536, -0.4904, 0.4619, -0.4157, 0.3536, -0.2778, 0.1913, -0.0975} - }; - - unsigned int x, y; - short tmp[64]; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - tmp[y * VL_BLOCK_WIDTH + x] = (short) - ( - src[y * VL_BLOCK_WIDTH + 0] * basis[x][0] + - src[y * VL_BLOCK_WIDTH + 1] * basis[x][1] + - src[y * VL_BLOCK_WIDTH + 2] * basis[x][2] + - src[y * VL_BLOCK_WIDTH + 3] * basis[x][3] + - src[y * VL_BLOCK_WIDTH + 4] * basis[x][4] + - src[y * VL_BLOCK_WIDTH + 5] * basis[x][5] + - src[y * VL_BLOCK_WIDTH + 6] * basis[x][6] + - src[y * VL_BLOCK_WIDTH + 7] * basis[x][7] - ); - - for (x = 0; x < VL_BLOCK_WIDTH; ++x) - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - { - dst[y * VL_BLOCK_WIDTH + x] = bias + (short) - ( - tmp[0 * VL_BLOCK_WIDTH + x] * basis[y][0] + - tmp[1 * VL_BLOCK_WIDTH + x] * basis[y][1] + - tmp[2 * VL_BLOCK_WIDTH + x] * basis[y][2] + - tmp[3 * VL_BLOCK_WIDTH + x] * basis[y][3] + - tmp[4 * VL_BLOCK_WIDTH + x] * basis[y][4] + - tmp[5 * VL_BLOCK_WIDTH + x] * basis[y][5] + - tmp[6 * VL_BLOCK_WIDTH + x] * basis[y][6] + - tmp[7 * VL_BLOCK_WIDTH + x] * basis[y][7] - ); - if (dst[y * VL_BLOCK_WIDTH + x] > 255) - dst[y * VL_BLOCK_WIDTH + x] = 255; - else if (bias > 0 && dst[y * VL_BLOCK_WIDTH + x] < 0) - dst[y * VL_BLOCK_WIDTH + x] = 0; - } - return 0; -} -#endif + struct vlSurface *sfc; + struct pipe_texture template; -static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) -{ - unsigned int y; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - memcpy - ( - dst + y * dst_pitch, - src + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - - return 0; -} + assert(screen); + assert(surface); -static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) -{ - unsigned int y; - - for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) - memcpy - ( - dst + y * dst_pitch * 2, - src + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - - dst += VL_BLOCK_HEIGHT * dst_pitch; - - for (; y < VL_BLOCK_HEIGHT; ++y) - memcpy - ( - dst + y * dst_pitch * 2, - src + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - - return 0; -} + sfc = calloc(1, sizeof(struct vlSurface)); -static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) -{ - unsigned int y; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - memset - ( - dst + y * dst_pitch, - 0, - VL_BLOCK_WIDTH * 2 - ); - - return 0; -} + if (!sfc) + return 1; -static int vlGrabBlocks -( - struct VL_CONTEXT *context, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - enum VL_SAMPLE_TYPE sample_type, - short *blocks -) -{ - struct pipe_surface *tex_surface; - short *texels; - unsigned int tex_pitch; - unsigned int tb, sb = 0; - - assert(context); - assert(blocks); - - tex_surface = context->pipe->screen->get_tex_surface - ( - context->pipe->screen, - context->states.mc.textures[0], - 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE - ); - - texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); - tex_pitch = tex_surface->stride / tex_surface->block.size; - - for (tb = 0; tb < 4; ++tb) - { - if ((coded_block_pattern >> (5 - tb)) & 1) - { - short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; - -#ifdef DO_IDCT - vlTransformBlock(cur_block, cur_block, sample_type == VL_FULL_SAMPLE ? 128 : 0); -#endif - - if (dct_type == VL_DCT_FRAME_CODED) - vlGrabFrameCodedBlock - ( - cur_block, - texels + tb * tex_pitch * VL_BLOCK_HEIGHT, - tex_pitch - ); - else - vlGrabFieldCodedBlock - ( - cur_block, - texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch, - tex_pitch - ); - - ++sb; - } - else - vlGrabNoBlock(texels + tb * tex_pitch * VL_BLOCK_HEIGHT, tex_pitch); - } - - pipe_surface_unmap(tex_surface); - - /* TODO: Implement 422, 444 */ - for (tb = 0; tb < 2; ++tb) - { - tex_surface = context->pipe->screen->get_tex_surface - ( - context->pipe->screen, - context->states.mc.textures[tb + 1], - 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE - ); - - texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); - tex_pitch = tex_surface->stride / tex_surface->block.size; - - if ((coded_block_pattern >> (1 - tb)) & 1) - { - short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; - -#ifdef DO_IDCT - vlTransformBlock(cur_block, cur_block, sample_type == VL_FULL_SAMPLE ? 128 : 0); -#endif - - vlGrabFrameCodedBlock - ( - cur_block, - texels, - tex_pitch - ); - - ++sb; - } - else - vlGrabNoBlock(texels, tex_pitch); - - pipe_surface_unmap(tex_surface); - } - - return 0; -} + sfc->screen = screen; + sfc->width = width; + sfc->height = height; + sfc->format = format; -int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface) -{ - struct pipe_context *pipe; - struct pipe_texture template; - struct VL_SURFACE *sfc; - - assert(context); - assert(surface); - - pipe = context->pipe; - - sfc = calloc(1, sizeof(struct VL_SURFACE)); - - sfc->context = context; - sfc->width = vlRoundUpPOT(context->video_width); - sfc->height = vlRoundUpPOT(context->video_height); - sfc->format = context->video_format; - memset(&template, 0, sizeof(struct pipe_texture)); template.target = PIPE_TEXTURE_2D; template.format = PIPE_FORMAT_A8R8G8B8_UNORM; template.last_level = 0; - template.width[0] = sfc->width; - template.height[0] = sfc->height; + template.width[0] = vlRoundUpPOT(sfc->width); + template.height[0] = vlRoundUpPOT(sfc->height); template.depth[0] = 1; template.compressed = 0; pf_get_block(template.format, &template.block); - /* XXX: Needed? */ template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; - - sfc->texture = pipe->screen->texture_create(pipe->screen, &template); - + + sfc->texture = vlGetPipeScreen(screen)->texture_create(vlGetPipeScreen(screen), &template); + *surface = sfc; - + return 0; } -int vlDestroySurface(struct VL_SURFACE *surface) +int vlDestroySurface +( + struct vlSurface *surface +) { assert(surface); + pipe_texture_release(&surface->texture); free(surface); - + return 0; } -int vlRenderIMacroBlock +int vlRenderMacroBlocksMpeg2 ( - enum VL_PICTURE picture_type, - enum VL_FIELD_ORDER field_order, - unsigned int mbx, - unsigned int mby, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - short *blocks, - struct VL_SURFACE *surface + struct vlMpeg2MacroBlockBatch *batch, + struct vlSurface *surface ) { - struct pipe_context *pipe; - struct VL_MC_VS_CONSTS *vs_consts; - - assert(blocks); + assert(batch); assert(surface); - - /* TODO: Implement interlaced rendering */ - if (picture_type != VL_FRAME_PICTURE) - return 0; - - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_FULL_SAMPLE, blocks); - - pipe = surface->context->pipe; - - vs_consts = pipe->winsys->buffer_map - ( - pipe->winsys, - surface->context->states.mc.vs_const_buf.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE - ); - - vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; - vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; - vs_consts->scale.z = 1.0f; - vs_consts->scale.w = 1.0f; - vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; - vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; - vs_consts->mb_pos_trans.z = 0.0f; - vs_consts->mb_pos_trans.w = 0.0f; - - pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); - - surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface + + surface->context->render->vlBegin(surface->context->render); + + surface->context->render->vlRenderMacroBlocksMpeg2 ( - pipe->screen, - surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + surface->context->render, + batch, + surface ); - pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); - pipe->set_sampler_textures(pipe, 3, surface->context->states.mc.textures); - pipe->bind_sampler_states(pipe, 3, (void**)surface->context->states.mc.samplers); - pipe->bind_vs_state(pipe, surface->context->states.mc.i_vs); - pipe->bind_fs_state(pipe, surface->context->states.mc.i_fs); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); - + + surface->context->render->vlEnd(surface->context->render); + return 0; } -int vlRenderPMacroBlock +int vlPutPicture ( - enum VL_PICTURE picture_type, - enum VL_FIELD_ORDER field_order, - unsigned int mbx, - unsigned int mby, - enum VL_MC_TYPE mc_type, - struct VL_MOTION_VECTOR *motion_vector, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - short *blocks, - struct VL_SURFACE *ref_surface, - struct VL_SURFACE *surface + struct vlSurface *surface, + vlNativeDrawable drawable, + int srcx, + int srcy, + int srcw, + int srch, + int destx, + int desty, + int destw, + int desth, + enum vlPictureType picture_type ) { + struct vlCSC *csc; struct pipe_context *pipe; - struct VL_MC_VS_CONSTS *vs_consts; - - assert(motion_vectors); - assert(blocks); - assert(ref_surface); + assert(surface); - - /* TODO: Implement interlaced rendering */ - if (picture_type != VL_FRAME_PICTURE) - return 0; - /* TODO: Implement other MC types */ - if (mc_type != VL_FRAME_MC && mc_type != VL_FIELD_MC) - return 0; - - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); - + assert(surface->context); + + csc = surface->context->csc; pipe = surface->context->pipe; - - vs_consts = pipe->winsys->buffer_map + + csc->vlResizeFrameBuffer(csc, destw, desth); + + csc->vlBegin(csc); + + csc->vlPutPicture ( - pipe->winsys, - surface->context->states.mc.vs_const_buf.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE + csc, + surface, + srcx, + srcy, + srcw, + srch, + destx, + desty, + destw, + desth, + picture_type ); - - vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; - vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; - vs_consts->scale.z = 1.0f; - vs_consts->scale.w = 1.0f; - vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; - vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; - vs_consts->mb_pos_trans.z = 0.0f; - vs_consts->mb_pos_trans.w = 0.0f; - vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector->top_field.x * 0.5f) / (float)surface->width; - vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector->top_field.y * 0.5f) / (float)surface->height; - vs_consts->mb_tc_trans[0].top_field.z = 0.0f; - vs_consts->mb_tc_trans[0].top_field.w = 0.0f; - - if (mc_type == VL_FIELD_MC) - { - vs_consts->denorm.x = (float)surface->width; - vs_consts->denorm.y = (float)surface->height; - - vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector->bottom_field.x * 0.5f) / (float)surface->width; - vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector->bottom_field.y * 0.5f) / (float)surface->height; - vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f; - vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f; - - pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs[1]); - pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs[1]); - } - else - { - pipe->bind_vs_state(pipe, surface->context->states.mc.p_vs[0]); - pipe->bind_fs_state(pipe, surface->context->states.mc.p_fs[0]); - } - - pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); - - surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface + + csc->vlEnd(csc); + + pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + bind_pipe_drawable(pipe, drawable); + /* TODO: Need to take destx, desty into consideration */ + pipe->winsys->flush_frontbuffer ( - pipe->screen, - surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + pipe->winsys, + csc->vlGetFrameBuffer(csc), + pipe->priv ); - pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); - - surface->context->states.mc.textures[3] = ref_surface->texture; - pipe->set_sampler_textures(pipe, 4, surface->context->states.mc.textures); - pipe->bind_sampler_states(pipe, 4, (void**)surface->context->states.mc.samplers); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); - + return 0; } -int vlRenderBMacroBlock +struct vlScreen* vlSurfaceGetScreen ( - enum VL_PICTURE picture_type, - enum VL_FIELD_ORDER field_order, - unsigned int mbx, - unsigned int mby, - enum VL_MC_TYPE mc_type, - struct VL_MOTION_VECTOR *motion_vector, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - short *blocks, - struct VL_SURFACE *past_surface, - struct VL_SURFACE *future_surface, - struct VL_SURFACE *surface + struct vlSurface *surface ) { - struct pipe_context *pipe; - struct VL_MC_VS_CONSTS *vs_consts; - - assert(motion_vectors); - assert(blocks); - assert(ref_surface); assert(surface); - - /* TODO: Implement interlaced rendering */ - if (picture_type != VL_FRAME_PICTURE) - return 0; - /* TODO: Implement other MC types */ - if (mc_type != VL_FRAME_MC && mc_type != VL_FIELD_MC) - return 0; - - vlGrabBlocks(surface->context, coded_block_pattern, dct_type, VL_DIFFERENCE_SAMPLE, blocks); - - pipe = surface->context->pipe; - - vs_consts = pipe->winsys->buffer_map - ( - pipe->winsys, - surface->context->states.mc.vs_const_buf.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE - ); - - vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->width; - vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->height; - vs_consts->scale.z = 1.0f; - vs_consts->scale.w = 1.0f; - vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->width; - vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->height; - vs_consts->mb_pos_trans.z = 0.0f; - vs_consts->mb_pos_trans.w = 0.0f; - vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[0].top_field.x * 0.5f) / (float)surface->width; - vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[0].top_field.y * 0.5f) / (float)surface->height; - vs_consts->mb_tc_trans[0].top_field.z = 0.0f; - vs_consts->mb_tc_trans[0].top_field.w = 0.0f; - vs_consts->mb_tc_trans[1].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[1].top_field.x * 0.5f) / (float)surface->width; - vs_consts->mb_tc_trans[1].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[1].top_field.y * 0.5f) / (float)surface->height; - vs_consts->mb_tc_trans[1].top_field.z = 0.0f; - vs_consts->mb_tc_trans[1].top_field.w = 0.0f; - - if (mc_type == VL_FIELD_MC) - { - vs_consts->denorm.x = (float)surface->width; - vs_consts->denorm.y = (float)surface->height; - - vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[0].bottom_field.x * 0.5f) / (float)surface->width; - vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[0].bottom_field.y * 0.5f) / (float)surface->height; - vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f; - vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f; - vs_consts->mb_tc_trans[1].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + motion_vector[1].bottom_field.x * 0.5f) / (float)surface->width; - vs_consts->mb_tc_trans[1].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + motion_vector[1].bottom_field.y * 0.5f) / (float)surface->height; - vs_consts->mb_tc_trans[1].bottom_field.z = 0.0f; - vs_consts->mb_tc_trans[1].bottom_field.w = 0.0f; - - pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs[1]); - pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs[1]); - } - else - { - pipe->bind_vs_state(pipe, surface->context->states.mc.b_vs[0]); - pipe->bind_fs_state(pipe, surface->context->states.mc.b_fs[0]); - } - - pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.mc.vs_const_buf.buffer); - - surface->context->states.mc.render_target.cbufs[0] = pipe->screen->get_tex_surface - ( - pipe->screen, - surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE - ); - pipe->set_framebuffer_state(pipe, &surface->context->states.mc.render_target); - - surface->context->states.mc.textures[3] = past_surface->texture; - surface->context->states.mc.textures[4] = future_surface->texture; - pipe->set_sampler_textures(pipe, 5, surface->context->states.mc.textures); - pipe->bind_sampler_states(pipe, 5, (void**)surface->context->states.mc.samplers); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); - - return 0; + + return surface->screen; } -int vlPutSurface +struct vlContext* vlBindToContext ( - struct VL_SURFACE *surface, - Drawable drawable, - unsigned int srcx, - unsigned int srcy, - unsigned int srcw, - unsigned int srch, - unsigned int destx, - unsigned int desty, - unsigned int destw, - unsigned int desth, - enum VL_PICTURE picture_type + struct vlSurface *surface, + struct vlContext *context ) { - unsigned int create_fb = 0; - struct pipe_context *pipe; - struct VL_CSC_VS_CONSTS *vs_consts; - + struct vlContext *old; + assert(surface); - - pipe = surface->context->pipe; - - if (!surface->context->states.csc.framebuffer.cbufs[0]) - create_fb = 1; - else if - ( - surface->context->states.csc.framebuffer.width != destw || - surface->context->states.csc.framebuffer.height != desth - ) - { - pipe->winsys->surface_release - ( - pipe->winsys, - &surface->context->states.csc.framebuffer.cbufs[0] - ); - - create_fb = 1; - } - - if (create_fb) - { - surface->context->states.csc.viewport.scale[0] = destw; - surface->context->states.csc.viewport.scale[1] = desth; - surface->context->states.csc.viewport.scale[2] = 1; - surface->context->states.csc.viewport.scale[3] = 1; - surface->context->states.csc.viewport.translate[0] = 0; - surface->context->states.csc.viewport.translate[1] = 0; - surface->context->states.csc.viewport.translate[2] = 0; - surface->context->states.csc.viewport.translate[3] = 0; - - surface->context->states.csc.framebuffer.width = destw; - surface->context->states.csc.framebuffer.height = desth; - surface->context->states.csc.framebuffer.cbufs[0] = pipe->winsys->surface_alloc(pipe->winsys); - pipe->winsys->surface_alloc_storage - ( - pipe->winsys, - surface->context->states.csc.framebuffer.cbufs[0], - destw, - desth, - PIPE_FORMAT_A8R8G8B8_UNORM, - /* XXX: SoftPipe doesn't change GPU usage to CPU like it does for textures */ - PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE, - 0 - ); - } - - vlEndRender(surface->context); - - vs_consts = pipe->winsys->buffer_map - ( - pipe->winsys, - surface->context->states.csc.vs_const_buf.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE - ); - - vs_consts->src_scale.x = srcw / (float)surface->width; - vs_consts->src_scale.y = srch / (float)surface->height; - vs_consts->src_scale.z = 1; - vs_consts->src_scale.w = 1; - vs_consts->src_trans.x = srcx / (float)surface->width; - vs_consts->src_trans.y = srcy / (float)surface->height; - vs_consts->src_trans.z = 0; - vs_consts->src_trans.w = 0; - - pipe->winsys->buffer_unmap(pipe->winsys, surface->context->states.csc.vs_const_buf.buffer); - - pipe->set_sampler_textures(pipe, 1, &surface->texture); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); - pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); - bind_pipe_drawable(pipe, drawable); - /* TODO: Need to take destx, desty into consideration */ - pipe->winsys->flush_frontbuffer - ( - pipe->winsys, - surface->context->states.csc.framebuffer.cbufs[0], - pipe->priv - ); - - vlBeginRender(surface->context); - - return 0; -} + old = surface->context; + surface->context = context; + + return old; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h index 9f56b77e1e7..b975e131fa6 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.h +++ b/src/gallium/state_trackers/g3dvl/vl_surface.h @@ -1,81 +1,66 @@ #ifndef vl_surface_h #define vl_surface_h -#include #include "vl_types.h" +#ifdef VL_INTERNAL struct pipe_texture; -struct VL_SURFACE +struct vlSurface { - struct VL_CONTEXT *context; + struct vlScreen *screen; + struct vlContext *context; unsigned int width; unsigned int height; - enum VL_FORMAT format; + enum vlFormat format; struct pipe_texture *texture; }; +#endif -int vlCreateSurface(struct VL_CONTEXT *context, struct VL_SURFACE **surface); +int vlCreateSurface +( + struct vlScreen *screen, + unsigned int width, + unsigned int height, + enum vlFormat format, + struct vlSurface **surface +); -int vlDestroySurface(struct VL_SURFACE *surface); +int vlDestroySurface +( + struct vlSurface *surface +); -int vlRenderIMacroBlock +int vlRenderMacroBlocksMpeg2 ( - enum VL_PICTURE picture_type, - enum VL_FIELD_ORDER field_order, - unsigned int mbx, - unsigned int mby, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - short *blocks, - struct VL_SURFACE *surface + struct vlMpeg2MacroBlockBatch *batch, + struct vlSurface *surface ); -int vlRenderPMacroBlock +int vlPutPicture ( - enum VL_PICTURE picture_type, - enum VL_FIELD_ORDER field_order, - unsigned int mbx, - unsigned int mby, - enum VL_MC_TYPE mc_type, - struct VL_MOTION_VECTOR *motion_vector, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - short *blocks, - struct VL_SURFACE *ref_surface, - struct VL_SURFACE *surface + struct vlSurface *surface, + vlNativeDrawable drawable, + int srcx, + int srcy, + int srcw, + int srch, + int destx, + int desty, + int destw, + int desth, + enum vlPictureType picture_type ); -int vlRenderBMacroBlock +struct vlScreen* vlSurfaceGetScreen ( - enum VL_PICTURE picture_type, - enum VL_FIELD_ORDER field_order, - unsigned int mbx, - unsigned int mby, - enum VL_MC_TYPE mc_type, - struct VL_MOTION_VECTOR *motion_vector, - unsigned int coded_block_pattern, - enum VL_DCT_TYPE dct_type, - short *blocks, - struct VL_SURFACE *past_surface, - struct VL_SURFACE *future_surface, - struct VL_SURFACE *surface + struct vlSurface *surface ); -int vlPutSurface +struct vlContext* vlBindToContext ( - struct VL_SURFACE *surface, - Drawable drawable, - unsigned int srcx, - unsigned int srcy, - unsigned int srcw, - unsigned int srch, - unsigned int destx, - unsigned int desty, - unsigned int destw, - unsigned int desth, - enum VL_PICTURE picture_type + struct vlSurface *surface, + struct vlContext *context ); #endif - diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h index 4d210c9e0aa..504ba8ac81a 100644 --- a/src/gallium/state_trackers/g3dvl/vl_types.h +++ b/src/gallium/state_trackers/g3dvl/vl_types.h @@ -1,102 +1,106 @@ #ifndef vl_types_h #define vl_types_h -enum VL_FORMAT -{ - VL_FORMAT_YCBCR_420, - VL_FORMAT_YCBCR_422, - VL_FORMAT_YCBCR_444 -}; +#if 1 /*#ifdef X11*/ +#include -enum VL_PICTURE -{ - VL_TOP_FIELD, - VL_BOTTOM_FIELD, - VL_FRAME_PICTURE -}; +typedef Display* vlNativeDisplay; +typedef Drawable vlNativeDrawable; +#endif + +struct vlDisplay; +struct vlScreen; +struct vlContext; +struct vlSurface; -enum VL_FIELD_ORDER +enum vlProfile { - VL_FIELD_FIRST, - VL_FIELD_SECOND + vlProfileMpeg2Simple, + vlProfileMpeg2Main, + + vlProfileCount }; -enum VL_DCT_TYPE +enum vlEntryPoint { - VL_DCT_FIELD_CODED, - VL_DCT_FRAME_CODED + vlEntryPointIDCT, + vlEntryPointMC, + vlEntryPointCSC, + + vlEntryPointCount }; -enum VL_SAMPLE_TYPE +enum vlFormat { - VL_FULL_SAMPLE, - VL_DIFFERENCE_SAMPLE + vlFormatYCbCr420, + vlFormatYCbCr422, + vlFormatYCbCr444 }; -enum VL_MC_TYPE +enum vlPictureType { - VL_FIELD_MC, - VL_FRAME_MC, - VL_DUAL_PRIME_MC, - VL_16x8_MC = VL_FRAME_MC + vlPictureTypeTopField, + vlPictureTypeBottomField, + vlPictureTypeFrame }; -struct VL_VERTEX4F +enum vlMotionType { - float x, y, z, w; + vlMotionTypeField, + vlMotionTypeFrame, + vlMotionTypeDualPrime, + vlMotionType16x8 }; -struct VL_VERTEX2F +enum vlFieldOrder { - float x, y; + vlFieldOrderFirst, + vlFieldOrderSecond }; -struct VL_TEXCOORD2F +enum vlDCTType { - float s, t; + vlDCTTypeFrameCoded, + vlDCTTypeFieldCoded }; -struct VL_MC_VS_CONSTS +struct vlVertex2f { - struct VL_VERTEX4F scale; - struct VL_VERTEX4F mb_pos_trans; - struct VL_VERTEX4F denorm; - struct - { - struct VL_VERTEX4F top_field; - struct VL_VERTEX4F bottom_field; - } mb_tc_trans[2]; + float x, y; }; -struct VL_MC_FS_CONSTS +struct vlVertex4f { - struct VL_VERTEX4F multiplier; - struct VL_VERTEX4F bias; - struct VL_VERTEX4F y_divider; + float x, y, z, w; }; -struct VL_CSC_VS_CONSTS +enum vlMacroBlockType { - struct VL_VERTEX4F src_scale; - struct VL_VERTEX4F src_trans; + vlMacroBlockTypeIntra, + vlMacroBlockTypeFwdPredicted, + vlMacroBlockTypeBkwdPredicted, + vlMacroBlockTypeBiPredicted }; -struct VL_CSC_FS_CONSTS +struct vlMpeg2MacroBlock { - struct VL_VERTEX4F bias; - float matrix[16]; + unsigned int mbx, mby; + enum vlMacroBlockType mb_type; + enum vlMotionType mo_type; + enum vlDCTType dct_type; + int PMV[2][2][2]; + unsigned int cbp; + short *blocks; }; -struct VL_MOTION_VECTOR +struct vlMpeg2MacroBlockBatch { - struct - { - int x, y; - } top_field, bottom_field; + struct vlSurface *past_surface; + struct vlSurface *future_surface; + enum vlPictureType picture_type; + enum vlFieldOrder field_order; + unsigned int num_macroblocks; + struct vlMpeg2MacroBlock *macroblocks; }; -struct VL_CONTEXT; -struct VL_SURFACE; - #endif - diff --git a/src/gallium/state_trackers/g3dvl/vl_util.c b/src/gallium/state_trackers/g3dvl/vl_util.c index 2421ae22101..50aa9af66f2 100644 --- a/src/gallium/state_trackers/g3dvl/vl_util.c +++ b/src/gallium/state_trackers/g3dvl/vl_util.c @@ -4,14 +4,13 @@ unsigned int vlRoundUpPOT(unsigned int x) { unsigned int i; - + assert(x > 0); - + --x; - + for (i = 1; i < sizeof(unsigned int) * 8; i <<= 1) x |= x >> i; - + return x + 1; } - diff --git a/src/gallium/state_trackers/g3dvl/vl_util.h b/src/gallium/state_trackers/g3dvl/vl_util.h index e4b72c4f870..bc98e79df47 100644 --- a/src/gallium/state_trackers/g3dvl/vl_util.h +++ b/src/gallium/state_trackers/g3dvl/vl_util.h @@ -4,4 +4,3 @@ unsigned int vlRoundUpPOT(unsigned int x); #endif - diff --git a/src/libXvMC/block.c b/src/libXvMC/block.c index deca305bdcf..328b035576b 100644 --- a/src/libXvMC/block.c +++ b/src/libXvMC/block.c @@ -2,83 +2,78 @@ #include #include #include +#include +#include #include -/* - * XvMC defines 64 element blocks (8x8 elements). - * Elements are 8 bits when they represent color values, - * 9 bits when they reprecent DCT coefficients, we - * store them in 2 bytes in either case. - */ #define BLOCK_SIZE (64 * 2) Status XvMCCreateBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks) { - struct vl_context *vl_ctx; - + struct vlContext *vl_ctx; + assert(display); - + if (!context) return XvMCBadContext; if (num_blocks == 0) return BadValue; - + assert(blocks); - + vl_ctx = context->privData; - assert(display == vl_ctx->display); + assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); blocks->context_id = context->context_id; blocks->num_blocks = num_blocks; blocks->blocks = malloc(BLOCK_SIZE * num_blocks); /* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */ blocks->privData = display; - + return Success; } Status XvMCDestroyBlocks(Display *display, XvMCBlockArray *blocks) -{ +{ assert(display); assert(blocks); assert(display == blocks->privData); free(blocks->blocks); - + return Success; } Status XvMCCreateMacroBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks) { - struct vl_context *vl_ctx; - + struct vlContext *vl_ctx; + assert(display); - + if (!context) return XvMCBadContext; if (num_blocks == 0) return BadValue; - + assert(blocks); - + vl_ctx = context->privData; - assert(display == vl_ctx->display); - + assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); + blocks->context_id = context->context_id; blocks->num_blocks = num_blocks; blocks->macro_blocks = malloc(sizeof(XvMCMacroBlock) * num_blocks); /* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */ blocks->privData = display; - + return Success; } Status XvMCDestroyMacroBlocks(Display *display, XvMCMacroBlockArray *blocks) -{ +{ assert(display); assert(blocks); assert(display == blocks->privData); free(blocks->macro_blocks); - + return Success; } - diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c index 9cf654d6bb2..760e012d1a8 100644 --- a/src/libXvMC/context.c +++ b/src/libXvMC/context.c @@ -1,10 +1,23 @@ #include #include #include +#include +#include +#include #include #include -static Status Validate(Display *display, XvPortID port, int surface_type_id, unsigned int width, unsigned int height, int flags, int *chroma_format) +static Status Validate +( + Display *display, + XvPortID port, + int surface_type_id, + unsigned int width, + unsigned int height, + int flags, + int *chroma_format, + int *mc_type +) { unsigned int found_port = 0; unsigned int found_surface = 0; @@ -14,13 +27,13 @@ static Status Validate(Display *display, XvPortID port, int surface_type_id, uns unsigned int max_width, max_height; Status ret; unsigned int i, j, k; - + assert(display && chroma_format); - + ret = XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info); if (ret != Success) return ret; - + /* Scan through all adaptors looking for this port and surface */ for (i = 0; i < num_adaptors && !found_port; ++i) { @@ -31,10 +44,10 @@ static Status Validate(Display *display, XvPortID port, int surface_type_id, uns if (adaptor_info[i].base_id + j == port) { XvMCSurfaceInfo *surface_info; - + found_port = 1; surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types); - + if (surface_info) { for (k = 0; k < num_types && !found_surface; ++k) @@ -45,9 +58,10 @@ static Status Validate(Display *display, XvPortID port, int surface_type_id, uns max_width = surface_info[k].max_width; max_height = surface_info[k].max_height; *chroma_format = surface_info[k].chroma_format; + *mc_type = surface_info[k].mc_type; } } - + XFree(surface_info); } else @@ -58,9 +72,9 @@ static Status Validate(Display *display, XvPortID port, int surface_type_id, uns } } } - + XvFreeAdaptorInfo(adaptor_info); - + if (!found_port) return XvBadPort; if (!found_surface) @@ -69,60 +83,86 @@ static Status Validate(Display *display, XvPortID port, int surface_type_id, uns return BadValue; if (flags != XVMC_DIRECT && flags != 0) return BadValue; - + return Success; } -static enum VL_FORMAT FormatToVL(int xvmc_format) +static enum vlProfile ProfileToVL(int xvmc_profile) +{ + if (xvmc_profile & XVMC_MPEG_1) + assert(0); + else if (xvmc_profile & XVMC_MPEG_2) + return vlProfileMpeg2Main; + else if (xvmc_profile & XVMC_H263) + assert(0); + else if (xvmc_profile & XVMC_MPEG_4) + assert(0); + else + assert(0); + + return -1; +} + +static enum vlEntryPoint EntryToVL(int xvmc_entry) +{ + return xvmc_entry & XVMC_IDCT ? vlEntryPointIDCT : vlEntryPointMC; +} + +static enum vlFormat FormatToVL(int xvmc_format) { - enum VL_FORMAT vl_format; - switch (xvmc_format) { case XVMC_CHROMA_FORMAT_420: - { - vl_format = VL_FORMAT_YCBCR_420; - break; - } + return vlFormatYCbCr420; case XVMC_CHROMA_FORMAT_422: - { - vl_format = VL_FORMAT_YCBCR_422; - break; - } + return vlFormatYCbCr422; case XVMC_CHROMA_FORMAT_444: - { - vl_format = VL_FORMAT_YCBCR_444; - break; - } + return vlFormatYCbCr444; default: assert(0); } - - return vl_format; + + return -1; } Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, int width, int height, int flags, XvMCContext *context) { int chroma_format; + int mc_type; Status ret; - struct VL_CONTEXT *vl_ctx; + struct vlDisplay *vl_dpy; + struct vlScreen *vl_scrn; + struct vlContext *vl_ctx; struct pipe_context *pipe; - + assert(display); - + if (!context) return XvMCBadContext; - - ret = Validate(display, port, surface_type_id, width, height, flags, &chroma_format); + + ret = Validate(display, port, surface_type_id, width, height, flags, &chroma_format, &mc_type); if (ret != Success) return ret; - + + /* XXX: Assumes default screen, should check which screen port is on */ pipe = create_pipe_context(display, XDefaultScreen(display)); - + assert(pipe); - - vlCreateContext(display, pipe, width, height, FormatToVL(chroma_format), &vl_ctx); - + + vlCreateDisplay(display, &vl_dpy); + vlCreateScreen(vl_dpy, XDefaultScreen(display), pipe->screen, &vl_scrn); + vlCreateContext + ( + vl_scrn, + pipe, + width, + height, + FormatToVL(chroma_format), + ProfileToVL(mc_type), + EntryToVL(mc_type), + &vl_ctx + ); + context->context_id = XAllocID(display); context->surface_type_id = surface_type_id; context->width = width; @@ -130,89 +170,27 @@ Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, i context->flags = flags; context->port = port; context->privData = vl_ctx; - + return Success; } Status XvMCDestroyContext(Display *display, XvMCContext *context) { - struct VL_CONTEXT *vl_ctx; + struct vlContext *vl_ctx; struct pipe_context *pipe; - + assert(display); - + if (!context) return XvMCBadContext; - + vl_ctx = context->privData; - - assert(display == vl_ctx->display); - - pipe = vl_ctx->pipe; - vlDestroyContext(vl_ctx); - destroy_pipe_context(pipe); - - return Success; -} -/* XXX: The following are here temporarily, need to be implemented in the DDX driver */ -/* TODO: Figure out which of these need to be in DDX, which are better off in DDX, which can stay */ + assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); -Bool XvMCQueryExtension(Display *display, int *event_base, int *err_base) -{ - *event_base = 0; - *err_base = 0; - - return True; -} + pipe = vlGetPipeContext(vl_ctx); + vlDestroyContext(vl_ctx); + destroy_pipe_context(pipe); -Status XvMCQueryVersion(Display *display, int *major, int *minor) -{ - *major = 1; - *minor = 0; - return Success; } - -XvMCSurfaceInfo* XvMCListSurfaceTypes(Display *display, XvPortID port, int *num) -{ - XvMCSurfaceInfo *surface_info = calloc(1, sizeof(XvMCSurfaceInfo)); - - *num = 1; - - surface_info->chroma_format = XVMC_CHROMA_FORMAT_420; - surface_info->max_width = 2048; - surface_info->max_height = 2048; - surface_info->subpicture_max_width = 2048; - surface_info->subpicture_max_height = 2048; - surface_info->mc_type = XVMC_IDCT | XVMC_MPEG_2; - surface_info->surface_type_id = 123; /* FIXME: XAllocID(display)*/; - surface_info->flags = XVMC_INTRA_UNSIGNED | XVMC_SUBPICTURE_INDEPENDENT_SCALING | XVMC_BACKEND_SUBPICTURE; - - return surface_info; -} - -XvImageFormatValues* XvMCListSubpictureTypes(Display* display, XvPortID port, int surface_type_id, int *count_return) -{ - XvImageFormatValues *image_formats = calloc(1, sizeof(XvImageFormatValues)); - - *count_return = 1; - - image_formats[0].id = 123; - image_formats[0].type = XvRGB; - image_formats[0].byte_order = LSBFirst; - image_formats[0].bits_per_pixel = 8; - image_formats[0].format = XvPacked; - image_formats[0].num_planes = 1; - image_formats[0].depth = 8; - image_formats[0].red_mask = 0x0000FF; - image_formats[0].green_mask = 0x00FF00; - image_formats[0].blue_mask = 0xFF0000; - image_formats[0].component_order[0] = 'R'; - image_formats[0].component_order[0] = 'G'; - image_formats[0].component_order[0] = 'B'; - image_formats[0].scanline_order = XvTopToBottom; - - return image_formats; -} - diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index 1c07220e844..038befc297b 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -1,90 +1,96 @@ #include #include #include +#include +#include #include #include +#include -static enum VL_PICTURE PictureToVL(int xvmc_pic) +static enum vlMacroBlockType TypeToVL(int xvmc_mb_type) +{ + if (xvmc_mb_type & XVMC_MB_TYPE_INTRA) + return vlMacroBlockTypeIntra; + if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_FORWARD) + return vlMacroBlockTypeFwdPredicted; + if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_BACKWARD) + return vlMacroBlockTypeBkwdPredicted; + if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) + return vlMacroBlockTypeBiPredicted; + + assert(0); + + return -1; +} + +static enum vlPictureType PictureToVL(int xvmc_pic) { - enum VL_PICTURE vl_pic; - switch (xvmc_pic) { case XVMC_TOP_FIELD: - { - vl_pic = VL_TOP_FIELD; - break; - } + return vlPictureTypeTopField; case XVMC_BOTTOM_FIELD: - { - vl_pic = VL_BOTTOM_FIELD; - break; - } + return vlPictureTypeBottomField; case XVMC_FRAME_PICTURE: - { - vl_pic = VL_FRAME_PICTURE; - break; - } + return vlPictureTypeFrame; default: assert(0); } - - return vl_pic; + + return -1; } -static enum VL_MC_TYPE MotionToVL(int xvmc_motion_type) +static enum vlMotionType MotionToVL(int xvmc_motion_type) { - enum VL_MC_TYPE vl_mc_type; - switch (xvmc_motion_type) { case XVMC_PREDICTION_FRAME: - { - vl_mc_type = VL_FRAME_MC; - break; - } + return vlMotionTypeFrame; case XVMC_PREDICTION_FIELD: - { - vl_mc_type = VL_FIELD_MC; - break; - } + return vlMotionTypeField; case XVMC_PREDICTION_DUAL_PRIME: - { - vl_mc_type = VL_DUAL_PRIME_MC; - break; - } + return vlMotionTypeDualPrime; default: assert(0); } - - return vl_mc_type; + + return -1; } Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *surface) { - struct VL_CONTEXT *vl_ctx; - struct VL_SURFACE *vl_sfc; - + struct vlContext *vl_ctx; + struct vlSurface *vl_sfc; + assert(display); - + if (!context) return XvMCBadContext; if (!surface) return XvMCBadSurface; - + vl_ctx = context->privData; - - assert(display == vl_ctx->display); - - vlCreateSurface(vl_ctx, &vl_sfc); - + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); + + vlCreateSurface + ( + vlContextGetScreen(vl_ctx), + context->width, + context->height, + vlGetPictureFormat(vl_ctx), + &vl_sfc + ); + + vlBindToContext(vl_sfc, vl_ctx); + surface->surface_id = XAllocID(display); surface->context_id = context->context_id; surface->surface_type_id = context->surface_type_id; surface->width = context->width; surface->height = context->height; surface->privData = vl_sfc; - + return Success; } @@ -103,19 +109,21 @@ Status XvMCRenderSurface XvMCBlockArray *blocks ) { - struct VL_CONTEXT *vl_ctx; - struct VL_SURFACE *target_vl_surface; - struct VL_SURFACE *past_vl_surface; - struct VL_SURFACE *future_vl_surface; - unsigned int i; - + struct vlContext *vl_ctx; + struct vlSurface *target_vl_surface; + struct vlSurface *past_vl_surface; + struct vlSurface *future_vl_surface; + struct vlMpeg2MacroBlockBatch batch; + struct vlMpeg2MacroBlock vl_macroblocks[num_macroblocks]; + unsigned int i; + assert(display); - + if (!context) return XvMCBadContext; if (!target_surface) return XvMCBadSurface; - + if ( picture_structure != XVMC_TOP_FIELD && @@ -125,178 +133,94 @@ Status XvMCRenderSurface return BadValue; if (future_surface && !past_surface) return BadMatch; - + vl_ctx = context->privData; - - assert(display == vl_ctx->display); - + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); + target_vl_surface = target_surface->privData; past_vl_surface = past_surface ? past_surface->privData : NULL; future_vl_surface = future_surface ? future_surface->privData : NULL; - - assert(vl_ctx == target_vl_surface->context); - assert(!past_vl_surface || vl_ctx == past_vl_surface->context); - assert(!future_vl_surface || vl_ctx == future_vl_surface->context); - + + assert(context->context_id == target_surface->context_id); + assert(!past_surface || context->context_id == past_surface->context_id); + assert(!future_surface || context->context_id == future_surface->context_id); + assert(macroblocks); assert(blocks); - + assert(macroblocks->context_id == context->context_id); assert(blocks->context_id == context->context_id); - + assert(flags == 0 || flags == XVMC_SECOND_FIELD); - - /* TODO: Batch macroblocks by type (I,P,B) */ - - for (i = first_macroblock; i < first_macroblock + num_macroblocks; ++i) - if (macroblocks->macro_blocks[i].macroblock_type & XVMC_MB_TYPE_INTRA) - vlRenderIMacroBlock - ( - PictureToVL(picture_structure), - flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, - macroblocks->macro_blocks[i].x, - macroblocks->macro_blocks[i].y, - macroblocks->macro_blocks[i].coded_block_pattern, - macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, - blocks->blocks + (macroblocks->macro_blocks[i].index * 64), - target_vl_surface - ); - else if - ( - (macroblocks->macro_blocks[i].macroblock_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) - == XVMC_MB_TYPE_MOTION_FORWARD - ) - { - struct VL_MOTION_VECTOR motion_vector = - { - { - macroblocks->macro_blocks[i].PMV[0][0][0], - macroblocks->macro_blocks[i].PMV[0][0][1], - }, - { - macroblocks->macro_blocks[i].PMV[1][0][0], - macroblocks->macro_blocks[i].PMV[1][0][1], - } - }; - - vlRenderPMacroBlock - ( - PictureToVL(picture_structure), - flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, - macroblocks->macro_blocks[i].x, - macroblocks->macro_blocks[i].y, - MotionToVL(macroblocks->macro_blocks[i].motion_type), - &motion_vector, - macroblocks->macro_blocks[i].coded_block_pattern, - macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, - blocks->blocks + (macroblocks->macro_blocks[i].index * 64), - past_vl_surface, - target_vl_surface - ); - } - else if - ( - (macroblocks->macro_blocks[i].macroblock_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) - == XVMC_MB_TYPE_MOTION_BACKWARD - ) - { - struct VL_MOTION_VECTOR motion_vector = - { - { - macroblocks->macro_blocks[i].PMV[0][1][0], - macroblocks->macro_blocks[i].PMV[0][1][1], - }, - { - macroblocks->macro_blocks[i].PMV[1][1][0], - macroblocks->macro_blocks[i].PMV[1][1][1], - } - }; - - vlRenderPMacroBlock - ( - PictureToVL(picture_structure), - flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, - macroblocks->macro_blocks[i].x, - macroblocks->macro_blocks[i].y, - MotionToVL(macroblocks->macro_blocks[i].motion_type), - &motion_vector, - macroblocks->macro_blocks[i].coded_block_pattern, - macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, - blocks->blocks + (macroblocks->macro_blocks[i].index * 64), - future_vl_surface, - target_vl_surface - ); - } - else if - ( - (macroblocks->macro_blocks[i].macroblock_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) - == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD) - ) - { - struct VL_MOTION_VECTOR motion_vector[2] = - { - { - { - macroblocks->macro_blocks[i].PMV[0][0][0], - macroblocks->macro_blocks[i].PMV[0][0][1], - }, - { - macroblocks->macro_blocks[i].PMV[1][0][0], - macroblocks->macro_blocks[i].PMV[1][0][1], - } - }, - { - { - macroblocks->macro_blocks[i].PMV[0][1][0], - macroblocks->macro_blocks[i].PMV[0][1][1], - }, - { - macroblocks->macro_blocks[i].PMV[1][1][0], - macroblocks->macro_blocks[i].PMV[1][1][1], - } - } - }; - - vlRenderBMacroBlock - ( - PictureToVL(picture_structure), - flags == XVMC_SECOND_FIELD ? VL_FIELD_SECOND : VL_FIELD_FIRST, - macroblocks->macro_blocks[i].x, - macroblocks->macro_blocks[i].y, - MotionToVL(macroblocks->macro_blocks[i].motion_type), - motion_vector, - macroblocks->macro_blocks[i].coded_block_pattern, - macroblocks->macro_blocks[i].dct_type == XVMC_DCT_TYPE_FIELD ? VL_DCT_FIELD_CODED : VL_DCT_FRAME_CODED, - blocks->blocks + (macroblocks->macro_blocks[i].index * 64), - past_vl_surface, - future_vl_surface, - target_vl_surface - ); - } - else - fprintf(stderr, "Unrecognized macroblock\n"); - + + batch.past_surface = past_vl_surface; + batch.future_surface = future_vl_surface; + batch.picture_type = PictureToVL(picture_structure); + batch.field_order = flags & XVMC_SECOND_FIELD ? vlFieldOrderSecond : vlFieldOrderFirst; + batch.num_macroblocks = num_macroblocks; + batch.macroblocks = vl_macroblocks; + + for (i = 0; i < num_macroblocks; ++i) + { + unsigned int j = first_macroblock + i; + + unsigned int k, l, m; + + batch.macroblocks[i].mbx = macroblocks->macro_blocks[j].x; + batch.macroblocks[i].mby = macroblocks->macro_blocks[j].y; + batch.macroblocks[i].mb_type = TypeToVL(macroblocks->macro_blocks[j].macroblock_type); + if (batch.macroblocks[i].mb_type != vlMacroBlockTypeIntra) + batch.macroblocks[i].mo_type = MotionToVL(macroblocks->macro_blocks[j].motion_type); + batch.macroblocks[i].dct_type = macroblocks->macro_blocks[j].dct_type & XVMC_DCT_TYPE_FIELD ? vlDCTTypeFieldCoded : vlDCTTypeFrameCoded; + + for (k = 0; k < 2; ++k) + for (l = 0; l < 2; ++l) + for (m = 0; m < 2; ++m) + batch.macroblocks[i].PMV[k][l][m] = macroblocks->macro_blocks[j].PMV[k][l][m]; + + batch.macroblocks[i].cbp = macroblocks->macro_blocks[j].coded_block_pattern; + batch.macroblocks[i].blocks = blocks->blocks + (macroblocks->macro_blocks[j].index * 64); + } + + vlRenderMacroBlocksMpeg2(&batch, target_vl_surface); + return Success; } Status XvMCFlushSurface(Display *display, XvMCSurface *surface) { + struct vlSurface *vl_sfc; + assert(display); - + if (!surface) return XvMCBadSurface; - - /* TODO: Check display & surface match */ + + vl_sfc = surface->privData; + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); + + /* TODO */ + return Success; } Status XvMCSyncSurface(Display *display, XvMCSurface *surface) { + struct vlSurface *vl_sfc; + assert(display); - + if (!surface) return XvMCBadSurface; - + + vl_sfc = surface->privData; + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); + + /* TODO */ + return Success; } @@ -321,92 +245,85 @@ Status XvMCPutSurface unsigned int width, height; unsigned int border_width; unsigned int depth; - struct VL_SURFACE *vl_sfc; - + struct vlSurface *vl_sfc; + assert(display); - + if (!surface) return XvMCBadSurface; - + if (XGetGeometry(display, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable) return BadDrawable; - + assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE); - + /* TODO: Correct for negative srcx,srcy & destx,desty by clipping */ - + assert(srcx + srcw - 1 < surface->width); assert(srcy + srch - 1 < surface->height); assert(destx + destw - 1 < width); assert(desty + desth - 1 < height); - + vl_sfc = surface->privData; - - vlPutSurface(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, PictureToVL(flags)); - + + vlPutPicture(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, PictureToVL(flags)); + return Success; } Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status) { - struct VL_CONTEXT *vl_ctx; - struct VL_SURFACE *vl_sfc; - + struct vlSurface *vl_sfc; + assert(display); - + if (!surface) return XvMCBadSurface; - + assert(status); - + vl_sfc = surface->privData; - vl_ctx = vl_sfc->context; - - assert(display == vl_ctx->display); - + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); + /* TODO */ *status = 0; - + return Success; } Status XvMCDestroySurface(Display *display, XvMCSurface *surface) { - struct VL_CONTEXT *vl_ctx; - struct VL_SURFACE *vl_sfc; - + struct vlSurface *vl_sfc; + assert(display); - + if (!surface) return XvMCBadSurface; - + vl_sfc = surface->privData; - vl_ctx = vl_sfc->context; - - assert(display == vl_ctx->display); - + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); + vlDestroySurface(vl_sfc); - + return Success; } Status XvMCHideSurface(Display *display, XvMCSurface *surface) { - struct VL_CONTEXT *vl_ctx; - struct VL_SURFACE *vl_sfc; - + struct vlSurface *vl_sfc; + assert(display); - + if (!surface) return XvMCBadSurface; - + vl_sfc = surface->privData; - vl_ctx = vl_sfc->context; - - assert(display == vl_ctx->display); - + + assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); + /* No op, only for overlaid rendering */ - + return Success; } - -- cgit v1.2.3 From f90e50dff9e5cdbad6e9bb74c0aeaaaa82242b25 Mon Sep 17 00:00:00 2001 From: Stephane Marchesin Date: Mon, 18 Aug 2008 03:00:25 +0200 Subject: nv30: add some opcodes. --- src/gallium/drivers/nv30/nv30_fragprog.c | 11 +++++++++++ src/gallium/drivers/nv30/nv30_vertprog.c | 6 ++++++ 2 files changed, 17 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index d3693fdf567..02f30ad9bad 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -526,6 +526,12 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, case TGSI_OPCODE_MUL: arith(fpc, sat, MUL, dst, mask, src[0], src[1], none); break; + case TGSI_OPCODE_NOISE1: + case TGSI_OPCODE_NOISE2: + case TGSI_OPCODE_NOISE3: + case TGSI_OPCODE_NOISE4: + arith(fpc, sat, SFL, dst, mask, none, none, none); + break; case TGSI_OPCODE_POW: arith(fpc, sat, POW, dst, mask, src[0], src[1], none); break; @@ -557,6 +563,9 @@ nv30_fragprog_parse_instruction(struct nv30_fpc *fpc, case TGSI_OPCODE_SGE: arith(fpc, sat, SGE, dst, mask, src[0], src[1], none); break; + case TGSI_OPCODE_SGT: + arith(fpc, sat, SGT, dst, mask, src[0], src[1], none); + break; case TGSI_OPCODE_SLT: arith(fpc, sat, SLT, dst, mask, src[0], src[1], none); break; @@ -730,6 +739,8 @@ nv30_fragprog_translate(struct nv30_context *nv30, struct tgsi_parse_context parse; struct nv30_fpc *fpc = NULL; + tgsi_dump(fp->pipe.tokens,0); + fpc = CALLOC(1, sizeof(struct nv30_fpc)); if (!fpc) return; diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c index 6c075cdb752..72824559e8b 100644 --- a/src/gallium/drivers/nv30/nv30_vertprog.c +++ b/src/gallium/drivers/nv30/nv30_vertprog.c @@ -4,6 +4,7 @@ #include "pipe/p_shader_tokens.h" #include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_dump.h" #include "nv30_context.h" #include "nv30_state.h" @@ -457,6 +458,9 @@ nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, case TGSI_OPCODE_SGE: arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); break; + case TGSI_OPCODE_SGT: + arith(vpc, 0, OP_SGT, dst, mask, src[0], src[1], none); + break; case TGSI_OPCODE_SLT: arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); break; @@ -570,6 +574,8 @@ nv30_vertprog_translate(struct nv30_context *nv30, struct tgsi_parse_context parse; struct nv30_vpc *vpc = NULL; + tgsi_dump(vp->pipe.tokens,0); + vpc = CALLOC(1, sizeof(struct nv30_vpc)); if (!vpc) return; -- cgit v1.2.3 From 7f100d04ddacf9f6517c9aff1e2de5257eb77fb0 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 18 Aug 2008 00:04:29 -0400 Subject: g3dvl: Use rotating buffers to avoid waiting for map(). --- src/driclient/src/Makefile | 2 +- src/gallium/state_trackers/g3dvl/Makefile | 4 +- src/gallium/state_trackers/g3dvl/vl_data.c | 76 ----------------------- src/gallium/state_trackers/g3dvl/vl_data.h | 27 -------- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c | 46 +++++++++----- src/gallium/winsys/g3dvl/nouveau/Makefile | 2 +- src/libXvMC/Makefile | 2 +- 7 files changed, 37 insertions(+), 122 deletions(-) delete mode 100644 src/gallium/state_trackers/g3dvl/vl_data.c delete mode 100644 src/gallium/state_trackers/g3dvl/vl_data.h diff --git a/src/driclient/src/Makefile b/src/driclient/src/Makefile index 89ffb84477f..5d913ad5fa5 100644 --- a/src/driclient/src/Makefile +++ b/src/driclient/src/Makefile @@ -2,7 +2,7 @@ TARGET = libdriclient.a OBJECTS = driclient.o XF86dri.o DRMDIR ?= /usr -CFLAGS += -g -Wall -fPIC -Werror -I../include -I${DRMDIR}/include -I${DRMDIR}/include/drm +CFLAGS += -g -Wall -fPIC -I../include -I${DRMDIR}/include -I${DRMDIR}/include/drm ############################################# diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile index 9995c554ab8..bd77c62bc58 100644 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -1,9 +1,9 @@ TARGET = libg3dvl.a -OBJECTS = vl_display.o vl_screen.o vl_context.o vl_surface.o vl_data.o vl_shader_build.o vl_util.o vl_basic_csc.o \ +OBJECTS = vl_display.o vl_screen.o vl_context.o vl_surface.o vl_shader_build.o vl_util.o vl_basic_csc.o \ vl_r16snorm_mc.o GALLIUMDIR = ../.. -CFLAGS += -g -Wall -fPIC -Werror -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary -I${GALLIUMDIR}/winsys/g3dvl +CFLAGS += -g -Wall -fPIC -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary -I${GALLIUMDIR}/winsys/g3dvl ############################################# diff --git a/src/gallium/state_trackers/g3dvl/vl_data.c b/src/gallium/state_trackers/g3dvl/vl_data.c deleted file mode 100644 index f2476dbf1ee..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_data.c +++ /dev/null @@ -1,76 +0,0 @@ -#include "vl_data.h" - -/* - * Represents 8 triangles (4 quads, 1 per block) in noormalized coords - * that render a macroblock. - * Need to be scaled to cover mbW*mbH macroblock pixels and translated into - * position on target surface. - */ -const struct vlVertex2f macroblock_verts[24] = -{ - {0.0f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.0f}, - {0.5f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.5f}, - - {0.5f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.0f}, - {1.0f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.5f}, - - {0.0f, 0.5f}, {0.0f, 1.0f}, {0.5f, 0.5f}, - {0.5f, 0.5f}, {0.0f, 1.0f}, {0.5f, 1.0f}, - - {0.5f, 0.5f}, {0.5f, 1.0f}, {1.0f, 0.5f}, - {1.0f, 0.5f}, {0.5f, 1.0f}, {1.0f, 1.0f} -}; - -/* - * Represents texcoords for the above for rendering 4 luma blocks arranged - * in a bW*(bH*4) texture. First luma block located at 0,0->bW,bH; second at - * 0,bH->bW,2bH; third at 0,2bH->bW,3bH; fourth at 0,3bH->bW,4bH. - */ -const struct vlVertex2f macroblock_luma_texcoords[24] = -{ - {0.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.0f}, - {1.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.25f}, - - {0.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.25f}, - {1.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.5f}, - - {0.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.5f}, - {1.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.75f}, - - {0.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 0.75f}, - {1.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 1.0f} -}; - -/* - * Represents texcoords for the above for rendering 1 chroma block. - * Straight forward 0,0->1,1 mapping so we can reuse the MB pos vectors. - */ -const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts; - -/* - * Represents texcoords for the above for rendering 2 chroma blocks arranged - * in a bW*(bH*2) texture. First chroma block located at 0,0->bW,bH; second at - * 0,bH->bW,2bH. We can render this with 0,0->1,1 mapping. - * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. - */ -const struct vlVertex2f *macroblock_chroma_422_texcoords = macroblock_verts; - -/* - * Represents texcoords for the above for rendering 4 chroma blocks. - * Same case as 4 luma blocks. - */ -const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texcoords; - -/* - * Used when rendering P and B macroblocks, multiplier is applied to the A channel, - * which is then added to the L channel, then the bias is subtracted from that to - * get back the differential. The differential is then added to the samples from the - * reference surface(s). - */ -#if 0 -const struct VL_MC_FS_CONSTS vl_mc_fs_consts = -{ - {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, - {0.5f, 2.0f, 0.0f, 0.0f} -}; -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_data.h b/src/gallium/state_trackers/g3dvl/vl_data.h deleted file mode 100644 index f0de2e976ce..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_data.h +++ /dev/null @@ -1,27 +0,0 @@ -#ifndef vl_data_h -#define vl_data_h - -#include "vl_types.h" - -/* TODO: Needs to be rolled into the appropriate stage */ - -extern const struct vlVertex2f macroblock_verts[24]; -extern const struct vlVertex2f macroblock_luma_texcoords[24]; -extern const struct vlVertex2f *macroblock_chroma_420_texcoords; -extern const struct vlVertex2f *macroblock_chroma_422_texcoords; -extern const struct vlVertex2f *macroblock_chroma_444_texcoords; - -extern const struct vlVertex2f surface_verts[4]; -extern const struct vlVertex2f *surface_texcoords; - -/* -extern const struct VL_MC_FS_CONSTS vl_mc_fs_consts; - -extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_identity; -extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601; -extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_601_full; -extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709; -extern const struct VL_CSC_FS_CONSTS vl_csc_fs_consts_709_full; -*/ - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c index 4fae2244310..80b09a6d1d0 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c @@ -15,6 +15,8 @@ #include "vl_types.h" #include "vl_defs.h" +#define NUM_BUFS 4 /* Number of rotating buffers to use */ + struct vlVertexShaderConsts { /*struct vlVertex4f scale; @@ -41,12 +43,13 @@ struct vlR16SnormMC unsigned int video_width, video_height; enum vlFormat video_format; + unsigned int cur_buf; struct pipe_context *pipe; struct pipe_viewport_state viewport; struct pipe_framebuffer_state render_target; struct pipe_sampler_state *samplers[5]; - struct pipe_texture *textures[5]; + struct pipe_texture *textures[NUM_BUFS][5]; void *i_vs, *p_vs[2], *b_vs[2]; void *i_fs, *p_fs[2], *b_fs[2]; struct pipe_vertex_buffer vertex_bufs[3]; @@ -230,7 +233,7 @@ static int vlGrabBlocks tex_surface = mc->pipe->screen->get_tex_surface ( mc->pipe->screen, - mc->textures[0], + mc->textures[mc->cur_buf % NUM_BUFS][0], 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE ); @@ -276,7 +279,7 @@ static int vlGrabBlocks tex_surface = mc->pipe->screen->get_tex_surface ( mc->pipe->screen, - mc->textures[tb + 1], + mc->textures[mc->cur_buf % NUM_BUFS][tb + 1], 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE ); @@ -361,12 +364,14 @@ int vlRenderIMacroBlock 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE ); pipe->set_framebuffer_state(pipe, &mc->render_target); - pipe->set_sampler_textures(pipe, 3, mc->textures); + pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUFS]); pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers); pipe->bind_vs_state(pipe, mc->i_vs); pipe->bind_fs_state(pipe, mc->i_fs); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + mc->cur_buf++; return 0; } @@ -458,11 +463,13 @@ int vlRenderPMacroBlock ); pipe->set_framebuffer_state(pipe, &mc->render_target); - mc->textures[3] = ref_surface->texture; - pipe->set_sampler_textures(pipe, 4, mc->textures); + mc->textures[mc->cur_buf % NUM_BUFS][3] = ref_surface->texture; + pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUFS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + mc->cur_buf++; return 0; } @@ -567,12 +574,14 @@ int vlRenderBMacroBlock ); pipe->set_framebuffer_state(pipe, &mc->render_target); - mc->textures[3] = past_surface->texture; - mc->textures[4] = future_surface->texture; - pipe->set_sampler_textures(pipe, 5, mc->textures); + mc->textures[mc->cur_buf % NUM_BUFS][3] = past_surface->texture; + mc->textures[mc->cur_buf % NUM_BUFS][4] = future_surface->texture; + pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUFS]); pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); + + mc->cur_buf++; return 0; } @@ -724,8 +733,12 @@ int vlDestroy pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[i].buffer); /* Textures 3 & 4 are not created directly, no need to release them here */ - for (i = 0; i < 3; ++i) - pipe_texture_release(&mc->textures[i]); + for (i = 0; i < NUM_BUFS; ++i) + { + pipe_texture_release(&mc->textures[i][0]); + pipe_texture_release(&mc->textures[i][1]); + pipe_texture_release(&mc->textures[i][2]); + } pipe->delete_vs_state(pipe, mc->i_vs); pipe->delete_fs_state(pipe, mc->i_fs); @@ -2252,7 +2265,8 @@ static int vlInit template.compressed = 0; pf_get_block(template.format, &template.block); - mc->textures[0] = pipe->screen->texture_create(pipe->screen, &template); + for (i = 0; i < NUM_BUFS; ++i) + mc->textures[i][0] = pipe->screen->texture_create(pipe->screen, &template); if (mc->video_format == vlFormatYCbCr420) template.height[0] = 8; @@ -2263,8 +2277,11 @@ static int vlInit else assert(0); - mc->textures[1] = pipe->screen->texture_create(pipe->screen, &template); - mc->textures[2] = pipe->screen->texture_create(pipe->screen, &template); + for (i = 0; i < NUM_BUFS; ++i) + { + mc->textures[i][1] = pipe->screen->texture_create(pipe->screen, &template); + mc->textures[i][2] = pipe->screen->texture_create(pipe->screen, &template); + } /* textures[3] & textures[4] are assigned from vlSurfaces for P and B macroblocks at render time */ @@ -2306,6 +2323,7 @@ int vlCreateR16SNormMC mc->pipe = pipe; mc->video_width = video_width; mc->video_height = video_height; + mc->cur_buf = 0; vlInit(mc); diff --git a/src/gallium/winsys/g3dvl/nouveau/Makefile b/src/gallium/winsys/g3dvl/nouveau/Makefile index 2861bd7db4f..7fa29d2f5fb 100644 --- a/src/gallium/winsys/g3dvl/nouveau/Makefile +++ b/src/gallium/winsys/g3dvl/nouveau/Makefile @@ -9,7 +9,7 @@ OBJECTS = nouveau_bo.o nouveau_fence.o nouveau_swapbuffers.o nouveau_channel.o nouveau_pushbuf.o nouveau_resource.o nouveau_screen.o nv04_surface.o \ nv50_surface.o #nouveau_winsys_softpipe.o -CFLAGS += -g -Wall -Werror -fPIC \ +CFLAGS += -g -Wall -fPIC \ -I${GALLIUMDIR}/include \ -I${GALLIUMDIR}/winsys/g3dvl \ -I${DRMDIR}/include \ diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile index c154e17dc35..b72bb16efbf 100644 --- a/src/libXvMC/Makefile +++ b/src/libXvMC/Makefile @@ -8,7 +8,7 @@ ifeq (${DRIVER}, softpipe) OBJECTS += ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o endif -CFLAGS += -g -fPIC -Wall -Werror \ +CFLAGS += -g -fPIC -Wall \ -I${GALLIUMDIR}/state_trackers/g3dvl \ -I${GALLIUMDIR}/winsys/g3dvl \ -I${GALLIUMDIR}/include \ -- cgit v1.2.3 From 4d9d192672508eaa9b2a70f84e933f11108bf09f Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 28 Aug 2008 23:25:13 -0400 Subject: g3dvl: Buffer the entire frame before rendering. --- src/gallium/state_trackers/g3dvl/Makefile | 2 +- src/gallium/state_trackers/g3dvl/vl_context.c | 4 +- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c | 45 +- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h | 4 +- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 2341 ++++++++++++++++++++ .../state_trackers/g3dvl/vl_r16snorm_mc_buf.h | 18 + src/gallium/state_trackers/g3dvl/vl_render.h | 5 + src/gallium/state_trackers/g3dvl/vl_surface.c | 2 + 8 files changed, 2400 insertions(+), 21 deletions(-) create mode 100644 src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c create mode 100644 src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile index bd77c62bc58..4f7a9534847 100644 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ b/src/gallium/state_trackers/g3dvl/Makefile @@ -1,6 +1,6 @@ TARGET = libg3dvl.a OBJECTS = vl_display.o vl_screen.o vl_context.o vl_surface.o vl_shader_build.o vl_util.o vl_basic_csc.o \ - vl_r16snorm_mc.o + vl_r16snorm_mc.o vl_r16snorm_mc_buf.o GALLIUMDIR = ../.. CFLAGS += -g -Wall -fPIC -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary -I${GALLIUMDIR}/winsys/g3dvl diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c index 56d360c05b9..fe107e406da 100644 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ b/src/gallium/state_trackers/g3dvl/vl_context.c @@ -6,6 +6,7 @@ #include #include "vl_render.h" #include "vl_r16snorm_mc.h" +#include "vl_r16snorm_mc_buf.h" #include "vl_csc.h" #include "vl_basic_csc.h" @@ -126,7 +127,8 @@ int vlCreateContext vlInitCommon(ctx); - vlCreateR16SNormMC(pipe, picture_width, picture_height, picture_format, &ctx->render); + /*vlCreateR16SNormMC(pipe, picture_width, picture_height, picture_format, &ctx->render);*/ + vlCreateR16SNormBufferedMC(pipe, picture_width, picture_height, picture_format, &ctx->render); vlCreateBasicCSC(pipe, &ctx->csc); *context = ctx; diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c index 80b09a6d1d0..3272220ef83 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c @@ -57,7 +57,7 @@ struct vlR16SnormMC struct pipe_constant_buffer vs_const_buf, fs_const_buf; }; -int vlBegin +static int vlBegin ( struct vlRender *render ) @@ -312,7 +312,7 @@ static int vlGrabBlocks return 0; } -int vlRenderIMacroBlock +static int vlRenderIMacroBlock ( struct vlR16SnormMC *mc, enum vlPictureType picture_type, @@ -370,13 +370,13 @@ int vlRenderIMacroBlock pipe->bind_fs_state(pipe, mc->i_fs); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); - + mc->cur_buf++; return 0; } -int vlRenderPMacroBlock +static int vlRenderPMacroBlock ( struct vlR16SnormMC *mc, enum vlPictureType picture_type, @@ -468,13 +468,13 @@ int vlRenderPMacroBlock pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); - + mc->cur_buf++; return 0; } -int vlRenderBMacroBlock +static int vlRenderBMacroBlock ( struct vlR16SnormMC *mc, enum vlPictureType picture_type, @@ -580,13 +580,13 @@ int vlRenderBMacroBlock pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers); pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24); - + mc->cur_buf++; return 0; } -int vlRenderMacroBlocksMpeg2R16Snorm +static int vlRenderMacroBlocksMpeg2R16Snorm ( struct vlRender *render, struct vlMpeg2MacroBlockBatch *batch, @@ -702,7 +702,17 @@ int vlRenderMacroBlocksMpeg2R16Snorm return 0; } -int vlEnd +static int vlEnd +( + struct vlRender *render +) +{ + assert(render); + + return 0; +} + +static int vlFlush ( struct vlRender *render ) @@ -712,7 +722,7 @@ int vlEnd return 0; } -int vlDestroy +static int vlDestroy ( struct vlRender *render ) @@ -765,7 +775,7 @@ int vlDestroy * Need to be scaled to cover mbW*mbH macroblock pixels and translated into * position on target surface. */ -const struct vlVertex2f macroblock_verts[24] = +static const struct vlVertex2f macroblock_verts[24] = { {0.0f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.0f}, {0.5f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.5f}, @@ -785,7 +795,7 @@ const struct vlVertex2f macroblock_verts[24] = * in a bW*(bH*4) texture. First luma block located at 0,0->bW,bH; second at * 0,bH->bW,2bH; third at 0,2bH->bW,3bH; fourth at 0,3bH->bW,4bH. */ -const struct vlVertex2f macroblock_luma_texcoords[24] = +static const struct vlVertex2f macroblock_luma_texcoords[24] = { {0.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.0f}, {1.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.25f}, @@ -804,7 +814,7 @@ const struct vlVertex2f macroblock_luma_texcoords[24] = * Represents texcoords for the above for rendering 1 chroma block. * Straight forward 0,0->1,1 mapping so we can reuse the MB pos vectors. */ -const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts; +static const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts; /* * Represents texcoords for the above for rendering 2 chroma blocks arranged @@ -812,13 +822,13 @@ const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts; * 0,bH->bW,2bH. We can render this with 0,0->1,1 mapping. * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors. */ -const struct vlVertex2f *macroblock_chroma_422_texcoords = macroblock_verts; +static const struct vlVertex2f *macroblock_chroma_422_texcoords = macroblock_verts; /* * Represents texcoords for the above for rendering 4 chroma blocks. * Same case as 4 luma blocks. */ -const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texcoords; +static const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texcoords; /* * Used when rendering P and B macroblocks, multiplier is applied to the A channel, @@ -826,7 +836,7 @@ const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texco * get back the differential. The differential is then added to the samples from the * reference surface(s). */ -const struct vlFragmentShaderConsts fs_consts = +static const struct vlFragmentShaderConsts fs_consts = { {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, {0.5f, 2.0f, 0.0f, 0.0f} @@ -2093,7 +2103,7 @@ static int vlCreateFragmentShaderFieldBMB return 0; } -int vlCreateDataBufs +static int vlCreateDataBufs ( struct vlR16SnormMC *mc ) @@ -2319,6 +2329,7 @@ int vlCreateR16SNormMC mc->base.vlBegin = &vlBegin; mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16Snorm; mc->base.vlEnd = &vlEnd; + mc->base.vlFlush = &vlFlush; mc->base.vlDestroy = &vlDestroy; mc->pipe = pipe; mc->video_width = video_width; diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h index a6eecf05b6c..9842926bf70 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h @@ -1,5 +1,5 @@ -#ifndef vl_mc_h -#define vl_mc_h +#ifndef vl_r16snorm_mc_h +#define vl_r16snorm_mc_h #include "vl_types.h" diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c new file mode 100644 index 00000000000..fc383cb8f6d --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -0,0 +1,2341 @@ +#define VL_INTERNAL +#include "vl_r16snorm_mc_buf.h" +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "vl_render.h" +#include "vl_shader_build.h" +#include "vl_surface.h" +#include "vl_util.h" +#include "vl_types.h" +#include "vl_defs.h" + +#define NUM_BUF_SETS 4 /* Number of rotating buffer sets to use */ + +enum vlMacroBlockTypeEx +{ + vlMacroBlockExTypeIntra, + vlMacroBlockExTypeFwdPredictedFrame, + vlMacroBlockExTypeFwdPredictedField, + vlMacroBlockExTypeBkwdPredictedFrame, + vlMacroBlockExTypeBkwdPredictedField, + vlMacroBlockExTypeBiPredictedFrame, + vlMacroBlockExTypeBiPredictedField, + + vlNumMacroBlockExTypes +}; + +struct vlVertexShaderConsts +{ + struct vlVertex4f denorm; +}; + +struct vlFragmentShaderConsts +{ + struct vlVertex4f multiplier; + struct vlVertex4f div; +}; + +struct vlR16SnormBufferedMC +{ + struct vlRender base; + + unsigned int video_width, video_height; + enum vlFormat video_format; + + unsigned int cur_buf; + struct vlSurface *buffered_surface; + struct vlSurface *past_surface, *future_surface; + struct vlVertex2f surface_tex_inv_size; + unsigned int num_macroblocks[vlNumMacroBlockExTypes]; + unsigned int total_num_macroblocks; + + struct pipe_context *pipe; + struct pipe_viewport_state viewport; + struct pipe_framebuffer_state render_target; + struct pipe_sampler_state *samplers[5]; + struct pipe_texture *textures[NUM_BUF_SETS][5]; + void *i_vs, *p_vs[2], *b_vs[2]; + void *i_fs, *p_fs[2], *b_fs[2]; + struct pipe_vertex_buffer vertex_bufs[NUM_BUF_SETS][vlNumMacroBlockExTypes][3]; + struct pipe_vertex_element vertex_elems[5]; + struct pipe_constant_buffer vs_const_buf, fs_const_buf; +}; + +static int vlBegin +( + struct vlRender *render +) +{ + assert(render); + + return 0; +} + +static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + memcpy + ( + dst + y * dst_pitch, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) + memcpy + ( + dst + y * dst_pitch * 2, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + dst += VL_BLOCK_HEIGHT * dst_pitch; + + for (; y < VL_BLOCK_HEIGHT; ++y) + memcpy + ( + dst + y * dst_pitch * 2, + src + y * VL_BLOCK_WIDTH, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) +{ + unsigned int y; + + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) + memset + ( + dst + y * dst_pitch, + 0, + VL_BLOCK_WIDTH * 2 + ); + + return 0; +} + +static int vlGrabBlocks +( + struct vlR16SnormBufferedMC *mc, + unsigned int mbx, + unsigned int mby, + enum vlDCTType dct_type, + unsigned int coded_block_pattern, + short *blocks +) +{ + struct pipe_surface *tex_surface; + short *texels; + unsigned int tex_pitch; + unsigned int x, y, tb = 0, sb = 0; + unsigned int mbpx = mbx * VL_MACROBLOCK_WIDTH, mbpy = mby * VL_MACROBLOCK_HEIGHT; + + assert(mc); + assert(blocks); + + tex_surface = mc->pipe->screen->get_tex_surface + ( + mc->pipe->screen, + mc->textures[mc->cur_buf % NUM_BUF_SETS][0], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); + + texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); + tex_pitch = tex_surface->stride / tex_surface->block.size; + + texels += mbpy * tex_pitch + mbpx; + + for (y = 0; y < 2; ++y) + { + for (x = 0; x < 2; ++x, ++tb) + { + if ((coded_block_pattern >> (5 - tb)) & 1) + { + short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; + + if (dct_type == vlDCTTypeFrameCoded) + { + vlGrabFrameCodedBlock + ( + cur_block, + texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, + tex_pitch + ); + } + else + { + vlGrabFieldCodedBlock + ( + cur_block, + texels + y * tex_pitch + x * VL_BLOCK_WIDTH, + tex_pitch + ); + } + + ++sb; + } + else + vlGrabNoBlock(texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, tex_pitch); + } + } + + pipe_surface_unmap(tex_surface); + + /* TODO: Implement 422, 444 */ + mbpx >>= 1; + mbpy >>= 1; + + for (tb = 0; tb < 2; ++tb) + { + tex_surface = mc->pipe->screen->get_tex_surface + ( + mc->pipe->screen, + mc->textures[mc->cur_buf % NUM_BUF_SETS][tb + 1], + 0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE + ); + + texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE); + tex_pitch = tex_surface->stride / tex_surface->block.size; + + texels += mbpy * tex_pitch + mbpx; + + if ((coded_block_pattern >> (1 - tb)) & 1) + { + short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; + + vlGrabFrameCodedBlock + ( + cur_block, + texels, + tex_pitch + ); + + ++sb; + } + else + vlGrabNoBlock(texels, tex_pitch); + + pipe_surface_unmap(tex_surface); + } + + return 0; +} + +#if 0 +static int vlGrabMacroBlock +( + struct vlR16SnormBufferedMC *mc, + struct vlMpeg2MacroBlock *macroblock +) +{ + const struct vlVertex2f unit = + { + mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH, + mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT + }; + const struct vlVertex2f half = + { + mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2), + mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2) + }; + + struct vlVertex2f *vb; + enum vlMacroBlockTypeEx mb_type_ex; + struct vlVertex2f mo_vec[2]; + unsigned int i; + + assert(mc); + assert(macroblock); + + switch (macroblock->mb_type) + { + case vlMacroBlockTypeIntra: + { + mb_type_ex = vlMacroBlockExTypeIntra; + break; + } + case vlMacroBlockTypeFwdPredicted: + { + mb_type_ex = macroblock->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField; + break; + } + case vlMacroBlockTypeBkwdPredicted: + { + mb_type_ex = macroblock->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField; + break; + } + case vlMacroBlockTypeBiPredicted: + { + mb_type_ex = macroblock->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField; + break; + } + default: + assert(0); + } + + switch (macroblock->mb_type) + { + case vlMacroBlockTypeBiPredicted: + { + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + ( + mc->pipe->winsys, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][2].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ) + mc->num_macroblocks[mb_type_ex] * 2 * 24; + + mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; + + if (macroblock->mo_type == vlMotionTypeFrame) + { + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + } + } + else + { + mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; + + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + vb[i + 1].x = mo_vec[1].x; + vb[i + 1].y = mo_vec[1].y; + } + } + + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][2].buffer); + + /* fall-through */ + } + case vlMacroBlockTypeFwdPredicted: + case vlMacroBlockTypeBkwdPredicted: + { + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + ( + mc->pipe->winsys, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][1].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ) + mc->num_macroblocks[mb_type_ex] * 2 * 24; + + if (macroblock->mb_type == vlMacroBlockTypeBkwdPredicted) + { + mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; + + if (macroblock->mo_type == vlMotionTypeField) + { + mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; + } + } + else + { + mo_vec[0].x = macroblock->PMV[0][0][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[0].y = macroblock->PMV[0][0][1] * 0.5f * mc->surface_tex_inv_size.y; + + if (macroblock->mo_type == vlMotionTypeField) + { + mo_vec[1].x = macroblock->PMV[1][0][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[1].y = macroblock->PMV[1][0][1] * 0.5f * mc->surface_tex_inv_size.y; + } + } + + if (macroblock->mo_type == vlMotionTypeFrame) + { + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + } + } + else + { + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + vb[i + 1].x = mo_vec[1].x; + vb[i + 1].y = mo_vec[1].y; + } + } + + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][1].buffer); + + /* fall-through */ + } + case vlMacroBlockTypeIntra: + { + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + ( + mc->pipe->winsys, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][0].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ) + mc->num_macroblocks[mb_type_ex] * 24; + + vb[0].x = macroblock->mbx * unit.x; vb[0].y = macroblock->mby * unit.y; + vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y + half.y; + vb[2].x = macroblock->mbx * unit.x + half.x; vb[2].y = macroblock->mby * unit.y; + + vb[3].x = macroblock->mbx * unit.x + half.x; vb[3].y = macroblock->mby * unit.y; + vb[4].x = macroblock->mbx * unit.x; vb[4].y = macroblock->mby * unit.y + half.y; + vb[5].x = macroblock->mbx * unit.x + half.x; vb[5].y = macroblock->mby * unit.y + half.y; + + vb[6].x = macroblock->mbx * unit.x + half.x; vb[6].y = macroblock->mby * unit.y; + vb[7].x = macroblock->mbx * unit.x + half.x; vb[7].y = macroblock->mby * unit.y + half.y; + vb[8].x = macroblock->mbx * unit.x + unit.x; vb[8].y = macroblock->mby * unit.y; + + vb[9].x = macroblock->mbx * unit.x + unit.x; vb[9].y = macroblock->mby * unit.y; + vb[10].x = macroblock->mbx * unit.x + half.x; vb[10].y = macroblock->mby * unit.y + half.y; + vb[11].x = macroblock->mbx * unit.x + unit.x; vb[11].y = macroblock->mby * unit.y + half.y; + + vb[12].x = macroblock->mbx * unit.x; vb[12].y = macroblock->mby * unit.y + half.y; + vb[13].x = macroblock->mbx * unit.x; vb[13].y = macroblock->mby * unit.y + unit.y; + vb[14].x = macroblock->mbx * unit.x + half.x; vb[14].y = macroblock->mby * unit.y + half.y; + + vb[15].x = macroblock->mbx * unit.x + half.x; vb[15].y = macroblock->mby * unit.y + half.y; + vb[16].x = macroblock->mbx * unit.x; vb[16].y = macroblock->mby * unit.y + unit.y; + vb[17].x = macroblock->mbx * unit.x + half.x; vb[17].y = macroblock->mby * unit.y + unit.y; + + vb[18].x = macroblock->mbx * unit.x + half.x; vb[18].y = macroblock->mby * unit.y + half.y; + vb[19].x = macroblock->mbx * unit.x + half.x; vb[19].y = macroblock->mby * unit.y + unit.y; + vb[20].x = macroblock->mbx * unit.x + unit.x; vb[20].y = macroblock->mby * unit.y + half.y; + + vb[21].x = macroblock->mbx * unit.x + unit.x; vb[21].y = macroblock->mby * unit.y + half.y; + vb[22].x = macroblock->mbx * unit.x + half.x; vb[22].y = macroblock->mby * unit.y + unit.y; + vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + unit.y; + + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][0].buffer); + + break; + } + default: + assert(0); + } + + vlGrabBlocks + ( + mc, + macroblock->mbx, + macroblock->mby, + macroblock->dct_type, + macroblock->cbp, + macroblock->blocks + ); + + mc->num_macroblocks[mb_type_ex]++; + mc->total_num_macroblocks++; + + return 0; +} +#else +static int vlGrabMacroBlock +( + struct vlR16SnormBufferedMC *mc, + struct vlMpeg2MacroBlock *macroblock +) +{ + const struct vlVertex2f unit = + { + mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH, + mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT + }; + const struct vlVertex2f half = + { + mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2), + mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2) + }; + + struct vlVertex2f *vb; + unsigned int mb_buf_id; + struct vlVertex2f mo_vec[2]; + unsigned int i; + + assert(mc); + assert(macroblock); + + switch (macroblock->mb_type) + { + case vlMacroBlockTypeIntra: + { + mb_buf_id = vlMacroBlockExTypeIntra; + break; + } + case vlMacroBlockTypeFwdPredicted: + { + mb_buf_id = macroblock->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField; + break; + } + case vlMacroBlockTypeBkwdPredicted: + { + mb_buf_id = macroblock->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField; + break; + } + case vlMacroBlockTypeBiPredicted: + { + mb_buf_id = macroblock->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField; + break; + } + default: + assert(0); + } + + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + ( + mc->pipe->winsys, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][0].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ) + mc->num_macroblocks[mb_buf_id] * 24; + + vb[0].x = macroblock->mbx * unit.x; vb[0].y = macroblock->mby * unit.y; + vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y + half.y; + vb[2].x = macroblock->mbx * unit.x + half.x; vb[2].y = macroblock->mby * unit.y; + + vb[3].x = macroblock->mbx * unit.x + half.x; vb[3].y = macroblock->mby * unit.y; + vb[4].x = macroblock->mbx * unit.x; vb[4].y = macroblock->mby * unit.y + half.y; + vb[5].x = macroblock->mbx * unit.x + half.x; vb[5].y = macroblock->mby * unit.y + half.y; + + vb[6].x = macroblock->mbx * unit.x + half.x; vb[6].y = macroblock->mby * unit.y; + vb[7].x = macroblock->mbx * unit.x + half.x; vb[7].y = macroblock->mby * unit.y + half.y; + vb[8].x = macroblock->mbx * unit.x + unit.x; vb[8].y = macroblock->mby * unit.y; + + vb[9].x = macroblock->mbx * unit.x + unit.x; vb[9].y = macroblock->mby * unit.y; + vb[10].x = macroblock->mbx * unit.x + half.x; vb[10].y = macroblock->mby * unit.y + half.y; + vb[11].x = macroblock->mbx * unit.x + unit.x; vb[11].y = macroblock->mby * unit.y + half.y; + + vb[12].x = macroblock->mbx * unit.x; vb[12].y = macroblock->mby * unit.y + half.y; + vb[13].x = macroblock->mbx * unit.x; vb[13].y = macroblock->mby * unit.y + unit.y; + vb[14].x = macroblock->mbx * unit.x + half.x; vb[14].y = macroblock->mby * unit.y + half.y; + + vb[15].x = macroblock->mbx * unit.x + half.x; vb[15].y = macroblock->mby * unit.y + half.y; + vb[16].x = macroblock->mbx * unit.x; vb[16].y = macroblock->mby * unit.y + unit.y; + vb[17].x = macroblock->mbx * unit.x + half.x; vb[17].y = macroblock->mby * unit.y + unit.y; + + vb[18].x = macroblock->mbx * unit.x + half.x; vb[18].y = macroblock->mby * unit.y + half.y; + vb[19].x = macroblock->mbx * unit.x + half.x; vb[19].y = macroblock->mby * unit.y + unit.y; + vb[20].x = macroblock->mbx * unit.x + unit.x; vb[20].y = macroblock->mby * unit.y + half.y; + + vb[21].x = macroblock->mbx * unit.x + unit.x; vb[21].y = macroblock->mby * unit.y + half.y; + vb[22].x = macroblock->mbx * unit.x + half.x; vb[22].y = macroblock->mby * unit.y + unit.y; + vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + unit.y; + + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][0].buffer); + + if (macroblock->mb_type == vlMacroBlockTypeIntra) + { + vlGrabBlocks + ( + mc, + macroblock->mbx, + macroblock->mby, + macroblock->dct_type, + macroblock->cbp, + macroblock->blocks + ); + + mc->num_macroblocks[mb_buf_id]++; + mc->total_num_macroblocks++; + return 0; + } + + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + ( + mc->pipe->winsys, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][1].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ) + mc->num_macroblocks[mb_buf_id] * 2 * 24; + + if (macroblock->mb_type == vlMacroBlockTypeFwdPredicted || macroblock->mb_type == vlMacroBlockTypeBiPredicted) + { + mo_vec[0].x = macroblock->PMV[0][0][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[0].y = macroblock->PMV[0][0][1] * 0.5f * mc->surface_tex_inv_size.y; + + if (macroblock->mo_type == vlMotionTypeField) + { + mo_vec[1].x = macroblock->PMV[1][0][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[1].y = macroblock->PMV[1][0][1] * 0.5f * mc->surface_tex_inv_size.y; + } + } + else + { + mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; + + if (macroblock->mo_type == vlMotionTypeField) + { + mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; + } + } + + if (macroblock->mo_type == vlMotionTypeFrame) + { + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + } + } + else + { + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + vb[i + 1].x = mo_vec[1].x; + vb[i + 1].y = mo_vec[1].y; + } + } + + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][1].buffer); + + if (macroblock->mb_type != vlMacroBlockTypeBiPredicted) + { + vlGrabBlocks + ( + mc, + macroblock->mbx, + macroblock->mby, + macroblock->dct_type, + macroblock->cbp, + macroblock->blocks + ); + + mc->num_macroblocks[mb_buf_id]++; + mc->total_num_macroblocks++; + return 0; + } + + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + ( + mc->pipe->winsys, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][2].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ) + mc->num_macroblocks[mb_buf_id] * 2 * 24; + + mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; + + if (macroblock->mo_type == vlMotionTypeFrame) + { + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + } + } + else + { + mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; + mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; + + for (i = 0; i < 24 * 2; i += 2) + { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + vb[i + 1].x = mo_vec[1].x; + vb[i + 1].y = mo_vec[1].y; + } + } + + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][2].buffer); + + vlGrabBlocks + ( + mc, + macroblock->mbx, + macroblock->mby, + macroblock->dct_type, + macroblock->cbp, + macroblock->blocks + ); + + mc->num_macroblocks[mb_buf_id]++; + mc->total_num_macroblocks++; + + return 0; +} +#endif + +static int vlFlush +( + struct vlRender *render +) +{ + struct vlR16SnormBufferedMC *mc; + struct pipe_context *pipe; + struct vlVertexShaderConsts *vs_consts; + + assert(mc); + + mc = (struct vlR16SnormBufferedMC*)render; + pipe = mc->pipe; + + mc->render_target.cbufs[0] = pipe->screen->get_tex_surface + ( + pipe->screen, + mc->buffered_surface->texture, + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + ); + + pipe->set_framebuffer_state(pipe, &mc->render_target); + pipe->set_viewport_state(pipe, &mc->viewport); + vs_consts = pipe->winsys->buffer_map + ( + pipe->winsys, + mc->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE + ); + + vs_consts->denorm.x = mc->buffered_surface->texture->width[0]; + vs_consts->denorm.y = mc->buffered_surface->texture->height[0]; + + pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer); + pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf); + + if (mc->num_macroblocks[vlMacroBlockExTypeIntra] > 0) + { + pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeIntra]); + pipe->set_vertex_elements(pipe, 1, mc->vertex_elems); + pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->i_vs); + pipe->bind_fs_state(pipe, mc->i_fs); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeIntra] * 24); + } + + if (mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0) + { + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeFwdPredictedFrame]); + pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; + pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->p_vs[0]); + pipe->bind_fs_state(pipe, mc->p_fs[0]); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24); + } + + if (mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0) + { + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeFwdPredictedField]); + pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; + pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->p_vs[1]); + pipe->bind_fs_state(pipe, mc->p_fs[1]); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24); + } + + if (mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0) + { + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBkwdPredictedFrame]); + pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; + pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->p_vs[0]); + pipe->bind_fs_state(pipe, mc->p_fs[0]); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24); + } + + if (mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0) + { + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBkwdPredictedField]); + pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; + pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->p_vs[1]); + pipe->bind_fs_state(pipe, mc->p_fs[1]); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24); + } + + if (mc->num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0) + { + pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBiPredictedFrame]); + pipe->set_vertex_elements(pipe, 5, mc->vertex_elems); + mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; + mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; + pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->b_vs[0]); + pipe->bind_fs_state(pipe, mc->b_fs[0]); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24); + } + + if (mc->num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0) + { + pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBiPredictedField]); + pipe->set_vertex_elements(pipe, 5, mc->vertex_elems); + mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; + mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; + pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]); + pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers); + pipe->bind_vs_state(pipe, mc->b_vs[1]); + pipe->bind_fs_state(pipe, mc->b_fs[1]); + + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24); + } + + memset(mc->num_macroblocks, 0, sizeof(unsigned int) * 7); + mc->total_num_macroblocks = 0; + + return 0; +} + +static int vlRenderMacroBlocksMpeg2R16SnormBuffered +( + struct vlRender *render, + struct vlMpeg2MacroBlockBatch *batch, + struct vlSurface *surface +) +{ + struct vlR16SnormBufferedMC *mc; + unsigned int i; + + assert(render); + + mc = (struct vlR16SnormBufferedMC*)render; + + if (mc->buffered_surface) + { + if + ( + mc->buffered_surface != surface /*|| + mc->past_surface != batch->past_surface || + mc->future_surface != batch->future_surface*/ + ) + { + vlFlush(&mc->base); + mc->buffered_surface = surface; + mc->past_surface = batch->past_surface; + mc->future_surface = batch->future_surface; + mc->surface_tex_inv_size.x = 1.0f / surface->texture->width[0]; + mc->surface_tex_inv_size.y = 1.0f / surface->texture->height[0]; + } + } + else + { + mc->buffered_surface = surface; + mc->past_surface = batch->past_surface; + mc->future_surface = batch->future_surface; + mc->surface_tex_inv_size.x = 1.0f / surface->texture->width[0]; + mc->surface_tex_inv_size.y = 1.0f / surface->texture->height[0]; + } + + for (i = 0; i < batch->num_macroblocks; ++i) + vlGrabMacroBlock(mc, &batch->macroblocks[i]); + + return 0; +} + +static int vlEnd +( + struct vlRender *render +) +{ + assert(render); + + return 0; +} + +static int vlDestroy +( + struct vlRender *render +) +{ + struct vlR16SnormBufferedMC *mc; + struct pipe_context *pipe; + unsigned int g, h, i; + + assert(render); + + mc = (struct vlR16SnormBufferedMC*)render; + pipe = mc->pipe; + + for (i = 0; i < 5; ++i) + pipe->delete_sampler_state(pipe, mc->samplers[i]); + + for (g = 0; g < NUM_BUF_SETS; ++g) + for (h = 0; h < 7; ++h) + for (i = 0; i < 3; ++i) + pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[g][h][i].buffer); + + /* Textures 3 & 4 are not created directly, no need to release them here */ + for (i = 0; i < NUM_BUF_SETS; ++i) + { + pipe_texture_release(&mc->textures[i][0]); + pipe_texture_release(&mc->textures[i][1]); + pipe_texture_release(&mc->textures[i][2]); + } + + pipe->delete_vs_state(pipe, mc->i_vs); + pipe->delete_fs_state(pipe, mc->i_fs); + + for (i = 0; i < 2; ++i) + { + pipe->delete_vs_state(pipe, mc->p_vs[i]); + pipe->delete_fs_state(pipe, mc->p_fs[i]); + pipe->delete_vs_state(pipe, mc->b_vs[i]); + pipe->delete_fs_state(pipe, mc->b_fs[i]); + } + + pipe->winsys->buffer_destroy(pipe->winsys, mc->vs_const_buf.buffer); + pipe->winsys->buffer_destroy(pipe->winsys, mc->fs_const_buf.buffer); + + free(mc); + + return 0; +} + +/* + * Muliplier renormalizes block samples from 16 bits to 12 bits. + * Divider is used when calculating Y % 2 for choosing top or bottom + * field for P or B macroblocks. + * TODO: Use immediates. + */ +static const struct vlFragmentShaderConsts fs_consts = +{ + {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, + {0.5f, 2.0f, 0.0f, 0.0f} +}; + +static int vlCreateVertexShaderIMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 50; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos, luma & chroma texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma/chroma texcoords + */ + for (i = 0; i < 2; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i0 ; Move input luma/chroma texcoords to output + */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->i_vs = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderIMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* decl i0 ; Luma/chroma texcoords */ + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + */ + for (i = 0; i < 3; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul o0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->i_fs = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFramePMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos, luma/chroma texcoords + * decl i1 ; Ref surface top field texcoords + * decl i2 ; Ref surface bottom field texcoords (unused, packed in the same stream) + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma/chroma texcoords + * decl o2 ; Ref macroblock texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i0 ; Move input luma/chroma texcoords to output + */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* add o2, i0, i1 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->p_vs[0] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFieldPMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos, luma/chroma texcoords + * decl i1 ; Ref surface top field texcoords + * decl i2 ; Ref surface bottom field texcoords + */ + for (i = 0; i < 3; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Texcoord denorm coefficients */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma/chroma texcoords + * decl o2 ; Top field ref macroblock texcoords + * decl o3 ; Bottom field ref macroblock texcoords + * decl o4 ; Denormalized vertex pos + */ + for (i = 0; i < 5; i++) + { + decl = vl_decl_output((i == 0 || i == 5) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i0 ; Move input luma/chroma texcoords to output + */ + for (i = 0; i < 3; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i == 0 ? 0 : i - 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* + * add o2, i0, i1 ; Translate vertex pos by motion vec to form top field macroblock texcoords + * add o3, i0, i2 ; Translate vertex pos by motion vec to form bottom field macroblock texcoords + */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul o4, i0, c0 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 5, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->p_vs[1] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFramePMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0, s1, s2 + * decl i1 ; Texcoords for s3 + */ + for (i = 0; i < 2; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ + for (i = 0; i < 4; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* tex2d t1, i1, s3 ; Read texel from ref macroblock */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 1, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->p_fs[0] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFieldPMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 200; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0, s1, s2 + * decl i1 ; Texcoords for s3 + * decl i2 ; Texcoords for s3 + * decl i3 ; Denormalized vertex pos + */ + for (i = 0; i < 4; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t4 */ + decl = vl_decl_temps(0, 4); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ + for (i = 0; i < 4; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i1, s3 ; Read texel from ref macroblock top field + * tex2d t2, i2, s3 ; Read texel from ref macroblock bottom field + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 1, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* XXX: Pos values off by 0.5? */ + /* sub t4, i3.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 3, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* floor t3, t3 ; Get rid of fractional part */ + inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t3, c1.y ; Multiply by 2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->p_fs[1] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFrameBMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos, luma/chroma texcoords + * decl i1 ; First ref surface top field texcoords + * decl i2 ; First ref surface bottom field texcoords (unused, packed in the same stream) + * decl i3 ; Second ref surface top field texcoords + * decl i4 ; Second ref surface bottom field texcoords (unused, packed in the same stream) + */ + for (i = 0; i < 5; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma/chroma texcoords + * decl o2 ; First ref macroblock texcoords + * decl o3 ; Second ref macroblock texcoords + */ + for (i = 0; i < 4; i++) + { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i0 ; Move input luma/chroma texcoords to output + */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* + * add o2, i0, i1 ; Translate vertex pos by motion vec to form first ref macroblock texcoords + * add o3, i0, i3 ; Translate vertex pos by motion vec to form second ref macroblock texcoords + */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i * 2 + 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->b_vs[0] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateVertexShaderFieldBMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos, Luma/chroma texcoords + * decl i1 ; First ref surface top field texcoords + * decl i2 ; First ref surface bottom field texcoords + * decl i3 ; Second ref surface top field texcoords + * decl i4 ; Second ref surface bottom field texcoords + */ + for (i = 0; i < 5; i++) + { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Denorm coefficients */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 6); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma/chroma texcoords + * decl o2 ; Top field past ref macroblock texcoords + * decl o3 ; Bottom field past ref macroblock texcoords + * decl o4 ; Top field future ref macroblock texcoords + * decl o5 ; Bottom field future ref macroblock texcoords + * decl o6 ; Denormalized vertex pos + */ + for (i = 0; i < 7; i++) + { + decl = vl_decl_output((i == 0 || i == 7) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i0 ; Move input luma/chroma texcoords to output + * mov o2, i1 ; Move past top field texcoords to output + * mov o3, i2 ; Move past bottom field texcoords to output + * mov o4, i3 ; Move future top field texcoords to output + * mov o5, i4 ; Move future bottom field texcoords to output + */ + for (i = 0; i < 6; ++i) + { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* + * add o2, i0, i1 ; Translate vertex pos by motion vec to form first top field macroblock texcoords + * add o3, i0, i2 ; Translate vertex pos by motion vec to form first bottom field macroblock texcoords + * add o4, i0, i3 ; Translate vertex pos by motion vec to form second top field macroblock texcoords + * add o5, i0, i4 ; Translate vertex pos by motion vec to form second bottom field macroblock texcoords + */ + for (i = 0; i < 4; ++i) + { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul o6, i0, c0 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 6, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + vs.tokens = tokens; + mc->b_vs[1] = pipe->create_vs_state(pipe, &vs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFrameBMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 100; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0, s1, s2 + * decl i1 ; Texcoords for s3 + * decl i2 ; Texcoords for s4 + */ + for (i = 0; i < 3; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constant 1/2 in .x channel to use as weight to blend past and future texels + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t2 */ + decl = vl_decl_temps(0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for past ref surface texture + * decl s4 ; Sampler for future ref surface texture + */ + for (i = 0; i < 5; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i1, s3 ; Read texel from past ref macroblock + * tex2d t2, i2, s4 ; Read texel from future ref macroblock + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 1, TGSI_FILE_SAMPLER, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->b_fs[0] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateFragmentShaderFieldBMB +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int max_tokens = 200; + + struct pipe_context *pipe; + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned int ti; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); + + /* Version */ + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + /* Header */ + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + /* Processor */ + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Texcoords for s0, s1, s2 + * decl i1 ; Texcoords for s3 + * decl i2 ; Texcoords for s3 + * decl i3 ; Texcoords for s4 + * decl i4 ; Texcoords for s4 + * decl i5 ; Denormalized vertex pos + */ + for (i = 0; i < 6; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels + * ; and for Y-mod-2 top/bottom field selection + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t5 */ + decl = vl_decl_temps(0, 5); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for past ref surface texture + * decl s4 ; Sampler for future ref surface texture + */ + for (i = 0; i < 5; ++i) + { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* XXX: Pos values off by 0.5? */ + /* sub t4, i5.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 5, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* floor t3, t3 ; Get rid of fractional part */ + inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* mul t3, t3, c1.y ; Multiply by 2 */ + inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); + inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; + inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i1, s3 ; Read texel from past ref macroblock top field + * tex2d t2, i2, s3 ; Read texel from past ref macroblock bottom field + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 1, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t4, i3, s4 ; Read texel from future ref macroblock top field + * tex2d t5, i4, s4 ; Read texel from future ref macroblock bottom field + */ + for (i = 0; i < 2; ++i) + { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 4); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* TODO: Move to conditional tex fetch on t3 instead of lerp */ + /* lerp t2, t3, t4, t5 ; Choose between top and bottom fields based on Y % 2 */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + fs.tokens = tokens; + mc->b_fs[1] = pipe->create_fs_state(pipe, &fs); + free(tokens); + + return 0; +} + +static int vlCreateDataBufs +( + struct vlR16SnormBufferedMC *mc +) +{ + const unsigned int mbw = align(mc->video_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH; + const unsigned int mbh = align(mc->video_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT; + const unsigned int num_mb_per_frame = mbw * mbh; + + struct pipe_context *pipe; + unsigned int g, h, i; + + assert(mc); + + pipe = mc->pipe; + + for (g = 0; g < NUM_BUF_SETS; ++g) + { + for (h = 0; h < 7; ++h) + { + /* Create our vertex buffer and vertex buffer element */ + mc->vertex_bufs[g][h][0].pitch = sizeof(struct vlVertex2f); + mc->vertex_bufs[g][h][0].max_index = 24 * num_mb_per_frame - 1; + mc->vertex_bufs[g][h][0].buffer_offset = 0; + mc->vertex_bufs[g][h][0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 24 * num_mb_per_frame + ); + } + } + + /* Position & block luma, block chroma texcoord element */ + mc->vertex_elems[0].src_offset = 0; + mc->vertex_elems[0].vertex_buffer_index = 0; + mc->vertex_elems[0].nr_components = 2; + mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + for (g = 0; g < NUM_BUF_SETS; ++g) + { + for (h = 0; h < 7; ++h) + { + for (i = 1; i < 3; ++i) + { + mc->vertex_bufs[g][h][i].pitch = sizeof(struct vlVertex2f) * 2; + mc->vertex_bufs[g][h][i].max_index = 24 * num_mb_per_frame - 1; + mc->vertex_bufs[g][h][i].buffer_offset = 0; + mc->vertex_bufs[g][h][i].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 2 * 24 * num_mb_per_frame + ); + } + } + } + + /* First ref surface top field texcoord element */ + mc->vertex_elems[1].src_offset = 0; + mc->vertex_elems[1].vertex_buffer_index = 1; + mc->vertex_elems[1].nr_components = 2; + mc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* First ref surface bottom field texcoord element */ + mc->vertex_elems[2].src_offset = sizeof(struct vlVertex2f); + mc->vertex_elems[2].vertex_buffer_index = 1; + mc->vertex_elems[2].nr_components = 2; + mc->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Second ref surface top field texcoord element */ + mc->vertex_elems[3].src_offset = 0; + mc->vertex_elems[3].vertex_buffer_index = 2; + mc->vertex_elems[3].nr_components = 2; + mc->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Second ref surface bottom field texcoord element */ + mc->vertex_elems[4].src_offset = sizeof(struct vlVertex2f); + mc->vertex_elems[4].vertex_buffer_index = 2; + mc->vertex_elems[4].nr_components = 2; + mc->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Create our constant buffer */ + mc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts); + mc->vs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + mc->vs_const_buf.size + ); + + mc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts); + mc->fs_const_buf.buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + mc->fs_const_buf.size + ); + + memcpy + ( + pipe->winsys->buffer_map(pipe->winsys, mc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + &fs_consts, + sizeof(struct vlFragmentShaderConsts) + ); + + pipe->winsys->buffer_unmap(pipe->winsys, mc->fs_const_buf.buffer); + + return 0; +} + +static int vlInit +( + struct vlR16SnormBufferedMC *mc +) +{ + struct pipe_context *pipe; + struct pipe_sampler_state sampler; + struct pipe_texture template; + unsigned int filters[5]; + unsigned int i; + + assert(mc); + + pipe = mc->pipe; + + /* For MC we render to textures, which are rounded up to nearest POT */ + mc->viewport.scale[0] = vlRoundUpPOT(mc->video_width); + mc->viewport.scale[1] = vlRoundUpPOT(mc->video_height); + mc->viewport.scale[2] = 1; + mc->viewport.scale[3] = 1; + mc->viewport.translate[0] = 0; + mc->viewport.translate[1] = 0; + mc->viewport.translate[2] = 0; + mc->viewport.translate[3] = 0; + + mc->render_target.width = vlRoundUpPOT(mc->video_width); + mc->render_target.height = vlRoundUpPOT(mc->video_height); + mc->render_target.num_cbufs = 1; + /* FB for MC stage is a vlSurface created by the user, set at render time */ + mc->render_target.zsbuf = NULL; + + filters[0] = PIPE_TEX_FILTER_NEAREST; + filters[1] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + filters[2] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + filters[3] = PIPE_TEX_FILTER_LINEAR; + filters[4] = PIPE_TEX_FILTER_LINEAR; + + for (i = 0; i < 5; ++i) + { + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = filters[i]; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = filters[i]; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.shadow_ambient = ;*/ + /*sampler.lod_bias = ;*/ + sampler.min_lod = 0; + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + mc->samplers[i] = pipe->create_sampler_state(pipe, &sampler); + } + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + template.format = PIPE_FORMAT_R16_SNORM; + template.last_level = 0; + template.width[0] = vlRoundUpPOT(mc->video_width); + template.height[0] = vlRoundUpPOT(mc->video_height); + template.depth[0] = 1; + template.compressed = 0; + pf_get_block(template.format, &template.block); + + for (i = 0; i < NUM_BUF_SETS; ++i) + mc->textures[i][0] = pipe->screen->texture_create(pipe->screen, &template); + + if (mc->video_format == vlFormatYCbCr420) + { + template.width[0] = vlRoundUpPOT(mc->video_width / 2); + template.height[0] = vlRoundUpPOT(mc->video_height / 2); + } + else if (mc->video_format == vlFormatYCbCr422) + template.height[0] = vlRoundUpPOT(mc->video_height / 2); + + for (i = 0; i < NUM_BUF_SETS; ++i) + { + mc->textures[i][1] = pipe->screen->texture_create(pipe->screen, &template); + mc->textures[i][2] = pipe->screen->texture_create(pipe->screen, &template); + } + + /* textures[3] & textures[4] are assigned from vlSurfaces for P and B macroblocks at render time */ + + vlCreateVertexShaderIMB(mc); + vlCreateFragmentShaderIMB(mc); + vlCreateVertexShaderFramePMB(mc); + vlCreateVertexShaderFieldPMB(mc); + vlCreateFragmentShaderFramePMB(mc); + vlCreateFragmentShaderFieldPMB(mc); + vlCreateVertexShaderFrameBMB(mc); + vlCreateVertexShaderFieldBMB(mc); + vlCreateFragmentShaderFrameBMB(mc); + vlCreateFragmentShaderFieldBMB(mc); + vlCreateDataBufs(mc); + + return 0; +} + +int vlCreateR16SNormBufferedMC +( + struct pipe_context *pipe, + unsigned int video_width, + unsigned int video_height, + enum vlFormat video_format, + struct vlRender **render +) +{ + struct vlR16SnormBufferedMC *mc; + + assert(pipe); + assert(render); + + mc = calloc(1, sizeof(struct vlR16SnormBufferedMC)); + + mc->base.vlBegin = &vlBegin; + mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16SnormBuffered; + mc->base.vlEnd = &vlEnd; + mc->base.vlFlush = &vlFlush; + mc->base.vlDestroy = &vlDestroy; + mc->pipe = pipe; + mc->video_width = video_width; + mc->video_height = video_height; + + mc->cur_buf = 0; + mc->buffered_surface = NULL; + mc->past_surface = NULL; + mc->future_surface = NULL; + memset(mc->num_macroblocks, 0, sizeof(unsigned int) * 7); + mc->total_num_macroblocks = 0; + + vlInit(mc); + + *render = &mc->base; + + return 0; +} diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h new file mode 100644 index 00000000000..30f67db3e7e --- /dev/null +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h @@ -0,0 +1,18 @@ +#ifndef vl_r16snorm_mc_buf_h +#define vl_r16snorm_mc_buf_h + +#include "vl_types.h" + +struct pipe_context; +struct vlRender; + +int vlCreateR16SNormBufferedMC +( + struct pipe_context *pipe, + unsigned int video_width, + unsigned int video_height, + enum vlFormat video_format, + struct vlRender **render +); + +#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_render.h b/src/gallium/state_trackers/g3dvl/vl_render.h index 63016b5cbe0..166030b4988 100644 --- a/src/gallium/state_trackers/g3dvl/vl_render.h +++ b/src/gallium/state_trackers/g3dvl/vl_render.h @@ -24,6 +24,11 @@ struct vlRender struct vlRender *render ); + int (*vlFlush) + ( + struct vlRender *render + ); + int (*vlDestroy) ( struct vlRender *render diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index ffc81221728..687fd1ec29b 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -113,6 +113,8 @@ int vlPutPicture assert(surface); assert(surface->context); + surface->context->render->vlFlush(surface->context->render); + csc = surface->context->csc; pipe = surface->context->pipe; -- cgit v1.2.3 From 248831e741602450fa957d7f63b1ff1e1395c412 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sat, 30 Aug 2008 22:30:03 +0200 Subject: nv30: activate fp texture units when needed, to get texturing --- src/gallium/drivers/nv30/nv30_state_emit.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 40fed621b24..9c96085408e 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -71,6 +71,10 @@ nv30_state_emit(struct nv30_context *nv30) state->dirty = 0; + /* FIXME/TODO: Try to find a way to reemit only when changed */ + BEGIN_RING(rankine, NV34TCL_TX_UNITS_ENABLE, 1); + OUT_RING(state->fp_samplers); + so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { if (!(samplers & (1 << i))) -- cgit v1.2.3 From 0c47bd0374e533f614ca3013f429fc32946e5be7 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sun, 31 Aug 2008 11:10:08 +0200 Subject: nv30: set fp samplers with fragprog generation --- src/gallium/drivers/nv30/nv30_fragprog.c | 4 +++- src/gallium/drivers/nv30/nv30_state_emit.c | 4 ---- 2 files changed, 3 insertions(+), 5 deletions(-) diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c index 02f30ad9bad..320ba3f4bf4 100644 --- a/src/gallium/drivers/nv30/nv30_fragprog.c +++ b/src/gallium/drivers/nv30/nv30_fragprog.c @@ -852,7 +852,7 @@ nv30_fragprog_validate(struct nv30_context *nv30) fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4); nv30_fragprog_upload(nv30, fp); - so = so_new(6, 1); + so = so_new(8, 1); so_method(so, nv30->screen->rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1); so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR, @@ -861,6 +861,8 @@ nv30_fragprog_validate(struct nv30_context *nv30) so_data (so, fp->fp_control); so_method(so, nv30->screen->rankine, NV34TCL_FP_REG_CONTROL, 1); so_data (so, fp->fp_reg_control); + so_method(so, nv30->screen->rankine, NV34TCL_TX_UNITS_ENABLE, 1); + so_data (so, fp->samplers); so_ref(so, &fp->so); update_constants: diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 9c96085408e..40fed621b24 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -71,10 +71,6 @@ nv30_state_emit(struct nv30_context *nv30) state->dirty = 0; - /* FIXME/TODO: Try to find a way to reemit only when changed */ - BEGIN_RING(rankine, NV34TCL_TX_UNITS_ENABLE, 1); - OUT_RING(state->fp_samplers); - so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { if (!(samplers & (1 << i))) -- cgit v1.2.3 From e959c23a31c11f0bcc5775e6e0eb48c5c3d70cf3 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Fri, 29 Aug 2008 00:22:09 -0400 Subject: g3dvl: Re-enable buffer rotation, disable high quality 420->444 conversion. Using linear interpolation when upscaling the chroma blocks causes some discoloration around the edges. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 238 +-------------------- 1 file changed, 4 insertions(+), 234 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index fc383cb8f6d..2d9558587cf 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -241,7 +241,6 @@ static int vlGrabBlocks return 0; } -#if 0 static int vlGrabMacroBlock ( struct vlR16SnormBufferedMC *mc, @@ -456,237 +455,6 @@ static int vlGrabMacroBlock return 0; } -#else -static int vlGrabMacroBlock -( - struct vlR16SnormBufferedMC *mc, - struct vlMpeg2MacroBlock *macroblock -) -{ - const struct vlVertex2f unit = - { - mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH, - mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT - }; - const struct vlVertex2f half = - { - mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2), - mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2) - }; - - struct vlVertex2f *vb; - unsigned int mb_buf_id; - struct vlVertex2f mo_vec[2]; - unsigned int i; - - assert(mc); - assert(macroblock); - - switch (macroblock->mb_type) - { - case vlMacroBlockTypeIntra: - { - mb_buf_id = vlMacroBlockExTypeIntra; - break; - } - case vlMacroBlockTypeFwdPredicted: - { - mb_buf_id = macroblock->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField; - break; - } - case vlMacroBlockTypeBkwdPredicted: - { - mb_buf_id = macroblock->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField; - break; - } - case vlMacroBlockTypeBiPredicted: - { - mb_buf_id = macroblock->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField; - break; - } - default: - assert(0); - } - - vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map - ( - mc->pipe->winsys, - mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][0].buffer, - PIPE_BUFFER_USAGE_CPU_WRITE - ) + mc->num_macroblocks[mb_buf_id] * 24; - - vb[0].x = macroblock->mbx * unit.x; vb[0].y = macroblock->mby * unit.y; - vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y + half.y; - vb[2].x = macroblock->mbx * unit.x + half.x; vb[2].y = macroblock->mby * unit.y; - - vb[3].x = macroblock->mbx * unit.x + half.x; vb[3].y = macroblock->mby * unit.y; - vb[4].x = macroblock->mbx * unit.x; vb[4].y = macroblock->mby * unit.y + half.y; - vb[5].x = macroblock->mbx * unit.x + half.x; vb[5].y = macroblock->mby * unit.y + half.y; - - vb[6].x = macroblock->mbx * unit.x + half.x; vb[6].y = macroblock->mby * unit.y; - vb[7].x = macroblock->mbx * unit.x + half.x; vb[7].y = macroblock->mby * unit.y + half.y; - vb[8].x = macroblock->mbx * unit.x + unit.x; vb[8].y = macroblock->mby * unit.y; - - vb[9].x = macroblock->mbx * unit.x + unit.x; vb[9].y = macroblock->mby * unit.y; - vb[10].x = macroblock->mbx * unit.x + half.x; vb[10].y = macroblock->mby * unit.y + half.y; - vb[11].x = macroblock->mbx * unit.x + unit.x; vb[11].y = macroblock->mby * unit.y + half.y; - - vb[12].x = macroblock->mbx * unit.x; vb[12].y = macroblock->mby * unit.y + half.y; - vb[13].x = macroblock->mbx * unit.x; vb[13].y = macroblock->mby * unit.y + unit.y; - vb[14].x = macroblock->mbx * unit.x + half.x; vb[14].y = macroblock->mby * unit.y + half.y; - - vb[15].x = macroblock->mbx * unit.x + half.x; vb[15].y = macroblock->mby * unit.y + half.y; - vb[16].x = macroblock->mbx * unit.x; vb[16].y = macroblock->mby * unit.y + unit.y; - vb[17].x = macroblock->mbx * unit.x + half.x; vb[17].y = macroblock->mby * unit.y + unit.y; - - vb[18].x = macroblock->mbx * unit.x + half.x; vb[18].y = macroblock->mby * unit.y + half.y; - vb[19].x = macroblock->mbx * unit.x + half.x; vb[19].y = macroblock->mby * unit.y + unit.y; - vb[20].x = macroblock->mbx * unit.x + unit.x; vb[20].y = macroblock->mby * unit.y + half.y; - - vb[21].x = macroblock->mbx * unit.x + unit.x; vb[21].y = macroblock->mby * unit.y + half.y; - vb[22].x = macroblock->mbx * unit.x + half.x; vb[22].y = macroblock->mby * unit.y + unit.y; - vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + unit.y; - - mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][0].buffer); - - if (macroblock->mb_type == vlMacroBlockTypeIntra) - { - vlGrabBlocks - ( - mc, - macroblock->mbx, - macroblock->mby, - macroblock->dct_type, - macroblock->cbp, - macroblock->blocks - ); - - mc->num_macroblocks[mb_buf_id]++; - mc->total_num_macroblocks++; - return 0; - } - - vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map - ( - mc->pipe->winsys, - mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][1].buffer, - PIPE_BUFFER_USAGE_CPU_WRITE - ) + mc->num_macroblocks[mb_buf_id] * 2 * 24; - - if (macroblock->mb_type == vlMacroBlockTypeFwdPredicted || macroblock->mb_type == vlMacroBlockTypeBiPredicted) - { - mo_vec[0].x = macroblock->PMV[0][0][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[0].y = macroblock->PMV[0][0][1] * 0.5f * mc->surface_tex_inv_size.y; - - if (macroblock->mo_type == vlMotionTypeField) - { - mo_vec[1].x = macroblock->PMV[1][0][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[1].y = macroblock->PMV[1][0][1] * 0.5f * mc->surface_tex_inv_size.y; - } - } - else - { - mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; - - if (macroblock->mo_type == vlMotionTypeField) - { - mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; - } - } - - if (macroblock->mo_type == vlMotionTypeFrame) - { - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - } - } - else - { - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - vb[i + 1].x = mo_vec[1].x; - vb[i + 1].y = mo_vec[1].y; - } - } - - mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][1].buffer); - - if (macroblock->mb_type != vlMacroBlockTypeBiPredicted) - { - vlGrabBlocks - ( - mc, - macroblock->mbx, - macroblock->mby, - macroblock->dct_type, - macroblock->cbp, - macroblock->blocks - ); - - mc->num_macroblocks[mb_buf_id]++; - mc->total_num_macroblocks++; - return 0; - } - - vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map - ( - mc->pipe->winsys, - mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][2].buffer, - PIPE_BUFFER_USAGE_CPU_WRITE - ) + mc->num_macroblocks[mb_buf_id] * 2 * 24; - - mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; - - if (macroblock->mo_type == vlMotionTypeFrame) - { - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - } - } - else - { - mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; - - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - vb[i + 1].x = mo_vec[1].x; - vb[i + 1].y = mo_vec[1].y; - } - } - - mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_buf_id][2].buffer); - - vlGrabBlocks - ( - mc, - macroblock->mbx, - macroblock->mby, - macroblock->dct_type, - macroblock->cbp, - macroblock->blocks - ); - - mc->num_macroblocks[mb_buf_id]++; - mc->total_num_macroblocks++; - - return 0; -} -#endif static int vlFlush ( @@ -818,6 +586,7 @@ static int vlFlush memset(mc->num_macroblocks, 0, sizeof(unsigned int) * 7); mc->total_num_macroblocks = 0; + mc->cur_buf++; return 0; } @@ -2231,8 +2000,9 @@ static int vlInit mc->render_target.zsbuf = NULL; filters[0] = PIPE_TEX_FILTER_NEAREST; - filters[1] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; - filters[2] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR; + /* FIXME: Linear causes discoloration around block edges */ + filters[1] = /*mc->video_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; + filters[2] = /*mc->video_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; filters[3] = PIPE_TEX_FILTER_LINEAR; filters[4] = PIPE_TEX_FILTER_LINEAR; -- cgit v1.2.3 From 29f876cc90605ad7de1141443d3b242395eed5ee Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 30 Aug 2008 13:26:06 -0400 Subject: g3dvl: Inline hint for relatively small, frequently called functions. --- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 2d9558587cf..08aed4542e7 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -79,7 +79,7 @@ static int vlBegin return 0; } -static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) +static inline int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) { unsigned int y; @@ -94,7 +94,7 @@ static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) return 0; } -static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) +static inline int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) { unsigned int y; @@ -119,7 +119,7 @@ static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) return 0; } -static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) +static inline int vlGrabNoBlock(short *dst, unsigned int dst_pitch) { unsigned int y; @@ -130,11 +130,11 @@ static int vlGrabNoBlock(short *dst, unsigned int dst_pitch) 0, VL_BLOCK_WIDTH * 2 ); - + return 0; } -static int vlGrabBlocks +static inline int vlGrabBlocks ( struct vlR16SnormBufferedMC *mc, unsigned int mbx, @@ -164,7 +164,7 @@ static int vlGrabBlocks tex_pitch = tex_surface->stride / tex_surface->block.size; texels += mbpy * tex_pitch + mbpx; - + for (y = 0; y < 2; ++y) { for (x = 0; x < 2; ++x, ++tb) @@ -241,7 +241,7 @@ static int vlGrabBlocks return 0; } -static int vlGrabMacroBlock +static inline int vlGrabMacroBlock ( struct vlR16SnormBufferedMC *mc, struct vlMpeg2MacroBlock *macroblock -- cgit v1.2.3 From 72bcb69459d336fe0c2cf9da57a9b98a933299ca Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sun, 31 Aug 2008 01:01:51 -0400 Subject: g3dvl: Some clean ups. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 27 +++++++++++++--------- src/gallium/state_trackers/g3dvl/vl_types.h | 4 +++- 2 files changed, 19 insertions(+), 12 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 08aed4542e7..900635228f7 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -17,6 +17,15 @@ #include "vl_types.h" #include "vl_defs.h" +/* + * TODO: Dynamically determine number of buf sets to use, based on + * video size and available mem, since we can easily run out of memory + * for high res videos. + * Note: Destroying previous frame's buffers and creating new ones + * doesn't work, since the buffer are not actually destroyed until their + * fence is signalled, and if we render fast enough we will create faster + * than we destroy. + */ #define NUM_BUF_SETS 4 /* Number of rotating buffer sets to use */ enum vlMacroBlockTypeEx @@ -55,7 +64,6 @@ struct vlR16SnormBufferedMC struct vlSurface *past_surface, *future_surface; struct vlVertex2f surface_tex_inv_size; unsigned int num_macroblocks[vlNumMacroBlockExTypes]; - unsigned int total_num_macroblocks; struct pipe_context *pipe; struct pipe_viewport_state viewport; @@ -130,7 +138,7 @@ static inline int vlGrabNoBlock(short *dst, unsigned int dst_pitch) 0, VL_BLOCK_WIDTH * 2 ); - + return 0; } @@ -164,7 +172,7 @@ static inline int vlGrabBlocks tex_pitch = tex_surface->stride / tex_surface->block.size; texels += mbpy * tex_pitch + mbpx; - + for (y = 0; y < 2; ++y) { for (x = 0; x < 2; ++x, ++tb) @@ -451,7 +459,6 @@ static inline int vlGrabMacroBlock ); mc->num_macroblocks[mb_type_ex]++; - mc->total_num_macroblocks++; return 0; } @@ -584,8 +591,7 @@ static int vlFlush pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24); } - memset(mc->num_macroblocks, 0, sizeof(unsigned int) * 7); - mc->total_num_macroblocks = 0; + memset(mc->num_macroblocks, 0, sizeof(unsigned int) * vlNumMacroBlockExTypes); mc->cur_buf++; return 0; @@ -665,7 +671,7 @@ static int vlDestroy pipe->delete_sampler_state(pipe, mc->samplers[i]); for (g = 0; g < NUM_BUF_SETS; ++g) - for (h = 0; h < 7; ++h) + for (h = 0; h < vlNumMacroBlockExTypes; ++h) for (i = 0; i < 3; ++i) pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[g][h][i].buffer); @@ -1871,7 +1877,7 @@ static int vlCreateDataBufs for (g = 0; g < NUM_BUF_SETS; ++g) { - for (h = 0; h < 7; ++h) + for (h = 0; h < vlNumMacroBlockExTypes; ++h) { /* Create our vertex buffer and vertex buffer element */ mc->vertex_bufs[g][h][0].pitch = sizeof(struct vlVertex2f); @@ -1895,7 +1901,7 @@ static int vlCreateDataBufs for (g = 0; g < NUM_BUF_SETS; ++g) { - for (h = 0; h < 7; ++h) + for (h = 0; h < vlNumMacroBlockExTypes; ++h) { for (i = 1; i < 3; ++i) { @@ -2100,8 +2106,7 @@ int vlCreateR16SNormBufferedMC mc->buffered_surface = NULL; mc->past_surface = NULL; mc->future_surface = NULL; - memset(mc->num_macroblocks, 0, sizeof(unsigned int) * 7); - mc->total_num_macroblocks = 0; + memset(mc->num_macroblocks, 0, sizeof(unsigned int) * vlNumMacroBlockExTypes); vlInit(mc); diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h index 504ba8ac81a..b432bfde935 100644 --- a/src/gallium/state_trackers/g3dvl/vl_types.h +++ b/src/gallium/state_trackers/g3dvl/vl_types.h @@ -79,7 +79,9 @@ enum vlMacroBlockType vlMacroBlockTypeIntra, vlMacroBlockTypeFwdPredicted, vlMacroBlockTypeBkwdPredicted, - vlMacroBlockTypeBiPredicted + vlMacroBlockTypeBiPredicted, + + vlNumMacroBlockTypes }; struct vlMpeg2MacroBlock -- cgit v1.2.3 From ff1a5066513fc75fb0fbbe7fe8a3f1ff27fbf6d3 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 1 Sep 2008 20:13:50 -0400 Subject: g3dvl: Use one VB for all MBs, sort MBs at flush to determine placement. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 278 ++++++++++++--------- 1 file changed, 166 insertions(+), 112 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 900635228f7..b3deab46566 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -63,7 +63,8 @@ struct vlR16SnormBufferedMC struct vlSurface *buffered_surface; struct vlSurface *past_surface, *future_surface; struct vlVertex2f surface_tex_inv_size; - unsigned int num_macroblocks[vlNumMacroBlockExTypes]; + unsigned int num_macroblocks; + struct vlMpeg2MacroBlock *macroblocks; struct pipe_context *pipe; struct pipe_viewport_state viewport; @@ -72,7 +73,7 @@ struct vlR16SnormBufferedMC struct pipe_texture *textures[NUM_BUF_SETS][5]; void *i_vs, *p_vs[2], *b_vs[2]; void *i_fs, *p_fs[2], *b_fs[2]; - struct pipe_vertex_buffer vertex_bufs[NUM_BUF_SETS][vlNumMacroBlockExTypes][3]; + struct pipe_vertex_buffer vertex_bufs[NUM_BUF_SETS][3]; struct pipe_vertex_element vertex_elems[5]; struct pipe_constant_buffer vs_const_buf, fs_const_buf; }; @@ -249,11 +250,77 @@ static inline int vlGrabBlocks return 0; } +static inline enum vlMacroBlockTypeEx vlGetMacroBlockTypeEx(struct vlMpeg2MacroBlock *mb) +{ + assert(mb); + + switch (mb->mb_type) + { + case vlMacroBlockTypeIntra: + return vlMacroBlockExTypeIntra; + case vlMacroBlockTypeFwdPredicted: + return mb->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField; + case vlMacroBlockTypeBkwdPredicted: + return mb->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField; + case vlMacroBlockTypeBiPredicted: + return mb->mo_type == vlMotionTypeFrame ? + vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField; + default: + assert(0); + } + + /* Unreachable */ + return -1; +} + static inline int vlGrabMacroBlock ( struct vlR16SnormBufferedMC *mc, struct vlMpeg2MacroBlock *macroblock ) +{ + assert(mc); + assert(macroblock); + + mc->macroblocks[mc->num_macroblocks].mbx = macroblock->mbx; + mc->macroblocks[mc->num_macroblocks].mby = macroblock->mby; + mc->macroblocks[mc->num_macroblocks].mb_type = macroblock->mb_type; + mc->macroblocks[mc->num_macroblocks].mo_type = macroblock->mo_type; + mc->macroblocks[mc->num_macroblocks].dct_type = macroblock->dct_type; + mc->macroblocks[mc->num_macroblocks].PMV[0][0][0] = macroblock->PMV[0][0][0]; + mc->macroblocks[mc->num_macroblocks].PMV[0][0][1] = macroblock->PMV[0][0][1]; + mc->macroblocks[mc->num_macroblocks].PMV[0][1][0] = macroblock->PMV[0][1][0]; + mc->macroblocks[mc->num_macroblocks].PMV[0][1][1] = macroblock->PMV[0][1][1]; + mc->macroblocks[mc->num_macroblocks].PMV[1][0][0] = macroblock->PMV[1][0][0]; + mc->macroblocks[mc->num_macroblocks].PMV[1][0][1] = macroblock->PMV[1][0][1]; + mc->macroblocks[mc->num_macroblocks].PMV[1][1][0] = macroblock->PMV[1][1][0]; + mc->macroblocks[mc->num_macroblocks].PMV[1][1][1] = macroblock->PMV[1][1][1]; + mc->macroblocks[mc->num_macroblocks].cbp = macroblock->cbp; + mc->macroblocks[mc->num_macroblocks].blocks = macroblock->blocks; + + vlGrabBlocks + ( + mc, + macroblock->mbx, + macroblock->mby, + macroblock->dct_type, + macroblock->cbp, + macroblock->blocks + ); + + mc->num_macroblocks++; + + return 0; +} + +static inline int vlGrabMacroBlockVB +( + struct vlR16SnormBufferedMC *mc, + struct vlMpeg2MacroBlock *macroblock, + unsigned int pos +) { const struct vlVertex2f unit = { @@ -267,42 +334,12 @@ static inline int vlGrabMacroBlock }; struct vlVertex2f *vb; - enum vlMacroBlockTypeEx mb_type_ex; struct vlVertex2f mo_vec[2]; unsigned int i; assert(mc); assert(macroblock); - switch (macroblock->mb_type) - { - case vlMacroBlockTypeIntra: - { - mb_type_ex = vlMacroBlockExTypeIntra; - break; - } - case vlMacroBlockTypeFwdPredicted: - { - mb_type_ex = macroblock->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField; - break; - } - case vlMacroBlockTypeBkwdPredicted: - { - mb_type_ex = macroblock->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField; - break; - } - case vlMacroBlockTypeBiPredicted: - { - mb_type_ex = macroblock->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField; - break; - } - default: - assert(0); - } - switch (macroblock->mb_type) { case vlMacroBlockTypeBiPredicted: @@ -310,9 +347,9 @@ static inline int vlGrabMacroBlock vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map ( mc->pipe->winsys, - mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][2].buffer, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][2].buffer, PIPE_BUFFER_USAGE_CPU_WRITE - ) + mc->num_macroblocks[mb_type_ex] * 2 * 24; + ) + pos * 2 * 24; mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; @@ -339,7 +376,7 @@ static inline int vlGrabMacroBlock } } - mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][2].buffer); + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][2].buffer); /* fall-through */ } @@ -349,9 +386,9 @@ static inline int vlGrabMacroBlock vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map ( mc->pipe->winsys, - mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][1].buffer, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE - ) + mc->num_macroblocks[mb_type_ex] * 2 * 24; + ) + pos * 2 * 24; if (macroblock->mb_type == vlMacroBlockTypeBkwdPredicted) { @@ -395,7 +432,7 @@ static inline int vlGrabMacroBlock } } - mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][1].buffer); + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][1].buffer); /* fall-through */ } @@ -404,9 +441,9 @@ static inline int vlGrabMacroBlock vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map ( mc->pipe->winsys, - mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][0].buffer, + mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE - ) + mc->num_macroblocks[mb_type_ex] * 24; + ) + pos * 24; vb[0].x = macroblock->mbx * unit.x; vb[0].y = macroblock->mby * unit.y; vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y + half.y; @@ -440,7 +477,7 @@ static inline int vlGrabMacroBlock vb[22].x = macroblock->mbx * unit.x + half.x; vb[22].y = macroblock->mby * unit.y + unit.y; vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + unit.y; - mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][mb_type_ex][0].buffer); + mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer); break; } @@ -448,18 +485,6 @@ static inline int vlGrabMacroBlock assert(0); } - vlGrabBlocks - ( - mc, - macroblock->mbx, - macroblock->mby, - macroblock->dct_type, - macroblock->cbp, - macroblock->blocks - ); - - mc->num_macroblocks[mb_type_ex]++; - return 0; } @@ -471,12 +496,37 @@ static int vlFlush struct vlR16SnormBufferedMC *mc; struct pipe_context *pipe; struct vlVertexShaderConsts *vs_consts; + unsigned int num_macroblocks[vlNumMacroBlockExTypes] = {0}; + unsigned int offset[vlNumMacroBlockExTypes]; + unsigned int vb_start = 0; + unsigned int i; - assert(mc); + assert(render); mc = (struct vlR16SnormBufferedMC*)render; pipe = mc->pipe; + for (i = 0; i < mc->num_macroblocks; ++i) + { + enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]); + + num_macroblocks[mb_type_ex]++; + } + + offset[0] = 0; + + for (i = 1; i < vlNumMacroBlockExTypes; ++i) + offset[i] = offset[i - 1] + num_macroblocks[i - 1]; + + for (i = 0; i < mc->num_macroblocks; ++i) + { + enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]); + + vlGrabMacroBlockVB(mc, &mc->macroblocks[i], offset[mb_type_ex]); + + offset[mb_type_ex]++; + } + mc->render_target.cbufs[0] = pipe->screen->get_tex_surface ( pipe->screen, @@ -497,23 +547,25 @@ static int vlFlush vs_consts->denorm.y = mc->buffered_surface->texture->height[0]; pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer); + pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &mc->vs_const_buf); pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf); - if (mc->num_macroblocks[vlMacroBlockExTypeIntra] > 0) + if (num_macroblocks[vlMacroBlockExTypeIntra] > 0) { - pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeIntra]); + pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 1, mc->vertex_elems); pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers); pipe->bind_vs_state(pipe, mc->i_vs); pipe->bind_fs_state(pipe, mc->i_fs); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeIntra] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeIntra] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeIntra] * 24; } - if (mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0) + if (num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0) { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeFwdPredictedFrame]); + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -521,12 +573,13 @@ static int vlFlush pipe->bind_vs_state(pipe, mc->p_vs[0]); pipe->bind_fs_state(pipe, mc->p_fs[0]); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24; } - if (mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0) + if (num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0) { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeFwdPredictedField]); + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -534,12 +587,13 @@ static int vlFlush pipe->bind_vs_state(pipe, mc->p_vs[1]); pipe->bind_fs_state(pipe, mc->p_fs[1]); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24; } - if (mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0) + if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0) { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBkwdPredictedFrame]); + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -547,12 +601,13 @@ static int vlFlush pipe->bind_vs_state(pipe, mc->p_vs[0]); pipe->bind_fs_state(pipe, mc->p_fs[0]); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24; } - if (mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0) + if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0) { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBkwdPredictedField]); + pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -560,12 +615,13 @@ static int vlFlush pipe->bind_vs_state(pipe, mc->p_vs[1]); pipe->bind_fs_state(pipe, mc->p_fs[1]); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24; } - if (mc->num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0) + if (num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0) { - pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBiPredictedFrame]); + pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 5, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; @@ -574,12 +630,13 @@ static int vlFlush pipe->bind_vs_state(pipe, mc->b_vs[0]); pipe->bind_fs_state(pipe, mc->b_fs[0]); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24; } - if (mc->num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0) + if (num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0) { - pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][vlMacroBlockExTypeBiPredictedField]); + pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); pipe->set_vertex_elements(pipe, 5, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; @@ -588,10 +645,11 @@ static int vlFlush pipe->bind_vs_state(pipe, mc->b_vs[1]); pipe->bind_fs_state(pipe, mc->b_fs[1]); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, mc->num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24); + pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24); + vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24; } - memset(mc->num_macroblocks, 0, sizeof(unsigned int) * vlNumMacroBlockExTypes); + mc->num_macroblocks = 0; mc->cur_buf++; return 0; @@ -660,7 +718,7 @@ static int vlDestroy { struct vlR16SnormBufferedMC *mc; struct pipe_context *pipe; - unsigned int g, h, i; + unsigned int h, i; assert(render); @@ -670,10 +728,9 @@ static int vlDestroy for (i = 0; i < 5; ++i) pipe->delete_sampler_state(pipe, mc->samplers[i]); - for (g = 0; g < NUM_BUF_SETS; ++g) - for (h = 0; h < vlNumMacroBlockExTypes; ++h) + for (h = 0; h < NUM_BUF_SETS; ++h) for (i = 0; i < 3; ++i) - pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[g][h][i].buffer); + pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[h][i].buffer); /* Textures 3 & 4 are not created directly, no need to release them here */ for (i = 0; i < NUM_BUF_SETS; ++i) @@ -697,6 +754,7 @@ static int vlDestroy pipe->winsys->buffer_destroy(pipe->winsys, mc->vs_const_buf.buffer); pipe->winsys->buffer_destroy(pipe->winsys, mc->fs_const_buf.buffer); + free(mc->macroblocks); free(mc); return 0; @@ -1869,28 +1927,25 @@ static int vlCreateDataBufs const unsigned int num_mb_per_frame = mbw * mbh; struct pipe_context *pipe; - unsigned int g, h, i; + unsigned int h, i; assert(mc); pipe = mc->pipe; - for (g = 0; g < NUM_BUF_SETS; ++g) + for (h = 0; h < NUM_BUF_SETS; ++h) { - for (h = 0; h < vlNumMacroBlockExTypes; ++h) - { - /* Create our vertex buffer and vertex buffer element */ - mc->vertex_bufs[g][h][0].pitch = sizeof(struct vlVertex2f); - mc->vertex_bufs[g][h][0].max_index = 24 * num_mb_per_frame - 1; - mc->vertex_bufs[g][h][0].buffer_offset = 0; - mc->vertex_bufs[g][h][0].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct vlVertex2f) * 24 * num_mb_per_frame - ); - } + /* Create our vertex buffer and vertex buffer element */ + mc->vertex_bufs[h][0].pitch = sizeof(struct vlVertex2f); + mc->vertex_bufs[h][0].max_index = 24 * num_mb_per_frame - 1; + mc->vertex_bufs[h][0].buffer_offset = 0; + mc->vertex_bufs[h][0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 24 * num_mb_per_frame + ); } /* Position & block luma, block chroma texcoord element */ @@ -1899,23 +1954,20 @@ static int vlCreateDataBufs mc->vertex_elems[0].nr_components = 2; mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - for (g = 0; g < NUM_BUF_SETS; ++g) + for (h = 0; h < NUM_BUF_SETS; ++h) { - for (h = 0; h < vlNumMacroBlockExTypes; ++h) + for (i = 1; i < 3; ++i) { - for (i = 1; i < 3; ++i) - { - mc->vertex_bufs[g][h][i].pitch = sizeof(struct vlVertex2f) * 2; - mc->vertex_bufs[g][h][i].max_index = 24 * num_mb_per_frame - 1; - mc->vertex_bufs[g][h][i].buffer_offset = 0; - mc->vertex_bufs[g][h][i].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct vlVertex2f) * 2 * 24 * num_mb_per_frame - ); - } + mc->vertex_bufs[h][i].pitch = sizeof(struct vlVertex2f) * 2; + mc->vertex_bufs[h][i].max_index = 24 * num_mb_per_frame - 1; + mc->vertex_bufs[h][i].buffer_offset = 0; + mc->vertex_bufs[h][i].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 2 * 24 * num_mb_per_frame + ); } } @@ -1971,6 +2023,8 @@ static int vlCreateDataBufs pipe->winsys->buffer_unmap(pipe->winsys, mc->fs_const_buf.buffer); + mc->macroblocks = malloc(sizeof(struct vlMpeg2MacroBlock) * num_mb_per_frame); + return 0; } @@ -2106,7 +2160,7 @@ int vlCreateR16SNormBufferedMC mc->buffered_surface = NULL; mc->past_surface = NULL; mc->future_surface = NULL; - memset(mc->num_macroblocks, 0, sizeof(unsigned int) * vlNumMacroBlockExTypes); + mc->num_macroblocks = 0; vlInit(mc); -- cgit v1.2.3 From 2ab5e69f16ed9e9b399ddd526f1370032d3d10a4 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 1 Sep 2008 20:22:41 -0400 Subject: g3dvl: Use consistent variable names. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 44 +++++++++++----------- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.h | 6 +-- 2 files changed, 25 insertions(+), 25 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index b3deab46566..13c6fd55683 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -56,8 +56,8 @@ struct vlR16SnormBufferedMC { struct vlRender base; - unsigned int video_width, video_height; - enum vlFormat video_format; + unsigned int picture_width, picture_height; + enum vlFormat picture_format; unsigned int cur_buf; struct vlSurface *buffered_surface; @@ -1922,8 +1922,8 @@ static int vlCreateDataBufs struct vlR16SnormBufferedMC *mc ) { - const unsigned int mbw = align(mc->video_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH; - const unsigned int mbh = align(mc->video_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT; + const unsigned int mbw = align(mc->picture_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH; + const unsigned int mbh = align(mc->picture_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT; const unsigned int num_mb_per_frame = mbw * mbh; struct pipe_context *pipe; @@ -2044,8 +2044,8 @@ static int vlInit pipe = mc->pipe; /* For MC we render to textures, which are rounded up to nearest POT */ - mc->viewport.scale[0] = vlRoundUpPOT(mc->video_width); - mc->viewport.scale[1] = vlRoundUpPOT(mc->video_height); + mc->viewport.scale[0] = vlRoundUpPOT(mc->picture_width); + mc->viewport.scale[1] = vlRoundUpPOT(mc->picture_height); mc->viewport.scale[2] = 1; mc->viewport.scale[3] = 1; mc->viewport.translate[0] = 0; @@ -2053,16 +2053,16 @@ static int vlInit mc->viewport.translate[2] = 0; mc->viewport.translate[3] = 0; - mc->render_target.width = vlRoundUpPOT(mc->video_width); - mc->render_target.height = vlRoundUpPOT(mc->video_height); + mc->render_target.width = vlRoundUpPOT(mc->picture_width); + mc->render_target.height = vlRoundUpPOT(mc->picture_height); mc->render_target.num_cbufs = 1; /* FB for MC stage is a vlSurface created by the user, set at render time */ mc->render_target.zsbuf = NULL; filters[0] = PIPE_TEX_FILTER_NEAREST; /* FIXME: Linear causes discoloration around block edges */ - filters[1] = /*mc->video_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; - filters[2] = /*mc->video_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; + filters[1] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; + filters[2] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; filters[3] = PIPE_TEX_FILTER_LINEAR; filters[4] = PIPE_TEX_FILTER_LINEAR; @@ -2091,8 +2091,8 @@ static int vlInit template.target = PIPE_TEXTURE_2D; template.format = PIPE_FORMAT_R16_SNORM; template.last_level = 0; - template.width[0] = vlRoundUpPOT(mc->video_width); - template.height[0] = vlRoundUpPOT(mc->video_height); + template.width[0] = vlRoundUpPOT(mc->picture_width); + template.height[0] = vlRoundUpPOT(mc->picture_height); template.depth[0] = 1; template.compressed = 0; pf_get_block(template.format, &template.block); @@ -2100,13 +2100,13 @@ static int vlInit for (i = 0; i < NUM_BUF_SETS; ++i) mc->textures[i][0] = pipe->screen->texture_create(pipe->screen, &template); - if (mc->video_format == vlFormatYCbCr420) + if (mc->picture_format == vlFormatYCbCr420) { - template.width[0] = vlRoundUpPOT(mc->video_width / 2); - template.height[0] = vlRoundUpPOT(mc->video_height / 2); + template.width[0] = vlRoundUpPOT(mc->picture_width / 2); + template.height[0] = vlRoundUpPOT(mc->picture_height / 2); } - else if (mc->video_format == vlFormatYCbCr422) - template.height[0] = vlRoundUpPOT(mc->video_height / 2); + else if (mc->picture_format == vlFormatYCbCr422) + template.height[0] = vlRoundUpPOT(mc->picture_height / 2); for (i = 0; i < NUM_BUF_SETS; ++i) { @@ -2134,9 +2134,9 @@ static int vlInit int vlCreateR16SNormBufferedMC ( struct pipe_context *pipe, - unsigned int video_width, - unsigned int video_height, - enum vlFormat video_format, + unsigned int picture_width, + unsigned int picture_height, + enum vlFormat picture_format, struct vlRender **render ) { @@ -2153,8 +2153,8 @@ int vlCreateR16SNormBufferedMC mc->base.vlFlush = &vlFlush; mc->base.vlDestroy = &vlDestroy; mc->pipe = pipe; - mc->video_width = video_width; - mc->video_height = video_height; + mc->picture_width = picture_width; + mc->picture_height = picture_height; mc->cur_buf = 0; mc->buffered_surface = NULL; diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h index 30f67db3e7e..27177d64cad 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h @@ -9,9 +9,9 @@ struct vlRender; int vlCreateR16SNormBufferedMC ( struct pipe_context *pipe, - unsigned int video_width, - unsigned int video_height, - enum vlFormat video_format, + unsigned int picture_width, + unsigned int picture_height, + enum vlFormat picture_format, struct vlRender **render ); -- cgit v1.2.3 From 72b90b7894afbcb39f72dc4d76cdea0678bf4bc3 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Mon, 1 Sep 2008 22:21:08 +0200 Subject: Intialize swizzled_surface object --- src/gallium/winsys/dri/nouveau/nv04_surface.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c index 0085b1c345d..8fa3d106c8c 100644 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ b/src/gallium/winsys/dri/nouveau/nv04_surface.c @@ -276,6 +276,10 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) } BIND_RING (chan, nvc->NvSwzSurf, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + BEGIN_RING(chan, nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1); + OUT_RING (chan, nvc->channel->vram->handle); if (chipset < 0x10) { class = NV04_SCALED_IMAGE_FROM_MEMORY; -- cgit v1.2.3 From 7891efdac125185d216a8da8b044db0f06f34f0e Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 3 Sep 2008 11:50:38 -0400 Subject: g3dvl: Define block texcoords for each vertex instead of reusing pos. This is needed for zero-block optimization. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 243 +++++++++++---------- 1 file changed, 130 insertions(+), 113 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 13c6fd55683..4d778e7fe3a 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -74,7 +74,7 @@ struct vlR16SnormBufferedMC void *i_vs, *p_vs[2], *b_vs[2]; void *i_fs, *p_fs[2], *b_fs[2]; struct pipe_vertex_buffer vertex_bufs[NUM_BUF_SETS][3]; - struct pipe_vertex_element vertex_elems[5]; + struct pipe_vertex_element vertex_elems[6]; struct pipe_constant_buffer vs_const_buf, fs_const_buf; }; @@ -443,39 +443,63 @@ static inline int vlGrabMacroBlockVB mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE - ) + pos * 24; + ) + pos * 2 * 24; vb[0].x = macroblock->mbx * unit.x; vb[0].y = macroblock->mby * unit.y; - vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y + half.y; - vb[2].x = macroblock->mbx * unit.x + half.x; vb[2].y = macroblock->mby * unit.y; - - vb[3].x = macroblock->mbx * unit.x + half.x; vb[3].y = macroblock->mby * unit.y; - vb[4].x = macroblock->mbx * unit.x; vb[4].y = macroblock->mby * unit.y + half.y; - vb[5].x = macroblock->mbx * unit.x + half.x; vb[5].y = macroblock->mby * unit.y + half.y; + vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y; + vb[2].x = macroblock->mbx * unit.x; vb[2].y = macroblock->mby * unit.y + half.y; + vb[3].x = macroblock->mbx * unit.x; vb[3].y = macroblock->mby * unit.y + half.y; + vb[4].x = macroblock->mbx * unit.x + half.x; vb[4].y = macroblock->mby * unit.y; + vb[5].x = macroblock->mbx * unit.x + half.x; vb[5].y = macroblock->mby * unit.y; vb[6].x = macroblock->mbx * unit.x + half.x; vb[6].y = macroblock->mby * unit.y; - vb[7].x = macroblock->mbx * unit.x + half.x; vb[7].y = macroblock->mby * unit.y + half.y; - vb[8].x = macroblock->mbx * unit.x + unit.x; vb[8].y = macroblock->mby * unit.y; - - vb[9].x = macroblock->mbx * unit.x + unit.x; vb[9].y = macroblock->mby * unit.y; + vb[7].x = macroblock->mbx * unit.x + half.x; vb[7].y = macroblock->mby * unit.y; + vb[8].x = macroblock->mbx * unit.x; vb[8].y = macroblock->mby * unit.y + half.y; + vb[9].x = macroblock->mbx * unit.x; vb[9].y = macroblock->mby * unit.y + half.y; vb[10].x = macroblock->mbx * unit.x + half.x; vb[10].y = macroblock->mby * unit.y + half.y; - vb[11].x = macroblock->mbx * unit.x + unit.x; vb[11].y = macroblock->mby * unit.y + half.y; + vb[11].x = macroblock->mbx * unit.x + half.x; vb[11].y = macroblock->mby * unit.y + half.y; - vb[12].x = macroblock->mbx * unit.x; vb[12].y = macroblock->mby * unit.y + half.y; - vb[13].x = macroblock->mbx * unit.x; vb[13].y = macroblock->mby * unit.y + unit.y; + vb[12].x = macroblock->mbx * unit.x + half.x; vb[12].y = macroblock->mby * unit.y; + vb[13].x = macroblock->mbx * unit.x + half.x; vb[13].y = macroblock->mby * unit.y; vb[14].x = macroblock->mbx * unit.x + half.x; vb[14].y = macroblock->mby * unit.y + half.y; - vb[15].x = macroblock->mbx * unit.x + half.x; vb[15].y = macroblock->mby * unit.y + half.y; - vb[16].x = macroblock->mbx * unit.x; vb[16].y = macroblock->mby * unit.y + unit.y; - vb[17].x = macroblock->mbx * unit.x + half.x; vb[17].y = macroblock->mby * unit.y + unit.y; - - vb[18].x = macroblock->mbx * unit.x + half.x; vb[18].y = macroblock->mby * unit.y + half.y; - vb[19].x = macroblock->mbx * unit.x + half.x; vb[19].y = macroblock->mby * unit.y + unit.y; - vb[20].x = macroblock->mbx * unit.x + unit.x; vb[20].y = macroblock->mby * unit.y + half.y; - - vb[21].x = macroblock->mbx * unit.x + unit.x; vb[21].y = macroblock->mby * unit.y + half.y; - vb[22].x = macroblock->mbx * unit.x + half.x; vb[22].y = macroblock->mby * unit.y + unit.y; - vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + unit.y; + vb[16].x = macroblock->mbx * unit.x + unit.x; vb[16].y = macroblock->mby * unit.y; + vb[17].x = macroblock->mbx * unit.x + unit.x; vb[17].y = macroblock->mby * unit.y; + + vb[18].x = macroblock->mbx * unit.x + unit.x; vb[18].y = macroblock->mby * unit.y; + vb[19].x = macroblock->mbx * unit.x + unit.x; vb[19].y = macroblock->mby * unit.y; + vb[20].x = macroblock->mbx * unit.x + half.x; vb[20].y = macroblock->mby * unit.y + half.y; + vb[21].x = macroblock->mbx * unit.x + half.x; vb[21].y = macroblock->mby * unit.y + half.y; + vb[22].x = macroblock->mbx * unit.x + unit.x; vb[22].y = macroblock->mby * unit.y + half.y; + vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + half.y; + + vb[24].x = macroblock->mbx * unit.x; vb[24].y = macroblock->mby * unit.y + half.y; + vb[25].x = macroblock->mbx * unit.x; vb[25].y = macroblock->mby * unit.y + half.y; + vb[26].x = macroblock->mbx * unit.x; vb[26].y = macroblock->mby * unit.y + unit.y; + vb[27].x = macroblock->mbx * unit.x; vb[27].y = macroblock->mby * unit.y + unit.y; + vb[28].x = macroblock->mbx * unit.x + half.x; vb[28].y = macroblock->mby * unit.y + half.y; + vb[29].x = macroblock->mbx * unit.x + half.x; vb[29].y = macroblock->mby * unit.y + half.y; + + vb[30].x = macroblock->mbx * unit.x + half.x; vb[30].y = macroblock->mby * unit.y + half.y; + vb[31].x = macroblock->mbx * unit.x + half.x; vb[31].y = macroblock->mby * unit.y + half.y; + vb[32].x = macroblock->mbx * unit.x; vb[32].y = macroblock->mby * unit.y + unit.y; + vb[33].x = macroblock->mbx * unit.x; vb[33].y = macroblock->mby * unit.y + unit.y; + vb[34].x = macroblock->mbx * unit.x + half.x; vb[34].y = macroblock->mby * unit.y + unit.y; + vb[35].x = macroblock->mbx * unit.x + half.x; vb[35].y = macroblock->mby * unit.y + unit.y; + + vb[36].x = macroblock->mbx * unit.x + half.x; vb[36].y = macroblock->mby * unit.y + half.y; + vb[37].x = macroblock->mbx * unit.x + half.x; vb[37].y = macroblock->mby * unit.y + half.y; + vb[38].x = macroblock->mbx * unit.x + half.x; vb[38].y = macroblock->mby * unit.y + unit.y; + vb[39].x = macroblock->mbx * unit.x + half.x; vb[39].y = macroblock->mby * unit.y + unit.y; + vb[40].x = macroblock->mbx * unit.x + unit.x; vb[40].y = macroblock->mby * unit.y + half.y; + vb[41].x = macroblock->mbx * unit.x + unit.x; vb[41].y = macroblock->mby * unit.y + half.y; + + vb[42].x = macroblock->mbx * unit.x + unit.x; vb[42].y = macroblock->mby * unit.y + half.y; + vb[43].x = macroblock->mbx * unit.x + unit.x; vb[43].y = macroblock->mby * unit.y + half.y; + vb[44].x = macroblock->mbx * unit.x + half.x; vb[44].y = macroblock->mby * unit.y + unit.y; + vb[45].x = macroblock->mbx * unit.x + half.x; vb[45].y = macroblock->mby * unit.y + unit.y; + vb[46].x = macroblock->mbx * unit.x + unit.x; vb[46].y = macroblock->mby * unit.y + unit.y; + vb[47].x = macroblock->mbx * unit.x + unit.x; vb[47].y = macroblock->mby * unit.y + unit.y; mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer); @@ -553,7 +577,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeIntra] > 0) { pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 1, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 2, mc->vertex_elems); pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers); pipe->bind_vs_state(pipe, mc->i_vs); @@ -566,7 +590,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -580,7 +604,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -594,7 +618,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -608,7 +632,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 3, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -622,7 +646,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0) { pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 5, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -637,7 +661,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0) { pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 5, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -806,9 +830,10 @@ static int vlCreateVertexShaderIMB ti = 3; /* - * decl i0 ; Vertex pos, luma & chroma texcoords + * decl i0 ; Vertex pos + * decl i1 ; Luma/chroma texcoords */ - for (i = 0; i < 3; i++) + for (i = 0; i < 2; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -826,11 +851,11 @@ static int vlCreateVertexShaderIMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i0 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma/chroma texcoords to output */ for (i = 0; i < 2; ++i) { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -976,11 +1001,12 @@ static int vlCreateVertexShaderFramePMB ti = 3; /* - * decl i0 ; Vertex pos, luma/chroma texcoords - * decl i1 ; Ref surface top field texcoords - * decl i2 ; Ref surface bottom field texcoords (unused, packed in the same stream) + * decl i0 ; Vertex pos + * decl i1 ; Luma/chroma texcoords + * decl i2 ; Ref surface top field texcoords + * decl i3 ; Ref surface bottom field texcoords (unused, packed in the same stream) */ - for (i = 0; i < 3; i++) + for (i = 0; i < 4; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -999,16 +1025,16 @@ static int vlCreateVertexShaderFramePMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i0 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma/chroma texcoords to output */ for (i = 0; i < 2; ++i) { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* add o2, i0, i1 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 1); + /* add o2, i0, i2 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* end */ @@ -1057,10 +1083,11 @@ static int vlCreateVertexShaderFieldPMB /* * decl i0 ; Vertex pos, luma/chroma texcoords - * decl i1 ; Ref surface top field texcoords - * decl i2 ; Ref surface bottom field texcoords + * decl i1 ; Texcoord denorm coefficients + * decl i2 ; Ref surface top field texcoords + * decl i3 ; Ref surface bottom field texcoords */ - for (i = 0; i < 3; i++) + for (i = 0; i < 4; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1085,21 +1112,21 @@ static int vlCreateVertexShaderFieldPMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i0 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma/chroma texcoords to output */ - for (i = 0; i < 3; ++i) + for (i = 0; i < 2; ++i) { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i == 0 ? 0 : i - 1); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* - * add o2, i0, i1 ; Translate vertex pos by motion vec to form top field macroblock texcoords - * add o3, i0, i2 ; Translate vertex pos by motion vec to form bottom field macroblock texcoords + * add o2, i0, i2 ; Translate vertex pos by motion vec to form top field macroblock texcoords + * add o3, i0, i3 ; Translate vertex pos by motion vec to form bottom field macroblock texcoords */ for (i = 0; i < 2; ++i) { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 1); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1430,12 +1457,13 @@ static int vlCreateVertexShaderFrameBMB /* * decl i0 ; Vertex pos, luma/chroma texcoords - * decl i1 ; First ref surface top field texcoords - * decl i2 ; First ref surface bottom field texcoords (unused, packed in the same stream) - * decl i3 ; Second ref surface top field texcoords - * decl i4 ; Second ref surface bottom field texcoords (unused, packed in the same stream) + * decl i1 ; Luma/chroma texcoords + * decl i2 ; First ref surface top field texcoords + * decl i3 ; First ref surface bottom field texcoords (unused, packed in the same stream) + * decl i4 ; Second ref surface top field texcoords + * decl i5 ; Second ref surface bottom field texcoords (unused, packed in the same stream) */ - for (i = 0; i < 5; i++) + for (i = 0; i < 6; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1455,21 +1483,21 @@ static int vlCreateVertexShaderFrameBMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i0 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma/chroma texcoords to output */ for (i = 0; i < 2; ++i) { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* - * add o2, i0, i1 ; Translate vertex pos by motion vec to form first ref macroblock texcoords - * add o3, i0, i3 ; Translate vertex pos by motion vec to form second ref macroblock texcoords + * add o2, i0, i2 ; Translate vertex pos by motion vec to form first ref macroblock texcoords + * add o3, i0, i4 ; Translate vertex pos by motion vec to form second ref macroblock texcoords */ for (i = 0; i < 2; ++i) { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i * 2 + 1); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 1) * 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1519,12 +1547,13 @@ static int vlCreateVertexShaderFieldBMB /* * decl i0 ; Vertex pos, Luma/chroma texcoords - * decl i1 ; First ref surface top field texcoords - * decl i2 ; First ref surface bottom field texcoords - * decl i3 ; Second ref surface top field texcoords - * decl i4 ; Second ref surface bottom field texcoords + * decl i1 ; Luma/chroma texcoords + * decl i2 ; First ref surface top field texcoords + * decl i3 ; First ref surface bottom field texcoords + * decl i4 ; Second ref surface top field texcoords + * decl i5 ; Second ref surface bottom field texcoords */ - for (i = 0; i < 5; i++) + for (i = 0; i < 6; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1555,27 +1584,23 @@ static int vlCreateVertexShaderFieldBMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i0 ; Move input luma/chroma texcoords to output - * mov o2, i1 ; Move past top field texcoords to output - * mov o3, i2 ; Move past bottom field texcoords to output - * mov o4, i3 ; Move future top field texcoords to output - * mov o5, i4 ; Move future bottom field texcoords to output + * mov o1, i1 ; Move input luma/chroma texcoords to output */ - for (i = 0; i < 6; ++i) + for (i = 0; i < 2; ++i) { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, 0); + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* - * add o2, i0, i1 ; Translate vertex pos by motion vec to form first top field macroblock texcoords - * add o3, i0, i2 ; Translate vertex pos by motion vec to form first bottom field macroblock texcoords - * add o4, i0, i3 ; Translate vertex pos by motion vec to form second top field macroblock texcoords - * add o5, i0, i4 ; Translate vertex pos by motion vec to form second bottom field macroblock texcoords + * add o2, i0, i2 ; Translate vertex pos by motion vec to form first top field macroblock texcoords + * add o3, i0, i3 ; Translate vertex pos by motion vec to form first bottom field macroblock texcoords + * add o4, i0, i4 ; Translate vertex pos by motion vec to form second top field macroblock texcoords + * add o5, i0, i5 ; Translate vertex pos by motion vec to form second bottom field macroblock texcoords */ for (i = 0; i < 4; ++i) { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 1); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1933,30 +1958,10 @@ static int vlCreateDataBufs pipe = mc->pipe; + /* Create our vertex buffers */ for (h = 0; h < NUM_BUF_SETS; ++h) { - /* Create our vertex buffer and vertex buffer element */ - mc->vertex_bufs[h][0].pitch = sizeof(struct vlVertex2f); - mc->vertex_bufs[h][0].max_index = 24 * num_mb_per_frame - 1; - mc->vertex_bufs[h][0].buffer_offset = 0; - mc->vertex_bufs[h][0].buffer = pipe->winsys->buffer_create - ( - pipe->winsys, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct vlVertex2f) * 24 * num_mb_per_frame - ); - } - - /* Position & block luma, block chroma texcoord element */ - mc->vertex_elems[0].src_offset = 0; - mc->vertex_elems[0].vertex_buffer_index = 0; - mc->vertex_elems[0].nr_components = 2; - mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - - for (h = 0; h < NUM_BUF_SETS; ++h) - { - for (i = 1; i < 3; ++i) + for (i = 0; i < 3; ++i) { mc->vertex_bufs[h][i].pitch = sizeof(struct vlVertex2f) * 2; mc->vertex_bufs[h][i].max_index = 24 * num_mb_per_frame - 1; @@ -1971,30 +1976,42 @@ static int vlCreateDataBufs } } - /* First ref surface top field texcoord element */ - mc->vertex_elems[1].src_offset = 0; - mc->vertex_elems[1].vertex_buffer_index = 1; + /* Position element */ + mc->vertex_elems[0].src_offset = 0; + mc->vertex_elems[0].vertex_buffer_index = 0; + mc->vertex_elems[0].nr_components = 2; + mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Block luma, block chroma texcoord element */ + mc->vertex_elems[1].src_offset = sizeof(struct vlVertex2f); + mc->vertex_elems[1].vertex_buffer_index = 0; mc->vertex_elems[1].nr_components = 2; mc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* First ref surface bottom field texcoord element */ - mc->vertex_elems[2].src_offset = sizeof(struct vlVertex2f); + /* First ref surface top field texcoord element */ + mc->vertex_elems[2].src_offset = 0; mc->vertex_elems[2].vertex_buffer_index = 1; mc->vertex_elems[2].nr_components = 2; mc->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* Second ref surface top field texcoord element */ - mc->vertex_elems[3].src_offset = 0; - mc->vertex_elems[3].vertex_buffer_index = 2; + /* First ref surface bottom field texcoord element */ + mc->vertex_elems[3].src_offset = sizeof(struct vlVertex2f); + mc->vertex_elems[3].vertex_buffer_index = 1; mc->vertex_elems[3].nr_components = 2; mc->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* Second ref surface bottom field texcoord element */ - mc->vertex_elems[4].src_offset = sizeof(struct vlVertex2f); + /* Second ref surface top field texcoord element */ + mc->vertex_elems[4].src_offset = 0; mc->vertex_elems[4].vertex_buffer_index = 2; mc->vertex_elems[4].nr_components = 2; mc->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT; + /* Second ref surface bottom field texcoord element */ + mc->vertex_elems[5].src_offset = sizeof(struct vlVertex2f); + mc->vertex_elems[5].vertex_buffer_index = 2; + mc->vertex_elems[5].nr_components = 2; + mc->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT; + /* Create our constant buffer */ mc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts); mc->vs_const_buf.buffer = pipe->winsys->buffer_create -- cgit v1.2.3 From 0bfbe834d35946fe75eb991d03ed777e115f418d Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 4 Sep 2008 00:16:51 -0400 Subject: g3dvl: Define texcoords seperately for luma, Cb, Cr textures. Need to be able to address each texture separately to do zero-block. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 566 +++++++++++++-------- 1 file changed, 343 insertions(+), 223 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 4d778e7fe3a..b5aa79cdb44 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -74,7 +74,7 @@ struct vlR16SnormBufferedMC void *i_vs, *p_vs[2], *b_vs[2]; void *i_fs, *p_fs[2], *b_fs[2]; struct pipe_vertex_buffer vertex_bufs[NUM_BUF_SETS][3]; - struct pipe_vertex_element vertex_elems[6]; + struct pipe_vertex_element vertex_elems[8]; struct pipe_constant_buffer vs_const_buf, fs_const_buf; }; @@ -315,6 +315,71 @@ static inline int vlGrabMacroBlock return 0; } +#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, zx, zy) \ + (vb)[0].pos.x = (mbx) * (unitx) + (ofsx); (vb)[0].pos.y = (mby) * (unity) + (ofsy); \ + (vb)[1].pos.x = (mbx) * (unitx) + (ofsx); (vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].pos.y = (mby) * (unity) + (ofsy); \ + (vb)[3].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].pos.y = (mby) * (unity) + (ofsy); \ + (vb)[4].pos.x = (mbx) * (unitx) + (ofsx); (vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy); \ + \ + /*if ((cbp) & (lm)) \ + {*/ \ + (vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].luma_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].luma_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].luma_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + /*} \ + else \ + { \ + (vb)[0].luma_tc.x = (zx); (vb)[0].luma_tc.y = (zy); \ + (vb)[1].luma_tc.x = (zx); (vb)[1].luma_tc.y = (zy) + (hy); \ + (vb)[2].luma_tc.x = (zx) + (hx); (vb)[2].luma_tc.y = (zy); \ + (vb)[3].luma_tc.x = (zx) + (hx); (vb)[3].luma_tc.y = (zy); \ + (vb)[4].luma_tc.x = (zx); (vb)[4].luma_tc.y = (zy) + (hy); \ + (vb)[5].luma_tc.x = ((zx) + (hx); (vb)[5].luma_tc.y = (zy) + (hy); \ + }*/ \ + \ + /*if ((cbp) & (cbm)) \ + {*/ \ + (vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cb_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cb_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cb_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + /*} \ + else \ + { \ + (vb)[0].cb_tc.x = (zx); (vb)[0].cb_tc.y = (zy); \ + (vb)[1].cb_tc.x = (zx); (vb)[1].cb_tc.y = (zy) + (hy); \ + (vb)[2].cb_tc.x = (zx) + (hx); (vb)[2].cb_tc.y = (zy); \ + (vb)[3].cb_tc.x = (zx) + (hx); (vb)[3].cb_tc.y = (zy); \ + (vb)[4].cb_tc.x = (zx); (vb)[4].cb_tc.y = (zy) + (hy); \ + (vb)[5].cb_tc.x = ((zx) + (hx); (vb)[5].cb_tc.y = (zy) + (hy); \ + }*/ \ + \ + /*if ((cbp) & (crm)) \ + {*/ \ + (vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cr_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cr_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cr_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + /*} \ + else \ + { \ + (vb)[0].cr_tc.x = (zx); (vb)[0].cb_tc.y = (zy); \ + (vb)[1].cr_tc.x = (zx); (vb)[1].cb_tc.y = (zy) + (hy); \ + (vb)[2].cr_tc.x = (zx) + (hx); (vb)[2].cb_tc.y = (zy); \ + (vb)[3].cr_tc.x = (zx) + (hx); (vb)[3].cb_tc.y = (zy); \ + (vb)[4].cr_tc.x = (zx); (vb)[4].cb_tc.y = (zy) + (hy); \ + (vb)[5].cr_tc.x = ((zx) + (hx); (vb)[5].cb_tc.y = (zy) + (hy); \ + }*/ + static inline int vlGrabMacroBlockVB ( struct vlR16SnormBufferedMC *mc, @@ -322,18 +387,6 @@ static inline int vlGrabMacroBlockVB unsigned int pos ) { - const struct vlVertex2f unit = - { - mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH, - mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT - }; - const struct vlVertex2f half = - { - mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2), - mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2) - }; - - struct vlVertex2f *vb; struct vlVertex2f mo_vec[2]; unsigned int i; @@ -344,6 +397,8 @@ static inline int vlGrabMacroBlockVB { case vlMacroBlockTypeBiPredicted: { + struct vlVertex2f *vb; + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map ( mc->pipe->winsys, @@ -383,6 +438,8 @@ static inline int vlGrabMacroBlockVB case vlMacroBlockTypeFwdPredicted: case vlMacroBlockTypeBkwdPredicted: { + struct vlVertex2f *vb; + vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map ( mc->pipe->winsys, @@ -438,68 +495,63 @@ static inline int vlGrabMacroBlockVB } case vlMacroBlockTypeIntra: { - vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map + const struct vlVertex2f unit = + { + mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH, + mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT + }; + const struct vlVertex2f half = + { + mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2), + mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2) + }; + + struct vlMacroBlockVertexStream0 + { + struct vlVertex2f pos; + struct vlVertex2f luma_tc; + struct vlVertex2f cb_tc; + struct vlVertex2f cr_tc; + } *vb; + + vb = (struct vlMacroBlockVertexStream0*)mc->pipe->winsys->buffer_map ( mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE - ) + pos * 2 * 24; + ) + pos * 24; - vb[0].x = macroblock->mbx * unit.x; vb[0].y = macroblock->mby * unit.y; - vb[1].x = macroblock->mbx * unit.x; vb[1].y = macroblock->mby * unit.y; - vb[2].x = macroblock->mbx * unit.x; vb[2].y = macroblock->mby * unit.y + half.y; - vb[3].x = macroblock->mbx * unit.x; vb[3].y = macroblock->mby * unit.y + half.y; - vb[4].x = macroblock->mbx * unit.x + half.x; vb[4].y = macroblock->mby * unit.y; - vb[5].x = macroblock->mbx * unit.x + half.x; vb[5].y = macroblock->mby * unit.y; - - vb[6].x = macroblock->mbx * unit.x + half.x; vb[6].y = macroblock->mby * unit.y; - vb[7].x = macroblock->mbx * unit.x + half.x; vb[7].y = macroblock->mby * unit.y; - vb[8].x = macroblock->mbx * unit.x; vb[8].y = macroblock->mby * unit.y + half.y; - vb[9].x = macroblock->mbx * unit.x; vb[9].y = macroblock->mby * unit.y + half.y; - vb[10].x = macroblock->mbx * unit.x + half.x; vb[10].y = macroblock->mby * unit.y + half.y; - vb[11].x = macroblock->mbx * unit.x + half.x; vb[11].y = macroblock->mby * unit.y + half.y; - - vb[12].x = macroblock->mbx * unit.x + half.x; vb[12].y = macroblock->mby * unit.y; - vb[13].x = macroblock->mbx * unit.x + half.x; vb[13].y = macroblock->mby * unit.y; - vb[14].x = macroblock->mbx * unit.x + half.x; vb[14].y = macroblock->mby * unit.y + half.y; - vb[15].x = macroblock->mbx * unit.x + half.x; vb[15].y = macroblock->mby * unit.y + half.y; - vb[16].x = macroblock->mbx * unit.x + unit.x; vb[16].y = macroblock->mby * unit.y; - vb[17].x = macroblock->mbx * unit.x + unit.x; vb[17].y = macroblock->mby * unit.y; - - vb[18].x = macroblock->mbx * unit.x + unit.x; vb[18].y = macroblock->mby * unit.y; - vb[19].x = macroblock->mbx * unit.x + unit.x; vb[19].y = macroblock->mby * unit.y; - vb[20].x = macroblock->mbx * unit.x + half.x; vb[20].y = macroblock->mby * unit.y + half.y; - vb[21].x = macroblock->mbx * unit.x + half.x; vb[21].y = macroblock->mby * unit.y + half.y; - vb[22].x = macroblock->mbx * unit.x + unit.x; vb[22].y = macroblock->mby * unit.y + half.y; - vb[23].x = macroblock->mbx * unit.x + unit.x; vb[23].y = macroblock->mby * unit.y + half.y; - - vb[24].x = macroblock->mbx * unit.x; vb[24].y = macroblock->mby * unit.y + half.y; - vb[25].x = macroblock->mbx * unit.x; vb[25].y = macroblock->mby * unit.y + half.y; - vb[26].x = macroblock->mbx * unit.x; vb[26].y = macroblock->mby * unit.y + unit.y; - vb[27].x = macroblock->mbx * unit.x; vb[27].y = macroblock->mby * unit.y + unit.y; - vb[28].x = macroblock->mbx * unit.x + half.x; vb[28].y = macroblock->mby * unit.y + half.y; - vb[29].x = macroblock->mbx * unit.x + half.x; vb[29].y = macroblock->mby * unit.y + half.y; - - vb[30].x = macroblock->mbx * unit.x + half.x; vb[30].y = macroblock->mby * unit.y + half.y; - vb[31].x = macroblock->mbx * unit.x + half.x; vb[31].y = macroblock->mby * unit.y + half.y; - vb[32].x = macroblock->mbx * unit.x; vb[32].y = macroblock->mby * unit.y + unit.y; - vb[33].x = macroblock->mbx * unit.x; vb[33].y = macroblock->mby * unit.y + unit.y; - vb[34].x = macroblock->mbx * unit.x + half.x; vb[34].y = macroblock->mby * unit.y + unit.y; - vb[35].x = macroblock->mbx * unit.x + half.x; vb[35].y = macroblock->mby * unit.y + unit.y; - - vb[36].x = macroblock->mbx * unit.x + half.x; vb[36].y = macroblock->mby * unit.y + half.y; - vb[37].x = macroblock->mbx * unit.x + half.x; vb[37].y = macroblock->mby * unit.y + half.y; - vb[38].x = macroblock->mbx * unit.x + half.x; vb[38].y = macroblock->mby * unit.y + unit.y; - vb[39].x = macroblock->mbx * unit.x + half.x; vb[39].y = macroblock->mby * unit.y + unit.y; - vb[40].x = macroblock->mbx * unit.x + unit.x; vb[40].y = macroblock->mby * unit.y + half.y; - vb[41].x = macroblock->mbx * unit.x + unit.x; vb[41].y = macroblock->mby * unit.y + half.y; - - vb[42].x = macroblock->mbx * unit.x + unit.x; vb[42].y = macroblock->mby * unit.y + half.y; - vb[43].x = macroblock->mbx * unit.x + unit.x; vb[43].y = macroblock->mby * unit.y + half.y; - vb[44].x = macroblock->mbx * unit.x + half.x; vb[44].y = macroblock->mby * unit.y + unit.y; - vb[45].x = macroblock->mbx * unit.x + half.x; vb[45].y = macroblock->mby * unit.y + unit.y; - vb[46].x = macroblock->mbx * unit.x + unit.x; vb[46].y = macroblock->mby * unit.y + unit.y; - vb[47].x = macroblock->mbx * unit.x + unit.x; vb[47].y = macroblock->mby * unit.y + unit.y; + SET_BLOCK + ( + vb, + macroblock->cbp, macroblock->mbx, macroblock->mby, + unit.x, unit.y, 0, 0, half.x, half.y, + 32, 2, 1, mc->zero_block.x, mc->zero_block.y + ); + + SET_BLOCK + ( + vb + 6, + macroblock->cbp, macroblock->mbx, macroblock->mby, + unit.x, unit.y, half.x, 0, half.x, half.y, + 16, 2, 1, mc->zero_block.x, mc->zero_block.y + ); + + SET_BLOCK + ( + vb + 12, + macroblock->cbp, macroblock->mbx, macroblock->mby, + unit.x, unit.y, 0, half.y, half.x, half.y, + 8, 2, 1, mc->zero_block.x, mc->zero_block.y + ); + + SET_BLOCK + ( + vb + 18, + macroblock->cbp, macroblock->mbx, macroblock->mby, + unit.x, unit.y, half.x, half.y, half.x, half.y, + 4, 2, 1, mc->zero_block.x, mc->zero_block.y + ); mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer); @@ -577,7 +629,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeIntra] > 0) { pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 2, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers); pipe->bind_vs_state(pipe, mc->i_vs); @@ -590,7 +642,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -604,7 +656,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -618,7 +670,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -632,7 +684,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0) { pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]); pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers); @@ -646,7 +698,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0) { pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 8, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -661,7 +713,7 @@ static int vlFlush if (num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0) { pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]); - pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); + pipe->set_vertex_elements(pipe, 8, mc->vertex_elems); mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture; mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture; pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]); @@ -831,9 +883,11 @@ static int vlCreateVertexShaderIMB /* * decl i0 ; Vertex pos - * decl i1 ; Luma/chroma texcoords + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords */ - for (i = 0; i < 2; i++) + for (i = 0; i < 4; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -841,9 +895,11 @@ static int vlCreateVertexShaderIMB /* * decl o0 ; Vertex pos - * decl o1 ; Luma/chroma texcoords + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords */ - for (i = 0; i < 2; i++) + for (i = 0; i < 4; i++) { decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -851,9 +907,11 @@ static int vlCreateVertexShaderIMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output */ - for (i = 0; i < 2; ++i) + for (i = 0; i < 4; ++i) { inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); @@ -903,9 +961,16 @@ static int vlCreateFragmentShaderIMB ti = 3; - /* decl i0 ; Luma/chroma texcoords */ - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + /* + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + */ + for (i = 0; i < 3; ++i) + { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); @@ -933,14 +998,14 @@ static int vlCreateFragmentShaderIMB /* * tex2d t1, i0, s0 ; Read texel from luma texture * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); @@ -1002,11 +1067,13 @@ static int vlCreateVertexShaderFramePMB /* * decl i0 ; Vertex pos - * decl i1 ; Luma/chroma texcoords - * decl i2 ; Ref surface top field texcoords - * decl i3 ; Ref surface bottom field texcoords (unused, packed in the same stream) + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + * decl i4 ; Ref surface top field texcoords + * decl i5 ; Ref surface bottom field texcoords (unused, packed in the same stream) */ - for (i = 0; i < 4; i++) + for (i = 0; i < 6; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1014,10 +1081,12 @@ static int vlCreateVertexShaderFramePMB /* * decl o0 ; Vertex pos - * decl o1 ; Luma/chroma texcoords - * decl o2 ; Ref macroblock texcoords + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + * decl o4 ; Ref macroblock texcoords */ - for (i = 0; i < 3; i++) + for (i = 0; i < 5; i++) { decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1025,16 +1094,18 @@ static int vlCreateVertexShaderFramePMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output */ - for (i = 0; i < 2; ++i) + for (i = 0; i < 4; ++i) { inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* add o2, i0, i2 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 2); + /* add o4, i0, i4 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 4); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* end */ @@ -1082,12 +1153,14 @@ static int vlCreateVertexShaderFieldPMB ti = 3; /* - * decl i0 ; Vertex pos, luma/chroma texcoords - * decl i1 ; Texcoord denorm coefficients - * decl i2 ; Ref surface top field texcoords - * decl i3 ; Ref surface bottom field texcoords + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + * decl i4 ; Ref macroblock top field texcoords + * decl i5 ; Ref macroblock bottom field texcoords */ - for (i = 0; i < 4; i++) + for (i = 0; i < 6; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1099,39 +1172,43 @@ static int vlCreateVertexShaderFieldPMB /* * decl o0 ; Vertex pos - * decl o1 ; Luma/chroma texcoords - * decl o2 ; Top field ref macroblock texcoords - * decl o3 ; Bottom field ref macroblock texcoords - * decl o4 ; Denormalized vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + * decl o4 ; Ref macroblock top field texcoords + * decl o5 ; Ref macroblock bottom field texcoords + * decl o6 ; Denormalized vertex pos */ - for (i = 0; i < 5; i++) + for (i = 0; i < 7; i++) { - decl = vl_decl_output((i == 0 || i == 5) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + decl = vl_decl_output((i == 0 || i == 6) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output */ - for (i = 0; i < 2; ++i) + for (i = 0; i < 4; ++i) { inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* - * add o2, i0, i2 ; Translate vertex pos by motion vec to form top field macroblock texcoords - * add o3, i0, i3 ; Translate vertex pos by motion vec to form bottom field macroblock texcoords + * add o4, i0, i4 ; Translate vertex pos by motion vec to form top field macroblock texcoords + * add o5, i0, i5 ; Translate vertex pos by motion vec to form bottom field macroblock texcoords */ for (i = 0; i < 2; ++i) { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 2); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul o4, i0, c0 ; Denorm vertex pos */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 5, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + /* mul o6, i0, c0 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 6, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* end */ @@ -1179,10 +1256,12 @@ static int vlCreateFragmentShaderFramePMB ti = 3; /* - * decl i0 ; Texcoords for s0, s1, s2 - * decl i1 ; Texcoords for s3 + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + * decl i3 ; Ref macroblock texcoords */ - for (i = 0; i < 2; ++i) + for (i = 0; i < 4; ++i) { decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1215,14 +1294,14 @@ static int vlCreateFragmentShaderFramePMB /* * tex2d t1, i0, s0 ; Read texel from luma texture * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); @@ -1238,8 +1317,8 @@ static int vlCreateFragmentShaderFramePMB inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - /* tex2d t1, i1, s3 ; Read texel from ref macroblock */ - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 1, TGSI_FILE_SAMPLER, 3); + /* tex2d t1, i3, s3 ; Read texel from ref macroblock */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 3, TGSI_FILE_SAMPLER, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* add o0, t0, t1 ; Add ref and differential to form final output */ @@ -1291,12 +1370,14 @@ static int vlCreateFragmentShaderFieldPMB ti = 3; /* - * decl i0 ; Texcoords for s0, s1, s2 - * decl i1 ; Texcoords for s3 - * decl i2 ; Texcoords for s3 - * decl i3 ; Denormalized vertex pos + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + * decl i3 ; Ref macroblock top field texcoords + * decl i4 ; Ref macroblock bottom field texcoords + * decl i5 ; Denormalized vertex pos */ - for (i = 0; i < 4; ++i) + for (i = 0; i < 6; ++i) { decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1332,14 +1413,14 @@ static int vlCreateFragmentShaderFieldPMB /* * tex2d t1, i0, s0 ; Read texel from luma texture * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); @@ -1356,18 +1437,18 @@ static int vlCreateFragmentShaderFieldPMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - * tex2d t1, i1, s3 ; Read texel from ref macroblock top field - * tex2d t2, i2, s3 ; Read texel from ref macroblock bottom field + * tex2d t1, i3, s3 ; Read texel from ref macroblock top field + * tex2d t2, i4, s3 ; Read texel from ref macroblock bottom field */ for (i = 0; i < 2; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 1, TGSI_FILE_SAMPLER, 3); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* XXX: Pos values off by 0.5? */ - /* sub t4, i3.y, c1.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 3, TGSI_FILE_CONSTANT, 1); + /* sub t4, i5.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 5, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; @@ -1456,14 +1537,16 @@ static int vlCreateVertexShaderFrameBMB ti = 3; /* - * decl i0 ; Vertex pos, luma/chroma texcoords - * decl i1 ; Luma/chroma texcoords - * decl i2 ; First ref surface top field texcoords - * decl i3 ; First ref surface bottom field texcoords (unused, packed in the same stream) - * decl i4 ; Second ref surface top field texcoords - * decl i5 ; Second ref surface bottom field texcoords (unused, packed in the same stream) + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + * decl i4 ; First ref macroblock top field texcoords + * decl i5 ; First ref macroblock bottom field texcoords (unused, packed in the same stream) + * decl i6 ; Second ref macroblock top field texcoords + * decl i7 ; Second ref macroblock bottom field texcoords (unused, packed in the same stream) */ - for (i = 0; i < 6; i++) + for (i = 0; i < 8; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1471,11 +1554,13 @@ static int vlCreateVertexShaderFrameBMB /* * decl o0 ; Vertex pos - * decl o1 ; Luma/chroma texcoords - * decl o2 ; First ref macroblock texcoords - * decl o3 ; Second ref macroblock texcoords + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + * decl o4 ; First ref macroblock texcoords + * decl o5 ; Second ref macroblock texcoords */ - for (i = 0; i < 4; i++) + for (i = 0; i < 6; i++) { decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1483,21 +1568,23 @@ static int vlCreateVertexShaderFrameBMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output */ - for (i = 0; i < 2; ++i) + for (i = 0; i < 4; ++i) { inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* - * add o2, i0, i2 ; Translate vertex pos by motion vec to form first ref macroblock texcoords - * add o3, i0, i4 ; Translate vertex pos by motion vec to form second ref macroblock texcoords + * add o4, i0, i4 ; Translate vertex pos by motion vec to form first ref macroblock texcoords + * add o5, i0, i6 ; Translate vertex pos by motion vec to form second ref macroblock texcoords */ for (i = 0; i < 2; ++i) { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 1) * 2); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 2) * 2); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1546,14 +1633,16 @@ static int vlCreateVertexShaderFieldBMB ti = 3; /* - * decl i0 ; Vertex pos, Luma/chroma texcoords - * decl i1 ; Luma/chroma texcoords - * decl i2 ; First ref surface top field texcoords - * decl i3 ; First ref surface bottom field texcoords - * decl i4 ; Second ref surface top field texcoords - * decl i5 ; Second ref surface bottom field texcoords + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + * decl i4 ; First ref macroblock top field texcoords + * decl i5 ; First ref macroblock bottom field texcoords + * decl i6 ; Second ref macroblock top field texcoords + * decl i7 ; Second ref macroblock bottom field texcoords */ - for (i = 0; i < 6; i++) + for (i = 0; i < 8; i++) { decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1565,16 +1654,18 @@ static int vlCreateVertexShaderFieldBMB /* * decl o0 ; Vertex pos - * decl o1 ; Luma/chroma texcoords - * decl o2 ; Top field past ref macroblock texcoords - * decl o3 ; Bottom field past ref macroblock texcoords - * decl o4 ; Top field future ref macroblock texcoords - * decl o5 ; Bottom field future ref macroblock texcoords - * decl o6 ; Denormalized vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + * decl o4 ; First ref macroblock top field texcoords + * decl o5 ; First ref macroblock Bottom field texcoords + * decl o6 ; Second ref macroblock top field texcoords + * decl o7 ; Second ref macroblock Bottom field texcoords + * decl o8 ; Denormalized vertex pos */ - for (i = 0; i < 7; i++) + for (i = 0; i < 9; i++) { - decl = vl_decl_output((i == 0 || i == 7) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + decl = vl_decl_output((i == 0 || i == 8) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } @@ -1584,28 +1675,30 @@ static int vlCreateVertexShaderFieldBMB /* * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma/chroma texcoords to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output */ - for (i = 0; i < 2; ++i) + for (i = 0; i < 4; ++i) { inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } /* - * add o2, i0, i2 ; Translate vertex pos by motion vec to form first top field macroblock texcoords - * add o3, i0, i3 ; Translate vertex pos by motion vec to form first bottom field macroblock texcoords - * add o4, i0, i4 ; Translate vertex pos by motion vec to form second top field macroblock texcoords - * add o5, i0, i5 ; Translate vertex pos by motion vec to form second bottom field macroblock texcoords + * add o4, i0, i4 ; Translate vertex pos by motion vec to form first top field macroblock texcoords + * add o5, i0, i5 ; Translate vertex pos by motion vec to form first bottom field macroblock texcoords + * add o6, i0, i6 ; Translate vertex pos by motion vec to form second top field macroblock texcoords + * add o7, i0, i7 ; Translate vertex pos by motion vec to form second bottom field macroblock texcoords */ for (i = 0; i < 4; ++i) { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 2, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 2); + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } - /* mul o6, i0, c0 ; Denorm vertex pos */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 6, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); + /* mul o8, i0, c0 ; Denorm vertex pos */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 8, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* end */ @@ -1653,11 +1746,13 @@ static int vlCreateFragmentShaderFrameBMB ti = 3; /* - * decl i0 ; Texcoords for s0, s1, s2 - * decl i1 ; Texcoords for s3 - * decl i2 ; Texcoords for s4 + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + * decl i3 ; First ref macroblock texcoords + * decl i4 ; Second ref macroblock texcoords */ - for (i = 0; i < 3; ++i) + for (i = 0; i < 5; ++i) { decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1682,8 +1777,8 @@ static int vlCreateFragmentShaderFrameBMB * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for past ref surface texture - * decl s4 ; Sampler for future ref surface texture + * decl s3 ; Sampler for first ref surface texture + * decl s4 ; Sampler for second ref surface texture */ for (i = 0; i < 5; ++i) { @@ -1694,14 +1789,14 @@ static int vlCreateFragmentShaderFrameBMB /* * tex2d t1, i0, s0 ; Read texel from luma texture * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); @@ -1718,12 +1813,12 @@ static int vlCreateFragmentShaderFrameBMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - * tex2d t1, i1, s3 ; Read texel from past ref macroblock - * tex2d t2, i2, s4 ; Read texel from future ref macroblock + * tex2d t1, i3, s3 ; Read texel from first ref macroblock + * tex2d t2, i4, s4 ; Read texel from second ref macroblock */ for (i = 0; i < 2; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 1, TGSI_FILE_SAMPLER, i + 3); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, i + 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1784,14 +1879,16 @@ static int vlCreateFragmentShaderFieldBMB ti = 3; /* - * decl i0 ; Texcoords for s0, s1, s2 - * decl i1 ; Texcoords for s3 - * decl i2 ; Texcoords for s3 - * decl i3 ; Texcoords for s4 - * decl i4 ; Texcoords for s4 - * decl i5 ; Denormalized vertex pos + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + * decl i3 ; First ref macroblock top field texcoords + * decl i4 ; First ref macroblock bottom field texcoords + * decl i5 ; Second ref macroblock top field texcoords + * decl i6 ; Second ref macroblock bottom field texcoords + * decl i7 ; Denormalized vertex pos */ - for (i = 0; i < 6; ++i) + for (i = 0; i < 8; ++i) { decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1817,8 +1914,8 @@ static int vlCreateFragmentShaderFieldBMB * decl s0 ; Sampler for luma texture * decl s1 ; Sampler for chroma Cb texture * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for past ref surface texture - * decl s4 ; Sampler for future ref surface texture + * decl s3 ; Sampler for first ref surface texture + * decl s4 ; Sampler for second ref surface texture */ for (i = 0; i < 5; ++i) { @@ -1829,14 +1926,14 @@ static int vlCreateFragmentShaderFieldBMB /* * tex2d t1, i0, s0 ; Read texel from luma texture * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i0, s1 ; Read texel from chroma Cb texture + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i0, s2 ; Read texel from chroma Cr texture + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture * mov t0.z, t1.x ; Move Cr sample into .z component */ for (i = 0; i < 3; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, i); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); @@ -1853,8 +1950,8 @@ static int vlCreateFragmentShaderFieldBMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* XXX: Pos values off by 0.5? */ - /* sub t4, i5.y, c1.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 5, TGSI_FILE_CONSTANT, 1); + /* sub t4, i7.y, c1.x ; Sub 0.5 from denormalized pos */ + inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 7, TGSI_FILE_CONSTANT, 1); inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; @@ -1890,12 +1987,12 @@ static int vlCreateFragmentShaderFieldBMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - * tex2d t1, i1, s3 ; Read texel from past ref macroblock top field - * tex2d t2, i2, s3 ; Read texel from past ref macroblock bottom field + * tex2d t1, i3, s3 ; Read texel from past ref macroblock top field + * tex2d t2, i4, s3 ; Read texel from past ref macroblock bottom field */ for (i = 0; i < 2; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 1, TGSI_FILE_SAMPLER, 3); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1905,12 +2002,12 @@ static int vlCreateFragmentShaderFieldBMB ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); /* - * tex2d t4, i3, s4 ; Read texel from future ref macroblock top field - * tex2d t5, i4, s4 ; Read texel from future ref macroblock bottom field + * tex2d t4, i5, s4 ; Read texel from future ref macroblock top field + * tex2d t5, i6, s4 ; Read texel from future ref macroblock bottom field */ for (i = 0; i < 2; ++i) { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 4); + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 5, TGSI_FILE_SAMPLER, 4); ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); } @@ -1961,7 +2058,18 @@ static int vlCreateDataBufs /* Create our vertex buffers */ for (h = 0; h < NUM_BUF_SETS; ++h) { - for (i = 0; i < 3; ++i) + mc->vertex_bufs[h][0].pitch = sizeof(struct vlVertex2f) * 4; + mc->vertex_bufs[h][0].max_index = 24 * num_mb_per_frame - 1; + mc->vertex_bufs[h][0].buffer_offset = 0; + mc->vertex_bufs[h][0].buffer = pipe->winsys->buffer_create + ( + pipe->winsys, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vlVertex2f) * 4 * 24 * num_mb_per_frame + ); + + for (i = 1; i < 3; ++i) { mc->vertex_bufs[h][i].pitch = sizeof(struct vlVertex2f) * 2; mc->vertex_bufs[h][i].max_index = 24 * num_mb_per_frame - 1; @@ -1982,36 +2090,48 @@ static int vlCreateDataBufs mc->vertex_elems[0].nr_components = 2; mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* Block luma, block chroma texcoord element */ + /* Luma, texcoord element */ mc->vertex_elems[1].src_offset = sizeof(struct vlVertex2f); mc->vertex_elems[1].vertex_buffer_index = 0; mc->vertex_elems[1].nr_components = 2; mc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* First ref surface top field texcoord element */ - mc->vertex_elems[2].src_offset = 0; - mc->vertex_elems[2].vertex_buffer_index = 1; + /* Chroma Cr texcoord element */ + mc->vertex_elems[2].src_offset = sizeof(struct vlVertex2f) * 2; + mc->vertex_elems[2].vertex_buffer_index = 0; mc->vertex_elems[2].nr_components = 2; mc->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* First ref surface bottom field texcoord element */ - mc->vertex_elems[3].src_offset = sizeof(struct vlVertex2f); - mc->vertex_elems[3].vertex_buffer_index = 1; + /* Chroma Cb texcoord element */ + mc->vertex_elems[3].src_offset = sizeof(struct vlVertex2f) * 3; + mc->vertex_elems[3].vertex_buffer_index = 0; mc->vertex_elems[3].nr_components = 2; mc->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* Second ref surface top field texcoord element */ + /* First ref surface top field texcoord element */ mc->vertex_elems[4].src_offset = 0; - mc->vertex_elems[4].vertex_buffer_index = 2; + mc->vertex_elems[4].vertex_buffer_index = 1; mc->vertex_elems[4].nr_components = 2; mc->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT; - /* Second ref surface bottom field texcoord element */ + /* First ref surface bottom field texcoord element */ mc->vertex_elems[5].src_offset = sizeof(struct vlVertex2f); - mc->vertex_elems[5].vertex_buffer_index = 2; + mc->vertex_elems[5].vertex_buffer_index = 1; mc->vertex_elems[5].nr_components = 2; mc->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT; + /* Second ref surface top field texcoord element */ + mc->vertex_elems[6].src_offset = 0; + mc->vertex_elems[6].vertex_buffer_index = 2; + mc->vertex_elems[6].nr_components = 2; + mc->vertex_elems[6].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Second ref surface bottom field texcoord element */ + mc->vertex_elems[7].src_offset = sizeof(struct vlVertex2f); + mc->vertex_elems[7].vertex_buffer_index = 2; + mc->vertex_elems[7].nr_components = 2; + mc->vertex_elems[7].src_format = PIPE_FORMAT_R32G32_FLOAT; + /* Create our constant buffer */ mc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts); mc->vs_const_buf.buffer = pipe->winsys->buffer_create -- cgit v1.2.3 From 2f41095e04036654259bc2efc3324a92f32c8ca1 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 4 Sep 2008 04:10:43 -0400 Subject: g3dvl: Zero-block optimization. For blocks whose contents are derived completely from ref surfaces, don't bother zero-ing the corresponding block in the luma/chroma textures, except for the first such luma, chroma Cb, and chroma Cr each frame. All later zero blocks are textured from that first zero block. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 92 +++++++++++++--------- 1 file changed, 54 insertions(+), 38 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index b5aa79cdb44..93839e8aa9b 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -63,6 +63,7 @@ struct vlR16SnormBufferedMC struct vlSurface *buffered_surface; struct vlSurface *past_surface, *future_surface; struct vlVertex2f surface_tex_inv_size; + struct vlVertex2f zero_block[3]; unsigned int num_macroblocks; struct vlMpeg2MacroBlock *macroblocks; @@ -203,8 +204,13 @@ static inline int vlGrabBlocks ++sb; } - else + else if (mc->zero_block[0].x < 0.0f) + { vlGrabNoBlock(texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, tex_pitch); + + mc->zero_block[0].x = (mbpx + x * 8) * mc->surface_tex_inv_size.x; + mc->zero_block[0].y = (mbpy + y * 8) * mc->surface_tex_inv_size.y; + } } } @@ -241,9 +247,14 @@ static inline int vlGrabBlocks ++sb; } - else + else if (mc->zero_block[tb + 1].x < 0.0f) + { vlGrabNoBlock(texels, tex_pitch); + mc->zero_block[tb + 1].x = (mbpx << 1) * mc->surface_tex_inv_size.x; + mc->zero_block[tb + 1].y = (mbpy << 1) * mc->surface_tex_inv_size.y; + } + pipe_surface_unmap(tex_surface); } @@ -315,7 +326,7 @@ static inline int vlGrabMacroBlock return 0; } -#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, zx, zy) \ +#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, zb) \ (vb)[0].pos.x = (mbx) * (unitx) + (ofsx); (vb)[0].pos.y = (mby) * (unity) + (ofsy); \ (vb)[1].pos.x = (mbx) * (unitx) + (ofsx); (vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].pos.y = (mby) * (unity) + (ofsy); \ @@ -323,62 +334,62 @@ static inline int vlGrabMacroBlock (vb)[4].pos.x = (mbx) * (unitx) + (ofsx); (vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy); \ \ - /*if ((cbp) & (lm)) \ - {*/ \ + if ((cbp) & (lm)) \ + { \ (vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].luma_tc.y = (mby) * (unity) + (ofsy); \ (vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].luma_tc.y = (mby) * (unity) + (ofsy); \ (vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].luma_tc.y = (mby) * (unity) + (ofsy); \ (vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - /*} \ + } \ else \ { \ - (vb)[0].luma_tc.x = (zx); (vb)[0].luma_tc.y = (zy); \ - (vb)[1].luma_tc.x = (zx); (vb)[1].luma_tc.y = (zy) + (hy); \ - (vb)[2].luma_tc.x = (zx) + (hx); (vb)[2].luma_tc.y = (zy); \ - (vb)[3].luma_tc.x = (zx) + (hx); (vb)[3].luma_tc.y = (zy); \ - (vb)[4].luma_tc.x = (zx); (vb)[4].luma_tc.y = (zy) + (hy); \ - (vb)[5].luma_tc.x = ((zx) + (hx); (vb)[5].luma_tc.y = (zy) + (hy); \ - }*/ \ + (vb)[0].luma_tc.x = (zb)[0].x; (vb)[0].luma_tc.y = (zb)[0].y; \ + (vb)[1].luma_tc.x = (zb)[0].x; (vb)[1].luma_tc.y = (zb)[0].y + (hy); \ + (vb)[2].luma_tc.x = (zb)[0].x + (hx); (vb)[2].luma_tc.y = (zb)[0].y; \ + (vb)[3].luma_tc.x = (zb)[0].x + (hx); (vb)[3].luma_tc.y = (zb)[0].y; \ + (vb)[4].luma_tc.x = (zb)[0].x; (vb)[4].luma_tc.y = (zb)[0].y + (hy); \ + (vb)[5].luma_tc.x = (zb)[0].x + (hx); (vb)[5].luma_tc.y = (zb)[0].y + (hy); \ + } \ \ - /*if ((cbp) & (cbm)) \ - {*/ \ + if ((cbp) & (cbm)) \ + { \ (vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cb_tc.y = (mby) * (unity) + (ofsy); \ (vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cb_tc.y = (mby) * (unity) + (ofsy); \ (vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cb_tc.y = (mby) * (unity) + (ofsy); \ (vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - /*} \ + } \ else \ { \ - (vb)[0].cb_tc.x = (zx); (vb)[0].cb_tc.y = (zy); \ - (vb)[1].cb_tc.x = (zx); (vb)[1].cb_tc.y = (zy) + (hy); \ - (vb)[2].cb_tc.x = (zx) + (hx); (vb)[2].cb_tc.y = (zy); \ - (vb)[3].cb_tc.x = (zx) + (hx); (vb)[3].cb_tc.y = (zy); \ - (vb)[4].cb_tc.x = (zx); (vb)[4].cb_tc.y = (zy) + (hy); \ - (vb)[5].cb_tc.x = ((zx) + (hx); (vb)[5].cb_tc.y = (zy) + (hy); \ - }*/ \ + (vb)[0].cb_tc.x = (zb)[1].x; (vb)[0].cb_tc.y = (zb)[1].y; \ + (vb)[1].cb_tc.x = (zb)[1].x; (vb)[1].cb_tc.y = (zb)[1].y + (hy); \ + (vb)[2].cb_tc.x = (zb)[1].x + (hx); (vb)[2].cb_tc.y = (zb)[1].y; \ + (vb)[3].cb_tc.x = (zb)[1].x + (hx); (vb)[3].cb_tc.y = (zb)[1].y; \ + (vb)[4].cb_tc.x = (zb)[1].x; (vb)[4].cb_tc.y = (zb)[1].y + (hy); \ + (vb)[5].cb_tc.x = (zb)[1].x + (hx); (vb)[5].cb_tc.y = (zb)[1].y + (hy); \ + } \ \ - /*if ((cbp) & (crm)) \ - {*/ \ + if ((cbp) & (crm)) \ + { \ (vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cr_tc.y = (mby) * (unity) + (ofsy); \ (vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cr_tc.y = (mby) * (unity) + (ofsy); \ (vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cr_tc.y = (mby) * (unity) + (ofsy); \ (vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ (vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - /*} \ + } \ else \ { \ - (vb)[0].cr_tc.x = (zx); (vb)[0].cb_tc.y = (zy); \ - (vb)[1].cr_tc.x = (zx); (vb)[1].cb_tc.y = (zy) + (hy); \ - (vb)[2].cr_tc.x = (zx) + (hx); (vb)[2].cb_tc.y = (zy); \ - (vb)[3].cr_tc.x = (zx) + (hx); (vb)[3].cb_tc.y = (zy); \ - (vb)[4].cr_tc.x = (zx); (vb)[4].cb_tc.y = (zy) + (hy); \ - (vb)[5].cr_tc.x = ((zx) + (hx); (vb)[5].cb_tc.y = (zy) + (hy); \ - }*/ + (vb)[0].cr_tc.x = (zb)[2].x; (vb)[0].cr_tc.y = (zb)[2].y; \ + (vb)[1].cr_tc.x = (zb)[2].x; (vb)[1].cr_tc.y = (zb)[2].y + (hy); \ + (vb)[2].cr_tc.x = (zb)[2].x + (hx); (vb)[2].cr_tc.y = (zb)[2].y; \ + (vb)[3].cr_tc.x = (zb)[2].x + (hx); (vb)[3].cr_tc.y = (zb)[2].y; \ + (vb)[4].cr_tc.x = (zb)[2].x; (vb)[4].cr_tc.y = (zb)[2].y + (hy); \ + (vb)[5].cr_tc.x = (zb)[2].x + (hx); (vb)[5].cr_tc.y = (zb)[2].y + (hy); \ + } static inline int vlGrabMacroBlockVB ( @@ -526,7 +537,7 @@ static inline int vlGrabMacroBlockVB vb, macroblock->cbp, macroblock->mbx, macroblock->mby, unit.x, unit.y, 0, 0, half.x, half.y, - 32, 2, 1, mc->zero_block.x, mc->zero_block.y + 32, 2, 1, mc->zero_block ); SET_BLOCK @@ -534,7 +545,7 @@ static inline int vlGrabMacroBlockVB vb + 6, macroblock->cbp, macroblock->mbx, macroblock->mby, unit.x, unit.y, half.x, 0, half.x, half.y, - 16, 2, 1, mc->zero_block.x, mc->zero_block.y + 16, 2, 1, mc->zero_block ); SET_BLOCK @@ -542,7 +553,7 @@ static inline int vlGrabMacroBlockVB vb + 12, macroblock->cbp, macroblock->mbx, macroblock->mby, unit.x, unit.y, 0, half.y, half.x, half.y, - 8, 2, 1, mc->zero_block.x, mc->zero_block.y + 8, 2, 1, mc->zero_block ); SET_BLOCK @@ -550,7 +561,7 @@ static inline int vlGrabMacroBlockVB vb + 18, macroblock->cbp, macroblock->mbx, macroblock->mby, unit.x, unit.y, half.x, half.y, half.x, half.y, - 4, 2, 1, mc->zero_block.x, mc->zero_block.y + 4, 2, 1, mc->zero_block ); mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer); @@ -725,6 +736,8 @@ static int vlFlush vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24; } + for (i = 0; i < 3; ++i) + mc->zero_block[i].x = -1.0f; mc->num_macroblocks = 0; mc->cur_buf++; @@ -2277,7 +2290,8 @@ int vlCreateR16SNormBufferedMC struct vlRender **render ) { - struct vlR16SnormBufferedMC *mc; + struct vlR16SnormBufferedMC *mc; + unsigned int i; assert(pipe); assert(render); @@ -2297,6 +2311,8 @@ int vlCreateR16SNormBufferedMC mc->buffered_surface = NULL; mc->past_surface = NULL; mc->future_surface = NULL; + for (i = 0; i < 3; ++i) + mc->zero_block[i].x = -1.0f; mc->num_macroblocks = 0; vlInit(mc); -- cgit v1.2.3 From 02025148c28d03d644e3d66dde1a423fe21e1c44 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Thu, 4 Sep 2008 04:43:34 -0400 Subject: g3dvl: Initialize DRI drawables correctly. --- src/driclient/src/driclient.c | 128 ++++++++++++++++++++++-------------------- 1 file changed, 68 insertions(+), 60 deletions(-) diff --git a/src/driclient/src/driclient.c b/src/driclient/src/driclient.c index 94d01aca4f9..2d7514166fa 100644 --- a/src/driclient/src/driclient.c +++ b/src/driclient/src/driclient.c @@ -12,40 +12,40 @@ int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri drm_handle_t sarea_handle; char *bus_id; dri_screen_t *dri_scrn; - + assert(display); assert(dri_screen); - + if (!XF86DRIQueryExtension(display, &evbase, &errbase)) return 1; - + dri_scrn = calloc(1, sizeof(dri_screen_t)); - + if (!dri_scrn) return 1; if (!XF86DRIQueryVersion(display, &dri_scrn->dri.major, &dri_scrn->dri.minor, &dri_scrn->dri.patch)) goto free_screen; - + dri_scrn->display = display; dri_scrn->num = screen; dri_scrn->draw_lock_id = 1; - + if (!XF86DRIOpenConnection(display, screen, &sarea_handle, &bus_id)) goto free_screen; - + dri_scrn->fd = -1; dri_scrn->fd = drmOpenOnce(NULL, bus_id, &newly_opened); XFree(bus_id); if (dri_scrn->fd < 0) goto close_connection; - + if (drmGetMagic(dri_scrn->fd, &magic)) goto close_drm; - + drm_version = drmGetVersion(dri_scrn->fd); - + if (!drm_version) goto close_drm; @@ -53,10 +53,10 @@ int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri dri_scrn->drm.minor = drm_version->version_minor; dri_scrn->drm.patch = drm_version->version_patchlevel; drmFreeVersion(drm_version); - + if (!XF86DRIAuthConnection(display, screen, magic)) goto close_drm; - + if (!XF86DRIGetClientDriverName ( display, @@ -67,15 +67,15 @@ int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri &driver_name )) goto close_drm; - + if (drmMap(dri_scrn->fd, sarea_handle, SAREA_MAX, (drmAddress)&dri_scrn->sarea)) goto close_drm; - + dri_scrn->drawable_hash = drmHashCreate(); - + if (!dri_scrn->drawable_hash) goto unmap_sarea; - + if (dri_framebuf) { if (!XF86DRIGetDeviceInfo @@ -90,9 +90,9 @@ int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri )) goto destroy_hash; } - + *dri_screen = dri_scrn; - + return 0; destroy_hash: @@ -105,20 +105,20 @@ close_connection: XF86DRICloseConnection(display, screen); free_screen: free(dri_scrn); - + return 1; } int driDestroyScreen(dri_screen_t *dri_screen) { assert(dri_screen); - + drmHashDestroy(dri_screen->drawable_hash); drmUnmap(dri_screen->sarea, SAREA_MAX); drmCloseOnce(dri_screen->fd); XF86DRICloseConnection(dri_screen->display, dri_screen->num); free(dri_screen); - + return 0; } @@ -126,73 +126,82 @@ int driCreateDrawable(dri_screen_t *dri_screen, Drawable drawable, dri_drawable_ { int evbase, errbase; dri_drawable_t *dri_draw; - + assert(dri_screen); assert(dri_drawable); - + if (!XF86DRIQueryExtension(dri_screen->display, &evbase, &errbase)) return 1; - + if (!drmHashLookup(dri_screen->drawable_hash, drawable, (void**)dri_drawable)) { /* Found */ (*dri_drawable)->refcount++; return 0; } - + dri_draw = calloc(1, sizeof(dri_drawable_t)); - + if (!dri_draw) return 1; - + if (!XF86DRICreateDrawable(dri_screen->display, 0, drawable, &dri_draw->drm_drawable)) { free(dri_draw); return 1; } - + dri_draw->x_drawable = drawable; dri_draw->sarea_index = 0; dri_draw->sarea_stamp = NULL; dri_draw->last_sarea_stamp = 0; dri_draw->dri_screen = dri_screen; dri_draw->refcount = 1; - + if (drmHashInsert(dri_screen->drawable_hash, drawable, dri_draw)) { XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable); free(dri_draw); return 1; } - - /* - * XXX: Need this to initialize sarea pointer and other stuff in dri_drawable_t - * to be able to use the DRI_VALIDATE_DRAWABLE_INFO macro, but is it safe to - * call without any sync? - */ - if (driUpdateDrawableInfo(dri_draw)) + + if (!dri_draw->sarea_stamp || *dri_draw->sarea_stamp != dri_draw->last_sarea_stamp) { - XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable); - free(dri_draw); - return 1; + DRM_SPINLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id); + + if (driUpdateDrawableInfo(dri_draw)) + { + XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable); + free(dri_draw); + DRM_SPINUNLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id); + return 1; + } + + DRM_SPINUNLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id); } - + *dri_drawable = dri_draw; - + return 0; } int driUpdateDrawableInfo(dri_drawable_t *dri_drawable) { assert(dri_drawable); - + if (dri_drawable->cliprects) + { XFree(dri_drawable->cliprects); + dri_drawable->cliprects = NULL; + } if (dri_drawable->back_cliprects) + { XFree(dri_drawable->back_cliprects); - + dri_drawable->back_cliprects = NULL; + } + DRM_SPINUNLOCK(&dri_drawable->dri_screen->sarea->drawable_lock, dri_drawable->dri_screen->draw_lock_id); - + if (!XF86DRIGetDrawableInfo ( dri_drawable->dri_screen->display, @@ -217,21 +226,21 @@ int driUpdateDrawableInfo(dri_drawable_t *dri_drawable) dri_drawable->cliprects = NULL; dri_drawable->num_back_cliprects = 0; dri_drawable->back_cliprects = 0; - + return 1; } else dri_drawable->sarea_stamp = &dri_drawable->dri_screen->sarea->drawableTable[dri_drawable->sarea_index].stamp; - + DRM_SPINLOCK(&dri_drawable->dri_screen->sarea->drawable_lock, dri_drawable->dri_screen->draw_lock_id); - + return 0; } int driDestroyDrawable(dri_drawable_t *dri_drawable) { assert(dri_drawable); - + if (--dri_drawable->refcount == 0) { if (dri_drawable->cliprects) @@ -242,7 +251,7 @@ int driDestroyDrawable(dri_drawable_t *dri_drawable) XF86DRIDestroyDrawable(dri_drawable->dri_screen->display, dri_drawable->dri_screen->num, dri_drawable->x_drawable); free(dri_drawable); } - + return 0; } @@ -250,38 +259,37 @@ int driCreateContext(dri_screen_t *dri_screen, Visual *visual, dri_context_t **d { int evbase, errbase; dri_context_t *dri_ctx; - + assert(dri_screen); assert(visual); assert(dri_context); - + if (!XF86DRIQueryExtension(dri_screen->display, &evbase, &errbase)) return 1; - + dri_ctx = calloc(1, sizeof(dri_context_t)); - + if (!dri_ctx) return 1; - + if (!XF86DRICreateContext(dri_screen->display, dri_screen->num, visual, &dri_ctx->id, &dri_ctx->drm_context)) { free(dri_ctx); return 1; } - - dri_ctx->dri_screen = dri_screen; + + dri_ctx->dri_screen = dri_screen; *dri_context = dri_ctx; - + return 0; } int driDestroyContext(dri_context_t *dri_context) { assert(dri_context); - + XF86DRIDestroyContext(dri_context->dri_screen->display, dri_context->dri_screen->num, dri_context->id); free(dri_context); - + return 0; } - -- cgit v1.2.3 From f302fca5eb63e4bca8af5b35c585451486143e6a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 11 Sep 2008 06:41:18 +1000 Subject: nouveau: gallium directory structure changed again.. --- src/gallium/drivers/nouveau/nouveau_stateobj.h | 1 - src/gallium/drivers/nv04/nv04_context.c | 1 - src/gallium/drivers/nv04/nv04_context.h | 4 + src/gallium/drivers/nv04/nv04_fragprog.c | 1 - src/gallium/drivers/nv04/nv04_miptree.c | 6 +- src/gallium/drivers/nv04/nv04_prim_vbuf.c | 5 +- src/gallium/drivers/nv04/nv04_screen.c | 1 - src/gallium/drivers/nv04/nv04_state.c | 1 - src/gallium/drivers/nv04/nv04_surface.c | 3 +- src/gallium/drivers/nv04/nv04_vbo.c | 1 - src/gallium/drivers/nv10/nv10_context.c | 1 - src/gallium/drivers/nv10/nv10_context.h | 4 + src/gallium/drivers/nv10/nv10_fragprog.c | 1 - src/gallium/drivers/nv10/nv10_miptree.c | 6 +- src/gallium/drivers/nv10/nv10_prim_vbuf.c | 6 +- src/gallium/drivers/nv10/nv10_screen.c | 1 - src/gallium/drivers/nv10/nv10_state.c | 1 - src/gallium/drivers/nv10/nv10_state_emit.c | 2 - src/gallium/drivers/nv10/nv10_surface.c | 3 +- src/gallium/drivers/nv10/nv10_vbo.c | 1 - src/gallium/drivers/nv30/nv30_context.c | 1 - src/gallium/drivers/nv30/nv30_context.h | 4 + src/gallium/drivers/nv30/nv30_draw.c | 1 - src/gallium/drivers/nv30/nv30_miptree.c | 9 +- src/gallium/drivers/nv30/nv30_screen.c | 1 - src/gallium/drivers/nv30/nv30_state.c | 1 - src/gallium/drivers/nv30/nv30_surface.c | 4 +- src/gallium/drivers/nv30/nv30_vbo.c | 1 - src/gallium/drivers/nv40/nv40_context.c | 1 - src/gallium/drivers/nv40/nv40_context.h | 4 + src/gallium/drivers/nv40/nv40_draw.c | 3 +- src/gallium/drivers/nv40/nv40_miptree.c | 11 +- src/gallium/drivers/nv40/nv40_screen.c | 1 - src/gallium/drivers/nv40/nv40_state.c | 1 - src/gallium/drivers/nv40/nv40_surface.c | 4 +- src/gallium/drivers/nv40/nv40_vbo.c | 1 - src/gallium/drivers/nv50/nv50_context.c | 1 - src/gallium/drivers/nv50/nv50_context.h | 4 + src/gallium/drivers/nv50/nv50_draw.c | 1 - src/gallium/drivers/nv50/nv50_miptree.c | 12 +- src/gallium/drivers/nv50/nv50_screen.c | 1 - src/gallium/drivers/nv50/nv50_state.c | 1 - src/gallium/drivers/nv50/nv50_surface.c | 12 +- src/gallium/drivers/nv50/nv50_vbo.c | 1 - src/gallium/winsys/dri/nouveau/Makefile | 45 -- src/gallium/winsys/dri/nouveau/nouveau_bo.c | 470 --------------------- src/gallium/winsys/dri/nouveau/nouveau_channel.c | 126 ------ src/gallium/winsys/dri/nouveau/nouveau_context.c | 346 --------------- src/gallium/winsys/dri/nouveau/nouveau_context.h | 113 ----- src/gallium/winsys/dri/nouveau/nouveau_device.c | 159 ------- src/gallium/winsys/dri/nouveau/nouveau_dma.c | 219 ---------- src/gallium/winsys/dri/nouveau/nouveau_dma.h | 143 ------- src/gallium/winsys/dri/nouveau/nouveau_dri.h | 28 -- src/gallium/winsys/dri/nouveau/nouveau_drmif.h | 310 -------------- src/gallium/winsys/dri/nouveau/nouveau_fence.c | 214 ---------- src/gallium/winsys/dri/nouveau/nouveau_grobj.c | 107 ----- src/gallium/winsys/dri/nouveau/nouveau_local.h | 117 ----- src/gallium/winsys/dri/nouveau/nouveau_lock.c | 94 ----- src/gallium/winsys/dri/nouveau/nouveau_notifier.c | 137 ------ src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c | 271 ------------ src/gallium/winsys/dri/nouveau/nouveau_resource.c | 116 ----- src/gallium/winsys/dri/nouveau/nouveau_screen.c | 310 -------------- src/gallium/winsys/dri/nouveau/nouveau_screen.h | 20 - .../winsys/dri/nouveau/nouveau_swapbuffers.c | 86 ---- .../winsys/dri/nouveau/nouveau_swapbuffers.h | 10 - src/gallium/winsys/dri/nouveau/nouveau_winsys.c | 158 ------- .../winsys/dri/nouveau/nouveau_winsys_pipe.c | 205 --------- .../winsys/dri/nouveau/nouveau_winsys_pipe.h | 34 -- .../winsys/dri/nouveau/nouveau_winsys_softpipe.c | 85 ---- src/gallium/winsys/dri/nouveau/nv04_surface.c | 314 -------------- src/gallium/winsys/dri/nouveau/nv50_surface.c | 194 --------- src/gallium/winsys/drm/nouveau/Makefile | 45 ++ src/gallium/winsys/drm/nouveau/nouveau_bo.c | 470 +++++++++++++++++++++ src/gallium/winsys/drm/nouveau/nouveau_channel.c | 126 ++++++ src/gallium/winsys/drm/nouveau/nouveau_context.c | 346 +++++++++++++++ src/gallium/winsys/drm/nouveau/nouveau_context.h | 113 +++++ src/gallium/winsys/drm/nouveau/nouveau_device.c | 159 +++++++ src/gallium/winsys/drm/nouveau/nouveau_dma.c | 219 ++++++++++ src/gallium/winsys/drm/nouveau/nouveau_dma.h | 143 +++++++ src/gallium/winsys/drm/nouveau/nouveau_dri.h | 28 ++ src/gallium/winsys/drm/nouveau/nouveau_drmif.h | 310 ++++++++++++++ src/gallium/winsys/drm/nouveau/nouveau_fence.c | 214 ++++++++++ src/gallium/winsys/drm/nouveau/nouveau_grobj.c | 107 +++++ src/gallium/winsys/drm/nouveau/nouveau_local.h | 117 +++++ src/gallium/winsys/drm/nouveau/nouveau_lock.c | 94 +++++ src/gallium/winsys/drm/nouveau/nouveau_notifier.c | 137 ++++++ src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c | 271 ++++++++++++ src/gallium/winsys/drm/nouveau/nouveau_resource.c | 116 +++++ src/gallium/winsys/drm/nouveau/nouveau_screen.c | 310 ++++++++++++++ src/gallium/winsys/drm/nouveau/nouveau_screen.h | 20 + .../winsys/drm/nouveau/nouveau_swapbuffers.c | 86 ++++ .../winsys/drm/nouveau/nouveau_swapbuffers.h | 10 + src/gallium/winsys/drm/nouveau/nouveau_winsys.c | 158 +++++++ .../winsys/drm/nouveau/nouveau_winsys_pipe.c | 206 +++++++++ .../winsys/drm/nouveau/nouveau_winsys_pipe.h | 34 ++ .../winsys/drm/nouveau/nouveau_winsys_softpipe.c | 85 ++++ src/gallium/winsys/drm/nouveau/nv04_surface.c | 314 ++++++++++++++ src/gallium/winsys/drm/nouveau/nv50_surface.c | 194 +++++++++ 98 files changed, 4487 insertions(+), 4507 deletions(-) delete mode 100644 src/gallium/winsys/dri/nouveau/Makefile delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_bo.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_channel.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_context.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_context.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_device.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_dma.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_dma.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_dri.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_drmif.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_fence.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_grobj.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_local.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_lock.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_notifier.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_resource.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_screen.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_screen.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h delete mode 100644 src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c delete mode 100644 src/gallium/winsys/dri/nouveau/nv04_surface.c delete mode 100644 src/gallium/winsys/dri/nouveau/nv50_surface.c create mode 100644 src/gallium/winsys/drm/nouveau/Makefile create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_bo.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_channel.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_context.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_context.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_device.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_dma.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_dma.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_dri.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_drmif.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_fence.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_grobj.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_local.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_lock.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_notifier.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_resource.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_screen.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_screen.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_winsys.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h create mode 100644 src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c create mode 100644 src/gallium/winsys/drm/nouveau/nv04_surface.c create mode 100644 src/gallium/winsys/drm/nouveau/nv50_surface.c diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index 998ec2d4ad4..729988b095e 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -1,7 +1,6 @@ #ifndef __NOUVEAU_STATEOBJ_H__ #define __NOUVEAU_STATEOBJ_H__ -#include "pipe/p_util.h" #include "pipe/p_debug.h" struct nouveau_stateobj_reloc { diff --git a/src/gallium/drivers/nv04/nv04_context.c b/src/gallium/drivers/nv04/nv04_context.c index 852a8edf5ff..9f75253363f 100644 --- a/src/gallium/drivers/nv04/nv04_context.c +++ b/src/gallium/drivers/nv04/nv04_context.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" -#include "pipe/p_util.h" #include "nv04_context.h" #include "nv04_screen.h" diff --git a/src/gallium/drivers/nv04/nv04_context.h b/src/gallium/drivers/nv04/nv04_context.h index 5ba1d4ecdc1..3e6a0852702 100644 --- a/src/gallium/drivers/nv04/nv04_context.h +++ b/src/gallium/drivers/nv04/nv04_context.h @@ -4,6 +4,10 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" #include "draw/draw_vertex.h" diff --git a/src/gallium/drivers/nv04/nv04_fragprog.c b/src/gallium/drivers/nv04/nv04_fragprog.c index 215974eec00..8a2af41fe06 100644 --- a/src/gallium/drivers/nv04/nv04_fragprog.c +++ b/src/gallium/drivers/nv04/nv04_fragprog.c @@ -1,7 +1,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi/tgsi_parse.h" diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c index 97f679731ea..02f7d210e31 100644 --- a/src/gallium/drivers/nv04/nv04_miptree.c +++ b/src/gallium/drivers/nv04/nv04_miptree.c @@ -1,6 +1,5 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "nv04_context.h" @@ -72,7 +71,6 @@ nv04_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) static void nv04_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) { - struct pipe_winsys *ws = screen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -80,7 +78,7 @@ nv04_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) struct nv04_miptree *nv04mt = (struct nv04_miptree *)mt; int l; - pipe_buffer_reference(ws, &nv04mt->buffer, NULL); + pipe_buffer_reference(screen, &nv04mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv04mt->level[l].image_offset) FREE(nv04mt->level[l].image_offset); @@ -101,7 +99,7 @@ nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps = ws->surface_alloc(ws); if (!ps) return NULL; - pipe_buffer_reference(ws, &ps->buffer, nv04mt->buffer); + pipe_buffer_reference(pscreen, &ps->buffer, nv04mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; diff --git a/src/gallium/drivers/nv04/nv04_prim_vbuf.c b/src/gallium/drivers/nv04/nv04_prim_vbuf.c index d3963d1f59d..19979fff795 100644 --- a/src/gallium/drivers/nv04/nv04_prim_vbuf.c +++ b/src/gallium/drivers/nv04/nv04_prim_vbuf.c @@ -1,9 +1,10 @@ -#include "draw/draw_vbuf.h" #include "pipe/p_debug.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" +#include "pipe/p_compiler.h" + +#include "draw/draw_vbuf.h" #include "nv04_context.h" #include "nv04_state.h" diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c index da09a3a5fe8..3966a29ffa9 100644 --- a/src/gallium/drivers/nv04/nv04_screen.c +++ b/src/gallium/drivers/nv04/nv04_screen.c @@ -1,5 +1,4 @@ #include "pipe/p_screen.h" -#include "pipe/p_util.h" #include "nv04_context.h" #include "nv04_screen.h" diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c index 668d875671f..ff1933b5508 100644 --- a/src/gallium/drivers/nv04/nv04_state.c +++ b/src/gallium/drivers/nv04/nv04_state.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi/tgsi_parse.h" diff --git a/src/gallium/drivers/nv04/nv04_surface.c b/src/gallium/drivers/nv04/nv04_surface.c index b13ebf9f9b5..57039483c62 100644 --- a/src/gallium/drivers/nv04/nv04_surface.c +++ b/src/gallium/drivers/nv04/nv04_surface.c @@ -28,10 +28,9 @@ #include "nv04_context.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" -#include "util/p_tile.h" +#include "util/u_tile.h" static void nv04_surface_copy(struct pipe_context *pipe, unsigned do_flip, diff --git a/src/gallium/drivers/nv04/nv04_vbo.c b/src/gallium/drivers/nv04/nv04_vbo.c index fbfe0cf406d..91f919d48ec 100644 --- a/src/gallium/drivers/nv04/nv04_vbo.c +++ b/src/gallium/drivers/nv04/nv04_vbo.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_context.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "nv04_context.h" #include "nv04_state.h" diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c index 9fcd0b0fc3a..e9b61daae7f 100644 --- a/src/gallium/drivers/nv10/nv10_context.c +++ b/src/gallium/drivers/nv10/nv10_context.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" -#include "pipe/p_util.h" #include "nv10_context.h" #include "nv10_screen.h" diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h index 2bdba53db8a..f3b56de25a7 100644 --- a/src/gallium/drivers/nv10/nv10_context.h +++ b/src/gallium/drivers/nv10/nv10_context.h @@ -4,6 +4,10 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" #include "draw/draw_vertex.h" diff --git a/src/gallium/drivers/nv10/nv10_fragprog.c b/src/gallium/drivers/nv10/nv10_fragprog.c index 137de9d53ef..698db5a16a9 100644 --- a/src/gallium/drivers/nv10/nv10_fragprog.c +++ b/src/gallium/drivers/nv10/nv10_fragprog.c @@ -1,7 +1,6 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi/tgsi_parse.h" diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index 9a68df29250..ad084e72b81 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -1,6 +1,5 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "nv10_context.h" @@ -79,7 +78,6 @@ nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) static void nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) { - struct pipe_winsys *ws = screen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -87,7 +85,7 @@ nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt; int l; - pipe_buffer_reference(ws, &nv10mt->buffer, NULL); + pipe_buffer_reference(screen, &nv10mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv10mt->level[l].image_offset) FREE(nv10mt->level[l].image_offset); @@ -115,7 +113,7 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, ps = ws->surface_alloc(ws); if (!ps) return NULL; - pipe_buffer_reference(ws, &ps->buffer, nv10mt->buffer); + pipe_buffer_reference(screen, &ps->buffer, nv10mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; diff --git a/src/gallium/drivers/nv10/nv10_prim_vbuf.c b/src/gallium/drivers/nv10/nv10_prim_vbuf.c index 930536b9468..62a8f6d89da 100644 --- a/src/gallium/drivers/nv10/nv10_prim_vbuf.c +++ b/src/gallium/drivers/nv10/nv10_prim_vbuf.c @@ -38,15 +38,14 @@ */ -#include "draw/draw_vbuf.h" #include "pipe/p_debug.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" #include "nv10_context.h" #include "nv10_state.h" +#include "draw/draw_vbuf.h" /** * Primitive renderer for nv10. @@ -180,10 +179,11 @@ nv10_vbuf_render_release_vertices( struct vbuf_render *render, struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); struct nv10_context *nv10 = nv10_render->nv10; struct pipe_winsys *winsys = nv10->pipe.winsys; + struct pipe_screen *pscreen = &nv10->screen->pipe; assert(nv10_render->buffer); winsys->buffer_unmap(winsys, nv10_render->buffer); - pipe_buffer_reference(winsys, &nv10_render->buffer, NULL); + pipe_buffer_reference(pscreen, &nv10_render->buffer, NULL); } diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c index 403f7b98cde..27a9edf9bba 100644 --- a/src/gallium/drivers/nv10/nv10_screen.c +++ b/src/gallium/drivers/nv10/nv10_screen.c @@ -1,5 +1,4 @@ #include "pipe/p_screen.h" -#include "pipe/p_util.h" #include "nv10_context.h" #include "nv10_screen.h" diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c index f902fd54b66..d2375aa2f64 100644 --- a/src/gallium/drivers/nv10/nv10_state.c +++ b/src/gallium/drivers/nv10/nv10_state.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" #include "tgsi/tgsi_parse.h" diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c index d21368d33f3..46c7e1d7536 100644 --- a/src/gallium/drivers/nv10/nv10_state_emit.c +++ b/src/gallium/drivers/nv10/nv10_state_emit.c @@ -1,5 +1,3 @@ -#include "pipe/p_util.h" - #include "nv10_context.h" #include "nv10_state.h" diff --git a/src/gallium/drivers/nv10/nv10_surface.c b/src/gallium/drivers/nv10/nv10_surface.c index 2e230ebbecc..875e4c58589 100644 --- a/src/gallium/drivers/nv10/nv10_surface.c +++ b/src/gallium/drivers/nv10/nv10_surface.c @@ -28,10 +28,9 @@ #include "nv10_context.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" -#include "util/p_tile.h" +#include "util/u_tile.h" static void nv10_surface_copy(struct pipe_context *pipe, unsigned do_flip, diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c index f024f534209..d0e788ac036 100644 --- a/src/gallium/drivers/nv10/nv10_vbo.c +++ b/src/gallium/drivers/nv10/nv10_vbo.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_context.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "nv10_context.h" #include "nv10_state.h" diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index eefc614e5b5..2bff28aca9c 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" -#include "pipe/p_util.h" #include "nv30_context.h" #include "nv30_screen.h" diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 823b34a7c3b..b9337697008 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -4,6 +4,10 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" #include "draw/draw_vertex.h" diff --git a/src/gallium/drivers/nv30/nv30_draw.c b/src/gallium/drivers/nv30/nv30_draw.c index aeeaf58f20a..74fc138c051 100644 --- a/src/gallium/drivers/nv30/nv30_draw.c +++ b/src/gallium/drivers/nv30/nv30_draw.c @@ -1,5 +1,4 @@ #include "draw/draw_pipe.h" -#include "pipe/p_util.h" #include "nv30_context.h" diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index a0e488c09b1..5c4f4da9481 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -1,6 +1,5 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "nv30_context.h" @@ -33,7 +32,7 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) if (swizzled) pitch = pt->nblocksx[l]; - pitch = align_int(pitch, 64); + pitch = align(pitch, 64); nv30mt->level[l].pitch = pitch * pt->block.size; nv30mt->level[l].image_offset = @@ -84,7 +83,6 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) static void nv30_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) { - struct pipe_winsys *ws = pscreen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -92,7 +90,7 @@ nv30_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) struct nv30_miptree *nv30mt = (struct nv30_miptree *)mt; int l; - pipe_buffer_reference(ws, &nv30mt->buffer, NULL); + pipe_buffer_reference(pscreen, &nv30mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv30mt->level[l].image_offset) FREE(nv30mt->level[l].image_offset); @@ -106,7 +104,6 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { - struct pipe_winsys *ws = pscreen->winsys; struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt; struct pipe_surface *ps; @@ -114,7 +111,7 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, if (!ps) return NULL; pipe_texture_reference(&ps->texture, pt); - pipe_buffer_reference(ws, &ps->buffer, nv30mt->buffer); + pipe_buffer_reference(pscreen, &ps->buffer, nv30mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index d5514c2aba5..a595e2eb225 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -1,5 +1,4 @@ #include "pipe/p_screen.h" -#include "pipe/p_util.h" #include "nv30_context.h" #include "nv30_screen.h" diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c index eceb5353152..fc66075c83f 100644 --- a/src/gallium/drivers/nv30/nv30_state.c +++ b/src/gallium/drivers/nv30/nv30_state.c @@ -1,6 +1,5 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "tgsi/tgsi_parse.h" diff --git a/src/gallium/drivers/nv30/nv30_surface.c b/src/gallium/drivers/nv30/nv30_surface.c index b22211ac86e..36f48877503 100644 --- a/src/gallium/drivers/nv30/nv30_surface.c +++ b/src/gallium/drivers/nv30/nv30_surface.c @@ -28,10 +28,10 @@ #include "nv30_context.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" -#include "util/p_tile.h" + +#include "util/u_tile.h" static void nv30_surface_copy(struct pipe_context *pipe, unsigned do_flip, diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c index b1c73793bf2..556f981d4a5 100644 --- a/src/gallium/drivers/nv30/nv30_vbo.c +++ b/src/gallium/drivers/nv30/nv30_vbo.c @@ -1,6 +1,5 @@ #include "pipe/p_context.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "nv30_context.h" #include "nv30_state.h" diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index a40f14895f4..cc63dd734bc 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -1,7 +1,6 @@ #include "draw/draw_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" -#include "pipe/p_util.h" #include "nv40_context.h" #include "nv40_screen.h" diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 8e60a81e68e..adcfbdd85a8 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -4,6 +4,10 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" #include "draw/draw_vertex.h" diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c index 2cf58e2950a..8e56cdc2fe0 100644 --- a/src/gallium/drivers/nv40/nv40_draw.c +++ b/src/gallium/drivers/nv40/nv40_draw.c @@ -1,6 +1,7 @@ -#include "pipe/p_util.h" #include "pipe/p_shader_tokens.h" +#include "util/u_pack_color.h" + #include "draw/draw_context.h" #include "draw/draw_vertex.h" #include "draw/draw_pipe.h" diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 38e1a5f04ca..6c54c37ef71 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -1,6 +1,5 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "nv40_context.h" @@ -33,7 +32,7 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) if (swizzled) pitch = pt->nblocksx[l]; - pitch = align_int(pitch, 64); + pitch = align(pitch, 64); nv40mt->level[l].pitch = pitch * pt->block.size; nv40mt->level[l].image_offset = @@ -84,7 +83,6 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) static void nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) { - struct pipe_winsys *ws = pscreen->winsys; struct pipe_texture *mt = *pt; *pt = NULL; @@ -92,7 +90,7 @@ nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) struct nv40_miptree *nv40mt = (struct nv40_miptree *)mt; int l; - pipe_buffer_reference(ws, &nv40mt->buffer, NULL); + pipe_buffer_reference(pscreen, &nv40mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { if (nv40mt->level[l].image_offset) FREE(nv40mt->level[l].image_offset); @@ -106,7 +104,6 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { - struct pipe_winsys *ws = pscreen->winsys; struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; struct pipe_surface *ps; @@ -114,7 +111,7 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, if (!ps) return NULL; pipe_texture_reference(&ps->texture, pt); - pipe_buffer_reference(ws, &ps->buffer, nv40mt->buffer); + pipe_buffer_reference(pscreen, &ps->buffer, nv40mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; @@ -148,7 +145,7 @@ nv40_miptree_surface_del(struct pipe_screen *pscreen, return; pipe_texture_reference(&ps->texture, NULL); - pipe_buffer_reference(pscreen->winsys, &ps->buffer, NULL); + pipe_buffer_reference(pscreen, &ps->buffer, NULL); FREE(ps); } diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index 0e1df89ee85..ada0238511d 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -1,5 +1,4 @@ #include "pipe/p_screen.h" -#include "pipe/p_util.h" #include "nv40_context.h" #include "nv40_screen.h" diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c index 63d0ecc9158..255c4b294d1 100644 --- a/src/gallium/drivers/nv40/nv40_state.c +++ b/src/gallium/drivers/nv40/nv40_state.c @@ -1,6 +1,5 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "draw/draw_context.h" diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c index 0916555d564..576af7c59ee 100644 --- a/src/gallium/drivers/nv40/nv40_surface.c +++ b/src/gallium/drivers/nv40/nv40_surface.c @@ -28,10 +28,10 @@ #include "nv40_context.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" -#include "util/p_tile.h" + +#include "util/u_tile.h" static void nv40_surface_copy(struct pipe_context *pipe, boolean do_flip, diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c index 755d5586b7b..09f6e79d32a 100644 --- a/src/gallium/drivers/nv40/nv40_vbo.c +++ b/src/gallium/drivers/nv40/nv40_vbo.c @@ -1,6 +1,5 @@ #include "pipe/p_context.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "nv40_context.h" #include "nv40_state.h" diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 07987c7d02d..b02c53f2095 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -23,7 +23,6 @@ #include "draw/draw_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" -#include "pipe/p_util.h" #include "nv50_context.h" #include "nv50_screen.h" diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 1c069f1625f..5d377f2d067 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -4,6 +4,10 @@ #include "pipe/p_context.h" #include "pipe/p_defines.h" #include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" #include "draw/draw_vertex.h" diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c index 4fd81bd27a1..2f6f607261e 100644 --- a/src/gallium/drivers/nv50/nv50_draw.c +++ b/src/gallium/drivers/nv50/nv50_draw.c @@ -21,7 +21,6 @@ */ #include "draw/draw_pipe.h" -#include "pipe/p_util.h" #include "nv50_context.h" diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index a02ad41885b..b0e8fe2f0b7 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -22,7 +22,6 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "nv50_context.h" @@ -62,7 +61,6 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) static void nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt) { - struct pipe_winsys *ws = pscreen->winsys; struct pipe_texture *pt = *ppt; *ppt = NULL; @@ -70,7 +68,7 @@ nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt) if (--pt->refcount <= 0) { struct nv50_miptree *mt = nv50_miptree(pt); - pipe_buffer_reference(ws, &mt->buffer, NULL); + pipe_buffer_reference(pscreen, &mt->buffer, NULL); FREE(mt); } } @@ -80,7 +78,6 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { - struct pipe_winsys *ws = pscreen->winsys; struct nv50_miptree *mt = nv50_miptree(pt); struct nv50_surface *s; struct pipe_surface *ps; @@ -91,7 +88,7 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps = &s->base; ps->refcount = 1; - ps->winsys = ws; + ps->winsys = pscreen->winsys; ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; @@ -104,7 +101,7 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps->status = PIPE_SURFACE_STATUS_DEFINED; pipe_texture_reference(&ps->texture, pt); - pipe_buffer_reference(ws, &ps->buffer, mt->buffer); + pipe_buffer_reference(pscreen, &ps->buffer, mt->buffer); return ps; } @@ -113,7 +110,6 @@ static void nv50_miptree_surface_del(struct pipe_screen *pscreen, struct pipe_surface **psurface) { - struct pipe_winsys *ws = pscreen->winsys; struct pipe_surface *ps = *psurface; struct nv50_surface *s = nv50_surface(ps); @@ -121,7 +117,7 @@ nv50_miptree_surface_del(struct pipe_screen *pscreen, if (--ps->refcount <= 0) { pipe_texture_reference(&ps->texture, NULL); - pipe_buffer_reference(ws, &ps->buffer, NULL); + pipe_buffer_reference(pscreen, &ps->buffer, NULL); FREE(s); } } diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index ec439239298..b5aef7dadd8 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -21,7 +21,6 @@ */ #include "pipe/p_screen.h" -#include "pipe/p_util.h" #include "nv50_context.h" #include "nv50_screen.h" diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 4055527d9f8..95f9d408b5e 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -22,7 +22,6 @@ #include "pipe/p_state.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" #include "tgsi/tgsi_parse.h" diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index a9daeee3694..5bf97d3a6bf 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -22,10 +22,10 @@ #include "nv50_context.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" -#include "util/p_tile.h" + +#include "util/u_tile.h" static void nv50_surface_copy(struct pipe_context *pipe, boolean flip, @@ -90,10 +90,10 @@ nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *ps, } static void -nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *ps) +nv50_surface_unmap(struct pipe_screen *pscreen, struct pipe_surface *ps) { - struct nouveau_winsys *nvws = nv50_screen(screen)->nvws; - struct pipe_winsys *ws = screen->winsys; + struct nouveau_winsys *nvws = nv50_screen(pscreen)->nvws; + struct pipe_winsys *ws = pscreen->winsys; struct nv50_surface *s = nv50_surface(ps); struct nv50_surface m = *s; @@ -104,7 +104,7 @@ nv50_surface_unmap(struct pipe_screen *screen, struct pipe_surface *ps) nvws->surface_copy(nvws, &s->base, 0, 0, &m.base, 0, 0, ps->width, ps->height); - pipe_buffer_reference(ws, &s->untiled, NULL); + pipe_buffer_reference(pscreen, &s->untiled, NULL); } void diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index c94531723bc..584336682e4 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -22,7 +22,6 @@ #include "pipe/p_context.h" #include "pipe/p_state.h" -#include "pipe/p_util.h" #include "nv50_context.h" diff --git a/src/gallium/winsys/dri/nouveau/Makefile b/src/gallium/winsys/dri/nouveau/Makefile deleted file mode 100644 index be630ff6d16..00000000000 --- a/src/gallium/winsys/dri/nouveau/Makefile +++ /dev/null @@ -1,45 +0,0 @@ - -TOP = ../../../../.. -include $(TOP)/configs/current - -LIBNAME = nouveau_dri.so - -MINIGLX_SOURCES = - -PIPE_DRIVERS = \ - $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ - $(TOP)/src/gallium/drivers/nv04/libnv04.a \ - $(TOP)/src/gallium/drivers/nv10/libnv10.a \ - $(TOP)/src/gallium/drivers/nv30/libnv30.a \ - $(TOP)/src/gallium/drivers/nv40/libnv40.a \ - $(TOP)/src/gallium/drivers/nv50/libnv50.a - -DRIVER_SOURCES = \ - nouveau_bo.c \ - nouveau_channel.c \ - nouveau_context.c \ - nouveau_device.c \ - nouveau_dma.c \ - nouveau_fence.c \ - nouveau_grobj.c \ - nouveau_lock.c \ - nouveau_notifier.c \ - nouveau_pushbuf.c \ - nouveau_resource.c \ - nouveau_screen.c \ - nouveau_swapbuffers.c \ - nouveau_winsys.c \ - nouveau_winsys_pipe.c \ - nouveau_winsys_softpipe.c \ - nv04_surface.c \ - nv50_surface.c - -C_SOURCES = \ - $(COMMON_GALLIUM_SOURCES) \ - $(DRIVER_SOURCES) - -ASM_SOURCES = - -include ../Makefile.template - -symlinks: diff --git a/src/gallium/winsys/dri/nouveau/nouveau_bo.c b/src/gallium/winsys/dri/nouveau/nouveau_bo.c deleted file mode 100644 index b5942994d94..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_bo.c +++ /dev/null @@ -1,470 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" -#include "nouveau_local.h" - -static void -nouveau_mem_free(struct nouveau_device *dev, struct drm_nouveau_mem_alloc *ma, - void **map) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct drm_nouveau_mem_free mf; - - if (map && *map) { - drmUnmap(*map, ma->size); - *map = NULL; - } - - if (ma->size) { - mf.offset = ma->offset; - mf.flags = ma->flags; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, - &mf, sizeof(mf)); - ma->size = 0; - } -} - -static int -nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, - uint32_t flags, struct drm_nouveau_mem_alloc *ma, void **map) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - int ret; - - ma->alignment = align; - ma->size = size; - ma->flags = flags; - if (map) - ma->flags |= NOUVEAU_MEM_MAPPED; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, ma, - sizeof(struct drm_nouveau_mem_alloc)); - if (ret) - return ret; - - if (map) { - ret = drmMap(nvdev->fd, ma->map_handle, ma->size, map); - if (ret) { - *map = NULL; - nouveau_mem_free(dev, ma, map); - return ret; - } - } - - return 0; -} - -static void -nouveau_bo_tmp_del(void *priv) -{ - struct nouveau_resource *r = priv; - - nouveau_fence_ref(NULL, (struct nouveau_fence **)&r->priv); - nouveau_resource_free(&r); -} - -static unsigned -nouveau_bo_tmp_max(struct nouveau_device_priv *nvdev) -{ - struct nouveau_resource *r = nvdev->sa_heap; - unsigned max = 0; - - while (r) { - if (r->in_use && !nouveau_fence(r->priv)->emitted) { - r = r->next; - continue; - } - - if (max < r->size) - max = r->size; - r = r->next; - } - - return max; -} - -static struct nouveau_resource * -nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, - struct nouveau_fence *fence) -{ - struct nouveau_device_priv *nvdev = nouveau_device(chan->device); - struct nouveau_resource *r = NULL; - struct nouveau_fence *ref = NULL; - - if (fence) - nouveau_fence_ref(fence, &ref); - else - nouveau_fence_new(chan, &ref); - assert(ref); - - while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) { - if (nouveau_bo_tmp_max(nvdev) < size) { - nouveau_fence_ref(NULL, &ref); - return NULL; - } - - nouveau_fence_flush(chan); - } - nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r); - - return r; -} - -int -nouveau_bo_init(struct nouveau_device *dev) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - int ret; - - ret = nouveau_mem_alloc(dev, 128*1024, 0, NOUVEAU_MEM_AGP | - NOUVEAU_MEM_PCI, &nvdev->sa, &nvdev->sa_map); - if (ret) - return ret; - - ret = nouveau_resource_init(&nvdev->sa_heap, 0, nvdev->sa.size); - if (ret) { - nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); - return ret; - } - - return 0; -} - -void -nouveau_bo_takedown(struct nouveau_device *dev) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); -} - -int -nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, - int size, struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo; - int ret; - - if (!dev || !bo || *bo) - return -EINVAL; - - nvbo = calloc(1, sizeof(struct nouveau_bo_priv)); - if (!nvbo) - return -ENOMEM; - nvbo->base.device = dev; - nvbo->base.size = size; - nvbo->base.handle = bo_to_ptr(nvbo); - nvbo->drm.alignment = align; - nvbo->refcount = 1; - - if (flags & NOUVEAU_BO_TILED) { - nvbo->tiled = 1; - if (flags & NOUVEAU_BO_ZTILE) - nvbo->tiled |= 2; - flags &= ~NOUVEAU_BO_TILED; - } - - ret = nouveau_bo_set_status(&nvbo->base, flags); - if (ret) { - free(nvbo); - return ret; - } - - *bo = &nvbo->base; - return 0; -} - -int -nouveau_bo_user(struct nouveau_device *dev, void *ptr, int size, - struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo; - - if (!dev || !bo || *bo) - return -EINVAL; - - nvbo = calloc(1, sizeof(*nvbo)); - if (!nvbo) - return -ENOMEM; - nvbo->base.device = dev; - - nvbo->sysmem = ptr; - nvbo->user = 1; - - nvbo->base.size = size; - nvbo->base.offset = nvbo->drm.offset; - nvbo->base.handle = bo_to_ptr(nvbo); - nvbo->refcount = 1; - *bo = &nvbo->base; - return 0; -} - -int -nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, - struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo = ptr_to_bo(handle); - - if (!dev || !bo || *bo) - return -EINVAL; - - nvbo->refcount++; - *bo = &nvbo->base; - return 0; -} - -static void -nouveau_bo_del_cb(void *priv) -{ - struct nouveau_bo_priv *nvbo = priv; - - nouveau_fence_ref(NULL, &nvbo->fence); - nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); - if (nvbo->sysmem && !nvbo->user) - free(nvbo->sysmem); - free(nvbo); -} - -void -nouveau_bo_del(struct nouveau_bo **bo) -{ - struct nouveau_bo_priv *nvbo; - - if (!bo || !*bo) - return; - nvbo = nouveau_bo(*bo); - *bo = NULL; - - if (--nvbo->refcount) - return; - - if (nvbo->pending) - nouveau_pushbuf_flush(nvbo->pending->channel, 0); - - if (nvbo->fence) - nouveau_fence_signal_cb(nvbo->fence, nouveau_bo_del_cb, nvbo); - else - nouveau_bo_del_cb(nvbo); -} - -int -nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - - if (!nvbo) - return -EINVAL; - - if (nvbo->pending && - (nvbo->pending->flags & NOUVEAU_BO_WR || flags & NOUVEAU_BO_WR)) { - nouveau_pushbuf_flush(nvbo->pending->channel, 0); - } - - if (flags & NOUVEAU_BO_WR) - nouveau_fence_wait(&nvbo->fence); - else - nouveau_fence_wait(&nvbo->wr_fence); - - if (nvbo->sysmem) - bo->map = nvbo->sysmem; - else - bo->map = nvbo->map; - return 0; -} - -void -nouveau_bo_unmap(struct nouveau_bo *bo) -{ - bo->map = NULL; -} - -static int -nouveau_bo_upload(struct nouveau_bo_priv *nvbo) -{ - if (nvbo->fence) - nouveau_fence_wait(&nvbo->fence); - memcpy(nvbo->map, nvbo->sysmem, nvbo->drm.size); - return 0; -} - -int -nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - struct drm_nouveau_mem_alloc new; - void *new_map = NULL, *new_sysmem = NULL; - unsigned new_flags = 0, ret; - - assert(!bo->map); - - /* Check current memtype vs requested, if they match do nothing */ - if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) - return 0; - if ((nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) && - (flags & NOUVEAU_BO_GART)) - return 0; - if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) - return 0; - - memset(&new, 0x00, sizeof(new)); - - /* Allocate new memory */ - if (flags & NOUVEAU_BO_VRAM) - new_flags |= NOUVEAU_MEM_FB; - else - if (flags & NOUVEAU_BO_GART) - new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); - - if (nvbo->tiled && flags) { - new_flags |= NOUVEAU_MEM_TILE; - if (nvbo->tiled & 2) - new_flags |= NOUVEAU_MEM_TILE_ZETA; - } - - if (new_flags) { - ret = nouveau_mem_alloc(bo->device, bo->size, - nvbo->drm.alignment, new_flags, - &new, &new_map); - if (ret) - return ret; - } else - if (!nvbo->user) { - new_sysmem = malloc(bo->size); - } - - /* Copy old -> new */ - /*XXX: use M2MF */ - if (nvbo->sysmem || nvbo->map) { - struct nouveau_pushbuf_bo *pbo = nvbo->pending; - nvbo->pending = NULL; - nouveau_bo_map(bo, NOUVEAU_BO_RD); - memcpy(new_map, bo->map, bo->size); - nouveau_bo_unmap(bo); - nvbo->pending = pbo; - } - - /* Free old memory */ - if (nvbo->fence) - nouveau_fence_wait(&nvbo->fence); - nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); - if (nvbo->sysmem && !nvbo->user) - free(nvbo->sysmem); - - nvbo->drm = new; - nvbo->map = new_map; - if (!nvbo->user) - nvbo->sysmem = new_sysmem; - bo->flags = flags; - bo->offset = nvbo->drm.offset; - return 0; -} - -static int -nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_device_priv *nvdev = nouveau_device(chan->device); - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - struct nouveau_resource *r; - - if (nvchan->user_charge + bo->size > nvdev->sa.size) - return 1; - - if (!(flags & NOUVEAU_BO_GART)) - return 1; - - r = nouveau_bo_tmp(chan, bo->size, fence); - if (!r) - return 1; - nvchan->user_charge += bo->size; - - memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); - - nvbo->offset = nvdev->sa.offset + r->start; - nvbo->flags = NOUVEAU_BO_GART; - return 0; -} - -static int -nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, - struct nouveau_fence *fence, uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - int ret; - - ret = nouveau_bo_set_status(bo, flags); - if (ret) { - nouveau_fence_flush(chan); - - ret = nouveau_bo_set_status(bo, flags); - if (ret) - return ret; - } - - if (nvbo->user) - nouveau_bo_upload(nvbo); - - nvbo->offset = nvbo->drm.offset; - if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) - nvbo->flags = NOUVEAU_BO_GART; - else - nvbo->flags = NOUVEAU_BO_VRAM; - - return 0; -} - -int -nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, - uint32_t flags) -{ - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - struct nouveau_fence *fence = nouveau_pushbuf(chan->pushbuf)->fence; - int ret; - - assert(bo->map == NULL); - - if (nvbo->user) { - ret = nouveau_bo_validate_user(chan, bo, fence, flags); - if (ret) { - ret = nouveau_bo_validate_bo(chan, bo, fence, flags); - if (ret) - return ret; - } - } else { - ret = nouveau_bo_validate_bo(chan, bo, fence, flags); - if (ret) - return ret; - } - - if (flags & NOUVEAU_BO_WR) - nouveau_fence_ref(fence, &nvbo->wr_fence); - nouveau_fence_ref(fence, &nvbo->fence); - return 0; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_channel.c b/src/gallium/winsys/dri/nouveau/nouveau_channel.c deleted file mode 100644 index 3b4dcd1ecf2..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_channel.c +++ /dev/null @@ -1,126 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" - -int -nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma, - uint32_t tt_ctxdma, struct nouveau_channel **chan) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct nouveau_channel_priv *nvchan; - int ret; - - if (!nvdev || !chan || *chan) - return -EINVAL; - - nvchan = calloc(1, sizeof(struct nouveau_channel_priv)); - if (!nvchan) - return -ENOMEM; - nvchan->base.device = dev; - - nvchan->drm.fb_ctxdma_handle = fb_ctxdma; - nvchan->drm.tt_ctxdma_handle = tt_ctxdma; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_CHANNEL_ALLOC, - &nvchan->drm, sizeof(nvchan->drm)); - if (ret) { - free(nvchan); - return ret; - } - - nvchan->base.id = nvchan->drm.channel; - if (nouveau_grobj_ref(&nvchan->base, nvchan->drm.fb_ctxdma_handle, - &nvchan->base.vram) || - nouveau_grobj_ref(&nvchan->base, nvchan->drm.tt_ctxdma_handle, - &nvchan->base.gart)) { - nouveau_channel_free((void *)&nvchan); - return -EINVAL; - } - - ret = drmMap(nvdev->fd, nvchan->drm.ctrl, nvchan->drm.ctrl_size, - (void*)&nvchan->user); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - nvchan->put = &nvchan->user[0x40/4]; - nvchan->get = &nvchan->user[0x44/4]; - nvchan->ref_cnt = &nvchan->user[0x48/4]; - - ret = drmMap(nvdev->fd, nvchan->drm.notifier, nvchan->drm.notifier_size, - (drmAddressPtr)&nvchan->notifier_block); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - - ret = drmMap(nvdev->fd, nvchan->drm.cmdbuf, nvchan->drm.cmdbuf_size, - (void*)&nvchan->pushbuf); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - - ret = nouveau_grobj_alloc(&nvchan->base, 0x00000000, 0x0030, - &nvchan->base.nullobj); - if (ret) { - nouveau_channel_free((void *)&nvchan); - return ret; - } - - nouveau_dma_channel_init(&nvchan->base); - nouveau_pushbuf_init(&nvchan->base); - - *chan = &nvchan->base; - return 0; -} - -void -nouveau_channel_free(struct nouveau_channel **chan) -{ - struct nouveau_channel_priv *nvchan; - struct nouveau_device_priv *nvdev; - struct drm_nouveau_channel_free cf; - - if (!chan || !*chan) - return; - nvchan = nouveau_channel(*chan); - *chan = NULL; - nvdev = nouveau_device(nvchan->base.device); - - FIRE_RING_CH(&nvchan->base); - - nouveau_grobj_free(&nvchan->base.vram); - nouveau_grobj_free(&nvchan->base.gart); - nouveau_grobj_free(&nvchan->base.nullobj); - - cf.channel = nvchan->drm.channel; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf)); - free(nvchan); -} - - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.c b/src/gallium/winsys/dri/nouveau/nouveau_context.c deleted file mode 100644 index 74413c408f3..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.c +++ /dev/null @@ -1,346 +0,0 @@ -#include "main/glheader.h" -#include "glapi/glthread.h" -#include -#include "utils.h" - -#include "state_tracker/st_public.h" -#include "state_tracker/st_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_context.h" -#include "pipe/p_screen.h" - -#include "nouveau_context.h" -#include "nouveau_dri.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_winsys_pipe.h" - -#ifdef DEBUG -static const struct dri_debug_control debug_control[] = { - { "bo", DEBUG_BO }, - { NULL, 0 } -}; -int __nouveau_debug = 0; -#endif - -static void -nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) -{ - nouveau_grobj_free(&nvc->NvCtxSurf2D); - nouveau_grobj_free(&nvc->NvImageBlit); - nouveau_grobj_free(&nvc->NvGdiRect); - nouveau_grobj_free(&nvc->NvM2MF); - nouveau_grobj_free(&nvc->Nv2D); - nouveau_grobj_free(&nvc->NvSwzSurf); - nouveau_grobj_free(&nvc->NvSIFM); - - nouveau_notifier_free(&nvc->sync_notifier); - - nouveau_channel_free(&nvc->channel); - - FREE(nvc); -} - -static struct nouveau_channel_context * -nouveau_channel_context_create(struct nouveau_device *dev) -{ - struct nouveau_channel_context *nvc; - int ret; - - nvc = CALLOC_STRUCT(nouveau_channel_context); - if (!nvc) - return NULL; - - if ((ret = nouveau_channel_alloc(dev, 0x8003d001, 0x8003d002, - &nvc->channel))) { - NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); - nouveau_channel_context_destroy(nvc); - return NULL; - } - - nvc->next_handle = 0x80000000; - - if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1, - &nvc->sync_notifier))) { - NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); - nouveau_channel_context_destroy(nvc); - return NULL; - } - - switch (dev->chipset & 0xf0) { - case 0x50: - case 0x80: - case 0x90: - ret = nouveau_surface_channel_create_nv50(nvc); - break; - default: - ret = nouveau_surface_channel_create_nv04(nvc); - break; - } - - if (ret) { - NOUVEAU_ERR("Error initialising surface objects: %d\n", ret); - nouveau_channel_context_destroy(nvc); - return NULL; - } - - return nvc; -} - -GLboolean -nouveau_context_create(const __GLcontextModes *glVis, - __DRIcontextPrivate *driContextPriv, - void *sharedContextPrivate) -{ - __DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv; - struct nouveau_screen *nv_screen = driScrnPriv->private; - struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); - struct pipe_context *pipe = NULL; - struct st_context *st_share = NULL; - struct nouveau_channel_context *nvc = NULL; - struct nouveau_device *dev = nv_screen->device; - int i; - - if (sharedContextPrivate) { - st_share = ((struct nouveau_context *)sharedContextPrivate)->st; - } - - switch (dev->chipset & 0xf0) { - case 0x10: - case 0x20: - /* NV10 */ - case 0x30: - /* NV30 */ - case 0x40: - case 0x60: - /* NV40 */ - case 0x50: - case 0x80: - case 0x90: - /* G80 */ - break; - default: - NOUVEAU_ERR("Unsupported chipset: NV%02x\n", dev->chipset); - return GL_FALSE; - } - - driContextPriv->driverPrivate = (void *)nv; - nv->nv_screen = nv_screen; - nv->dri_screen = driScrnPriv; - - { - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - nvdev->ctx = driContextPriv->hHWContext; - nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; - } - - driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, - nv->dri_screen->myNum, "nouveau"); -#ifdef DEBUG - __nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"), - debug_control); -#endif - - /*XXX: Hack up a fake region and buffer object for front buffer. - * This will go away with TTM, replaced with a simple reference - * of the front buffer handle passed to us by the DDX. - */ - { - struct pipe_surface *fb_surf; - struct nouveau_pipe_buffer *fb_buf; - struct nouveau_bo_priv *fb_bo; - - fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); - fb_bo->drm.offset = nv_screen->front_offset; - fb_bo->drm.flags = NOUVEAU_MEM_FB; - fb_bo->drm.size = nv_screen->front_pitch * - nv_screen->front_height; - fb_bo->refcount = 1; - fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; - fb_bo->base.offset = fb_bo->drm.offset; - fb_bo->base.handle = (unsigned long)fb_bo; - fb_bo->base.size = fb_bo->drm.size; - fb_bo->base.device = nv_screen->device; - - fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer)); - fb_buf->bo = &fb_bo->base; - - fb_surf = calloc(1, sizeof(struct pipe_surface)); - if (nv_screen->front_cpp == 2) - fb_surf->format = PIPE_FORMAT_R5G6B5_UNORM; - else - fb_surf->format = PIPE_FORMAT_A8R8G8B8_UNORM; - pf_get_block(fb_surf->format, &fb_surf->block); - fb_surf->width = nv_screen->front_pitch / nv_screen->front_cpp; - fb_surf->height = nv_screen->front_height; - fb_surf->stride = fb_surf->width * fb_surf->block.size; - fb_surf->refcount = 1; - fb_surf->buffer = &fb_buf->base; - - nv->frontbuffer = fb_surf; - } - - /* Attempt to share a single channel between multiple contexts from - * a single process. - */ - nvc = nv_screen->nvc; - if (!nvc && st_share) { - struct nouveau_context *snv = st_share->pipe->priv; - if (snv) { - nvc = snv->nvc; - } - } - - /*XXX: temporary - disable multi-context/single-channel on pre-NV4x */ - switch (dev->chipset & 0xf0) { - case 0x40: - case 0x60: - /* NV40 class */ - case 0x50: - case 0x80: - case 0x90: - /* G80 class */ - break; - default: - nvc = NULL; - break; - } - - if (!nvc) { - nvc = nouveau_channel_context_create(dev); - if (!nvc) { - NOUVEAU_ERR("Failed initialising GPU context\n"); - return GL_FALSE; - } - nv_screen->nvc = nvc; - } - - nvc->refcount++; - nv->nvc = nvc; - - /* Find a free slot for a pipe context, allocate a new one if needed */ - nv->pctx_id = -1; - for (i = 0; i < nvc->nr_pctx; i++) { - if (nvc->pctx[i] == NULL) { - nv->pctx_id = i; - break; - } - } - - if (nv->pctx_id < 0) { - nv->pctx_id = nvc->nr_pctx++; - nvc->pctx = - realloc(nvc->pctx, - sizeof(struct pipe_context *) * nvc->nr_pctx); - } - - /* Create pipe */ - switch (dev->chipset & 0xf0) { - case 0x50: - case 0x80: - case 0x90: - if (nouveau_surface_init_nv50(nv)) - return GL_FALSE; - break; - default: - if (nouveau_surface_init_nv04(nv)) - return GL_FALSE; - break; - } - - if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { - struct pipe_screen *pscreen; - - pipe = nouveau_pipe_create(nv); - if (!pipe) - NOUVEAU_ERR("Couldn't create hw pipe\n"); - pscreen = nvc->pscreen; - - nv->cap.hw_vertex_buffer = - pscreen->get_param(pscreen, NOUVEAU_CAP_HW_VTXBUF); - nv->cap.hw_index_buffer = - pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF); - } - - if (!pipe) { - NOUVEAU_MSG("Using softpipe\n"); - pipe = nouveau_create_softpipe(nv); - if (!pipe) { - NOUVEAU_ERR("Error creating pipe, bailing\n"); - return GL_FALSE; - } - } - - pipe->priv = nv; - nv->st = st_create_context(pipe, glVis, st_share); - return GL_TRUE; -} - -void -nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) -{ - struct nouveau_context *nv = driContextPriv->driverPrivate; - struct nouveau_channel_context *nvc = nv->nvc; - - assert(nv); - - st_finish(nv->st); - st_destroy_context(nv->st); - - if (nv->pctx_id >= 0) { - nvc->pctx[nv->pctx_id] = NULL; - if (--nvc->refcount <= 0) { - nouveau_channel_context_destroy(nvc); - nv->nv_screen->nvc = NULL; - } - } - - free(nv); -} - -GLboolean -nouveau_context_bind(__DRIcontextPrivate *driContextPriv, - __DRIdrawablePrivate *driDrawPriv, - __DRIdrawablePrivate *driReadPriv) -{ - struct nouveau_context *nv; - struct nouveau_framebuffer *draw, *read; - - if (!driContextPriv) { - st_make_current(NULL, NULL, NULL); - return GL_TRUE; - } - - nv = driContextPriv->driverPrivate; - draw = driDrawPriv->driverPrivate; - read = driReadPriv->driverPrivate; - - st_make_current(nv->st, draw->stfb, read->stfb); - - if ((nv->dri_drawable != driDrawPriv) || - (nv->last_stamp != driDrawPriv->lastStamp)) { - nv->dri_drawable = driDrawPriv; - st_resize_framebuffer(draw->stfb, driDrawPriv->w, - driDrawPriv->h); - nv->last_stamp = driDrawPriv->lastStamp; - } - - if (driDrawPriv != driReadPriv) { - st_resize_framebuffer(read->stfb, driReadPriv->w, - driReadPriv->h); - } - - return GL_TRUE; -} - -GLboolean -nouveau_context_unbind(__DRIcontextPrivate *driContextPriv) -{ - struct nouveau_context *nv = driContextPriv->driverPrivate; - (void)nv; - - st_flush(nv->st, 0, NULL); - return GL_TRUE; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_context.h b/src/gallium/winsys/dri/nouveau/nouveau_context.h deleted file mode 100644 index 77e2147a2c7..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_context.h +++ /dev/null @@ -1,113 +0,0 @@ -#ifndef __NOUVEAU_CONTEXT_H__ -#define __NOUVEAU_CONTEXT_H__ - -#include "dri_util.h" -#include "xmlconfig.h" - -#include "nouveau/nouveau_winsys.h" -#include "nouveau_drmif.h" -#include "nouveau_dma.h" - -struct nouveau_framebuffer { - struct st_framebuffer *stfb; -}; - -struct nouveau_channel_context { - struct pipe_screen *pscreen; - int refcount; - - unsigned cur_pctx; - unsigned nr_pctx; - struct pipe_context **pctx; - - struct nouveau_channel *channel; - - struct nouveau_notifier *sync_notifier; - - /* Common */ - struct nouveau_grobj *NvM2MF; - /* NV04-NV40 */ - struct nouveau_grobj *NvCtxSurf2D; - struct nouveau_grobj *NvSwzSurf; - struct nouveau_grobj *NvImageBlit; - struct nouveau_grobj *NvGdiRect; - struct nouveau_grobj *NvSIFM; - /* G80 */ - struct nouveau_grobj *Nv2D; - - uint32_t next_handle; - uint32_t next_subchannel; - uint32_t next_sequence; -}; - -struct nouveau_context { - struct st_context *st; - - /* DRI stuff */ - __DRIscreenPrivate *dri_screen; - __DRIdrawablePrivate *dri_drawable; - unsigned int last_stamp; - driOptionCache dri_option_cache; - drm_context_t drm_context; - drmLock drm_lock; - GLboolean locked; - struct nouveau_screen *nv_screen; - struct pipe_surface *frontbuffer; - - struct { - int hw_vertex_buffer; - int hw_index_buffer; - } cap; - - /* Hardware context */ - struct nouveau_channel_context *nvc; - int pctx_id; - - /* pipe_surface accel */ - struct pipe_surface *surf_src, *surf_dst; - unsigned surf_src_offset, surf_dst_offset; - int (*surface_copy_prep)(struct nouveau_context *, - struct pipe_surface *dst, - struct pipe_surface *src); - void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h); - void (*surface_copy_done)(struct nouveau_context *); - int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, - unsigned, unsigned, unsigned, unsigned, unsigned); -}; - -extern GLboolean nouveau_context_create(const __GLcontextModes *, - __DRIcontextPrivate *, void *); -extern void nouveau_context_destroy(__DRIcontextPrivate *); -extern GLboolean nouveau_context_bind(__DRIcontextPrivate *, - __DRIdrawablePrivate *draw, - __DRIdrawablePrivate *read); -extern GLboolean nouveau_context_unbind(__DRIcontextPrivate *); - -#ifdef DEBUG -extern int __nouveau_debug; - -#define DEBUG_BO (1 << 0) - -#define DBG(flag, ...) do { \ - if (__nouveau_debug & (DEBUG_##flag)) \ - NOUVEAU_ERR(__VA_ARGS__); \ -} while(0) -#else -#define DBG(flag, ...) -#endif - -extern void LOCK_HARDWARE(struct nouveau_context *); -extern void UNLOCK_HARDWARE(struct nouveau_context *); - -extern int -nouveau_surface_channel_create_nv04(struct nouveau_channel_context *); -extern int -nouveau_surface_channel_create_nv50(struct nouveau_channel_context *); -extern int nouveau_surface_init_nv04(struct nouveau_context *); -extern int nouveau_surface_init_nv50(struct nouveau_context *); - -extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int); -extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *); - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_device.c b/src/gallium/winsys/dri/nouveau/nouveau_device.c deleted file mode 100644 index 0b452fcd02d..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_device.c +++ /dev/null @@ -1,159 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" - -int -nouveau_device_open_existing(struct nouveau_device **dev, int close, - int fd, drm_context_t ctx) -{ - struct nouveau_device_priv *nvdev; - int ret; - - if (!dev || *dev) - return -EINVAL; - - nvdev = calloc(1, sizeof(*nvdev)); - if (!nvdev) - return -ENOMEM; - nvdev->fd = fd; - nvdev->ctx = ctx; - nvdev->needs_close = close; - - drmCommandNone(nvdev->fd, DRM_NOUVEAU_CARD_INIT); - - if ((ret = nouveau_bo_init(&nvdev->base))) { - nouveau_device_close((void *)&nvdev); - return ret; - } - - { - uint64_t value; - - ret = nouveau_device_get_param(&nvdev->base, - NOUVEAU_GETPARAM_CHIPSET_ID, - &value); - if (ret) { - nouveau_device_close((void *)&nvdev); - return ret; - } - nvdev->base.chipset = value; - } - - *dev = &nvdev->base; - return 0; -} - -int -nouveau_device_open(struct nouveau_device **dev, const char *busid) -{ - drm_context_t ctx; - int fd, ret; - - if (!dev || *dev) - return -EINVAL; - - fd = drmOpen("nouveau", busid); - if (fd < 0) - return -EINVAL; - - ret = drmCreateContext(fd, &ctx); - if (ret) { - drmClose(fd); - return ret; - } - - ret = nouveau_device_open_existing(dev, 1, fd, ctx); - if (ret) { - drmDestroyContext(fd, ctx); - drmClose(fd); - return ret; - } - - return 0; -} - -void -nouveau_device_close(struct nouveau_device **dev) -{ - struct nouveau_device_priv *nvdev; - - if (dev || !*dev) - return; - nvdev = nouveau_device(*dev); - *dev = NULL; - - nouveau_bo_takedown(&nvdev->base); - - if (nvdev->needs_close) { - drmDestroyContext(nvdev->fd, nvdev->ctx); - drmClose(nvdev->fd); - } - free(nvdev); -} - -int -nouveau_device_get_param(struct nouveau_device *dev, - uint64_t param, uint64_t *value) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct drm_nouveau_getparam g; - int ret; - - if (!nvdev || !value) - return -EINVAL; - - g.param = param; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_GETPARAM, - &g, sizeof(g)); - if (ret) - return ret; - - *value = g.value; - return 0; -} - -int -nouveau_device_set_param(struct nouveau_device *dev, - uint64_t param, uint64_t value) -{ - struct nouveau_device_priv *nvdev = nouveau_device(dev); - struct drm_nouveau_setparam s; - int ret; - - if (!nvdev) - return -EINVAL; - - s.param = param; - s.value = value; - ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_SETPARAM, - &s, sizeof(s)); - if (ret) - return ret; - - return 0; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_dma.c b/src/gallium/winsys/dri/nouveau/nouveau_dma.c deleted file mode 100644 index f8a8ba04f6d..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_dma.c +++ /dev/null @@ -1,219 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" -#include "nouveau_local.h" - -static inline uint32_t -READ_GET(struct nouveau_channel_priv *nvchan) -{ - return *nvchan->get; -} - -static inline void -WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) -{ - uint32_t put = ((val << 2) + nvchan->dma->base); - volatile int dum; - - NOUVEAU_DMA_BARRIER; - dum = READ_GET(nvchan); - - *nvchan->put = put; - nvchan->dma->put = val; -#ifdef NOUVEAU_DMA_TRACE - NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put); -#endif - - NOUVEAU_DMA_BARRIER; -} - -static inline int -LOCAL_GET(struct nouveau_dma_priv *dma, uint32_t *val) -{ - uint32_t get = *val; - - if (get >= dma->base && get <= (dma->base + (dma->max << 2))) { - *val = (get - dma->base) >> 2; - return 1; - } - - return 0; -} - -void -nouveau_dma_channel_init(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - int i; - - nvchan->dma = &nvchan->dma_master; - nvchan->dma->base = nvchan->drm.put_base; - nvchan->dma->cur = nvchan->dma->put = 0; - nvchan->dma->max = (nvchan->drm.cmdbuf_size >> 2) - 2; - nvchan->dma->free = nvchan->dma->max - nvchan->dma->cur; - - RING_SPACE_CH(chan, RING_SKIPS); - for (i = 0; i < RING_SKIPS; i++) - OUT_RING_CH(chan, 0); -} - -#define CHECK_TIMEOUT() do { \ - if ((NOUVEAU_TIME_MSEC() - t_start) > NOUVEAU_DMA_TIMEOUT) \ - return - EBUSY; \ -} while(0) - -int -nouveau_dma_wait(struct nouveau_channel *chan, int size) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - uint32_t get, t_start; - - FIRE_RING_CH(chan); - - t_start = NOUVEAU_TIME_MSEC(); - while (dma->free < size) { - CHECK_TIMEOUT(); - - get = READ_GET(nvchan); - if (!LOCAL_GET(dma, &get)) - continue; - - if (dma->put >= get) { - dma->free = dma->max - dma->cur; - - if (dma->free < size) { -#ifdef NOUVEAU_DMA_DEBUG - dma->push_free = 1; -#endif - OUT_RING_CH(chan, 0x20000000 | dma->base); - if (get <= RING_SKIPS) { - /*corner case - will be idle*/ - if (dma->put <= RING_SKIPS) - WRITE_PUT(nvchan, - RING_SKIPS + 1); - - do { - CHECK_TIMEOUT(); - get = READ_GET(nvchan); - if (!LOCAL_GET(dma, &get)) - get = 0; - } while (get <= RING_SKIPS); - } - - WRITE_PUT(nvchan, RING_SKIPS); - dma->cur = dma->put = RING_SKIPS; - dma->free = get - (RING_SKIPS + 1); - } - } else { - dma->free = get - dma->cur - 1; - } - } - - return 0; -} - -#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF -static void -nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - unsigned mthd_count = 0; - - while (get != put) { - uint32_t gpuget = (get << 2) + nvchan->drm.put_base; - uint32_t data; - - if (get < 0 || get >= nvchan->drm.cmdbuf_size) { - NOUVEAU_ERR("DMA_PT 0x%08x\n", gpuget); - assert(0); - } - data = nvchan->pushbuf[get++]; - - if (mthd_count) { - NOUVEAU_MSG("0x%08x 0x%08x\n", gpuget, data); - mthd_count--; - continue; - } - - switch (data & 0x60000000) { - case 0x00000000: - mthd_count = (data >> 18) & 0x7ff; - NOUVEAU_MSG("0x%08x 0x%08x MTHD " - "Sc %d Mthd 0x%04x Size %d\n", - gpuget, data, (data>>13) & 7, data & 0x1ffc, - mthd_count); - break; - case 0x20000000: - get = (data & 0x1ffffffc) >> 2; - NOUVEAU_MSG("0x%08x 0x%08x JUMP 0x%08x\n", - gpuget, data, data & 0x1ffffffc); - continue; - case 0x40000000: - mthd_count = (data >> 18) & 0x7ff; - NOUVEAU_MSG("0x%08x 0x%08x NINC " - "Sc %d Mthd 0x%04x Size %d\n", - gpuget, data, (data>>13) & 7, data & 0x1ffc, - mthd_count); - break; - case 0x60000000: - /* DMA_OPCODE_CALL apparently, doesn't seem to work on - * my NV40 at least.. - */ - /* fall-through */ - default: - NOUVEAU_MSG("DMA_PUSHER 0x%08x 0x%08x\n", - gpuget, data); - assert(0); - } - } -} -#endif - -void -nouveau_dma_kickoff(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - - if (dma->cur == dma->put) - return; - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free) { - NOUVEAU_ERR("Packet incomplete: %d left\n", dma->push_free); - return; - } -#endif - -#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF - nouveau_dma_parse_pushbuf(chan, dma->put, dma->cur); -#endif - - WRITE_PUT(nvchan, dma->cur); -} diff --git a/src/gallium/winsys/dri/nouveau/nouveau_dma.h b/src/gallium/winsys/dri/nouveau/nouveau_dma.h deleted file mode 100644 index cfa6d26e828..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_dma.h +++ /dev/null @@ -1,143 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_DMA_H__ -#define __NOUVEAU_DMA_H__ - -#include -#include "nouveau_drmif.h" -#include "nouveau_local.h" - -#define RING_SKIPS 8 - -extern int nouveau_dma_wait(struct nouveau_channel *chan, int size); -extern void nouveau_dma_subc_bind(struct nouveau_grobj *); -extern void nouveau_dma_channel_init(struct nouveau_channel *); -extern void nouveau_dma_kickoff(struct nouveau_channel *); - -#ifdef NOUVEAU_DMA_DEBUG -static char faulty[1024]; -#endif - -static inline void -nouveau_dma_out(struct nouveau_channel *chan, uint32_t data) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free == 0) { - NOUVEAU_ERR("No space left in packet at %s\n", faulty); - return; - } - dma->push_free--; -#endif -#ifdef NOUVEAU_DMA_TRACE - { - uint32_t offset = (dma->cur << 2) + dma->base; - NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", - nvchan->drm.channel, offset, data); - } -#endif - nvchan->pushbuf[dma->cur + (dma->base - nvchan->drm.put_base)/4] = data; - dma->cur++; -} - -static inline void -nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - (void)dma; - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free < size) { - NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", - dma->push_free, size); - return; - } -#endif -#ifdef NOUVEAU_DMA_TRACE - while (size--) { - nouveau_dma_out(chan, *ptr); - ptr++; - } -#else - memcpy(&nvchan->pushbuf[dma->cur], ptr, size << 2); -#ifdef NOUVEAU_DMA_DEBUG - dma->push_free -= size; -#endif - dma->cur += size; -#endif -} - -static inline void -nouveau_dma_space(struct nouveau_channel *chan, int size) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - - if (dma->free < size) { - if (nouveau_dma_wait(chan, size) && chan->hang_notify) - chan->hang_notify(chan); - } - dma->free -= size; -#ifdef NOUVEAU_DMA_DEBUG - dma->push_free = size; -#endif -} - -static inline void -nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, - int method, int size, const char* file, int line) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *dma = nvchan->dma; - (void)dma; - -#ifdef NOUVEAU_DMA_TRACE - NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, - grobj->handle, grobj->subc, method, size); -#endif - -#ifdef NOUVEAU_DMA_DEBUG - if (dma->push_free) { - NOUVEAU_ERR("Previous packet incomplete: %d left at %s\n", - dma->push_free, faulty); - return; - } - sprintf(faulty,"%s:%d",file,line); -#endif - - nouveau_dma_space(chan, (size + 1)); - nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); -} - -#define RING_SPACE_CH(ch,sz) nouveau_dma_space((ch), (sz)) -#define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) -#define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) -#define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ - (dwords)) -#define FIRE_RING_CH(ch) nouveau_dma_kickoff((ch)) -#define WAIT_RING_CH(ch,sz) nouveau_dma_wait((ch), (sz)) - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_dri.h b/src/gallium/winsys/dri/nouveau/nouveau_dri.h deleted file mode 100644 index 1207c2d609c..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_dri.h +++ /dev/null @@ -1,28 +0,0 @@ -#ifndef _NOUVEAU_DRI_ -#define _NOUVEAU_DRI_ - -#include "xf86drm.h" -#include "drm.h" -#include "nouveau_drm.h" - -struct nouveau_dri { - uint32_t device_id; /**< \brief PCI device ID */ - uint32_t width; /**< \brief width in pixels of display */ - uint32_t height; /**< \brief height in scanlines of display */ - uint32_t depth; /**< \brief depth of display (8, 15, 16, 24) */ - uint32_t bpp; /**< \brief bit depth of display (8, 16, 24, 32) */ - - uint32_t bus_type; /**< \brief ths bus type */ - uint32_t bus_mode; /**< \brief bus mode (used for AGP, maybe also for PCI-E ?) */ - - uint32_t front_offset; /**< \brief front buffer offset */ - uint32_t front_pitch; /**< \brief front buffer pitch */ - uint32_t back_offset; /**< \brief private back buffer offset */ - uint32_t back_pitch; /**< \brief private back buffer pitch */ - uint32_t depth_offset; /**< \brief private depth buffer offset */ - uint32_t depth_pitch; /**< \brief private depth buffer pitch */ - -}; - -#endif - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h b/src/gallium/winsys/dri/nouveau/nouveau_drmif.h deleted file mode 100644 index dcd6a5eb0a4..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_drmif.h +++ /dev/null @@ -1,310 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef __NOUVEAU_DRMIF_H__ -#define __NOUVEAU_DRMIF_H__ - -#include -#include -#include - -#include "nouveau/nouveau_device.h" -#include "nouveau/nouveau_channel.h" -#include "nouveau/nouveau_grobj.h" -#include "nouveau/nouveau_notifier.h" -#include "nouveau/nouveau_bo.h" -#include "nouveau/nouveau_resource.h" -#include "nouveau/nouveau_pushbuf.h" - -struct nouveau_device_priv { - struct nouveau_device base; - - int fd; - drm_context_t ctx; - drmLock *lock; - int needs_close; - - struct drm_nouveau_mem_alloc sa; - void *sa_map; - struct nouveau_resource *sa_heap; -}; -#define nouveau_device(n) ((struct nouveau_device_priv *)(n)) - -extern int -nouveau_device_open_existing(struct nouveau_device **, int close, - int fd, drm_context_t ctx); - -extern int -nouveau_device_open(struct nouveau_device **, const char *busid); - -extern void -nouveau_device_close(struct nouveau_device **); - -extern int -nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v); - -extern int -nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val); - -struct nouveau_fence { - struct nouveau_channel *channel; -}; - -struct nouveau_fence_cb { - struct nouveau_fence_cb *next; - void (*func)(void *); - void *priv; -}; - -struct nouveau_fence_priv { - struct nouveau_fence base; - int refcount; - - struct nouveau_fence *next; - struct nouveau_fence_cb *signal_cb; - - uint32_t sequence; - int emitted; - int signalled; -}; -#define nouveau_fence(n) ((struct nouveau_fence_priv *)(n)) - -extern int -nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **); - -extern int -nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); - -extern int -nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *); - -extern void -nouveau_fence_emit(struct nouveau_fence *); - -extern int -nouveau_fence_wait(struct nouveau_fence **); - -extern void -nouveau_fence_flush(struct nouveau_channel *); - -struct nouveau_pushbuf_reloc { - struct nouveau_pushbuf_bo *pbbo; - uint32_t *ptr; - uint32_t flags; - uint32_t data; - uint32_t vor; - uint32_t tor; -}; - -struct nouveau_pushbuf_bo { - struct nouveau_channel *channel; - struct nouveau_bo *bo; - unsigned flags; - unsigned handled; -}; - -#define NOUVEAU_PUSHBUF_MAX_BUFFERS 1024 -#define NOUVEAU_PUSHBUF_MAX_RELOCS 1024 -struct nouveau_pushbuf_priv { - struct nouveau_pushbuf base; - - struct nouveau_fence *fence; - - unsigned nop_jump; - unsigned start; - unsigned size; - - struct nouveau_pushbuf_bo *buffers; - unsigned nr_buffers; - struct nouveau_pushbuf_reloc *relocs; - unsigned nr_relocs; -}; -#define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) - -#define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o)) -#define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h)) -#define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o)) -#define ptr_to_pbrel(h) ((struct nouveau_pushbuf_reloc *)(unsigned long)(h)) -#define bo_to_ptr(o) ((uint64_t)(unsigned long)(o)) -#define ptr_to_bo(h) ((struct nouveau_bo_priv *)(unsigned long)(h)) - -extern int -nouveau_pushbuf_init(struct nouveau_channel *); - -extern int -nouveau_pushbuf_flush(struct nouveau_channel *, unsigned min); - -extern int -nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr, - struct nouveau_bo *, uint32_t data, uint32_t flags, - uint32_t vor, uint32_t tor); - -struct nouveau_dma_priv { - uint32_t base; - uint32_t max; - uint32_t cur; - uint32_t put; - uint32_t free; - - int push_free; -} dma; - -struct nouveau_channel_priv { - struct nouveau_channel base; - - struct drm_nouveau_channel_alloc drm; - - uint32_t *pushbuf; - void *notifier_block; - - volatile uint32_t *user; - volatile uint32_t *put; - volatile uint32_t *get; - volatile uint32_t *ref_cnt; - - struct nouveau_dma_priv dma_master; - struct nouveau_dma_priv dma_bufmgr; - struct nouveau_dma_priv *dma; - - struct nouveau_fence *fence_head; - struct nouveau_fence *fence_tail; - uint32_t fence_sequence; - - struct nouveau_pushbuf_priv pb; - - unsigned user_charge; -}; -#define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) - -extern int -nouveau_channel_alloc(struct nouveau_device *, uint32_t fb, uint32_t tt, - struct nouveau_channel **); - -extern void -nouveau_channel_free(struct nouveau_channel **); - -struct nouveau_grobj_priv { - struct nouveau_grobj base; -}; -#define nouveau_grobj(n) ((struct nouveau_grobj_priv *)(n)) - -extern int nouveau_grobj_alloc(struct nouveau_channel *, uint32_t handle, - int class, struct nouveau_grobj **); -extern int nouveau_grobj_ref(struct nouveau_channel *, uint32_t handle, - struct nouveau_grobj **); -extern void nouveau_grobj_free(struct nouveau_grobj **); - - -struct nouveau_notifier_priv { - struct nouveau_notifier base; - - struct drm_nouveau_notifierobj_alloc drm; - volatile void *map; -}; -#define nouveau_notifier(n) ((struct nouveau_notifier_priv *)(n)) - -extern int -nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, int count, - struct nouveau_notifier **); - -extern void -nouveau_notifier_free(struct nouveau_notifier **); - -extern void -nouveau_notifier_reset(struct nouveau_notifier *, int id); - -extern uint32_t -nouveau_notifier_status(struct nouveau_notifier *, int id); - -extern uint32_t -nouveau_notifier_return_val(struct nouveau_notifier *, int id); - -extern int -nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, - int timeout); - -struct nouveau_bo_priv { - struct nouveau_bo base; - - struct nouveau_pushbuf_bo *pending; - struct nouveau_fence *fence; - struct nouveau_fence *wr_fence; - - struct drm_nouveau_mem_alloc drm; - void *map; - - void *sysmem; - int user; - - int refcount; - - uint64_t offset; - uint64_t flags; - int tiled; -}; -#define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) - -extern int -nouveau_bo_init(struct nouveau_device *); - -extern void -nouveau_bo_takedown(struct nouveau_device *); - -extern int -nouveau_bo_new(struct nouveau_device *, uint32_t flags, int align, int size, - struct nouveau_bo **); - -extern int -nouveau_bo_user(struct nouveau_device *, void *ptr, int size, - struct nouveau_bo **); - -extern int -nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); - -extern int -nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags); - -extern void -nouveau_bo_del(struct nouveau_bo **); - -extern int -nouveau_bo_map(struct nouveau_bo *, uint32_t flags); - -extern void -nouveau_bo_unmap(struct nouveau_bo *); - -extern int -nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, - uint32_t flags); - -extern int -nouveau_resource_init(struct nouveau_resource **heap, unsigned start, - unsigned size); - -extern int -nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, - struct nouveau_resource **); - -extern void -nouveau_resource_free(struct nouveau_resource **); - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_fence.c b/src/gallium/winsys/dri/nouveau/nouveau_fence.c deleted file mode 100644 index e7b0b4ff079..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_fence.c +++ /dev/null @@ -1,214 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" -#include "nouveau_local.h" - -static void -nouveau_fence_del_unsignalled(struct nouveau_fence *fence) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); - struct nouveau_fence *le; - - if (nvchan->fence_head == fence) { - nvchan->fence_head = nouveau_fence(fence)->next; - if (nvchan->fence_head == NULL) - nvchan->fence_tail = NULL; - return; - } - - le = nvchan->fence_head; - while (le && nouveau_fence(le)->next != fence) - le = nouveau_fence(le)->next; - assert(le && nouveau_fence(le)->next == fence); - nouveau_fence(le)->next = nouveau_fence(fence)->next; - if (nvchan->fence_tail == fence) - nvchan->fence_tail = le; -} - -static void -nouveau_fence_del(struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!fence || !*fence) - return; - nvfence = nouveau_fence(*fence); - *fence = NULL; - - if (--nvfence->refcount) - return; - - if (nvfence->emitted && !nvfence->signalled) { - if (nvfence->signal_cb) { - nvfence->refcount++; - nouveau_fence_wait((void *)&nvfence); - return; - } - - nouveau_fence_del_unsignalled(&nvfence->base); - } - free(nvfence); -} - -int -nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!chan || !fence || *fence) - return -EINVAL; - - nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); - if (!nvfence) - return -ENOMEM; - nvfence->base.channel = chan; - nvfence->refcount = 1; - - *fence = &nvfence->base; - return 0; -} - -int -nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!fence) - return -EINVAL; - - if (*fence) { - nouveau_fence_del(fence); - *fence = NULL; - } - - if (ref) { - nvfence = nouveau_fence(ref); - nvfence->refcount++; - *fence = &nvfence->base; - } - - return 0; -} - -int -nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *), - void *priv) -{ - struct nouveau_fence_priv *nvfence = nouveau_fence(fence); - struct nouveau_fence_cb *cb; - - if (!nvfence || !func) - return -EINVAL; - - cb = malloc(sizeof(struct nouveau_fence_cb)); - if (!cb) - return -ENOMEM; - - cb->func = func; - cb->priv = priv; - cb->next = nvfence->signal_cb; - nvfence->signal_cb = cb; - return 0; -} - -void -nouveau_fence_emit(struct nouveau_fence *fence) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); - struct nouveau_fence_priv *nvfence = nouveau_fence(fence); - - nvfence->emitted = 1; - nvfence->sequence = ++nvchan->fence_sequence; - if (nvfence->sequence == 0xffffffff) - NOUVEAU_ERR("AII wrap unhandled\n"); - - /*XXX: assumes subc 0 is populated */ - RING_SPACE_CH(fence->channel, 2); - OUT_RING_CH (fence->channel, 0x00040050); - OUT_RING_CH (fence->channel, nvfence->sequence); - - if (nvchan->fence_tail) { - nouveau_fence(nvchan->fence_tail)->next = fence; - } else { - nvchan->fence_head = fence; - } - nvchan->fence_tail = fence; -} - -void -nouveau_fence_flush(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - uint32_t sequence = *nvchan->ref_cnt; - - while (nvchan->fence_head) { - struct nouveau_fence_priv *nvfence; - - nvfence = nouveau_fence(nvchan->fence_head); - if (nvfence->sequence > sequence) - break; - nouveau_fence_del_unsignalled(&nvfence->base); - nvfence->signalled = 1; - - if (nvfence->signal_cb) { - struct nouveau_fence *fence = NULL; - - nouveau_fence_ref(&nvfence->base, &fence); - - while (nvfence->signal_cb) { - struct nouveau_fence_cb *cb; - - cb = nvfence->signal_cb; - nvfence->signal_cb = cb->next; - cb->func(cb->priv); - free(cb); - } - - nouveau_fence_ref(NULL, &fence); - } - } -} - -int -nouveau_fence_wait(struct nouveau_fence **fence) -{ - struct nouveau_fence_priv *nvfence; - - if (!fence || !*fence) - return -EINVAL; - nvfence = nouveau_fence(*fence); - - if (nvfence->emitted) { - while (!nvfence->signalled) - nouveau_fence_flush(nvfence->base.channel); - } - nouveau_fence_ref(NULL, fence); - - return 0; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_grobj.c b/src/gallium/winsys/dri/nouveau/nouveau_grobj.c deleted file mode 100644 index 51523897d58..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_grobj.c +++ /dev/null @@ -1,107 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include - -#include "nouveau_drmif.h" - -int -nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle, - int class, struct nouveau_grobj **grobj) -{ - struct nouveau_device_priv *nvdev = nouveau_device(chan->device); - struct nouveau_grobj_priv *nvgrobj; - struct drm_nouveau_grobj_alloc g; - int ret; - - if (!nvdev || !grobj || *grobj) - return -EINVAL; - - nvgrobj = calloc(1, sizeof(*nvgrobj)); - if (!nvgrobj) - return -ENOMEM; - nvgrobj->base.channel = chan; - nvgrobj->base.handle = handle; - nvgrobj->base.grclass = class; - - g.channel = chan->id; - g.handle = handle; - g.class = class; - ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC, - &g, sizeof(g)); - if (ret) { - nouveau_grobj_free((void *)&nvgrobj); - return ret; - } - - *grobj = &nvgrobj->base; - return 0; -} - -int -nouveau_grobj_ref(struct nouveau_channel *chan, uint32_t handle, - struct nouveau_grobj **grobj) -{ - struct nouveau_grobj_priv *nvgrobj; - - if (!chan || !grobj || *grobj) - return -EINVAL; - - nvgrobj = calloc(1, sizeof(struct nouveau_grobj_priv)); - if (!nvgrobj) - return -ENOMEM; - nvgrobj->base.channel = chan; - nvgrobj->base.handle = handle; - nvgrobj->base.grclass = 0; - - *grobj = &nvgrobj->base; - return 0; -} - -void -nouveau_grobj_free(struct nouveau_grobj **grobj) -{ - struct nouveau_device_priv *nvdev; - struct nouveau_channel_priv *chan; - struct nouveau_grobj_priv *nvgrobj; - - if (!grobj || !*grobj) - return; - nvgrobj = nouveau_grobj(*grobj); - *grobj = NULL; - - - chan = nouveau_channel(nvgrobj->base.channel); - nvdev = nouveau_device(chan->base.device); - - if (nvgrobj->base.grclass) { - struct drm_nouveau_gpuobj_free f; - - f.channel = chan->drm.channel; - f.handle = nvgrobj->base.handle; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, - &f, sizeof(f)); - } - free(nvgrobj); -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_local.h b/src/gallium/winsys/dri/nouveau/nouveau_local.h deleted file mode 100644 index e878a408037..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_local.h +++ /dev/null @@ -1,117 +0,0 @@ -#ifndef __NOUVEAU_LOCAL_H__ -#define __NOUVEAU_LOCAL_H__ - -#include "pipe/p_compiler.h" -#include "nouveau_winsys_pipe.h" -#include - -struct pipe_buffer; - -/* Debug output */ -#define NOUVEAU_MSG(fmt, args...) do { \ - fprintf(stdout, "nouveau: "fmt, ##args); \ - fflush(stdout); \ -} while(0) - -#define NOUVEAU_ERR(fmt, args...) do { \ - fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); \ - fflush(stderr); \ -} while(0) - -#define NOUVEAU_TIME_MSEC() 0 - -/* User FIFO control */ -//#define NOUVEAU_DMA_TRACE -//#define NOUVEAU_DMA_DEBUG -//#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF -#define NOUVEAU_DMA_BARRIER -#define NOUVEAU_DMA_TIMEOUT 2000 - -/* Push buffer access macros */ -static INLINE void -OUT_RING(struct nouveau_channel *chan, unsigned data) -{ - *(chan->pushbuf->cur++) = (data); -} - -static INLINE void -OUT_RINGp(struct nouveau_channel *chan, uint32_t *data, unsigned size) -{ - memcpy(chan->pushbuf->cur, data, size * 4); - chan->pushbuf->cur += size; -} - -static INLINE void -OUT_RINGf(struct nouveau_channel *chan, float f) -{ - union { uint32_t i; float f; } c; - c.f = f; - OUT_RING(chan, c.i); -} - -static INLINE void -BEGIN_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, - unsigned mthd, unsigned size) -{ - if (chan->pushbuf->remaining < (size + 1)) - nouveau_pushbuf_flush(chan, (size + 1)); - OUT_RING(chan, (gr->subc << 13) | (size << 18) | mthd); - chan->pushbuf->remaining -= (size + 1); -} - -static INLINE void -FIRE_RING(struct nouveau_channel *chan) -{ - nouveau_pushbuf_flush(chan, 0); -} - -static INLINE void -BIND_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, unsigned subc) -{ - gr->subc = subc; - BEGIN_RING(chan, gr, 0x0000, 1); - OUT_RING (chan, gr->handle); -} - -static INLINE void -OUT_RELOC(struct nouveau_channel *chan, struct nouveau_bo *bo, - unsigned data, unsigned flags, unsigned vor, unsigned tor) -{ - nouveau_pushbuf_emit_reloc(chan, chan->pushbuf->cur++, bo, - data, flags, vor, tor); -} - -/* Raw data + flags depending on FB/TT buffer */ -static INLINE void -OUT_RELOCd(struct nouveau_channel *chan, struct nouveau_bo *bo, - unsigned data, unsigned flags, unsigned vor, unsigned tor) -{ - OUT_RELOC(chan, bo, data, flags | NOUVEAU_BO_OR, vor, tor); -} - -/* FB/TT object handle */ -static INLINE void -OUT_RELOCo(struct nouveau_channel *chan, struct nouveau_bo *bo, - unsigned flags) -{ - OUT_RELOC(chan, bo, 0, flags | NOUVEAU_BO_OR, - chan->vram->handle, chan->gart->handle); -} - -/* Low 32-bits of offset */ -static INLINE void -OUT_RELOCl(struct nouveau_channel *chan, struct nouveau_bo *bo, - unsigned delta, unsigned flags) -{ - OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_LOW, 0, 0); -} - -/* High 32-bits of offset */ -static INLINE void -OUT_RELOCh(struct nouveau_channel *chan, struct nouveau_bo *bo, - unsigned delta, unsigned flags) -{ - OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_HIGH, 0, 0); -} - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_lock.c b/src/gallium/winsys/dri/nouveau/nouveau_lock.c deleted file mode 100644 index 9adb9ac8547..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_lock.c +++ /dev/null @@ -1,94 +0,0 @@ -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "main/glheader.h" -#include "glapi/glthread.h" -#include - -#include "nouveau_context.h" -#include "nouveau_screen.h" - -_glthread_DECLARE_STATIC_MUTEX( lockMutex ); - -static void -nouveau_contended_lock(struct nouveau_context *nv, GLuint flags) -{ - __DRIdrawablePrivate *dPriv = nv->dri_drawable; - __DRIscreenPrivate *sPriv = nv->dri_screen; - struct nouveau_screen *nv_screen = nv->nv_screen; - struct nouveau_device *dev = nv_screen->device; - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - drmGetLock(nvdev->fd, nvdev->ctx, flags); - - /* If the window moved, may need to set a new cliprect now. - * - * NOTE: This releases and regains the hw lock, so all state - * checking must be done *after* this call: - */ - if (dPriv) - DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv); -} - -/* Lock the hardware and validate our state. - */ -void -LOCK_HARDWARE(struct nouveau_context *nv) -{ - struct nouveau_screen *nv_screen = nv->nv_screen; - struct nouveau_device *dev = nv_screen->device; - struct nouveau_device_priv *nvdev = nouveau_device(dev); - char __ret=0; - - _glthread_LOCK_MUTEX(lockMutex); - assert(!nv->locked); - - DRM_CAS(nvdev->lock, nvdev->ctx, - (DRM_LOCK_HELD | nvdev->ctx), __ret); - - if (__ret) - nouveau_contended_lock(nv, 0); - nv->locked = GL_TRUE; -} - - - /* Unlock the hardware using the global current context - */ -void -UNLOCK_HARDWARE(struct nouveau_context *nv) -{ - struct nouveau_screen *nv_screen = nv->nv_screen; - struct nouveau_device *dev = nv_screen->device; - struct nouveau_device_priv *nvdev = nouveau_device(dev); - - assert(nv->locked); - nv->locked = GL_FALSE; - - DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx); - - _glthread_UNLOCK_MUTEX(lockMutex); -} diff --git a/src/gallium/winsys/dri/nouveau/nouveau_notifier.c b/src/gallium/winsys/dri/nouveau/nouveau_notifier.c deleted file mode 100644 index 01e8f38440e..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_notifier.c +++ /dev/null @@ -1,137 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_local.h" - -#define NOTIFIER(__v) \ - struct nouveau_notifier_priv *nvnotify = nouveau_notifier(notifier); \ - volatile uint32_t *__v = (void*)nvnotify->map + (id * 32) - -int -nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, - int count, struct nouveau_notifier **notifier) -{ - struct nouveau_notifier_priv *nvnotify; - int ret; - - if (!chan || !notifier || *notifier) - return -EINVAL; - - nvnotify = calloc(1, sizeof(struct nouveau_notifier_priv)); - if (!nvnotify) - return -ENOMEM; - nvnotify->base.channel = chan; - nvnotify->base.handle = handle; - - nvnotify->drm.channel = chan->id; - nvnotify->drm.handle = handle; - nvnotify->drm.count = count; - if ((ret = drmCommandWriteRead(nouveau_device(chan->device)->fd, - DRM_NOUVEAU_NOTIFIEROBJ_ALLOC, - &nvnotify->drm, - sizeof(nvnotify->drm)))) { - nouveau_notifier_free((void *)&nvnotify); - return ret; - } - - nvnotify->map = (void *)nouveau_channel(chan)->notifier_block + - nvnotify->drm.offset; - *notifier = &nvnotify->base; - return 0; -} - -void -nouveau_notifier_free(struct nouveau_notifier **notifier) -{ - - struct nouveau_notifier_priv *nvnotify; - struct nouveau_channel_priv *nvchan; - struct nouveau_device_priv *nvdev; - struct drm_nouveau_gpuobj_free f; - - if (!notifier || !*notifier) - return; - nvnotify = nouveau_notifier(*notifier); - *notifier = NULL; - - nvchan = nouveau_channel(nvnotify->base.channel); - nvdev = nouveau_device(nvchan->base.device); - - f.channel = nvchan->drm.channel; - f.handle = nvnotify->base.handle; - drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); - free(nvnotify); -} - -void -nouveau_notifier_reset(struct nouveau_notifier *notifier, int id) -{ - NOTIFIER(n); - - n[NV_NOTIFY_TIME_0 /4] = 0x00000000; - n[NV_NOTIFY_TIME_1 /4] = 0x00000000; - n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000; - n[NV_NOTIFY_STATE /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS << - NV_NOTIFY_STATE_STATUS_SHIFT); -} - -uint32_t -nouveau_notifier_status(struct nouveau_notifier *notifier, int id) -{ - NOTIFIER(n); - - return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; -} - -uint32_t -nouveau_notifier_return_val(struct nouveau_notifier *notifier, int id) -{ - NOTIFIER(n); - - return n[NV_NOTIFY_RETURN_VALUE/4]; -} - -int -nouveau_notifier_wait_status(struct nouveau_notifier *notifier, int id, - int status, int timeout) -{ - NOTIFIER(n); - uint32_t time = 0, t_start = NOUVEAU_TIME_MSEC(); - - while (time <= timeout) { - uint32_t v; - - v = n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; - if (v == status) - return 0; - - if (timeout) - time = NOUVEAU_TIME_MSEC() - t_start; - } - - return -EBUSY; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c deleted file mode 100644 index 815046ba85f..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_pushbuf.c +++ /dev/null @@ -1,271 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_dma.h" - -#define PB_BUFMGR_DWORDS (4096 / 2) -#define PB_MIN_USER_DWORDS 2048 - -static int -nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; - - assert((min + 1) <= nvchan->dma->max); - - /* Wait for enough space in push buffer */ - min = min < PB_MIN_USER_DWORDS ? PB_MIN_USER_DWORDS : min; - min += 1; /* a bit extra for the NOP */ - if (nvchan->dma->free < min) - WAIT_RING_CH(chan, min); - - /* Insert NOP, may turn into a jump later */ - RING_SPACE_CH(chan, 1); - nvpb->nop_jump = nvchan->dma->cur; - OUT_RING_CH(chan, 0); - - /* Any remaining space is available to the user */ - nvpb->start = nvchan->dma->cur; - nvpb->size = nvchan->dma->free; - nvpb->base.channel = chan; - nvpb->base.remaining = nvpb->size; - nvpb->base.cur = &nvchan->pushbuf[nvpb->start]; - - /* Create a new fence object for this "frame" */ - nouveau_fence_ref(NULL, &nvpb->fence); - nouveau_fence_new(chan, &nvpb->fence); - - return 0; -} - -int -nouveau_pushbuf_init(struct nouveau_channel *chan) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_dma_priv *m = &nvchan->dma_master; - struct nouveau_dma_priv *b = &nvchan->dma_bufmgr; - int i; - - if (!nvchan) - return -EINVAL; - - /* Reassign last bit of push buffer for a "separate" bufmgr - * ring buffer - */ - m->max -= PB_BUFMGR_DWORDS; - m->free -= PB_BUFMGR_DWORDS; - - b->base = m->base + ((m->max + 2) << 2); - b->max = PB_BUFMGR_DWORDS - 2; - b->cur = b->put = 0; - b->free = b->max - b->cur; - - /* Some NOPs just to be safe - *XXX: RING_SKIPS - */ - nvchan->dma = b; - RING_SPACE_CH(chan, 8); - for (i = 0; i < 8; i++) - OUT_RING_CH(chan, 0); - nvchan->dma = m; - - nouveau_pushbuf_space(chan, 0); - chan->pushbuf = &nvchan->pb.base; - - nvchan->pb.buffers = calloc(NOUVEAU_PUSHBUF_MAX_BUFFERS, - sizeof(struct nouveau_pushbuf_bo)); - nvchan->pb.relocs = calloc(NOUVEAU_PUSHBUF_MAX_RELOCS, - sizeof(struct nouveau_pushbuf_reloc)); - return 0; -} - -static uint32_t -nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo, - struct nouveau_pushbuf_reloc *r) -{ - uint32_t push; - - if (r->flags & NOUVEAU_BO_LOW) { - push = bo->offset + r->data; - } else - if (r->flags & NOUVEAU_BO_HIGH) { - push = (bo->offset + r->data) >> 32; - } else { - push = r->data; - } - - if (r->flags & NOUVEAU_BO_OR) { - if (bo->flags & NOUVEAU_BO_VRAM) - push |= r->vor; - else - push |= r->tor; - } - - return push; -} - -/* This would be our TTM "superioctl" */ -int -nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) -{ - struct nouveau_channel_priv *nvchan = nouveau_channel(chan); - struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; - int ret, i; - - if (nvpb->base.remaining == nvpb->size) - return 0; - - nouveau_fence_flush(chan); - - nvpb->size -= nvpb->base.remaining; - nvchan->dma->cur += nvpb->size; - nvchan->dma->free -= nvpb->size; - assert(nvchan->dma->cur <= nvchan->dma->max); - - nvchan->dma = &nvchan->dma_bufmgr; - nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 | - (nvchan->dma->base + (nvchan->dma->cur << 2)); - - /* Validate buffers + apply relocations */ - nvchan->user_charge = 0; - for (i = 0; i < nvpb->nr_relocs; i++) { - struct nouveau_pushbuf_reloc *r = &nvpb->relocs[i]; - struct nouveau_pushbuf_bo *pbbo = r->pbbo; - struct nouveau_bo *bo = pbbo->bo; - - /* Validated, mem matches presumed, no relocation necessary */ - if (pbbo->handled & 2) { - if (!(pbbo->handled & 1)) - assert(0); - continue; - } - - /* Not yet validated, do it now */ - if (!(pbbo->handled & 1)) { - ret = nouveau_bo_validate(chan, bo, pbbo->flags); - if (ret) { - assert(0); - return ret; - } - pbbo->handled |= 1; - - if (bo->offset == nouveau_bo(bo)->offset && - bo->flags == nouveau_bo(bo)->flags) { - pbbo->handled |= 2; - continue; - } - bo->offset = nouveau_bo(bo)->offset; - bo->flags = nouveau_bo(bo)->flags; - } - - /* Apply the relocation */ - *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); - } - nvpb->nr_relocs = 0; - - /* Dereference all buffers on validate list */ - for (i = 0; i < nvpb->nr_buffers; i++) { - struct nouveau_pushbuf_bo *pbbo = &nvpb->buffers[i]; - - nouveau_bo(pbbo->bo)->pending = NULL; - nouveau_bo_del(&pbbo->bo); - } - nvpb->nr_buffers = 0; - - /* Switch back to user's ring */ - RING_SPACE_CH(chan, 1); - OUT_RING_CH(chan, 0x20000000 | ((nvpb->start << 2) + - nvchan->dma_master.base)); - nvchan->dma = &nvchan->dma_master; - - /* Fence + kickoff */ - nouveau_fence_emit(nvpb->fence); - FIRE_RING_CH(chan); - - /* Allocate space for next push buffer */ - ret = nouveau_pushbuf_space(chan, min); - assert(!ret); - - return 0; -} - -static struct nouveau_pushbuf_bo * -nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) -{ - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); - struct nouveau_bo_priv *nvbo = nouveau_bo(bo); - struct nouveau_pushbuf_bo *pbbo; - - if (nvbo->pending) - return nvbo->pending; - - if (nvpb->nr_buffers >= NOUVEAU_PUSHBUF_MAX_BUFFERS) - return NULL; - pbbo = nvpb->buffers + nvpb->nr_buffers++; - nvbo->pending = pbbo; - - nouveau_bo_ref(bo->device, bo->handle, &pbbo->bo); - pbbo->channel = chan; - pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; - pbbo->handled = 0; - return pbbo; -} - -int -nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, - struct nouveau_bo *bo, uint32_t data, uint32_t flags, - uint32_t vor, uint32_t tor) -{ - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); - struct nouveau_pushbuf_bo *pbbo; - struct nouveau_pushbuf_reloc *r; - - if (nvpb->nr_relocs >= NOUVEAU_PUSHBUF_MAX_RELOCS) - return -ENOMEM; - - pbbo = nouveau_pushbuf_emit_buffer(chan, bo); - if (!pbbo) - return -ENOMEM; - pbbo->flags |= (flags & NOUVEAU_BO_RDWR); - pbbo->flags &= (flags | NOUVEAU_BO_RDWR); - - r = nvpb->relocs + nvpb->nr_relocs++; - r->pbbo = pbbo; - r->ptr = ptr; - r->flags = flags; - r->data = data; - r->vor = vor; - r->tor = tor; - - if (flags & NOUVEAU_BO_DUMMY) - *(uint32_t *)ptr = 0; - else - *(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r); - return 0; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_resource.c b/src/gallium/winsys/dri/nouveau/nouveau_resource.c deleted file mode 100644 index 3bbcb5c45e0..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_resource.c +++ /dev/null @@ -1,116 +0,0 @@ -/* - * Copyright 2007 Nouveau Project - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF - * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include -#include - -#include "nouveau_drmif.h" -#include "nouveau_local.h" - -int -nouveau_resource_init(struct nouveau_resource **heap, - unsigned start, unsigned size) -{ - struct nouveau_resource *r; - - r = calloc(1, sizeof(struct nouveau_resource)); - if (!r) - return 1; - - r->start = start; - r->size = size; - *heap = r; - return 0; -} - -int -nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, - struct nouveau_resource **res) -{ - struct nouveau_resource *r; - - if (!heap || !size || !res || *res) - return 1; - - while (heap) { - if (!heap->in_use && heap->size >= size) { - r = calloc(1, sizeof(struct nouveau_resource)); - if (!r) - return 1; - - r->start = (heap->start + heap->size) - size; - r->size = size; - r->in_use = 1; - r->priv = priv; - - heap->size -= size; - - r->next = heap->next; - if (heap->next) - heap->next->prev = r; - r->prev = heap; - heap->next = r; - - *res = r; - return 0; - } - - heap = heap->next; - } - - return 1; -} - -void -nouveau_resource_free(struct nouveau_resource **res) -{ - struct nouveau_resource *r; - - if (!res || !*res) - return; - r = *res; - *res = NULL; - - r->in_use = 0; - - if (r->next && !r->next->in_use) { - struct nouveau_resource *new = r->next; - - new->prev = r->prev; - if (r->prev) - r->prev->next = new; - new->size += r->size; - new->start = r->start; - - free(r); - r = new; - } - - if (r->prev && !r->prev->in_use) { - r->prev->next = r->next; - if (r->next) - r->next->prev = r->prev; - r->prev->size += r->size; - free(r); - } - -} diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.c b/src/gallium/winsys/dri/nouveau/nouveau_screen.c deleted file mode 100644 index df1fe7e69b4..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.c +++ /dev/null @@ -1,310 +0,0 @@ -#include "utils.h" -#include "vblank.h" -#include "xmlpool.h" - -#include "pipe/p_context.h" -#include "state_tracker/st_public.h" -#include "state_tracker/st_cb_fbo.h" - -#include "nouveau_context.h" -#include "nouveau_drm.h" -#include "nouveau_dri.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_swapbuffers.h" - -#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11 -#error nouveau_drm.h version does not match expected version -#endif - -/* Extension stuff, enabling of extensions handled by Gallium's GL state - * tracker. But, we still need to define the entry points we want. - */ -#define need_GL_ARB_fragment_program -#define need_GL_ARB_multisample -#define need_GL_ARB_occlusion_query -#define need_GL_ARB_point_parameters -#define need_GL_ARB_shader_objects -#define need_GL_ARB_texture_compression -#define need_GL_ARB_vertex_program -#define need_GL_ARB_vertex_shader -#define need_GL_ARB_vertex_buffer_object -#define need_GL_EXT_compiled_vertex_array -#define need_GL_EXT_fog_coord -#define need_GL_EXT_secondary_color -#define need_GL_EXT_framebuffer_object -#define need_GL_VERSION_2_0 -#define need_GL_VERSION_2_1 -#include "extension_helper.h" - -const struct dri_extension card_extensions[] = -{ - { "GL_ARB_multisample", GL_ARB_multisample_functions }, - { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, - { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, - { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, - { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, - { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, - { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, - { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, - { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, - { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, - { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, - { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, - { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, - { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, - { NULL, 0 } -}; - -PUBLIC const char __driConfigOptions[] = -DRI_CONF_BEGIN -DRI_CONF_END; -static const GLuint __driNConfigOptions = 0; - -extern const struct dri_extension common_extensions[]; -extern const struct dri_extension nv40_extensions[]; - -static GLboolean -nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) -{ - struct nouveau_dri *nv_dri = driScrnPriv->pDevPriv; - struct nouveau_screen *nv_screen; - int ret; - - if (driScrnPriv->devPrivSize != sizeof(struct nouveau_dri)) { - NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); - return GL_FALSE; - } - - nv_screen = CALLOC_STRUCT(nouveau_screen); - if (!nv_screen) - return GL_FALSE; - nv_screen->driScrnPriv = driScrnPriv; - driScrnPriv->private = (void *)nv_screen; - - driParseOptionInfo(&nv_screen->option_cache, - __driConfigOptions, __driNConfigOptions); - - if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, - driScrnPriv->fd, 0))) { - NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); - return GL_FALSE; - } - - nv_screen->front_offset = nv_dri->front_offset; - nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); - nv_screen->front_cpp = nv_dri->bpp / 8; - nv_screen->front_height = nv_dri->height; - - return GL_TRUE; -} - -static void -nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) -{ - struct nouveau_screen *nv_screen = driScrnPriv->private; - - driScrnPriv->private = NULL; - FREE(nv_screen); -} - -static GLboolean -nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, - __DRIdrawablePrivate * driDrawPriv, - const __GLcontextModes *glVis, GLboolean pixmapBuffer) -{ - struct nouveau_framebuffer *nvfb; - enum pipe_format colour, depth, stencil; - - if (pixmapBuffer) - return GL_FALSE; - - nvfb = CALLOC_STRUCT(nouveau_framebuffer); - if (!nvfb) - return GL_FALSE; - - if (glVis->redBits == 5) - colour = PIPE_FORMAT_R5G6B5_UNORM; - else - colour = PIPE_FORMAT_A8R8G8B8_UNORM; - - if (glVis->depthBits == 16) - depth = PIPE_FORMAT_Z16_UNORM; - else if (glVis->depthBits == 24) - depth = PIPE_FORMAT_Z24S8_UNORM; - else - depth = PIPE_FORMAT_NONE; - - if (glVis->stencilBits == 8) - stencil = PIPE_FORMAT_Z24S8_UNORM; - else - stencil = PIPE_FORMAT_NONE; - - nvfb->stfb = st_create_framebuffer(glVis, colour, depth, stencil, - driDrawPriv->w, driDrawPriv->h, - (void*)nvfb); - if (!nvfb->stfb) { - free(nvfb); - return GL_FALSE; - } - - driDrawPriv->driverPrivate = (void *)nvfb; - return GL_TRUE; -} - -static void -nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) -{ - struct nouveau_framebuffer *nvfb; - - nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate; - st_unreference_framebuffer(&nvfb->stfb); - free(nvfb); -} - -static struct __DriverAPIRec -nouveau_api = { - .InitDriver = nouveau_screen_create, - .DestroyScreen = nouveau_screen_destroy, - .CreateContext = nouveau_context_create, - .DestroyContext = nouveau_context_destroy, - .CreateBuffer = nouveau_create_buffer, - .DestroyBuffer = nouveau_destroy_buffer, - .SwapBuffers = nouveau_swap_buffers, - .MakeCurrent = nouveau_context_bind, - .UnbindContext = nouveau_context_unbind, - .GetSwapInfo = NULL, - .GetMSC = NULL, - .WaitForMSC = NULL, - .WaitForSBC = NULL, - .SwapBuffersMSC = NULL, - .CopySubBuffer = nouveau_copy_sub_buffer, - .setTexOffset = NULL -}; - -static __GLcontextModes * -nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, - unsigned stencil_bits, GLboolean have_back_buffer) -{ - __GLcontextModes * modes; - __GLcontextModes * m; - unsigned num_modes; - unsigned depth_buffer_factor; - unsigned back_buffer_factor; - int i; - - static const struct { - GLenum format; - GLenum type; - } fb_format_array[] = { - { GL_RGB , GL_UNSIGNED_SHORT_5_6_5 }, - { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV }, - { GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV }, - }; - - /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't - * support pageflipping at all. - */ - static const GLenum back_buffer_modes[] = { - GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML - }; - - uint8_t depth_bits_array[4] = { 0, 16, 24, 24 }; - uint8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; - uint8_t msaa_samples_array[1] = { 0 }; - - depth_buffer_factor = 4; - back_buffer_factor = (have_back_buffer) ? 3 : 1; - - num_modes = ((pixel_bits==16) ? 1 : 2) * - depth_buffer_factor * back_buffer_factor * 4; - modes = (*dri_interface->createContextModes)(num_modes, - sizeof(__GLcontextModes)); - m = modes; - - for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) { - if (!driFillInModes(&m, fb_format_array[i].format, - fb_format_array[i].type, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - msaa_samples_array, 1, - GLX_TRUE_COLOR)) { - fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", - __func__, __LINE__ ); - return NULL; - } - - if (!driFillInModes(&m, fb_format_array[i].format, - fb_format_array[i].type, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - msaa_samples_array, 1, - GLX_DIRECT_COLOR)) { - fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", - __func__, __LINE__ ); - return NULL; - } - } - - return modes; -} -PUBLIC void * -__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, - __DRIscreen *psc, const __GLcontextModes * modes, - const __DRIversion * ddx_version, - const __DRIversion * dri_version, - const __DRIversion * drm_version, - const __DRIframebuffer * frame_buffer, - void * pSAREA, int fd, int internal_api_version, - const __DRIinterfaceMethods * interface, - __GLcontextModes ** driver_modes) -{ - __DRIscreenPrivate *psp; - static const __DRIversion ddx_expected = - { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; - static const __DRIversion dri_expected = { 4, 0, 0 }; - static const __DRIversion drm_expected = - { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; - struct nouveau_dri *nv_dri = NULL; - - dri_interface = interface; - - if (!driCheckDriDdxDrmVersions2("nouveau", - dri_version, &dri_expected, - ddx_version, &ddx_expected, - drm_version, &drm_expected)) { - return NULL; - } - - if (drm_expected.patch != drm_version->patch) { - fprintf(stderr, "Incompatible DRM patch level.\n" - "Expected: %d\n" "Current : %d\n", - drm_expected.patch, drm_version->patch); - return NULL; - } - - psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, - ddx_version, dri_version, drm_version, - frame_buffer, pSAREA, fd, - internal_api_version, - &nouveau_api); - if (psp == NULL) - return NULL; - nv_dri = psp->pDevPriv; - - *driver_modes = nouveau_fill_in_modes(nv_dri->bpp, - (nv_dri->bpp == 16) ? 16 : 24, - (nv_dri->bpp == 16) ? 0 : 8, - 1); - - driInitExtensions(NULL, card_extensions, GL_FALSE); - - return (void *)psp; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_screen.h b/src/gallium/winsys/dri/nouveau/nouveau_screen.h deleted file mode 100644 index 388d6be9bbc..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_screen.h +++ /dev/null @@ -1,20 +0,0 @@ -#ifndef __NOUVEAU_SCREEN_H__ -#define __NOUVEAU_SCREEN_H__ - -#include "xmlconfig.h" - -struct nouveau_screen { - __DRIscreenPrivate *driScrnPriv; - driOptionCache option_cache; - - struct nouveau_device *device; - - uint32_t front_offset; - uint32_t front_pitch; - uint32_t front_cpp; - uint32_t front_height; - - void *nvc; -}; - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c deleted file mode 100644 index 70e0104e83b..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.c +++ /dev/null @@ -1,86 +0,0 @@ -#include "main/glheader.h" -#include "glapi/glthread.h" -#include - -#include "pipe/p_context.h" -#include "state_tracker/st_public.h" -#include "state_tracker/st_context.h" -#include "state_tracker/st_cb_fbo.h" - -#include "nouveau_context.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_swapbuffers.h" - -void -nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, - const drm_clip_rect_t *rect) -{ - struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; - drm_clip_rect_t *pbox; - int nbox, i; - - LOCK_HARDWARE(nv); - if (!dPriv->numClipRects) { - UNLOCK_HARDWARE(nv); - return; - } - pbox = dPriv->pClipRects; - nbox = dPriv->numClipRects; - - nv->surface_copy_prep(nv, nv->frontbuffer, surf); - for (i = 0; i < nbox; i++, pbox++) { - int sx, sy, dx, dy, w, h; - - sx = pbox->x1 - dPriv->x; - sy = pbox->y1 - dPriv->y; - dx = pbox->x1; - dy = pbox->y1; - w = pbox->x2 - pbox->x1; - h = pbox->y2 - pbox->y1; - - nv->surface_copy(nv, dx, dy, sx, sy, w, h); - } - - FIRE_RING(nv->nvc->channel); - UNLOCK_HARDWARE(nv); - - if (nv->last_stamp != dPriv->lastStamp) { - struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; - st_resize_framebuffer(nvfb->stfb, dPriv->w, dPriv->h); - nv->last_stamp = dPriv->lastStamp; - } -} - -void -nouveau_copy_sub_buffer(__DRIdrawablePrivate *dPriv, int x, int y, int w, int h) -{ - struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; - struct pipe_surface *surf; - - surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); - if (surf) { - drm_clip_rect_t rect; - rect.x1 = x; - rect.y1 = y; - rect.x2 = x + w; - rect.y2 = y + h; - - st_notify_swapbuffers(nvfb->stfb); - nouveau_copy_buffer(dPriv, surf, &rect); - } -} - -void -nouveau_swap_buffers(__DRIdrawablePrivate *dPriv) -{ - struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; - struct pipe_surface *surf; - - surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); - if (surf) { - st_notify_swapbuffers(nvfb->stfb); - nouveau_copy_buffer(dPriv, surf, NULL); - } -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h b/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h deleted file mode 100644 index 825d3da6da5..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_swapbuffers.h +++ /dev/null @@ -1,10 +0,0 @@ -#ifndef __NOUVEAU_SWAPBUFFERS_H__ -#define __NOUVEAU_SWAPBUFFERS_H__ - -extern void nouveau_copy_buffer(__DRIdrawablePrivate *, struct pipe_surface *, - const drm_clip_rect_t *); -extern void nouveau_copy_sub_buffer(__DRIdrawablePrivate *, - int x, int y, int w, int h); -extern void nouveau_swap_buffers(__DRIdrawablePrivate *); - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys.c deleted file mode 100644 index 5eabbc88939..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys.c +++ /dev/null @@ -1,158 +0,0 @@ -#include "pipe/p_util.h" - -#include "nouveau_context.h" -#include "nouveau_screen.h" -#include "nouveau_winsys_pipe.h" - -#include "nouveau/nouveau_winsys.h" - -static int -nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, - struct nouveau_notifier **notify) -{ - struct nouveau_context *nv = nvws->nv; - - return nouveau_notifier_alloc(nv->nvc->channel, nv->nvc->next_handle++, - count, notify); -} - -static int -nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, - struct nouveau_grobj **grobj) -{ - struct nouveau_context *nv = nvws->nv; - struct nouveau_channel *chan = nv->nvc->channel; - int ret; - - ret = nouveau_grobj_alloc(chan, nv->nvc->next_handle++, - grclass, grobj); - if (ret) - return ret; - - assert(nv->nvc->next_subchannel < 7); - BIND_RING(chan, *grobj, nv->nvc->next_subchannel++); - return 0; -} - -static int -nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst, - unsigned dx, unsigned dy, struct pipe_surface *src, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct nouveau_context *nv = nvws->nv; - - if (nv->surface_copy_prep(nv, dst, src)) - return 1; - nv->surface_copy(nv, dx, dy, sx, sy, w, h); - nv->surface_copy_done(nv); - - return 0; -} - -static int -nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - if (nvws->nv->surface_fill(nvws->nv, dst, dx, dy, w, h, value)) - return 1; - return 0; -} - -static int -nouveau_pipe_push_reloc(struct nouveau_winsys *nvws, void *ptr, - struct pipe_buffer *buf, uint32_t data, - uint32_t flags, uint32_t vor, uint32_t tor) -{ - return nouveau_pushbuf_emit_reloc(nvws->channel, ptr, - nouveau_buffer(buf)->bo, - data, flags, vor, tor); -} - -static int -nouveau_pipe_push_flush(struct nouveau_winsys *nvws, unsigned size, - struct pipe_fence_handle **fence) -{ - if (fence) { - struct nouveau_pushbuf *pb = nvws->channel->pushbuf; - struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(pb); - struct nouveau_fence *ref = NULL; - - nouveau_fence_ref(nvpb->fence, &ref); - *fence = (struct pipe_fence_handle *)ref; - } - - return nouveau_pushbuf_flush(nvws->channel, size); -} - -struct pipe_context * -nouveau_pipe_create(struct nouveau_context *nv) -{ - struct nouveau_channel_context *nvc = nv->nvc; - struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); - struct pipe_screen *(*hws_create)(struct pipe_winsys *, - struct nouveau_winsys *); - struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned); - struct pipe_winsys *ws; - unsigned chipset = nv->nv_screen->device->chipset; - - if (!nvws) - return NULL; - - switch (chipset & 0xf0) { - case 0x10: - case 0x20: - hws_create = nv10_screen_create; - hw_create = nv10_create; - break; - case 0x30: - hws_create = nv30_screen_create; - hw_create = nv30_create; - break; - case 0x40: - case 0x60: - hws_create = nv40_screen_create; - hw_create = nv40_create; - break; - case 0x50: - case 0x80: - case 0x90: - hws_create = nv50_screen_create; - hw_create = nv50_create; - break; - default: - NOUVEAU_ERR("Unknown chipset NV%02x\n", chipset); - return NULL; - } - - nvws->nv = nv; - nvws->channel = nv->nvc->channel; - - nvws->res_init = nouveau_resource_init; - nvws->res_alloc = nouveau_resource_alloc; - nvws->res_free = nouveau_resource_free; - - nvws->push_reloc = nouveau_pipe_push_reloc; - nvws->push_flush = nouveau_pipe_push_flush; - - nvws->grobj_alloc = nouveau_pipe_grobj_alloc; - nvws->grobj_free = nouveau_grobj_free; - - nvws->notifier_alloc = nouveau_pipe_notifier_alloc; - nvws->notifier_free = nouveau_notifier_free; - nvws->notifier_reset = nouveau_notifier_reset; - nvws->notifier_status = nouveau_notifier_status; - nvws->notifier_retval = nouveau_notifier_return_val; - nvws->notifier_wait = nouveau_notifier_wait_status; - - nvws->surface_copy = nouveau_pipe_surface_copy; - nvws->surface_fill = nouveau_pipe_surface_fill; - - ws = nouveau_create_pipe_winsys(nv); - - if (!nvc->pscreen) - nvc->pscreen = hws_create(ws, nvws); - nvc->pctx[nv->pctx_id] = hw_create(nvc->pscreen, nv->pctx_id); - return nvc->pctx[nv->pctx_id]; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c deleted file mode 100644 index 8a2870a2ff2..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.c +++ /dev/null @@ -1,205 +0,0 @@ -#include "pipe/p_winsys.h" -#include "pipe/p_defines.h" -#include "pipe/p_util.h" -#include "pipe/p_inlines.h" - -#include "nouveau_context.h" -#include "nouveau_local.h" -#include "nouveau_screen.h" -#include "nouveau_swapbuffers.h" -#include "nouveau_winsys_pipe.h" - -static void -nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, - void *context_private) -{ - struct nouveau_context *nv = context_private; - __DRIdrawablePrivate *dPriv = nv->dri_drawable; - - nouveau_copy_buffer(dPriv, surf, NULL); -} - -static const char * -nouveau_get_name(struct pipe_winsys *pws) -{ - return "Nouveau/DRI"; -} - -static struct pipe_buffer * -nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, - unsigned usage, unsigned size) -{ - struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; - struct nouveau_context *nv = nvpws->nv; - struct nouveau_device *dev = nv->nv_screen->device; - struct nouveau_pipe_buffer *nvbuf; - uint32_t flags; - - nvbuf = calloc(1, sizeof(*nvbuf)); - if (!nvbuf) - return NULL; - nvbuf->base.refcount = 1; - nvbuf->base.alignment = alignment; - nvbuf->base.usage = usage; - nvbuf->base.size = size; - - flags = NOUVEAU_BO_LOCAL; - - if (usage & PIPE_BUFFER_USAGE_PIXEL) { - if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE) - flags |= NOUVEAU_BO_GART; - flags |= NOUVEAU_BO_VRAM; - - switch (dev->chipset & 0xf0) { - case 0x50: - case 0x80: - case 0x90: - flags |= NOUVEAU_BO_TILED; - if (usage & NOUVEAU_BUFFER_USAGE_ZETA) - flags |= NOUVEAU_BO_ZTILE; - break; - default: - break; - } - } - - if (usage & PIPE_BUFFER_USAGE_VERTEX) { - if (nv->cap.hw_vertex_buffer) - flags |= NOUVEAU_BO_GART; - } - - if (usage & PIPE_BUFFER_USAGE_INDEX) { - if (nv->cap.hw_index_buffer) - flags |= NOUVEAU_BO_GART; - } - - if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { - free(nvbuf); - return NULL; - } - - return &nvbuf->base; -} - -static struct pipe_buffer * -nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) -{ - struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; - struct nouveau_device *dev = nvpws->nv->nv_screen->device; - struct nouveau_pipe_buffer *nvbuf; - - nvbuf = calloc(1, sizeof(*nvbuf)); - if (!nvbuf) - return NULL; - nvbuf->base.refcount = 1; - nvbuf->base.size = bytes; - - if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) { - free(nvbuf); - return NULL; - } - - return &nvbuf->base; -} - -static void -nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) -{ - struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - - nouveau_bo_del(&nvbuf->bo); - free(nvbuf); -} - -static void * -nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf, - unsigned flags) -{ - struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - uint32_t map_flags = 0; - - if (flags & PIPE_BUFFER_USAGE_CPU_READ) - map_flags |= NOUVEAU_BO_RD; - if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) - map_flags |= NOUVEAU_BO_WR; - - if (nouveau_bo_map(nvbuf->bo, map_flags)) - return NULL; - return nvbuf->bo->map; -} - -static void -nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) -{ - struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); - - nouveau_bo_unmap(nvbuf->bo); -} - -static INLINE struct nouveau_fence * -nouveau_pipe_fence(struct pipe_fence_handle *pfence) -{ - return (struct nouveau_fence *)pfence; -} - -static void -nouveau_pipe_fence_reference(struct pipe_winsys *ws, - struct pipe_fence_handle **ptr, - struct pipe_fence_handle *pfence) -{ - nouveau_fence_ref((void *)pfence, (void *)ptr); -} - -static int -nouveau_pipe_fence_signalled(struct pipe_winsys *ws, - struct pipe_fence_handle *pfence, unsigned flag) -{ - struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)ws; - struct nouveau_fence *fence = nouveau_pipe_fence(pfence); - - if (nouveau_fence(fence)->signalled == 0) - nouveau_fence_flush(nvpws->nv->nvc->channel); - - return !nouveau_fence(fence)->signalled; -} - -static int -nouveau_pipe_fence_finish(struct pipe_winsys *ws, - struct pipe_fence_handle *pfence, unsigned flag) -{ - struct nouveau_fence *fence = nouveau_pipe_fence(pfence); - struct nouveau_fence *ref = NULL; - - nouveau_fence_ref(fence, &ref); - return nouveau_fence_wait(&ref); -} - -struct pipe_winsys * -nouveau_create_pipe_winsys(struct nouveau_context *nv) -{ - struct nouveau_pipe_winsys *nvpws; - struct pipe_winsys *pws; - - nvpws = CALLOC_STRUCT(nouveau_pipe_winsys); - if (!nvpws) - return NULL; - nvpws->nv = nv; - pws = &nvpws->pws; - - pws->flush_frontbuffer = nouveau_flush_frontbuffer; - - pws->buffer_create = nouveau_pipe_bo_create; - pws->buffer_destroy = nouveau_pipe_bo_del; - pws->user_buffer_create = nouveau_pipe_bo_user_create; - pws->buffer_map = nouveau_pipe_bo_map; - pws->buffer_unmap = nouveau_pipe_bo_unmap; - - pws->fence_reference = nouveau_pipe_fence_reference; - pws->fence_signalled = nouveau_pipe_fence_signalled; - pws->fence_finish = nouveau_pipe_fence_finish; - - pws->get_name = nouveau_get_name; - - return &nvpws->pws; -} - diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h deleted file mode 100644 index 6a03ac0d773..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_pipe.h +++ /dev/null @@ -1,34 +0,0 @@ -#ifndef NOUVEAU_PIPE_WINSYS_H -#define NOUVEAU_PIPE_WINSYS_H - -#include "pipe/p_context.h" -#include "pipe/p_winsys.h" -#include "nouveau_context.h" - -struct nouveau_pipe_buffer { - struct pipe_buffer base; - struct nouveau_bo *bo; -}; - -static inline struct nouveau_pipe_buffer * -nouveau_buffer(struct pipe_buffer *buf) -{ - return (struct nouveau_pipe_buffer *)buf; -} - -struct nouveau_pipe_winsys { - struct pipe_winsys pws; - - struct nouveau_context *nv; -}; - -extern struct pipe_winsys * -nouveau_create_pipe_winsys(struct nouveau_context *nv); - -struct pipe_context * -nouveau_create_softpipe(struct nouveau_context *nv); - -struct pipe_context * -nouveau_pipe_create(struct nouveau_context *nv); - -#endif diff --git a/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c deleted file mode 100644 index 704f6c77506..00000000000 --- a/src/gallium/winsys/dri/nouveau/nouveau_winsys_softpipe.c +++ /dev/null @@ -1,85 +0,0 @@ -/************************************************************************** - * - * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * - **************************************************************************/ -/* - * Authors: Keith Whitwell - */ - -#include "imports.h" - -#include "pipe/p_defines.h" -#include "pipe/p_format.h" -#include "softpipe/sp_winsys.h" - -#include "nouveau_context.h" -#include "nouveau_winsys_pipe.h" - -struct nouveau_softpipe_winsys { - struct softpipe_winsys sws; - struct nouveau_context *nv; -}; - -/** - * Return list of surface formats supported by this driver. - */ -static boolean -nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) -{ - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return TRUE; - default: - break; - }; - - return FALSE; -} - -struct pipe_context * -nouveau_create_softpipe(struct nouveau_context *nv) -{ - struct nouveau_softpipe_winsys *nvsws; - struct pipe_screen *pscreen; - struct pipe_winsys *ws; - - ws = nouveau_create_pipe_winsys(nv); - if (!ws) - return NULL; - pscreen = softpipe_create_screen(ws); - - nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); - if (!nvsws) - return NULL; - - nvsws->sws.is_format_supported = nouveau_is_format_supported; - nvsws->nv = nv; - - return softpipe_create(pscreen, ws, &nvsws->sws); -} - diff --git a/src/gallium/winsys/dri/nouveau/nv04_surface.c b/src/gallium/winsys/dri/nouveau/nv04_surface.c deleted file mode 100644 index 8fa3d106c8c..00000000000 --- a/src/gallium/winsys/dri/nouveau/nv04_surface.c +++ /dev/null @@ -1,314 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_format.h" - -#include "nouveau_context.h" - -static INLINE int -nv04_surface_format(enum pipe_format format) -{ - switch (format) { - case PIPE_FORMAT_A8_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; - case PIPE_FORMAT_R5G6B5_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; - default: - return -1; - } -} - -static INLINE int -nv04_rect_format(enum pipe_format format) -{ - switch (format) { - case PIPE_FORMAT_A8_UNORM: - return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - case PIPE_FORMAT_R5G6B5_UNORM: - return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; - default: - return -1; - } -} - -static void -nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct nouveau_channel *chan = nv->nvc->channel; - struct pipe_surface *dst = nv->surf_dst; - struct pipe_surface *src = nv->surf_src; - unsigned dst_offset, src_offset; - - dst_offset = dst->offset + (dy * dst->stride) + (dx * dst->block.size); - src_offset = src->offset + (sy * src->stride) + (sx * src->block.size); - - while (h) { - int count = (h > 2047) ? 2047 : h; - - BEGIN_RING(chan, nv->nvc->NvM2MF, - NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); - OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src_offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst_offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); - OUT_RING (chan, src->stride); - OUT_RING (chan, dst->stride); - OUT_RING (chan, w * src->block.size); - OUT_RING (chan, count); - OUT_RING (chan, 0x0101); - OUT_RING (chan, 0); - - h -= count; - src_offset += src->stride * count; - dst_offset += dst->stride * count; - } -} - -static void -nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct nouveau_channel *chan = nv->nvc->channel; - - BEGIN_RING(chan, nv->nvc->NvImageBlit, 0x0300, 3); - OUT_RING (chan, (sy << 16) | sx); - OUT_RING (chan, (dy << 16) | dx); - OUT_RING (chan, ( h << 16) | w); -} - -static int -nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, - struct pipe_surface *src) -{ - struct nouveau_channel *chan = nv->nvc->channel; - int format; - - if (src->format != dst->format) - return 1; - - /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback - * to NV_MEMORY_TO_MEMORY_FORMAT in this case. - */ - if ((src->offset & 63) || (dst->offset & 63)) { - BEGIN_RING(nv->nvc->channel, nv->nvc->NvM2MF, - NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); - OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, - NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - nv->surface_copy = nv04_surface_copy_m2mf; - nv->surf_dst = dst; - nv->surf_src = src; - return 0; - - } - - if ((format = nv04_surface_format(dst->format)) < 0) { - NOUVEAU_ERR("Bad surface format 0x%x\n", dst->format); - return 1; - } - nv->surface_copy = nv04_surface_copy_blit; - - BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, - NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, - NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (chan, format); - OUT_RING (chan, (dst->stride << 16) | src->stride); - OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); - OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - return 0; -} - -static void -nv04_surface_copy_done(struct nouveau_context *nv) -{ - FIRE_RING(nv->nvc->channel); -} - -static int -nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - struct nouveau_channel *chan = nv->nvc->channel; - struct nouveau_grobj *surf2d = nv->nvc->NvCtxSurf2D; - struct nouveau_grobj *rect = nv->nvc->NvGdiRect; - int cs2d_format, gdirect_format; - - if ((cs2d_format = nv04_surface_format(dst->format)) < 0) { - NOUVEAU_ERR("Bad format = %d\n", dst->format); - return 1; - } - - if ((gdirect_format = nv04_rect_format(dst->format)) < 0) { - NOUVEAU_ERR("Bad format = %d\n", dst->format); - return 1; - } - - BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); - OUT_RING (chan, cs2d_format); - OUT_RING (chan, (dst->stride << 16) | dst->stride); - OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, - NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); - OUT_RING (chan, gdirect_format); - BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); - OUT_RING (chan, value); - BEGIN_RING(chan, rect, - NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); - OUT_RING (chan, (dx << 16) | dy); - OUT_RING (chan, ( w << 16) | h); - - FIRE_RING(chan); - return 0; -} - -int -nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) -{ - struct nouveau_channel *chan = nvc->channel; - unsigned chipset = nvc->channel->device->chipset, class; - int ret; - - if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, 0x39, - &nvc->NvM2MF))) { - NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); - return 1; - } - BIND_RING (chan, nvc->NvM2MF, nvc->next_subchannel++); - BEGIN_RING(chan, nvc->NvM2MF, - NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); - OUT_RING (chan, nvc->sync_notifier->handle); - - class = chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : - NV10_CONTEXT_SURFACES_2D; - if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, - &nvc->NvCtxSurf2D))) { - NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); - return 1; - } - BIND_RING (chan, nvc->NvCtxSurf2D, nvc->next_subchannel++); - BEGIN_RING(chan, nvc->NvCtxSurf2D, - NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); - OUT_RING (chan, nvc->channel->vram->handle); - OUT_RING (chan, nvc->channel->vram->handle); - - class = chipset < 0x10 ? NV04_IMAGE_BLIT : NV12_IMAGE_BLIT; - if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, - &nvc->NvImageBlit))) { - NOUVEAU_ERR("Error creating blit object: %d\n", ret); - return 1; - } - BIND_RING (chan, nvc->NvImageBlit, nvc->next_subchannel++); - BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); - OUT_RING (chan, nvc->sync_notifier->handle); - BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); - OUT_RING (chan, nvc->NvCtxSurf2D->handle); - BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); - OUT_RING (chan, NV04_IMAGE_BLIT_OPERATION_SRCCOPY); - - class = NV04_GDI_RECTANGLE_TEXT; - if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, - &nvc->NvGdiRect))) { - NOUVEAU_ERR("Error creating rect object: %d\n", ret); - return 1; - } - BIND_RING (chan, nvc->NvGdiRect, nvc->next_subchannel++); - BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); - OUT_RING (chan, nvc->sync_notifier->handle); - BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); - OUT_RING (chan, nvc->NvCtxSurf2D->handle); - BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); - OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); - BEGIN_RING(chan, nvc->NvGdiRect, - NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); - OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); - - switch (chipset & 0xf0) { - case 0x00: - case 0x10: - class = NV04_SWIZZLED_SURFACE; - break; - case 0x20: - class = NV20_SWIZZLED_SURFACE; - break; - case 0x30: - class = NV30_SWIZZLED_SURFACE; - break; - case 0x40: - case 0x60: - class = NV40_SWIZZLED_SURFACE; - break; - default: - /* Famous last words: this really can't happen.. */ - assert(0); - break; - } - - ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, - &nvc->NvSwzSurf); - if (ret) { - NOUVEAU_ERR("Error creating swizzled surface: %d\n", ret); - return 1; - } - - BIND_RING (chan, nvc->NvSwzSurf, nvc->next_subchannel++); - BEGIN_RING(chan, nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_NOTIFY, 1); - OUT_RING (chan, nvc->sync_notifier->handle); - BEGIN_RING(chan, nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1); - OUT_RING (chan, nvc->channel->vram->handle); - - if (chipset < 0x10) { - class = NV04_SCALED_IMAGE_FROM_MEMORY; - } else - if (chipset < 0x40) { - class = NV10_SCALED_IMAGE_FROM_MEMORY; - } else { - class = NV40_SCALED_IMAGE_FROM_MEMORY; - } - - ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, - &nvc->NvSIFM); - if (ret) { - NOUVEAU_ERR("Error creating scaled image object: %d\n", ret); - return 1; - } - - BIND_RING (chan, nvc->NvSIFM, nvc->next_subchannel++); - - return 0; -} - -int -nouveau_surface_init_nv04(struct nouveau_context *nv) -{ - nv->surface_copy_prep = nv04_surface_copy_prep; - nv->surface_copy = nv04_surface_copy_blit; - nv->surface_copy_done = nv04_surface_copy_done; - nv->surface_fill = nv04_surface_fill; - return 0; -} - diff --git a/src/gallium/winsys/dri/nouveau/nv50_surface.c b/src/gallium/winsys/dri/nouveau/nv50_surface.c deleted file mode 100644 index c8ab7f690f3..00000000000 --- a/src/gallium/winsys/dri/nouveau/nv50_surface.c +++ /dev/null @@ -1,194 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_format.h" - -#include "nouveau_context.h" - -static INLINE int -nv50_format(enum pipe_format format) -{ - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - return NV50_2D_DST_FORMAT_32BPP; - case PIPE_FORMAT_X8R8G8B8_UNORM: - return NV50_2D_DST_FORMAT_24BPP; - case PIPE_FORMAT_R5G6B5_UNORM: - return NV50_2D_DST_FORMAT_16BPP; - case PIPE_FORMAT_A8_UNORM: - return NV50_2D_DST_FORMAT_8BPP; - default: - return -1; - } -} - -static int -nv50_surface_set(struct nouveau_context *nv, struct pipe_surface *surf, int dst) -{ - struct nouveau_channel *chan = nv->nvc->channel; - struct nouveau_grobj *eng2d = nv->nvc->Nv2D; - struct nouveau_bo *bo = nouveau_buffer(surf->buffer)->bo; - int surf_format, mthd = dst ? NV50_2D_DST_FORMAT : NV50_2D_SRC_FORMAT; - int flags = NOUVEAU_BO_VRAM | (dst ? NOUVEAU_BO_WR : NOUVEAU_BO_RD); - - surf_format = nv50_format(surf->format); - if (surf_format < 0) - return 1; - - if (!nouveau_bo(bo)->tiled) { - BEGIN_RING(chan, eng2d, mthd, 2); - OUT_RING (chan, surf_format); - OUT_RING (chan, 1); - BEGIN_RING(chan, eng2d, mthd + 0x14, 5); - OUT_RING (chan, surf->stride); - OUT_RING (chan, surf->width); - OUT_RING (chan, surf->height); - OUT_RELOCh(chan, bo, surf->offset, flags); - OUT_RELOCl(chan, bo, surf->offset, flags); - } else { - BEGIN_RING(chan, eng2d, mthd, 5); - OUT_RING (chan, surf_format); - OUT_RING (chan, 0); - OUT_RING (chan, 0); - OUT_RING (chan, 1); - OUT_RING (chan, 0); - BEGIN_RING(chan, eng2d, mthd + 0x18, 4); - OUT_RING (chan, surf->width); - OUT_RING (chan, surf->height); - OUT_RELOCh(chan, bo, surf->offset, flags); - OUT_RELOCl(chan, bo, surf->offset, flags); - } - -#if 0 - if (dst) { - BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); - OUT_RING (chan, 0); - OUT_RING (chan, 0); - OUT_RING (chan, surf->width); - OUT_RING (chan, surf->height); - } -#endif - - return 0; -} - -static int -nv50_surface_copy_prep(struct nouveau_context *nv, - struct pipe_surface *dst, struct pipe_surface *src) -{ - int ret; - - assert(src->format == dst->format); - - ret = nv50_surface_set(nv, dst, 1); - if (ret) - return ret; - - ret = nv50_surface_set(nv, src, 0); - if (ret) - return ret; - - return 0; -} - -static void -nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, - unsigned sx, unsigned sy, unsigned w, unsigned h) -{ - struct nouveau_channel *chan = nv->nvc->channel; - struct nouveau_grobj *eng2d = nv->nvc->Nv2D; - - BEGIN_RING(chan, eng2d, 0x088c, 1); - OUT_RING (chan, 0); - BEGIN_RING(chan, eng2d, NV50_2D_BLIT_DST_X, 4); - OUT_RING (chan, dx); - OUT_RING (chan, dy); - OUT_RING (chan, w); - OUT_RING (chan, h); - BEGIN_RING(chan, eng2d, 0x08c0, 4); - OUT_RING (chan, 0); - OUT_RING (chan, 1); - OUT_RING (chan, 0); - OUT_RING (chan, 1); - BEGIN_RING(chan, eng2d, 0x08d0, 4); - OUT_RING (chan, 0); - OUT_RING (chan, sx); - OUT_RING (chan, 0); - OUT_RING (chan, sy); -} - -static void -nv50_surface_copy_done(struct nouveau_context *nv) -{ - FIRE_RING(nv->nvc->channel); -} - -static int -nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, - unsigned dx, unsigned dy, unsigned w, unsigned h, - unsigned value) -{ - struct nouveau_channel *chan = nv->nvc->channel; - struct nouveau_grobj *eng2d = nv->nvc->Nv2D; - int rect_format, ret; - - rect_format = nv50_format(dst->format); - if (rect_format < 0) - return 1; - - ret = nv50_surface_set(nv, dst, 1); - if (ret) - return ret; - - BEGIN_RING(chan, eng2d, 0x0580, 3); - OUT_RING (chan, 4); - OUT_RING (chan, rect_format); - OUT_RING (chan, value); - - BEGIN_RING(chan, eng2d, NV50_2D_RECT_X1, 4); - OUT_RING (chan, dx); - OUT_RING (chan, dy); - OUT_RING (chan, dx + w); - OUT_RING (chan, dy + h); - - FIRE_RING(chan); - return 0; -} - -int -nouveau_surface_channel_create_nv50(struct nouveau_channel_context *nvc) -{ - struct nouveau_channel *chan = nvc->channel; - struct nouveau_grobj *eng2d = NULL; - int ret; - - ret = nouveau_grobj_alloc(chan, nvc->next_handle++, NV50_2D, &eng2d); - if (ret) - return ret; - nvc->Nv2D = eng2d; - - BIND_RING (chan, eng2d, nvc->next_subchannel++); - BEGIN_RING(chan, eng2d, NV50_2D_DMA_NOTIFY, 4); - OUT_RING (chan, nvc->sync_notifier->handle); - OUT_RING (chan, chan->vram->handle); - OUT_RING (chan, chan->vram->handle); - OUT_RING (chan, chan->vram->handle); - BEGIN_RING(chan, eng2d, NV50_2D_OPERATION, 1); - OUT_RING (chan, NV50_2D_OPERATION_SRCCOPY); - BEGIN_RING(chan, eng2d, 0x0290, 1); - OUT_RING (chan, 0); - BEGIN_RING(chan, eng2d, 0x0888, 1); - OUT_RING (chan, 1); - - return 0; -} - -int -nouveau_surface_init_nv50(struct nouveau_context *nv) -{ - nv->surface_copy_prep = nv50_surface_copy_prep; - nv->surface_copy = nv50_surface_copy; - nv->surface_copy_done = nv50_surface_copy_done; - nv->surface_fill = nv50_surface_fill; - return 0; -} - diff --git a/src/gallium/winsys/drm/nouveau/Makefile b/src/gallium/winsys/drm/nouveau/Makefile new file mode 100644 index 00000000000..be630ff6d16 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/Makefile @@ -0,0 +1,45 @@ + +TOP = ../../../../.. +include $(TOP)/configs/current + +LIBNAME = nouveau_dri.so + +MINIGLX_SOURCES = + +PIPE_DRIVERS = \ + $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ + $(TOP)/src/gallium/drivers/nv04/libnv04.a \ + $(TOP)/src/gallium/drivers/nv10/libnv10.a \ + $(TOP)/src/gallium/drivers/nv30/libnv30.a \ + $(TOP)/src/gallium/drivers/nv40/libnv40.a \ + $(TOP)/src/gallium/drivers/nv50/libnv50.a + +DRIVER_SOURCES = \ + nouveau_bo.c \ + nouveau_channel.c \ + nouveau_context.c \ + nouveau_device.c \ + nouveau_dma.c \ + nouveau_fence.c \ + nouveau_grobj.c \ + nouveau_lock.c \ + nouveau_notifier.c \ + nouveau_pushbuf.c \ + nouveau_resource.c \ + nouveau_screen.c \ + nouveau_swapbuffers.c \ + nouveau_winsys.c \ + nouveau_winsys_pipe.c \ + nouveau_winsys_softpipe.c \ + nv04_surface.c \ + nv50_surface.c + +C_SOURCES = \ + $(COMMON_GALLIUM_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../Makefile.template + +symlinks: diff --git a/src/gallium/winsys/drm/nouveau/nouveau_bo.c b/src/gallium/winsys/drm/nouveau/nouveau_bo.c new file mode 100644 index 00000000000..b5942994d94 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_bo.c @@ -0,0 +1,470 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +static void +nouveau_mem_free(struct nouveau_device *dev, struct drm_nouveau_mem_alloc *ma, + void **map) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_mem_free mf; + + if (map && *map) { + drmUnmap(*map, ma->size); + *map = NULL; + } + + if (ma->size) { + mf.offset = ma->offset; + mf.flags = ma->flags; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE, + &mf, sizeof(mf)); + ma->size = 0; + } +} + +static int +nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align, + uint32_t flags, struct drm_nouveau_mem_alloc *ma, void **map) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + int ret; + + ma->alignment = align; + ma->size = size; + ma->flags = flags; + if (map) + ma->flags |= NOUVEAU_MEM_MAPPED; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, ma, + sizeof(struct drm_nouveau_mem_alloc)); + if (ret) + return ret; + + if (map) { + ret = drmMap(nvdev->fd, ma->map_handle, ma->size, map); + if (ret) { + *map = NULL; + nouveau_mem_free(dev, ma, map); + return ret; + } + } + + return 0; +} + +static void +nouveau_bo_tmp_del(void *priv) +{ + struct nouveau_resource *r = priv; + + nouveau_fence_ref(NULL, (struct nouveau_fence **)&r->priv); + nouveau_resource_free(&r); +} + +static unsigned +nouveau_bo_tmp_max(struct nouveau_device_priv *nvdev) +{ + struct nouveau_resource *r = nvdev->sa_heap; + unsigned max = 0; + + while (r) { + if (r->in_use && !nouveau_fence(r->priv)->emitted) { + r = r->next; + continue; + } + + if (max < r->size) + max = r->size; + r = r->next; + } + + return max; +} + +static struct nouveau_resource * +nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size, + struct nouveau_fence *fence) +{ + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_resource *r = NULL; + struct nouveau_fence *ref = NULL; + + if (fence) + nouveau_fence_ref(fence, &ref); + else + nouveau_fence_new(chan, &ref); + assert(ref); + + while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) { + if (nouveau_bo_tmp_max(nvdev) < size) { + nouveau_fence_ref(NULL, &ref); + return NULL; + } + + nouveau_fence_flush(chan); + } + nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r); + + return r; +} + +int +nouveau_bo_init(struct nouveau_device *dev) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + int ret; + + ret = nouveau_mem_alloc(dev, 128*1024, 0, NOUVEAU_MEM_AGP | + NOUVEAU_MEM_PCI, &nvdev->sa, &nvdev->sa_map); + if (ret) + return ret; + + ret = nouveau_resource_init(&nvdev->sa_heap, 0, nvdev->sa.size); + if (ret) { + nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); + return ret; + } + + return 0; +} + +void +nouveau_bo_takedown(struct nouveau_device *dev) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map); +} + +int +nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align, + int size, struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo; + int ret; + + if (!dev || !bo || *bo) + return -EINVAL; + + nvbo = calloc(1, sizeof(struct nouveau_bo_priv)); + if (!nvbo) + return -ENOMEM; + nvbo->base.device = dev; + nvbo->base.size = size; + nvbo->base.handle = bo_to_ptr(nvbo); + nvbo->drm.alignment = align; + nvbo->refcount = 1; + + if (flags & NOUVEAU_BO_TILED) { + nvbo->tiled = 1; + if (flags & NOUVEAU_BO_ZTILE) + nvbo->tiled |= 2; + flags &= ~NOUVEAU_BO_TILED; + } + + ret = nouveau_bo_set_status(&nvbo->base, flags); + if (ret) { + free(nvbo); + return ret; + } + + *bo = &nvbo->base; + return 0; +} + +int +nouveau_bo_user(struct nouveau_device *dev, void *ptr, int size, + struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo; + + if (!dev || !bo || *bo) + return -EINVAL; + + nvbo = calloc(1, sizeof(*nvbo)); + if (!nvbo) + return -ENOMEM; + nvbo->base.device = dev; + + nvbo->sysmem = ptr; + nvbo->user = 1; + + nvbo->base.size = size; + nvbo->base.offset = nvbo->drm.offset; + nvbo->base.handle = bo_to_ptr(nvbo); + nvbo->refcount = 1; + *bo = &nvbo->base; + return 0; +} + +int +nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle, + struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo = ptr_to_bo(handle); + + if (!dev || !bo || *bo) + return -EINVAL; + + nvbo->refcount++; + *bo = &nvbo->base; + return 0; +} + +static void +nouveau_bo_del_cb(void *priv) +{ + struct nouveau_bo_priv *nvbo = priv; + + nouveau_fence_ref(NULL, &nvbo->fence); + nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem && !nvbo->user) + free(nvbo->sysmem); + free(nvbo); +} + +void +nouveau_bo_del(struct nouveau_bo **bo) +{ + struct nouveau_bo_priv *nvbo; + + if (!bo || !*bo) + return; + nvbo = nouveau_bo(*bo); + *bo = NULL; + + if (--nvbo->refcount) + return; + + if (nvbo->pending) + nouveau_pushbuf_flush(nvbo->pending->channel, 0); + + if (nvbo->fence) + nouveau_fence_signal_cb(nvbo->fence, nouveau_bo_del_cb, nvbo); + else + nouveau_bo_del_cb(nvbo); +} + +int +nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + + if (!nvbo) + return -EINVAL; + + if (nvbo->pending && + (nvbo->pending->flags & NOUVEAU_BO_WR || flags & NOUVEAU_BO_WR)) { + nouveau_pushbuf_flush(nvbo->pending->channel, 0); + } + + if (flags & NOUVEAU_BO_WR) + nouveau_fence_wait(&nvbo->fence); + else + nouveau_fence_wait(&nvbo->wr_fence); + + if (nvbo->sysmem) + bo->map = nvbo->sysmem; + else + bo->map = nvbo->map; + return 0; +} + +void +nouveau_bo_unmap(struct nouveau_bo *bo) +{ + bo->map = NULL; +} + +static int +nouveau_bo_upload(struct nouveau_bo_priv *nvbo) +{ + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + memcpy(nvbo->map, nvbo->sysmem, nvbo->drm.size); + return 0; +} + +int +nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct drm_nouveau_mem_alloc new; + void *new_map = NULL, *new_sysmem = NULL; + unsigned new_flags = 0, ret; + + assert(!bo->map); + + /* Check current memtype vs requested, if they match do nothing */ + if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM)) + return 0; + if ((nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) && + (flags & NOUVEAU_BO_GART)) + return 0; + if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL)) + return 0; + + memset(&new, 0x00, sizeof(new)); + + /* Allocate new memory */ + if (flags & NOUVEAU_BO_VRAM) + new_flags |= NOUVEAU_MEM_FB; + else + if (flags & NOUVEAU_BO_GART) + new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI); + + if (nvbo->tiled && flags) { + new_flags |= NOUVEAU_MEM_TILE; + if (nvbo->tiled & 2) + new_flags |= NOUVEAU_MEM_TILE_ZETA; + } + + if (new_flags) { + ret = nouveau_mem_alloc(bo->device, bo->size, + nvbo->drm.alignment, new_flags, + &new, &new_map); + if (ret) + return ret; + } else + if (!nvbo->user) { + new_sysmem = malloc(bo->size); + } + + /* Copy old -> new */ + /*XXX: use M2MF */ + if (nvbo->sysmem || nvbo->map) { + struct nouveau_pushbuf_bo *pbo = nvbo->pending; + nvbo->pending = NULL; + nouveau_bo_map(bo, NOUVEAU_BO_RD); + memcpy(new_map, bo->map, bo->size); + nouveau_bo_unmap(bo); + nvbo->pending = pbo; + } + + /* Free old memory */ + if (nvbo->fence) + nouveau_fence_wait(&nvbo->fence); + nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map); + if (nvbo->sysmem && !nvbo->user) + free(nvbo->sysmem); + + nvbo->drm = new; + nvbo->map = new_map; + if (!nvbo->user) + nvbo->sysmem = new_sysmem; + bo->flags = flags; + bo->offset = nvbo->drm.offset; + return 0; +} + +static int +nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_resource *r; + + if (nvchan->user_charge + bo->size > nvdev->sa.size) + return 1; + + if (!(flags & NOUVEAU_BO_GART)) + return 1; + + r = nouveau_bo_tmp(chan, bo->size, fence); + if (!r) + return 1; + nvchan->user_charge += bo->size; + + memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size); + + nvbo->offset = nvdev->sa.offset + r->start; + nvbo->flags = NOUVEAU_BO_GART; + return 0; +} + +static int +nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo, + struct nouveau_fence *fence, uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + int ret; + + ret = nouveau_bo_set_status(bo, flags); + if (ret) { + nouveau_fence_flush(chan); + + ret = nouveau_bo_set_status(bo, flags); + if (ret) + return ret; + } + + if (nvbo->user) + nouveau_bo_upload(nvbo); + + nvbo->offset = nvbo->drm.offset; + if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) + nvbo->flags = NOUVEAU_BO_GART; + else + nvbo->flags = NOUVEAU_BO_VRAM; + + return 0; +} + +int +nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo, + uint32_t flags) +{ + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_fence *fence = nouveau_pushbuf(chan->pushbuf)->fence; + int ret; + + assert(bo->map == NULL); + + if (nvbo->user) { + ret = nouveau_bo_validate_user(chan, bo, fence, flags); + if (ret) { + ret = nouveau_bo_validate_bo(chan, bo, fence, flags); + if (ret) + return ret; + } + } else { + ret = nouveau_bo_validate_bo(chan, bo, fence, flags); + if (ret) + return ret; + } + + if (flags & NOUVEAU_BO_WR) + nouveau_fence_ref(fence, &nvbo->wr_fence); + nouveau_fence_ref(fence, &nvbo->fence); + return 0; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_channel.c b/src/gallium/winsys/drm/nouveau/nouveau_channel.c new file mode 100644 index 00000000000..3b4dcd1ecf2 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_channel.c @@ -0,0 +1,126 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +int +nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma, + uint32_t tt_ctxdma, struct nouveau_channel **chan) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct nouveau_channel_priv *nvchan; + int ret; + + if (!nvdev || !chan || *chan) + return -EINVAL; + + nvchan = calloc(1, sizeof(struct nouveau_channel_priv)); + if (!nvchan) + return -ENOMEM; + nvchan->base.device = dev; + + nvchan->drm.fb_ctxdma_handle = fb_ctxdma; + nvchan->drm.tt_ctxdma_handle = tt_ctxdma; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_CHANNEL_ALLOC, + &nvchan->drm, sizeof(nvchan->drm)); + if (ret) { + free(nvchan); + return ret; + } + + nvchan->base.id = nvchan->drm.channel; + if (nouveau_grobj_ref(&nvchan->base, nvchan->drm.fb_ctxdma_handle, + &nvchan->base.vram) || + nouveau_grobj_ref(&nvchan->base, nvchan->drm.tt_ctxdma_handle, + &nvchan->base.gart)) { + nouveau_channel_free((void *)&nvchan); + return -EINVAL; + } + + ret = drmMap(nvdev->fd, nvchan->drm.ctrl, nvchan->drm.ctrl_size, + (void*)&nvchan->user); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + nvchan->put = &nvchan->user[0x40/4]; + nvchan->get = &nvchan->user[0x44/4]; + nvchan->ref_cnt = &nvchan->user[0x48/4]; + + ret = drmMap(nvdev->fd, nvchan->drm.notifier, nvchan->drm.notifier_size, + (drmAddressPtr)&nvchan->notifier_block); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + + ret = drmMap(nvdev->fd, nvchan->drm.cmdbuf, nvchan->drm.cmdbuf_size, + (void*)&nvchan->pushbuf); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + + ret = nouveau_grobj_alloc(&nvchan->base, 0x00000000, 0x0030, + &nvchan->base.nullobj); + if (ret) { + nouveau_channel_free((void *)&nvchan); + return ret; + } + + nouveau_dma_channel_init(&nvchan->base); + nouveau_pushbuf_init(&nvchan->base); + + *chan = &nvchan->base; + return 0; +} + +void +nouveau_channel_free(struct nouveau_channel **chan) +{ + struct nouveau_channel_priv *nvchan; + struct nouveau_device_priv *nvdev; + struct drm_nouveau_channel_free cf; + + if (!chan || !*chan) + return; + nvchan = nouveau_channel(*chan); + *chan = NULL; + nvdev = nouveau_device(nvchan->base.device); + + FIRE_RING_CH(&nvchan->base); + + nouveau_grobj_free(&nvchan->base.vram); + nouveau_grobj_free(&nvchan->base.gart); + nouveau_grobj_free(&nvchan->base.nullobj); + + cf.channel = nvchan->drm.channel; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf)); + free(nvchan); +} + + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_context.c b/src/gallium/winsys/drm/nouveau/nouveau_context.c new file mode 100644 index 00000000000..74413c408f3 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_context.c @@ -0,0 +1,346 @@ +#include "main/glheader.h" +#include "glapi/glthread.h" +#include +#include "utils.h" + +#include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_context.h" +#include "pipe/p_screen.h" + +#include "nouveau_context.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_winsys_pipe.h" + +#ifdef DEBUG +static const struct dri_debug_control debug_control[] = { + { "bo", DEBUG_BO }, + { NULL, 0 } +}; +int __nouveau_debug = 0; +#endif + +static void +nouveau_channel_context_destroy(struct nouveau_channel_context *nvc) +{ + nouveau_grobj_free(&nvc->NvCtxSurf2D); + nouveau_grobj_free(&nvc->NvImageBlit); + nouveau_grobj_free(&nvc->NvGdiRect); + nouveau_grobj_free(&nvc->NvM2MF); + nouveau_grobj_free(&nvc->Nv2D); + nouveau_grobj_free(&nvc->NvSwzSurf); + nouveau_grobj_free(&nvc->NvSIFM); + + nouveau_notifier_free(&nvc->sync_notifier); + + nouveau_channel_free(&nvc->channel); + + FREE(nvc); +} + +static struct nouveau_channel_context * +nouveau_channel_context_create(struct nouveau_device *dev) +{ + struct nouveau_channel_context *nvc; + int ret; + + nvc = CALLOC_STRUCT(nouveau_channel_context); + if (!nvc) + return NULL; + + if ((ret = nouveau_channel_alloc(dev, 0x8003d001, 0x8003d002, + &nvc->channel))) { + NOUVEAU_ERR("Error creating GPU channel: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + nvc->next_handle = 0x80000000; + + if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1, + &nvc->sync_notifier))) { + NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + switch (dev->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + ret = nouveau_surface_channel_create_nv50(nvc); + break; + default: + ret = nouveau_surface_channel_create_nv04(nvc); + break; + } + + if (ret) { + NOUVEAU_ERR("Error initialising surface objects: %d\n", ret); + nouveau_channel_context_destroy(nvc); + return NULL; + } + + return nvc; +} + +GLboolean +nouveau_context_create(const __GLcontextModes *glVis, + __DRIcontextPrivate *driContextPriv, + void *sharedContextPrivate) +{ + __DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv; + struct nouveau_screen *nv_screen = driScrnPriv->private; + struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context); + struct pipe_context *pipe = NULL; + struct st_context *st_share = NULL; + struct nouveau_channel_context *nvc = NULL; + struct nouveau_device *dev = nv_screen->device; + int i; + + if (sharedContextPrivate) { + st_share = ((struct nouveau_context *)sharedContextPrivate)->st; + } + + switch (dev->chipset & 0xf0) { + case 0x10: + case 0x20: + /* NV10 */ + case 0x30: + /* NV30 */ + case 0x40: + case 0x60: + /* NV40 */ + case 0x50: + case 0x80: + case 0x90: + /* G80 */ + break; + default: + NOUVEAU_ERR("Unsupported chipset: NV%02x\n", dev->chipset); + return GL_FALSE; + } + + driContextPriv->driverPrivate = (void *)nv; + nv->nv_screen = nv_screen; + nv->dri_screen = driScrnPriv; + + { + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + nvdev->ctx = driContextPriv->hHWContext; + nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock; + } + + driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache, + nv->dri_screen->myNum, "nouveau"); +#ifdef DEBUG + __nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"), + debug_control); +#endif + + /*XXX: Hack up a fake region and buffer object for front buffer. + * This will go away with TTM, replaced with a simple reference + * of the front buffer handle passed to us by the DDX. + */ + { + struct pipe_surface *fb_surf; + struct nouveau_pipe_buffer *fb_buf; + struct nouveau_bo_priv *fb_bo; + + fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); + fb_bo->drm.offset = nv_screen->front_offset; + fb_bo->drm.flags = NOUVEAU_MEM_FB; + fb_bo->drm.size = nv_screen->front_pitch * + nv_screen->front_height; + fb_bo->refcount = 1; + fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; + fb_bo->base.offset = fb_bo->drm.offset; + fb_bo->base.handle = (unsigned long)fb_bo; + fb_bo->base.size = fb_bo->drm.size; + fb_bo->base.device = nv_screen->device; + + fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer)); + fb_buf->bo = &fb_bo->base; + + fb_surf = calloc(1, sizeof(struct pipe_surface)); + if (nv_screen->front_cpp == 2) + fb_surf->format = PIPE_FORMAT_R5G6B5_UNORM; + else + fb_surf->format = PIPE_FORMAT_A8R8G8B8_UNORM; + pf_get_block(fb_surf->format, &fb_surf->block); + fb_surf->width = nv_screen->front_pitch / nv_screen->front_cpp; + fb_surf->height = nv_screen->front_height; + fb_surf->stride = fb_surf->width * fb_surf->block.size; + fb_surf->refcount = 1; + fb_surf->buffer = &fb_buf->base; + + nv->frontbuffer = fb_surf; + } + + /* Attempt to share a single channel between multiple contexts from + * a single process. + */ + nvc = nv_screen->nvc; + if (!nvc && st_share) { + struct nouveau_context *snv = st_share->pipe->priv; + if (snv) { + nvc = snv->nvc; + } + } + + /*XXX: temporary - disable multi-context/single-channel on pre-NV4x */ + switch (dev->chipset & 0xf0) { + case 0x40: + case 0x60: + /* NV40 class */ + case 0x50: + case 0x80: + case 0x90: + /* G80 class */ + break; + default: + nvc = NULL; + break; + } + + if (!nvc) { + nvc = nouveau_channel_context_create(dev); + if (!nvc) { + NOUVEAU_ERR("Failed initialising GPU context\n"); + return GL_FALSE; + } + nv_screen->nvc = nvc; + } + + nvc->refcount++; + nv->nvc = nvc; + + /* Find a free slot for a pipe context, allocate a new one if needed */ + nv->pctx_id = -1; + for (i = 0; i < nvc->nr_pctx; i++) { + if (nvc->pctx[i] == NULL) { + nv->pctx_id = i; + break; + } + } + + if (nv->pctx_id < 0) { + nv->pctx_id = nvc->nr_pctx++; + nvc->pctx = + realloc(nvc->pctx, + sizeof(struct pipe_context *) * nvc->nr_pctx); + } + + /* Create pipe */ + switch (dev->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + if (nouveau_surface_init_nv50(nv)) + return GL_FALSE; + break; + default: + if (nouveau_surface_init_nv04(nv)) + return GL_FALSE; + break; + } + + if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) { + struct pipe_screen *pscreen; + + pipe = nouveau_pipe_create(nv); + if (!pipe) + NOUVEAU_ERR("Couldn't create hw pipe\n"); + pscreen = nvc->pscreen; + + nv->cap.hw_vertex_buffer = + pscreen->get_param(pscreen, NOUVEAU_CAP_HW_VTXBUF); + nv->cap.hw_index_buffer = + pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF); + } + + if (!pipe) { + NOUVEAU_MSG("Using softpipe\n"); + pipe = nouveau_create_softpipe(nv); + if (!pipe) { + NOUVEAU_ERR("Error creating pipe, bailing\n"); + return GL_FALSE; + } + } + + pipe->priv = nv; + nv->st = st_create_context(pipe, glVis, st_share); + return GL_TRUE; +} + +void +nouveau_context_destroy(__DRIcontextPrivate *driContextPriv) +{ + struct nouveau_context *nv = driContextPriv->driverPrivate; + struct nouveau_channel_context *nvc = nv->nvc; + + assert(nv); + + st_finish(nv->st); + st_destroy_context(nv->st); + + if (nv->pctx_id >= 0) { + nvc->pctx[nv->pctx_id] = NULL; + if (--nvc->refcount <= 0) { + nouveau_channel_context_destroy(nvc); + nv->nv_screen->nvc = NULL; + } + } + + free(nv); +} + +GLboolean +nouveau_context_bind(__DRIcontextPrivate *driContextPriv, + __DRIdrawablePrivate *driDrawPriv, + __DRIdrawablePrivate *driReadPriv) +{ + struct nouveau_context *nv; + struct nouveau_framebuffer *draw, *read; + + if (!driContextPriv) { + st_make_current(NULL, NULL, NULL); + return GL_TRUE; + } + + nv = driContextPriv->driverPrivate; + draw = driDrawPriv->driverPrivate; + read = driReadPriv->driverPrivate; + + st_make_current(nv->st, draw->stfb, read->stfb); + + if ((nv->dri_drawable != driDrawPriv) || + (nv->last_stamp != driDrawPriv->lastStamp)) { + nv->dri_drawable = driDrawPriv; + st_resize_framebuffer(draw->stfb, driDrawPriv->w, + driDrawPriv->h); + nv->last_stamp = driDrawPriv->lastStamp; + } + + if (driDrawPriv != driReadPriv) { + st_resize_framebuffer(read->stfb, driReadPriv->w, + driReadPriv->h); + } + + return GL_TRUE; +} + +GLboolean +nouveau_context_unbind(__DRIcontextPrivate *driContextPriv) +{ + struct nouveau_context *nv = driContextPriv->driverPrivate; + (void)nv; + + st_flush(nv->st, 0, NULL); + return GL_TRUE; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_context.h b/src/gallium/winsys/drm/nouveau/nouveau_context.h new file mode 100644 index 00000000000..77e2147a2c7 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_context.h @@ -0,0 +1,113 @@ +#ifndef __NOUVEAU_CONTEXT_H__ +#define __NOUVEAU_CONTEXT_H__ + +#include "dri_util.h" +#include "xmlconfig.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +struct nouveau_framebuffer { + struct st_framebuffer *stfb; +}; + +struct nouveau_channel_context { + struct pipe_screen *pscreen; + int refcount; + + unsigned cur_pctx; + unsigned nr_pctx; + struct pipe_context **pctx; + + struct nouveau_channel *channel; + + struct nouveau_notifier *sync_notifier; + + /* Common */ + struct nouveau_grobj *NvM2MF; + /* NV04-NV40 */ + struct nouveau_grobj *NvCtxSurf2D; + struct nouveau_grobj *NvSwzSurf; + struct nouveau_grobj *NvImageBlit; + struct nouveau_grobj *NvGdiRect; + struct nouveau_grobj *NvSIFM; + /* G80 */ + struct nouveau_grobj *Nv2D; + + uint32_t next_handle; + uint32_t next_subchannel; + uint32_t next_sequence; +}; + +struct nouveau_context { + struct st_context *st; + + /* DRI stuff */ + __DRIscreenPrivate *dri_screen; + __DRIdrawablePrivate *dri_drawable; + unsigned int last_stamp; + driOptionCache dri_option_cache; + drm_context_t drm_context; + drmLock drm_lock; + GLboolean locked; + struct nouveau_screen *nv_screen; + struct pipe_surface *frontbuffer; + + struct { + int hw_vertex_buffer; + int hw_index_buffer; + } cap; + + /* Hardware context */ + struct nouveau_channel_context *nvc; + int pctx_id; + + /* pipe_surface accel */ + struct pipe_surface *surf_src, *surf_dst; + unsigned surf_src_offset, surf_dst_offset; + int (*surface_copy_prep)(struct nouveau_context *, + struct pipe_surface *dst, + struct pipe_surface *src); + void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h); + void (*surface_copy_done)(struct nouveau_context *); + int (*surface_fill)(struct nouveau_context *, struct pipe_surface *, + unsigned, unsigned, unsigned, unsigned, unsigned); +}; + +extern GLboolean nouveau_context_create(const __GLcontextModes *, + __DRIcontextPrivate *, void *); +extern void nouveau_context_destroy(__DRIcontextPrivate *); +extern GLboolean nouveau_context_bind(__DRIcontextPrivate *, + __DRIdrawablePrivate *draw, + __DRIdrawablePrivate *read); +extern GLboolean nouveau_context_unbind(__DRIcontextPrivate *); + +#ifdef DEBUG +extern int __nouveau_debug; + +#define DEBUG_BO (1 << 0) + +#define DBG(flag, ...) do { \ + if (__nouveau_debug & (DEBUG_##flag)) \ + NOUVEAU_ERR(__VA_ARGS__); \ +} while(0) +#else +#define DBG(flag, ...) +#endif + +extern void LOCK_HARDWARE(struct nouveau_context *); +extern void UNLOCK_HARDWARE(struct nouveau_context *); + +extern int +nouveau_surface_channel_create_nv04(struct nouveau_channel_context *); +extern int +nouveau_surface_channel_create_nv50(struct nouveau_channel_context *); +extern int nouveau_surface_init_nv04(struct nouveau_context *); +extern int nouveau_surface_init_nv50(struct nouveau_context *); + +extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int); +extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *); + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_device.c b/src/gallium/winsys/drm/nouveau/nouveau_device.c new file mode 100644 index 00000000000..0b452fcd02d --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_device.c @@ -0,0 +1,159 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" + +int +nouveau_device_open_existing(struct nouveau_device **dev, int close, + int fd, drm_context_t ctx) +{ + struct nouveau_device_priv *nvdev; + int ret; + + if (!dev || *dev) + return -EINVAL; + + nvdev = calloc(1, sizeof(*nvdev)); + if (!nvdev) + return -ENOMEM; + nvdev->fd = fd; + nvdev->ctx = ctx; + nvdev->needs_close = close; + + drmCommandNone(nvdev->fd, DRM_NOUVEAU_CARD_INIT); + + if ((ret = nouveau_bo_init(&nvdev->base))) { + nouveau_device_close((void *)&nvdev); + return ret; + } + + { + uint64_t value; + + ret = nouveau_device_get_param(&nvdev->base, + NOUVEAU_GETPARAM_CHIPSET_ID, + &value); + if (ret) { + nouveau_device_close((void *)&nvdev); + return ret; + } + nvdev->base.chipset = value; + } + + *dev = &nvdev->base; + return 0; +} + +int +nouveau_device_open(struct nouveau_device **dev, const char *busid) +{ + drm_context_t ctx; + int fd, ret; + + if (!dev || *dev) + return -EINVAL; + + fd = drmOpen("nouveau", busid); + if (fd < 0) + return -EINVAL; + + ret = drmCreateContext(fd, &ctx); + if (ret) { + drmClose(fd); + return ret; + } + + ret = nouveau_device_open_existing(dev, 1, fd, ctx); + if (ret) { + drmDestroyContext(fd, ctx); + drmClose(fd); + return ret; + } + + return 0; +} + +void +nouveau_device_close(struct nouveau_device **dev) +{ + struct nouveau_device_priv *nvdev; + + if (dev || !*dev) + return; + nvdev = nouveau_device(*dev); + *dev = NULL; + + nouveau_bo_takedown(&nvdev->base); + + if (nvdev->needs_close) { + drmDestroyContext(nvdev->fd, nvdev->ctx); + drmClose(nvdev->fd); + } + free(nvdev); +} + +int +nouveau_device_get_param(struct nouveau_device *dev, + uint64_t param, uint64_t *value) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_getparam g; + int ret; + + if (!nvdev || !value) + return -EINVAL; + + g.param = param; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_GETPARAM, + &g, sizeof(g)); + if (ret) + return ret; + + *value = g.value; + return 0; +} + +int +nouveau_device_set_param(struct nouveau_device *dev, + uint64_t param, uint64_t value) +{ + struct nouveau_device_priv *nvdev = nouveau_device(dev); + struct drm_nouveau_setparam s; + int ret; + + if (!nvdev) + return -EINVAL; + + s.param = param; + s.value = value; + ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_SETPARAM, + &s, sizeof(s)); + if (ret) + return ret; + + return 0; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_dma.c b/src/gallium/winsys/drm/nouveau/nouveau_dma.c new file mode 100644 index 00000000000..f8a8ba04f6d --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_dma.c @@ -0,0 +1,219 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +static inline uint32_t +READ_GET(struct nouveau_channel_priv *nvchan) +{ + return *nvchan->get; +} + +static inline void +WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val) +{ + uint32_t put = ((val << 2) + nvchan->dma->base); + volatile int dum; + + NOUVEAU_DMA_BARRIER; + dum = READ_GET(nvchan); + + *nvchan->put = put; + nvchan->dma->put = val; +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put); +#endif + + NOUVEAU_DMA_BARRIER; +} + +static inline int +LOCAL_GET(struct nouveau_dma_priv *dma, uint32_t *val) +{ + uint32_t get = *val; + + if (get >= dma->base && get <= (dma->base + (dma->max << 2))) { + *val = (get - dma->base) >> 2; + return 1; + } + + return 0; +} + +void +nouveau_dma_channel_init(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + int i; + + nvchan->dma = &nvchan->dma_master; + nvchan->dma->base = nvchan->drm.put_base; + nvchan->dma->cur = nvchan->dma->put = 0; + nvchan->dma->max = (nvchan->drm.cmdbuf_size >> 2) - 2; + nvchan->dma->free = nvchan->dma->max - nvchan->dma->cur; + + RING_SPACE_CH(chan, RING_SKIPS); + for (i = 0; i < RING_SKIPS; i++) + OUT_RING_CH(chan, 0); +} + +#define CHECK_TIMEOUT() do { \ + if ((NOUVEAU_TIME_MSEC() - t_start) > NOUVEAU_DMA_TIMEOUT) \ + return - EBUSY; \ +} while(0) + +int +nouveau_dma_wait(struct nouveau_channel *chan, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + uint32_t get, t_start; + + FIRE_RING_CH(chan); + + t_start = NOUVEAU_TIME_MSEC(); + while (dma->free < size) { + CHECK_TIMEOUT(); + + get = READ_GET(nvchan); + if (!LOCAL_GET(dma, &get)) + continue; + + if (dma->put >= get) { + dma->free = dma->max - dma->cur; + + if (dma->free < size) { +#ifdef NOUVEAU_DMA_DEBUG + dma->push_free = 1; +#endif + OUT_RING_CH(chan, 0x20000000 | dma->base); + if (get <= RING_SKIPS) { + /*corner case - will be idle*/ + if (dma->put <= RING_SKIPS) + WRITE_PUT(nvchan, + RING_SKIPS + 1); + + do { + CHECK_TIMEOUT(); + get = READ_GET(nvchan); + if (!LOCAL_GET(dma, &get)) + get = 0; + } while (get <= RING_SKIPS); + } + + WRITE_PUT(nvchan, RING_SKIPS); + dma->cur = dma->put = RING_SKIPS; + dma->free = get - (RING_SKIPS + 1); + } + } else { + dma->free = get - dma->cur - 1; + } + } + + return 0; +} + +#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF +static void +nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + unsigned mthd_count = 0; + + while (get != put) { + uint32_t gpuget = (get << 2) + nvchan->drm.put_base; + uint32_t data; + + if (get < 0 || get >= nvchan->drm.cmdbuf_size) { + NOUVEAU_ERR("DMA_PT 0x%08x\n", gpuget); + assert(0); + } + data = nvchan->pushbuf[get++]; + + if (mthd_count) { + NOUVEAU_MSG("0x%08x 0x%08x\n", gpuget, data); + mthd_count--; + continue; + } + + switch (data & 0x60000000) { + case 0x00000000: + mthd_count = (data >> 18) & 0x7ff; + NOUVEAU_MSG("0x%08x 0x%08x MTHD " + "Sc %d Mthd 0x%04x Size %d\n", + gpuget, data, (data>>13) & 7, data & 0x1ffc, + mthd_count); + break; + case 0x20000000: + get = (data & 0x1ffffffc) >> 2; + NOUVEAU_MSG("0x%08x 0x%08x JUMP 0x%08x\n", + gpuget, data, data & 0x1ffffffc); + continue; + case 0x40000000: + mthd_count = (data >> 18) & 0x7ff; + NOUVEAU_MSG("0x%08x 0x%08x NINC " + "Sc %d Mthd 0x%04x Size %d\n", + gpuget, data, (data>>13) & 7, data & 0x1ffc, + mthd_count); + break; + case 0x60000000: + /* DMA_OPCODE_CALL apparently, doesn't seem to work on + * my NV40 at least.. + */ + /* fall-through */ + default: + NOUVEAU_MSG("DMA_PUSHER 0x%08x 0x%08x\n", + gpuget, data); + assert(0); + } + } +} +#endif + +void +nouveau_dma_kickoff(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + + if (dma->cur == dma->put) + return; + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free) { + NOUVEAU_ERR("Packet incomplete: %d left\n", dma->push_free); + return; + } +#endif + +#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF + nouveau_dma_parse_pushbuf(chan, dma->put, dma->cur); +#endif + + WRITE_PUT(nvchan, dma->cur); +} diff --git a/src/gallium/winsys/drm/nouveau/nouveau_dma.h b/src/gallium/winsys/drm/nouveau/nouveau_dma.h new file mode 100644 index 00000000000..cfa6d26e828 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_dma.h @@ -0,0 +1,143 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DMA_H__ +#define __NOUVEAU_DMA_H__ + +#include +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +#define RING_SKIPS 8 + +extern int nouveau_dma_wait(struct nouveau_channel *chan, int size); +extern void nouveau_dma_subc_bind(struct nouveau_grobj *); +extern void nouveau_dma_channel_init(struct nouveau_channel *); +extern void nouveau_dma_kickoff(struct nouveau_channel *); + +#ifdef NOUVEAU_DMA_DEBUG +static char faulty[1024]; +#endif + +static inline void +nouveau_dma_out(struct nouveau_channel *chan, uint32_t data) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free == 0) { + NOUVEAU_ERR("No space left in packet at %s\n", faulty); + return; + } + dma->push_free--; +#endif +#ifdef NOUVEAU_DMA_TRACE + { + uint32_t offset = (dma->cur << 2) + dma->base; + NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n", + nvchan->drm.channel, offset, data); + } +#endif + nvchan->pushbuf[dma->cur + (dma->base - nvchan->drm.put_base)/4] = data; + dma->cur++; +} + +static inline void +nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + (void)dma; + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free < size) { + NOUVEAU_ERR("Packet too small. Free=%d, Need=%d\n", + dma->push_free, size); + return; + } +#endif +#ifdef NOUVEAU_DMA_TRACE + while (size--) { + nouveau_dma_out(chan, *ptr); + ptr++; + } +#else + memcpy(&nvchan->pushbuf[dma->cur], ptr, size << 2); +#ifdef NOUVEAU_DMA_DEBUG + dma->push_free -= size; +#endif + dma->cur += size; +#endif +} + +static inline void +nouveau_dma_space(struct nouveau_channel *chan, int size) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + + if (dma->free < size) { + if (nouveau_dma_wait(chan, size) && chan->hang_notify) + chan->hang_notify(chan); + } + dma->free -= size; +#ifdef NOUVEAU_DMA_DEBUG + dma->push_free = size; +#endif +} + +static inline void +nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj, + int method, int size, const char* file, int line) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *dma = nvchan->dma; + (void)dma; + +#ifdef NOUVEAU_DMA_TRACE + NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel, + grobj->handle, grobj->subc, method, size); +#endif + +#ifdef NOUVEAU_DMA_DEBUG + if (dma->push_free) { + NOUVEAU_ERR("Previous packet incomplete: %d left at %s\n", + dma->push_free, faulty); + return; + } + sprintf(faulty,"%s:%d",file,line); +#endif + + nouveau_dma_space(chan, (size + 1)); + nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method); +} + +#define RING_SPACE_CH(ch,sz) nouveau_dma_space((ch), (sz)) +#define BEGIN_RING_CH(ch,gr,m,sz) nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ ) +#define OUT_RING_CH(ch, data) nouveau_dma_out((ch), (data)) +#define OUT_RINGp_CH(ch,ptr,dwords) nouveau_dma_outp((ch), (void*)(ptr), \ + (dwords)) +#define FIRE_RING_CH(ch) nouveau_dma_kickoff((ch)) +#define WAIT_RING_CH(ch,sz) nouveau_dma_wait((ch), (sz)) + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_dri.h b/src/gallium/winsys/drm/nouveau/nouveau_dri.h new file mode 100644 index 00000000000..1207c2d609c --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_dri.h @@ -0,0 +1,28 @@ +#ifndef _NOUVEAU_DRI_ +#define _NOUVEAU_DRI_ + +#include "xf86drm.h" +#include "drm.h" +#include "nouveau_drm.h" + +struct nouveau_dri { + uint32_t device_id; /**< \brief PCI device ID */ + uint32_t width; /**< \brief width in pixels of display */ + uint32_t height; /**< \brief height in scanlines of display */ + uint32_t depth; /**< \brief depth of display (8, 15, 16, 24) */ + uint32_t bpp; /**< \brief bit depth of display (8, 16, 24, 32) */ + + uint32_t bus_type; /**< \brief ths bus type */ + uint32_t bus_mode; /**< \brief bus mode (used for AGP, maybe also for PCI-E ?) */ + + uint32_t front_offset; /**< \brief front buffer offset */ + uint32_t front_pitch; /**< \brief front buffer pitch */ + uint32_t back_offset; /**< \brief private back buffer offset */ + uint32_t back_pitch; /**< \brief private back buffer pitch */ + uint32_t depth_offset; /**< \brief private depth buffer offset */ + uint32_t depth_pitch; /**< \brief private depth buffer pitch */ + +}; + +#endif + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_drmif.h b/src/gallium/winsys/drm/nouveau/nouveau_drmif.h new file mode 100644 index 00000000000..dcd6a5eb0a4 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_drmif.h @@ -0,0 +1,310 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_DRMIF_H__ +#define __NOUVEAU_DRMIF_H__ + +#include +#include +#include + +#include "nouveau/nouveau_device.h" +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_grobj.h" +#include "nouveau/nouveau_notifier.h" +#include "nouveau/nouveau_bo.h" +#include "nouveau/nouveau_resource.h" +#include "nouveau/nouveau_pushbuf.h" + +struct nouveau_device_priv { + struct nouveau_device base; + + int fd; + drm_context_t ctx; + drmLock *lock; + int needs_close; + + struct drm_nouveau_mem_alloc sa; + void *sa_map; + struct nouveau_resource *sa_heap; +}; +#define nouveau_device(n) ((struct nouveau_device_priv *)(n)) + +extern int +nouveau_device_open_existing(struct nouveau_device **, int close, + int fd, drm_context_t ctx); + +extern int +nouveau_device_open(struct nouveau_device **, const char *busid); + +extern void +nouveau_device_close(struct nouveau_device **); + +extern int +nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v); + +extern int +nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val); + +struct nouveau_fence { + struct nouveau_channel *channel; +}; + +struct nouveau_fence_cb { + struct nouveau_fence_cb *next; + void (*func)(void *); + void *priv; +}; + +struct nouveau_fence_priv { + struct nouveau_fence base; + int refcount; + + struct nouveau_fence *next; + struct nouveau_fence_cb *signal_cb; + + uint32_t sequence; + int emitted; + int signalled; +}; +#define nouveau_fence(n) ((struct nouveau_fence_priv *)(n)) + +extern int +nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **); + +extern int +nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **); + +extern int +nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *); + +extern void +nouveau_fence_emit(struct nouveau_fence *); + +extern int +nouveau_fence_wait(struct nouveau_fence **); + +extern void +nouveau_fence_flush(struct nouveau_channel *); + +struct nouveau_pushbuf_reloc { + struct nouveau_pushbuf_bo *pbbo; + uint32_t *ptr; + uint32_t flags; + uint32_t data; + uint32_t vor; + uint32_t tor; +}; + +struct nouveau_pushbuf_bo { + struct nouveau_channel *channel; + struct nouveau_bo *bo; + unsigned flags; + unsigned handled; +}; + +#define NOUVEAU_PUSHBUF_MAX_BUFFERS 1024 +#define NOUVEAU_PUSHBUF_MAX_RELOCS 1024 +struct nouveau_pushbuf_priv { + struct nouveau_pushbuf base; + + struct nouveau_fence *fence; + + unsigned nop_jump; + unsigned start; + unsigned size; + + struct nouveau_pushbuf_bo *buffers; + unsigned nr_buffers; + struct nouveau_pushbuf_reloc *relocs; + unsigned nr_relocs; +}; +#define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n)) + +#define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h)) +#define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_pbrel(h) ((struct nouveau_pushbuf_reloc *)(unsigned long)(h)) +#define bo_to_ptr(o) ((uint64_t)(unsigned long)(o)) +#define ptr_to_bo(h) ((struct nouveau_bo_priv *)(unsigned long)(h)) + +extern int +nouveau_pushbuf_init(struct nouveau_channel *); + +extern int +nouveau_pushbuf_flush(struct nouveau_channel *, unsigned min); + +extern int +nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr, + struct nouveau_bo *, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor); + +struct nouveau_dma_priv { + uint32_t base; + uint32_t max; + uint32_t cur; + uint32_t put; + uint32_t free; + + int push_free; +} dma; + +struct nouveau_channel_priv { + struct nouveau_channel base; + + struct drm_nouveau_channel_alloc drm; + + uint32_t *pushbuf; + void *notifier_block; + + volatile uint32_t *user; + volatile uint32_t *put; + volatile uint32_t *get; + volatile uint32_t *ref_cnt; + + struct nouveau_dma_priv dma_master; + struct nouveau_dma_priv dma_bufmgr; + struct nouveau_dma_priv *dma; + + struct nouveau_fence *fence_head; + struct nouveau_fence *fence_tail; + uint32_t fence_sequence; + + struct nouveau_pushbuf_priv pb; + + unsigned user_charge; +}; +#define nouveau_channel(n) ((struct nouveau_channel_priv *)(n)) + +extern int +nouveau_channel_alloc(struct nouveau_device *, uint32_t fb, uint32_t tt, + struct nouveau_channel **); + +extern void +nouveau_channel_free(struct nouveau_channel **); + +struct nouveau_grobj_priv { + struct nouveau_grobj base; +}; +#define nouveau_grobj(n) ((struct nouveau_grobj_priv *)(n)) + +extern int nouveau_grobj_alloc(struct nouveau_channel *, uint32_t handle, + int class, struct nouveau_grobj **); +extern int nouveau_grobj_ref(struct nouveau_channel *, uint32_t handle, + struct nouveau_grobj **); +extern void nouveau_grobj_free(struct nouveau_grobj **); + + +struct nouveau_notifier_priv { + struct nouveau_notifier base; + + struct drm_nouveau_notifierobj_alloc drm; + volatile void *map; +}; +#define nouveau_notifier(n) ((struct nouveau_notifier_priv *)(n)) + +extern int +nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, int count, + struct nouveau_notifier **); + +extern void +nouveau_notifier_free(struct nouveau_notifier **); + +extern void +nouveau_notifier_reset(struct nouveau_notifier *, int id); + +extern uint32_t +nouveau_notifier_status(struct nouveau_notifier *, int id); + +extern uint32_t +nouveau_notifier_return_val(struct nouveau_notifier *, int id); + +extern int +nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status, + int timeout); + +struct nouveau_bo_priv { + struct nouveau_bo base; + + struct nouveau_pushbuf_bo *pending; + struct nouveau_fence *fence; + struct nouveau_fence *wr_fence; + + struct drm_nouveau_mem_alloc drm; + void *map; + + void *sysmem; + int user; + + int refcount; + + uint64_t offset; + uint64_t flags; + int tiled; +}; +#define nouveau_bo(n) ((struct nouveau_bo_priv *)(n)) + +extern int +nouveau_bo_init(struct nouveau_device *); + +extern void +nouveau_bo_takedown(struct nouveau_device *); + +extern int +nouveau_bo_new(struct nouveau_device *, uint32_t flags, int align, int size, + struct nouveau_bo **); + +extern int +nouveau_bo_user(struct nouveau_device *, void *ptr, int size, + struct nouveau_bo **); + +extern int +nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **); + +extern int +nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags); + +extern void +nouveau_bo_del(struct nouveau_bo **); + +extern int +nouveau_bo_map(struct nouveau_bo *, uint32_t flags); + +extern void +nouveau_bo_unmap(struct nouveau_bo *); + +extern int +nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *, + uint32_t flags); + +extern int +nouveau_resource_init(struct nouveau_resource **heap, unsigned start, + unsigned size); + +extern int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **); + +extern void +nouveau_resource_free(struct nouveau_resource **); + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_fence.c b/src/gallium/winsys/drm/nouveau/nouveau_fence.c new file mode 100644 index 00000000000..e7b0b4ff079 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_fence.c @@ -0,0 +1,214 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" +#include "nouveau_local.h" + +static void +nouveau_fence_del_unsignalled(struct nouveau_fence *fence) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); + struct nouveau_fence *le; + + if (nvchan->fence_head == fence) { + nvchan->fence_head = nouveau_fence(fence)->next; + if (nvchan->fence_head == NULL) + nvchan->fence_tail = NULL; + return; + } + + le = nvchan->fence_head; + while (le && nouveau_fence(le)->next != fence) + le = nouveau_fence(le)->next; + assert(le && nouveau_fence(le)->next == fence); + nouveau_fence(le)->next = nouveau_fence(fence)->next; + if (nvchan->fence_tail == fence) + nvchan->fence_tail = le; +} + +static void +nouveau_fence_del(struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence || !*fence) + return; + nvfence = nouveau_fence(*fence); + *fence = NULL; + + if (--nvfence->refcount) + return; + + if (nvfence->emitted && !nvfence->signalled) { + if (nvfence->signal_cb) { + nvfence->refcount++; + nouveau_fence_wait((void *)&nvfence); + return; + } + + nouveau_fence_del_unsignalled(&nvfence->base); + } + free(nvfence); +} + +int +nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!chan || !fence || *fence) + return -EINVAL; + + nvfence = calloc(1, sizeof(struct nouveau_fence_priv)); + if (!nvfence) + return -ENOMEM; + nvfence->base.channel = chan; + nvfence->refcount = 1; + + *fence = &nvfence->base; + return 0; +} + +int +nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence) + return -EINVAL; + + if (*fence) { + nouveau_fence_del(fence); + *fence = NULL; + } + + if (ref) { + nvfence = nouveau_fence(ref); + nvfence->refcount++; + *fence = &nvfence->base; + } + + return 0; +} + +int +nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *), + void *priv) +{ + struct nouveau_fence_priv *nvfence = nouveau_fence(fence); + struct nouveau_fence_cb *cb; + + if (!nvfence || !func) + return -EINVAL; + + cb = malloc(sizeof(struct nouveau_fence_cb)); + if (!cb) + return -ENOMEM; + + cb->func = func; + cb->priv = priv; + cb->next = nvfence->signal_cb; + nvfence->signal_cb = cb; + return 0; +} + +void +nouveau_fence_emit(struct nouveau_fence *fence) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel); + struct nouveau_fence_priv *nvfence = nouveau_fence(fence); + + nvfence->emitted = 1; + nvfence->sequence = ++nvchan->fence_sequence; + if (nvfence->sequence == 0xffffffff) + NOUVEAU_ERR("AII wrap unhandled\n"); + + /*XXX: assumes subc 0 is populated */ + RING_SPACE_CH(fence->channel, 2); + OUT_RING_CH (fence->channel, 0x00040050); + OUT_RING_CH (fence->channel, nvfence->sequence); + + if (nvchan->fence_tail) { + nouveau_fence(nvchan->fence_tail)->next = fence; + } else { + nvchan->fence_head = fence; + } + nvchan->fence_tail = fence; +} + +void +nouveau_fence_flush(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + uint32_t sequence = *nvchan->ref_cnt; + + while (nvchan->fence_head) { + struct nouveau_fence_priv *nvfence; + + nvfence = nouveau_fence(nvchan->fence_head); + if (nvfence->sequence > sequence) + break; + nouveau_fence_del_unsignalled(&nvfence->base); + nvfence->signalled = 1; + + if (nvfence->signal_cb) { + struct nouveau_fence *fence = NULL; + + nouveau_fence_ref(&nvfence->base, &fence); + + while (nvfence->signal_cb) { + struct nouveau_fence_cb *cb; + + cb = nvfence->signal_cb; + nvfence->signal_cb = cb->next; + cb->func(cb->priv); + free(cb); + } + + nouveau_fence_ref(NULL, &fence); + } + } +} + +int +nouveau_fence_wait(struct nouveau_fence **fence) +{ + struct nouveau_fence_priv *nvfence; + + if (!fence || !*fence) + return -EINVAL; + nvfence = nouveau_fence(*fence); + + if (nvfence->emitted) { + while (!nvfence->signalled) + nouveau_fence_flush(nvfence->base.channel); + } + nouveau_fence_ref(NULL, fence); + + return 0; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_grobj.c b/src/gallium/winsys/drm/nouveau/nouveau_grobj.c new file mode 100644 index 00000000000..51523897d58 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_grobj.c @@ -0,0 +1,107 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" + +int +nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle, + int class, struct nouveau_grobj **grobj) +{ + struct nouveau_device_priv *nvdev = nouveau_device(chan->device); + struct nouveau_grobj_priv *nvgrobj; + struct drm_nouveau_grobj_alloc g; + int ret; + + if (!nvdev || !grobj || *grobj) + return -EINVAL; + + nvgrobj = calloc(1, sizeof(*nvgrobj)); + if (!nvgrobj) + return -ENOMEM; + nvgrobj->base.channel = chan; + nvgrobj->base.handle = handle; + nvgrobj->base.grclass = class; + + g.channel = chan->id; + g.handle = handle; + g.class = class; + ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC, + &g, sizeof(g)); + if (ret) { + nouveau_grobj_free((void *)&nvgrobj); + return ret; + } + + *grobj = &nvgrobj->base; + return 0; +} + +int +nouveau_grobj_ref(struct nouveau_channel *chan, uint32_t handle, + struct nouveau_grobj **grobj) +{ + struct nouveau_grobj_priv *nvgrobj; + + if (!chan || !grobj || *grobj) + return -EINVAL; + + nvgrobj = calloc(1, sizeof(struct nouveau_grobj_priv)); + if (!nvgrobj) + return -ENOMEM; + nvgrobj->base.channel = chan; + nvgrobj->base.handle = handle; + nvgrobj->base.grclass = 0; + + *grobj = &nvgrobj->base; + return 0; +} + +void +nouveau_grobj_free(struct nouveau_grobj **grobj) +{ + struct nouveau_device_priv *nvdev; + struct nouveau_channel_priv *chan; + struct nouveau_grobj_priv *nvgrobj; + + if (!grobj || !*grobj) + return; + nvgrobj = nouveau_grobj(*grobj); + *grobj = NULL; + + + chan = nouveau_channel(nvgrobj->base.channel); + nvdev = nouveau_device(chan->base.device); + + if (nvgrobj->base.grclass) { + struct drm_nouveau_gpuobj_free f; + + f.channel = chan->drm.channel; + f.handle = nvgrobj->base.handle; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, + &f, sizeof(f)); + } + free(nvgrobj); +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_local.h b/src/gallium/winsys/drm/nouveau/nouveau_local.h new file mode 100644 index 00000000000..e878a408037 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_local.h @@ -0,0 +1,117 @@ +#ifndef __NOUVEAU_LOCAL_H__ +#define __NOUVEAU_LOCAL_H__ + +#include "pipe/p_compiler.h" +#include "nouveau_winsys_pipe.h" +#include + +struct pipe_buffer; + +/* Debug output */ +#define NOUVEAU_MSG(fmt, args...) do { \ + fprintf(stdout, "nouveau: "fmt, ##args); \ + fflush(stdout); \ +} while(0) + +#define NOUVEAU_ERR(fmt, args...) do { \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); \ + fflush(stderr); \ +} while(0) + +#define NOUVEAU_TIME_MSEC() 0 + +/* User FIFO control */ +//#define NOUVEAU_DMA_TRACE +//#define NOUVEAU_DMA_DEBUG +//#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF +#define NOUVEAU_DMA_BARRIER +#define NOUVEAU_DMA_TIMEOUT 2000 + +/* Push buffer access macros */ +static INLINE void +OUT_RING(struct nouveau_channel *chan, unsigned data) +{ + *(chan->pushbuf->cur++) = (data); +} + +static INLINE void +OUT_RINGp(struct nouveau_channel *chan, uint32_t *data, unsigned size) +{ + memcpy(chan->pushbuf->cur, data, size * 4); + chan->pushbuf->cur += size; +} + +static INLINE void +OUT_RINGf(struct nouveau_channel *chan, float f) +{ + union { uint32_t i; float f; } c; + c.f = f; + OUT_RING(chan, c.i); +} + +static INLINE void +BEGIN_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, + unsigned mthd, unsigned size) +{ + if (chan->pushbuf->remaining < (size + 1)) + nouveau_pushbuf_flush(chan, (size + 1)); + OUT_RING(chan, (gr->subc << 13) | (size << 18) | mthd); + chan->pushbuf->remaining -= (size + 1); +} + +static INLINE void +FIRE_RING(struct nouveau_channel *chan) +{ + nouveau_pushbuf_flush(chan, 0); +} + +static INLINE void +BIND_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, unsigned subc) +{ + gr->subc = subc; + BEGIN_RING(chan, gr, 0x0000, 1); + OUT_RING (chan, gr->handle); +} + +static INLINE void +OUT_RELOC(struct nouveau_channel *chan, struct nouveau_bo *bo, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + nouveau_pushbuf_emit_reloc(chan, chan->pushbuf->cur++, bo, + data, flags, vor, tor); +} + +/* Raw data + flags depending on FB/TT buffer */ +static INLINE void +OUT_RELOCd(struct nouveau_channel *chan, struct nouveau_bo *bo, + unsigned data, unsigned flags, unsigned vor, unsigned tor) +{ + OUT_RELOC(chan, bo, data, flags | NOUVEAU_BO_OR, vor, tor); +} + +/* FB/TT object handle */ +static INLINE void +OUT_RELOCo(struct nouveau_channel *chan, struct nouveau_bo *bo, + unsigned flags) +{ + OUT_RELOC(chan, bo, 0, flags | NOUVEAU_BO_OR, + chan->vram->handle, chan->gart->handle); +} + +/* Low 32-bits of offset */ +static INLINE void +OUT_RELOCl(struct nouveau_channel *chan, struct nouveau_bo *bo, + unsigned delta, unsigned flags) +{ + OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_LOW, 0, 0); +} + +/* High 32-bits of offset */ +static INLINE void +OUT_RELOCh(struct nouveau_channel *chan, struct nouveau_bo *bo, + unsigned delta, unsigned flags) +{ + OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_HIGH, 0, 0); +} + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_lock.c b/src/gallium/winsys/drm/nouveau/nouveau_lock.c new file mode 100644 index 00000000000..9adb9ac8547 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_lock.c @@ -0,0 +1,94 @@ +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "main/glheader.h" +#include "glapi/glthread.h" +#include + +#include "nouveau_context.h" +#include "nouveau_screen.h" + +_glthread_DECLARE_STATIC_MUTEX( lockMutex ); + +static void +nouveau_contended_lock(struct nouveau_context *nv, GLuint flags) +{ + __DRIdrawablePrivate *dPriv = nv->dri_drawable; + __DRIscreenPrivate *sPriv = nv->dri_screen; + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + drmGetLock(nvdev->fd, nvdev->ctx, flags); + + /* If the window moved, may need to set a new cliprect now. + * + * NOTE: This releases and regains the hw lock, so all state + * checking must be done *after* this call: + */ + if (dPriv) + DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv); +} + +/* Lock the hardware and validate our state. + */ +void +LOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + char __ret=0; + + _glthread_LOCK_MUTEX(lockMutex); + assert(!nv->locked); + + DRM_CAS(nvdev->lock, nvdev->ctx, + (DRM_LOCK_HELD | nvdev->ctx), __ret); + + if (__ret) + nouveau_contended_lock(nv, 0); + nv->locked = GL_TRUE; +} + + + /* Unlock the hardware using the global current context + */ +void +UNLOCK_HARDWARE(struct nouveau_context *nv) +{ + struct nouveau_screen *nv_screen = nv->nv_screen; + struct nouveau_device *dev = nv_screen->device; + struct nouveau_device_priv *nvdev = nouveau_device(dev); + + assert(nv->locked); + nv->locked = GL_FALSE; + + DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx); + + _glthread_UNLOCK_MUTEX(lockMutex); +} diff --git a/src/gallium/winsys/drm/nouveau/nouveau_notifier.c b/src/gallium/winsys/drm/nouveau/nouveau_notifier.c new file mode 100644 index 00000000000..01e8f38440e --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_notifier.c @@ -0,0 +1,137 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +#define NOTIFIER(__v) \ + struct nouveau_notifier_priv *nvnotify = nouveau_notifier(notifier); \ + volatile uint32_t *__v = (void*)nvnotify->map + (id * 32) + +int +nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, + int count, struct nouveau_notifier **notifier) +{ + struct nouveau_notifier_priv *nvnotify; + int ret; + + if (!chan || !notifier || *notifier) + return -EINVAL; + + nvnotify = calloc(1, sizeof(struct nouveau_notifier_priv)); + if (!nvnotify) + return -ENOMEM; + nvnotify->base.channel = chan; + nvnotify->base.handle = handle; + + nvnotify->drm.channel = chan->id; + nvnotify->drm.handle = handle; + nvnotify->drm.count = count; + if ((ret = drmCommandWriteRead(nouveau_device(chan->device)->fd, + DRM_NOUVEAU_NOTIFIEROBJ_ALLOC, + &nvnotify->drm, + sizeof(nvnotify->drm)))) { + nouveau_notifier_free((void *)&nvnotify); + return ret; + } + + nvnotify->map = (void *)nouveau_channel(chan)->notifier_block + + nvnotify->drm.offset; + *notifier = &nvnotify->base; + return 0; +} + +void +nouveau_notifier_free(struct nouveau_notifier **notifier) +{ + + struct nouveau_notifier_priv *nvnotify; + struct nouveau_channel_priv *nvchan; + struct nouveau_device_priv *nvdev; + struct drm_nouveau_gpuobj_free f; + + if (!notifier || !*notifier) + return; + nvnotify = nouveau_notifier(*notifier); + *notifier = NULL; + + nvchan = nouveau_channel(nvnotify->base.channel); + nvdev = nouveau_device(nvchan->base.device); + + f.channel = nvchan->drm.channel; + f.handle = nvnotify->base.handle; + drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f)); + free(nvnotify); +} + +void +nouveau_notifier_reset(struct nouveau_notifier *notifier, int id) +{ + NOTIFIER(n); + + n[NV_NOTIFY_TIME_0 /4] = 0x00000000; + n[NV_NOTIFY_TIME_1 /4] = 0x00000000; + n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000; + n[NV_NOTIFY_STATE /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS << + NV_NOTIFY_STATE_STATUS_SHIFT); +} + +uint32_t +nouveau_notifier_status(struct nouveau_notifier *notifier, int id) +{ + NOTIFIER(n); + + return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; +} + +uint32_t +nouveau_notifier_return_val(struct nouveau_notifier *notifier, int id) +{ + NOTIFIER(n); + + return n[NV_NOTIFY_RETURN_VALUE/4]; +} + +int +nouveau_notifier_wait_status(struct nouveau_notifier *notifier, int id, + int status, int timeout) +{ + NOTIFIER(n); + uint32_t time = 0, t_start = NOUVEAU_TIME_MSEC(); + + while (time <= timeout) { + uint32_t v; + + v = n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT; + if (v == status) + return 0; + + if (timeout) + time = NOUVEAU_TIME_MSEC() - t_start; + } + + return -EBUSY; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c new file mode 100644 index 00000000000..815046ba85f --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c @@ -0,0 +1,271 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_dma.h" + +#define PB_BUFMGR_DWORDS (4096 / 2) +#define PB_MIN_USER_DWORDS 2048 + +static int +nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; + + assert((min + 1) <= nvchan->dma->max); + + /* Wait for enough space in push buffer */ + min = min < PB_MIN_USER_DWORDS ? PB_MIN_USER_DWORDS : min; + min += 1; /* a bit extra for the NOP */ + if (nvchan->dma->free < min) + WAIT_RING_CH(chan, min); + + /* Insert NOP, may turn into a jump later */ + RING_SPACE_CH(chan, 1); + nvpb->nop_jump = nvchan->dma->cur; + OUT_RING_CH(chan, 0); + + /* Any remaining space is available to the user */ + nvpb->start = nvchan->dma->cur; + nvpb->size = nvchan->dma->free; + nvpb->base.channel = chan; + nvpb->base.remaining = nvpb->size; + nvpb->base.cur = &nvchan->pushbuf[nvpb->start]; + + /* Create a new fence object for this "frame" */ + nouveau_fence_ref(NULL, &nvpb->fence); + nouveau_fence_new(chan, &nvpb->fence); + + return 0; +} + +int +nouveau_pushbuf_init(struct nouveau_channel *chan) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_dma_priv *m = &nvchan->dma_master; + struct nouveau_dma_priv *b = &nvchan->dma_bufmgr; + int i; + + if (!nvchan) + return -EINVAL; + + /* Reassign last bit of push buffer for a "separate" bufmgr + * ring buffer + */ + m->max -= PB_BUFMGR_DWORDS; + m->free -= PB_BUFMGR_DWORDS; + + b->base = m->base + ((m->max + 2) << 2); + b->max = PB_BUFMGR_DWORDS - 2; + b->cur = b->put = 0; + b->free = b->max - b->cur; + + /* Some NOPs just to be safe + *XXX: RING_SKIPS + */ + nvchan->dma = b; + RING_SPACE_CH(chan, 8); + for (i = 0; i < 8; i++) + OUT_RING_CH(chan, 0); + nvchan->dma = m; + + nouveau_pushbuf_space(chan, 0); + chan->pushbuf = &nvchan->pb.base; + + nvchan->pb.buffers = calloc(NOUVEAU_PUSHBUF_MAX_BUFFERS, + sizeof(struct nouveau_pushbuf_bo)); + nvchan->pb.relocs = calloc(NOUVEAU_PUSHBUF_MAX_RELOCS, + sizeof(struct nouveau_pushbuf_reloc)); + return 0; +} + +static uint32_t +nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo, + struct nouveau_pushbuf_reloc *r) +{ + uint32_t push; + + if (r->flags & NOUVEAU_BO_LOW) { + push = bo->offset + r->data; + } else + if (r->flags & NOUVEAU_BO_HIGH) { + push = (bo->offset + r->data) >> 32; + } else { + push = r->data; + } + + if (r->flags & NOUVEAU_BO_OR) { + if (bo->flags & NOUVEAU_BO_VRAM) + push |= r->vor; + else + push |= r->tor; + } + + return push; +} + +/* This would be our TTM "superioctl" */ +int +nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min) +{ + struct nouveau_channel_priv *nvchan = nouveau_channel(chan); + struct nouveau_pushbuf_priv *nvpb = &nvchan->pb; + int ret, i; + + if (nvpb->base.remaining == nvpb->size) + return 0; + + nouveau_fence_flush(chan); + + nvpb->size -= nvpb->base.remaining; + nvchan->dma->cur += nvpb->size; + nvchan->dma->free -= nvpb->size; + assert(nvchan->dma->cur <= nvchan->dma->max); + + nvchan->dma = &nvchan->dma_bufmgr; + nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 | + (nvchan->dma->base + (nvchan->dma->cur << 2)); + + /* Validate buffers + apply relocations */ + nvchan->user_charge = 0; + for (i = 0; i < nvpb->nr_relocs; i++) { + struct nouveau_pushbuf_reloc *r = &nvpb->relocs[i]; + struct nouveau_pushbuf_bo *pbbo = r->pbbo; + struct nouveau_bo *bo = pbbo->bo; + + /* Validated, mem matches presumed, no relocation necessary */ + if (pbbo->handled & 2) { + if (!(pbbo->handled & 1)) + assert(0); + continue; + } + + /* Not yet validated, do it now */ + if (!(pbbo->handled & 1)) { + ret = nouveau_bo_validate(chan, bo, pbbo->flags); + if (ret) { + assert(0); + return ret; + } + pbbo->handled |= 1; + + if (bo->offset == nouveau_bo(bo)->offset && + bo->flags == nouveau_bo(bo)->flags) { + pbbo->handled |= 2; + continue; + } + bo->offset = nouveau_bo(bo)->offset; + bo->flags = nouveau_bo(bo)->flags; + } + + /* Apply the relocation */ + *r->ptr = nouveau_pushbuf_calc_reloc(bo, r); + } + nvpb->nr_relocs = 0; + + /* Dereference all buffers on validate list */ + for (i = 0; i < nvpb->nr_buffers; i++) { + struct nouveau_pushbuf_bo *pbbo = &nvpb->buffers[i]; + + nouveau_bo(pbbo->bo)->pending = NULL; + nouveau_bo_del(&pbbo->bo); + } + nvpb->nr_buffers = 0; + + /* Switch back to user's ring */ + RING_SPACE_CH(chan, 1); + OUT_RING_CH(chan, 0x20000000 | ((nvpb->start << 2) + + nvchan->dma_master.base)); + nvchan->dma = &nvchan->dma_master; + + /* Fence + kickoff */ + nouveau_fence_emit(nvpb->fence); + FIRE_RING_CH(chan); + + /* Allocate space for next push buffer */ + ret = nouveau_pushbuf_space(chan, min); + assert(!ret); + + return 0; +} + +static struct nouveau_pushbuf_bo * +nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo) +{ + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); + struct nouveau_bo_priv *nvbo = nouveau_bo(bo); + struct nouveau_pushbuf_bo *pbbo; + + if (nvbo->pending) + return nvbo->pending; + + if (nvpb->nr_buffers >= NOUVEAU_PUSHBUF_MAX_BUFFERS) + return NULL; + pbbo = nvpb->buffers + nvpb->nr_buffers++; + nvbo->pending = pbbo; + + nouveau_bo_ref(bo->device, bo->handle, &pbbo->bo); + pbbo->channel = chan; + pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART; + pbbo->handled = 0; + return pbbo; +} + +int +nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr, + struct nouveau_bo *bo, uint32_t data, uint32_t flags, + uint32_t vor, uint32_t tor) +{ + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf); + struct nouveau_pushbuf_bo *pbbo; + struct nouveau_pushbuf_reloc *r; + + if (nvpb->nr_relocs >= NOUVEAU_PUSHBUF_MAX_RELOCS) + return -ENOMEM; + + pbbo = nouveau_pushbuf_emit_buffer(chan, bo); + if (!pbbo) + return -ENOMEM; + pbbo->flags |= (flags & NOUVEAU_BO_RDWR); + pbbo->flags &= (flags | NOUVEAU_BO_RDWR); + + r = nvpb->relocs + nvpb->nr_relocs++; + r->pbbo = pbbo; + r->ptr = ptr; + r->flags = flags; + r->data = data; + r->vor = vor; + r->tor = tor; + + if (flags & NOUVEAU_BO_DUMMY) + *(uint32_t *)ptr = 0; + else + *(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r); + return 0; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_resource.c b/src/gallium/winsys/drm/nouveau/nouveau_resource.c new file mode 100644 index 00000000000..3bbcb5c45e0 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_resource.c @@ -0,0 +1,116 @@ +/* + * Copyright 2007 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include +#include + +#include "nouveau_drmif.h" +#include "nouveau_local.h" + +int +nouveau_resource_init(struct nouveau_resource **heap, + unsigned start, unsigned size) +{ + struct nouveau_resource *r; + + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = start; + r->size = size; + *heap = r; + return 0; +} + +int +nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv, + struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!heap || !size || !res || *res) + return 1; + + while (heap) { + if (!heap->in_use && heap->size >= size) { + r = calloc(1, sizeof(struct nouveau_resource)); + if (!r) + return 1; + + r->start = (heap->start + heap->size) - size; + r->size = size; + r->in_use = 1; + r->priv = priv; + + heap->size -= size; + + r->next = heap->next; + if (heap->next) + heap->next->prev = r; + r->prev = heap; + heap->next = r; + + *res = r; + return 0; + } + + heap = heap->next; + } + + return 1; +} + +void +nouveau_resource_free(struct nouveau_resource **res) +{ + struct nouveau_resource *r; + + if (!res || !*res) + return; + r = *res; + *res = NULL; + + r->in_use = 0; + + if (r->next && !r->next->in_use) { + struct nouveau_resource *new = r->next; + + new->prev = r->prev; + if (r->prev) + r->prev->next = new; + new->size += r->size; + new->start = r->start; + + free(r); + r = new; + } + + if (r->prev && !r->prev->in_use) { + r->prev->next = r->next; + if (r->next) + r->next->prev = r->prev; + r->prev->size += r->size; + free(r); + } + +} diff --git a/src/gallium/winsys/drm/nouveau/nouveau_screen.c b/src/gallium/winsys/drm/nouveau/nouveau_screen.c new file mode 100644 index 00000000000..df1fe7e69b4 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_screen.c @@ -0,0 +1,310 @@ +#include "utils.h" +#include "vblank.h" +#include "xmlpool.h" + +#include "pipe/p_context.h" +#include "state_tracker/st_public.h" +#include "state_tracker/st_cb_fbo.h" + +#include "nouveau_context.h" +#include "nouveau_drm.h" +#include "nouveau_dri.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11 +#error nouveau_drm.h version does not match expected version +#endif + +/* Extension stuff, enabling of extensions handled by Gallium's GL state + * tracker. But, we still need to define the entry points we want. + */ +#define need_GL_ARB_fragment_program +#define need_GL_ARB_multisample +#define need_GL_ARB_occlusion_query +#define need_GL_ARB_point_parameters +#define need_GL_ARB_shader_objects +#define need_GL_ARB_texture_compression +#define need_GL_ARB_vertex_program +#define need_GL_ARB_vertex_shader +#define need_GL_ARB_vertex_buffer_object +#define need_GL_EXT_compiled_vertex_array +#define need_GL_EXT_fog_coord +#define need_GL_EXT_secondary_color +#define need_GL_EXT_framebuffer_object +#define need_GL_VERSION_2_0 +#define need_GL_VERSION_2_1 +#include "extension_helper.h" + +const struct dri_extension card_extensions[] = +{ + { "GL_ARB_multisample", GL_ARB_multisample_functions }, + { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, + { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, + { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, + { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, + { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, + { "GL_ARB_texture_compression", GL_ARB_texture_compression_functions }, + { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, + { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, + { "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions }, + { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions }, + { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, + { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions }, + { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, + { NULL, 0 } +}; + +PUBLIC const char __driConfigOptions[] = +DRI_CONF_BEGIN +DRI_CONF_END; +static const GLuint __driNConfigOptions = 0; + +extern const struct dri_extension common_extensions[]; +extern const struct dri_extension nv40_extensions[]; + +static GLboolean +nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_dri *nv_dri = driScrnPriv->pDevPriv; + struct nouveau_screen *nv_screen; + int ret; + + if (driScrnPriv->devPrivSize != sizeof(struct nouveau_dri)) { + NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); + return GL_FALSE; + } + + nv_screen = CALLOC_STRUCT(nouveau_screen); + if (!nv_screen) + return GL_FALSE; + nv_screen->driScrnPriv = driScrnPriv; + driScrnPriv->private = (void *)nv_screen; + + driParseOptionInfo(&nv_screen->option_cache, + __driConfigOptions, __driNConfigOptions); + + if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, + driScrnPriv->fd, 0))) { + NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); + return GL_FALSE; + } + + nv_screen->front_offset = nv_dri->front_offset; + nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); + nv_screen->front_cpp = nv_dri->bpp / 8; + nv_screen->front_height = nv_dri->height; + + return GL_TRUE; +} + +static void +nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_screen *nv_screen = driScrnPriv->private; + + driScrnPriv->private = NULL; + FREE(nv_screen); +} + +static GLboolean +nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, + __DRIdrawablePrivate * driDrawPriv, + const __GLcontextModes *glVis, GLboolean pixmapBuffer) +{ + struct nouveau_framebuffer *nvfb; + enum pipe_format colour, depth, stencil; + + if (pixmapBuffer) + return GL_FALSE; + + nvfb = CALLOC_STRUCT(nouveau_framebuffer); + if (!nvfb) + return GL_FALSE; + + if (glVis->redBits == 5) + colour = PIPE_FORMAT_R5G6B5_UNORM; + else + colour = PIPE_FORMAT_A8R8G8B8_UNORM; + + if (glVis->depthBits == 16) + depth = PIPE_FORMAT_Z16_UNORM; + else if (glVis->depthBits == 24) + depth = PIPE_FORMAT_Z24S8_UNORM; + else + depth = PIPE_FORMAT_NONE; + + if (glVis->stencilBits == 8) + stencil = PIPE_FORMAT_Z24S8_UNORM; + else + stencil = PIPE_FORMAT_NONE; + + nvfb->stfb = st_create_framebuffer(glVis, colour, depth, stencil, + driDrawPriv->w, driDrawPriv->h, + (void*)nvfb); + if (!nvfb->stfb) { + free(nvfb); + return GL_FALSE; + } + + driDrawPriv->driverPrivate = (void *)nvfb; + return GL_TRUE; +} + +static void +nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) +{ + struct nouveau_framebuffer *nvfb; + + nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate; + st_unreference_framebuffer(&nvfb->stfb); + free(nvfb); +} + +static struct __DriverAPIRec +nouveau_api = { + .InitDriver = nouveau_screen_create, + .DestroyScreen = nouveau_screen_destroy, + .CreateContext = nouveau_context_create, + .DestroyContext = nouveau_context_destroy, + .CreateBuffer = nouveau_create_buffer, + .DestroyBuffer = nouveau_destroy_buffer, + .SwapBuffers = nouveau_swap_buffers, + .MakeCurrent = nouveau_context_bind, + .UnbindContext = nouveau_context_unbind, + .GetSwapInfo = NULL, + .GetMSC = NULL, + .WaitForMSC = NULL, + .WaitForSBC = NULL, + .SwapBuffersMSC = NULL, + .CopySubBuffer = nouveau_copy_sub_buffer, + .setTexOffset = NULL +}; + +static __GLcontextModes * +nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, + unsigned stencil_bits, GLboolean have_back_buffer) +{ + __GLcontextModes * modes; + __GLcontextModes * m; + unsigned num_modes; + unsigned depth_buffer_factor; + unsigned back_buffer_factor; + int i; + + static const struct { + GLenum format; + GLenum type; + } fb_format_array[] = { + { GL_RGB , GL_UNSIGNED_SHORT_5_6_5 }, + { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV }, + { GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV }, + }; + + /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't + * support pageflipping at all. + */ + static const GLenum back_buffer_modes[] = { + GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML + }; + + uint8_t depth_bits_array[4] = { 0, 16, 24, 24 }; + uint8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; + uint8_t msaa_samples_array[1] = { 0 }; + + depth_buffer_factor = 4; + back_buffer_factor = (have_back_buffer) ? 3 : 1; + + num_modes = ((pixel_bits==16) ? 1 : 2) * + depth_buffer_factor * back_buffer_factor * 4; + modes = (*dri_interface->createContextModes)(num_modes, + sizeof(__GLcontextModes)); + m = modes; + + for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) { + if (!driFillInModes(&m, fb_format_array[i].format, + fb_format_array[i].type, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + msaa_samples_array, 1, + GLX_TRUE_COLOR)) { + fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__ ); + return NULL; + } + + if (!driFillInModes(&m, fb_format_array[i].format, + fb_format_array[i].type, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + msaa_samples_array, 1, + GLX_DIRECT_COLOR)) { + fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__ ); + return NULL; + } + } + + return modes; +} +PUBLIC void * +__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, + __DRIscreen *psc, const __GLcontextModes * modes, + const __DRIversion * ddx_version, + const __DRIversion * dri_version, + const __DRIversion * drm_version, + const __DRIframebuffer * frame_buffer, + void * pSAREA, int fd, int internal_api_version, + const __DRIinterfaceMethods * interface, + __GLcontextModes ** driver_modes) +{ + __DRIscreenPrivate *psp; + static const __DRIversion ddx_expected = + { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; + static const __DRIversion dri_expected = { 4, 0, 0 }; + static const __DRIversion drm_expected = + { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; + struct nouveau_dri *nv_dri = NULL; + + dri_interface = interface; + + if (!driCheckDriDdxDrmVersions2("nouveau", + dri_version, &dri_expected, + ddx_version, &ddx_expected, + drm_version, &drm_expected)) { + return NULL; + } + + if (drm_expected.patch != drm_version->patch) { + fprintf(stderr, "Incompatible DRM patch level.\n" + "Expected: %d\n" "Current : %d\n", + drm_expected.patch, drm_version->patch); + return NULL; + } + + psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, + ddx_version, dri_version, drm_version, + frame_buffer, pSAREA, fd, + internal_api_version, + &nouveau_api); + if (psp == NULL) + return NULL; + nv_dri = psp->pDevPriv; + + *driver_modes = nouveau_fill_in_modes(nv_dri->bpp, + (nv_dri->bpp == 16) ? 16 : 24, + (nv_dri->bpp == 16) ? 0 : 8, + 1); + + driInitExtensions(NULL, card_extensions, GL_FALSE); + + return (void *)psp; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_screen.h b/src/gallium/winsys/drm/nouveau/nouveau_screen.h new file mode 100644 index 00000000000..388d6be9bbc --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_screen.h @@ -0,0 +1,20 @@ +#ifndef __NOUVEAU_SCREEN_H__ +#define __NOUVEAU_SCREEN_H__ + +#include "xmlconfig.h" + +struct nouveau_screen { + __DRIscreenPrivate *driScrnPriv; + driOptionCache option_cache; + + struct nouveau_device *device; + + uint32_t front_offset; + uint32_t front_pitch; + uint32_t front_cpp; + uint32_t front_height; + + void *nvc; +}; + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c new file mode 100644 index 00000000000..70e0104e83b --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c @@ -0,0 +1,86 @@ +#include "main/glheader.h" +#include "glapi/glthread.h" +#include + +#include "pipe/p_context.h" +#include "state_tracker/st_public.h" +#include "state_tracker/st_context.h" +#include "state_tracker/st_cb_fbo.h" + +#include "nouveau_context.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" + +void +nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf, + const drm_clip_rect_t *rect) +{ + struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate; + drm_clip_rect_t *pbox; + int nbox, i; + + LOCK_HARDWARE(nv); + if (!dPriv->numClipRects) { + UNLOCK_HARDWARE(nv); + return; + } + pbox = dPriv->pClipRects; + nbox = dPriv->numClipRects; + + nv->surface_copy_prep(nv, nv->frontbuffer, surf); + for (i = 0; i < nbox; i++, pbox++) { + int sx, sy, dx, dy, w, h; + + sx = pbox->x1 - dPriv->x; + sy = pbox->y1 - dPriv->y; + dx = pbox->x1; + dy = pbox->y1; + w = pbox->x2 - pbox->x1; + h = pbox->y2 - pbox->y1; + + nv->surface_copy(nv, dx, dy, sx, sy, w, h); + } + + FIRE_RING(nv->nvc->channel); + UNLOCK_HARDWARE(nv); + + if (nv->last_stamp != dPriv->lastStamp) { + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + st_resize_framebuffer(nvfb->stfb, dPriv->w, dPriv->h); + nv->last_stamp = dPriv->lastStamp; + } +} + +void +nouveau_copy_sub_buffer(__DRIdrawablePrivate *dPriv, int x, int y, int w, int h) +{ + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + struct pipe_surface *surf; + + surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + drm_clip_rect_t rect; + rect.x1 = x; + rect.y1 = y; + rect.x2 = x + w; + rect.y2 = y + h; + + st_notify_swapbuffers(nvfb->stfb); + nouveau_copy_buffer(dPriv, surf, &rect); + } +} + +void +nouveau_swap_buffers(__DRIdrawablePrivate *dPriv) +{ + struct nouveau_framebuffer *nvfb = dPriv->driverPrivate; + struct pipe_surface *surf; + + surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT); + if (surf) { + st_notify_swapbuffers(nvfb->stfb); + nouveau_copy_buffer(dPriv, surf, NULL); + } +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h new file mode 100644 index 00000000000..825d3da6da5 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h @@ -0,0 +1,10 @@ +#ifndef __NOUVEAU_SWAPBUFFERS_H__ +#define __NOUVEAU_SWAPBUFFERS_H__ + +extern void nouveau_copy_buffer(__DRIdrawablePrivate *, struct pipe_surface *, + const drm_clip_rect_t *); +extern void nouveau_copy_sub_buffer(__DRIdrawablePrivate *, + int x, int y, int w, int h); +extern void nouveau_swap_buffers(__DRIdrawablePrivate *); + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys.c new file mode 100644 index 00000000000..0878840dcc0 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys.c @@ -0,0 +1,158 @@ +#include "util/u_memory.h" + +#include "nouveau_context.h" +#include "nouveau_screen.h" +#include "nouveau_winsys_pipe.h" + +#include "nouveau/nouveau_winsys.h" + +static int +nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count, + struct nouveau_notifier **notify) +{ + struct nouveau_context *nv = nvws->nv; + + return nouveau_notifier_alloc(nv->nvc->channel, nv->nvc->next_handle++, + count, notify); +} + +static int +nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass, + struct nouveau_grobj **grobj) +{ + struct nouveau_context *nv = nvws->nv; + struct nouveau_channel *chan = nv->nvc->channel; + int ret; + + ret = nouveau_grobj_alloc(chan, nv->nvc->next_handle++, + grclass, grobj); + if (ret) + return ret; + + assert(nv->nvc->next_subchannel < 7); + BIND_RING(chan, *grobj, nv->nvc->next_subchannel++); + return 0; +} + +static int +nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, struct pipe_surface *src, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct nouveau_context *nv = nvws->nv; + + if (nv->surface_copy_prep(nv, dst, src)) + return 1; + nv->surface_copy(nv, dx, dy, sx, sy, w, h); + nv->surface_copy_done(nv); + + return 0; +} + +static int +nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + if (nvws->nv->surface_fill(nvws->nv, dst, dx, dy, w, h, value)) + return 1; + return 0; +} + +static int +nouveau_pipe_push_reloc(struct nouveau_winsys *nvws, void *ptr, + struct pipe_buffer *buf, uint32_t data, + uint32_t flags, uint32_t vor, uint32_t tor) +{ + return nouveau_pushbuf_emit_reloc(nvws->channel, ptr, + nouveau_buffer(buf)->bo, + data, flags, vor, tor); +} + +static int +nouveau_pipe_push_flush(struct nouveau_winsys *nvws, unsigned size, + struct pipe_fence_handle **fence) +{ + if (fence) { + struct nouveau_pushbuf *pb = nvws->channel->pushbuf; + struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(pb); + struct nouveau_fence *ref = NULL; + + nouveau_fence_ref(nvpb->fence, &ref); + *fence = (struct pipe_fence_handle *)ref; + } + + return nouveau_pushbuf_flush(nvws->channel, size); +} + +struct pipe_context * +nouveau_pipe_create(struct nouveau_context *nv) +{ + struct nouveau_channel_context *nvc = nv->nvc; + struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys); + struct pipe_screen *(*hws_create)(struct pipe_winsys *, + struct nouveau_winsys *); + struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned); + struct pipe_winsys *ws; + unsigned chipset = nv->nv_screen->device->chipset; + + if (!nvws) + return NULL; + + switch (chipset & 0xf0) { + case 0x10: + case 0x20: + hws_create = nv10_screen_create; + hw_create = nv10_create; + break; + case 0x30: + hws_create = nv30_screen_create; + hw_create = nv30_create; + break; + case 0x40: + case 0x60: + hws_create = nv40_screen_create; + hw_create = nv40_create; + break; + case 0x50: + case 0x80: + case 0x90: + hws_create = nv50_screen_create; + hw_create = nv50_create; + break; + default: + NOUVEAU_ERR("Unknown chipset NV%02x\n", chipset); + return NULL; + } + + nvws->nv = nv; + nvws->channel = nv->nvc->channel; + + nvws->res_init = nouveau_resource_init; + nvws->res_alloc = nouveau_resource_alloc; + nvws->res_free = nouveau_resource_free; + + nvws->push_reloc = nouveau_pipe_push_reloc; + nvws->push_flush = nouveau_pipe_push_flush; + + nvws->grobj_alloc = nouveau_pipe_grobj_alloc; + nvws->grobj_free = nouveau_grobj_free; + + nvws->notifier_alloc = nouveau_pipe_notifier_alloc; + nvws->notifier_free = nouveau_notifier_free; + nvws->notifier_reset = nouveau_notifier_reset; + nvws->notifier_status = nouveau_notifier_status; + nvws->notifier_retval = nouveau_notifier_return_val; + nvws->notifier_wait = nouveau_notifier_wait_status; + + nvws->surface_copy = nouveau_pipe_surface_copy; + nvws->surface_fill = nouveau_pipe_surface_fill; + + ws = nouveau_create_pipe_winsys(nv); + + if (!nvc->pscreen) + nvc->pscreen = hws_create(ws, nvws); + nvc->pctx[nv->pctx_id] = hw_create(nvc->pscreen, nv->pctx_id); + return nvc->pctx[nv->pctx_id]; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c new file mode 100644 index 00000000000..5276806de6b --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c @@ -0,0 +1,206 @@ +#include "pipe/p_winsys.h" +#include "pipe/p_defines.h" +#include "pipe/p_inlines.h" + +#include "util/u_memory.h" + +#include "nouveau_context.h" +#include "nouveau_local.h" +#include "nouveau_screen.h" +#include "nouveau_swapbuffers.h" +#include "nouveau_winsys_pipe.h" + +static void +nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf, + void *context_private) +{ + struct nouveau_context *nv = context_private; + __DRIdrawablePrivate *dPriv = nv->dri_drawable; + + nouveau_copy_buffer(dPriv, surf, NULL); +} + +static const char * +nouveau_get_name(struct pipe_winsys *pws) +{ + return "Nouveau/DRI"; +} + +static struct pipe_buffer * +nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment, + unsigned usage, unsigned size) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_context *nv = nvpws->nv; + struct nouveau_device *dev = nv->nv_screen->device; + struct nouveau_pipe_buffer *nvbuf; + uint32_t flags; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.alignment = alignment; + nvbuf->base.usage = usage; + nvbuf->base.size = size; + + flags = NOUVEAU_BO_LOCAL; + + if (usage & PIPE_BUFFER_USAGE_PIXEL) { + if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE) + flags |= NOUVEAU_BO_GART; + flags |= NOUVEAU_BO_VRAM; + + switch (dev->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + flags |= NOUVEAU_BO_TILED; + if (usage & NOUVEAU_BUFFER_USAGE_ZETA) + flags |= NOUVEAU_BO_ZTILE; + break; + default: + break; + } + } + + if (usage & PIPE_BUFFER_USAGE_VERTEX) { + if (nv->cap.hw_vertex_buffer) + flags |= NOUVEAU_BO_GART; + } + + if (usage & PIPE_BUFFER_USAGE_INDEX) { + if (nv->cap.hw_index_buffer) + flags |= NOUVEAU_BO_GART; + } + + if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) { + free(nvbuf); + return NULL; + } + + return &nvbuf->base; +} + +static struct pipe_buffer * +nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws; + struct nouveau_device *dev = nvpws->nv->nv_screen->device; + struct nouveau_pipe_buffer *nvbuf; + + nvbuf = calloc(1, sizeof(*nvbuf)); + if (!nvbuf) + return NULL; + nvbuf->base.refcount = 1; + nvbuf->base.size = bytes; + + if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) { + free(nvbuf); + return NULL; + } + + return &nvbuf->base; +} + +static void +nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + + nouveau_bo_del(&nvbuf->bo); + free(nvbuf); +} + +static void * +nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf, + unsigned flags) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + uint32_t map_flags = 0; + + if (flags & PIPE_BUFFER_USAGE_CPU_READ) + map_flags |= NOUVEAU_BO_RD; + if (flags & PIPE_BUFFER_USAGE_CPU_WRITE) + map_flags |= NOUVEAU_BO_WR; + + if (nouveau_bo_map(nvbuf->bo, map_flags)) + return NULL; + return nvbuf->bo->map; +} + +static void +nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf) +{ + struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf); + + nouveau_bo_unmap(nvbuf->bo); +} + +static INLINE struct nouveau_fence * +nouveau_pipe_fence(struct pipe_fence_handle *pfence) +{ + return (struct nouveau_fence *)pfence; +} + +static void +nouveau_pipe_fence_reference(struct pipe_winsys *ws, + struct pipe_fence_handle **ptr, + struct pipe_fence_handle *pfence) +{ + nouveau_fence_ref((void *)pfence, (void *)ptr); +} + +static int +nouveau_pipe_fence_signalled(struct pipe_winsys *ws, + struct pipe_fence_handle *pfence, unsigned flag) +{ + struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)ws; + struct nouveau_fence *fence = nouveau_pipe_fence(pfence); + + if (nouveau_fence(fence)->signalled == 0) + nouveau_fence_flush(nvpws->nv->nvc->channel); + + return !nouveau_fence(fence)->signalled; +} + +static int +nouveau_pipe_fence_finish(struct pipe_winsys *ws, + struct pipe_fence_handle *pfence, unsigned flag) +{ + struct nouveau_fence *fence = nouveau_pipe_fence(pfence); + struct nouveau_fence *ref = NULL; + + nouveau_fence_ref(fence, &ref); + return nouveau_fence_wait(&ref); +} + +struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv) +{ + struct nouveau_pipe_winsys *nvpws; + struct pipe_winsys *pws; + + nvpws = CALLOC_STRUCT(nouveau_pipe_winsys); + if (!nvpws) + return NULL; + nvpws->nv = nv; + pws = &nvpws->pws; + + pws->flush_frontbuffer = nouveau_flush_frontbuffer; + + pws->buffer_create = nouveau_pipe_bo_create; + pws->buffer_destroy = nouveau_pipe_bo_del; + pws->user_buffer_create = nouveau_pipe_bo_user_create; + pws->buffer_map = nouveau_pipe_bo_map; + pws->buffer_unmap = nouveau_pipe_bo_unmap; + + pws->fence_reference = nouveau_pipe_fence_reference; + pws->fence_signalled = nouveau_pipe_fence_signalled; + pws->fence_finish = nouveau_pipe_fence_finish; + + pws->get_name = nouveau_get_name; + + return &nvpws->pws; +} + diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h new file mode 100644 index 00000000000..6a03ac0d773 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h @@ -0,0 +1,34 @@ +#ifndef NOUVEAU_PIPE_WINSYS_H +#define NOUVEAU_PIPE_WINSYS_H + +#include "pipe/p_context.h" +#include "pipe/p_winsys.h" +#include "nouveau_context.h" + +struct nouveau_pipe_buffer { + struct pipe_buffer base; + struct nouveau_bo *bo; +}; + +static inline struct nouveau_pipe_buffer * +nouveau_buffer(struct pipe_buffer *buf) +{ + return (struct nouveau_pipe_buffer *)buf; +} + +struct nouveau_pipe_winsys { + struct pipe_winsys pws; + + struct nouveau_context *nv; +}; + +extern struct pipe_winsys * +nouveau_create_pipe_winsys(struct nouveau_context *nv); + +struct pipe_context * +nouveau_create_softpipe(struct nouveau_context *nv); + +struct pipe_context * +nouveau_pipe_create(struct nouveau_context *nv); + +#endif diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c new file mode 100644 index 00000000000..704f6c77506 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c @@ -0,0 +1,85 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * + **************************************************************************/ +/* + * Authors: Keith Whitwell + */ + +#include "imports.h" + +#include "pipe/p_defines.h" +#include "pipe/p_format.h" +#include "softpipe/sp_winsys.h" + +#include "nouveau_context.h" +#include "nouveau_winsys_pipe.h" + +struct nouveau_softpipe_winsys { + struct softpipe_winsys sws; + struct nouveau_context *nv; +}; + +/** + * Return list of surface formats supported by this driver. + */ +static boolean +nouveau_is_format_supported(struct softpipe_winsys *sws, uint format) +{ + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return TRUE; + default: + break; + }; + + return FALSE; +} + +struct pipe_context * +nouveau_create_softpipe(struct nouveau_context *nv) +{ + struct nouveau_softpipe_winsys *nvsws; + struct pipe_screen *pscreen; + struct pipe_winsys *ws; + + ws = nouveau_create_pipe_winsys(nv); + if (!ws) + return NULL; + pscreen = softpipe_create_screen(ws); + + nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys); + if (!nvsws) + return NULL; + + nvsws->sws.is_format_supported = nouveau_is_format_supported; + nvsws->nv = nv; + + return softpipe_create(pscreen, ws, &nvsws->sws); +} + diff --git a/src/gallium/winsys/drm/nouveau/nv04_surface.c b/src/gallium/winsys/drm/nouveau/nv04_surface.c new file mode 100644 index 00000000000..8fa3d106c8c --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nv04_surface.c @@ -0,0 +1,314 @@ +#include "pipe/p_context.h" +#include "pipe/p_format.h" + +#include "nouveau_context.h" + +static INLINE int +nv04_surface_format(enum pipe_format format) +{ + switch (format) { + case PIPE_FORMAT_A8_UNORM: + return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8; + case PIPE_FORMAT_R5G6B5_UNORM: + return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5; + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32; + default: + return -1; + } +} + +static INLINE int +nv04_rect_format(enum pipe_format format) +{ + switch (format) { + case PIPE_FORMAT_A8_UNORM: + return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + case PIPE_FORMAT_R5G6B5_UNORM: + return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5; + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8; + default: + return -1; + } +} + +static void +nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct nouveau_channel *chan = nv->nvc->channel; + struct pipe_surface *dst = nv->surf_dst; + struct pipe_surface *src = nv->surf_src; + unsigned dst_offset, src_offset; + + dst_offset = dst->offset + (dy * dst->stride) + (dx * dst->block.size); + src_offset = src->offset + (sy * src->stride) + (sx * src->block.size); + + while (h) { + int count = (h > 2047) ? 2047 : h; + + BEGIN_RING(chan, nv->nvc->NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src_offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst_offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); + OUT_RING (chan, src->stride); + OUT_RING (chan, dst->stride); + OUT_RING (chan, w * src->block.size); + OUT_RING (chan, count); + OUT_RING (chan, 0x0101); + OUT_RING (chan, 0); + + h -= count; + src_offset += src->stride * count; + dst_offset += dst->stride * count; + } +} + +static void +nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct nouveau_channel *chan = nv->nvc->channel; + + BEGIN_RING(chan, nv->nvc->NvImageBlit, 0x0300, 3); + OUT_RING (chan, (sy << 16) | sx); + OUT_RING (chan, (dy << 16) | dx); + OUT_RING (chan, ( h << 16) | w); +} + +static int +nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, + struct pipe_surface *src) +{ + struct nouveau_channel *chan = nv->nvc->channel; + int format; + + if (src->format != dst->format) + return 1; + + /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback + * to NV_MEMORY_TO_MEMORY_FORMAT in this case. + */ + if ((src->offset & 63) || (dst->offset & 63)) { + BEGIN_RING(nv->nvc->channel, nv->nvc->NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + nv->surface_copy = nv04_surface_copy_m2mf; + nv->surf_dst = dst; + nv->surf_src = src; + return 0; + + } + + if ((format = nv04_surface_format(dst->format)) < 0) { + NOUVEAU_ERR("Bad surface format 0x%x\n", dst->format); + return 1; + } + nv->surface_copy = nv04_surface_copy_blit; + + BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, + NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(chan, nv->nvc->NvCtxSurf2D, + NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (chan, format); + OUT_RING (chan, (dst->stride << 16) | src->stride); + OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + return 0; +} + +static void +nv04_surface_copy_done(struct nouveau_context *nv) +{ + FIRE_RING(nv->nvc->channel); +} + +static int +nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *surf2d = nv->nvc->NvCtxSurf2D; + struct nouveau_grobj *rect = nv->nvc->NvGdiRect; + int cs2d_format, gdirect_format; + + if ((cs2d_format = nv04_surface_format(dst->format)) < 0) { + NOUVEAU_ERR("Bad format = %d\n", dst->format); + return 1; + } + + if ((gdirect_format = nv04_rect_format(dst->format)) < 0) { + NOUVEAU_ERR("Bad format = %d\n", dst->format); + return 1; + } + + BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4); + OUT_RING (chan, cs2d_format); + OUT_RING (chan, (dst->stride << 16) | dst->stride); + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset, + NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1); + OUT_RING (chan, gdirect_format); + BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1); + OUT_RING (chan, value); + BEGIN_RING(chan, rect, + NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2); + OUT_RING (chan, (dx << 16) | dy); + OUT_RING (chan, ( w << 16) | h); + + FIRE_RING(chan); + return 0; +} + +int +nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) +{ + struct nouveau_channel *chan = nvc->channel; + unsigned chipset = nvc->channel->device->chipset, class; + int ret; + + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, 0x39, + &nvc->NvM2MF))) { + NOUVEAU_ERR("Error creating m2mf object: %d\n", ret); + return 1; + } + BIND_RING (chan, nvc->NvM2MF, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + + class = chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D : + NV10_CONTEXT_SURFACES_2D; + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvCtxSurf2D))) { + NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret); + return 1; + } + BIND_RING (chan, nvc->NvCtxSurf2D, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvCtxSurf2D, + NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); + OUT_RING (chan, nvc->channel->vram->handle); + OUT_RING (chan, nvc->channel->vram->handle); + + class = chipset < 0x10 ? NV04_IMAGE_BLIT : NV12_IMAGE_BLIT; + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvImageBlit))) { + NOUVEAU_ERR("Error creating blit object: %d\n", ret); + return 1; + } + BIND_RING (chan, nvc->NvImageBlit, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1); + OUT_RING (chan, nvc->NvCtxSurf2D->handle); + BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1); + OUT_RING (chan, NV04_IMAGE_BLIT_OPERATION_SRCCOPY); + + class = NV04_GDI_RECTANGLE_TEXT; + if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvGdiRect))) { + NOUVEAU_ERR("Error creating rect object: %d\n", ret); + return 1; + } + BIND_RING (chan, nvc->NvGdiRect, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1); + OUT_RING (chan, nvc->NvCtxSurf2D->handle); + BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1); + OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY); + BEGIN_RING(chan, nvc->NvGdiRect, + NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1); + OUT_RING (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE); + + switch (chipset & 0xf0) { + case 0x00: + case 0x10: + class = NV04_SWIZZLED_SURFACE; + break; + case 0x20: + class = NV20_SWIZZLED_SURFACE; + break; + case 0x30: + class = NV30_SWIZZLED_SURFACE; + break; + case 0x40: + case 0x60: + class = NV40_SWIZZLED_SURFACE; + break; + default: + /* Famous last words: this really can't happen.. */ + assert(0); + break; + } + + ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvSwzSurf); + if (ret) { + NOUVEAU_ERR("Error creating swizzled surface: %d\n", ret); + return 1; + } + + BIND_RING (chan, nvc->NvSwzSurf, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_NOTIFY, 1); + OUT_RING (chan, nvc->sync_notifier->handle); + BEGIN_RING(chan, nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1); + OUT_RING (chan, nvc->channel->vram->handle); + + if (chipset < 0x10) { + class = NV04_SCALED_IMAGE_FROM_MEMORY; + } else + if (chipset < 0x40) { + class = NV10_SCALED_IMAGE_FROM_MEMORY; + } else { + class = NV40_SCALED_IMAGE_FROM_MEMORY; + } + + ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class, + &nvc->NvSIFM); + if (ret) { + NOUVEAU_ERR("Error creating scaled image object: %d\n", ret); + return 1; + } + + BIND_RING (chan, nvc->NvSIFM, nvc->next_subchannel++); + + return 0; +} + +int +nouveau_surface_init_nv04(struct nouveau_context *nv) +{ + nv->surface_copy_prep = nv04_surface_copy_prep; + nv->surface_copy = nv04_surface_copy_blit; + nv->surface_copy_done = nv04_surface_copy_done; + nv->surface_fill = nv04_surface_fill; + return 0; +} + diff --git a/src/gallium/winsys/drm/nouveau/nv50_surface.c b/src/gallium/winsys/drm/nouveau/nv50_surface.c new file mode 100644 index 00000000000..c8ab7f690f3 --- /dev/null +++ b/src/gallium/winsys/drm/nouveau/nv50_surface.c @@ -0,0 +1,194 @@ +#include "pipe/p_context.h" +#include "pipe/p_format.h" + +#include "nouveau_context.h" + +static INLINE int +nv50_format(enum pipe_format format) +{ + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + return NV50_2D_DST_FORMAT_32BPP; + case PIPE_FORMAT_X8R8G8B8_UNORM: + return NV50_2D_DST_FORMAT_24BPP; + case PIPE_FORMAT_R5G6B5_UNORM: + return NV50_2D_DST_FORMAT_16BPP; + case PIPE_FORMAT_A8_UNORM: + return NV50_2D_DST_FORMAT_8BPP; + default: + return -1; + } +} + +static int +nv50_surface_set(struct nouveau_context *nv, struct pipe_surface *surf, int dst) +{ + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *eng2d = nv->nvc->Nv2D; + struct nouveau_bo *bo = nouveau_buffer(surf->buffer)->bo; + int surf_format, mthd = dst ? NV50_2D_DST_FORMAT : NV50_2D_SRC_FORMAT; + int flags = NOUVEAU_BO_VRAM | (dst ? NOUVEAU_BO_WR : NOUVEAU_BO_RD); + + surf_format = nv50_format(surf->format); + if (surf_format < 0) + return 1; + + if (!nouveau_bo(bo)->tiled) { + BEGIN_RING(chan, eng2d, mthd, 2); + OUT_RING (chan, surf_format); + OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, mthd + 0x14, 5); + OUT_RING (chan, surf->stride); + OUT_RING (chan, surf->width); + OUT_RING (chan, surf->height); + OUT_RELOCh(chan, bo, surf->offset, flags); + OUT_RELOCl(chan, bo, surf->offset, flags); + } else { + BEGIN_RING(chan, eng2d, mthd, 5); + OUT_RING (chan, surf_format); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, eng2d, mthd + 0x18, 4); + OUT_RING (chan, surf->width); + OUT_RING (chan, surf->height); + OUT_RELOCh(chan, bo, surf->offset, flags); + OUT_RELOCl(chan, bo, surf->offset, flags); + } + +#if 0 + if (dst) { + BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, surf->width); + OUT_RING (chan, surf->height); + } +#endif + + return 0; +} + +static int +nv50_surface_copy_prep(struct nouveau_context *nv, + struct pipe_surface *dst, struct pipe_surface *src) +{ + int ret; + + assert(src->format == dst->format); + + ret = nv50_surface_set(nv, dst, 1); + if (ret) + return ret; + + ret = nv50_surface_set(nv, src, 0); + if (ret) + return ret; + + return 0; +} + +static void +nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy, + unsigned sx, unsigned sy, unsigned w, unsigned h) +{ + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *eng2d = nv->nvc->Nv2D; + + BEGIN_RING(chan, eng2d, 0x088c, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, eng2d, NV50_2D_BLIT_DST_X, 4); + OUT_RING (chan, dx); + OUT_RING (chan, dy); + OUT_RING (chan, w); + OUT_RING (chan, h); + BEGIN_RING(chan, eng2d, 0x08c0, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + BEGIN_RING(chan, eng2d, 0x08d0, 4); + OUT_RING (chan, 0); + OUT_RING (chan, sx); + OUT_RING (chan, 0); + OUT_RING (chan, sy); +} + +static void +nv50_surface_copy_done(struct nouveau_context *nv) +{ + FIRE_RING(nv->nvc->channel); +} + +static int +nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst, + unsigned dx, unsigned dy, unsigned w, unsigned h, + unsigned value) +{ + struct nouveau_channel *chan = nv->nvc->channel; + struct nouveau_grobj *eng2d = nv->nvc->Nv2D; + int rect_format, ret; + + rect_format = nv50_format(dst->format); + if (rect_format < 0) + return 1; + + ret = nv50_surface_set(nv, dst, 1); + if (ret) + return ret; + + BEGIN_RING(chan, eng2d, 0x0580, 3); + OUT_RING (chan, 4); + OUT_RING (chan, rect_format); + OUT_RING (chan, value); + + BEGIN_RING(chan, eng2d, NV50_2D_RECT_X1, 4); + OUT_RING (chan, dx); + OUT_RING (chan, dy); + OUT_RING (chan, dx + w); + OUT_RING (chan, dy + h); + + FIRE_RING(chan); + return 0; +} + +int +nouveau_surface_channel_create_nv50(struct nouveau_channel_context *nvc) +{ + struct nouveau_channel *chan = nvc->channel; + struct nouveau_grobj *eng2d = NULL; + int ret; + + ret = nouveau_grobj_alloc(chan, nvc->next_handle++, NV50_2D, &eng2d); + if (ret) + return ret; + nvc->Nv2D = eng2d; + + BIND_RING (chan, eng2d, nvc->next_subchannel++); + BEGIN_RING(chan, eng2d, NV50_2D_DMA_NOTIFY, 4); + OUT_RING (chan, nvc->sync_notifier->handle); + OUT_RING (chan, chan->vram->handle); + OUT_RING (chan, chan->vram->handle); + OUT_RING (chan, chan->vram->handle); + BEGIN_RING(chan, eng2d, NV50_2D_OPERATION, 1); + OUT_RING (chan, NV50_2D_OPERATION_SRCCOPY); + BEGIN_RING(chan, eng2d, 0x0290, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, eng2d, 0x0888, 1); + OUT_RING (chan, 1); + + return 0; +} + +int +nouveau_surface_init_nv50(struct nouveau_context *nv) +{ + nv->surface_copy_prep = nv50_surface_copy_prep; + nv->surface_copy = nv50_surface_copy; + nv->surface_copy_done = nv50_surface_copy_done; + nv->surface_fill = nv50_surface_fill; + return 0; +} + -- cgit v1.2.3 From d64da83a2c21d17f2f0a6b9b4c8e90df0e993559 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Fri, 12 Sep 2008 23:51:39 +1000 Subject: nouveau: rework nouveau_screen for latest and greatest changes --- src/gallium/winsys/drm/nouveau/nouveau_screen.c | 267 ++++++++++-------------- 1 file changed, 110 insertions(+), 157 deletions(-) diff --git a/src/gallium/winsys/drm/nouveau/nouveau_screen.c b/src/gallium/winsys/drm/nouveau/nouveau_screen.c index df1fe7e69b4..94896451519 100644 --- a/src/gallium/winsys/drm/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/drm/nouveau/nouveau_screen.c @@ -64,50 +64,6 @@ static const GLuint __driNConfigOptions = 0; extern const struct dri_extension common_extensions[]; extern const struct dri_extension nv40_extensions[]; -static GLboolean -nouveau_screen_create(__DRIscreenPrivate *driScrnPriv) -{ - struct nouveau_dri *nv_dri = driScrnPriv->pDevPriv; - struct nouveau_screen *nv_screen; - int ret; - - if (driScrnPriv->devPrivSize != sizeof(struct nouveau_dri)) { - NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); - return GL_FALSE; - } - - nv_screen = CALLOC_STRUCT(nouveau_screen); - if (!nv_screen) - return GL_FALSE; - nv_screen->driScrnPriv = driScrnPriv; - driScrnPriv->private = (void *)nv_screen; - - driParseOptionInfo(&nv_screen->option_cache, - __driConfigOptions, __driNConfigOptions); - - if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, - driScrnPriv->fd, 0))) { - NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); - return GL_FALSE; - } - - nv_screen->front_offset = nv_dri->front_offset; - nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); - nv_screen->front_cpp = nv_dri->bpp / 8; - nv_screen->front_height = nv_dri->height; - - return GL_TRUE; -} - -static void -nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) -{ - struct nouveau_screen *nv_screen = driScrnPriv->private; - - driScrnPriv->private = NULL; - FREE(nv_screen); -} - static GLboolean nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv, __DRIdrawablePrivate * driDrawPriv, @@ -162,149 +118,146 @@ nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) free(nvfb); } -static struct __DriverAPIRec -nouveau_api = { - .InitDriver = nouveau_screen_create, - .DestroyScreen = nouveau_screen_destroy, - .CreateContext = nouveau_context_create, - .DestroyContext = nouveau_context_destroy, - .CreateBuffer = nouveau_create_buffer, - .DestroyBuffer = nouveau_destroy_buffer, - .SwapBuffers = nouveau_swap_buffers, - .MakeCurrent = nouveau_context_bind, - .UnbindContext = nouveau_context_unbind, - .GetSwapInfo = NULL, - .GetMSC = NULL, - .WaitForMSC = NULL, - .WaitForSBC = NULL, - .SwapBuffersMSC = NULL, - .CopySubBuffer = nouveau_copy_sub_buffer, - .setTexOffset = NULL -}; - -static __GLcontextModes * -nouveau_fill_in_modes(unsigned pixel_bits, unsigned depth_bits, +static __DRIconfig ** +nouveau_fill_in_modes(__DRIscreenPrivate *psp, + unsigned pixel_bits, unsigned depth_bits, unsigned stencil_bits, GLboolean have_back_buffer) { - __GLcontextModes * modes; - __GLcontextModes * m; - unsigned num_modes; + __DRIconfig **configs; unsigned depth_buffer_factor; unsigned back_buffer_factor; - int i; - - static const struct { - GLenum format; - GLenum type; - } fb_format_array[] = { - { GL_RGB , GL_UNSIGNED_SHORT_5_6_5 }, - { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV }, - { GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV }, - }; + GLenum fb_format; + GLenum fb_type; - /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't - * support pageflipping at all. - */ static const GLenum back_buffer_modes[] = { - GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML + GLX_NONE, GLX_SWAP_UNDEFINED_OML, }; - uint8_t depth_bits_array[4] = { 0, 16, 24, 24 }; - uint8_t stencil_bits_array[4] = { 0, 0, 0, 8 }; - uint8_t msaa_samples_array[1] = { 0 }; - - depth_buffer_factor = 4; - back_buffer_factor = (have_back_buffer) ? 3 : 1; - - num_modes = ((pixel_bits==16) ? 1 : 2) * - depth_buffer_factor * back_buffer_factor * 4; - modes = (*dri_interface->createContextModes)(num_modes, - sizeof(__GLcontextModes)); - m = modes; - - for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) { - if (!driFillInModes(&m, fb_format_array[i].format, - fb_format_array[i].type, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - msaa_samples_array, 1, - GLX_TRUE_COLOR)) { - fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", - __func__, __LINE__ ); - return NULL; - } - - if (!driFillInModes(&m, fb_format_array[i].format, - fb_format_array[i].type, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - msaa_samples_array, 1, - GLX_DIRECT_COLOR)) { - fprintf( stderr, "[%s:%u] Error creating FBConfig!\n", - __func__, __LINE__ ); + u_int8_t depth_bits_array[3]; + u_int8_t stencil_bits_array[3]; + + depth_bits_array[0] = 0; + depth_bits_array[1] = depth_bits; + depth_bits_array[2] = depth_bits; + + /* Just like with the accumulation buffer, always provide some modes + * with a stencil buffer. It will be a sw fallback, but some apps won't + * care about that. + */ + stencil_bits_array[0] = 0; + stencil_bits_array[1] = 0; + if (depth_bits == 24) + stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits; + + stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits; + + depth_buffer_factor = + ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1; + back_buffer_factor = (have_back_buffer) ? 3 : 1; + + if (pixel_bits == 16) { + fb_format = GL_RGB; + fb_type = GL_UNSIGNED_SHORT_5_6_5; + } + else { + fb_format = GL_BGRA; + fb_type = GL_UNSIGNED_INT_8_8_8_8_REV; + } + + configs = driCreateConfigs(fb_format, fb_type, + depth_bits_array, stencil_bits_array, + depth_buffer_factor, back_buffer_modes, + back_buffer_factor); + if (configs == NULL) { + fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", + __func__, __LINE__); return NULL; - } } - return modes; + return configs; } -PUBLIC void * -__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn, - __DRIscreen *psc, const __GLcontextModes * modes, - const __DRIversion * ddx_version, - const __DRIversion * dri_version, - const __DRIversion * drm_version, - const __DRIframebuffer * frame_buffer, - void * pSAREA, int fd, int internal_api_version, - const __DRIinterfaceMethods * interface, - __GLcontextModes ** driver_modes) + +static const __DRIconfig ** +nouveau_screen_create(__DRIscreenPrivate *psp) { - __DRIscreenPrivate *psp; + struct nouveau_dri *nv_dri = psp->pDevPriv; + struct nouveau_screen *nv_screen; static const __DRIversion ddx_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; - struct nouveau_dri *nv_dri = NULL; - - dri_interface = interface; + int ret; if (!driCheckDriDdxDrmVersions2("nouveau", - dri_version, &dri_expected, - ddx_version, &ddx_expected, - drm_version, &drm_expected)) { + &psp->dri_version, &dri_expected, + &psp->ddx_version, &ddx_expected, + &psp->drm_version, &drm_expected)) { return NULL; } - if (drm_expected.patch != drm_version->patch) { + if (drm_expected.patch != psp->drm_version.patch) { fprintf(stderr, "Incompatible DRM patch level.\n" "Expected: %d\n" "Current : %d\n", - drm_expected.patch, drm_version->patch); + drm_expected.patch, psp->drm_version.patch); return NULL; } - psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, - ddx_version, dri_version, drm_version, - frame_buffer, pSAREA, fd, - internal_api_version, - &nouveau_api); - if (psp == NULL) - return NULL; - nv_dri = psp->pDevPriv; + driInitExtensions(NULL, card_extensions, GL_FALSE); - *driver_modes = nouveau_fill_in_modes(nv_dri->bpp, - (nv_dri->bpp == 16) ? 16 : 24, - (nv_dri->bpp == 16) ? 0 : 8, - 1); + if (psp->devPrivSize != sizeof(struct nouveau_dri)) { + NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n"); + return GL_FALSE; + } - driInitExtensions(NULL, card_extensions, GL_FALSE); + nv_screen = CALLOC_STRUCT(nouveau_screen); + if (!nv_screen) + return GL_FALSE; + nv_screen->driScrnPriv = psp; + psp->private = (void *)nv_screen; + + driParseOptionInfo(&nv_screen->option_cache, + __driConfigOptions, __driNConfigOptions); + + if ((ret = nouveau_device_open_existing(&nv_screen->device, 0, + psp->fd, 0))) { + NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret); + return GL_FALSE; + } - return (void *)psp; + nv_screen->front_offset = nv_dri->front_offset; + nv_screen->front_pitch = nv_dri->front_pitch * (nv_dri->bpp / 8); + nv_screen->front_cpp = nv_dri->bpp / 8; + nv_screen->front_height = nv_dri->height; + + return (const __DRIconfig **) + nouveau_fill_in_modes(psp, nv_dri->bpp, + (nv_dri->bpp == 16) ? 16 : 24, + (nv_dri->bpp == 16) ? 0 : 8, 1); } +static void +nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv) +{ + struct nouveau_screen *nv_screen = driScrnPriv->private; + + driScrnPriv->private = NULL; + FREE(nv_screen); +} + +const struct __DriverAPIRec +driDriverAPI = { + .InitScreen = nouveau_screen_create, + .DestroyScreen = nouveau_screen_destroy, + .CreateContext = nouveau_context_create, + .DestroyContext = nouveau_context_destroy, + .CreateBuffer = nouveau_create_buffer, + .DestroyBuffer = nouveau_destroy_buffer, + .SwapBuffers = nouveau_swap_buffers, + .MakeCurrent = nouveau_context_bind, + .UnbindContext = nouveau_context_unbind, + .CopySubBuffer = nouveau_copy_sub_buffer, + + .InitScreen2 = NULL, /* one day, I promise! */ +}; + -- cgit v1.2.3 From 6b50fd27b85bc0d060ead13e6c656dd64c9a2978 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 8 Sep 2008 21:05:49 -0400 Subject: g3dvl: Clean up Makefile deps. --- src/driclient/src/Makefile | 8 +- src/driclient/src/test.c | 41 ---------- src/driclient/src/xf86dri.h | 119 ------------------------------ src/gallium/winsys/g3dvl/nouveau/Makefile | 9 ++- src/libXvMC/Makefile | 21 ++++-- 5 files changed, 22 insertions(+), 176 deletions(-) delete mode 100644 src/driclient/src/test.c delete mode 100644 src/driclient/src/xf86dri.h diff --git a/src/driclient/src/Makefile b/src/driclient/src/Makefile index 5d913ad5fa5..34435a2086e 100644 --- a/src/driclient/src/Makefile +++ b/src/driclient/src/Makefile @@ -8,16 +8,12 @@ CFLAGS += -g -Wall -fPIC -I../include -I${DRMDIR}/include -I${DRMDIR}/include/d .PHONY = all clean -all: ${TARGET} test +all: ${TARGET} ${TARGET}: ${OBJECTS} ar rcs $@ $^ if ! test -d ../lib; then mkdir ../lib; fi cp ${TARGET} ../lib -test: test.o - $(CC) -L../lib -L${DRMDIR}/lib ${LDFLAGS} -o $@ $^ -ldriclient -lX11 -lXext -ldrm - clean: - rm -rf ${OBJECTS} ${TARGET} ../lib/${TARGET} test test.o - + rm -rf ${OBJECTS} ${TARGET} ../lib/${TARGET} diff --git a/src/driclient/src/test.c b/src/driclient/src/test.c deleted file mode 100644 index 15f75d928b6..00000000000 --- a/src/driclient/src/test.c +++ /dev/null @@ -1,41 +0,0 @@ -#include -#include -#include "driclient.h" - -int main(int argc, char **argv) -{ - Display *dpy; - Window root, window; - - dri_screen_t *screen; - dri_drawable_t *dri_drawable; - dri_context_t *context; - - dpy = XOpenDisplay(NULL); - root = XDefaultRootWindow(dpy); - window = XCreateSimpleWindow(dpy, root, 0, 0, 100, 100, 0, 0, 0); - - XSelectInput(dpy, window, 0); - XMapWindow(dpy, window); - XSync(dpy, 0); - - assert(driCreateScreen(dpy, 0, &screen, NULL) == 0); - assert(driCreateDrawable(screen, window, &dri_drawable) == 0); - assert(driCreateContext(screen, XDefaultVisual(dpy, 0), &context) == 0); - assert(driUpdateDrawableInfo(dri_drawable) == 0); - - DRI_VALIDATE_DRAWABLE_INFO(screen, dri_drawable); - - assert(drmGetLock(screen->fd, context->drm_context, 0) == 0); - assert(drmUnlock(screen->fd, context->drm_context) == 0); - - assert(driDestroyContext(context) == 0); - assert(driDestroyDrawable(dri_drawable) == 0); - assert(driDestroyScreen(screen) == 0); - - XDestroyWindow(dpy, window); - XCloseDisplay(dpy); - - return 0; -} - diff --git a/src/driclient/src/xf86dri.h b/src/driclient/src/xf86dri.h deleted file mode 100644 index baf80a7a9dd..00000000000 --- a/src/driclient/src/xf86dri.h +++ /dev/null @@ -1,119 +0,0 @@ -/************************************************************************** - -Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas. -Copyright 2000 VA Linux Systems, Inc. -All Rights Reserved. - -Permission is hereby granted, free of charge, to any person obtaining a -copy of this software and associated documentation files (the -"Software"), to deal in the Software without restriction, including -without limitation the rights to use, copy, modify, merge, publish, -distribute, sub license, and/or sell copies of the Software, and to -permit persons to whom the Software is furnished to do so, subject to -the following conditions: - -The above copyright notice and this permission notice (including the -next paragraph) shall be included in all copies or substantial portions -of the Software. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS -OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF -MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. -IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR -ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, -TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE -SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - -**************************************************************************/ - -/** - * \file xf86dri.h - * Protocol numbers and function prototypes for DRI X protocol. - * - * \author Kevin E. Martin - * \author Jens Owen - * \author Rickard E. (Rik) Faith - */ - -#ifndef _XF86DRI_H_ -#define _XF86DRI_H_ - -#include -#include - -#define X_XF86DRIQueryVersion 0 -#define X_XF86DRIQueryDirectRenderingCapable 1 -#define X_XF86DRIOpenConnection 2 -#define X_XF86DRICloseConnection 3 -#define X_XF86DRIGetClientDriverName 4 -#define X_XF86DRICreateContext 5 -#define X_XF86DRIDestroyContext 6 -#define X_XF86DRICreateDrawable 7 -#define X_XF86DRIDestroyDrawable 8 -#define X_XF86DRIGetDrawableInfo 9 -#define X_XF86DRIGetDeviceInfo 10 -#define X_XF86DRIAuthConnection 11 -#define X_XF86DRIOpenFullScreen 12 /* Deprecated */ -#define X_XF86DRICloseFullScreen 13 /* Deprecated */ - -#define XF86DRINumberEvents 0 - -#define XF86DRIClientNotLocal 0 -#define XF86DRIOperationNotSupported 1 -#define XF86DRINumberErrors (XF86DRIOperationNotSupported + 1) - -#ifndef _XF86DRI_SERVER_ - -_XFUNCPROTOBEGIN - -Bool XF86DRIQueryExtension( Display *dpy, int *event_base, int *error_base ); - -Bool XF86DRIQueryVersion( Display *dpy, int *majorVersion, int *minorVersion, - int *patchVersion ); - -Bool XF86DRIQueryDirectRenderingCapable( Display *dpy, int screen, - Bool *isCapable ); - -Bool XF86DRIOpenConnection( Display *dpy, int screen, drm_handle_t *hSAREA, - char **busIDString ); - -Bool XF86DRIAuthConnection( Display *dpy, int screen, drm_magic_t magic ); - -Bool XF86DRICloseConnection( Display *dpy, int screen ); - -Bool XF86DRIGetClientDriverName( Display *dpy, int screen, - int *ddxDriverMajorVersion, int *ddxDriverMinorVersion, - int *ddxDriverPatchVersion, char **clientDriverName ); - -Bool XF86DRICreateContext( Display *dpy, int screen, Visual *visual, - XID *ptr_to_returned_context_id, drm_context_t *hHWContext ); - -Bool XF86DRICreateContextWithConfig( Display *dpy, int screen, int configID, - XID *ptr_to_returned_context_id, drm_context_t *hHWContext ); - -Bool XF86DRIDestroyContext( Display *dpy, int screen, - XID context_id ); - -Bool XF86DRICreateDrawable( Display *dpy, int screen, - Drawable drawable, drm_drawable_t *hHWDrawable ); - -Bool XF86DRIDestroyDrawable( Display *dpy, int screen, - Drawable drawable); - -Bool XF86DRIGetDrawableInfo( Display *dpy, int screen, Drawable drawable, - unsigned int *index, unsigned int *stamp, - int *X, int *Y, int *W, int *H, - int *numClipRects, drm_clip_rect_t ** pClipRects, - int *backX, int *backY, - int *numBackClipRects, drm_clip_rect_t **pBackClipRects ); - -Bool XF86DRIGetDeviceInfo( Display *dpy, int screen, - drm_handle_t *hFrameBuffer, int *fbOrigin, int *fbSize, - int *fbStride, int *devPrivateSize, void **pDevPrivate ); - -_XFUNCPROTOEND - -#endif /* _XF86DRI_SERVER_ */ - -#endif /* _XF86DRI_H_ */ - diff --git a/src/gallium/winsys/g3dvl/nouveau/Makefile b/src/gallium/winsys/g3dvl/nouveau/Makefile index 7fa29d2f5fb..5d11bde322b 100644 --- a/src/gallium/winsys/g3dvl/nouveau/Makefile +++ b/src/gallium/winsys/g3dvl/nouveau/Makefile @@ -34,15 +34,16 @@ LIBS += -ldriclient -ldrm -lnv10 -lnv30 -lnv40 -lnv50 -ldraw -ltgsi -ltranslate ############################################# -.PHONY = all clean +.PHONY = all clean libdriclient all: ${TARGET} -${TARGET}: ${OBJECTS} +${TARGET}: ${OBJECTS} libdriclient + $(CC) ${LDFLAGS} -shared -o $@ ${OBJECTS} ${LIBS} + +libdriclient: cd ${DRIDIR}/src; ${MAKE} - $(CC) ${LDFLAGS} -shared -o $@ $^ ${LIBS} clean: cd ${DRIDIR}/src; ${MAKE} clean rm -rf ${OBJECTS} ${TARGET} - diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile index b72bb16efbf..8565d271c70 100644 --- a/src/libXvMC/Makefile +++ b/src/libXvMC/Makefile @@ -38,27 +38,36 @@ endif ############################################# -.PHONY = all clean +ifeq (${DRIVER}, softpipe) +.PHONY = all clean g3dvl +else +.PHONY = all clean g3dvl nouveau_winsys +endif all: ${TARGET} ifeq (${DRIVER}, softpipe) -${TARGET}: ${OBJECTS} +${TARGET}: g3dvl + $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS} + +g3dvl: cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} - $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ $^ ${LIBS} clean: cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean rm -rf ${OBJECTS} ${TARGET} else -${TARGET}: ${OBJECTS} +${TARGET}: ${OBJECTS} g3dvl nouveau_winsys + $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS} + +g3dvl: cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} + +nouveau_winsys: cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} - $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ $^ ${LIBS} clean: cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} clean rm -rf ${OBJECTS} ${TARGET} endif - -- cgit v1.2.3 From ca0efa84edf2f099f6797959a20186b6eb3451b4 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Tue, 9 Sep 2008 00:28:37 -0400 Subject: g3dvl: Cleanup. --- src/driclient/src/.gitignore | 1 - 1 file changed, 1 deletion(-) delete mode 100644 src/driclient/src/.gitignore diff --git a/src/driclient/src/.gitignore b/src/driclient/src/.gitignore deleted file mode 100644 index 9daeafb9864..00000000000 --- a/src/driclient/src/.gitignore +++ /dev/null @@ -1 +0,0 @@ -test -- cgit v1.2.3 From 59f23e92e24b93eb48f4e0552dd8e397aefd1714 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Fri, 12 Sep 2008 13:22:43 -0400 Subject: g3dvl: Update softlinks to nouveau winsys files, related fixes. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_context.c | 37 ++++---- src/gallium/winsys/g3dvl/nouveau/nouveau_device.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_local.h | 2 +- .../winsys/g3dvl/nouveau/nouveau_notifier.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c | 2 +- .../winsys/g3dvl/nouveau/nouveau_resource.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c | 13 ++- src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c | 2 +- .../winsys/g3dvl/nouveau/nouveau_winsys_pipe.c | 11 +-- .../winsys/g3dvl/nouveau/nouveau_winsys_pipe.h | 2 +- .../winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nv04_surface.c | 2 +- src/gallium/winsys/g3dvl/nouveau/nv50_surface.c | 2 +- src/gallium/winsys/g3dvl/xsp_winsys.c | 105 ++++++++++----------- src/libXvMC/Makefile | 2 +- 24 files changed, 101 insertions(+), 105 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 93839e8aa9b..cfbad08038b 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -6,10 +6,10 @@ #include #include #include -#include #include #include #include +#include #include "vl_render.h" #include "vl_shader_build.h" #include "vl_surface.h" diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c index 73ac4a4171a..1005282dd4d 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_bo.c \ No newline at end of file +../../drm/nouveau/nouveau_bo.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c index 6c9b2c48d84..5af82029508 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_channel.c \ No newline at end of file +../../drm/nouveau/nouveau_channel.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c index 5e173c76728..06a61fcda3e 100644 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c @@ -1,7 +1,7 @@ #include "pipe/p_defines.h" #include "pipe/p_context.h" #include "pipe/p_screen.h" -#include "pipe/p_util.h" +#include "util/u_memory.h" #include "nouveau_context.h" #include "nouveau_dri.h" @@ -150,7 +150,7 @@ nouveau_context_create(dri_context_t *dri_context) fb_bo = calloc(1, sizeof(struct nouveau_bo_priv)); fb_bo->drm.offset = nv_screen->front_offset; fb_bo->drm.flags = NOUVEAU_MEM_FB; - fb_bo->drm.size = nv_screen->front_pitch * + fb_bo->drm.size = nv_screen->front_pitch * nv_screen->front_height; fb_bo->refcount = 1; fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM; @@ -280,7 +280,7 @@ nouveau_context_bind(struct nouveau_context *nv, dri_drawable_t *dri_drawable) { assert(nv); assert(dri_drawable); - + if (nv->dri_drawable != dri_drawable) { nv->dri_drawable = dri_drawable; @@ -294,9 +294,9 @@ int nouveau_context_unbind(struct nouveau_context *nv) { assert(nv); - + nv->dri_drawable = NULL; - + return 0; } @@ -306,20 +306,20 @@ int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable) { struct nouveau_context *nv; dri_drawable_t *dri_drawable; - + nv = pipe->priv; - + driCreateDrawable(nv->nv_screen->dri_screen, drawable, &dri_drawable); - + nouveau_context_bind(nv, dri_drawable); - + return 0; } int unbind_pipe_drawable(struct pipe_context *pipe) { nouveau_context_unbind(pipe->priv); - + return 0; } @@ -329,15 +329,15 @@ struct pipe_context* create_pipe_context(Display *display, int screen) dri_framebuffer_t dri_framebuf; dri_context_t *dri_context; struct nouveau_context *nv; - + driCreateScreen(display, screen, &dri_screen, &dri_framebuf); driCreateContext(dri_screen, XDefaultVisual(display, screen), &dri_context); - + nouveau_screen_create(dri_screen, &dri_framebuf); nouveau_context_create(dri_context); - + nv = dri_context->private; - + return nv->nvc->pctx[nv->pctx_id]; } @@ -348,15 +348,15 @@ int destroy_pipe_context(struct pipe_context *pipe) struct nouveau_context *nv; dri_screen_t *dri_screen; dri_context_t *dri_context; - + assert(pipe); - + screen = pipe->screen; winsys = pipe->winsys; nv = pipe->priv; dri_context = nv->dri_context; dri_screen = dri_context->dri_screen; - + pipe->destroy(pipe); screen->destroy(screen); free(winsys); @@ -365,7 +365,6 @@ int destroy_pipe_context(struct pipe_context *pipe) nouveau_screen_destroy(dri_screen); driDestroyContext(dri_context); driDestroyScreen(dri_screen); - + return 0; } - diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c index 47d52dafa89..63f1fa040c7 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_device.c \ No newline at end of file +../../drm/nouveau/nouveau_device.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c index 45078c964f4..cd0d32e5373 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_dma.c \ No newline at end of file +../../drm/nouveau/nouveau_dma.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h index 6b9ec777417..e6c7d4bc965 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h @@ -1 +1 @@ -../../dri/nouveau/nouveau_dma.h \ No newline at end of file +../../drm/nouveau/nouveau_dma.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h index 0e6c9fce355..c8f9dbdc3a8 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h @@ -1 +1 @@ -../../dri/nouveau/nouveau_dri.h \ No newline at end of file +../../drm/nouveau/nouveau_dri.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h index 473b7d4812d..27082c9d34c 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h @@ -1 +1 @@ -../../dri/nouveau/nouveau_drmif.h \ No newline at end of file +../../drm/nouveau/nouveau_drmif.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c index ef1f0c6afe1..51a50527e6d 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_fence.c \ No newline at end of file +../../drm/nouveau/nouveau_fence.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c index 428186544c6..db17c72e6d7 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_grobj.c \ No newline at end of file +../../drm/nouveau/nouveau_grobj.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h index 6b878aad224..4e9d3042cb5 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h @@ -1 +1 @@ -../../dri/nouveau/nouveau_local.h \ No newline at end of file +../../drm/nouveau/nouveau_local.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c index 3024a612a7a..703bc3ce4ab 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_notifier.c \ No newline at end of file +../../drm/nouveau/nouveau_notifier.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c index dae31d97998..4ac137cadad 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_pushbuf.c \ No newline at end of file +../../drm/nouveau/nouveau_pushbuf.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c index e0d71e9d2bf..2241af328ff 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_resource.c \ No newline at end of file +../../drm/nouveau/nouveau_resource.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c index daea3fff680..f80d00050c8 100644 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c @@ -1,5 +1,5 @@ #include "pipe/p_context.h" -#include "pipe/p_util.h" +#include "util/u_memory.h" #include "nouveau_context.h" #include #include "nouveau_dri.h" @@ -23,11 +23,11 @@ int nouveau_check_dri_drm_ddx(dri_version_t *dri, dri_version_t *drm, dri_versio static const dri_version_t ddx_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL}; static const dri_version_t dri_expected = {4, 0, 0}; static const dri_version_t drm_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL}; - + assert(dri); assert(drm); assert(ddx); - + if (dri->major != dri_expected.major || dri->minor < dri_expected.minor) { NOUVEAU_ERR("Unexpected DRI version.\n"); @@ -43,17 +43,17 @@ int nouveau_check_dri_drm_ddx(dri_version_t *dri, dri_version_t *drm, dri_versio NOUVEAU_ERR("Unexpected DDX version.\n"); return 1; } - + return 0; } int nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf) -{ +{ struct nouveau_dri *nv_dri = dri_framebuf->private; struct nouveau_screen *nv_screen; int ret; - + if (nouveau_check_dri_drm_ddx(&dri_screen->dri, &dri_screen->drm, &dri_screen->ddx)) return 1; @@ -89,4 +89,3 @@ nouveau_screen_destroy(dri_screen_t *dri_screen) FREE(nv_screen); } - diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c index 43de49b98bb..ce4052d5572 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_winsys.c \ No newline at end of file +../../drm/nouveau/nouveau_winsys.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c index b835bd57606..5d5380c2502 100644 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c @@ -1,7 +1,7 @@ #include "pipe/p_winsys.h" #include "pipe/p_defines.h" -#include "pipe/p_util.h" #include "pipe/p_inlines.h" +#include "util/u_memory.h" #include "nouveau_context.h" #include "nouveau_local.h" @@ -29,7 +29,7 @@ static struct pipe_surface * nouveau_surface_alloc(struct pipe_winsys *ws) { struct pipe_surface *surf; - + surf = CALLOC_STRUCT(pipe_surface); if (!surf) return NULL; @@ -59,10 +59,10 @@ nouveau_surface_alloc_storage ) { const unsigned int ALIGNMENT = 256; - + assert(pws); assert(surface); - + surface->width = width; surface->height = height; surface->format = format; @@ -72,7 +72,7 @@ nouveau_surface_alloc_storage surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT); surface->usage = flags; surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * surface->nblocksy); - + return 0; } @@ -258,4 +258,3 @@ nouveau_create_pipe_winsys(struct nouveau_context *nv) return &nvpws->pws; } - diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h index 264716fef04..9d9460883e0 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h @@ -1 +1 @@ -../../dri/nouveau/nouveau_winsys_pipe.h \ No newline at end of file +../../drm/nouveau/nouveau_winsys_pipe.h \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c index 83faccde969..ec613ecbf65 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c @@ -1 +1 @@ -../../dri/nouveau/nouveau_winsys_softpipe.c \ No newline at end of file +../../drm/nouveau/nouveau_winsys_softpipe.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c b/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c index e05f0671d63..4455d8f9243 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c +++ b/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c @@ -1 +1 @@ -../../dri/nouveau/nv04_surface.c \ No newline at end of file +../../drm/nouveau/nv04_surface.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c b/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c index 3850748229d..19f102001e9 120000 --- a/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c +++ b/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c @@ -1 +1 @@ -../../dri/nouveau/nv50_surface.c \ No newline at end of file +../../drm/nouveau/nv50_surface.c \ No newline at end of file diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index 0100fe37bdb..42e7841df85 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -2,8 +2,8 @@ #include #include #include -#include #include +#include #include /* pipe_winsys implementation */ @@ -33,79 +33,79 @@ struct xsp_buffer static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size) { struct xsp_buffer *buffer; - + assert(pws); - + buffer = calloc(1, sizeof(struct xsp_buffer)); buffer->base.refcount = 1; buffer->base.alignment = alignment; buffer->base.usage = usage; buffer->base.size = size; buffer->data = align_malloc(size, alignment); - + return (struct pipe_buffer*)buffer; } static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size) { struct xsp_buffer *buffer; - + assert(pws); - + buffer = calloc(1, sizeof(struct xsp_buffer)); buffer->base.refcount = 1; buffer->base.size = size; buffer->is_user_buffer = TRUE; buffer->data = data; - + return (struct pipe_buffer*)buffer; } static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags) { struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; - + assert(pws); assert(buffer); - + xsp_buf->mapped_data = xsp_buf->data; - + return xsp_buf->mapped_data; } static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer) { struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; - + assert(pws); assert(buffer); - + xsp_buf->mapped_data = NULL; } static void xsp_buffer_destroy(struct pipe_winsys *pws, struct pipe_buffer *buffer) { struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; - + assert(pws); assert(buffer); - + if (!xsp_buf->is_user_buffer) align_free(xsp_buf->data); - + free(xsp_buf); } static struct pipe_surface* xsp_surface_alloc(struct pipe_winsys *pws) { struct pipe_surface *surface; - + assert(pws); - + surface = calloc(1, sizeof(struct pipe_surface)); surface->refcount = 1; surface->winsys = pws; - + return surface; } @@ -127,10 +127,10 @@ static int xsp_surface_alloc_storage ) { const unsigned int ALIGNMENT = 1; - + assert(pws); assert(surface); - + surface->width = width; surface->height = height; surface->format = format; @@ -140,28 +140,28 @@ static int xsp_surface_alloc_storage surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT); surface->usage = flags; surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * surface->nblocksy); - + return 0; } static void xsp_surface_release(struct pipe_winsys *pws, struct pipe_surface **surface) { struct pipe_surface *s; - + assert(pws); assert(surface); assert(*surface); - + s = *surface; - + s->refcount--; - + if (s->refcount == 0) { pipe_buffer_reference(pws, &s->buffer, NULL); free(s); } - + *surface = NULL; } @@ -176,7 +176,7 @@ static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle { assert(pws); assert(fence); - + return 0; } @@ -184,30 +184,30 @@ static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *f { assert(pws); assert(fence); - + return 0; } static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private) -{ +{ struct xsp_pipe_winsys *xsp_winsys; struct xsp_context *xsp_context; - + assert(pws); assert(surface); assert(context_private); - + xsp_winsys = (struct xsp_pipe_winsys*)pws; xsp_context = (struct xsp_context*)context_private; - + if (!xsp_context->drawable_bound) return; - + xsp_winsys->fbimage.width = surface->width; xsp_winsys->fbimage.height = surface->height; xsp_winsys->fbimage.bytes_per_line = surface->width * (xsp_winsys->fbimage.bits_per_pixel >> 3); xsp_winsys->fbimage.data = pipe_surface_map(surface, 0); - + XPutImage ( xsp_context->display, @@ -236,25 +236,25 @@ static const char* xsp_get_name(struct pipe_winsys *pws) int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable) { struct xsp_context *xsp_context; - + assert(pipe); - + xsp_context = pipe->priv; xsp_context->drawable = drawable; xsp_context->drawable_bound = 1; - + return 0; } int unbind_pipe_drawable(struct pipe_context *pipe) { struct xsp_context *xsp_context; - + assert(pipe); - + xsp_context = pipe->priv; xsp_context->drawable_bound = 0; - + return 0; } @@ -264,9 +264,9 @@ struct pipe_context* create_pipe_context(Display *display, int screen) struct xsp_context *xsp_context; struct pipe_screen *sp_screen; struct pipe_context *sp_pipe; - + assert(display); - + xsp_winsys = calloc(1, sizeof(struct xsp_pipe_winsys)); xsp_winsys->base.buffer_create = xsp_buffer_create; xsp_winsys->base.user_buffer_create = xsp_user_buffer_create; @@ -281,7 +281,7 @@ struct pipe_context* create_pipe_context(Display *display, int screen) xsp_winsys->base.fence_finish = xsp_fence_finish; xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer; xsp_winsys->base.get_name = xsp_get_name; - + { /* XXX: Can't use the returned XImage* directly, since we don't have control over winsys destruction @@ -299,22 +299,22 @@ struct pipe_context* create_pipe_context(Display *display, int screen) 32, 0 ); - + memcpy(&xsp_winsys->fbimage, template, sizeof(XImage)); XInitImage(&xsp_winsys->fbimage); - + XDestroyImage(template); } - + sp_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); sp_pipe = softpipe_create(sp_screen, (struct pipe_winsys*)xsp_winsys, NULL); - + xsp_context = calloc(1, sizeof(struct xsp_context)); xsp_context->display = display; xsp_context->screen = screen; - + sp_pipe->priv = xsp_context; - + return sp_pipe; } @@ -322,16 +322,15 @@ int destroy_pipe_context(struct pipe_context *pipe) { struct pipe_screen *screen; struct pipe_winsys *winsys; - + assert(pipe); - + screen = pipe->screen; winsys = pipe->winsys; free(pipe->priv); pipe->destroy(pipe); screen->destroy(screen); free(winsys); - + return 0; } - diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile index 8565d271c70..4de26e96202 100644 --- a/src/libXvMC/Makefile +++ b/src/libXvMC/Makefile @@ -47,7 +47,7 @@ endif all: ${TARGET} ifeq (${DRIVER}, softpipe) -${TARGET}: g3dvl +${TARGET}: ${OBJECTS} g3dvl $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS} g3dvl: -- cgit v1.2.3 From 42a42dec3dbb5e150584b3d0b2e14e9b555a4ac1 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Tue, 9 Sep 2008 02:27:05 -0400 Subject: g3dvl: Fullscreen fixes. Respect src & dst coords during scaling, clear FB. --- src/gallium/state_trackers/g3dvl/vl_basic_csc.c | 83 ++++++++++++++----------- src/gallium/state_trackers/g3dvl/vl_surface.c | 4 +- src/gallium/state_trackers/g3dvl/vl_surface.h | 2 + src/libXvMC/surface.c | 2 +- 4 files changed, 54 insertions(+), 37 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c index ea003a31d17..9f9dafc8a95 100644 --- a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c +++ b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c @@ -14,6 +14,8 @@ struct vlVertexShaderConsts { + struct vlVertex4f dst_scale; + struct vlVertex4f dst_trans; struct vlVertex4f src_scale; struct vlVertex4f src_trans; }; @@ -87,6 +89,9 @@ static int vlResizeFrameBuffer 0 ); + /* Clear to black, in case video doesn't fill the entire window */ + pipe->clear(pipe, basic_csc->framebuffer.cbufs[0], 0); + return 0; } @@ -149,6 +154,15 @@ static int vlPutPictureCSC PIPE_BUFFER_USAGE_CPU_WRITE ); + vs_consts->dst_scale.x = destw / (float)basic_csc->framebuffer.cbufs[0]->width; + vs_consts->dst_scale.y = desth / (float)basic_csc->framebuffer.cbufs[0]->height; + vs_consts->dst_scale.z = 1; + vs_consts->dst_scale.w = 1; + vs_consts->dst_trans.x = destx / (float)basic_csc->framebuffer.cbufs[0]->width; + vs_consts->dst_trans.y = desty / (float)basic_csc->framebuffer.cbufs[0]->height; + vs_consts->dst_trans.z = 0; + vs_consts->dst_trans.w = 0; + vs_consts->src_scale.x = srcw / (float)surface->texture->width[0]; vs_consts->src_scale.y = srch / (float)surface->texture->height[0]; vs_consts->src_scale.z = 1; @@ -376,10 +390,12 @@ static int vlCreateVertexShader } /* - * decl c0 ; Scaling vector to scale texcoord rect to source size - * decl c1 ; Translation vector to move texcoord rect into position + * decl c0 ; Scaling vector to scale vertex pos rect to destination size + * decl c1 ; Translation vector to move vertex pos rect into position + * decl c2 ; Scaling vector to scale texcoord rect to source size + * decl c3 ; Translation vector to move texcoord rect into position */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* @@ -392,21 +408,19 @@ static int vlCreateVertexShader ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* decl t0 */ - decl = vl_decl_temps(0, 0); + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - /* mov o0, i0 ; Move pos in to pos out */ - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_INPUT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t0, i1, c0 ; Scale unit texcoord rect to source size */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 1, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o1, t0, c1 ; Translate texcoord rect into position */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + /* + * madd o0, i0, c0, c1 ; Scale and translate unit output rect to destination size and pos + * madd o1, i1, c2, c3 ; Scale and translate unit texcoord rect to source size and pos + */ + for (i = 0; i < 2; ++i) + { + inst = vl_inst4(TGSI_OPCODE_MADD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } /* end */ inst = vl_end(); @@ -487,9 +501,8 @@ static int vlCreateFragmentShader * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix * dp4 o0.y, t0, c2 * dp4 o0.z, t0, c3 - * dp4 o0.w, t0, c4 ; XXX: Don't need 4th coefficient */ - for (i = 0; i < 4; ++i) + for (i = 0; i < 3; ++i) { inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; @@ -519,11 +532,11 @@ static int vlCreateDataBufs pipe = csc->pipe; /* - Create our vertex buffer and vertex buffer element - VB contains 4 vertices that render a quad covering the entire window - to display a rendered surface - Quad is rendered as a tri strip - */ + * Create our vertex buffer and vertex buffer element + * VB contains 4 vertices that render a quad covering the entire window + * to display a rendered surface + * Quad is rendered as a tri strip + */ csc->vertex_bufs[0].pitch = sizeof(struct vlVertex2f); csc->vertex_bufs[0].max_index = 3; csc->vertex_bufs[0].buffer_offset = 0; @@ -550,9 +563,9 @@ static int vlCreateDataBufs csc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; /* - Create our texcoord buffer and texcoord buffer element - Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices - */ + * Create our texcoord buffer and texcoord buffer element + * Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices + */ csc->vertex_bufs[1].pitch = sizeof(struct vlVertex2f); csc->vertex_bufs[1].max_index = 3; csc->vertex_bufs[1].buffer_offset = 0; @@ -579,9 +592,9 @@ static int vlCreateDataBufs csc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; /* - Create our vertex shader's constant buffer - Const buffer contains scaling and translation vectors - */ + * Create our vertex shader's constant buffer + * Const buffer contains scaling and translation vectors + */ csc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts); csc->vs_const_buf.buffer = pipe->winsys->buffer_create ( @@ -592,9 +605,9 @@ static int vlCreateDataBufs ); /* - Create our fragment shader's constant buffer - Const buffer contains the color conversion matrix and bias vectors - */ + * Create our fragment shader's constant buffer + * Const buffer contains the color conversion matrix and bias vectors + */ csc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts); csc->fs_const_buf.buffer = pipe->winsys->buffer_create ( @@ -605,9 +618,9 @@ static int vlCreateDataBufs ); /* - TODO: Refactor this into a seperate function, - allow changing the CSC matrix at runtime to switch between regular & full versions - */ + * TODO: Refactor this into a seperate function, + * allow changing the CSC matrix at runtime to switch between regular & full versions + */ memcpy ( pipe->winsys->buffer_map(pipe->winsys, csc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 687fd1ec29b..07a9be2a990 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -104,6 +104,8 @@ int vlPutPicture int desty, int destw, int desth, + int drawable_w, + int drawable_h, enum vlPictureType picture_type ) { @@ -118,7 +120,7 @@ int vlPutPicture csc = surface->context->csc; pipe = surface->context->pipe; - csc->vlResizeFrameBuffer(csc, destw, desth); + csc->vlResizeFrameBuffer(csc, drawable_w, drawable_h); csc->vlBegin(csc); diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h index b975e131fa6..3e050503450 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.h +++ b/src/gallium/state_trackers/g3dvl/vl_surface.h @@ -49,6 +49,8 @@ int vlPutPicture int desty, int destw, int desth, + int drawable_w, + int drawable_h, enum vlPictureType picture_type ); diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index 038befc297b..b8459c03cf4 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -266,7 +266,7 @@ Status XvMCPutSurface vl_sfc = surface->privData; - vlPutPicture(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, PictureToVL(flags)); + vlPutPicture(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, width, height, PictureToVL(flags)); return Success; } -- cgit v1.2.3 From f3f449a49136ae2fd2dc3cf62d2c24dd42505e7d Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 10 Sep 2008 19:37:56 -0400 Subject: g3dvl: Implement surface sync functions. --- .../state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 20 ++++---- src/gallium/state_trackers/g3dvl/vl_surface.c | 59 +++++++++++++++++++++- src/gallium/state_trackers/g3dvl/vl_surface.h | 30 ++++++++--- src/gallium/state_trackers/g3dvl/vl_types.h | 7 +++ src/libXvMC/surface.c | 31 ++++++++++-- 5 files changed, 123 insertions(+), 24 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index cfbad08038b..75d326b36e8 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -591,6 +591,10 @@ static int vlFlush assert(render); mc = (struct vlR16SnormBufferedMC*)render; + + if (!mc->buffered_surface) + return 0; + pipe = mc->pipe; for (i = 0; i < mc->num_macroblocks; ++i) @@ -736,8 +740,12 @@ static int vlFlush vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24; } + pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &mc->buffered_surface->render_fence); + for (i = 0; i < 3; ++i) mc->zero_block[i].x = -1.0f; + + mc->buffered_surface = NULL; mc->num_macroblocks = 0; mc->cur_buf++; @@ -760,12 +768,7 @@ static int vlRenderMacroBlocksMpeg2R16SnormBuffered if (mc->buffered_surface) { - if - ( - mc->buffered_surface != surface /*|| - mc->past_surface != batch->past_surface || - mc->future_surface != batch->future_surface*/ - ) + if (mc->buffered_surface != surface) { vlFlush(&mc->base); mc->buffered_surface = surface; @@ -1027,7 +1030,6 @@ static int vlCreateFragmentShaderIMB inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } /* mul o0, t0, c0 ; Rescale texel to correct range */ @@ -1323,7 +1325,6 @@ static int vlCreateFragmentShaderFramePMB inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } /* mul t0, t0, c0 ; Rescale texel to correct range */ @@ -1442,7 +1443,6 @@ static int vlCreateFragmentShaderFieldPMB inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } /* mul t0, t0, c0 ; Rescale texel to correct range */ @@ -1818,7 +1818,6 @@ static int vlCreateFragmentShaderFrameBMB inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } /* mul t0, t0, c0 ; Rescale texel to correct range */ @@ -1955,7 +1954,6 @@ static int vlCreateFragmentShaderFieldBMB inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } /* mul t0, t0, c0 ; Rescale texel to correct range */ diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 07a9be2a990..6648133ef83 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -77,6 +77,7 @@ int vlRenderMacroBlocksMpeg2 { assert(batch); assert(surface); + assert(surface->context); surface->context->render->vlBegin(surface->context->render); @@ -141,9 +142,10 @@ int vlPutPicture csc->vlEnd(csc); - pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &surface->disp_fence); + bind_pipe_drawable(pipe, drawable); - /* TODO: Need to take destx, desty into consideration */ + pipe->winsys->flush_frontbuffer ( pipe->winsys, @@ -154,6 +156,59 @@ int vlPutPicture return 0; } +int vlSurfaceGetStatus +( + struct vlSurface *surface, + enum vlResourceStatus *status +) +{ + assert(surface); + assert(surface->context); + assert(status); + + if (surface->render_fence && !surface->context->pipe->winsys->fence_signalled(surface->context->pipe->winsys, surface->render_fence, 0)) + { + *status = vlResourceStatusRendering; + return 0; + } + + if (surface->disp_fence && !surface->context->pipe->winsys->fence_signalled(surface->context->pipe->winsys, surface->disp_fence, 0)) + { + *status = vlResourceStatusDisplaying; + return 0; + } + + *status = vlResourceStatusFree; + return 0; +} + +int vlSurfaceFlush +( + struct vlSurface *surface +) +{ + assert(surface); + assert(surface->context); + + surface->context->render->vlFlush(surface->context->render); + + return 0; +} + +int vlSurfaceSync +( + struct vlSurface *surface +) +{ + assert(surface); + assert(surface->context); + assert(surface->render_fence); + + surface->context->pipe->winsys->fence_finish(surface->context->pipe->winsys, surface->render_fence, 0); + + return 0; +} + struct vlScreen* vlSurfaceGetScreen ( struct vlSurface *surface diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h index 3e050503450..133e1515ef3 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.h +++ b/src/gallium/state_trackers/g3dvl/vl_surface.h @@ -8,12 +8,14 @@ struct pipe_texture; struct vlSurface { - struct vlScreen *screen; - struct vlContext *context; - unsigned int width; - unsigned int height; - enum vlFormat format; - struct pipe_texture *texture; + struct vlScreen *screen; + struct vlContext *context; + unsigned int width; + unsigned int height; + enum vlFormat format; + struct pipe_texture *texture; + struct pipe_fence_handle *render_fence; + struct pipe_fence_handle *disp_fence; }; #endif @@ -54,6 +56,22 @@ int vlPutPicture enum vlPictureType picture_type ); +int vlSurfaceGetStatus +( + struct vlSurface *surface, + enum vlResourceStatus *status +); + +int vlSurfaceFlush +( + struct vlSurface *surface +); + +int vlSurfaceSync +( + struct vlSurface *surface +); + struct vlScreen* vlSurfaceGetScreen ( struct vlSurface *surface diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h index b432bfde935..274e1f74377 100644 --- a/src/gallium/state_trackers/g3dvl/vl_types.h +++ b/src/gallium/state_trackers/g3dvl/vl_types.h @@ -13,6 +13,13 @@ struct vlScreen; struct vlContext; struct vlSurface; +enum vlResourceStatus +{ + vlResourceStatusFree, + vlResourceStatusRendering, + vlResourceStatusDisplaying +}; + enum vlProfile { vlProfileMpeg2Simple, diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index b8459c03cf4..a920da52f91 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -201,7 +201,7 @@ Status XvMCFlushSurface(Display *display, XvMCSurface *surface) assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - /* TODO */ + vlSurfaceFlush(vl_sfc); return Success; } @@ -219,7 +219,7 @@ Status XvMCSyncSurface(Display *display, XvMCSurface *surface) assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - /* TODO */ + vlSurfaceSync(vl_sfc); return Success; } @@ -273,7 +273,8 @@ Status XvMCPutSurface Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status) { - struct vlSurface *vl_sfc; + struct vlSurface *vl_sfc; + enum vlResourceStatus res_status; assert(display); @@ -286,8 +287,28 @@ Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status) assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - /* TODO */ - *status = 0; + vlSurfaceGetStatus(vl_sfc, &res_status); + + switch (res_status) + { + case vlResourceStatusFree: + { + *status = 0; + break; + } + case vlResourceStatusRendering: + { + *status = XVMC_RENDERING; + break; + } + case vlResourceStatusDisplaying: + { + *status = XVMC_DISPLAYING; + break; + } + default: + assert(0); + } return Success; } -- cgit v1.2.3 From 56c22687e139e8fede6f7ea4d524699364dda07c Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Fri, 12 Sep 2008 15:10:43 -0400 Subject: g3dvl: Use winsys_buffer_reference() instead of pipe_buffer_reference(). Old pipe_buffer_reference() renamed winsys_buffer_reference(), new pipe_buffer_reference() expects a screen. --- src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c | 2 +- src/gallium/winsys/g3dvl/xsp_winsys.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c index 5d5380c2502..4f6ac9cad0b 100644 --- a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c +++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c @@ -84,7 +84,7 @@ nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s) *s = NULL; if (--surf->refcount <= 0) { if (surf->buffer) - pipe_buffer_reference(ws, &surf->buffer, NULL); + winsys_buffer_reference(ws, &surf->buffer, NULL); free(surf); } } diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c index 42e7841df85..68be2c2ea3a 100644 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ b/src/gallium/winsys/g3dvl/xsp_winsys.c @@ -158,7 +158,7 @@ static void xsp_surface_release(struct pipe_winsys *pws, struct pipe_surface **s if (s->refcount == 0) { - pipe_buffer_reference(pws, &s->buffer, NULL); + winsys_buffer_reference(pws, &s->buffer, NULL); free(s); } -- cgit v1.2.3 From 3122f2bebe8d76568916b8cddff542f52466055e Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 13 Sep 2008 01:35:14 -0400 Subject: g3dvl: Fix field coded block copy. --- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 18 ++++-------------- src/libXvMC/surface.c | 8 ++++---- 2 files changed, 8 insertions(+), 18 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 75d326b36e8..888f0040bf7 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -108,17 +108,7 @@ static inline int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst { unsigned int y; - for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y) - memcpy - ( - dst + y * dst_pitch * 2, - src + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - - dst += VL_BLOCK_HEIGHT * dst_pitch; - - for (; y < VL_BLOCK_HEIGHT; ++y) + for (y = 0; y < VL_BLOCK_HEIGHT; ++y) memcpy ( dst + y * dst_pitch * 2, @@ -1181,7 +1171,7 @@ static int vlCreateVertexShaderFieldPMB ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* decl c0 ; Texcoord denorm coefficients */ + /* decl c0 ; Render target dimensions */ decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); @@ -1661,8 +1651,8 @@ static int vlCreateVertexShaderFieldBMB ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } - /* decl c0 ; Denorm coefficients */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 6); + /* decl c0 ; Render target dimensions */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); /* diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c index a920da52f91..6031b39d108 100644 --- a/src/libXvMC/surface.c +++ b/src/libXvMC/surface.c @@ -40,12 +40,12 @@ static enum vlPictureType PictureToVL(int xvmc_pic) return -1; } -static enum vlMotionType MotionToVL(int xvmc_motion_type) +static enum vlMotionType MotionToVL(int xvmc_motion_type, int xvmc_dct_type) { switch (xvmc_motion_type) { case XVMC_PREDICTION_FRAME: - return vlMotionTypeFrame; + return xvmc_dct_type == XVMC_DCT_TYPE_FIELD ? vlMotionType16x8 : vlMotionTypeFrame; case XVMC_PREDICTION_FIELD: return vlMotionTypeField; case XVMC_PREDICTION_DUAL_PRIME: @@ -171,8 +171,8 @@ Status XvMCRenderSurface batch.macroblocks[i].mby = macroblocks->macro_blocks[j].y; batch.macroblocks[i].mb_type = TypeToVL(macroblocks->macro_blocks[j].macroblock_type); if (batch.macroblocks[i].mb_type != vlMacroBlockTypeIntra) - batch.macroblocks[i].mo_type = MotionToVL(macroblocks->macro_blocks[j].motion_type); - batch.macroblocks[i].dct_type = macroblocks->macro_blocks[j].dct_type & XVMC_DCT_TYPE_FIELD ? vlDCTTypeFieldCoded : vlDCTTypeFrameCoded; + batch.macroblocks[i].mo_type = MotionToVL(macroblocks->macro_blocks[j].motion_type, macroblocks->macro_blocks[j].dct_type); + batch.macroblocks[i].dct_type = macroblocks->macro_blocks[j].dct_type == XVMC_DCT_TYPE_FIELD ? vlDCTTypeFieldCoded : vlDCTTypeFrameCoded; for (k = 0; k < 2; ++k) for (l = 0; l < 2; ++l) -- cgit v1.2.3 From 6fca18696d0e6a243f6fbb5a30de45100a8e5fa0 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Wed, 17 Sep 2008 22:00:37 -0400 Subject: g3dvl: Update XvMC unit tests. --- src/libXvMC/context.c | 19 ++- src/libXvMC/tests/.gitignore | 2 +- src/libXvMC/tests/Makefile | 10 +- src/libXvMC/tests/test_blocks.c | 29 ++-- src/libXvMC/tests/test_context.c | 24 ++- src/libXvMC/tests/test_rendering.c | 322 ++++++++++++++++++++++++++----------- src/libXvMC/tests/testlib.c | 44 ++++- src/libXvMC/tests/testlib.h | 6 +- src/libXvMC/tests/xvmc_bench.c | 271 +++++++++++++++++++++++++++++++ 9 files changed, 581 insertions(+), 146 deletions(-) create mode 100644 src/libXvMC/tests/xvmc_bench.c diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c index 760e012d1a8..698410d4c2e 100644 --- a/src/libXvMC/context.c +++ b/src/libXvMC/context.c @@ -15,11 +15,11 @@ static Status Validate unsigned int width, unsigned int height, int flags, + int *found_port, int *chroma_format, int *mc_type ) { - unsigned int found_port = 0; unsigned int found_surface = 0; XvAdaptorInfo *adaptor_info; unsigned int num_adaptors; @@ -30,22 +30,24 @@ static Status Validate assert(display && chroma_format); + *found_port = 0; + ret = XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info); if (ret != Success) return ret; /* Scan through all adaptors looking for this port and surface */ - for (i = 0; i < num_adaptors && !found_port; ++i) + for (i = 0; i < num_adaptors && !*found_port; ++i) { /* Scan through all ports of this adaptor looking for our port */ - for (j = 0; j < adaptor_info[i].num_ports && !found_port; ++j) + for (j = 0; j < adaptor_info[i].num_ports && !*found_port; ++j) { /* If this is our port, scan through all its surfaces looking for our surface */ if (adaptor_info[i].base_id + j == port) { XvMCSurfaceInfo *surface_info; - found_port = 1; + *found_port = 1; surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types); if (surface_info) @@ -75,7 +77,7 @@ static Status Validate XvFreeAdaptorInfo(adaptor_info); - if (!found_port) + if (!*found_port) return XvBadPort; if (!found_surface) return BadMatch; @@ -127,6 +129,7 @@ static enum vlFormat FormatToVL(int xvmc_format) Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, int width, int height, int flags, XvMCContext *context) { + int found_port; int chroma_format; int mc_type; Status ret; @@ -140,8 +143,10 @@ Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, i if (!context) return XvMCBadContext; - ret = Validate(display, port, surface_type_id, width, height, flags, &chroma_format, &mc_type); - if (ret != Success) + ret = Validate(display, port, surface_type_id, width, height, flags, &found_port, &chroma_format, &mc_type); + + /* XXX: Success and XvBadPort have the same value */ + if (ret != Success || !found_port) return ret; /* XXX: Assumes default screen, should check which screen port is on */ diff --git a/src/libXvMC/tests/.gitignore b/src/libXvMC/tests/.gitignore index eb1ef8a0761..e1d2f9023df 100644 --- a/src/libXvMC/tests/.gitignore +++ b/src/libXvMC/tests/.gitignore @@ -2,4 +2,4 @@ test_context test_surface test_blocks test_rendering - +xvmc_bench diff --git a/src/libXvMC/tests/Makefile b/src/libXvMC/tests/Makefile index 2cbc97e88ba..de095161d29 100644 --- a/src/libXvMC/tests/Makefile +++ b/src/libXvMC/tests/Makefile @@ -1,4 +1,4 @@ -CFLAGS += -g -Wall -Werror +CFLAGS += -g -Wall LDFLAGS += LIBS += -lXvMCW -lXvMC -lXv @@ -6,7 +6,7 @@ LIBS += -lXvMCW -lXvMC -lXv .PHONY = all clean -all: test_context test_surface test_blocks test_rendering +all: test_context test_surface test_blocks test_rendering xvmc_bench test_context: test_context.o testlib.o $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} @@ -20,6 +20,8 @@ test_blocks: test_blocks.o testlib.o test_rendering: test_rendering.o testlib.o $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} -clean: - rm -rf *.o test_context test_surface test_blocks test_rendering +xvmc_bench: xvmc_bench.o testlib.o + $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} +clean: + rm -rf *.o test_context test_surface test_blocks test_rendering xvmc_bench diff --git a/src/libXvMC/tests/test_blocks.c b/src/libXvMC/tests/test_blocks.c index 0b895ee7737..dc80adfa652 100644 --- a/src/libXvMC/tests/test_blocks.c +++ b/src/libXvMC/tests/test_blocks.c @@ -7,7 +7,7 @@ int main(int argc, char **argv) const unsigned int width = 16, height = 16; const unsigned int min_required_blocks = 1, min_required_macroblocks = 1; const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; - + Display *display; XvPortID port_num; int surface_type_id; @@ -17,9 +17,9 @@ int main(int argc, char **argv) XvMCSurface surface; XvMCBlockArray blocks = {0}; XvMCMacroBlockArray macroblocks = {0}; - + display = XOpenDisplay(NULL); - + if (!GetPort ( display, @@ -37,25 +37,20 @@ int main(int argc, char **argv) XCloseDisplay(display); error(1, 0, "Error, unable to find a good port.\n"); } - + if (is_overlay) { Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); } - + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); assert(XvMCCreateSurface(display, &context, &surface) == Success); - + /* Test NULL context */ assert(XvMCCreateBlocks(display, NULL, 1, &blocks) == XvMCBadContext); /* Test 0 blocks */ assert(XvMCCreateBlocks(display, &context, 0, &blocks) == BadValue); - /* Test too many blocks */ - /*assert(XvMCCreateBlocks(display, &context, 16384, &blocks) == BadAlloc);*/ - - /* Note: No XvMCBadBlock(s) error in spec */ - /* Test valid params */ assert(XvMCCreateBlocks(display, &context, min_required_blocks, &blocks) == Success); /* Test context id assigned and correct */ @@ -68,11 +63,6 @@ int main(int argc, char **argv) assert(XvMCCreateMacroBlocks(display, NULL, 1, ¯oblocks) == XvMCBadContext); /* Test 0 macroblocks */ assert(XvMCCreateMacroBlocks(display, &context, 0, ¯oblocks) == BadValue); - /* Test too many macroblocks */ - /*assert(XvMCCreateMacroBlocks(display, &context, 16384, ¯oblocks) == BadAlloc);*/ - - /* Note: No XvMCBadMacroBlock(s) error in spec */ - /* Test valid params */ assert(XvMCCreateMacroBlocks(display, &context, min_required_macroblocks, ¯oblocks) == Success); /* Test context id assigned and correct */ @@ -83,13 +73,12 @@ int main(int argc, char **argv) assert(XvMCDestroyMacroBlocks(display, ¯oblocks) == Success); /* Test valid params */ assert(XvMCDestroyBlocks(display, &blocks) == Success); - + assert(XvMCDestroySurface(display, &surface) == Success); assert(XvMCDestroyContext(display, &context) == Success); - + XvUngrabPort(display, port_num, CurrentTime); XCloseDisplay(display); - + return 0; } - diff --git a/src/libXvMC/tests/test_context.c b/src/libXvMC/tests/test_context.c index 22afb7ada4e..53f7449cd0b 100644 --- a/src/libXvMC/tests/test_context.c +++ b/src/libXvMC/tests/test_context.c @@ -6,16 +6,16 @@ int main(int argc, char **argv) { const unsigned int width = 16, height = 16; const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; - + Display *display; XvPortID port_num; int surface_type_id; unsigned int is_overlay, intra_unsigned; int colorkey; XvMCContext context = {0}; - + display = XOpenDisplay(NULL); - + if (!GetPort ( display, @@ -33,22 +33,21 @@ int main(int argc, char **argv) XCloseDisplay(display); error(1, 0, "Error, unable to find a good port.\n"); } - + if (is_overlay) { Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); } - - /* Note: XvMCBadContext not a valid return for XvMCCreateContext in the XvMC API, but openChrome driver returns it */ - /* Note: Nvidia binary driver segfaults on NULL context, halts with debug output on bad port */ - + /* Test NULL context */ + /* XXX: XvMCBadContext not a valid return for XvMCCreateContext in the XvMC API, but openChrome driver returns it */ assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, NULL) == XvMCBadContext); /* Test invalid port */ - assert(XvMCCreateContext(display, port_num + 1, surface_type_id, width, height, XVMC_DIRECT, &context) == XvBadPort); + /* XXX: Success and XvBadPort have the same value, if this call actually gets passed the validation step as of now we'll crash later */ + assert(XvMCCreateContext(display, -1, surface_type_id, width, height, XVMC_DIRECT, &context) == XvBadPort); /* Test invalid surface */ - assert(XvMCCreateContext(display, port_num, surface_type_id + 1, width, height, XVMC_DIRECT, &context) == BadMatch); + assert(XvMCCreateContext(display, port_num, -1, width, height, XVMC_DIRECT, &context) == BadMatch); /* Test invalid flags */ assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, -1, &context) == BadValue); /* Test huge width */ @@ -85,10 +84,9 @@ int main(int argc, char **argv) assert(XvMCCreateContext(display, port_num, surface_type_id, width + 1, height + 1, XVMC_DIRECT, &context) == Success); assert(context.width >= width + 1 && context.height >= height + 1); assert(XvMCDestroyContext(display, &context) == Success); - + XvUngrabPort(display, port_num, CurrentTime); XCloseDisplay(display); - + return 0; } - diff --git a/src/libXvMC/tests/test_rendering.c b/src/libXvMC/tests/test_rendering.c index 1914b1fb628..1e7467a3aa6 100644 --- a/src/libXvMC/tests/test_rendering.c +++ b/src/libXvMC/tests/test_rendering.c @@ -1,38 +1,126 @@ #include #include +#include #include #include "testlib.h" +#define BLOCK_WIDTH 8 +#define BLOCK_HEIGHT 8 +#define BLOCK_SIZE (BLOCK_WIDTH * BLOCK_HEIGHT) +#define MACROBLOCK_WIDTH 16 +#define MACROBLOCK_HEIGHT 16 +#define MACROBLOCK_WIDTH_IN_BLOCKS (MACROBLOCK_WIDTH / BLOCK_WIDTH) +#define MACROBLOCK_HEIGHT_IN_BLOCKS (MACROBLOCK_HEIGHT / BLOCK_HEIGHT) +#define BLOCKS_PER_MACROBLOCK 6 + +#define INPUT_WIDTH 16 +#define INPUT_HEIGHT 16 +#define INPUT_WIDTH_IN_MACROBLOCKS (INPUT_WIDTH / MACROBLOCK_WIDTH) +#define INPUT_HEIGHT_IN_MACROBLOCKS (INPUT_HEIGHT / MACROBLOCK_HEIGHT) +#define NUM_MACROBLOCKS (INPUT_WIDTH_IN_MACROBLOCKS * INPUT_HEIGHT_IN_MACROBLOCKS) + +#define DEFAULT_OUTPUT_WIDTH INPUT_WIDTH +#define DEFAULT_OUTPUT_HEIGHT INPUT_HEIGHT +#define DEFAULT_ACCEPTABLE_ERR 0.01 + +void ParseArgs(int argc, char **argv, unsigned int *output_width, unsigned int *output_height, double *acceptable_error, int *prompt) +{ + int fail = 0; + int i; + + *output_width = DEFAULT_OUTPUT_WIDTH; + *output_height = DEFAULT_OUTPUT_WIDTH; + *acceptable_error = DEFAULT_ACCEPTABLE_ERR; + *prompt = 1; + + for (i = 1; i < argc && !fail; ++i) + { + if (!strcmp(argv[i], "-w")) + { + if (sscanf(argv[++i], "%u", output_width) != 1) + fail = 1; + } + else if (!strcmp(argv[i], "-h")) + { + if (sscanf(argv[++i], "%u", output_height) != 1) + fail = 1; + } + else if (!strcmp(argv[i], "-e")) + { + if (sscanf(argv[++i], "%lf", acceptable_error) != 1) + fail = 1; + } + else if (strcmp(argv[i], "-n")) + *prompt = 0; + else + fail = 1; + } + + if (fail) + error + ( + 1, 0, + "Bad argument.\n" + "\n" + "Usage: %s [options]\n" + "\t-w \tOutput width\n" + "\t-h \tOutput height\n" + "\t-e \tAcceptable margin of error per pixel, from 0 to 1\n" + "\t-n\tDon't prompt for quit\n", + argv[0] + ); +} + +void Gradient(short *block, unsigned int start, unsigned int stop, int horizontal) +{ + unsigned int x, y; + unsigned int range = stop - start; + + if (horizontal) + { + for (y = 0; y < BLOCK_HEIGHT; ++y) + for (x = 0; x < BLOCK_WIDTH; ++x) + block[y * BLOCK_WIDTH + x] = (short)(start + range * (x / (float)(BLOCK_WIDTH - 1))); + } + else + { + for (y = 0; y < BLOCK_HEIGHT; ++y) + for (x = 0; x < BLOCK_WIDTH; ++x) + block[y * BLOCK_WIDTH + x] = (short)(start + range * (y / (float)(BLOCK_HEIGHT - 1))); + } +} + int main(int argc, char **argv) { - const unsigned int width = 32, height = 32; - const unsigned int mwidth = width / 16, mheight = height / 16; - const unsigned int num_macroblocks = mwidth * mheight; - const unsigned int num_blocks = num_macroblocks * 6; - const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; - - int quit = 0; + unsigned int output_width; + unsigned int output_height; + double acceptable_error; + int prompt; Display *display; Window root, window; - Pixmap framebuffer; - XEvent event; + const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; XvPortID port_num; int surface_type_id; unsigned int is_overlay, intra_unsigned; int colorkey; XvMCContext context; XvMCSurface surface; - XvMCBlockArray blocks; - XvMCMacroBlockArray macroblocks; - unsigned int b, x, y; - + XvMCBlockArray block_array; + XvMCMacroBlockArray mb_array; + int mbx, mby, bx, by; + XvMCMacroBlock *mb; + short *blocks; + int quit = 0; + + ParseArgs(argc, argv, &output_width, &output_height, &acceptable_error, &prompt); + display = XOpenDisplay(NULL); - + if (!GetPort ( display, - width, - height, + INPUT_WIDTH, + INPUT_HEIGHT, XVMC_CHROMA_FORMAT_420, mc_types, 2, @@ -45,109 +133,155 @@ int main(int argc, char **argv) XCloseDisplay(display); error(1, 0, "Error, unable to find a good port.\n"); } - + if (is_overlay) { Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); } - + root = XDefaultRootWindow(display); - window = XCreateSimpleWindow(display, root, 0, 0, width, height, 0, 0, colorkey); - framebuffer = XCreatePixmap(display, root, width, height, 24); - - XSelectInput(display, window, ExposureMask | KeyPressMask); - XMapWindow(display, window); - XSync(display, 0); - - assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); + window = XCreateSimpleWindow(display, root, 0, 0, output_width, output_height, 0, 0, colorkey); + + assert(XvMCCreateContext(display, port_num, surface_type_id, INPUT_WIDTH, INPUT_HEIGHT, XVMC_DIRECT, &context) == Success); assert(XvMCCreateSurface(display, &context, &surface) == Success); - assert(XvMCCreateBlocks(display, &context, num_blocks, &blocks) == Success); - assert(XvMCCreateMacroBlocks(display, &context, num_macroblocks, ¯oblocks) == Success); - - for (b = 0; b < 6; ++b) - { - for (y = 0; y < 8; ++y) - { - for (x = 0; x < 8; ++x) - { - blocks.blocks[b * 64 + y * 8 + x] = 0xFFFF; - } - } - } - - for (y = 0; y < mheight; ++y) - { - for (x = 0; x < mwidth; ++x) + assert(XvMCCreateBlocks(display, &context, NUM_MACROBLOCKS * BLOCKS_PER_MACROBLOCK, &block_array) == Success); + assert(XvMCCreateMacroBlocks(display, &context, NUM_MACROBLOCKS, &mb_array) == Success); + + mb = mb_array.macro_blocks; + blocks = block_array.blocks; + + for (mby = 0; mby < INPUT_HEIGHT_IN_MACROBLOCKS; ++mby) + for (mbx = 0; mbx < INPUT_WIDTH_IN_MACROBLOCKS; ++mbx) { - macroblocks.macro_blocks[y * mwidth + x].x = x; - macroblocks.macro_blocks[y * mwidth + x].y = y; - macroblocks.macro_blocks[y * mwidth + x].index = (y * mwidth + x) * 6; - macroblocks.macro_blocks[y * mwidth + x].macroblock_type = XVMC_MB_TYPE_INTRA; - macroblocks.macro_blocks[y * mwidth + x].coded_block_pattern = 0x3F; - macroblocks.macro_blocks[y * mwidth + x].dct_type = XVMC_DCT_TYPE_FRAME; + mb->x = mbx; + mb->y = mby; + mb->macroblock_type = XVMC_MB_TYPE_INTRA; + /*mb->motion_type = ;*/ + /*mb->motion_vertical_field_select = ;*/ + mb->dct_type = XVMC_DCT_TYPE_FRAME; + /*mb->PMV[0][0][0] = ; + mb->PMV[0][0][1] = ; + mb->PMV[0][1][0] = ; + mb->PMV[0][1][1] = ; + mb->PMV[1][0][0] = ; + mb->PMV[1][0][1] = ; + mb->PMV[1][1][0] = ; + mb->PMV[1][1][1] = ;*/ + mb->index = (mby * INPUT_WIDTH_IN_MACROBLOCKS + mbx) * BLOCKS_PER_MACROBLOCK; + mb->coded_block_pattern = 0x3F; + + mb++; + + for (by = 0; by < MACROBLOCK_HEIGHT_IN_BLOCKS; ++by) + for (bx = 0; bx < MACROBLOCK_WIDTH_IN_BLOCKS; ++bx) + { + const int start = 16, stop = 235, range = stop - start; + + Gradient + ( + blocks, + (short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH) / (float)(INPUT_WIDTH - 1))), + (short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH + BLOCK_WIDTH - 1) / (float)(INPUT_WIDTH - 1))), + 1 + ); + + blocks += BLOCK_SIZE; + } + + for (by = 0; by < MACROBLOCK_HEIGHT_IN_BLOCKS / 2; ++by) + for (bx = 0; bx < MACROBLOCK_WIDTH_IN_BLOCKS / 2; ++bx) + { + const int start = 16, stop = 240, range = stop - start; + + Gradient + ( + blocks, + (short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH) / (float)(INPUT_WIDTH - 1))), + (short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH + BLOCK_WIDTH - 1) / (float)(INPUT_WIDTH - 1))), + 1 + ); + + blocks += BLOCK_SIZE; + + Gradient + ( + blocks, + (short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH) / (float)(INPUT_WIDTH - 1))), + (short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH + BLOCK_WIDTH - 1) / (float)(INPUT_WIDTH - 1))), + 1 + ); + + blocks += BLOCK_SIZE; + } } - } - + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + /* Test NULL context */ - assert(XvMCRenderSurface(display, NULL, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, 1, 0, ¯oblocks, &blocks) == XvMCBadContext); + assert(XvMCRenderSurface(display, NULL, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == XvMCBadContext); /* Test NULL surface */ - assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, NULL, NULL, NULL, 0, 1, 0, ¯oblocks, &blocks) == XvMCBadSurface); + assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, NULL, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == XvMCBadSurface); /* Test bad picture structure */ - assert(XvMCRenderSurface(display, &context, 0, &surface, NULL, NULL, 0, 1, 0, ¯oblocks, &blocks) == BadValue); + assert(XvMCRenderSurface(display, &context, 0, &surface, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == BadValue); /* Test valid params */ - assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, num_macroblocks, 0, ¯oblocks, &blocks) == Success); - + assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == Success); + /* Test NULL surface */ - assert(XvMCPutSurface(display, NULL, window, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == XvMCBadSurface); + assert(XvMCPutSurface(display, NULL, window, 0, 0, INPUT_WIDTH, INPUT_HEIGHT, 0, 0, output_width, output_height, XVMC_FRAME_PICTURE) == XvMCBadSurface); /* Test bad window */ - /* X halts with a bad drawable for some reason, doesn't return BadDrawable as expected */ + /* XXX: X halts with a bad drawable for some reason, doesn't return BadDrawable as expected */ /*assert(XvMCPutSurface(display, &surface, 0, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == BadDrawable);*/ - /* Test valid params */ - assert(XvMCPutSurface(display, &surface, framebuffer, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == Success); - - puts("Press any key to continue..."); - - while (!quit) + + if (prompt) { - XNextEvent(display, &event); - switch (event.type) + puts("Press any button to quit..."); + + while (!quit) { - case Expose: - { - XCopyArea - ( - display, - framebuffer, - window, - XDefaultGC(display, XDefaultScreen(display)), - 0, - 0, - width, - height, - 0, - 0 - ); - break; - } - case KeyPress: + if (XPending(display) > 0) { - quit = 1; - break; + XEvent event; + + XNextEvent(display, &event); + + switch (event.type) + { + case Expose: + { + /* Test valid params */ + assert + ( + XvMCPutSurface + ( + display, &surface, window, + 0, 0, INPUT_WIDTH, INPUT_HEIGHT, + 0, 0, output_width, output_height, + XVMC_FRAME_PICTURE + ) == Success + ); + break; + } + case KeyPress: + { + quit = 1; + break; + } + } } } } - - assert(XvMCDestroyBlocks(display, &blocks) == Success); - assert(XvMCDestroyMacroBlocks(display, ¯oblocks) == Success); - assert(XvMCDestroySurface(display, &surface) == Success); + + assert(XvMCDestroyBlocks(display, &block_array) == Success); + assert(XvMCDestroyMacroBlocks(display, &mb_array) == Success); + assert(XvMCDestroySurface(display, &surface) == Success); assert(XvMCDestroyContext(display, &context) == Success); - - XFreePixmap(display, framebuffer); + XvUngrabPort(display, port_num, CurrentTime); XDestroyWindow(display, window); XCloseDisplay(display); - + return 0; } - diff --git a/src/libXvMC/tests/testlib.c b/src/libXvMC/tests/testlib.c index 8672aa99998..59a03ca813f 100644 --- a/src/libXvMC/tests/testlib.c +++ b/src/libXvMC/tests/testlib.c @@ -32,18 +32,18 @@ int GetPort int num_types; int ev_base, err_base; unsigned int i, j, k, l; - + if (!XvMCQueryExtension(display, &ev_base, &err_base)) return 0; if (XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info) != Success) return 0; - + for (i = 0; i < num_adaptors && !found_port; ++i) { if (adaptor_info[i].type & XvImageMask) { XvMCSurfaceInfo *surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types); - + if (surface_info) { for (j = 0; j < num_types && !found_port; ++j) @@ -74,14 +74,46 @@ int GetPort } } } - + XFree(surface_info); } } } - + XvFreeAdaptorInfo(adaptor_info); - + return found_port; } +unsigned int align(unsigned int value, unsigned int alignment) +{ + return (value + alignment - 1) & ~(alignment - 1); +} + +/* From the glibc manual */ +int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y) +{ + /* Perform the carry for the later subtraction by updating y. */ + if (x->tv_usec < y->tv_usec) + { + int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1; + y->tv_usec -= 1000000 * nsec; + y->tv_sec += nsec; + } + if (x->tv_usec - y->tv_usec > 1000000) + { + int nsec = (x->tv_usec - y->tv_usec) / 1000000; + y->tv_usec += 1000000 * nsec; + y->tv_sec -= nsec; + } + + /* + * Compute the time remaining to wait. + * tv_usec is certainly positive. + */ + result->tv_sec = x->tv_sec - y->tv_sec; + result->tv_usec = x->tv_usec - y->tv_usec; + + /* Return 1 if result is negative. */ + return x->tv_sec < y->tv_sec; +} diff --git a/src/libXvMC/tests/testlib.h b/src/libXvMC/tests/testlib.h index c73845807ec..af71ad74e17 100644 --- a/src/libXvMC/tests/testlib.h +++ b/src/libXvMC/tests/testlib.h @@ -7,6 +7,7 @@ void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line); */ +#include #include #include @@ -34,5 +35,8 @@ int GetPort unsigned int *intra_unsigned ); -#endif +unsigned int align(unsigned int value, unsigned int alignment); + +int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y); +#endif diff --git a/src/libXvMC/tests/xvmc_bench.c b/src/libXvMC/tests/xvmc_bench.c new file mode 100644 index 00000000000..16f6c21ae31 --- /dev/null +++ b/src/libXvMC/tests/xvmc_bench.c @@ -0,0 +1,271 @@ +#include +#include +#include +#include +#include +#include "testlib.h" + +#define MACROBLOCK_WIDTH 16 +#define MACROBLOCK_HEIGHT 16 +#define BLOCKS_PER_MACROBLOCK 6 + +#define DEFAULT_INPUT_WIDTH 720 +#define DEFAULT_INPUT_HEIGHT 480 +#define DEFAULT_REPS 100 + +#define PIPELINE_STEP_MC 1 +#define PIPELINE_STEP_CSC 2 +#define PIPELINE_STEP_SWAP 4 + +#define MB_TYPE_I 1 +#define MB_TYPE_P 2 +#define MB_TYPE_B 4 + +struct Config +{ + unsigned int input_width; + unsigned int input_height; + unsigned int output_width; + unsigned int output_height; + unsigned int pipeline; + unsigned int mb_types; + unsigned int reps; +}; + +void ParseArgs(int argc, char **argv, struct Config *config) +{ + int fail = 0; + int i; + + config->input_width = DEFAULT_INPUT_WIDTH; + config->input_height = DEFAULT_INPUT_HEIGHT; + config->output_width = 0; + config->output_height = 0; + config->pipeline = 0; + config->mb_types = 0; + config->reps = DEFAULT_REPS; + + for (i = 1; i < argc && !fail; ++i) + { + if (!strcmp(argv[i], "-iw")) + { + if (sscanf(argv[++i], "%u", &config->input_width) != 1) + fail = 1; + } + else if (!strcmp(argv[i], "-ih")) + { + if (sscanf(argv[++i], "%u", &config->input_height) != 1) + fail = 1; + } + else if (!strcmp(argv[i], "-ow")) + { + if (sscanf(argv[++i], "%u", &config->output_width) != 1) + fail = 1; + } + else if (!strcmp(argv[i], "-oh")) + { + if (sscanf(argv[++i], "%u", &config->output_height) != 1) + fail = 1; + } + else if (!strcmp(argv[i], "-p")) + { + char *token = strtok(argv[++i], ","); + + while (token && !fail) + { + if (!strcmp(token, "mc")) + config->pipeline |= PIPELINE_STEP_MC; + else if (!strcmp(token, "csc")) + config->pipeline |= PIPELINE_STEP_CSC; + else if (!strcmp(token, "swp")) + config->pipeline |= PIPELINE_STEP_SWAP; + else + fail = 1; + + if (!fail) + token = strtok(NULL, ","); + } + } + else if (!strcmp(argv[i], "-mb")) + { + char *token = strtok(argv[++i], ","); + + while (token && !fail) + { + if (strcmp(token, "i")) + config->mb_types |= MB_TYPE_I; + else if (strcmp(token, "p")) + config->mb_types |= MB_TYPE_P; + else if (strcmp(token, "b")) + config->mb_types |= MB_TYPE_B; + else + fail = 1; + + if (!fail) + token = strtok(NULL, ","); + } + } + else if (!strcmp(argv[i], "-r")) + { + if (sscanf(argv[++i], "%u", &config->reps) != 1) + fail = 1; + } + else + fail = 1; + } + + if (fail) + error + ( + 1, 0, + "Bad argument.\n" + "\n" + "Usage: %s [options]\n" + "\t-iw \tInput width\n" + "\t-ih \tInput height\n" + "\t-ow \tOutput width\n" + "\t-oh \tOutput height\n" + "\t-p \tPipeline to test\n" + "\t-mb \tMacroBlock types to use\n" + "\t-r \tRepetitions\n\n" + "\tPipeline steps: mc,csc,swap\n" + "\tMB types: i,p,b\n", + argv[0] + ); + + if (config->output_width == 0) + config->output_width = config->input_width; + if (config->output_height == 0) + config->output_height = config->input_height; + if (!config->pipeline) + config->pipeline = PIPELINE_STEP_MC | PIPELINE_STEP_CSC | PIPELINE_STEP_SWAP; + if (!config->mb_types) + config->mb_types = MB_TYPE_I | MB_TYPE_P | MB_TYPE_B; +} + +int main(int argc, char **argv) +{ + struct Config config; + Display *display; + Window root, window; + const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; + XvPortID port_num; + int surface_type_id; + unsigned int is_overlay, intra_unsigned; + int colorkey; + XvMCContext context; + XvMCSurface surface; + XvMCBlockArray block_array; + XvMCMacroBlockArray mb_array; + unsigned int mbw, mbh; + unsigned int mbx, mby; + unsigned int reps; + struct timeval start, stop, diff; + double diff_secs; + + ParseArgs(argc, argv, &config); + + mbw = align(config.input_width, MACROBLOCK_WIDTH) / MACROBLOCK_WIDTH; + mbh = align(config.input_height, MACROBLOCK_HEIGHT) / MACROBLOCK_HEIGHT; + + display = XOpenDisplay(NULL); + + if (!GetPort + ( + display, + config.input_width, + config.input_height, + XVMC_CHROMA_FORMAT_420, + mc_types, + 2, + &port_num, + &surface_type_id, + &is_overlay, + &intra_unsigned + )) + { + XCloseDisplay(display); + error(1, 0, "Error, unable to find a good port.\n"); + } + + if (is_overlay) + { + Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); + XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); + } + + root = XDefaultRootWindow(display); + window = XCreateSimpleWindow(display, root, 0, 0, config.output_width, config.output_height, 0, 0, colorkey); + + assert(XvMCCreateContext(display, port_num, surface_type_id, config.input_width, config.input_width, XVMC_DIRECT, &context) == Success); + assert(XvMCCreateSurface(display, &context, &surface) == Success); + assert(XvMCCreateBlocks(display, &context, mbw * mbh * BLOCKS_PER_MACROBLOCK, &block_array) == Success); + assert(XvMCCreateMacroBlocks(display, &context, mbw * mbh, &mb_array) == Success); + + for (mby = 0; mby < mbh; ++mby) + for (mbx = 0; mbx < mbw; ++mbx) + { + mb_array.macro_blocks[mby * mbw + mbx].x = mbx; + mb_array.macro_blocks[mby * mbw + mbx].y = mby; + mb_array.macro_blocks[mby * mbw + mbx].macroblock_type = XVMC_MB_TYPE_INTRA; + /*mb->motion_type = ;*/ + /*mb->motion_vertical_field_select = ;*/ + mb_array.macro_blocks[mby * mbw + mbx].dct_type = XVMC_DCT_TYPE_FRAME; + /*mb->PMV[0][0][0] = ; + mb->PMV[0][0][1] = ; + mb->PMV[0][1][0] = ; + mb->PMV[0][1][1] = ; + mb->PMV[1][0][0] = ; + mb->PMV[1][0][1] = ; + mb->PMV[1][1][0] = ; + mb->PMV[1][1][1] = ;*/ + mb_array.macro_blocks[mby * mbw + mbx].index = (mby * mbw + mbx) * BLOCKS_PER_MACROBLOCK; + mb_array.macro_blocks[mby * mbw + mbx].coded_block_pattern = 0x3F; + } + + XSelectInput(display, window, ExposureMask | KeyPressMask); + XMapWindow(display, window); + XSync(display, 0); + + gettimeofday(&start, NULL); + + for (reps = 0; reps < config.reps; ++reps) + { + if (config.pipeline & PIPELINE_STEP_MC) + { + assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, mbw * mbh, 0, &mb_array, &block_array) == Success); + assert(XvMCFlushSurface(display, &surface) == Success); + } + if (config.pipeline & PIPELINE_STEP_CSC) + assert(XvMCPutSurface(display, &surface, window, 0, 0, config.input_width, config.input_height, 0, 0, config.output_width, config.output_height, XVMC_FRAME_PICTURE) == Success); + } + + gettimeofday(&stop, NULL); + + timeval_subtract(&diff, &stop, &start); + diff_secs = (double)diff.tv_sec + (double)diff.tv_usec / 1000000.0; + + printf("XvMC Benchmark\n"); + printf("Input: %u,%u\nOutput: %u,%u\n", config.input_width, config.input_height, config.output_width, config.output_height); + printf("Pipeline: "); + if (config.pipeline & PIPELINE_STEP_MC) + printf("|mc|"); + if (config.pipeline & PIPELINE_STEP_CSC) + printf("|csc|"); + if (config.pipeline & PIPELINE_STEP_SWAP) + printf("|swap|"); + printf("\n"); + printf("Reps: %u\n", config.reps); + printf("Total time: %.2lf (%.2lf reps / sec)\n", diff_secs, config.reps / diff_secs); + + assert(XvMCDestroyBlocks(display, &block_array) == Success); + assert(XvMCDestroyMacroBlocks(display, &mb_array) == Success); + assert(XvMCDestroySurface(display, &surface) == Success); + assert(XvMCDestroyContext(display, &context) == Success); + + XvUngrabPort(display, port_num, CurrentTime); + XDestroyWindow(display, window); + XCloseDisplay(display); + + return 0; +} -- cgit v1.2.3 From 562b31195cc8140d3bd0ad7ed8532200e9df520a Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 18 Sep 2008 13:15:55 +1000 Subject: nv50: hack surface alloc a bit for now --- src/gallium/drivers/nv50/nv50_miptree.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index b0e8fe2f0b7..28a8bdc0fab 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -48,6 +48,8 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) } pitch = ((pt->width[0] + 63) & ~63) * pt->block.size; + /*XXX*/ + pitch *= 2; mt->buffer = ws->buffer_create(ws, 256, usage, pitch * pt->height[0]); if (!mt->buffer) { -- cgit v1.2.3 From ed0c308f39f7d7a5cb05571e37440469eacbd624 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Thu, 18 Sep 2008 13:39:52 +1000 Subject: nv50: use 3D engine clears, 2D engine doesn't understand zeta formats --- src/gallium/drivers/nv50/nv50_clear.c | 57 ++++++++++++++++++++++++++++++++++- 1 file changed, 56 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nv50/nv50_clear.c b/src/gallium/drivers/nv50/nv50_clear.c index fbc6cd09cee..a31a42d6b54 100644 --- a/src/gallium/drivers/nv50/nv50_clear.c +++ b/src/gallium/drivers/nv50/nv50_clear.c @@ -30,6 +30,61 @@ void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { - pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); + struct nv50_context *nv50 = nv50_context(pipe); + struct pipe_framebuffer_state fb, s_fb = nv50->framebuffer; + struct pipe_scissor_state sc, s_sc = nv50->scissor; + unsigned dirty = nv50->dirty; + + nv50->dirty = 0; + + if (ps->format == PIPE_FORMAT_Z24S8_UNORM || + ps->format == PIPE_FORMAT_Z16_UNORM) { + fb.num_cbufs = 0; + fb.zsbuf = ps; + } else { + fb.num_cbufs = 1; + fb.cbufs[0] = ps; + fb.zsbuf = NULL; + } + fb.width = ps->width; + fb.height = ps->height; + pipe->set_framebuffer_state(pipe, &fb); + + sc.minx = sc.miny = 0; + sc.maxx = fb.width; + sc.maxy = fb.height; + pipe->set_scissor_state(pipe, &sc); + + nv50_state_validate(nv50); + + switch (ps->format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + BEGIN_RING(tesla, 0x0d80, 4); + OUT_RINGf (ubyte_to_float((clearValue >> 16) & 0xff)); + OUT_RINGf (ubyte_to_float((clearValue >> 8) & 0xff)); + OUT_RINGf (ubyte_to_float((clearValue >> 0) & 0xff)); + OUT_RINGf (ubyte_to_float((clearValue >> 24) & 0xff)); + BEGIN_RING(tesla, 0x19d0, 1); + OUT_RING (0x3c); + break; + case PIPE_FORMAT_Z24S8_UNORM: + BEGIN_RING(tesla, 0x0d90, 1); + OUT_RINGf ((float)(clearValue >> 8) * (1.0 / 16777215.0)); + BEGIN_RING(tesla, 0x0da0, 1); + OUT_RING (clearValue & 0xff); + BEGIN_RING(tesla, 0x19d0, 1); + OUT_RING (0x03); + break; + default: + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, + clearValue); + break; + } + + pipe->set_framebuffer_state(pipe, &s_fb); + pipe->set_scissor_state(pipe, &s_sc); + nv50->dirty |= dirty; + ps->status = PIPE_SURFACE_STATUS_CLEAR; } + -- cgit v1.2.3 From 70071484d95bed8c2c932d1c79e20230bcdbc1dc Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sun, 21 Sep 2008 14:02:30 +0200 Subject: nouveau: add flag for swizzled surface upload --- src/gallium/drivers/nouveau/nouveau_bo.h | 1 + src/gallium/drivers/nv30/nv30_fragtex.c | 34 ++++++++++++++++++-------------- 2 files changed, 20 insertions(+), 15 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_bo.h b/src/gallium/drivers/nouveau/nouveau_bo.h index 65b138283c4..0ed3367815a 100644 --- a/src/gallium/drivers/nouveau/nouveau_bo.h +++ b/src/gallium/drivers/nouveau/nouveau_bo.h @@ -37,6 +37,7 @@ #define NOUVEAU_BO_LOCAL (1 << 9) #define NOUVEAU_BO_TILED (1 << 10) #define NOUVEAU_BO_ZTILE (1 << 11) +#define NOUVEAU_BO_SWIZZLED (1 << 12) #define NOUVEAU_BO_DUMMY (1 << 31) struct nouveau_bo { diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c index 91246f5ca0a..9e6f746a420 100644 --- a/src/gallium/drivers/nv30/nv30_fragtex.c +++ b/src/gallium/drivers/nv30/nv30_fragtex.c @@ -26,7 +26,7 @@ static INLINE int log2i(int i) return r; } -#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \ +#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w,swsurf) \ { \ TRUE, \ PIPE_FORMAT_##m, \ @@ -35,6 +35,7 @@ static INLINE int log2i(int i) NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w | \ NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y | \ NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w), \ + swsurf \ } struct nv30_texture_format { @@ -42,24 +43,25 @@ struct nv30_texture_format { uint pipe; int format; int swizzle; + int swizzled_surface; }; static struct nv30_texture_format nv30_texture_formats[] = { - _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W), - _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W), - _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W), - _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W), - _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X), - _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X), - _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X), - _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y), -// _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X), -// _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X), - _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W), - _(DXT1_RGBA , DXT1 , S1, S1, S1, S1, X, Y, Z, W), - _(DXT3_RGBA , DXT3 , S1, S1, S1, S1, X, Y, Z, W), - _(DXT5_RGBA , DXT5 , S1, S1, S1, S1, X, Y, Z, W), + _(A8R8G8B8_UNORM, A8R8G8B8, S1, S1, S1, S1, X, Y, Z, W, 1), + _(A1R5G5B5_UNORM, A1R5G5B5, S1, S1, S1, S1, X, Y, Z, W, 1), + _(A4R4G4B4_UNORM, A4R4G4B4, S1, S1, S1, S1, X, Y, Z, W, 1), + _(R5G6B5_UNORM , R5G6B5 , S1, S1, S1, ONE, X, Y, Z, W, 1), + _(L8_UNORM , L8 , S1, S1, S1, ONE, X, X, X, X, 1), + _(A8_UNORM , L8 , ZERO, ZERO, ZERO, S1, X, X, X, X, 1), + _(I8_UNORM , L8 , S1, S1, S1, S1, X, X, X, X, 1), + _(A8L8_UNORM , A8L8 , S1, S1, S1, S1, X, X, X, Y, 1), +// _(Z16_UNORM , Z16 , S1, S1, S1, ONE, X, X, X, X, 0), +// _(Z24S8_UNORM , Z24 , S1, S1, S1, ONE, X, X, X, X, 0), + _(DXT1_RGB , DXT1 , S1, S1, S1, ONE, X, Y, Z, W, 0), + _(DXT1_RGBA , DXT1 , S1, S1, S1, S1, X, Y, Z, W, 0), + _(DXT3_RGBA , DXT3 , S1, S1, S1, S1, X, Y, Z, W, 0), + _(DXT5_RGBA , DXT5 , S1, S1, S1, S1, X, Y, Z, W, 0), {}, }; @@ -96,6 +98,8 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit) if (!tf) assert(0); + tex_flags |= (tf->swizzled_surface ? NOUVEAU_BO_SWIZZLED : 0); + txf = tf->format; txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0); txf |= log2i(pt->width[0]) << 20; -- cgit v1.2.3 From da85a94d99cfbd8f094a475fb2272efb4e87b9a9 Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Sun, 21 Sep 2008 14:12:04 +0200 Subject: nouveau: init scaled_image_from_memory object for swizzled surface upload --- src/gallium/winsys/drm/nouveau/nv04_surface.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/src/gallium/winsys/drm/nouveau/nv04_surface.c b/src/gallium/winsys/drm/nouveau/nv04_surface.c index 8fa3d106c8c..36c913a7329 100644 --- a/src/gallium/winsys/drm/nouveau/nv04_surface.c +++ b/src/gallium/winsys/drm/nouveau/nv04_surface.c @@ -298,6 +298,22 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) } BIND_RING (chan, nvc->NvSIFM, nvc->next_subchannel++); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE, 1); + OUT_RING (chan, nvc->channel->vram->handle); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE, 1); + OUT_RING (chan, nvc->NvSwzSurf); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_ROP, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_BETA1, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_BETA4, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION, 1); + OUT_RING (chan, NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY); return 0; } -- cgit v1.2.3 From 2882a2db7a766c60bb231978ea829632438dd8a4 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Tue, 30 Sep 2008 12:06:56 +1000 Subject: nouveau: some small API changes --- src/gallium/winsys/drm/nouveau/nouveau_screen.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/src/gallium/winsys/drm/nouveau/nouveau_screen.c b/src/gallium/winsys/drm/nouveau/nouveau_screen.c index 94896451519..c6d0c535887 100644 --- a/src/gallium/winsys/drm/nouveau/nouveau_screen.c +++ b/src/gallium/winsys/drm/nouveau/nouveau_screen.c @@ -114,7 +114,7 @@ nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv) struct nouveau_framebuffer *nvfb; nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate; - st_unreference_framebuffer(&nvfb->stfb); + st_unreference_framebuffer(nvfb->stfb); free(nvfb); } @@ -133,8 +133,9 @@ nouveau_fill_in_modes(__DRIscreenPrivate *psp, GLX_NONE, GLX_SWAP_UNDEFINED_OML, }; - u_int8_t depth_bits_array[3]; - u_int8_t stencil_bits_array[3]; + uint8_t depth_bits_array[3]; + uint8_t stencil_bits_array[3]; + uint8_t msaa_samples_array[1]; depth_bits_array[0] = 0; depth_bits_array[1] = depth_bits; @@ -148,9 +149,10 @@ nouveau_fill_in_modes(__DRIscreenPrivate *psp, stencil_bits_array[1] = 0; if (depth_bits == 24) stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits; - stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits; + msaa_samples_array[0] = 0; + depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1; back_buffer_factor = (have_back_buffer) ? 3 : 1; @@ -167,7 +169,7 @@ nouveau_fill_in_modes(__DRIscreenPrivate *psp, configs = driCreateConfigs(fb_format, fb_type, depth_bits_array, stencil_bits_array, depth_buffer_factor, back_buffer_modes, - back_buffer_factor); + back_buffer_factor, msaa_samples_array, 1); if (configs == NULL) { fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__, __LINE__); -- cgit v1.2.3 From 111b8f6dd9c97cd30979c8d5f56244e1e6ed60a2 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 22 Sep 2008 12:13:23 -0400 Subject: g3dvl: Bad semantic index in shader decl. --- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 888f0040bf7..62107803ac0 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -974,7 +974,7 @@ static int vlCreateFragmentShaderIMB */ for (i = 0; i < 3; ++i) { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, i, i, TGSI_INTERPOLATE_LINEAR); + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } @@ -1186,7 +1186,7 @@ static int vlCreateVertexShaderFieldPMB */ for (i = 0; i < 7; i++) { - decl = vl_decl_output((i == 0 || i == 6) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } @@ -1668,7 +1668,7 @@ static int vlCreateVertexShaderFieldBMB */ for (i = 0; i < 9; i++) { - decl = vl_decl_output((i == 0 || i == 8) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); } -- cgit v1.2.3 From d008fb178631aecacc07aeec66299748470fd8c7 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Mon, 22 Sep 2008 12:26:13 -0400 Subject: g3dvl: Ignore client flush requests unless entire frame is buffered. --- src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c | 10 ++++++++++ src/gallium/state_trackers/g3dvl/vl_surface.c | 1 + 2 files changed, 11 insertions(+) diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c index 62107803ac0..e7a070ef4df 100644 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c @@ -576,6 +576,9 @@ static int vlFlush unsigned int num_macroblocks[vlNumMacroBlockExTypes] = {0}; unsigned int offset[vlNumMacroBlockExTypes]; unsigned int vb_start = 0; + unsigned int mbw; + unsigned int mbh; + unsigned int num_mb_per_frame; unsigned int i; assert(render); @@ -585,6 +588,13 @@ static int vlFlush if (!mc->buffered_surface) return 0; + mbw = align(mc->picture_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH; + mbh = align(mc->picture_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT; + num_mb_per_frame = mbw * mbh; + + if (mc->num_macroblocks < num_mb_per_frame) + return 0; + pipe = mc->pipe; for (i = 0; i < mc->num_macroblocks; ++i) diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c index 6648133ef83..076bd40d414 100644 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ b/src/gallium/state_trackers/g3dvl/vl_surface.c @@ -179,6 +179,7 @@ int vlSurfaceGetStatus } *status = vlResourceStatusFree; + return 0; } -- cgit v1.2.3 From af59f767a03da4dd434a6d655e3d0bc05feb298a Mon Sep 17 00:00:00 2001 From: Patrice Mandin Date: Tue, 7 Oct 2008 21:47:45 +0200 Subject: nouveau: first draft of swizzle texture upload to vram --- src/gallium/winsys/drm/nouveau/nv04_surface.c | 83 +++++++++++++++++++++++++++ 1 file changed, 83 insertions(+) diff --git a/src/gallium/winsys/drm/nouveau/nv04_surface.c b/src/gallium/winsys/drm/nouveau/nv04_surface.c index 36c913a7329..5bf89e1952e 100644 --- a/src/gallium/winsys/drm/nouveau/nv04_surface.c +++ b/src/gallium/winsys/drm/nouveau/nv04_surface.c @@ -3,6 +3,32 @@ #include "nouveau_context.h" +static INLINE int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + static INLINE int nv04_surface_format(enum pipe_format format) { @@ -81,6 +107,56 @@ nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy, OUT_RING (chan, ( h << 16) | w); } +static int +nv04_surface_copy_prep_swizzled(struct nouveau_context *nv, + struct pipe_surface *dst, + struct pipe_surface *src) +{ + struct nouveau_channel *chan = nv->nvc->channel; + + BEGIN_RING(chan, nv->nvc->NvSwzSurf, + NV04_SWIZZLED_SURFACE_FORMAT, 2); + /* FIXME: read destination format from somewhere */ + OUT_RING (chan, + NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8 + | (log2i(dst->width)<height)<buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + BEGIN_RING(chan, nv->nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION, 13); + OUT_RING (chan, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE); + /* FIXME: read source format from somewhere */ + OUT_RING (chan, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8); + OUT_RING (chan, NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY); + OUT_RING (chan, 0); + OUT_RING (chan, (src->height<<16) | src->width); + OUT_RING (chan, 0); + OUT_RING (chan, (src->height<<16) | src->width); + OUT_RING (chan, 1<<20); + OUT_RING (chan, 1<<20); + OUT_RING (chan, (src->height<<16) | src->width); + OUT_RING (chan, + src->stride + | NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER + | NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + OUT_RING (chan, 0); + + BEGIN_RING(chan, nv->nvc->NvM2MF, + NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); + OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); + OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo, + NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + nv->surface_copy = nv04_surface_copy_m2mf; + nv->surf_dst = dst; + nv->surf_src = src; + return 0; +} + static int nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, struct pipe_surface *src) @@ -91,6 +167,13 @@ nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst, if (src->format != dst->format) return 1; + /* Setup transfer to swizzle the texture to vram if needed */ + /* FIXME/TODO: check proper limits of this operation */ + if (nouveau_buffer(dst->buffer)->bo->flags & NOUVEAU_BO_SWIZZLED) { + /* FIXME: Disable it for the moment */ + /*return nv04_surface_copy_prep_swizzled(nv, dst, src);*/ + } + /* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback * to NV_MEMORY_TO_MEMORY_FORMAT in this case. */ -- cgit v1.2.3 From e52630e296807282592a50a3a4c7a9b50a56adb0 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Sat, 4 Oct 2008 15:38:12 -0400 Subject: g3dvl: Bad test, use proper picture height. --- src/libXvMC/tests/xvmc_bench.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/libXvMC/tests/xvmc_bench.c b/src/libXvMC/tests/xvmc_bench.c index 16f6c21ae31..d5a39ecf17d 100644 --- a/src/libXvMC/tests/xvmc_bench.c +++ b/src/libXvMC/tests/xvmc_bench.c @@ -197,7 +197,7 @@ int main(int argc, char **argv) root = XDefaultRootWindow(display); window = XCreateSimpleWindow(display, root, 0, 0, config.output_width, config.output_height, 0, 0, colorkey); - assert(XvMCCreateContext(display, port_num, surface_type_id, config.input_width, config.input_width, XVMC_DIRECT, &context) == Success); + assert(XvMCCreateContext(display, port_num, surface_type_id, config.input_width, config.input_height, XVMC_DIRECT, &context) == Success); assert(XvMCCreateSurface(display, &context, &surface) == Success); assert(XvMCCreateBlocks(display, &context, mbw * mbh * BLOCKS_PER_MACROBLOCK, &block_array) == Success); assert(XvMCCreateMacroBlocks(display, &context, mbw * mbh, &mb_array) == Success); -- cgit v1.2.3 From 3354e668f0d4eb7ad1d92607031c1fc2e785e8d1 Mon Sep 17 00:00:00 2001 From: Younes Manton Date: Fri, 17 Oct 2008 17:20:03 -0400 Subject: g3dvl: Get rid of state tracker unit tests, stale and not useful atm. --- src/gallium/state_trackers/g3dvl/tests/.gitignore | 6 - src/gallium/state_trackers/g3dvl/tests/Makefile | 45 ---- .../state_trackers/g3dvl/tests/test_b_rendering.c | 226 --------------------- .../state_trackers/g3dvl/tests/test_context.c | 22 -- .../state_trackers/g3dvl/tests/test_i_rendering.c | 137 ------------- .../state_trackers/g3dvl/tests/test_p_rendering.c | 214 ------------------- .../state_trackers/g3dvl/tests/test_pf_rendering.c | 214 ------------------- .../state_trackers/g3dvl/tests/test_surface.c | 26 --- 8 files changed, 890 deletions(-) delete mode 100644 src/gallium/state_trackers/g3dvl/tests/.gitignore delete mode 100644 src/gallium/state_trackers/g3dvl/tests/Makefile delete mode 100644 src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c delete mode 100644 src/gallium/state_trackers/g3dvl/tests/test_context.c delete mode 100644 src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c delete mode 100644 src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c delete mode 100644 src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c delete mode 100644 src/gallium/state_trackers/g3dvl/tests/test_surface.c diff --git a/src/gallium/state_trackers/g3dvl/tests/.gitignore b/src/gallium/state_trackers/g3dvl/tests/.gitignore deleted file mode 100644 index 9b1ec4e2122..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/.gitignore +++ /dev/null @@ -1,6 +0,0 @@ -test_context -test_surface -test_i_rendering -test_p_rendering -test_pf_rendering -test_b_rendering diff --git a/src/gallium/state_trackers/g3dvl/tests/Makefile b/src/gallium/state_trackers/g3dvl/tests/Makefile deleted file mode 100644 index 45cefa2e570..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/Makefile +++ /dev/null @@ -1,45 +0,0 @@ -GALLIUMDIR = ../../.. - -CFLAGS += -g -Wall -Werror \ - -I${GALLIUMDIR}/state_trackers/g3dvl \ - -I${GALLIUMDIR}/winsys/g3dvl \ - -I${GALLIUMDIR}/include \ - -I${GALLIUMDIR}/auxiliary \ - -I${GALLIUMDIR}/drivers -LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ - -L${GALLIUMDIR}/drivers/softpipe \ - -L${GALLIUMDIR}/auxiliary/tgsi \ - -L${GALLIUMDIR}/auxiliary/draw \ - -L${GALLIUMDIR}/auxiliary/util \ - -L${GALLIUMDIR}/auxiliary/translate \ - -L${GALLIUMDIR}/auxiliary/cso_cache \ - -L${GALLIUMDIR}/auxiliary/rtasm -LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lX11 -lm - -############################################# - -.PHONY = all clean - -all: test_context test_surface test_i_rendering test_p_rendering test_pf_rendering test_b_rendering - -test_context: test_context.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_surface: test_surface.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_i_rendering: test_i_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_p_rendering: test_p_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_pf_rendering: test_pf_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_b_rendering: test_b_rendering.o ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -clean: - rm -rf *.o test_context test_surface test_i_rendering test_p_rendering test_pf_rendering test_b_rendering - diff --git a/src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c deleted file mode 100644 index b78cc851ae4..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/test_b_rendering.c +++ /dev/null @@ -1,226 +0,0 @@ -#include -#include -#include -#include -#include - -static const unsigned short ycbcr16x16_420[8*8*6] = -{ - 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, - - 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, - - 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E -}; - -static const signed short ycbcr16x16_420_2[8*8*6] = -{ - -0x00A5,-0x00A5,-0x00A5,-0x0072,-0x00A5,-0x0072,-0x0072,-0x0072, - -0x0072,-0x00A5,-0x0072,-0x0072,-0x00A5,-0x0072,-0x0072,-0x0072, - -0x0072,-0x00A5,-0x0072,-0x0072,-0x00A5,-0x00A5,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x00A5, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x00A5,-0x00A5, - -0x0072,-0x0072,-0x0072,-0x0072,-0x0072,-0x00A5,-0x00A5,-0x00A5, - -0x0072,-0x0072,-0x0072,-0x0072,-0x00A5,-0x00A5,-0x00A5,-0x00A5, - - -0x004F,-0x004F,-0x004F,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, - -0x00B2,-0x00B2,-0x004F,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, - -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, - -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, - -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, - -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2, - -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x004F, - -0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x00B2,-0x004F,-0x004F, - - -0x003E,-0x003E,-0x003E,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, - -0x0060,-0x0060,-0x003E,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, - -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, - -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, - -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, - -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060, - -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x003E, - -0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x0060,-0x003E,-0x003E -}; - -int main(int argc, char **argv) -{ - const unsigned int video_width = 32, video_height = 32; - const unsigned int window_width = video_width * 2, window_height = video_height * 2; - int quit = 0; - Display *display; - Window root, window; - Pixmap framebuffer; - XEvent event; - struct pipe_context *pipe; - struct VL_CONTEXT *ctx; - struct VL_SURFACE *sfc, *past_sfc, *future_sfc; - struct VL_MOTION_VECTOR motion_vector[2] = - { - { - {0, 0}, {0, 0} - }, - { - {0, 0}, {0, 0} - } - }; - - display = XOpenDisplay(NULL); - root = XDefaultRootWindow(display); - window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); - framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); - - XSelectInput(display, window, ExposureMask | KeyPressMask); - XMapWindow(display, window); - XSync(display, 0); - - pipe = create_pipe_context(display); - vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); - vlCreateSurface(ctx, &sfc); - vlCreateSurface(ctx, &past_sfc); - vlCreateSurface(ctx, &future_sfc); - - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, past_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, future_sfc); - vlRenderBMacroBlock - ( - VL_FRAME_PICTURE, - VL_FIELD_FIRST, - 0, - 0, - VL_FRAME_MC, - motion_vector, - 0x3F, - VL_DCT_FRAME_CODED, - (short*)ycbcr16x16_420_2, - past_sfc, - future_sfc, - sfc - ); - vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); - - puts("Press any key to continue..."); - - while (!quit) - { - XNextEvent(display, &event); - switch (event.type) - { - case Expose: - { - XCopyArea - ( - display, - framebuffer, - window, - XDefaultGC(display, XDefaultScreen(display)), - 0, - 0, - window_width, - window_height, - 0, - 0 - ); - break; - } - case KeyPress: - { - quit = 1; - break; - } - } - } - - vlDestroySurface(sfc); - vlDestroySurface(past_sfc); - vlDestroySurface(future_sfc); - vlDestroyContext(ctx); - - XFreePixmap(display, framebuffer); - XDestroyWindow(display, window); - XCloseDisplay(display); - - return 0; -} - diff --git a/src/gallium/state_trackers/g3dvl/tests/test_context.c b/src/gallium/state_trackers/g3dvl/tests/test_context.c deleted file mode 100644 index 2002977ee24..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/test_context.c +++ /dev/null @@ -1,22 +0,0 @@ -#include -#include - -int main(int argc, char **argv) -{ - const unsigned int video_width = 32, video_height = 32; - - Display *display; - struct pipe_context *pipe; - struct VL_CONTEXT *ctx; - - display = XOpenDisplay(NULL); - pipe = create_pipe_context(display); - - vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); - vlDestroyContext(ctx); - - XCloseDisplay(display); - - return 0; -} - diff --git a/src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c deleted file mode 100644 index 1f964711308..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/test_i_rendering.c +++ /dev/null @@ -1,137 +0,0 @@ -#include -#include -#include -#include -#include - -static const unsigned short ycbcr16x16_420[8*8*6] = -{ - 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, - - 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, - - 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E -}; - -int main(int argc, char **argv) -{ - const unsigned int video_width = 32, video_height = 32; - const unsigned int window_width = video_width * 2, window_height = video_height * 2; - int quit = 0; - Display *display; - Window root, window; - Pixmap framebuffer; - XEvent event; - struct pipe_context *pipe; - struct VL_CONTEXT *ctx; - struct VL_SURFACE *sfc; - - display = XOpenDisplay(NULL); - root = XDefaultRootWindow(display); - window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); - framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); - - XSelectInput(display, window, ExposureMask | KeyPressMask); - XMapWindow(display, window); - XSync(display, 0); - - pipe = create_pipe_context(display); - vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); - vlCreateSurface(ctx, &sfc); - - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, sfc); - vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); - - puts("Press any key to continue..."); - - while (!quit) - { - XNextEvent(display, &event); - switch (event.type) - { - case Expose: - { - XCopyArea - ( - display, - framebuffer, - window, - XDefaultGC(display, XDefaultScreen(display)), - 0, - 0, - window_width, - window_height, - 0, - 0 - ); - break; - } - case KeyPress: - { - quit = 1; - break; - } - } - } - - vlDestroySurface(sfc); - vlDestroyContext(ctx); - - XFreePixmap(display, framebuffer); - XDestroyWindow(display, window); - XCloseDisplay(display); - - return 0; -} - diff --git a/src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c deleted file mode 100644 index 2203349784f..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/test_p_rendering.c +++ /dev/null @@ -1,214 +0,0 @@ -#include -#include -#include -#include -#include - -static const unsigned short ycbcr16x16_420[8*8*6] = -{ - 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, - - 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, - - 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E -}; - -static const signed short ycbcr16x16_420_2[8*8*6] = -{ - -51,-51,-51, 0,-51, 0, 0, 0, - 0,-51, 0, 0,-51, 0, 0, 0, - 0,-51, 0, 0,-51,-51, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 99, 99, 99, 0, 0, 0, 0, 0, - 0, 0, 99, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 33, 33, 33, 0, 0, 0, 0, 0, - 0, 0, 33, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0 -}; - -int main(int argc, char **argv) -{ - const unsigned int video_width = 32, video_height = 32; - const unsigned int window_width = video_width * 2, window_height = video_height * 2; - int quit = 0; - Display *display; - Window root, window; - Pixmap framebuffer; - XEvent event; - struct pipe_context *pipe; - struct VL_CONTEXT *ctx; - struct VL_SURFACE *sfc, *ref_sfc; - struct VL_MOTION_VECTOR motion_vector = - { - {0, 0}, {0, 0} - }; - - display = XOpenDisplay(NULL); - root = XDefaultRootWindow(display); - window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); - framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); - - XSelectInput(display, window, ExposureMask | KeyPressMask); - XMapWindow(display, window); - XSync(display, 0); - - pipe = create_pipe_context(display); - vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); - vlCreateSurface(ctx, &sfc); - vlCreateSurface(ctx, &ref_sfc); - - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderPMacroBlock - ( - VL_FRAME_PICTURE, - VL_FIELD_FIRST, - 0, - 0, - VL_FRAME_MC, - &motion_vector, - 0x3F, - VL_DCT_FRAME_CODED, - (short*)ycbcr16x16_420_2, - ref_sfc, - sfc - ); - vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); - - puts("Press any key to continue..."); - - while (!quit) - { - XNextEvent(display, &event); - switch (event.type) - { - case Expose: - { - XCopyArea - ( - display, - framebuffer, - window, - XDefaultGC(display, XDefaultScreen(display)), - 0, - 0, - window_width, - window_height, - 0, - 0 - ); - break; - } - case KeyPress: - { - quit = 1; - break; - } - } - } - - vlDestroySurface(sfc); - vlDestroySurface(ref_sfc); - vlDestroyContext(ctx); - - XFreePixmap(display, framebuffer); - XDestroyWindow(display, window); - XCloseDisplay(display); - - return 0; -} - diff --git a/src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c b/src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c deleted file mode 100644 index 43586fc553c..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/test_pf_rendering.c +++ /dev/null @@ -1,214 +0,0 @@ -#include -#include -#include -#include -#include - -static const unsigned short ycbcr16x16_420[8*8*6] = -{ - 0x00A5,0x00A5,0x00A5,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x0072,0x0072,0x0072, - 0x0072,0x00A5,0x0072,0x0072,0x00A5,0x00A5,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5, - 0x0072,0x0072,0x0072,0x0072,0x00A5,0x00A5,0x00A5,0x00A5, - - 0x004F,0x004F,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x004F,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F, - 0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x00B2,0x004F,0x004F, - - 0x003E,0x003E,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x003E,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E, - 0x0060,0x0060,0x0060,0x0060,0x0060,0x0060,0x003E,0x003E -}; - -static const signed short ycbcr16x16_420_2[8*8*6] = -{ - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0 -}; - -int main(int argc, char **argv) -{ - const unsigned int video_width = 32, video_height = 32; - const unsigned int window_width = video_width * 2, window_height = video_height * 2; - int quit = 0; - Display *display; - Window root, window; - Pixmap framebuffer; - XEvent event; - struct pipe_context *pipe; - struct VL_CONTEXT *ctx; - struct VL_SURFACE *sfc, *ref_sfc; - struct VL_MOTION_VECTOR motion_vector = - { - {0, 0}, {32, 32} - }; - - display = XOpenDisplay(NULL); - root = XDefaultRootWindow(display); - window = XCreateSimpleWindow(display, root, 0, 0, window_width, window_height, 0, 0, 0); - framebuffer = XCreatePixmap(display, root, window_width, window_height, 24); - - XSelectInput(display, window, ExposureMask | KeyPressMask); - XMapWindow(display, window); - XSync(display, 0); - - pipe = create_pipe_context(display); - vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); - vlCreateSurface(ctx, &sfc); - vlCreateSurface(ctx, &ref_sfc); - - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 0, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 0, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderIMacroBlock(VL_FRAME_PICTURE, VL_FIELD_FIRST, 1, 1, 0x3F, VL_DCT_FRAME_CODED, (short*)ycbcr16x16_420, ref_sfc); - vlRenderPMacroBlock - ( - VL_FRAME_PICTURE, - VL_FIELD_FIRST, - 0, - 0, - VL_FIELD_MC, - &motion_vector, - 0x3F, - VL_DCT_FRAME_CODED, - (short*)ycbcr16x16_420_2, - ref_sfc, - sfc - ); - vlPutSurface(sfc, framebuffer, 0, 0, video_width, video_height, 0, 0, window_width, window_height, VL_FRAME_PICTURE); - - puts("Press any key to continue..."); - - while (!quit) - { - XNextEvent(display, &event); - switch (event.type) - { - case Expose: - { - XCopyArea - ( - display, - framebuffer, - window, - XDefaultGC(display, XDefaultScreen(display)), - 0, - 0, - window_width, - window_height, - 0, - 0 - ); - break; - } - case KeyPress: - { - quit = 1; - break; - } - } - } - - vlDestroySurface(sfc); - vlDestroySurface(ref_sfc); - vlDestroyContext(ctx); - - XFreePixmap(display, framebuffer); - XDestroyWindow(display, window); - XCloseDisplay(display); - - return 0; -} - diff --git a/src/gallium/state_trackers/g3dvl/tests/test_surface.c b/src/gallium/state_trackers/g3dvl/tests/test_surface.c deleted file mode 100644 index 4d1946396af..00000000000 --- a/src/gallium/state_trackers/g3dvl/tests/test_surface.c +++ /dev/null @@ -1,26 +0,0 @@ -#include -#include -#include - -int main(int argc, char **argv) -{ - const unsigned int video_width = 32, video_height = 32; - - Display *display; - struct pipe_context *pipe; - struct VL_CONTEXT *ctx; - struct VL_SURFACE *sfc; - - display = XOpenDisplay(NULL); - pipe = create_pipe_context(display); - - vlCreateContext(display, pipe, video_width, video_height, VL_FORMAT_YCBCR_420, &ctx); - vlCreateSurface(ctx, &sfc); - vlDestroySurface(sfc); - vlDestroyContext(ctx); - - XCloseDisplay(display); - - return 0; -} - -- cgit v1.2.3 From 92674bc8889e10e580c630cf85c106fa6eb34d7b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Wed, 29 Oct 2008 18:12:48 +1100 Subject: nv40: rename some vars in texture layout code --- src/gallium/drivers/nv40/nv40_miptree.c | 52 +++++++++++++++++---------------- 1 file changed, 27 insertions(+), 25 deletions(-) diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index 6c54c37ef71..f8d4497bf75 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -5,9 +5,9 @@ #include "nv40_context.h" static void -nv40_miptree_layout(struct nv40_miptree *nv40mt) +nv40_miptree_layout(struct nv40_miptree *mt) { - struct pipe_texture *pt = &nv40mt->base; + struct pipe_texture *pt = &mt->base; boolean swizzled = FALSE; uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0]; uint offset = 0; @@ -34,8 +34,8 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) pitch = pt->nblocksx[l]; pitch = align(pitch, 64); - nv40mt->level[l].pitch = pitch * pt->block.size; - nv40mt->level[l].image_offset = + mt->level[l].pitch = pitch * pt->block.size; + mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = MAX2(1, width >> 1); @@ -45,12 +45,12 @@ nv40_miptree_layout(struct nv40_miptree *nv40mt) for (f = 0; f < nr_faces; f++) { for (l = 0; l <= pt->last_level; l++) { - nv40mt->level[l].image_offset[f] = offset; - offset += nv40mt->level[l].pitch * pt->height[l]; + mt->level[l].image_offset[f] = offset; + offset += mt->level[l].pitch * pt->height[l]; } } - nv40mt->total_size = offset; + mt->total_size = offset; } static struct pipe_texture * @@ -81,22 +81,24 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) } static void -nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **pt) +nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt) { - struct pipe_texture *mt = *pt; + struct pipe_texture *pt = *ppt; + struct nv40_miptree *mt = (struct nv40_miptree *)pt; + int l; - *pt = NULL; - if (--mt->refcount <= 0) { - struct nv40_miptree *nv40mt = (struct nv40_miptree *)mt; - int l; + *ppt = NULL; + if (--pt->refcount) + return; - pipe_buffer_reference(pscreen, &nv40mt->buffer, NULL); - for (l = 0; l <= mt->last_level; l++) { - if (nv40mt->level[l].image_offset) - FREE(nv40mt->level[l].image_offset); - } - FREE(nv40mt); + + pipe_buffer_reference(pscreen, &mt->buffer, NULL); + for (l = 0; l <= pt->last_level; l++) { + if (mt->level[l].image_offset) + FREE(mt->level[l].image_offset); } + + FREE(mt); } static struct pipe_surface * @@ -104,31 +106,31 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { - struct nv40_miptree *nv40mt = (struct nv40_miptree *)pt; + struct nv40_miptree *mt = (struct nv40_miptree *)pt; struct pipe_surface *ps; ps = CALLOC_STRUCT(pipe_surface); if (!ps) return NULL; pipe_texture_reference(&ps->texture, pt); - pipe_buffer_reference(pscreen, &ps->buffer, nv40mt->buffer); + pipe_buffer_reference(pscreen, &ps->buffer, mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; ps->block = pt->block; ps->nblocksx = pt->nblocksx[level]; ps->nblocksy = pt->nblocksy[level]; - ps->stride = nv40mt->level[level].pitch; + ps->stride = mt->level[level].pitch; ps->usage = flags; ps->status = PIPE_SURFACE_STATUS_DEFINED; if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv40mt->level[level].image_offset[face]; + ps->offset = mt->level[level].image_offset[face]; } else if (pt->target == PIPE_TEXTURE_3D) { - ps->offset = nv40mt->level[level].image_offset[zslice]; + ps->offset = mt->level[level].image_offset[zslice]; } else { - ps->offset = nv40mt->level[level].image_offset[0]; + ps->offset = mt->level[level].image_offset[0]; } return ps; -- cgit v1.2.3 From 0344b0e32e397ebcc6ed236295eb9b6a06a0dae2 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sat, 8 Nov 2008 17:04:45 +0200 Subject: Nouveau: start nv20 by copying the nv10 tree. Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv20/Makefile | 28 ++ src/gallium/drivers/nv20/nv10_clear.c | 12 + src/gallium/drivers/nv20/nv10_context.c | 296 +++++++++++++++ src/gallium/drivers/nv20/nv10_context.h | 153 ++++++++ src/gallium/drivers/nv20/nv10_fragprog.c | 21 ++ src/gallium/drivers/nv20/nv10_fragtex.c | 149 ++++++++ src/gallium/drivers/nv20/nv10_miptree.c | 147 ++++++++ src/gallium/drivers/nv20/nv10_prim_vbuf.c | 240 ++++++++++++ src/gallium/drivers/nv20/nv10_screen.c | 208 ++++++++++ src/gallium/drivers/nv20/nv10_screen.h | 22 ++ src/gallium/drivers/nv20/nv10_state.c | 588 +++++++++++++++++++++++++++++ src/gallium/drivers/nv20/nv10_state.h | 139 +++++++ src/gallium/drivers/nv20/nv10_state_emit.c | 303 +++++++++++++++ src/gallium/drivers/nv20/nv10_surface.c | 64 ++++ src/gallium/drivers/nv20/nv10_vbo.c | 77 ++++ 15 files changed, 2447 insertions(+) create mode 100644 src/gallium/drivers/nv20/Makefile create mode 100644 src/gallium/drivers/nv20/nv10_clear.c create mode 100644 src/gallium/drivers/nv20/nv10_context.c create mode 100644 src/gallium/drivers/nv20/nv10_context.h create mode 100644 src/gallium/drivers/nv20/nv10_fragprog.c create mode 100644 src/gallium/drivers/nv20/nv10_fragtex.c create mode 100644 src/gallium/drivers/nv20/nv10_miptree.c create mode 100644 src/gallium/drivers/nv20/nv10_prim_vbuf.c create mode 100644 src/gallium/drivers/nv20/nv10_screen.c create mode 100644 src/gallium/drivers/nv20/nv10_screen.h create mode 100644 src/gallium/drivers/nv20/nv10_state.c create mode 100644 src/gallium/drivers/nv20/nv10_state.h create mode 100644 src/gallium/drivers/nv20/nv10_state_emit.c create mode 100644 src/gallium/drivers/nv20/nv10_surface.c create mode 100644 src/gallium/drivers/nv20/nv10_vbo.c diff --git a/src/gallium/drivers/nv20/Makefile b/src/gallium/drivers/nv20/Makefile new file mode 100644 index 00000000000..4ba7ce586d6 --- /dev/null +++ b/src/gallium/drivers/nv20/Makefile @@ -0,0 +1,28 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = nv10 + +DRIVER_SOURCES = \ + nv10_clear.c \ + nv10_context.c \ + nv10_fragprog.c \ + nv10_fragtex.c \ + nv10_miptree.c \ + nv10_prim_vbuf.c \ + nv10_screen.c \ + nv10_state.c \ + nv10_state_emit.c \ + nv10_surface.c \ + nv10_vbo.c + +C_SOURCES = \ + $(COMMON_SOURCES) \ + $(DRIVER_SOURCES) + +ASM_SOURCES = + +include ../../Makefile.template + +symlinks: + diff --git a/src/gallium/drivers/nv20/nv10_clear.c b/src/gallium/drivers/nv20/nv10_clear.c new file mode 100644 index 00000000000..be7e09cf4b0 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv10_context.h" + +void +nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv20/nv10_context.c b/src/gallium/drivers/nv20/nv10_context.c new file mode 100644 index 00000000000..e9b61daae7f --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_context.c @@ -0,0 +1,296 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static void +nv10_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_flush(nv10->draw); + + FIRE_RING(fence); +} + +static void +nv10_destroy(struct pipe_context *pipe) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + if (nv10->draw) + draw_destroy(nv10->draw); + + FREE(nv10); +} + +static void nv10_init_hwctx(struct nv10_context *nv10) +{ + struct nv10_screen *screen = nv10->screen; + struct nouveau_winsys *nvws = screen->nvws; + int i; + float projectionmatrix[16]; + + BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); + OUT_RING (screen->sync->handle); + BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY2, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 2); + OUT_RING (0); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1); + OUT_RING ((0x7ff<<16)|0x800); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1); + OUT_RING ((0x7ff<<16)|0x800); + + for (i=1;i<8;i++) { + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(i), 1); + OUT_RING (0); + } + + BEGIN_RING(celsius, 0x290, 1); + OUT_RING ((0x10<<16)|1); + BEGIN_RING(celsius, 0x3f4, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + if (nv10->screen->celsius->grclass != NV10TCL) { + /* For nv11, nv17 */ + BEGIN_RING(celsius, 0x120, 3); + OUT_RING (0); + OUT_RING (1); + OUT_RING (2); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + } + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + /* Set state */ + BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 2); + OUT_RING (0x207); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2); + OUT_RING (0); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_RC_IN_ALPHA(0), 12); + OUT_RING (0x30141010); + OUT_RING (0); + OUT_RING (0x20040000); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x00000c00); + OUT_RING (0); + OUT_RING (0x00000c00); + OUT_RING (0x18000000); + OUT_RING (0x300e0300); + OUT_RING (0x0c091c80); + + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2); + OUT_RING (1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VERTEX_WEIGHT_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_SRC, 4); + OUT_RING (1); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x8006); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 8); + OUT_RING (0xff); + OUT_RING (0x207); + OUT_RING (0); + OUT_RING (0xff); + OUT_RING (0x1e00); + OUT_RING (0x1e00); + OUT_RING (0x1e00); + OUT_RING (0x1d01); + BEGIN_RING(celsius, NV10TCL_NORMALIZE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LIGHT_MODEL, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_COLOR_CONTROL, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ENABLED_LIGHTS, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + OUT_RING (0x201); + BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_FACTOR, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (8); + BEGIN_RING(celsius, NV10TCL_POINT_PARAMETERS_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LINE_WIDTH, 1); + OUT_RING (8); + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (0x1b02); + OUT_RING (0x1b02); + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (0x405); + OUT_RING (0x901); + BEGIN_RING(celsius, NV10TCL_POLYGON_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_CLIP_PLANE_ENABLE(0), 8); + for (i=0;i<8;i++) { + OUT_RING (0); + } + BEGIN_RING(celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3); + OUT_RING (0x3fc00000); /* -1.50 */ + OUT_RING (0xbdb8aa0a); /* -0.09 */ + OUT_RING (0); /* 0.00 */ + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_FOG_MODE, 2); + OUT_RING (0x802); + OUT_RING (2); + /* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when + * using texturing, except when using the texture matrix + */ + BEGIN_RING(celsius, NV10TCL_VIEW_MATRIX_ENABLE, 1); + OUT_RING (6); + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (0x01010101); + + /* Set vertex component */ + BEGIN_RING(celsius, NV10TCL_VERTEX_COL_4F_R, 4); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_COL2_3F_R, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VERTEX_NOR_3F_X, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_TX0_4F_S, 4); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_TX1_4F_S, 4); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_FOG_1F, 1); + OUT_RINGf (0.0); + BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1); + OUT_RING (1); + + memset(projectionmatrix, 0, sizeof(projectionmatrix)); + BEGIN_RING(celsius, NV10TCL_PROJECTION_MATRIX(0), 16); + projectionmatrix[0*4+0] = 1.0; + projectionmatrix[1*4+1] = 1.0; + projectionmatrix[2*4+2] = 1.0; + projectionmatrix[3*4+3] = 1.0; + for (i=0;i<16;i++) { + OUT_RINGf (projectionmatrix[i]); + } + + BEGIN_RING(celsius, NV10TCL_DEPTH_RANGE_NEAR, 2); + OUT_RING (0.0); + OUT_RINGf (16777216.0); + + BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); + OUT_RINGf (-2048.0); + OUT_RINGf (-2048.0); + OUT_RINGf (16777215.0 * 0.5); + OUT_RING (0); + + FIRE_RING (NULL); +} + +static void +nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + +struct pipe_context * +nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) +{ + struct nv10_screen *screen = nv10_screen(pscreen); + struct pipe_winsys *ws = pscreen->winsys; + struct nv10_context *nv10; + struct nouveau_winsys *nvws = screen->nvws; + + nv10 = CALLOC(1, sizeof(struct nv10_context)); + if (!nv10) + return NULL; + nv10->screen = screen; + nv10->pctx_id = pctx_id; + + nv10->nvws = nvws; + + nv10->pipe.winsys = ws; + nv10->pipe.screen = pscreen; + nv10->pipe.destroy = nv10_destroy; + nv10->pipe.set_edgeflags = nv10_set_edgeflags; + nv10->pipe.draw_arrays = nv10_draw_arrays; + nv10->pipe.draw_elements = nv10_draw_elements; + nv10->pipe.clear = nv10_clear; + nv10->pipe.flush = nv10_flush; + + nv10_init_surface_functions(nv10); + nv10_init_state_functions(nv10); + + nv10->draw = draw_create(); + assert(nv10->draw); + draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10)); + + nv10_init_hwctx(nv10); + + return &nv10->pipe; +} + diff --git a/src/gallium/drivers/nv20/nv10_context.h b/src/gallium/drivers/nv20/nv10_context.h new file mode 100644 index 00000000000..f3b56de25a7 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_context.h @@ -0,0 +1,153 @@ +#ifndef __NV10_CONTEXT_H__ +#define __NV10_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv10_screen *ctx = nv10->screen +#include "nouveau/nouveau_push.h" + +#include "nv10_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV10_NEW_VERTPROG (1 << 0) +#define NV10_NEW_FRAGPROG (1 << 1) +#define NV10_NEW_VTXARRAYS (1 << 2) +#define NV10_NEW_BLEND (1 << 3) +#define NV10_NEW_BLENDCOL (1 << 4) +#define NV10_NEW_RAST (1 << 5) +#define NV10_NEW_DSA (1 << 6) +#define NV10_NEW_VIEWPORT (1 << 7) +#define NV10_NEW_SCISSOR (1 << 8) +#define NV10_NEW_FRAMEBUFFER (1 << 9) + +#include "nv10_screen.h" + +struct nv10_context { + struct pipe_context pipe; + + struct nouveau_winsys *nvws; + struct nv10_screen *screen; + unsigned pctx_id; + + struct draw_context *draw; + + uint32_t dirty; + + struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv10_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + uint32_t rt_enable; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; + uint32_t lma_offset; + + struct nv10_blend_state *blend; + struct pipe_blend_color *blend_color; + struct nv10_rasterizer_state *rast; + struct nv10_depth_stencil_alpha_state *dsa; + struct pipe_viewport_state *viewport; + struct pipe_scissor_state *scissor; + struct pipe_framebuffer_state *framebuffer; + + //struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + float *constbuf[PIPE_SHADER_TYPES][32][4]; + unsigned constbuf_nr[PIPE_SHADER_TYPES]; + + struct vertex_info vertex_info; + + struct { + struct pipe_buffer *buffer; + uint32_t format; + } tex[2]; + + unsigned vb_enable; + struct { + struct pipe_buffer *buffer; + unsigned delta; + } vb[16]; + +/* struct { + + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + + struct nv10_vertex_program *active; + + struct nv10_vertex_program *current; + } vertprog; +*/ + struct { + struct nv10_fragment_program *active; + + struct nv10_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; +}; + +static INLINE struct nv10_context * +nv10_context(struct pipe_context *pipe) +{ + return (struct nv10_context *)pipe; +} + +extern void nv10_init_state_functions(struct nv10_context *nv10); +extern void nv10_init_surface_functions(struct nv10_context *nv10); + +extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen); + +/* nv10_clear.c */ +extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +/* nv10_draw.c */ +extern struct draw_stage *nv10_draw_render_stage(struct nv10_context *nv10); + +/* nv10_fragprog.c */ +extern void nv10_fragprog_bind(struct nv10_context *, + struct nv10_fragment_program *); +extern void nv10_fragprog_destroy(struct nv10_context *, + struct nv10_fragment_program *); + +/* nv10_fragtex.c */ +extern void nv10_fragtex_bind(struct nv10_context *); + +/* nv10_prim_vbuf.c */ +struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ); +extern void nv10_vtxbuf_bind(struct nv10_context* nv10); + +/* nv10_state.c and friends */ +extern void nv10_emit_hw_state(struct nv10_context *nv10); +extern void nv10_state_tex_update(struct nv10_context *nv10); + +/* nv10_vbo.c */ +extern boolean nv10_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv10_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count); + + +#endif diff --git a/src/gallium/drivers/nv20/nv10_fragprog.c b/src/gallium/drivers/nv20/nv10_fragprog.c new file mode 100644 index 00000000000..698db5a16a9 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_fragprog.c @@ -0,0 +1,21 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" + +#include "nv10_context.h" + +void +nv10_fragprog_bind(struct nv10_context *nv10, struct nv10_fragment_program *fp) +{ +} + +void +nv10_fragprog_destroy(struct nv10_context *nv10, + struct nv10_fragment_program *fp) +{ +} + diff --git a/src/gallium/drivers/nv20/nv10_fragtex.c b/src/gallium/drivers/nv20/nv10_fragtex.c new file mode 100644 index 00000000000..238634d0bb4 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_fragtex.c @@ -0,0 +1,149 @@ +#include "nv10_context.h" + +static INLINE int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + +#define _(m,tf) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV10TCL_TX_FORMAT_FORMAT_##tf, \ +} + +struct nv10_texture_format { + boolean defined; + uint pipe; + int format; +}; + +static struct nv10_texture_format +nv10_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8), + _(A1R5G5B5_UNORM, A1R5G5B5), + _(A4R4G4B4_UNORM, A4R4G4B4), + _(L8_UNORM , L8 ), + _(A8_UNORM , A8 ), + _(A8L8_UNORM , A8L8 ), +// _(RGB_DXT1 , DXT1, ), +// _(RGBA_DXT1 , DXT1, ), +// _(RGBA_DXT3 , DXT3, ), +// _(RGBA_DXT5 , DXT5, ), + {}, +}; + +static struct nv10_texture_format * +nv10_fragtex_format(uint pipe_format) +{ + struct nv10_texture_format *tf = nv10_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv10_fragtex_build(struct nv10_context *nv10, int unit) +{ +#if 0 + struct nv10_sampler_state *ps = nv10->tex_sampler[unit]; + struct nv10_miptree *nv10mt = nv10->tex_miptree[unit]; + struct pipe_texture *pt = &nv10mt->base; + struct nv10_texture_format *tf; + uint32_t txf, txs, txp; + + tf = nv10_fragtex_format(pt->format); + if (!tf || !tf->defined) { + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); + return; + } + + txf = tf->format << 8; + txf |= (pt->last_level + 1) << 16; + txf |= log2i(pt->width[0]) << 20; + txf |= log2i(pt->height[0]) << 24; + txf |= log2i(pt->depth[0]) << 28; + txf |= 8; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE: + txf |= NV10TCL_TX_FORMAT_CUBE_MAP; + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= (2<<4); + break; + case PIPE_TEXTURE_1D: + txf |= (1<<4); + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(unit), 8); + OUT_RELOCl(nv10mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv10mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RING (ps->wrap); + OUT_RING (0x40000000); /* enable */ + OUT_RING (txs); + OUT_RING (ps->filt | 0x2000 /* magic */); + OUT_RING ((pt->width[0] << 16) | pt->height[0]); + OUT_RING (ps->bcol); +#endif +} + +void +nv10_fragtex_bind(struct nv10_context *nv10) +{ +#if 0 + struct nv10_fragment_program *fp = nv10->fragprog.active; + unsigned samplers, unit; + + samplers = nv10->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(celsius, NV10TCL_TX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv10->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv10_fragtex_build(nv10, unit); + } + + nv10->fp_samplers = fp->samplers; +#endif +} + diff --git a/src/gallium/drivers/nv20/nv10_miptree.c b/src/gallium/drivers/nv20/nv10_miptree.c new file mode 100644 index 00000000000..ad084e72b81 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_miptree.c @@ -0,0 +1,147 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_inlines.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static void +nv10_miptree_layout(struct nv10_miptree *nv10mt) +{ + struct pipe_texture *pt = &nv10mt->base; + boolean swizzled = FALSE; + uint width = pt->width[0], height = pt->height[0]; + uint offset = 0; + int nr_faces, l, f; + + if (pt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else { + nr_faces = 1; + } + + for (l = 0; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); + pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); + + if (swizzled) + nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; + else + nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; + nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; + + nv10mt->level[l].image_offset = + CALLOC(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = 0; l <= pt->last_level; l++) { + nv10mt->level[l].image_offset[f] = offset; + offset += nv10mt->level[l].pitch * pt->height[l]; + } + } + + nv10mt->total_size = offset; +} + +static struct pipe_texture * +nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv10_miptree *mt; + + mt = MALLOC(sizeof(struct nv10_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + mt->base.refcount = 1; + mt->base.screen = screen; + + nv10_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + FREE(mt); + return NULL; + } + + return &mt->base; +} + +static void +nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) +{ + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt; + int l; + + pipe_buffer_reference(screen, &nv10mt->buffer, NULL); + for (l = 0; l <= mt->last_level; l++) { + if (nv10mt->level[l].image_offset) + FREE(nv10mt->level[l].image_offset); + } + FREE(nv10mt); + } +} + +static void +nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, + uint face, uint levels) +{ +} + + +static struct pipe_surface * +nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice, + unsigned flags) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(screen, &ps->buffer, nv10mt->buffer); + ps->format = pt->format; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->block = pt->block; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->stride = nv10mt->level[level].pitch; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv10mt->level[level].image_offset[face]; + } else { + ps->offset = nv10mt->level[level].image_offset[0]; + } + + return ps; +} + +static void +nv10_miptree_surface_release(struct pipe_screen *screen, + struct pipe_surface **surface) +{ +} + +void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv10_miptree_create; + pscreen->texture_release = nv10_miptree_release; + pscreen->get_tex_surface = nv10_miptree_surface_get; + pscreen->tex_surface_release = nv10_miptree_surface_release; +} + diff --git a/src/gallium/drivers/nv20/nv10_prim_vbuf.c b/src/gallium/drivers/nv20/nv10_prim_vbuf.c new file mode 100644 index 00000000000..62a8f6d89da --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_prim_vbuf.c @@ -0,0 +1,240 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * \file + * Build post-transformation, post-clipping vertex buffers and element + * lists by hooking into the end of the primitive pipeline and + * manipulating the vertex_id field in the vertex headers. + * + * XXX: work in progress + * + * \author José Fonseca + * \author Keith Whitwell + */ + + +#include "pipe/p_debug.h" +#include "pipe/p_inlines.h" +#include "pipe/p_winsys.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +#include "draw/draw_vbuf.h" + +/** + * Primitive renderer for nv10. + */ +struct nv10_vbuf_render { + struct vbuf_render base; + + struct nv10_context *nv10; + + /** Vertex buffer */ + struct pipe_buffer* buffer; + + /** Vertex size in bytes */ + unsigned vertex_size; + + /** Hardware primitive */ + unsigned hwprim; +}; + + +void nv10_vtxbuf_bind( struct nv10_context* nv10 ) +{ + int i; + for(i = 0; i < 8; i++) { + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1); + OUT_RING(0/*nv10->vtxbuf*/); + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1); + OUT_RING(0/*XXX*/); + } +} + +/** + * Basically a cast wrapper. + */ +static INLINE struct nv10_vbuf_render * +nv10_vbuf_render( struct vbuf_render *render ) +{ + assert(render); + return (struct nv10_vbuf_render *)render; +} + + +static const struct vertex_info * +nv10_vbuf_render_get_vertex_info( struct vbuf_render *render ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + + nv10_emit_hw_state(nv10); + + return &nv10->vertex_info; +} + + +static void * +nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, + ushort vertex_size, + ushort nr_vertices ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + struct pipe_winsys *winsys = nv10->pipe.winsys; + size_t size = (size_t)vertex_size * (size_t)nr_vertices; + + assert(!nv10_render->buffer); + nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); + + nv10->dirty |= NV10_NEW_VTXARRAYS; + + return winsys->buffer_map(winsys, + nv10_render->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); +} + + +static void +nv10_vbuf_render_set_primitive( struct vbuf_render *render, + unsigned prim ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + nv10_render->hwprim = prim + 1; +} + + +static void +nv10_vbuf_render_draw( struct vbuf_render *render, + const ushort *indices, + uint nr_indices) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + int push, i; + + nv10_emit_hw_state(nv10); + + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); + OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + + BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING(nv10_render->hwprim); + + if (nr_indices & 1) { + BEGIN_RING(celsius, NV10TCL_VB_ELEMENT_U32, 1); + OUT_RING (indices[0]); + indices++; nr_indices--; + } + + while (nr_indices) { + // XXX too big/small ? check the size + push = MIN2(nr_indices, 1200 * 2); + + BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((indices[i+1] << 16) | indices[i]); + + nr_indices -= push; + indices += push; + } + + BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING (0); +} + + +static void +nv10_vbuf_render_release_vertices( struct vbuf_render *render, + void *vertices, + unsigned vertex_size, + unsigned vertices_used ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + struct pipe_winsys *winsys = nv10->pipe.winsys; + struct pipe_screen *pscreen = &nv10->screen->pipe; + + assert(nv10_render->buffer); + winsys->buffer_unmap(winsys, nv10_render->buffer); + pipe_buffer_reference(pscreen, &nv10_render->buffer, NULL); +} + + +static void +nv10_vbuf_render_destroy( struct vbuf_render *render ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + FREE(nv10_render); +} + + +/** + * Create a new primitive render. + */ +static struct vbuf_render * +nv10_vbuf_render_create( struct nv10_context *nv10 ) +{ + struct nv10_vbuf_render *nv10_render = CALLOC_STRUCT(nv10_vbuf_render); + + nv10_render->nv10 = nv10; + + nv10_render->base.max_vertex_buffer_bytes = 16*1024; + nv10_render->base.max_indices = 1024; + nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info; + nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices; + nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive; + nv10_render->base.draw = nv10_vbuf_render_draw; + nv10_render->base.release_vertices = nv10_vbuf_render_release_vertices; + nv10_render->base.destroy = nv10_vbuf_render_destroy; + + return &nv10_render->base; +} + + +/** + * Create a new primitive vbuf/render stage. + */ +struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ) +{ + struct vbuf_render *render; + struct draw_stage *stage; + + render = nv10_vbuf_render_create(nv10); + if(!render) + return NULL; + + stage = draw_vbuf_stage( nv10->draw, render ); + if(!stage) { + render->destroy(render); + return NULL; + } + + return stage; +} diff --git a/src/gallium/drivers/nv20/nv10_screen.c b/src/gallium/drivers/nv20/nv10_screen.c new file mode 100644 index 00000000000..27a9edf9bba --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_screen.c @@ -0,0 +1,208 @@ +#include "pipe/p_screen.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static const char * +nv10_screen_get_name(struct pipe_screen *screen) +{ + struct nv10_screen *nv10screen = nv10_screen(screen); + struct nouveau_device *dev = nv10screen->nvws->channel->device; + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); + return buffer; +} + +static const char * +nv10_screen_get_vendor(struct pipe_screen *screen) +{ + return "nouveau"; +} + +static int +nv10_screen_get_param(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 2; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 0; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 1; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 12; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 0; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 12; + case NOUVEAU_CAP_HW_VTXBUF: + case NOUVEAU_CAP_HW_IDXBUF: + return 0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv10_screen_get_paramf(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 2.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static boolean +nv10_screen_is_format_supported(struct pipe_screen *screen, + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) +{ + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + } else { + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + case PIPE_FORMAT_I8_UNORM: + return TRUE; + default: + break; + } + } + + return FALSE; +} + +static void * +nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + +static void +nv10_screen_destroy(struct pipe_screen *pscreen) +{ + struct nv10_screen *screen = nv10_screen(pscreen); + struct nouveau_winsys *nvws = screen->nvws; + + nvws->notifier_free(&screen->sync); + nvws->grobj_free(&screen->celsius); + + FREE(pscreen); +} + +struct pipe_screen * +nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) +{ + struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen); + unsigned celsius_class; + unsigned chipset = nvws->channel->device->chipset; + int ret; + + if (!screen) + return NULL; + screen->nvws = nvws; + + /* 3D object */ + if (chipset>=0x20) + celsius_class=NV11TCL; + else if (chipset>=0x17) + celsius_class=NV17TCL; + else if (chipset>=0x11) + celsius_class=NV11TCL; + else + celsius_class=NV10TCL; + + if (!celsius_class) { + NOUVEAU_ERR("Unknown nv1x chipset: nv%02x\n", chipset); + return NULL; + } + + ret = nvws->grobj_alloc(nvws, celsius_class, &screen->celsius); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv10_screen_destroy(&screen->pipe); + return NULL; + } + + screen->pipe.winsys = ws; + screen->pipe.destroy = nv10_screen_destroy; + + screen->pipe.get_name = nv10_screen_get_name; + screen->pipe.get_vendor = nv10_screen_get_vendor; + screen->pipe.get_param = nv10_screen_get_param; + screen->pipe.get_paramf = nv10_screen_get_paramf; + + screen->pipe.is_format_supported = nv10_screen_is_format_supported; + + screen->pipe.surface_map = nv10_surface_map; + screen->pipe.surface_unmap = nv10_surface_unmap; + + nv10_screen_init_miptree_functions(&screen->pipe); + + return &screen->pipe; +} + diff --git a/src/gallium/drivers/nv20/nv10_screen.h b/src/gallium/drivers/nv20/nv10_screen.h new file mode 100644 index 00000000000..3f8750a13f7 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_screen.h @@ -0,0 +1,22 @@ +#ifndef __NV10_SCREEN_H__ +#define __NV10_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv10_screen { + struct pipe_screen pipe; + + struct nouveau_winsys *nvws; + + /* HW graphics objects */ + struct nouveau_grobj *celsius; + struct nouveau_notifier *sync; +}; + +static INLINE struct nv10_screen * +nv10_screen(struct pipe_screen *screen) +{ + return (struct nv10_screen *)screen; +} + +#endif diff --git a/src/gallium/drivers/nv20/nv10_state.c b/src/gallium/drivers/nv20/nv10_state.c new file mode 100644 index 00000000000..d2375aa2f64 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_state.c @@ -0,0 +1,588 @@ +#include "draw/draw_context.h" +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_shader_tokens.h" + +#include "tgsi/tgsi_parse.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +static void * +nv10_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv10_blend_state *cb; + + cb = MALLOC(sizeof(struct nv10_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + + cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + + cb->d_enable = cso->dither ? 1 : 0; + + return (void *)cb; +} + +static void +nv10_blend_state_bind(struct pipe_context *pipe, void *blend) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->blend = (struct nv10_blend_state*)blend; + + nv10->dirty |= NV10_NEW_BLEND; +} + +static void +nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + case PIPE_TEX_WRAP_MIRROR_CLAMP: + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; + break; + } + + return ret >> NV10TCL_TX_FORMAT_WRAP_S_SHIFT; +} + +static void * +nv10_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv10_sampler_state *ps; + uint32_t filter = 0; + + ps = MALLOC(sizeof(struct nv10_sampler_state)); + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy > 1.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + +/* if (cso->max_anisotropy >= 16.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_16X; + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_12X; + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_10X; + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_6X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_4X; + } else { + ps->en |= NV10TCL_TX_ENABLE_ANISO_2X; + }*/ + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MAGNIFY_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV10TCL_TX_FILTER_MAGNIFY_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST; + break; + } + break; + } + + ps->filt = filter; + +/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER; + break; + case PIPE_FUNC_GREATER: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER; + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL; + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL; + break; + case PIPE_FUNC_LESS: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS; + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL; + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL; + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS; + break; + default: + break; + } + }*/ + + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) +{ + struct nv10_context *nv10 = nv10_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv10->tex_sampler[unit] = sampler[unit]; + nv10->dirty_samplers |= (1 << unit); + } +} + +static void +nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + +static void +nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **miptree) +{ + struct nv10_context *nv10 = nv10_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit]; + nv10->dirty_samplers |= (1 << unit); + } +} + +static void * +nv10_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv10_rasterizer_state *rs; + int i; + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + rs = MALLOC(sizeof(struct nv10_rasterizer_state)); + + rs->templ = cso; + + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; + + rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; + rs->line_smooth_en = cso->line_smooth ? 1 : 0; + + rs->point_size = *(uint32_t*)&cso->point_size; + + rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; + + if (cso->front_winding == PIPE_WINDING_CCW) { + rs->front_face = NV10TCL_FRONT_FACE_CCW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + } else { + rs->front_face = NV10TCL_FRONT_FACE_CW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + } + + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CCW) + rs->cull_face = NV10TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV10TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_CW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CW) + rs->cull_face = NV10TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV10TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_BOTH: + rs->cull_face_en = 1; + rs->cull_face = NV10TCL_CULL_FACE_FRONT_AND_BACK; + break; + case PIPE_WINDING_NONE: + default: + rs->cull_face_en = 0; + rs->cull_face = 0; + break; + } + + if (cso->point_sprite) { + rs->point_sprite = (1 << 0); + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + rs->point_sprite |= (1 << (8 + i)); + } + } else { + rs->point_sprite = 0; + } + + return (void *)rs; +} + +static void +nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->rast = (struct nv10_rasterizer_state*)rast; + + draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL)); + + nv10->dirty |= NV10_NEW_RAST; +} + +static void +nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + +static void * +nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv10_depth_stencil_alpha_state *hw; + + hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state)); + + hw->depth.func = nvgl_comparison_op(cso->depth.func); + hw->depth.write_enable = cso->depth.writemask ? 1 : 0; + hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + + hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0; + hw->stencil.wmask = cso->stencil[0].write_mask; + hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func); + hw->stencil.ref = cso->stencil[0].ref_value; + hw->stencil.vmask = cso->stencil[0].value_mask; + hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op); + hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op); + hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op); + + hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; + hw->alpha.func = nvgl_comparison_op(cso->alpha.func); + hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + + return (void *)hw; +} + +static void +nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa; + + nv10->dirty |= NV10_NEW_DSA; +} + +static void +nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + +static void * +nv10_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *templ) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + return draw_create_vertex_shader(nv10->draw, templ); +} + +static void +nv10_vp_state_bind(struct pipe_context *pipe, void *shader) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); + + nv10->dirty |= NV10_NEW_VERTPROG; +} + +static void +nv10_vp_state_delete(struct pipe_context *pipe, void *shader) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); +} + +static void * +nv10_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv10_fragment_program *fp; + + fp = CALLOC(1, sizeof(struct nv10_fragment_program)); + fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); + + tgsi_scan_shader(cso->tokens, &fp->info); + + return (void *)fp; +} + +static void +nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_fragment_program *fp = hwcso; + + nv10->fragprog.current = fp; + nv10->dirty |= NV10_NEW_FRAGPROG; +} + +static void +nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_fragment_program *fp = hwcso; + + nv10_fragprog_destroy(nv10, fp); + FREE((void*)fp->pipe.tokens); + FREE(fp); +} + +static void +nv10_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->blend_color = (struct pipe_blend_color*)bcol; + + nv10->dirty |= NV10_NEW_BLENDCOL; +} + +static void +nv10_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_set_clip_state(nv10->draw, clip); +} + +static void +nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + + assert(shader < PIPE_SHADER_TYPES); + assert(index == 0); + + if (buf) { + void *mapped; + if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ))) + { + memcpy(nv10->constbuf[shader], mapped, buf->size); + nv10->constbuf_nr[shader] = + buf->size / (4 * sizeof(float)); + ws->buffer_unmap(ws, buf->buffer); + } + } +} + +static void +nv10_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->framebuffer = (struct pipe_framebuffer_state*)fb; + + nv10->dirty |= NV10_NEW_FRAMEBUFFER; +} + +static void +nv10_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + NOUVEAU_ERR("line stipple hahaha\n"); +} + +static void +nv10_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->scissor = (struct pipe_scissor_state*)s; + + nv10->dirty |= NV10_NEW_SCISSOR; +} + +static void +nv10_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->viewport = (struct pipe_viewport_state*)vpt; + + draw_set_viewport_state(nv10->draw, nv10->viewport); + + nv10->dirty |= NV10_NEW_VIEWPORT; +} + +static void +nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *vb) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); + nv10->dirty |= NV10_NEW_VTXARRAYS; + + draw_set_vertex_buffers(nv10->draw, count, vb); +} + +static void +nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *ve) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); + nv10->dirty |= NV10_NEW_VTXARRAYS; + + draw_set_vertex_elements(nv10->draw, count, ve); +} + +void +nv10_init_state_functions(struct nv10_context *nv10) +{ + nv10->pipe.create_blend_state = nv10_blend_state_create; + nv10->pipe.bind_blend_state = nv10_blend_state_bind; + nv10->pipe.delete_blend_state = nv10_blend_state_delete; + + nv10->pipe.create_sampler_state = nv10_sampler_state_create; + nv10->pipe.bind_sampler_states = nv10_sampler_state_bind; + nv10->pipe.delete_sampler_state = nv10_sampler_state_delete; + nv10->pipe.set_sampler_textures = nv10_set_sampler_texture; + + nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create; + nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind; + nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete; + + nv10->pipe.create_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_create; + nv10->pipe.bind_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_bind; + nv10->pipe.delete_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_delete; + + nv10->pipe.create_vs_state = nv10_vp_state_create; + nv10->pipe.bind_vs_state = nv10_vp_state_bind; + nv10->pipe.delete_vs_state = nv10_vp_state_delete; + + nv10->pipe.create_fs_state = nv10_fp_state_create; + nv10->pipe.bind_fs_state = nv10_fp_state_bind; + nv10->pipe.delete_fs_state = nv10_fp_state_delete; + + nv10->pipe.set_blend_color = nv10_set_blend_color; + nv10->pipe.set_clip_state = nv10_set_clip_state; + nv10->pipe.set_constant_buffer = nv10_set_constant_buffer; + nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state; + nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple; + nv10->pipe.set_scissor_state = nv10_set_scissor_state; + nv10->pipe.set_viewport_state = nv10_set_viewport_state; + + nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers; + nv10->pipe.set_vertex_elements = nv10_set_vertex_elements; +} + diff --git a/src/gallium/drivers/nv20/nv10_state.h b/src/gallium/drivers/nv20/nv10_state.h new file mode 100644 index 00000000000..3a3fd0d4f4f --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_state.h @@ -0,0 +1,139 @@ +#ifndef __NV10_STATE_H__ +#define __NV10_STATE_H__ + +#include "pipe/p_state.h" +#include "tgsi/tgsi_scan.h" + +struct nv10_blend_state { + uint32_t b_enable; + uint32_t b_srcfunc; + uint32_t b_dstfunc; + + uint32_t c_mask; + + uint32_t d_enable; +}; + +struct nv10_sampler_state { + uint32_t wrap; + uint32_t en; + uint32_t filt; + uint32_t bcol; +}; + +struct nv10_rasterizer_state { + uint32_t shade_model; + + uint32_t line_width; + uint32_t line_smooth_en; + + uint32_t point_size; + + uint32_t poly_smooth_en; + + uint32_t poly_mode_front; + uint32_t poly_mode_back; + + uint32_t front_face; + uint32_t cull_face; + uint32_t cull_face_en; + + uint32_t point_sprite; + + const struct pipe_rasterizer_state *templ; +}; + +struct nv10_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv10_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + +struct nv10_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + struct nv10_vertex_program_exec *insns; + unsigned nr_insns; + struct nv10_vertex_program_data *consts; + unsigned nr_consts; + + struct nouveau_resource *exec; + unsigned exec_start; + struct nouveau_resource *data; + unsigned data_start; + unsigned data_start_min; + + uint32_t ir; + uint32_t or; +}; + +struct nv10_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv10_fragment_program { + struct pipe_shader_state pipe; + struct tgsi_shader_info info; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv10_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + uint32_t fp_reg_control; +}; + + +struct nv10_depth_stencil_alpha_state { + struct { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; + } depth; + + struct { + uint32_t enable; + uint32_t wmask; + uint32_t func; + uint32_t ref; + uint32_t vmask; + uint32_t fail; + uint32_t zfail; + uint32_t zpass; + } stencil; + + struct { + uint32_t enabled; + uint32_t func; + uint32_t ref; + } alpha; +}; + +struct nv10_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +#endif diff --git a/src/gallium/drivers/nv20/nv10_state_emit.c b/src/gallium/drivers/nv20/nv10_state_emit.c new file mode 100644 index 00000000000..46c7e1d7536 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_state_emit.c @@ -0,0 +1,303 @@ +#include "nv10_context.h" +#include "nv10_state.h" + +static void nv10_state_emit_blend(struct nv10_context* nv10) +{ + struct nv10_blend_state *b = nv10->blend; + + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); + OUT_RING (b->d_enable); + + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); + OUT_RING (b->b_enable); + OUT_RING (b->b_srcfunc); + OUT_RING (b->b_dstfunc); + + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (b->c_mask); +} + +static void nv10_state_emit_blend_color(struct nv10_context* nv10) +{ + struct pipe_blend_color *c = nv10->blend_color; + + BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); + OUT_RING ((float_to_ubyte(c->color[3]) << 24)| + (float_to_ubyte(c->color[0]) << 16)| + (float_to_ubyte(c->color[1]) << 8) | + (float_to_ubyte(c->color[2]) << 0)); +} + +static void nv10_state_emit_rast(struct nv10_context* nv10) +{ + struct nv10_rasterizer_state *r = nv10->rast; + + BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); + OUT_RING (r->shade_model); + OUT_RING (r->line_width); + + + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (r->point_size); + + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (r->poly_mode_front); + OUT_RING (r->poly_mode_back); + + + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (r->cull_face); + OUT_RING (r->front_face); + + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); + OUT_RING (r->line_smooth_en); + OUT_RING (r->poly_smooth_en); + + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (r->cull_face_en); +} + +static void nv10_state_emit_dsa(struct nv10_context* nv10) +{ + struct nv10_depth_stencil_alpha_state *d = nv10->dsa; + + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + OUT_RING (d->depth.func); + + BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + OUT_RING (d->depth.write_enable); + + BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + OUT_RING (d->depth.test_enable); + +#if 0 + BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); + OUT_RING (d->stencil.enable); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); + OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7); +#endif + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (d->alpha.enabled); + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 1); + OUT_RING (d->alpha.func); + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_REF, 1); + OUT_RING (d->alpha.ref); +} + +static void nv10_state_emit_viewport(struct nv10_context* nv10) +{ +} + +static void nv10_state_emit_scissor(struct nv10_context* nv10) +{ + // XXX this is so not working +/* struct pipe_scissor_state *s = nv10->scissor; + BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ +} + +static void nv10_state_emit_framebuffer(struct nv10_context* nv10) +{ + struct pipe_framebuffer_state* fb = nv10->framebuffer; + struct pipe_surface *rt, *zeta = NULL; + uint32_t rt_format, w, h; + int colour_format = 0, zeta_format = 0; + + w = fb->cbufs[0]->width; + h = fb->cbufs[0]->height; + colour_format = fb->cbufs[0]->format; + rt = fb->cbufs[0]; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + if (zeta) { + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING (rt->stride | (zeta->stride << 16)); + } else { + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING (rt->stride | (rt->stride << 16)); + } + + nv10->rt[0] = rt->buffer; + + if (zeta_format) + { + nv10->zeta = zeta->buffer; + } + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((w - 1) << 16) | 0 | 0x08000800); + OUT_RING (((h - 1) << 16) | 0 | 0x08000800); +} + +static void nv10_vertex_layout(struct nv10_context *nv10) +{ + struct nv10_fragment_program *fp = nv10->fragprog.current; + uint32_t src = 0; + int i; + struct vertex_info vinfo; + + memset(&vinfo, 0, sizeof(vinfo)); + + for (i = 0; i < fp->info.num_inputs; i++) { + switch (fp->info.input_semantic_name[i]) { + case TGSI_SEMANTIC_POSITION: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + case TGSI_SEMANTIC_COLOR: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + default: + case TGSI_SEMANTIC_GENERIC: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + case TGSI_SEMANTIC_FOG: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + } + } + draw_compute_vertex_size(&vinfo); +} + +void +nv10_emit_hw_state(struct nv10_context *nv10) +{ + int i; + + if (nv10->dirty & NV10_NEW_VERTPROG) { + //nv10_vertprog_bind(nv10, nv10->vertprog.current); + nv10->dirty &= ~NV10_NEW_VERTPROG; + } + + if (nv10->dirty & NV10_NEW_FRAGPROG) { + nv10_fragprog_bind(nv10, nv10->fragprog.current); + /*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */ + nv10->dirty_samplers |= (1<<10); + nv10->dirty_samplers = 0; + } + + if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) { + nv10_fragtex_bind(nv10); + nv10->dirty &= ~NV10_NEW_FRAGPROG; + } + + if (nv10->dirty & NV10_NEW_VTXARRAYS) { + nv10->dirty &= ~NV10_NEW_VTXARRAYS; + nv10_vertex_layout(nv10); + nv10_vtxbuf_bind(nv10); + } + + if (nv10->dirty & NV10_NEW_BLEND) { + nv10->dirty &= ~NV10_NEW_BLEND; + nv10_state_emit_blend(nv10); + } + + if (nv10->dirty & NV10_NEW_BLENDCOL) { + nv10->dirty &= ~NV10_NEW_BLENDCOL; + nv10_state_emit_blend_color(nv10); + } + + if (nv10->dirty & NV10_NEW_RAST) { + nv10->dirty &= ~NV10_NEW_RAST; + nv10_state_emit_rast(nv10); + } + + if (nv10->dirty & NV10_NEW_DSA) { + nv10->dirty &= ~NV10_NEW_DSA; + nv10_state_emit_dsa(nv10); + } + + if (nv10->dirty & NV10_NEW_VIEWPORT) { + nv10->dirty &= ~NV10_NEW_VIEWPORT; + nv10_state_emit_viewport(nv10); + } + + if (nv10->dirty & NV10_NEW_SCISSOR) { + nv10->dirty &= ~NV10_NEW_SCISSOR; + nv10_state_emit_scissor(nv10); + } + + if (nv10->dirty & NV10_NEW_FRAMEBUFFER) { + nv10->dirty &= ~NV10_NEW_FRAMEBUFFER; + nv10_state_emit_framebuffer(nv10); + } + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv10_vbo.c + */ + + /* Render target */ +// XXX figre out who's who for NV10TCL_DMA_* and fill accordingly +// BEGIN_RING(celsius, NV10TCL_DMA_COLOR0, 1); +// OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + if (nv10->zeta) { +// XXX +// BEGIN_RING(celsius, NV10TCL_DMA_ZETA, 1); +// OUT_RELOCo(nv10->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + /* XXX for when we allocate LMA on nv17 */ +/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1); + OUT_RELOCl(nv10->zeta + lma_offset);*/ + } + + /* Vertex buffer */ + BEGIN_RING(celsius, NV10TCL_DMA_VTXBUF0, 1); + OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + /* Texture images */ + for (i = 0; i < 2; i++) { + if (!(nv10->fp_samplers & (1 << i))) + continue; + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 1); + OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + BEGIN_RING(celsius, NV10TCL_TX_FORMAT(i), 1); + OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0, + NV10TCL_TX_FORMAT_DMA1); + } +} + diff --git a/src/gallium/drivers/nv20/nv10_surface.c b/src/gallium/drivers/nv20/nv10_surface.c new file mode 100644 index 00000000000..875e4c58589 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_surface.c @@ -0,0 +1,64 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv10_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/u_tile.h" + +static void +nv10_surface_copy(struct pipe_context *pipe, unsigned do_flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv10_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv10_init_surface_functions(struct nv10_context *nv10) +{ + nv10->pipe.surface_copy = nv10_surface_copy; + nv10->pipe.surface_fill = nv10_surface_fill; +} diff --git a/src/gallium/drivers/nv20/nv10_vbo.c b/src/gallium/drivers/nv20/nv10_vbo.c new file mode 100644 index 00000000000..d0e788ac036 --- /dev/null +++ b/src/gallium/drivers/nv20/nv10_vbo.c @@ -0,0 +1,77 @@ +#include "draw/draw_context.h" +#include "pipe/p_context.h" +#include "pipe/p_state.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_pushbuf.h" + +boolean nv10_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count) +{ + struct nv10_context *nv10 = nv10_context( pipe ); + struct draw_context *draw = nv10->draw; + unsigned i; + + nv10_emit_hw_state(nv10); + + /* + * Map vertex buffers + */ + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (nv10->vtxbuf[i].buffer) { + void *buf + = pipe->winsys->buffer_map(pipe->winsys, + nv10->vtxbuf[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(draw, i, buf); + } + } + /* Map index buffer, if present */ + if (indexBuffer) { + void *mapped_indexes + = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes); + } + else { + /* no index/element buffer */ + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + draw_set_mapped_constant_buffer(draw, + nv10->constbuf[PIPE_SHADER_VERTEX], + nv10->constbuf_nr[PIPE_SHADER_VERTEX]); + + /* draw! */ + draw_arrays(nv10->draw, prim, start, count); + + /* + * unmap vertex/index buffers + */ + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (nv10->vtxbuf[i].buffer) { + pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer); + draw_set_mapped_vertex_buffer(draw, i, NULL); + } + } + if (indexBuffer) { + pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + return TRUE; +} + +boolean nv10_draw_arrays( struct pipe_context *pipe, + unsigned prim, unsigned start, unsigned count) +{ + return nv10_draw_elements(pipe, NULL, 0, prim, start, count); +} + + + -- cgit v1.2.3 From f910371b3736e7d09b82d42e0dd1295482817883 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sat, 8 Nov 2008 17:09:16 +0200 Subject: Nouveau: copy nv30_vertprog.c to nv20. Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv20/nv30_vertprog.c | 838 +++++++++++++++++++++++++++++++ 1 file changed, 838 insertions(+) create mode 100644 src/gallium/drivers/nv20/nv30_vertprog.c diff --git a/src/gallium/drivers/nv20/nv30_vertprog.c b/src/gallium/drivers/nv20/nv30_vertprog.c new file mode 100644 index 00000000000..72824559e8b --- /dev/null +++ b/src/gallium/drivers/nv20/nv30_vertprog.c @@ -0,0 +1,838 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_dump.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +/* TODO (at least...): + * 1. Indexed consts + ARL + * 2. Arb. swz/negation + * 3. NV_vp11, NV_vp2, NV_vp3 features + * - extra arith opcodes + * - branching + * - texture sampling + * - indexed attribs + * - indexed results + * 4. bugs + */ + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 8 +#define MASK_Y 4 +#define MASK_Z 2 +#define MASK_W 1 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE 0 +#define DEF_CTEST 0 +#include "nv30_shader.h" + +#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv30_sr_neg((s)) +#define abs(s) nv30_sr_abs((s)) + +struct nv30_vpc { + struct nv30_vertex_program *vp; + + struct nv30_vertex_program_exec *vpi; + + unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; + + int high_temp; + int temp_temp_count; + + struct nv30_sreg *imm; + unsigned nr_imm; +}; + +static struct nv30_sreg +temp(struct nv30_vpc *vpc) +{ + int idx; + + idx = vpc->temp_temp_count++; + idx += vpc->high_temp + 1; + return nv30_sr(NV30SR_TEMP, idx); +} + +static struct nv30_sreg +constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) +{ + struct nv30_vertex_program *vp = vpc->vp; + struct nv30_vertex_program_data *vpd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < vp->nr_consts; idx++) { + if (vp->consts[idx].index == pipe) + return nv30_sr(NV30SR_CONST, idx); + } + } + + idx = vp->nr_consts++; + vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); + vpd = &vp->consts[idx]; + + vpd->index = pipe; + vpd->value[0] = x; + vpd->value[1] = y; + vpd->value[2] = z; + vpd->value[3] = w; + return nv30_sr(NV30SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) + +static void +emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t sr = 0; + + switch (src.type) { + case NV30SR_TEMP: + sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT); + sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV30SR_INPUT: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + vp->ir |= (1 << src.index); + hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT); + break; + case NV30SR_CONST: + sr |= (NV30_VP_SRC_REG_TYPE_CONST << + NV30_VP_SRC_REG_TYPE_SHIFT); + assert(vpc->vpi->const_index == -1 || + vpc->vpi->const_index == src.index); + vpc->vpi->const_index = src.index; + break; + case NV30SR_NONE: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV30_VP_SRC_NEGATE; + + if (src.abs) + hw[0] |= (1 << (21 + pos)); + + sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) | + (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) | + (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) | + (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT)); + +/* + * |VVV| + * d�.�b + * \u/ + * + */ + + switch (pos) { + case 0: + hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >> + NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT; + hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) << + NV30_VP_INST_SRC0L_SHIFT; + break; + case 1: + hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT; + break; + case 2: + hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >> + NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT; + hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) << + NV30_VP_INST_SRC2L_SHIFT; + break; + default: + assert(0); + } +} + +static void +emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) +{ + struct nv30_vertex_program *vp = vpc->vp; + + switch (dst.type) { + case NV30SR_TEMP: + hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT); + break; + case NV30SR_OUTPUT: + switch (dst.index) { + case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; + case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; + case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; + case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; + case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; + case NV30_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; + case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; + case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; + case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; + case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; + case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; + case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; + case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; + case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + default: + break; + } + + hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); + hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); + + /*XXX: no way this is entirely correct, someone needs to + * figure out what exactly it is. + */ + hw[3] |= 0x800; + break; + default: + assert(0); + } +} + +static void +nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, + struct nv30_sreg s2) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t *hw; + + vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); + vpc->vpi = &vp->insns[vp->nr_insns - 1]; + memset(vpc->vpi, 0, sizeof(*vpc->vpi)); + vpc->vpi->const_index = -1; + + hw = vpc->vpi->data; + + hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT); + hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) | + (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) | + (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) | + (3 << NV30_VP_INST_COND_SWZ_W_SHIFT)); + + hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT); +// hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; +// hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); + + if (dst.type == NV30SR_OUTPUT) { + if (slot) + hw[3] |= (mask << NV30_VP_INST_SDEST_WRITEMASK_SHIFT); + else + hw[3] |= (mask << NV30_VP_INST_VDEST_WRITEMASK_SHIFT); + } else { + if (slot) + hw[3] |= (mask << NV30_VP_INST_STEMP_WRITEMASK_SHIFT); + else + hw[3] |= (mask << NV30_VP_INST_VTEMP_WRITEMASK_SHIFT); + } + + emit_dst(vpc, hw, slot, dst); + emit_src(vpc, hw, 0, s0); + emit_src(vpc, hw, 1, s1); + emit_src(vpc, hw, 2, s2); +} + +static INLINE struct nv30_sreg +tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv30_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); + break; + case TGSI_FILE_CONSTANT: + src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + break; + case TGSI_FILE_IMMEDIATE: + src = vpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + if (vpc->high_temp < fsrc->SrcRegister.Index) + vpc->high_temp = fsrc->SrcRegister.Index; + src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv30_sreg +tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + struct nv30_sreg dst; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + dst = nv30_sr(NV30SR_OUTPUT, + vpc->output_map[fdst->DstRegister.Index]); + + break; + case TGSI_FILE_TEMPORARY: + dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); + if (vpc->high_temp < dst.index) + vpc->high_temp = dst.index; + break; + default: + NOUVEAU_ERR("bad dst file\n"); + break; + } + + return dst; +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, + const struct tgsi_full_instruction *finst) +{ + struct nv30_sreg src[3], dst, tmp; + struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + int mask; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + vpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(vpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + /*XXX: index comparison is broken now that consts come from + * two different register files. + */ + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); + break; + case TGSI_OPCODE_ARL: + arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DST: + arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_EXP: + arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_FLR: + arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_LG2: + arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LIT: + arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LOG: + arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_MAD: + arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(vpc); + arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, + swz(src[0], X, X, X, X)); + arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(vpc, 1, OP_EX2, dst, mask, none, none, + swz(tmp, X, X, X, X)); + break; + case TGSI_OPCODE_RCP: + arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_RET: + break; + case TGSI_OPCODE_RSQ: + arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_SGE: + arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SGT: + arith(vpc, 0, OP_SGT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); + break; + case TGSI_OPCODE_XPD: + tmp = temp(vpc); + arith(vpc, 0, OP_MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV30_VP_INST_DEST_POS; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_BCOLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_BFC0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_BFC1; + } else { + NOUVEAU_ERR("bad bcolour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV30_VP_INST_DEST_FOGC; + break; + case TGSI_SEMANTIC_PSIZE: + hw = NV30_VP_INST_DEST_PSZ; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + vpc->output_map[fdec->DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv30_vertprog_prepare(struct nv30_vpc *vpc) +{ + struct tgsi_parse_context p; + int nr_imm = 0; + + tgsi_parse_init(&p, vpc->vp->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch(tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + nr_imm++; + break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (nr_imm) { + vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); + assert(vpc->imm); + } + + return TRUE; +} + +static void +nv30_vertprog_translate(struct nv30_context *nv30, + struct nv30_vertex_program *vp) +{ + struct tgsi_parse_context parse; + struct nv30_vpc *vpc = NULL; + + tgsi_dump(vp->pipe.tokens,0); + + vpc = CALLOC(1, sizeof(struct nv30_vpc)); + if (!vpc) + return; + vpc->vp = vp; + vpc->high_temp = -1; + + if (!nv30_vertprog_prepare(vpc)) { + FREE(vpc); + return; + } + + tgsi_parse_init(&parse, vp->pipe.tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_OUTPUT: + if (!nv30_vertprog_parse_decl_output(vpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); +// assert(imm->Immediate.Size == 4); + vpc->imm[vpc->nr_imm++] = + constant(vpc, -1, + imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + finst = &parse.FullToken.FullInstruction; + if (!nv30_vertprog_parse_instruction(vpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST; + vp->translated = TRUE; +out_err: + tgsi_parse_free(&parse); + FREE(vpc); +} + +static boolean +nv30_vertprog_validate(struct nv30_context *nv30) +{ + struct nouveau_winsys *nvws = nv30->nvws; + struct pipe_winsys *ws = nv30->pipe.winsys; + struct nouveau_grobj *rankine = nv30->screen->rankine; + struct nv30_vertex_program *vp; + struct pipe_buffer *constbuf; + boolean upload_code = FALSE, upload_data = FALSE; + int i; + + vp = nv30->vertprog; + constbuf = nv30->constbuf[PIPE_SHADER_VERTEX]; + + /* Translate TGSI shader into hw bytecode */ + if (!vp->translated) { + nv30_vertprog_translate(nv30, vp); + if (!vp->translated) + return FALSE; + } + + /* Allocate hw vtxprog exec slots */ + if (!vp->exec) { + struct nouveau_resource *heap = nv30->screen->vp_exec_heap; + struct nouveau_stateobj *so; + uint vplen = vp->nr_insns; + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { + while (heap->next && heap->size < vplen) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->exec); + } + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) + assert(0); + } + + so = so_new(2, 0); + so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); + so_data (so, vp->exec->start); + so_ref(so, &vp->so); + + upload_code = TRUE; + } + + /* Allocate hw vtxprog const slots */ + if (vp->nr_consts && !vp->data) { + struct nouveau_resource *heap = nv30->screen->vp_data_heap; + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { + while (heap->next && heap->size < vp->nr_consts) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) + assert(0); + } + + /*XXX: handle this some day */ + assert(vp->data->start >= vp->data_start_min); + + upload_data = TRUE; + if (vp->data_start != vp->data->start) + upload_code = TRUE; + } + + /* If exec or data segments moved we need to patch the program to + * fixup offsets and register IDs. + */ + if (vp->exec_start != vp->exec->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->has_branch_offset) { + assert(0); + } + } + + vp->exec_start = vp->exec->start; + } + + if (vp->nr_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->const_index >= 0) { + vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; + vpi->data[1] |= + (vpi->const_index + vp->data->start) << + NV30_VP_INST_CONST_SRC_SHIFT; + + } + } + + vp->data_start = vp->data->start; + } + + /* Update + Upload constant values */ + if (vp->nr_consts) { + float *map = NULL; + + if (constbuf) { + map = ws->buffer_map(ws, constbuf, + PIPE_BUFFER_USAGE_CPU_READ); + } + + for (i = 0; i < vp->nr_consts; i++) { + struct nv30_vertex_program_data *vpd = &vp->consts[i]; + + if (vpd->index >= 0) { + if (!upload_data && + !memcmp(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float))) + continue; + memcpy(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float)); + } + + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (i + vp->data->start); + OUT_RINGp ((uint32_t *)vpd->value, 4); + } + + if (constbuf) { + ws->buffer_unmap(ws, constbuf); + } + } + + /* Upload vtxprog */ + if (upload_code) { +#if 0 + for (i = 0; i < vp->nr_insns; i++) { + NOUVEAU_MSG("VP inst %d: 0x%08x 0x%08x 0x%08x 0x%08x\n", + i, vp->insns[i].data[0], vp->insns[i].data[1], + vp->insns[i].data[2], vp->insns[i].data[3]); + } +#endif + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (vp->exec->start); + for (i = 0; i < vp->nr_insns; i++) { + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (vp->insns[i].data, 4); + } + } + + if (vp->so != nv30->state.hw[NV30_STATE_VERTPROG]) { + so_ref(vp->so, &nv30->state.hw[NV30_STATE_VERTPROG]); + return TRUE; + } + + return FALSE; +} + +void +nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) +{ + struct nouveau_winsys *nvws = nv30->screen->nvws; + + vp->translated = FALSE; + + if (vp->nr_insns) { + FREE(vp->insns); + vp->insns = NULL; + vp->nr_insns = 0; + } + + if (vp->nr_consts) { + FREE(vp->consts); + vp->consts = NULL; + vp->nr_consts = 0; + } + + nvws->res_free(&vp->exec); + vp->exec_start = 0; + nvws->res_free(&vp->data); + vp->data_start = 0; + vp->data_start_min = 0; + + vp->ir = vp->or = 0; + so_ref(NULL, &vp->so); +} + +struct nv30_state_entry nv30_state_vertprog = { + .validate = nv30_vertprog_validate, + .dirty = { + .pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/, + .hw = NV30_STATE_VERTPROG, + } +}; -- cgit v1.2.3 From 83bb81856066101dff85fdebea32df55ed8de4c5 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sat, 8 Nov 2008 17:20:19 +0200 Subject: Nouveau: Rename nv20/ files as nv20. Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv20/Makefile | 25 +- src/gallium/drivers/nv20/nv10_clear.c | 12 - src/gallium/drivers/nv20/nv10_context.c | 296 ---------- src/gallium/drivers/nv20/nv10_context.h | 153 ------ src/gallium/drivers/nv20/nv10_fragprog.c | 21 - src/gallium/drivers/nv20/nv10_fragtex.c | 149 ----- src/gallium/drivers/nv20/nv10_miptree.c | 147 ----- src/gallium/drivers/nv20/nv10_prim_vbuf.c | 240 --------- src/gallium/drivers/nv20/nv10_screen.c | 208 ------- src/gallium/drivers/nv20/nv10_screen.h | 22 - src/gallium/drivers/nv20/nv10_state.c | 588 -------------------- src/gallium/drivers/nv20/nv10_state.h | 139 ----- src/gallium/drivers/nv20/nv10_state_emit.c | 303 ----------- src/gallium/drivers/nv20/nv10_surface.c | 64 --- src/gallium/drivers/nv20/nv10_vbo.c | 77 --- src/gallium/drivers/nv20/nv20_clear.c | 12 + src/gallium/drivers/nv20/nv20_context.c | 296 ++++++++++ src/gallium/drivers/nv20/nv20_context.h | 153 ++++++ src/gallium/drivers/nv20/nv20_fragprog.c | 21 + src/gallium/drivers/nv20/nv20_fragtex.c | 149 +++++ src/gallium/drivers/nv20/nv20_miptree.c | 147 +++++ src/gallium/drivers/nv20/nv20_prim_vbuf.c | 240 +++++++++ src/gallium/drivers/nv20/nv20_screen.c | 208 +++++++ src/gallium/drivers/nv20/nv20_screen.h | 22 + src/gallium/drivers/nv20/nv20_state.c | 588 ++++++++++++++++++++ src/gallium/drivers/nv20/nv20_state.h | 139 +++++ src/gallium/drivers/nv20/nv20_state_emit.c | 303 +++++++++++ src/gallium/drivers/nv20/nv20_surface.c | 64 +++ src/gallium/drivers/nv20/nv20_vbo.c | 77 +++ src/gallium/drivers/nv20/nv20_vertprog.c | 838 +++++++++++++++++++++++++++++ src/gallium/drivers/nv20/nv30_vertprog.c | 838 ----------------------------- 31 files changed, 3270 insertions(+), 3269 deletions(-) delete mode 100644 src/gallium/drivers/nv20/nv10_clear.c delete mode 100644 src/gallium/drivers/nv20/nv10_context.c delete mode 100644 src/gallium/drivers/nv20/nv10_context.h delete mode 100644 src/gallium/drivers/nv20/nv10_fragprog.c delete mode 100644 src/gallium/drivers/nv20/nv10_fragtex.c delete mode 100644 src/gallium/drivers/nv20/nv10_miptree.c delete mode 100644 src/gallium/drivers/nv20/nv10_prim_vbuf.c delete mode 100644 src/gallium/drivers/nv20/nv10_screen.c delete mode 100644 src/gallium/drivers/nv20/nv10_screen.h delete mode 100644 src/gallium/drivers/nv20/nv10_state.c delete mode 100644 src/gallium/drivers/nv20/nv10_state.h delete mode 100644 src/gallium/drivers/nv20/nv10_state_emit.c delete mode 100644 src/gallium/drivers/nv20/nv10_surface.c delete mode 100644 src/gallium/drivers/nv20/nv10_vbo.c create mode 100644 src/gallium/drivers/nv20/nv20_clear.c create mode 100644 src/gallium/drivers/nv20/nv20_context.c create mode 100644 src/gallium/drivers/nv20/nv20_context.h create mode 100644 src/gallium/drivers/nv20/nv20_fragprog.c create mode 100644 src/gallium/drivers/nv20/nv20_fragtex.c create mode 100644 src/gallium/drivers/nv20/nv20_miptree.c create mode 100644 src/gallium/drivers/nv20/nv20_prim_vbuf.c create mode 100644 src/gallium/drivers/nv20/nv20_screen.c create mode 100644 src/gallium/drivers/nv20/nv20_screen.h create mode 100644 src/gallium/drivers/nv20/nv20_state.c create mode 100644 src/gallium/drivers/nv20/nv20_state.h create mode 100644 src/gallium/drivers/nv20/nv20_state_emit.c create mode 100644 src/gallium/drivers/nv20/nv20_surface.c create mode 100644 src/gallium/drivers/nv20/nv20_vbo.c create mode 100644 src/gallium/drivers/nv20/nv20_vertprog.c delete mode 100644 src/gallium/drivers/nv20/nv30_vertprog.c diff --git a/src/gallium/drivers/nv20/Makefile b/src/gallium/drivers/nv20/Makefile index 4ba7ce586d6..76aafbe8f0b 100644 --- a/src/gallium/drivers/nv20/Makefile +++ b/src/gallium/drivers/nv20/Makefile @@ -1,20 +1,21 @@ TOP = ../../../.. include $(TOP)/configs/current -LIBNAME = nv10 +LIBNAME = nv20 DRIVER_SOURCES = \ - nv10_clear.c \ - nv10_context.c \ - nv10_fragprog.c \ - nv10_fragtex.c \ - nv10_miptree.c \ - nv10_prim_vbuf.c \ - nv10_screen.c \ - nv10_state.c \ - nv10_state_emit.c \ - nv10_surface.c \ - nv10_vbo.c + nv20_clear.c \ + nv20_context.c \ + nv20_fragprog.c \ + nv20_fragtex.c \ + nv20_miptree.c \ + nv20_prim_vbuf.c \ + nv20_screen.c \ + nv20_state.c \ + nv20_state_emit.c \ + nv20_surface.c \ + nv20_vbo.c \ + nv20_vertprog.c C_SOURCES = \ $(COMMON_SOURCES) \ diff --git a/src/gallium/drivers/nv20/nv10_clear.c b/src/gallium/drivers/nv20/nv10_clear.c deleted file mode 100644 index be7e09cf4b0..00000000000 --- a/src/gallium/drivers/nv20/nv10_clear.c +++ /dev/null @@ -1,12 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "nv10_context.h" - -void -nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue) -{ - pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); -} diff --git a/src/gallium/drivers/nv20/nv10_context.c b/src/gallium/drivers/nv20/nv10_context.c deleted file mode 100644 index e9b61daae7f..00000000000 --- a/src/gallium/drivers/nv20/nv10_context.c +++ /dev/null @@ -1,296 +0,0 @@ -#include "draw/draw_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" - -#include "nv10_context.h" -#include "nv10_screen.h" - -static void -nv10_flush(struct pipe_context *pipe, unsigned flags, - struct pipe_fence_handle **fence) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - draw_flush(nv10->draw); - - FIRE_RING(fence); -} - -static void -nv10_destroy(struct pipe_context *pipe) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - if (nv10->draw) - draw_destroy(nv10->draw); - - FREE(nv10); -} - -static void nv10_init_hwctx(struct nv10_context *nv10) -{ - struct nv10_screen *screen = nv10->screen; - struct nouveau_winsys *nvws = screen->nvws; - int i; - float projectionmatrix[16]; - - BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); - OUT_RING (screen->sync->handle); - BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY2, 2); - OUT_RING (nvws->channel->vram->handle); - OUT_RING (nvws->channel->vram->handle); - - BEGIN_RING(celsius, NV10TCL_NOP, 1); - OUT_RING (0); - - BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 2); - OUT_RING (0); - OUT_RING (0); - - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1); - OUT_RING ((0x7ff<<16)|0x800); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1); - OUT_RING ((0x7ff<<16)|0x800); - - for (i=1;i<8;i++) { - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(i), 1); - OUT_RING (0); - } - - BEGIN_RING(celsius, 0x290, 1); - OUT_RING ((0x10<<16)|1); - BEGIN_RING(celsius, 0x3f4, 1); - OUT_RING (0); - - BEGIN_RING(celsius, NV10TCL_NOP, 1); - OUT_RING (0); - - if (nv10->screen->celsius->grclass != NV10TCL) { - /* For nv11, nv17 */ - BEGIN_RING(celsius, 0x120, 3); - OUT_RING (0); - OUT_RING (1); - OUT_RING (2); - - BEGIN_RING(celsius, NV10TCL_NOP, 1); - OUT_RING (0); - } - - BEGIN_RING(celsius, NV10TCL_NOP, 1); - OUT_RING (0); - - /* Set state */ - BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 2); - OUT_RING (0x207); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2); - OUT_RING (0); - OUT_RING (0); - - BEGIN_RING(celsius, NV10TCL_RC_IN_ALPHA(0), 12); - OUT_RING (0x30141010); - OUT_RING (0); - OUT_RING (0x20040000); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0x00000c00); - OUT_RING (0); - OUT_RING (0x00000c00); - OUT_RING (0x18000000); - OUT_RING (0x300e0300); - OUT_RING (0x0c091c80); - - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2); - OUT_RING (1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VERTEX_WEIGHT_ENABLE, 2); - OUT_RING (0); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_SRC, 4); - OUT_RING (1); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0x8006); - BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 8); - OUT_RING (0xff); - OUT_RING (0x207); - OUT_RING (0); - OUT_RING (0xff); - OUT_RING (0x1e00); - OUT_RING (0x1e00); - OUT_RING (0x1e00); - OUT_RING (0x1d01); - BEGIN_RING(celsius, NV10TCL_NORMALIZE_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 2); - OUT_RING (0); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_LIGHT_MODEL, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_COLOR_CONTROL, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_ENABLED_LIGHTS, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); - OUT_RING (0x201); - BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_FACTOR, 2); - OUT_RING (0); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); - OUT_RING (8); - BEGIN_RING(celsius, NV10TCL_POINT_PARAMETERS_ENABLE, 2); - OUT_RING (0); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_LINE_WIDTH, 1); - OUT_RING (8); - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); - OUT_RING (0x1b02); - OUT_RING (0x1b02); - BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); - OUT_RING (0x405); - OUT_RING (0x901); - BEGIN_RING(celsius, NV10TCL_POLYGON_SMOOTH_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_CLIP_PLANE_ENABLE(0), 8); - for (i=0;i<8;i++) { - OUT_RING (0); - } - BEGIN_RING(celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3); - OUT_RING (0x3fc00000); /* -1.50 */ - OUT_RING (0xbdb8aa0a); /* -0.09 */ - OUT_RING (0); /* 0.00 */ - - BEGIN_RING(celsius, NV10TCL_NOP, 1); - OUT_RING (0); - - BEGIN_RING(celsius, NV10TCL_FOG_MODE, 2); - OUT_RING (0x802); - OUT_RING (2); - /* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when - * using texturing, except when using the texture matrix - */ - BEGIN_RING(celsius, NV10TCL_VIEW_MATRIX_ENABLE, 1); - OUT_RING (6); - BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); - OUT_RING (0x01010101); - - /* Set vertex component */ - BEGIN_RING(celsius, NV10TCL_VERTEX_COL_4F_R, 4); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_COL2_3F_R, 3); - OUT_RING (0); - OUT_RING (0); - OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VERTEX_NOR_3F_X, 3); - OUT_RING (0); - OUT_RING (0); - OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_TX0_4F_S, 4); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_TX1_4F_S, 4); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (0.0); - OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_FOG_1F, 1); - OUT_RINGf (0.0); - BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1); - OUT_RING (1); - - memset(projectionmatrix, 0, sizeof(projectionmatrix)); - BEGIN_RING(celsius, NV10TCL_PROJECTION_MATRIX(0), 16); - projectionmatrix[0*4+0] = 1.0; - projectionmatrix[1*4+1] = 1.0; - projectionmatrix[2*4+2] = 1.0; - projectionmatrix[3*4+3] = 1.0; - for (i=0;i<16;i++) { - OUT_RINGf (projectionmatrix[i]); - } - - BEGIN_RING(celsius, NV10TCL_DEPTH_RANGE_NEAR, 2); - OUT_RING (0.0); - OUT_RINGf (16777216.0); - - BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); - OUT_RINGf (-2048.0); - OUT_RINGf (-2048.0); - OUT_RINGf (16777215.0 * 0.5); - OUT_RING (0); - - FIRE_RING (NULL); -} - -static void -nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) -{ -} - -struct pipe_context * -nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) -{ - struct nv10_screen *screen = nv10_screen(pscreen); - struct pipe_winsys *ws = pscreen->winsys; - struct nv10_context *nv10; - struct nouveau_winsys *nvws = screen->nvws; - - nv10 = CALLOC(1, sizeof(struct nv10_context)); - if (!nv10) - return NULL; - nv10->screen = screen; - nv10->pctx_id = pctx_id; - - nv10->nvws = nvws; - - nv10->pipe.winsys = ws; - nv10->pipe.screen = pscreen; - nv10->pipe.destroy = nv10_destroy; - nv10->pipe.set_edgeflags = nv10_set_edgeflags; - nv10->pipe.draw_arrays = nv10_draw_arrays; - nv10->pipe.draw_elements = nv10_draw_elements; - nv10->pipe.clear = nv10_clear; - nv10->pipe.flush = nv10_flush; - - nv10_init_surface_functions(nv10); - nv10_init_state_functions(nv10); - - nv10->draw = draw_create(); - assert(nv10->draw); - draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10)); - - nv10_init_hwctx(nv10); - - return &nv10->pipe; -} - diff --git a/src/gallium/drivers/nv20/nv10_context.h b/src/gallium/drivers/nv20/nv10_context.h deleted file mode 100644 index f3b56de25a7..00000000000 --- a/src/gallium/drivers/nv20/nv10_context.h +++ /dev/null @@ -1,153 +0,0 @@ -#ifndef __NV10_CONTEXT_H__ -#define __NV10_CONTEXT_H__ - -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" -#include "pipe/p_compiler.h" - -#include "util/u_memory.h" -#include "util/u_math.h" - -#include "draw/draw_vertex.h" - -#include "nouveau/nouveau_winsys.h" -#include "nouveau/nouveau_gldefs.h" - -#define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv10_screen *ctx = nv10->screen -#include "nouveau/nouveau_push.h" - -#include "nv10_state.h" - -#define NOUVEAU_ERR(fmt, args...) \ - fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); -#define NOUVEAU_MSG(fmt, args...) \ - fprintf(stderr, "nouveau: "fmt, ##args); - -#define NV10_NEW_VERTPROG (1 << 0) -#define NV10_NEW_FRAGPROG (1 << 1) -#define NV10_NEW_VTXARRAYS (1 << 2) -#define NV10_NEW_BLEND (1 << 3) -#define NV10_NEW_BLENDCOL (1 << 4) -#define NV10_NEW_RAST (1 << 5) -#define NV10_NEW_DSA (1 << 6) -#define NV10_NEW_VIEWPORT (1 << 7) -#define NV10_NEW_SCISSOR (1 << 8) -#define NV10_NEW_FRAMEBUFFER (1 << 9) - -#include "nv10_screen.h" - -struct nv10_context { - struct pipe_context pipe; - - struct nouveau_winsys *nvws; - struct nv10_screen *screen; - unsigned pctx_id; - - struct draw_context *draw; - - uint32_t dirty; - - struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct nv10_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; - unsigned dirty_samplers; - unsigned fp_samplers; - unsigned vp_samplers; - - uint32_t rt_enable; - struct pipe_buffer *rt[4]; - struct pipe_buffer *zeta; - uint32_t lma_offset; - - struct nv10_blend_state *blend; - struct pipe_blend_color *blend_color; - struct nv10_rasterizer_state *rast; - struct nv10_depth_stencil_alpha_state *dsa; - struct pipe_viewport_state *viewport; - struct pipe_scissor_state *scissor; - struct pipe_framebuffer_state *framebuffer; - - //struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; - float *constbuf[PIPE_SHADER_TYPES][32][4]; - unsigned constbuf_nr[PIPE_SHADER_TYPES]; - - struct vertex_info vertex_info; - - struct { - struct pipe_buffer *buffer; - uint32_t format; - } tex[2]; - - unsigned vb_enable; - struct { - struct pipe_buffer *buffer; - unsigned delta; - } vb[16]; - -/* struct { - - struct nouveau_resource *exec_heap; - struct nouveau_resource *data_heap; - - struct nv10_vertex_program *active; - - struct nv10_vertex_program *current; - } vertprog; -*/ - struct { - struct nv10_fragment_program *active; - - struct nv10_fragment_program *current; - struct pipe_buffer *constant_buf; - } fragprog; - - struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; - struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; -}; - -static INLINE struct nv10_context * -nv10_context(struct pipe_context *pipe) -{ - return (struct nv10_context *)pipe; -} - -extern void nv10_init_state_functions(struct nv10_context *nv10); -extern void nv10_init_surface_functions(struct nv10_context *nv10); - -extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen); - -/* nv10_clear.c */ -extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, - unsigned clearValue); - -/* nv10_draw.c */ -extern struct draw_stage *nv10_draw_render_stage(struct nv10_context *nv10); - -/* nv10_fragprog.c */ -extern void nv10_fragprog_bind(struct nv10_context *, - struct nv10_fragment_program *); -extern void nv10_fragprog_destroy(struct nv10_context *, - struct nv10_fragment_program *); - -/* nv10_fragtex.c */ -extern void nv10_fragtex_bind(struct nv10_context *); - -/* nv10_prim_vbuf.c */ -struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ); -extern void nv10_vtxbuf_bind(struct nv10_context* nv10); - -/* nv10_state.c and friends */ -extern void nv10_emit_hw_state(struct nv10_context *nv10); -extern void nv10_state_tex_update(struct nv10_context *nv10); - -/* nv10_vbo.c */ -extern boolean nv10_draw_arrays(struct pipe_context *, unsigned mode, - unsigned start, unsigned count); -extern boolean nv10_draw_elements( struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, - unsigned indexSize, - unsigned prim, unsigned start, unsigned count); - - -#endif diff --git a/src/gallium/drivers/nv20/nv10_fragprog.c b/src/gallium/drivers/nv20/nv10_fragprog.c deleted file mode 100644 index 698db5a16a9..00000000000 --- a/src/gallium/drivers/nv20/nv10_fragprog.c +++ /dev/null @@ -1,21 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/p_shader_tokens.h" -#include "tgsi/tgsi_parse.h" -#include "tgsi/tgsi_util.h" - -#include "nv10_context.h" - -void -nv10_fragprog_bind(struct nv10_context *nv10, struct nv10_fragment_program *fp) -{ -} - -void -nv10_fragprog_destroy(struct nv10_context *nv10, - struct nv10_fragment_program *fp) -{ -} - diff --git a/src/gallium/drivers/nv20/nv10_fragtex.c b/src/gallium/drivers/nv20/nv10_fragtex.c deleted file mode 100644 index 238634d0bb4..00000000000 --- a/src/gallium/drivers/nv20/nv10_fragtex.c +++ /dev/null @@ -1,149 +0,0 @@ -#include "nv10_context.h" - -static INLINE int log2i(int i) -{ - int r = 0; - - if (i & 0xffff0000) { - i >>= 16; - r += 16; - } - if (i & 0x0000ff00) { - i >>= 8; - r += 8; - } - if (i & 0x000000f0) { - i >>= 4; - r += 4; - } - if (i & 0x0000000c) { - i >>= 2; - r += 2; - } - if (i & 0x00000002) { - r += 1; - } - return r; -} - -#define _(m,tf) \ -{ \ - TRUE, \ - PIPE_FORMAT_##m, \ - NV10TCL_TX_FORMAT_FORMAT_##tf, \ -} - -struct nv10_texture_format { - boolean defined; - uint pipe; - int format; -}; - -static struct nv10_texture_format -nv10_texture_formats[] = { - _(A8R8G8B8_UNORM, A8R8G8B8), - _(A1R5G5B5_UNORM, A1R5G5B5), - _(A4R4G4B4_UNORM, A4R4G4B4), - _(L8_UNORM , L8 ), - _(A8_UNORM , A8 ), - _(A8L8_UNORM , A8L8 ), -// _(RGB_DXT1 , DXT1, ), -// _(RGBA_DXT1 , DXT1, ), -// _(RGBA_DXT3 , DXT3, ), -// _(RGBA_DXT5 , DXT5, ), - {}, -}; - -static struct nv10_texture_format * -nv10_fragtex_format(uint pipe_format) -{ - struct nv10_texture_format *tf = nv10_texture_formats; - - while (tf->defined) { - if (tf->pipe == pipe_format) - return tf; - tf++; - } - - return NULL; -} - - -static void -nv10_fragtex_build(struct nv10_context *nv10, int unit) -{ -#if 0 - struct nv10_sampler_state *ps = nv10->tex_sampler[unit]; - struct nv10_miptree *nv10mt = nv10->tex_miptree[unit]; - struct pipe_texture *pt = &nv10mt->base; - struct nv10_texture_format *tf; - uint32_t txf, txs, txp; - - tf = nv10_fragtex_format(pt->format); - if (!tf || !tf->defined) { - NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); - return; - } - - txf = tf->format << 8; - txf |= (pt->last_level + 1) << 16; - txf |= log2i(pt->width[0]) << 20; - txf |= log2i(pt->height[0]) << 24; - txf |= log2i(pt->depth[0]) << 28; - txf |= 8; - - switch (pt->target) { - case PIPE_TEXTURE_CUBE: - txf |= NV10TCL_TX_FORMAT_CUBE_MAP; - /* fall-through */ - case PIPE_TEXTURE_2D: - txf |= (2<<4); - break; - case PIPE_TEXTURE_1D: - txf |= (1<<4); - break; - default: - NOUVEAU_ERR("Unknown target %d\n", pt->target); - return; - } - - BEGIN_RING(celsius, NV10TCL_TX_OFFSET(unit), 8); - OUT_RELOCl(nv10mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv10mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); - OUT_RING (ps->wrap); - OUT_RING (0x40000000); /* enable */ - OUT_RING (txs); - OUT_RING (ps->filt | 0x2000 /* magic */); - OUT_RING ((pt->width[0] << 16) | pt->height[0]); - OUT_RING (ps->bcol); -#endif -} - -void -nv10_fragtex_bind(struct nv10_context *nv10) -{ -#if 0 - struct nv10_fragment_program *fp = nv10->fragprog.active; - unsigned samplers, unit; - - samplers = nv10->fp_samplers & ~fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - BEGIN_RING(celsius, NV10TCL_TX_ENABLE(unit), 1); - OUT_RING (0); - } - - samplers = nv10->dirty_samplers & fp->samplers; - while (samplers) { - unit = ffs(samplers) - 1; - samplers &= ~(1 << unit); - - nv10_fragtex_build(nv10, unit); - } - - nv10->fp_samplers = fp->samplers; -#endif -} - diff --git a/src/gallium/drivers/nv20/nv10_miptree.c b/src/gallium/drivers/nv20/nv10_miptree.c deleted file mode 100644 index ad084e72b81..00000000000 --- a/src/gallium/drivers/nv20/nv10_miptree.c +++ /dev/null @@ -1,147 +0,0 @@ -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_inlines.h" - -#include "nv10_context.h" -#include "nv10_screen.h" - -static void -nv10_miptree_layout(struct nv10_miptree *nv10mt) -{ - struct pipe_texture *pt = &nv10mt->base; - boolean swizzled = FALSE; - uint width = pt->width[0], height = pt->height[0]; - uint offset = 0; - int nr_faces, l, f; - - if (pt->target == PIPE_TEXTURE_CUBE) { - nr_faces = 6; - } else { - nr_faces = 1; - } - - for (l = 0; l <= pt->last_level; l++) { - pt->width[l] = width; - pt->height[l] = height; - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - - if (swizzled) - nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; - else - nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; - nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; - - nv10mt->level[l].image_offset = - CALLOC(nr_faces, sizeof(unsigned)); - - width = MAX2(1, width >> 1); - height = MAX2(1, height >> 1); - - } - - for (f = 0; f < nr_faces; f++) { - for (l = 0; l <= pt->last_level; l++) { - nv10mt->level[l].image_offset[f] = offset; - offset += nv10mt->level[l].pitch * pt->height[l]; - } - } - - nv10mt->total_size = offset; -} - -static struct pipe_texture * -nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) -{ - struct pipe_winsys *ws = screen->winsys; - struct nv10_miptree *mt; - - mt = MALLOC(sizeof(struct nv10_miptree)); - if (!mt) - return NULL; - mt->base = *pt; - mt->base.refcount = 1; - mt->base.screen = screen; - - nv10_miptree_layout(mt); - - mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, - mt->total_size); - if (!mt->buffer) { - FREE(mt); - return NULL; - } - - return &mt->base; -} - -static void -nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) -{ - struct pipe_texture *mt = *pt; - - *pt = NULL; - if (--mt->refcount <= 0) { - struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt; - int l; - - pipe_buffer_reference(screen, &nv10mt->buffer, NULL); - for (l = 0; l <= mt->last_level; l++) { - if (nv10mt->level[l].image_offset) - FREE(nv10mt->level[l].image_offset); - } - FREE(nv10mt); - } -} - -static void -nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, - uint face, uint levels) -{ -} - - -static struct pipe_surface * -nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, - unsigned face, unsigned level, unsigned zslice, - unsigned flags) -{ - struct pipe_winsys *ws = screen->winsys; - struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt; - struct pipe_surface *ps; - - ps = ws->surface_alloc(ws); - if (!ps) - return NULL; - pipe_buffer_reference(screen, &ps->buffer, nv10mt->buffer); - ps->format = pt->format; - ps->width = pt->width[level]; - ps->height = pt->height[level]; - ps->block = pt->block; - ps->nblocksx = pt->nblocksx[level]; - ps->nblocksy = pt->nblocksy[level]; - ps->stride = nv10mt->level[level].pitch; - - if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv10mt->level[level].image_offset[face]; - } else { - ps->offset = nv10mt->level[level].image_offset[0]; - } - - return ps; -} - -static void -nv10_miptree_surface_release(struct pipe_screen *screen, - struct pipe_surface **surface) -{ -} - -void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) -{ - pscreen->texture_create = nv10_miptree_create; - pscreen->texture_release = nv10_miptree_release; - pscreen->get_tex_surface = nv10_miptree_surface_get; - pscreen->tex_surface_release = nv10_miptree_surface_release; -} - diff --git a/src/gallium/drivers/nv20/nv10_prim_vbuf.c b/src/gallium/drivers/nv20/nv10_prim_vbuf.c deleted file mode 100644 index 62a8f6d89da..00000000000 --- a/src/gallium/drivers/nv20/nv10_prim_vbuf.c +++ /dev/null @@ -1,240 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/** - * \file - * Build post-transformation, post-clipping vertex buffers and element - * lists by hooking into the end of the primitive pipeline and - * manipulating the vertex_id field in the vertex headers. - * - * XXX: work in progress - * - * \author José Fonseca - * \author Keith Whitwell - */ - - -#include "pipe/p_debug.h" -#include "pipe/p_inlines.h" -#include "pipe/p_winsys.h" - -#include "nv10_context.h" -#include "nv10_state.h" - -#include "draw/draw_vbuf.h" - -/** - * Primitive renderer for nv10. - */ -struct nv10_vbuf_render { - struct vbuf_render base; - - struct nv10_context *nv10; - - /** Vertex buffer */ - struct pipe_buffer* buffer; - - /** Vertex size in bytes */ - unsigned vertex_size; - - /** Hardware primitive */ - unsigned hwprim; -}; - - -void nv10_vtxbuf_bind( struct nv10_context* nv10 ) -{ - int i; - for(i = 0; i < 8; i++) { - BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1); - OUT_RING(0/*nv10->vtxbuf*/); - BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1); - OUT_RING(0/*XXX*/); - } -} - -/** - * Basically a cast wrapper. - */ -static INLINE struct nv10_vbuf_render * -nv10_vbuf_render( struct vbuf_render *render ) -{ - assert(render); - return (struct nv10_vbuf_render *)render; -} - - -static const struct vertex_info * -nv10_vbuf_render_get_vertex_info( struct vbuf_render *render ) -{ - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; - - nv10_emit_hw_state(nv10); - - return &nv10->vertex_info; -} - - -static void * -nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, - ushort vertex_size, - ushort nr_vertices ) -{ - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; - struct pipe_winsys *winsys = nv10->pipe.winsys; - size_t size = (size_t)vertex_size * (size_t)nr_vertices; - - assert(!nv10_render->buffer); - nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); - - nv10->dirty |= NV10_NEW_VTXARRAYS; - - return winsys->buffer_map(winsys, - nv10_render->buffer, - PIPE_BUFFER_USAGE_CPU_WRITE); -} - - -static void -nv10_vbuf_render_set_primitive( struct vbuf_render *render, - unsigned prim ) -{ - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - nv10_render->hwprim = prim + 1; -} - - -static void -nv10_vbuf_render_draw( struct vbuf_render *render, - const ushort *indices, - uint nr_indices) -{ - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; - int push, i; - - nv10_emit_hw_state(nv10); - - BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); - OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - - BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); - OUT_RING(nv10_render->hwprim); - - if (nr_indices & 1) { - BEGIN_RING(celsius, NV10TCL_VB_ELEMENT_U32, 1); - OUT_RING (indices[0]); - indices++; nr_indices--; - } - - while (nr_indices) { - // XXX too big/small ? check the size - push = MIN2(nr_indices, 1200 * 2); - - BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1); - for (i = 0; i < push; i+=2) - OUT_RING((indices[i+1] << 16) | indices[i]); - - nr_indices -= push; - indices += push; - } - - BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); - OUT_RING (0); -} - - -static void -nv10_vbuf_render_release_vertices( struct vbuf_render *render, - void *vertices, - unsigned vertex_size, - unsigned vertices_used ) -{ - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; - struct pipe_winsys *winsys = nv10->pipe.winsys; - struct pipe_screen *pscreen = &nv10->screen->pipe; - - assert(nv10_render->buffer); - winsys->buffer_unmap(winsys, nv10_render->buffer); - pipe_buffer_reference(pscreen, &nv10_render->buffer, NULL); -} - - -static void -nv10_vbuf_render_destroy( struct vbuf_render *render ) -{ - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - FREE(nv10_render); -} - - -/** - * Create a new primitive render. - */ -static struct vbuf_render * -nv10_vbuf_render_create( struct nv10_context *nv10 ) -{ - struct nv10_vbuf_render *nv10_render = CALLOC_STRUCT(nv10_vbuf_render); - - nv10_render->nv10 = nv10; - - nv10_render->base.max_vertex_buffer_bytes = 16*1024; - nv10_render->base.max_indices = 1024; - nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info; - nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices; - nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive; - nv10_render->base.draw = nv10_vbuf_render_draw; - nv10_render->base.release_vertices = nv10_vbuf_render_release_vertices; - nv10_render->base.destroy = nv10_vbuf_render_destroy; - - return &nv10_render->base; -} - - -/** - * Create a new primitive vbuf/render stage. - */ -struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ) -{ - struct vbuf_render *render; - struct draw_stage *stage; - - render = nv10_vbuf_render_create(nv10); - if(!render) - return NULL; - - stage = draw_vbuf_stage( nv10->draw, render ); - if(!stage) { - render->destroy(render); - return NULL; - } - - return stage; -} diff --git a/src/gallium/drivers/nv20/nv10_screen.c b/src/gallium/drivers/nv20/nv10_screen.c deleted file mode 100644 index 27a9edf9bba..00000000000 --- a/src/gallium/drivers/nv20/nv10_screen.c +++ /dev/null @@ -1,208 +0,0 @@ -#include "pipe/p_screen.h" - -#include "nv10_context.h" -#include "nv10_screen.h" - -static const char * -nv10_screen_get_name(struct pipe_screen *screen) -{ - struct nv10_screen *nv10screen = nv10_screen(screen); - struct nouveau_device *dev = nv10screen->nvws->channel->device; - static char buffer[128]; - - snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); - return buffer; -} - -static const char * -nv10_screen_get_vendor(struct pipe_screen *screen) -{ - return "nouveau"; -} - -static int -nv10_screen_get_param(struct pipe_screen *screen, int param) -{ - switch (param) { - case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: - return 2; - case PIPE_CAP_NPOT_TEXTURES: - return 0; - case PIPE_CAP_TWO_SIDED_STENCIL: - return 0; - case PIPE_CAP_GLSL: - return 0; - case PIPE_CAP_S3TC: - return 0; - case PIPE_CAP_ANISOTROPIC_FILTER: - return 1; - case PIPE_CAP_POINT_SPRITE: - return 0; - case PIPE_CAP_MAX_RENDER_TARGETS: - return 1; - case PIPE_CAP_OCCLUSION_QUERY: - return 0; - case PIPE_CAP_TEXTURE_SHADOW_MAP: - return 0; - case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: - return 12; - case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 0; - case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: - return 12; - case NOUVEAU_CAP_HW_VTXBUF: - case NOUVEAU_CAP_HW_IDXBUF: - return 0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0; - } -} - -static float -nv10_screen_get_paramf(struct pipe_screen *screen, int param) -{ - switch (param) { - case PIPE_CAP_MAX_LINE_WIDTH: - case PIPE_CAP_MAX_LINE_WIDTH_AA: - return 10.0; - case PIPE_CAP_MAX_POINT_WIDTH: - case PIPE_CAP_MAX_POINT_WIDTH_AA: - return 64.0; - case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: - return 2.0; - case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: - return 4.0; - default: - NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); - return 0.0; - } -} - -static boolean -nv10_screen_is_format_supported(struct pipe_screen *screen, - enum pipe_format format, - enum pipe_texture_target target, - unsigned tex_usage, unsigned geom_flags) -{ - if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_Z24S8_UNORM: - case PIPE_FORMAT_Z16_UNORM: - return TRUE; - default: - break; - } - } else { - switch (format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A1R5G5B5_UNORM: - case PIPE_FORMAT_A4R4G4B4_UNORM: - case PIPE_FORMAT_R5G6B5_UNORM: - case PIPE_FORMAT_L8_UNORM: - case PIPE_FORMAT_A8_UNORM: - case PIPE_FORMAT_I8_UNORM: - return TRUE; - default: - break; - } - } - - return FALSE; -} - -static void * -nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, - unsigned flags ) -{ - struct pipe_winsys *ws = screen->winsys; - void *map; - - map = ws->buffer_map(ws, surface->buffer, flags); - if (!map) - return NULL; - - return map + surface->offset; -} - -static void -nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) -{ - struct pipe_winsys *ws = screen->winsys; - - ws->buffer_unmap(ws, surface->buffer); -} - -static void -nv10_screen_destroy(struct pipe_screen *pscreen) -{ - struct nv10_screen *screen = nv10_screen(pscreen); - struct nouveau_winsys *nvws = screen->nvws; - - nvws->notifier_free(&screen->sync); - nvws->grobj_free(&screen->celsius); - - FREE(pscreen); -} - -struct pipe_screen * -nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) -{ - struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen); - unsigned celsius_class; - unsigned chipset = nvws->channel->device->chipset; - int ret; - - if (!screen) - return NULL; - screen->nvws = nvws; - - /* 3D object */ - if (chipset>=0x20) - celsius_class=NV11TCL; - else if (chipset>=0x17) - celsius_class=NV17TCL; - else if (chipset>=0x11) - celsius_class=NV11TCL; - else - celsius_class=NV10TCL; - - if (!celsius_class) { - NOUVEAU_ERR("Unknown nv1x chipset: nv%02x\n", chipset); - return NULL; - } - - ret = nvws->grobj_alloc(nvws, celsius_class, &screen->celsius); - if (ret) { - NOUVEAU_ERR("Error creating 3D object: %d\n", ret); - return FALSE; - } - - /* Notifier for sync purposes */ - ret = nvws->notifier_alloc(nvws, 1, &screen->sync); - if (ret) { - NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv10_screen_destroy(&screen->pipe); - return NULL; - } - - screen->pipe.winsys = ws; - screen->pipe.destroy = nv10_screen_destroy; - - screen->pipe.get_name = nv10_screen_get_name; - screen->pipe.get_vendor = nv10_screen_get_vendor; - screen->pipe.get_param = nv10_screen_get_param; - screen->pipe.get_paramf = nv10_screen_get_paramf; - - screen->pipe.is_format_supported = nv10_screen_is_format_supported; - - screen->pipe.surface_map = nv10_surface_map; - screen->pipe.surface_unmap = nv10_surface_unmap; - - nv10_screen_init_miptree_functions(&screen->pipe); - - return &screen->pipe; -} - diff --git a/src/gallium/drivers/nv20/nv10_screen.h b/src/gallium/drivers/nv20/nv10_screen.h deleted file mode 100644 index 3f8750a13f7..00000000000 --- a/src/gallium/drivers/nv20/nv10_screen.h +++ /dev/null @@ -1,22 +0,0 @@ -#ifndef __NV10_SCREEN_H__ -#define __NV10_SCREEN_H__ - -#include "pipe/p_screen.h" - -struct nv10_screen { - struct pipe_screen pipe; - - struct nouveau_winsys *nvws; - - /* HW graphics objects */ - struct nouveau_grobj *celsius; - struct nouveau_notifier *sync; -}; - -static INLINE struct nv10_screen * -nv10_screen(struct pipe_screen *screen) -{ - return (struct nv10_screen *)screen; -} - -#endif diff --git a/src/gallium/drivers/nv20/nv10_state.c b/src/gallium/drivers/nv20/nv10_state.c deleted file mode 100644 index d2375aa2f64..00000000000 --- a/src/gallium/drivers/nv20/nv10_state.c +++ /dev/null @@ -1,588 +0,0 @@ -#include "draw/draw_context.h" -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "pipe/p_shader_tokens.h" - -#include "tgsi/tgsi_parse.h" - -#include "nv10_context.h" -#include "nv10_state.h" - -static void * -nv10_blend_state_create(struct pipe_context *pipe, - const struct pipe_blend_state *cso) -{ - struct nv10_blend_state *cb; - - cb = MALLOC(sizeof(struct nv10_blend_state)); - - cb->b_enable = cso->blend_enable ? 1 : 0; - cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | - (nvgl_blend_func(cso->rgb_src_factor))); - cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | - (nvgl_blend_func(cso->rgb_dst_factor))); - - cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | - ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | - ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | - ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); - - cb->d_enable = cso->dither ? 1 : 0; - - return (void *)cb; -} - -static void -nv10_blend_state_bind(struct pipe_context *pipe, void *blend) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->blend = (struct nv10_blend_state*)blend; - - nv10->dirty |= NV10_NEW_BLEND; -} - -static void -nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso) -{ - FREE(hwcso); -} - - -static INLINE unsigned -wrap_mode(unsigned wrap) { - unsigned ret; - - switch (wrap) { - case PIPE_TEX_WRAP_REPEAT: - ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; - break; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - ret = NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT; - break; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE; - break; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER; - break; - case PIPE_TEX_WRAP_CLAMP: - ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP; - break; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - case PIPE_TEX_WRAP_MIRROR_CLAMP: - default: - NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); - ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; - break; - } - - return ret >> NV10TCL_TX_FORMAT_WRAP_S_SHIFT; -} - -static void * -nv10_sampler_state_create(struct pipe_context *pipe, - const struct pipe_sampler_state *cso) -{ - struct nv10_sampler_state *ps; - uint32_t filter = 0; - - ps = MALLOC(sizeof(struct nv10_sampler_state)); - - ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) | - (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT)); - - ps->en = 0; - if (cso->max_anisotropy > 1.0) { - /* no idea, binary driver sets it, works without it.. meh.. */ - ps->wrap |= (1 << 5); - -/* if (cso->max_anisotropy >= 16.0) { - ps->en |= NV10TCL_TX_ENABLE_ANISO_16X; - } else - if (cso->max_anisotropy >= 12.0) { - ps->en |= NV10TCL_TX_ENABLE_ANISO_12X; - } else - if (cso->max_anisotropy >= 10.0) { - ps->en |= NV10TCL_TX_ENABLE_ANISO_10X; - } else - if (cso->max_anisotropy >= 8.0) { - ps->en |= NV10TCL_TX_ENABLE_ANISO_8X; - } else - if (cso->max_anisotropy >= 6.0) { - ps->en |= NV10TCL_TX_ENABLE_ANISO_6X; - } else - if (cso->max_anisotropy >= 4.0) { - ps->en |= NV10TCL_TX_ENABLE_ANISO_4X; - } else { - ps->en |= NV10TCL_TX_ENABLE_ANISO_2X; - }*/ - } - - switch (cso->mag_img_filter) { - case PIPE_TEX_FILTER_LINEAR: - filter |= NV10TCL_TX_FILTER_MAGNIFY_LINEAR; - break; - case PIPE_TEX_FILTER_NEAREST: - default: - filter |= NV10TCL_TX_FILTER_MAGNIFY_NEAREST; - break; - } - - switch (cso->min_img_filter) { - case PIPE_TEX_FILTER_LINEAR: - switch (cso->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NEAREST: - filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; - break; - case PIPE_TEX_MIPFILTER_LINEAR: - filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; - break; - case PIPE_TEX_MIPFILTER_NONE: - default: - filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR; - break; - } - break; - case PIPE_TEX_FILTER_NEAREST: - default: - switch (cso->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NEAREST: - filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; - break; - case PIPE_TEX_MIPFILTER_LINEAR: - filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; - break; - case PIPE_TEX_MIPFILTER_NONE: - default: - filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST; - break; - } - break; - } - - ps->filt = filter; - -/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - switch (cso->compare_func) { - case PIPE_FUNC_NEVER: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER; - break; - case PIPE_FUNC_GREATER: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER; - break; - case PIPE_FUNC_EQUAL: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL; - break; - case PIPE_FUNC_GEQUAL: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL; - break; - case PIPE_FUNC_LESS: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS; - break; - case PIPE_FUNC_NOTEQUAL: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL; - break; - case PIPE_FUNC_LEQUAL: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL; - break; - case PIPE_FUNC_ALWAYS: - ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS; - break; - default: - break; - } - }*/ - - ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | - (float_to_ubyte(cso->border_color[0]) << 16) | - (float_to_ubyte(cso->border_color[1]) << 8) | - (float_to_ubyte(cso->border_color[2]) << 0)); - - return (void *)ps; -} - -static void -nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) -{ - struct nv10_context *nv10 = nv10_context(pipe); - unsigned unit; - - for (unit = 0; unit < nr; unit++) { - nv10->tex_sampler[unit] = sampler[unit]; - nv10->dirty_samplers |= (1 << unit); - } -} - -static void -nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso) -{ - FREE(hwcso); -} - -static void -nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr, - struct pipe_texture **miptree) -{ - struct nv10_context *nv10 = nv10_context(pipe); - unsigned unit; - - for (unit = 0; unit < nr; unit++) { - nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit]; - nv10->dirty_samplers |= (1 << unit); - } -} - -static void * -nv10_rasterizer_state_create(struct pipe_context *pipe, - const struct pipe_rasterizer_state *cso) -{ - struct nv10_rasterizer_state *rs; - int i; - - /*XXX: ignored: - * light_twoside - * offset_cw/ccw -nohw - * scissor - * point_smooth -nohw - * multisample - * offset_units / offset_scale - */ - rs = MALLOC(sizeof(struct nv10_rasterizer_state)); - - rs->templ = cso; - - rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; - - rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; - rs->line_smooth_en = cso->line_smooth ? 1 : 0; - - rs->point_size = *(uint32_t*)&cso->point_size; - - rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; - - if (cso->front_winding == PIPE_WINDING_CCW) { - rs->front_face = NV10TCL_FRONT_FACE_CCW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); - } else { - rs->front_face = NV10TCL_FRONT_FACE_CW; - rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); - rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); - } - - switch (cso->cull_mode) { - case PIPE_WINDING_CCW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CCW) - rs->cull_face = NV10TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV10TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_CW: - rs->cull_face_en = 1; - if (cso->front_winding == PIPE_WINDING_CW) - rs->cull_face = NV10TCL_CULL_FACE_FRONT; - else - rs->cull_face = NV10TCL_CULL_FACE_BACK; - break; - case PIPE_WINDING_BOTH: - rs->cull_face_en = 1; - rs->cull_face = NV10TCL_CULL_FACE_FRONT_AND_BACK; - break; - case PIPE_WINDING_NONE: - default: - rs->cull_face_en = 0; - rs->cull_face = 0; - break; - } - - if (cso->point_sprite) { - rs->point_sprite = (1 << 0); - for (i = 0; i < 8; i++) { - if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) - rs->point_sprite |= (1 << (8 + i)); - } - } else { - rs->point_sprite = 0; - } - - return (void *)rs; -} - -static void -nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->rast = (struct nv10_rasterizer_state*)rast; - - draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL)); - - nv10->dirty |= NV10_NEW_RAST; -} - -static void -nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) -{ - FREE(hwcso); -} - -static void * -nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, - const struct pipe_depth_stencil_alpha_state *cso) -{ - struct nv10_depth_stencil_alpha_state *hw; - - hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state)); - - hw->depth.func = nvgl_comparison_op(cso->depth.func); - hw->depth.write_enable = cso->depth.writemask ? 1 : 0; - hw->depth.test_enable = cso->depth.enabled ? 1 : 0; - - hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0; - hw->stencil.wmask = cso->stencil[0].write_mask; - hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func); - hw->stencil.ref = cso->stencil[0].ref_value; - hw->stencil.vmask = cso->stencil[0].value_mask; - hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op); - hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op); - hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op); - - hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; - hw->alpha.func = nvgl_comparison_op(cso->alpha.func); - hw->alpha.ref = float_to_ubyte(cso->alpha.ref); - - return (void *)hw; -} - -static void -nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa; - - nv10->dirty |= NV10_NEW_DSA; -} - -static void -nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) -{ - FREE(hwcso); -} - -static void * -nv10_vp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *templ) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - return draw_create_vertex_shader(nv10->draw, templ); -} - -static void -nv10_vp_state_bind(struct pipe_context *pipe, void *shader) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); - - nv10->dirty |= NV10_NEW_VERTPROG; -} - -static void -nv10_vp_state_delete(struct pipe_context *pipe, void *shader) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); -} - -static void * -nv10_fp_state_create(struct pipe_context *pipe, - const struct pipe_shader_state *cso) -{ - struct nv10_fragment_program *fp; - - fp = CALLOC(1, sizeof(struct nv10_fragment_program)); - fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); - - tgsi_scan_shader(cso->tokens, &fp->info); - - return (void *)fp; -} - -static void -nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso) -{ - struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_fragment_program *fp = hwcso; - - nv10->fragprog.current = fp; - nv10->dirty |= NV10_NEW_FRAGPROG; -} - -static void -nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) -{ - struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_fragment_program *fp = hwcso; - - nv10_fragprog_destroy(nv10, fp); - FREE((void*)fp->pipe.tokens); - FREE(fp); -} - -static void -nv10_set_blend_color(struct pipe_context *pipe, - const struct pipe_blend_color *bcol) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->blend_color = (struct pipe_blend_color*)bcol; - - nv10->dirty |= NV10_NEW_BLENDCOL; -} - -static void -nv10_set_clip_state(struct pipe_context *pipe, - const struct pipe_clip_state *clip) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - draw_set_clip_state(nv10->draw, clip); -} - -static void -nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, - const struct pipe_constant_buffer *buf ) -{ - struct nv10_context *nv10 = nv10_context(pipe); - struct pipe_winsys *ws = pipe->winsys; - - assert(shader < PIPE_SHADER_TYPES); - assert(index == 0); - - if (buf) { - void *mapped; - if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ))) - { - memcpy(nv10->constbuf[shader], mapped, buf->size); - nv10->constbuf_nr[shader] = - buf->size / (4 * sizeof(float)); - ws->buffer_unmap(ws, buf->buffer); - } - } -} - -static void -nv10_set_framebuffer_state(struct pipe_context *pipe, - const struct pipe_framebuffer_state *fb) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->framebuffer = (struct pipe_framebuffer_state*)fb; - - nv10->dirty |= NV10_NEW_FRAMEBUFFER; -} - -static void -nv10_set_polygon_stipple(struct pipe_context *pipe, - const struct pipe_poly_stipple *stipple) -{ - NOUVEAU_ERR("line stipple hahaha\n"); -} - -static void -nv10_set_scissor_state(struct pipe_context *pipe, - const struct pipe_scissor_state *s) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->scissor = (struct pipe_scissor_state*)s; - - nv10->dirty |= NV10_NEW_SCISSOR; -} - -static void -nv10_set_viewport_state(struct pipe_context *pipe, - const struct pipe_viewport_state *vpt) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - nv10->viewport = (struct pipe_viewport_state*)vpt; - - draw_set_viewport_state(nv10->draw, nv10->viewport); - - nv10->dirty |= NV10_NEW_VIEWPORT; -} - -static void -nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, - const struct pipe_vertex_buffer *vb) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); - nv10->dirty |= NV10_NEW_VTXARRAYS; - - draw_set_vertex_buffers(nv10->draw, count, vb); -} - -static void -nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, - const struct pipe_vertex_element *ve) -{ - struct nv10_context *nv10 = nv10_context(pipe); - - memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); - nv10->dirty |= NV10_NEW_VTXARRAYS; - - draw_set_vertex_elements(nv10->draw, count, ve); -} - -void -nv10_init_state_functions(struct nv10_context *nv10) -{ - nv10->pipe.create_blend_state = nv10_blend_state_create; - nv10->pipe.bind_blend_state = nv10_blend_state_bind; - nv10->pipe.delete_blend_state = nv10_blend_state_delete; - - nv10->pipe.create_sampler_state = nv10_sampler_state_create; - nv10->pipe.bind_sampler_states = nv10_sampler_state_bind; - nv10->pipe.delete_sampler_state = nv10_sampler_state_delete; - nv10->pipe.set_sampler_textures = nv10_set_sampler_texture; - - nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create; - nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind; - nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete; - - nv10->pipe.create_depth_stencil_alpha_state = - nv10_depth_stencil_alpha_state_create; - nv10->pipe.bind_depth_stencil_alpha_state = - nv10_depth_stencil_alpha_state_bind; - nv10->pipe.delete_depth_stencil_alpha_state = - nv10_depth_stencil_alpha_state_delete; - - nv10->pipe.create_vs_state = nv10_vp_state_create; - nv10->pipe.bind_vs_state = nv10_vp_state_bind; - nv10->pipe.delete_vs_state = nv10_vp_state_delete; - - nv10->pipe.create_fs_state = nv10_fp_state_create; - nv10->pipe.bind_fs_state = nv10_fp_state_bind; - nv10->pipe.delete_fs_state = nv10_fp_state_delete; - - nv10->pipe.set_blend_color = nv10_set_blend_color; - nv10->pipe.set_clip_state = nv10_set_clip_state; - nv10->pipe.set_constant_buffer = nv10_set_constant_buffer; - nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state; - nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple; - nv10->pipe.set_scissor_state = nv10_set_scissor_state; - nv10->pipe.set_viewport_state = nv10_set_viewport_state; - - nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers; - nv10->pipe.set_vertex_elements = nv10_set_vertex_elements; -} - diff --git a/src/gallium/drivers/nv20/nv10_state.h b/src/gallium/drivers/nv20/nv10_state.h deleted file mode 100644 index 3a3fd0d4f4f..00000000000 --- a/src/gallium/drivers/nv20/nv10_state.h +++ /dev/null @@ -1,139 +0,0 @@ -#ifndef __NV10_STATE_H__ -#define __NV10_STATE_H__ - -#include "pipe/p_state.h" -#include "tgsi/tgsi_scan.h" - -struct nv10_blend_state { - uint32_t b_enable; - uint32_t b_srcfunc; - uint32_t b_dstfunc; - - uint32_t c_mask; - - uint32_t d_enable; -}; - -struct nv10_sampler_state { - uint32_t wrap; - uint32_t en; - uint32_t filt; - uint32_t bcol; -}; - -struct nv10_rasterizer_state { - uint32_t shade_model; - - uint32_t line_width; - uint32_t line_smooth_en; - - uint32_t point_size; - - uint32_t poly_smooth_en; - - uint32_t poly_mode_front; - uint32_t poly_mode_back; - - uint32_t front_face; - uint32_t cull_face; - uint32_t cull_face_en; - - uint32_t point_sprite; - - const struct pipe_rasterizer_state *templ; -}; - -struct nv10_vertex_program_exec { - uint32_t data[4]; - boolean has_branch_offset; - int const_index; -}; - -struct nv10_vertex_program_data { - int index; /* immediates == -1 */ - float value[4]; -}; - -struct nv10_vertex_program { - const struct pipe_shader_state *pipe; - - boolean translated; - struct nv10_vertex_program_exec *insns; - unsigned nr_insns; - struct nv10_vertex_program_data *consts; - unsigned nr_consts; - - struct nouveau_resource *exec; - unsigned exec_start; - struct nouveau_resource *data; - unsigned data_start; - unsigned data_start_min; - - uint32_t ir; - uint32_t or; -}; - -struct nv10_fragment_program_data { - unsigned offset; - unsigned index; -}; - -struct nv10_fragment_program { - struct pipe_shader_state pipe; - struct tgsi_shader_info info; - - boolean translated; - boolean on_hw; - unsigned samplers; - - uint32_t *insn; - int insn_len; - - struct nv10_fragment_program_data *consts; - unsigned nr_consts; - - struct pipe_buffer *buffer; - - uint32_t fp_control; - uint32_t fp_reg_control; -}; - - -struct nv10_depth_stencil_alpha_state { - struct { - uint32_t func; - uint32_t write_enable; - uint32_t test_enable; - } depth; - - struct { - uint32_t enable; - uint32_t wmask; - uint32_t func; - uint32_t ref; - uint32_t vmask; - uint32_t fail; - uint32_t zfail; - uint32_t zpass; - } stencil; - - struct { - uint32_t enabled; - uint32_t func; - uint32_t ref; - } alpha; -}; - -struct nv10_miptree { - struct pipe_texture base; - - struct pipe_buffer *buffer; - uint total_size; - - struct { - uint pitch; - uint *image_offset; - } level[PIPE_MAX_TEXTURE_LEVELS]; -}; - -#endif diff --git a/src/gallium/drivers/nv20/nv10_state_emit.c b/src/gallium/drivers/nv20/nv10_state_emit.c deleted file mode 100644 index 46c7e1d7536..00000000000 --- a/src/gallium/drivers/nv20/nv10_state_emit.c +++ /dev/null @@ -1,303 +0,0 @@ -#include "nv10_context.h" -#include "nv10_state.h" - -static void nv10_state_emit_blend(struct nv10_context* nv10) -{ - struct nv10_blend_state *b = nv10->blend; - - BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); - OUT_RING (b->d_enable); - - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); - OUT_RING (b->b_enable); - OUT_RING (b->b_srcfunc); - OUT_RING (b->b_dstfunc); - - BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); - OUT_RING (b->c_mask); -} - -static void nv10_state_emit_blend_color(struct nv10_context* nv10) -{ - struct pipe_blend_color *c = nv10->blend_color; - - BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); - OUT_RING ((float_to_ubyte(c->color[3]) << 24)| - (float_to_ubyte(c->color[0]) << 16)| - (float_to_ubyte(c->color[1]) << 8) | - (float_to_ubyte(c->color[2]) << 0)); -} - -static void nv10_state_emit_rast(struct nv10_context* nv10) -{ - struct nv10_rasterizer_state *r = nv10->rast; - - BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); - OUT_RING (r->shade_model); - OUT_RING (r->line_width); - - - BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); - OUT_RING (r->point_size); - - BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); - OUT_RING (r->poly_mode_front); - OUT_RING (r->poly_mode_back); - - - BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); - OUT_RING (r->cull_face); - OUT_RING (r->front_face); - - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); - OUT_RING (r->line_smooth_en); - OUT_RING (r->poly_smooth_en); - - BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); - OUT_RING (r->cull_face_en); -} - -static void nv10_state_emit_dsa(struct nv10_context* nv10) -{ - struct nv10_depth_stencil_alpha_state *d = nv10->dsa; - - BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); - OUT_RING (d->depth.func); - - BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); - OUT_RING (d->depth.write_enable); - - BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); - OUT_RING (d->depth.test_enable); - -#if 0 - BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); - OUT_RING (d->stencil.enable); - BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); - OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7); -#endif - - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); - OUT_RING (d->alpha.enabled); - - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 1); - OUT_RING (d->alpha.func); - - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_REF, 1); - OUT_RING (d->alpha.ref); -} - -static void nv10_state_emit_viewport(struct nv10_context* nv10) -{ -} - -static void nv10_state_emit_scissor(struct nv10_context* nv10) -{ - // XXX this is so not working -/* struct pipe_scissor_state *s = nv10->scissor; - BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); - OUT_RING (((s->maxx - s->minx) << 16) | s->minx); - OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ -} - -static void nv10_state_emit_framebuffer(struct nv10_context* nv10) -{ - struct pipe_framebuffer_state* fb = nv10->framebuffer; - struct pipe_surface *rt, *zeta = NULL; - uint32_t rt_format, w, h; - int colour_format = 0, zeta_format = 0; - - w = fb->cbufs[0]->width; - h = fb->cbufs[0]->height; - colour_format = fb->cbufs[0]->format; - rt = fb->cbufs[0]; - - if (fb->zsbuf) { - if (colour_format) { - assert(w == fb->zsbuf->width); - assert(h == fb->zsbuf->height); - } else { - w = fb->zsbuf->width; - h = fb->zsbuf->height; - } - - zeta_format = fb->zsbuf->format; - zeta = fb->zsbuf; - } - - rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR; - - switch (colour_format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - case 0: - rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8; - break; - case PIPE_FORMAT_R5G6B5_UNORM: - rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5; - break; - default: - assert(0); - } - - if (zeta) { - BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING (rt->stride | (zeta->stride << 16)); - } else { - BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); - OUT_RING (rt->stride | (rt->stride << 16)); - } - - nv10->rt[0] = rt->buffer; - - if (zeta_format) - { - nv10->zeta = zeta->buffer; - } - - BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); - OUT_RING ((w << 16) | 0); - OUT_RING ((h << 16) | 0); - OUT_RING (rt_format); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); - OUT_RING (((w - 1) << 16) | 0 | 0x08000800); - OUT_RING (((h - 1) << 16) | 0 | 0x08000800); -} - -static void nv10_vertex_layout(struct nv10_context *nv10) -{ - struct nv10_fragment_program *fp = nv10->fragprog.current; - uint32_t src = 0; - int i; - struct vertex_info vinfo; - - memset(&vinfo, 0, sizeof(vinfo)); - - for (i = 0; i < fp->info.num_inputs; i++) { - switch (fp->info.input_semantic_name[i]) { - case TGSI_SEMANTIC_POSITION: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); - break; - case TGSI_SEMANTIC_COLOR: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); - break; - default: - case TGSI_SEMANTIC_GENERIC: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); - break; - case TGSI_SEMANTIC_FOG: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); - break; - } - } - draw_compute_vertex_size(&vinfo); -} - -void -nv10_emit_hw_state(struct nv10_context *nv10) -{ - int i; - - if (nv10->dirty & NV10_NEW_VERTPROG) { - //nv10_vertprog_bind(nv10, nv10->vertprog.current); - nv10->dirty &= ~NV10_NEW_VERTPROG; - } - - if (nv10->dirty & NV10_NEW_FRAGPROG) { - nv10_fragprog_bind(nv10, nv10->fragprog.current); - /*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */ - nv10->dirty_samplers |= (1<<10); - nv10->dirty_samplers = 0; - } - - if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) { - nv10_fragtex_bind(nv10); - nv10->dirty &= ~NV10_NEW_FRAGPROG; - } - - if (nv10->dirty & NV10_NEW_VTXARRAYS) { - nv10->dirty &= ~NV10_NEW_VTXARRAYS; - nv10_vertex_layout(nv10); - nv10_vtxbuf_bind(nv10); - } - - if (nv10->dirty & NV10_NEW_BLEND) { - nv10->dirty &= ~NV10_NEW_BLEND; - nv10_state_emit_blend(nv10); - } - - if (nv10->dirty & NV10_NEW_BLENDCOL) { - nv10->dirty &= ~NV10_NEW_BLENDCOL; - nv10_state_emit_blend_color(nv10); - } - - if (nv10->dirty & NV10_NEW_RAST) { - nv10->dirty &= ~NV10_NEW_RAST; - nv10_state_emit_rast(nv10); - } - - if (nv10->dirty & NV10_NEW_DSA) { - nv10->dirty &= ~NV10_NEW_DSA; - nv10_state_emit_dsa(nv10); - } - - if (nv10->dirty & NV10_NEW_VIEWPORT) { - nv10->dirty &= ~NV10_NEW_VIEWPORT; - nv10_state_emit_viewport(nv10); - } - - if (nv10->dirty & NV10_NEW_SCISSOR) { - nv10->dirty &= ~NV10_NEW_SCISSOR; - nv10_state_emit_scissor(nv10); - } - - if (nv10->dirty & NV10_NEW_FRAMEBUFFER) { - nv10->dirty &= ~NV10_NEW_FRAMEBUFFER; - nv10_state_emit_framebuffer(nv10); - } - - /* Emit relocs for every referenced buffer. - * This is to ensure the bufmgr has an accurate idea of how - * the buffer is used. This isn't very efficient, but we don't - * seem to take a significant performance hit. Will be improved - * at some point. Vertex arrays are emitted by nv10_vbo.c - */ - - /* Render target */ -// XXX figre out who's who for NV10TCL_DMA_* and fill accordingly -// BEGIN_RING(celsius, NV10TCL_DMA_COLOR0, 1); -// OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); - OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - if (nv10->zeta) { -// XXX -// BEGIN_RING(celsius, NV10TCL_DMA_ZETA, 1); -// OUT_RELOCo(nv10->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - /* XXX for when we allocate LMA on nv17 */ -/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1); - OUT_RELOCl(nv10->zeta + lma_offset);*/ - } - - /* Vertex buffer */ - BEGIN_RING(celsius, NV10TCL_DMA_VTXBUF0, 1); - OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); - OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - - /* Texture images */ - for (i = 0; i < 2; i++) { - if (!(nv10->fp_samplers & (1 << i))) - continue; - BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 1); - OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM | - NOUVEAU_BO_GART | NOUVEAU_BO_RD); - BEGIN_RING(celsius, NV10TCL_TX_FORMAT(i), 1); - OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, - NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | - NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0, - NV10TCL_TX_FORMAT_DMA1); - } -} - diff --git a/src/gallium/drivers/nv20/nv10_surface.c b/src/gallium/drivers/nv20/nv10_surface.c deleted file mode 100644 index 875e4c58589..00000000000 --- a/src/gallium/drivers/nv20/nv10_surface.c +++ /dev/null @@ -1,64 +0,0 @@ - -/************************************************************************** - * - * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "nv10_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_winsys.h" -#include "pipe/p_inlines.h" -#include "util/u_tile.h" - -static void -nv10_surface_copy(struct pipe_context *pipe, unsigned do_flip, - struct pipe_surface *dest, unsigned destx, unsigned desty, - struct pipe_surface *src, unsigned srcx, unsigned srcy, - unsigned width, unsigned height) -{ - struct nv10_context *nv10 = nv10_context(pipe); - struct nouveau_winsys *nvws = nv10->nvws; - - nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, - width, height); -} - -static void -nv10_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, - unsigned destx, unsigned desty, unsigned width, - unsigned height, unsigned value) -{ - struct nv10_context *nv10 = nv10_context(pipe); - struct nouveau_winsys *nvws = nv10->nvws; - - nvws->surface_fill(nvws, dest, destx, desty, width, height, value); -} - -void -nv10_init_surface_functions(struct nv10_context *nv10) -{ - nv10->pipe.surface_copy = nv10_surface_copy; - nv10->pipe.surface_fill = nv10_surface_fill; -} diff --git a/src/gallium/drivers/nv20/nv10_vbo.c b/src/gallium/drivers/nv20/nv10_vbo.c deleted file mode 100644 index d0e788ac036..00000000000 --- a/src/gallium/drivers/nv20/nv10_vbo.c +++ /dev/null @@ -1,77 +0,0 @@ -#include "draw/draw_context.h" -#include "pipe/p_context.h" -#include "pipe/p_state.h" - -#include "nv10_context.h" -#include "nv10_state.h" - -#include "nouveau/nouveau_channel.h" -#include "nouveau/nouveau_pushbuf.h" - -boolean nv10_draw_elements( struct pipe_context *pipe, - struct pipe_buffer *indexBuffer, - unsigned indexSize, - unsigned prim, unsigned start, unsigned count) -{ - struct nv10_context *nv10 = nv10_context( pipe ); - struct draw_context *draw = nv10->draw; - unsigned i; - - nv10_emit_hw_state(nv10); - - /* - * Map vertex buffers - */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (nv10->vtxbuf[i].buffer) { - void *buf - = pipe->winsys->buffer_map(pipe->winsys, - nv10->vtxbuf[i].buffer, - PIPE_BUFFER_USAGE_CPU_READ); - draw_set_mapped_vertex_buffer(draw, i, buf); - } - } - /* Map index buffer, if present */ - if (indexBuffer) { - void *mapped_indexes - = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, - PIPE_BUFFER_USAGE_CPU_READ); - draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes); - } - else { - /* no index/element buffer */ - draw_set_mapped_element_buffer(draw, 0, NULL); - } - - draw_set_mapped_constant_buffer(draw, - nv10->constbuf[PIPE_SHADER_VERTEX], - nv10->constbuf_nr[PIPE_SHADER_VERTEX]); - - /* draw! */ - draw_arrays(nv10->draw, prim, start, count); - - /* - * unmap vertex/index buffers - */ - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (nv10->vtxbuf[i].buffer) { - pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer); - draw_set_mapped_vertex_buffer(draw, i, NULL); - } - } - if (indexBuffer) { - pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); - draw_set_mapped_element_buffer(draw, 0, NULL); - } - - return TRUE; -} - -boolean nv10_draw_arrays( struct pipe_context *pipe, - unsigned prim, unsigned start, unsigned count) -{ - return nv10_draw_elements(pipe, NULL, 0, prim, start, count); -} - - - diff --git a/src/gallium/drivers/nv20/nv20_clear.c b/src/gallium/drivers/nv20/nv20_clear.c new file mode 100644 index 00000000000..be7e09cf4b0 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_clear.c @@ -0,0 +1,12 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "nv10_context.h" + +void +nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue) +{ + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); +} diff --git a/src/gallium/drivers/nv20/nv20_context.c b/src/gallium/drivers/nv20/nv20_context.c new file mode 100644 index 00000000000..e9b61daae7f --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_context.c @@ -0,0 +1,296 @@ +#include "draw/draw_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static void +nv10_flush(struct pipe_context *pipe, unsigned flags, + struct pipe_fence_handle **fence) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_flush(nv10->draw); + + FIRE_RING(fence); +} + +static void +nv10_destroy(struct pipe_context *pipe) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + if (nv10->draw) + draw_destroy(nv10->draw); + + FREE(nv10); +} + +static void nv10_init_hwctx(struct nv10_context *nv10) +{ + struct nv10_screen *screen = nv10->screen; + struct nouveau_winsys *nvws = screen->nvws; + int i; + float projectionmatrix[16]; + + BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); + OUT_RING (screen->sync->handle); + BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->gart->handle); + BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY2, 2); + OUT_RING (nvws->channel->vram->handle); + OUT_RING (nvws->channel->vram->handle); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 2); + OUT_RING (0); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1); + OUT_RING ((0x7ff<<16)|0x800); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1); + OUT_RING ((0x7ff<<16)|0x800); + + for (i=1;i<8;i++) { + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(i), 1); + OUT_RING (0); + } + + BEGIN_RING(celsius, 0x290, 1); + OUT_RING ((0x10<<16)|1); + BEGIN_RING(celsius, 0x3f4, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + if (nv10->screen->celsius->grclass != NV10TCL) { + /* For nv11, nv17 */ + BEGIN_RING(celsius, 0x120, 3); + OUT_RING (0); + OUT_RING (1); + OUT_RING (2); + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + } + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + /* Set state */ + BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 2); + OUT_RING (0x207); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2); + OUT_RING (0); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_RC_IN_ALPHA(0), 12); + OUT_RING (0x30141010); + OUT_RING (0); + OUT_RING (0x20040000); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x00000c00); + OUT_RING (0); + OUT_RING (0x00000c00); + OUT_RING (0x18000000); + OUT_RING (0x300e0300); + OUT_RING (0x0c091c80); + + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2); + OUT_RING (1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VERTEX_WEIGHT_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_SRC, 4); + OUT_RING (1); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0x8006); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 8); + OUT_RING (0xff); + OUT_RING (0x207); + OUT_RING (0); + OUT_RING (0xff); + OUT_RING (0x1e00); + OUT_RING (0x1e00); + OUT_RING (0x1e00); + OUT_RING (0x1d01); + BEGIN_RING(celsius, NV10TCL_NORMALIZE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LIGHT_MODEL, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_COLOR_CONTROL, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_ENABLED_LIGHTS, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + OUT_RING (0x201); + BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_FACTOR, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (8); + BEGIN_RING(celsius, NV10TCL_POINT_PARAMETERS_ENABLE, 2); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_LINE_WIDTH, 1); + OUT_RING (8); + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (0x1b02); + OUT_RING (0x1b02); + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (0x405); + OUT_RING (0x901); + BEGIN_RING(celsius, NV10TCL_POLYGON_SMOOTH_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_CLIP_PLANE_ENABLE(0), 8); + for (i=0;i<8;i++) { + OUT_RING (0); + } + BEGIN_RING(celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3); + OUT_RING (0x3fc00000); /* -1.50 */ + OUT_RING (0xbdb8aa0a); /* -0.09 */ + OUT_RING (0); /* 0.00 */ + + BEGIN_RING(celsius, NV10TCL_NOP, 1); + OUT_RING (0); + + BEGIN_RING(celsius, NV10TCL_FOG_MODE, 2); + OUT_RING (0x802); + OUT_RING (2); + /* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when + * using texturing, except when using the texture matrix + */ + BEGIN_RING(celsius, NV10TCL_VIEW_MATRIX_ENABLE, 1); + OUT_RING (6); + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (0x01010101); + + /* Set vertex component */ + BEGIN_RING(celsius, NV10TCL_VERTEX_COL_4F_R, 4); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_COL2_3F_R, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RING (0); + BEGIN_RING(celsius, NV10TCL_VERTEX_NOR_3F_X, 3); + OUT_RING (0); + OUT_RING (0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_TX0_4F_S, 4); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_TX1_4F_S, 4); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (0.0); + OUT_RINGf (1.0); + BEGIN_RING(celsius, NV10TCL_VERTEX_FOG_1F, 1); + OUT_RINGf (0.0); + BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1); + OUT_RING (1); + + memset(projectionmatrix, 0, sizeof(projectionmatrix)); + BEGIN_RING(celsius, NV10TCL_PROJECTION_MATRIX(0), 16); + projectionmatrix[0*4+0] = 1.0; + projectionmatrix[1*4+1] = 1.0; + projectionmatrix[2*4+2] = 1.0; + projectionmatrix[3*4+3] = 1.0; + for (i=0;i<16;i++) { + OUT_RINGf (projectionmatrix[i]); + } + + BEGIN_RING(celsius, NV10TCL_DEPTH_RANGE_NEAR, 2); + OUT_RING (0.0); + OUT_RINGf (16777216.0); + + BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); + OUT_RINGf (-2048.0); + OUT_RINGf (-2048.0); + OUT_RINGf (16777215.0 * 0.5); + OUT_RING (0); + + FIRE_RING (NULL); +} + +static void +nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +{ +} + +struct pipe_context * +nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) +{ + struct nv10_screen *screen = nv10_screen(pscreen); + struct pipe_winsys *ws = pscreen->winsys; + struct nv10_context *nv10; + struct nouveau_winsys *nvws = screen->nvws; + + nv10 = CALLOC(1, sizeof(struct nv10_context)); + if (!nv10) + return NULL; + nv10->screen = screen; + nv10->pctx_id = pctx_id; + + nv10->nvws = nvws; + + nv10->pipe.winsys = ws; + nv10->pipe.screen = pscreen; + nv10->pipe.destroy = nv10_destroy; + nv10->pipe.set_edgeflags = nv10_set_edgeflags; + nv10->pipe.draw_arrays = nv10_draw_arrays; + nv10->pipe.draw_elements = nv10_draw_elements; + nv10->pipe.clear = nv10_clear; + nv10->pipe.flush = nv10_flush; + + nv10_init_surface_functions(nv10); + nv10_init_state_functions(nv10); + + nv10->draw = draw_create(); + assert(nv10->draw); + draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10)); + + nv10_init_hwctx(nv10); + + return &nv10->pipe; +} + diff --git a/src/gallium/drivers/nv20/nv20_context.h b/src/gallium/drivers/nv20/nv20_context.h new file mode 100644 index 00000000000..f3b56de25a7 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_context.h @@ -0,0 +1,153 @@ +#ifndef __NV10_CONTEXT_H__ +#define __NV10_CONTEXT_H__ + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" +#include "pipe/p_compiler.h" + +#include "util/u_memory.h" +#include "util/u_math.h" + +#include "draw/draw_vertex.h" + +#include "nouveau/nouveau_winsys.h" +#include "nouveau/nouveau_gldefs.h" + +#define NOUVEAU_PUSH_CONTEXT(ctx) \ + struct nv10_screen *ctx = nv10->screen +#include "nouveau/nouveau_push.h" + +#include "nv10_state.h" + +#define NOUVEAU_ERR(fmt, args...) \ + fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); +#define NOUVEAU_MSG(fmt, args...) \ + fprintf(stderr, "nouveau: "fmt, ##args); + +#define NV10_NEW_VERTPROG (1 << 0) +#define NV10_NEW_FRAGPROG (1 << 1) +#define NV10_NEW_VTXARRAYS (1 << 2) +#define NV10_NEW_BLEND (1 << 3) +#define NV10_NEW_BLENDCOL (1 << 4) +#define NV10_NEW_RAST (1 << 5) +#define NV10_NEW_DSA (1 << 6) +#define NV10_NEW_VIEWPORT (1 << 7) +#define NV10_NEW_SCISSOR (1 << 8) +#define NV10_NEW_FRAMEBUFFER (1 << 9) + +#include "nv10_screen.h" + +struct nv10_context { + struct pipe_context pipe; + + struct nouveau_winsys *nvws; + struct nv10_screen *screen; + unsigned pctx_id; + + struct draw_context *draw; + + uint32_t dirty; + + struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv10_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + unsigned dirty_samplers; + unsigned fp_samplers; + unsigned vp_samplers; + + uint32_t rt_enable; + struct pipe_buffer *rt[4]; + struct pipe_buffer *zeta; + uint32_t lma_offset; + + struct nv10_blend_state *blend; + struct pipe_blend_color *blend_color; + struct nv10_rasterizer_state *rast; + struct nv10_depth_stencil_alpha_state *dsa; + struct pipe_viewport_state *viewport; + struct pipe_scissor_state *scissor; + struct pipe_framebuffer_state *framebuffer; + + //struct pipe_buffer *constbuf[PIPE_SHADER_TYPES]; + float *constbuf[PIPE_SHADER_TYPES][32][4]; + unsigned constbuf_nr[PIPE_SHADER_TYPES]; + + struct vertex_info vertex_info; + + struct { + struct pipe_buffer *buffer; + uint32_t format; + } tex[2]; + + unsigned vb_enable; + struct { + struct pipe_buffer *buffer; + unsigned delta; + } vb[16]; + +/* struct { + + struct nouveau_resource *exec_heap; + struct nouveau_resource *data_heap; + + struct nv10_vertex_program *active; + + struct nv10_vertex_program *current; + } vertprog; +*/ + struct { + struct nv10_fragment_program *active; + + struct nv10_fragment_program *current; + struct pipe_buffer *constant_buf; + } fragprog; + + struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS]; + struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; +}; + +static INLINE struct nv10_context * +nv10_context(struct pipe_context *pipe) +{ + return (struct nv10_context *)pipe; +} + +extern void nv10_init_state_functions(struct nv10_context *nv10); +extern void nv10_init_surface_functions(struct nv10_context *nv10); + +extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen); + +/* nv10_clear.c */ +extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, + unsigned clearValue); + +/* nv10_draw.c */ +extern struct draw_stage *nv10_draw_render_stage(struct nv10_context *nv10); + +/* nv10_fragprog.c */ +extern void nv10_fragprog_bind(struct nv10_context *, + struct nv10_fragment_program *); +extern void nv10_fragprog_destroy(struct nv10_context *, + struct nv10_fragment_program *); + +/* nv10_fragtex.c */ +extern void nv10_fragtex_bind(struct nv10_context *); + +/* nv10_prim_vbuf.c */ +struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ); +extern void nv10_vtxbuf_bind(struct nv10_context* nv10); + +/* nv10_state.c and friends */ +extern void nv10_emit_hw_state(struct nv10_context *nv10); +extern void nv10_state_tex_update(struct nv10_context *nv10); + +/* nv10_vbo.c */ +extern boolean nv10_draw_arrays(struct pipe_context *, unsigned mode, + unsigned start, unsigned count); +extern boolean nv10_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count); + + +#endif diff --git a/src/gallium/drivers/nv20/nv20_fragprog.c b/src/gallium/drivers/nv20/nv20_fragprog.c new file mode 100644 index 00000000000..698db5a16a9 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_fragprog.c @@ -0,0 +1,21 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_util.h" + +#include "nv10_context.h" + +void +nv10_fragprog_bind(struct nv10_context *nv10, struct nv10_fragment_program *fp) +{ +} + +void +nv10_fragprog_destroy(struct nv10_context *nv10, + struct nv10_fragment_program *fp) +{ +} + diff --git a/src/gallium/drivers/nv20/nv20_fragtex.c b/src/gallium/drivers/nv20/nv20_fragtex.c new file mode 100644 index 00000000000..238634d0bb4 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_fragtex.c @@ -0,0 +1,149 @@ +#include "nv10_context.h" + +static INLINE int log2i(int i) +{ + int r = 0; + + if (i & 0xffff0000) { + i >>= 16; + r += 16; + } + if (i & 0x0000ff00) { + i >>= 8; + r += 8; + } + if (i & 0x000000f0) { + i >>= 4; + r += 4; + } + if (i & 0x0000000c) { + i >>= 2; + r += 2; + } + if (i & 0x00000002) { + r += 1; + } + return r; +} + +#define _(m,tf) \ +{ \ + TRUE, \ + PIPE_FORMAT_##m, \ + NV10TCL_TX_FORMAT_FORMAT_##tf, \ +} + +struct nv10_texture_format { + boolean defined; + uint pipe; + int format; +}; + +static struct nv10_texture_format +nv10_texture_formats[] = { + _(A8R8G8B8_UNORM, A8R8G8B8), + _(A1R5G5B5_UNORM, A1R5G5B5), + _(A4R4G4B4_UNORM, A4R4G4B4), + _(L8_UNORM , L8 ), + _(A8_UNORM , A8 ), + _(A8L8_UNORM , A8L8 ), +// _(RGB_DXT1 , DXT1, ), +// _(RGBA_DXT1 , DXT1, ), +// _(RGBA_DXT3 , DXT3, ), +// _(RGBA_DXT5 , DXT5, ), + {}, +}; + +static struct nv10_texture_format * +nv10_fragtex_format(uint pipe_format) +{ + struct nv10_texture_format *tf = nv10_texture_formats; + + while (tf->defined) { + if (tf->pipe == pipe_format) + return tf; + tf++; + } + + return NULL; +} + + +static void +nv10_fragtex_build(struct nv10_context *nv10, int unit) +{ +#if 0 + struct nv10_sampler_state *ps = nv10->tex_sampler[unit]; + struct nv10_miptree *nv10mt = nv10->tex_miptree[unit]; + struct pipe_texture *pt = &nv10mt->base; + struct nv10_texture_format *tf; + uint32_t txf, txs, txp; + + tf = nv10_fragtex_format(pt->format); + if (!tf || !tf->defined) { + NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); + return; + } + + txf = tf->format << 8; + txf |= (pt->last_level + 1) << 16; + txf |= log2i(pt->width[0]) << 20; + txf |= log2i(pt->height[0]) << 24; + txf |= log2i(pt->depth[0]) << 28; + txf |= 8; + + switch (pt->target) { + case PIPE_TEXTURE_CUBE: + txf |= NV10TCL_TX_FORMAT_CUBE_MAP; + /* fall-through */ + case PIPE_TEXTURE_2D: + txf |= (2<<4); + break; + case PIPE_TEXTURE_1D: + txf |= (1<<4); + break; + default: + NOUVEAU_ERR("Unknown target %d\n", pt->target); + return; + } + + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(unit), 8); + OUT_RELOCl(nv10mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv10mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + OUT_RING (ps->wrap); + OUT_RING (0x40000000); /* enable */ + OUT_RING (txs); + OUT_RING (ps->filt | 0x2000 /* magic */); + OUT_RING ((pt->width[0] << 16) | pt->height[0]); + OUT_RING (ps->bcol); +#endif +} + +void +nv10_fragtex_bind(struct nv10_context *nv10) +{ +#if 0 + struct nv10_fragment_program *fp = nv10->fragprog.active; + unsigned samplers, unit; + + samplers = nv10->fp_samplers & ~fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + BEGIN_RING(celsius, NV10TCL_TX_ENABLE(unit), 1); + OUT_RING (0); + } + + samplers = nv10->dirty_samplers & fp->samplers; + while (samplers) { + unit = ffs(samplers) - 1; + samplers &= ~(1 << unit); + + nv10_fragtex_build(nv10, unit); + } + + nv10->fp_samplers = fp->samplers; +#endif +} + diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c new file mode 100644 index 00000000000..ad084e72b81 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_miptree.c @@ -0,0 +1,147 @@ +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_inlines.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static void +nv10_miptree_layout(struct nv10_miptree *nv10mt) +{ + struct pipe_texture *pt = &nv10mt->base; + boolean swizzled = FALSE; + uint width = pt->width[0], height = pt->height[0]; + uint offset = 0; + int nr_faces, l, f; + + if (pt->target == PIPE_TEXTURE_CUBE) { + nr_faces = 6; + } else { + nr_faces = 1; + } + + for (l = 0; l <= pt->last_level; l++) { + pt->width[l] = width; + pt->height[l] = height; + pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); + pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); + + if (swizzled) + nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; + else + nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; + nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; + + nv10mt->level[l].image_offset = + CALLOC(nr_faces, sizeof(unsigned)); + + width = MAX2(1, width >> 1); + height = MAX2(1, height >> 1); + + } + + for (f = 0; f < nr_faces; f++) { + for (l = 0; l <= pt->last_level; l++) { + nv10mt->level[l].image_offset[f] = offset; + offset += nv10mt->level[l].pitch * pt->height[l]; + } + } + + nv10mt->total_size = offset; +} + +static struct pipe_texture * +nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv10_miptree *mt; + + mt = MALLOC(sizeof(struct nv10_miptree)); + if (!mt) + return NULL; + mt->base = *pt; + mt->base.refcount = 1; + mt->base.screen = screen; + + nv10_miptree_layout(mt); + + mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, + mt->total_size); + if (!mt->buffer) { + FREE(mt); + return NULL; + } + + return &mt->base; +} + +static void +nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) +{ + struct pipe_texture *mt = *pt; + + *pt = NULL; + if (--mt->refcount <= 0) { + struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt; + int l; + + pipe_buffer_reference(screen, &nv10mt->buffer, NULL); + for (l = 0; l <= mt->last_level; l++) { + if (nv10mt->level[l].image_offset) + FREE(nv10mt->level[l].image_offset); + } + FREE(nv10mt); + } +} + +static void +nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, + uint face, uint levels) +{ +} + + +static struct pipe_surface * +nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, + unsigned face, unsigned level, unsigned zslice, + unsigned flags) +{ + struct pipe_winsys *ws = screen->winsys; + struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt; + struct pipe_surface *ps; + + ps = ws->surface_alloc(ws); + if (!ps) + return NULL; + pipe_buffer_reference(screen, &ps->buffer, nv10mt->buffer); + ps->format = pt->format; + ps->width = pt->width[level]; + ps->height = pt->height[level]; + ps->block = pt->block; + ps->nblocksx = pt->nblocksx[level]; + ps->nblocksy = pt->nblocksy[level]; + ps->stride = nv10mt->level[level].pitch; + + if (pt->target == PIPE_TEXTURE_CUBE) { + ps->offset = nv10mt->level[level].image_offset[face]; + } else { + ps->offset = nv10mt->level[level].image_offset[0]; + } + + return ps; +} + +static void +nv10_miptree_surface_release(struct pipe_screen *screen, + struct pipe_surface **surface) +{ +} + +void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) +{ + pscreen->texture_create = nv10_miptree_create; + pscreen->texture_release = nv10_miptree_release; + pscreen->get_tex_surface = nv10_miptree_surface_get; + pscreen->tex_surface_release = nv10_miptree_surface_release; +} + diff --git a/src/gallium/drivers/nv20/nv20_prim_vbuf.c b/src/gallium/drivers/nv20/nv20_prim_vbuf.c new file mode 100644 index 00000000000..62a8f6d89da --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_prim_vbuf.c @@ -0,0 +1,240 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * \file + * Build post-transformation, post-clipping vertex buffers and element + * lists by hooking into the end of the primitive pipeline and + * manipulating the vertex_id field in the vertex headers. + * + * XXX: work in progress + * + * \author José Fonseca + * \author Keith Whitwell + */ + + +#include "pipe/p_debug.h" +#include "pipe/p_inlines.h" +#include "pipe/p_winsys.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +#include "draw/draw_vbuf.h" + +/** + * Primitive renderer for nv10. + */ +struct nv10_vbuf_render { + struct vbuf_render base; + + struct nv10_context *nv10; + + /** Vertex buffer */ + struct pipe_buffer* buffer; + + /** Vertex size in bytes */ + unsigned vertex_size; + + /** Hardware primitive */ + unsigned hwprim; +}; + + +void nv10_vtxbuf_bind( struct nv10_context* nv10 ) +{ + int i; + for(i = 0; i < 8; i++) { + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1); + OUT_RING(0/*nv10->vtxbuf*/); + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1); + OUT_RING(0/*XXX*/); + } +} + +/** + * Basically a cast wrapper. + */ +static INLINE struct nv10_vbuf_render * +nv10_vbuf_render( struct vbuf_render *render ) +{ + assert(render); + return (struct nv10_vbuf_render *)render; +} + + +static const struct vertex_info * +nv10_vbuf_render_get_vertex_info( struct vbuf_render *render ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + + nv10_emit_hw_state(nv10); + + return &nv10->vertex_info; +} + + +static void * +nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, + ushort vertex_size, + ushort nr_vertices ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + struct pipe_winsys *winsys = nv10->pipe.winsys; + size_t size = (size_t)vertex_size * (size_t)nr_vertices; + + assert(!nv10_render->buffer); + nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); + + nv10->dirty |= NV10_NEW_VTXARRAYS; + + return winsys->buffer_map(winsys, + nv10_render->buffer, + PIPE_BUFFER_USAGE_CPU_WRITE); +} + + +static void +nv10_vbuf_render_set_primitive( struct vbuf_render *render, + unsigned prim ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + nv10_render->hwprim = prim + 1; +} + + +static void +nv10_vbuf_render_draw( struct vbuf_render *render, + const ushort *indices, + uint nr_indices) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + int push, i; + + nv10_emit_hw_state(nv10); + + BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); + OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + + BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING(nv10_render->hwprim); + + if (nr_indices & 1) { + BEGIN_RING(celsius, NV10TCL_VB_ELEMENT_U32, 1); + OUT_RING (indices[0]); + indices++; nr_indices--; + } + + while (nr_indices) { + // XXX too big/small ? check the size + push = MIN2(nr_indices, 1200 * 2); + + BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1); + for (i = 0; i < push; i+=2) + OUT_RING((indices[i+1] << 16) | indices[i]); + + nr_indices -= push; + indices += push; + } + + BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING (0); +} + + +static void +nv10_vbuf_render_release_vertices( struct vbuf_render *render, + void *vertices, + unsigned vertex_size, + unsigned vertices_used ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + struct nv10_context *nv10 = nv10_render->nv10; + struct pipe_winsys *winsys = nv10->pipe.winsys; + struct pipe_screen *pscreen = &nv10->screen->pipe; + + assert(nv10_render->buffer); + winsys->buffer_unmap(winsys, nv10_render->buffer); + pipe_buffer_reference(pscreen, &nv10_render->buffer, NULL); +} + + +static void +nv10_vbuf_render_destroy( struct vbuf_render *render ) +{ + struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); + FREE(nv10_render); +} + + +/** + * Create a new primitive render. + */ +static struct vbuf_render * +nv10_vbuf_render_create( struct nv10_context *nv10 ) +{ + struct nv10_vbuf_render *nv10_render = CALLOC_STRUCT(nv10_vbuf_render); + + nv10_render->nv10 = nv10; + + nv10_render->base.max_vertex_buffer_bytes = 16*1024; + nv10_render->base.max_indices = 1024; + nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info; + nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices; + nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive; + nv10_render->base.draw = nv10_vbuf_render_draw; + nv10_render->base.release_vertices = nv10_vbuf_render_release_vertices; + nv10_render->base.destroy = nv10_vbuf_render_destroy; + + return &nv10_render->base; +} + + +/** + * Create a new primitive vbuf/render stage. + */ +struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ) +{ + struct vbuf_render *render; + struct draw_stage *stage; + + render = nv10_vbuf_render_create(nv10); + if(!render) + return NULL; + + stage = draw_vbuf_stage( nv10->draw, render ); + if(!stage) { + render->destroy(render); + return NULL; + } + + return stage; +} diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c new file mode 100644 index 00000000000..27a9edf9bba --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_screen.c @@ -0,0 +1,208 @@ +#include "pipe/p_screen.h" + +#include "nv10_context.h" +#include "nv10_screen.h" + +static const char * +nv10_screen_get_name(struct pipe_screen *screen) +{ + struct nv10_screen *nv10screen = nv10_screen(screen); + struct nouveau_device *dev = nv10screen->nvws->channel->device; + static char buffer[128]; + + snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); + return buffer; +} + +static const char * +nv10_screen_get_vendor(struct pipe_screen *screen) +{ + return "nouveau"; +} + +static int +nv10_screen_get_param(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: + return 2; + case PIPE_CAP_NPOT_TEXTURES: + return 0; + case PIPE_CAP_TWO_SIDED_STENCIL: + return 0; + case PIPE_CAP_GLSL: + return 0; + case PIPE_CAP_S3TC: + return 0; + case PIPE_CAP_ANISOTROPIC_FILTER: + return 1; + case PIPE_CAP_POINT_SPRITE: + return 0; + case PIPE_CAP_MAX_RENDER_TARGETS: + return 1; + case PIPE_CAP_OCCLUSION_QUERY: + return 0; + case PIPE_CAP_TEXTURE_SHADOW_MAP: + return 0; + case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: + return 12; + case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: + return 0; + case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: + return 12; + case NOUVEAU_CAP_HW_VTXBUF: + case NOUVEAU_CAP_HW_IDXBUF: + return 0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0; + } +} + +static float +nv10_screen_get_paramf(struct pipe_screen *screen, int param) +{ + switch (param) { + case PIPE_CAP_MAX_LINE_WIDTH: + case PIPE_CAP_MAX_LINE_WIDTH_AA: + return 10.0; + case PIPE_CAP_MAX_POINT_WIDTH: + case PIPE_CAP_MAX_POINT_WIDTH_AA: + return 64.0; + case PIPE_CAP_MAX_TEXTURE_ANISOTROPY: + return 2.0; + case PIPE_CAP_MAX_TEXTURE_LOD_BIAS: + return 4.0; + default: + NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param); + return 0.0; + } +} + +static boolean +nv10_screen_is_format_supported(struct pipe_screen *screen, + enum pipe_format format, + enum pipe_texture_target target, + unsigned tex_usage, unsigned geom_flags) +{ + if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) { + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_Z16_UNORM: + return TRUE; + default: + break; + } + } else { + switch (format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_A1R5G5B5_UNORM: + case PIPE_FORMAT_A4R4G4B4_UNORM: + case PIPE_FORMAT_R5G6B5_UNORM: + case PIPE_FORMAT_L8_UNORM: + case PIPE_FORMAT_A8_UNORM: + case PIPE_FORMAT_I8_UNORM: + return TRUE; + default: + break; + } + } + + return FALSE; +} + +static void * +nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, + unsigned flags ) +{ + struct pipe_winsys *ws = screen->winsys; + void *map; + + map = ws->buffer_map(ws, surface->buffer, flags); + if (!map) + return NULL; + + return map + surface->offset; +} + +static void +nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +{ + struct pipe_winsys *ws = screen->winsys; + + ws->buffer_unmap(ws, surface->buffer); +} + +static void +nv10_screen_destroy(struct pipe_screen *pscreen) +{ + struct nv10_screen *screen = nv10_screen(pscreen); + struct nouveau_winsys *nvws = screen->nvws; + + nvws->notifier_free(&screen->sync); + nvws->grobj_free(&screen->celsius); + + FREE(pscreen); +} + +struct pipe_screen * +nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) +{ + struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen); + unsigned celsius_class; + unsigned chipset = nvws->channel->device->chipset; + int ret; + + if (!screen) + return NULL; + screen->nvws = nvws; + + /* 3D object */ + if (chipset>=0x20) + celsius_class=NV11TCL; + else if (chipset>=0x17) + celsius_class=NV17TCL; + else if (chipset>=0x11) + celsius_class=NV11TCL; + else + celsius_class=NV10TCL; + + if (!celsius_class) { + NOUVEAU_ERR("Unknown nv1x chipset: nv%02x\n", chipset); + return NULL; + } + + ret = nvws->grobj_alloc(nvws, celsius_class, &screen->celsius); + if (ret) { + NOUVEAU_ERR("Error creating 3D object: %d\n", ret); + return FALSE; + } + + /* Notifier for sync purposes */ + ret = nvws->notifier_alloc(nvws, 1, &screen->sync); + if (ret) { + NOUVEAU_ERR("Error creating notifier object: %d\n", ret); + nv10_screen_destroy(&screen->pipe); + return NULL; + } + + screen->pipe.winsys = ws; + screen->pipe.destroy = nv10_screen_destroy; + + screen->pipe.get_name = nv10_screen_get_name; + screen->pipe.get_vendor = nv10_screen_get_vendor; + screen->pipe.get_param = nv10_screen_get_param; + screen->pipe.get_paramf = nv10_screen_get_paramf; + + screen->pipe.is_format_supported = nv10_screen_is_format_supported; + + screen->pipe.surface_map = nv10_surface_map; + screen->pipe.surface_unmap = nv10_surface_unmap; + + nv10_screen_init_miptree_functions(&screen->pipe); + + return &screen->pipe; +} + diff --git a/src/gallium/drivers/nv20/nv20_screen.h b/src/gallium/drivers/nv20/nv20_screen.h new file mode 100644 index 00000000000..3f8750a13f7 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_screen.h @@ -0,0 +1,22 @@ +#ifndef __NV10_SCREEN_H__ +#define __NV10_SCREEN_H__ + +#include "pipe/p_screen.h" + +struct nv10_screen { + struct pipe_screen pipe; + + struct nouveau_winsys *nvws; + + /* HW graphics objects */ + struct nouveau_grobj *celsius; + struct nouveau_notifier *sync; +}; + +static INLINE struct nv10_screen * +nv10_screen(struct pipe_screen *screen) +{ + return (struct nv10_screen *)screen; +} + +#endif diff --git a/src/gallium/drivers/nv20/nv20_state.c b/src/gallium/drivers/nv20/nv20_state.c new file mode 100644 index 00000000000..d2375aa2f64 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_state.c @@ -0,0 +1,588 @@ +#include "draw/draw_context.h" +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/p_shader_tokens.h" + +#include "tgsi/tgsi_parse.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +static void * +nv10_blend_state_create(struct pipe_context *pipe, + const struct pipe_blend_state *cso) +{ + struct nv10_blend_state *cb; + + cb = MALLOC(sizeof(struct nv10_blend_state)); + + cb->b_enable = cso->blend_enable ? 1 : 0; + cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | + (nvgl_blend_func(cso->rgb_src_factor))); + cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) | + (nvgl_blend_func(cso->rgb_dst_factor))); + + cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) | + ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) | + ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) | + ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0)); + + cb->d_enable = cso->dither ? 1 : 0; + + return (void *)cb; +} + +static void +nv10_blend_state_bind(struct pipe_context *pipe, void *blend) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->blend = (struct nv10_blend_state*)blend; + + nv10->dirty |= NV10_NEW_BLEND; +} + +static void +nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + + +static INLINE unsigned +wrap_mode(unsigned wrap) { + unsigned ret; + + switch (wrap) { + case PIPE_TEX_WRAP_REPEAT: + ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; + break; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + ret = NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT; + break; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE; + break; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER; + break; + case PIPE_TEX_WRAP_CLAMP: + ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP; + break; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + case PIPE_TEX_WRAP_MIRROR_CLAMP: + default: + NOUVEAU_ERR("unknown wrap mode: %d\n", wrap); + ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT; + break; + } + + return ret >> NV10TCL_TX_FORMAT_WRAP_S_SHIFT; +} + +static void * +nv10_sampler_state_create(struct pipe_context *pipe, + const struct pipe_sampler_state *cso) +{ + struct nv10_sampler_state *ps; + uint32_t filter = 0; + + ps = MALLOC(sizeof(struct nv10_sampler_state)); + + ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) | + (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT)); + + ps->en = 0; + if (cso->max_anisotropy > 1.0) { + /* no idea, binary driver sets it, works without it.. meh.. */ + ps->wrap |= (1 << 5); + +/* if (cso->max_anisotropy >= 16.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_16X; + } else + if (cso->max_anisotropy >= 12.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_12X; + } else + if (cso->max_anisotropy >= 10.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_10X; + } else + if (cso->max_anisotropy >= 8.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_8X; + } else + if (cso->max_anisotropy >= 6.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_6X; + } else + if (cso->max_anisotropy >= 4.0) { + ps->en |= NV10TCL_TX_ENABLE_ANISO_4X; + } else { + ps->en |= NV10TCL_TX_ENABLE_ANISO_2X; + }*/ + } + + switch (cso->mag_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MAGNIFY_LINEAR; + break; + case PIPE_TEX_FILTER_NEAREST: + default: + filter |= NV10TCL_TX_FILTER_MAGNIFY_NEAREST; + break; + } + + switch (cso->min_img_filter) { + case PIPE_TEX_FILTER_LINEAR: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR; + break; + } + break; + case PIPE_TEX_FILTER_NEAREST: + default: + switch (cso->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NEAREST: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST; + break; + case PIPE_TEX_MIPFILTER_LINEAR: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR; + break; + case PIPE_TEX_MIPFILTER_NONE: + default: + filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST; + break; + } + break; + } + + ps->filt = filter; + +/* if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { + switch (cso->compare_func) { + case PIPE_FUNC_NEVER: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER; + break; + case PIPE_FUNC_GREATER: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER; + break; + case PIPE_FUNC_EQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL; + break; + case PIPE_FUNC_GEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL; + break; + case PIPE_FUNC_LESS: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS; + break; + case PIPE_FUNC_NOTEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL; + break; + case PIPE_FUNC_LEQUAL: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL; + break; + case PIPE_FUNC_ALWAYS: + ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS; + break; + default: + break; + } + }*/ + + ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) | + (float_to_ubyte(cso->border_color[0]) << 16) | + (float_to_ubyte(cso->border_color[1]) << 8) | + (float_to_ubyte(cso->border_color[2]) << 0)); + + return (void *)ps; +} + +static void +nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) +{ + struct nv10_context *nv10 = nv10_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv10->tex_sampler[unit] = sampler[unit]; + nv10->dirty_samplers |= (1 << unit); + } +} + +static void +nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + +static void +nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr, + struct pipe_texture **miptree) +{ + struct nv10_context *nv10 = nv10_context(pipe); + unsigned unit; + + for (unit = 0; unit < nr; unit++) { + nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit]; + nv10->dirty_samplers |= (1 << unit); + } +} + +static void * +nv10_rasterizer_state_create(struct pipe_context *pipe, + const struct pipe_rasterizer_state *cso) +{ + struct nv10_rasterizer_state *rs; + int i; + + /*XXX: ignored: + * light_twoside + * offset_cw/ccw -nohw + * scissor + * point_smooth -nohw + * multisample + * offset_units / offset_scale + */ + rs = MALLOC(sizeof(struct nv10_rasterizer_state)); + + rs->templ = cso; + + rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01; + + rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff; + rs->line_smooth_en = cso->line_smooth ? 1 : 0; + + rs->point_size = *(uint32_t*)&cso->point_size; + + rs->poly_smooth_en = cso->poly_smooth ? 1 : 0; + + if (cso->front_winding == PIPE_WINDING_CCW) { + rs->front_face = NV10TCL_FRONT_FACE_CCW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_cw); + } else { + rs->front_face = NV10TCL_FRONT_FACE_CW; + rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw); + rs->poly_mode_back = nvgl_polygon_mode(cso->fill_ccw); + } + + switch (cso->cull_mode) { + case PIPE_WINDING_CCW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CCW) + rs->cull_face = NV10TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV10TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_CW: + rs->cull_face_en = 1; + if (cso->front_winding == PIPE_WINDING_CW) + rs->cull_face = NV10TCL_CULL_FACE_FRONT; + else + rs->cull_face = NV10TCL_CULL_FACE_BACK; + break; + case PIPE_WINDING_BOTH: + rs->cull_face_en = 1; + rs->cull_face = NV10TCL_CULL_FACE_FRONT_AND_BACK; + break; + case PIPE_WINDING_NONE: + default: + rs->cull_face_en = 0; + rs->cull_face = 0; + break; + } + + if (cso->point_sprite) { + rs->point_sprite = (1 << 0); + for (i = 0; i < 8; i++) { + if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE) + rs->point_sprite |= (1 << (8 + i)); + } + } else { + rs->point_sprite = 0; + } + + return (void *)rs; +} + +static void +nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->rast = (struct nv10_rasterizer_state*)rast; + + draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL)); + + nv10->dirty |= NV10_NEW_RAST; +} + +static void +nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + +static void * +nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, + const struct pipe_depth_stencil_alpha_state *cso) +{ + struct nv10_depth_stencil_alpha_state *hw; + + hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state)); + + hw->depth.func = nvgl_comparison_op(cso->depth.func); + hw->depth.write_enable = cso->depth.writemask ? 1 : 0; + hw->depth.test_enable = cso->depth.enabled ? 1 : 0; + + hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0; + hw->stencil.wmask = cso->stencil[0].write_mask; + hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func); + hw->stencil.ref = cso->stencil[0].ref_value; + hw->stencil.vmask = cso->stencil[0].value_mask; + hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op); + hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op); + hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op); + + hw->alpha.enabled = cso->alpha.enabled ? 1 : 0; + hw->alpha.func = nvgl_comparison_op(cso->alpha.func); + hw->alpha.ref = float_to_ubyte(cso->alpha.ref); + + return (void *)hw; +} + +static void +nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa; + + nv10->dirty |= NV10_NEW_DSA; +} + +static void +nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +{ + FREE(hwcso); +} + +static void * +nv10_vp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *templ) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + return draw_create_vertex_shader(nv10->draw, templ); +} + +static void +nv10_vp_state_bind(struct pipe_context *pipe, void *shader) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); + + nv10->dirty |= NV10_NEW_VERTPROG; +} + +static void +nv10_vp_state_delete(struct pipe_context *pipe, void *shader) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); +} + +static void * +nv10_fp_state_create(struct pipe_context *pipe, + const struct pipe_shader_state *cso) +{ + struct nv10_fragment_program *fp; + + fp = CALLOC(1, sizeof(struct nv10_fragment_program)); + fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); + + tgsi_scan_shader(cso->tokens, &fp->info); + + return (void *)fp; +} + +static void +nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_fragment_program *fp = hwcso; + + nv10->fragprog.current = fp; + nv10->dirty |= NV10_NEW_FRAGPROG; +} + +static void +nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nv10_fragment_program *fp = hwcso; + + nv10_fragprog_destroy(nv10, fp); + FREE((void*)fp->pipe.tokens); + FREE(fp); +} + +static void +nv10_set_blend_color(struct pipe_context *pipe, + const struct pipe_blend_color *bcol) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->blend_color = (struct pipe_blend_color*)bcol; + + nv10->dirty |= NV10_NEW_BLENDCOL; +} + +static void +nv10_set_clip_state(struct pipe_context *pipe, + const struct pipe_clip_state *clip) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + draw_set_clip_state(nv10->draw, clip); +} + +static void +nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, + const struct pipe_constant_buffer *buf ) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct pipe_winsys *ws = pipe->winsys; + + assert(shader < PIPE_SHADER_TYPES); + assert(index == 0); + + if (buf) { + void *mapped; + if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ))) + { + memcpy(nv10->constbuf[shader], mapped, buf->size); + nv10->constbuf_nr[shader] = + buf->size / (4 * sizeof(float)); + ws->buffer_unmap(ws, buf->buffer); + } + } +} + +static void +nv10_set_framebuffer_state(struct pipe_context *pipe, + const struct pipe_framebuffer_state *fb) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->framebuffer = (struct pipe_framebuffer_state*)fb; + + nv10->dirty |= NV10_NEW_FRAMEBUFFER; +} + +static void +nv10_set_polygon_stipple(struct pipe_context *pipe, + const struct pipe_poly_stipple *stipple) +{ + NOUVEAU_ERR("line stipple hahaha\n"); +} + +static void +nv10_set_scissor_state(struct pipe_context *pipe, + const struct pipe_scissor_state *s) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->scissor = (struct pipe_scissor_state*)s; + + nv10->dirty |= NV10_NEW_SCISSOR; +} + +static void +nv10_set_viewport_state(struct pipe_context *pipe, + const struct pipe_viewport_state *vpt) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + nv10->viewport = (struct pipe_viewport_state*)vpt; + + draw_set_viewport_state(nv10->draw, nv10->viewport); + + nv10->dirty |= NV10_NEW_VIEWPORT; +} + +static void +nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_buffer *vb) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); + nv10->dirty |= NV10_NEW_VTXARRAYS; + + draw_set_vertex_buffers(nv10->draw, count, vb); +} + +static void +nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, + const struct pipe_vertex_element *ve) +{ + struct nv10_context *nv10 = nv10_context(pipe); + + memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); + nv10->dirty |= NV10_NEW_VTXARRAYS; + + draw_set_vertex_elements(nv10->draw, count, ve); +} + +void +nv10_init_state_functions(struct nv10_context *nv10) +{ + nv10->pipe.create_blend_state = nv10_blend_state_create; + nv10->pipe.bind_blend_state = nv10_blend_state_bind; + nv10->pipe.delete_blend_state = nv10_blend_state_delete; + + nv10->pipe.create_sampler_state = nv10_sampler_state_create; + nv10->pipe.bind_sampler_states = nv10_sampler_state_bind; + nv10->pipe.delete_sampler_state = nv10_sampler_state_delete; + nv10->pipe.set_sampler_textures = nv10_set_sampler_texture; + + nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create; + nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind; + nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete; + + nv10->pipe.create_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_create; + nv10->pipe.bind_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_bind; + nv10->pipe.delete_depth_stencil_alpha_state = + nv10_depth_stencil_alpha_state_delete; + + nv10->pipe.create_vs_state = nv10_vp_state_create; + nv10->pipe.bind_vs_state = nv10_vp_state_bind; + nv10->pipe.delete_vs_state = nv10_vp_state_delete; + + nv10->pipe.create_fs_state = nv10_fp_state_create; + nv10->pipe.bind_fs_state = nv10_fp_state_bind; + nv10->pipe.delete_fs_state = nv10_fp_state_delete; + + nv10->pipe.set_blend_color = nv10_set_blend_color; + nv10->pipe.set_clip_state = nv10_set_clip_state; + nv10->pipe.set_constant_buffer = nv10_set_constant_buffer; + nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state; + nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple; + nv10->pipe.set_scissor_state = nv10_set_scissor_state; + nv10->pipe.set_viewport_state = nv10_set_viewport_state; + + nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers; + nv10->pipe.set_vertex_elements = nv10_set_vertex_elements; +} + diff --git a/src/gallium/drivers/nv20/nv20_state.h b/src/gallium/drivers/nv20/nv20_state.h new file mode 100644 index 00000000000..3a3fd0d4f4f --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_state.h @@ -0,0 +1,139 @@ +#ifndef __NV10_STATE_H__ +#define __NV10_STATE_H__ + +#include "pipe/p_state.h" +#include "tgsi/tgsi_scan.h" + +struct nv10_blend_state { + uint32_t b_enable; + uint32_t b_srcfunc; + uint32_t b_dstfunc; + + uint32_t c_mask; + + uint32_t d_enable; +}; + +struct nv10_sampler_state { + uint32_t wrap; + uint32_t en; + uint32_t filt; + uint32_t bcol; +}; + +struct nv10_rasterizer_state { + uint32_t shade_model; + + uint32_t line_width; + uint32_t line_smooth_en; + + uint32_t point_size; + + uint32_t poly_smooth_en; + + uint32_t poly_mode_front; + uint32_t poly_mode_back; + + uint32_t front_face; + uint32_t cull_face; + uint32_t cull_face_en; + + uint32_t point_sprite; + + const struct pipe_rasterizer_state *templ; +}; + +struct nv10_vertex_program_exec { + uint32_t data[4]; + boolean has_branch_offset; + int const_index; +}; + +struct nv10_vertex_program_data { + int index; /* immediates == -1 */ + float value[4]; +}; + +struct nv10_vertex_program { + const struct pipe_shader_state *pipe; + + boolean translated; + struct nv10_vertex_program_exec *insns; + unsigned nr_insns; + struct nv10_vertex_program_data *consts; + unsigned nr_consts; + + struct nouveau_resource *exec; + unsigned exec_start; + struct nouveau_resource *data; + unsigned data_start; + unsigned data_start_min; + + uint32_t ir; + uint32_t or; +}; + +struct nv10_fragment_program_data { + unsigned offset; + unsigned index; +}; + +struct nv10_fragment_program { + struct pipe_shader_state pipe; + struct tgsi_shader_info info; + + boolean translated; + boolean on_hw; + unsigned samplers; + + uint32_t *insn; + int insn_len; + + struct nv10_fragment_program_data *consts; + unsigned nr_consts; + + struct pipe_buffer *buffer; + + uint32_t fp_control; + uint32_t fp_reg_control; +}; + + +struct nv10_depth_stencil_alpha_state { + struct { + uint32_t func; + uint32_t write_enable; + uint32_t test_enable; + } depth; + + struct { + uint32_t enable; + uint32_t wmask; + uint32_t func; + uint32_t ref; + uint32_t vmask; + uint32_t fail; + uint32_t zfail; + uint32_t zpass; + } stencil; + + struct { + uint32_t enabled; + uint32_t func; + uint32_t ref; + } alpha; +}; + +struct nv10_miptree { + struct pipe_texture base; + + struct pipe_buffer *buffer; + uint total_size; + + struct { + uint pitch; + uint *image_offset; + } level[PIPE_MAX_TEXTURE_LEVELS]; +}; + +#endif diff --git a/src/gallium/drivers/nv20/nv20_state_emit.c b/src/gallium/drivers/nv20/nv20_state_emit.c new file mode 100644 index 00000000000..46c7e1d7536 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_state_emit.c @@ -0,0 +1,303 @@ +#include "nv10_context.h" +#include "nv10_state.h" + +static void nv10_state_emit_blend(struct nv10_context* nv10) +{ + struct nv10_blend_state *b = nv10->blend; + + BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); + OUT_RING (b->d_enable); + + BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); + OUT_RING (b->b_enable); + OUT_RING (b->b_srcfunc); + OUT_RING (b->b_dstfunc); + + BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + OUT_RING (b->c_mask); +} + +static void nv10_state_emit_blend_color(struct nv10_context* nv10) +{ + struct pipe_blend_color *c = nv10->blend_color; + + BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); + OUT_RING ((float_to_ubyte(c->color[3]) << 24)| + (float_to_ubyte(c->color[0]) << 16)| + (float_to_ubyte(c->color[1]) << 8) | + (float_to_ubyte(c->color[2]) << 0)); +} + +static void nv10_state_emit_rast(struct nv10_context* nv10) +{ + struct nv10_rasterizer_state *r = nv10->rast; + + BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); + OUT_RING (r->shade_model); + OUT_RING (r->line_width); + + + BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + OUT_RING (r->point_size); + + BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + OUT_RING (r->poly_mode_front); + OUT_RING (r->poly_mode_back); + + + BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + OUT_RING (r->cull_face); + OUT_RING (r->front_face); + + BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); + OUT_RING (r->line_smooth_en); + OUT_RING (r->poly_smooth_en); + + BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + OUT_RING (r->cull_face_en); +} + +static void nv10_state_emit_dsa(struct nv10_context* nv10) +{ + struct nv10_depth_stencil_alpha_state *d = nv10->dsa; + + BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + OUT_RING (d->depth.func); + + BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + OUT_RING (d->depth.write_enable); + + BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + OUT_RING (d->depth.test_enable); + +#if 0 + BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); + OUT_RING (d->stencil.enable); + BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); + OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7); +#endif + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + OUT_RING (d->alpha.enabled); + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 1); + OUT_RING (d->alpha.func); + + BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_REF, 1); + OUT_RING (d->alpha.ref); +} + +static void nv10_state_emit_viewport(struct nv10_context* nv10) +{ +} + +static void nv10_state_emit_scissor(struct nv10_context* nv10) +{ + // XXX this is so not working +/* struct pipe_scissor_state *s = nv10->scissor; + BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); + OUT_RING (((s->maxx - s->minx) << 16) | s->minx); + OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ +} + +static void nv10_state_emit_framebuffer(struct nv10_context* nv10) +{ + struct pipe_framebuffer_state* fb = nv10->framebuffer; + struct pipe_surface *rt, *zeta = NULL; + uint32_t rt_format, w, h; + int colour_format = 0, zeta_format = 0; + + w = fb->cbufs[0]->width; + h = fb->cbufs[0]->height; + colour_format = fb->cbufs[0]->format; + rt = fb->cbufs[0]; + + if (fb->zsbuf) { + if (colour_format) { + assert(w == fb->zsbuf->width); + assert(h == fb->zsbuf->height); + } else { + w = fb->zsbuf->width; + h = fb->zsbuf->height; + } + + zeta_format = fb->zsbuf->format; + zeta = fb->zsbuf; + } + + rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR; + + switch (colour_format) { + case PIPE_FORMAT_A8R8G8B8_UNORM: + case 0: + rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8; + break; + case PIPE_FORMAT_R5G6B5_UNORM: + rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5; + break; + default: + assert(0); + } + + if (zeta) { + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING (rt->stride | (zeta->stride << 16)); + } else { + BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + OUT_RING (rt->stride | (rt->stride << 16)); + } + + nv10->rt[0] = rt->buffer; + + if (zeta_format) + { + nv10->zeta = zeta->buffer; + } + + BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); + OUT_RING ((w << 16) | 0); + OUT_RING ((h << 16) | 0); + OUT_RING (rt_format); + BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); + OUT_RING (((w - 1) << 16) | 0 | 0x08000800); + OUT_RING (((h - 1) << 16) | 0 | 0x08000800); +} + +static void nv10_vertex_layout(struct nv10_context *nv10) +{ + struct nv10_fragment_program *fp = nv10->fragprog.current; + uint32_t src = 0; + int i; + struct vertex_info vinfo; + + memset(&vinfo, 0, sizeof(vinfo)); + + for (i = 0; i < fp->info.num_inputs; i++) { + switch (fp->info.input_semantic_name[i]) { + case TGSI_SEMANTIC_POSITION: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + case TGSI_SEMANTIC_COLOR: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + break; + default: + case TGSI_SEMANTIC_GENERIC: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + case TGSI_SEMANTIC_FOG: + draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + break; + } + } + draw_compute_vertex_size(&vinfo); +} + +void +nv10_emit_hw_state(struct nv10_context *nv10) +{ + int i; + + if (nv10->dirty & NV10_NEW_VERTPROG) { + //nv10_vertprog_bind(nv10, nv10->vertprog.current); + nv10->dirty &= ~NV10_NEW_VERTPROG; + } + + if (nv10->dirty & NV10_NEW_FRAGPROG) { + nv10_fragprog_bind(nv10, nv10->fragprog.current); + /*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */ + nv10->dirty_samplers |= (1<<10); + nv10->dirty_samplers = 0; + } + + if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) { + nv10_fragtex_bind(nv10); + nv10->dirty &= ~NV10_NEW_FRAGPROG; + } + + if (nv10->dirty & NV10_NEW_VTXARRAYS) { + nv10->dirty &= ~NV10_NEW_VTXARRAYS; + nv10_vertex_layout(nv10); + nv10_vtxbuf_bind(nv10); + } + + if (nv10->dirty & NV10_NEW_BLEND) { + nv10->dirty &= ~NV10_NEW_BLEND; + nv10_state_emit_blend(nv10); + } + + if (nv10->dirty & NV10_NEW_BLENDCOL) { + nv10->dirty &= ~NV10_NEW_BLENDCOL; + nv10_state_emit_blend_color(nv10); + } + + if (nv10->dirty & NV10_NEW_RAST) { + nv10->dirty &= ~NV10_NEW_RAST; + nv10_state_emit_rast(nv10); + } + + if (nv10->dirty & NV10_NEW_DSA) { + nv10->dirty &= ~NV10_NEW_DSA; + nv10_state_emit_dsa(nv10); + } + + if (nv10->dirty & NV10_NEW_VIEWPORT) { + nv10->dirty &= ~NV10_NEW_VIEWPORT; + nv10_state_emit_viewport(nv10); + } + + if (nv10->dirty & NV10_NEW_SCISSOR) { + nv10->dirty &= ~NV10_NEW_SCISSOR; + nv10_state_emit_scissor(nv10); + } + + if (nv10->dirty & NV10_NEW_FRAMEBUFFER) { + nv10->dirty &= ~NV10_NEW_FRAMEBUFFER; + nv10_state_emit_framebuffer(nv10); + } + + /* Emit relocs for every referenced buffer. + * This is to ensure the bufmgr has an accurate idea of how + * the buffer is used. This isn't very efficient, but we don't + * seem to take a significant performance hit. Will be improved + * at some point. Vertex arrays are emitted by nv10_vbo.c + */ + + /* Render target */ +// XXX figre out who's who for NV10TCL_DMA_* and fill accordingly +// BEGIN_RING(celsius, NV10TCL_DMA_COLOR0, 1); +// OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + if (nv10->zeta) { +// XXX +// BEGIN_RING(celsius, NV10TCL_DMA_ZETA, 1); +// OUT_RELOCo(nv10->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + /* XXX for when we allocate LMA on nv17 */ +/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1); + OUT_RELOCl(nv10->zeta + lma_offset);*/ + } + + /* Vertex buffer */ + BEGIN_RING(celsius, NV10TCL_DMA_VTXBUF0, 1); + OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + + /* Texture images */ + for (i = 0; i < 2; i++) { + if (!(nv10->fp_samplers & (1 << i))) + continue; + BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 1); + OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + NOUVEAU_BO_GART | NOUVEAU_BO_RD); + BEGIN_RING(celsius, NV10TCL_TX_FORMAT(i), 1); + OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, + NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | + NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0, + NV10TCL_TX_FORMAT_DMA1); + } +} + diff --git a/src/gallium/drivers/nv20/nv20_surface.c b/src/gallium/drivers/nv20/nv20_surface.c new file mode 100644 index 00000000000..875e4c58589 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_surface.c @@ -0,0 +1,64 @@ + +/************************************************************************** + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "nv10_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_winsys.h" +#include "pipe/p_inlines.h" +#include "util/u_tile.h" + +static void +nv10_surface_copy(struct pipe_context *pipe, unsigned do_flip, + struct pipe_surface *dest, unsigned destx, unsigned desty, + struct pipe_surface *src, unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, + width, height); +} + +static void +nv10_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, + unsigned destx, unsigned desty, unsigned width, + unsigned height, unsigned value) +{ + struct nv10_context *nv10 = nv10_context(pipe); + struct nouveau_winsys *nvws = nv10->nvws; + + nvws->surface_fill(nvws, dest, destx, desty, width, height, value); +} + +void +nv10_init_surface_functions(struct nv10_context *nv10) +{ + nv10->pipe.surface_copy = nv10_surface_copy; + nv10->pipe.surface_fill = nv10_surface_fill; +} diff --git a/src/gallium/drivers/nv20/nv20_vbo.c b/src/gallium/drivers/nv20/nv20_vbo.c new file mode 100644 index 00000000000..d0e788ac036 --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_vbo.c @@ -0,0 +1,77 @@ +#include "draw/draw_context.h" +#include "pipe/p_context.h" +#include "pipe/p_state.h" + +#include "nv10_context.h" +#include "nv10_state.h" + +#include "nouveau/nouveau_channel.h" +#include "nouveau/nouveau_pushbuf.h" + +boolean nv10_draw_elements( struct pipe_context *pipe, + struct pipe_buffer *indexBuffer, + unsigned indexSize, + unsigned prim, unsigned start, unsigned count) +{ + struct nv10_context *nv10 = nv10_context( pipe ); + struct draw_context *draw = nv10->draw; + unsigned i; + + nv10_emit_hw_state(nv10); + + /* + * Map vertex buffers + */ + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (nv10->vtxbuf[i].buffer) { + void *buf + = pipe->winsys->buffer_map(pipe->winsys, + nv10->vtxbuf[i].buffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_vertex_buffer(draw, i, buf); + } + } + /* Map index buffer, if present */ + if (indexBuffer) { + void *mapped_indexes + = pipe->winsys->buffer_map(pipe->winsys, indexBuffer, + PIPE_BUFFER_USAGE_CPU_READ); + draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes); + } + else { + /* no index/element buffer */ + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + draw_set_mapped_constant_buffer(draw, + nv10->constbuf[PIPE_SHADER_VERTEX], + nv10->constbuf_nr[PIPE_SHADER_VERTEX]); + + /* draw! */ + draw_arrays(nv10->draw, prim, start, count); + + /* + * unmap vertex/index buffers + */ + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (nv10->vtxbuf[i].buffer) { + pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer); + draw_set_mapped_vertex_buffer(draw, i, NULL); + } + } + if (indexBuffer) { + pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer); + draw_set_mapped_element_buffer(draw, 0, NULL); + } + + return TRUE; +} + +boolean nv10_draw_arrays( struct pipe_context *pipe, + unsigned prim, unsigned start, unsigned count) +{ + return nv10_draw_elements(pipe, NULL, 0, prim, start, count); +} + + + diff --git a/src/gallium/drivers/nv20/nv20_vertprog.c b/src/gallium/drivers/nv20/nv20_vertprog.c new file mode 100644 index 00000000000..72824559e8b --- /dev/null +++ b/src/gallium/drivers/nv20/nv20_vertprog.c @@ -0,0 +1,838 @@ +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "pipe/p_state.h" + +#include "pipe/p_shader_tokens.h" +#include "tgsi/tgsi_parse.h" +#include "tgsi/tgsi_dump.h" + +#include "nv30_context.h" +#include "nv30_state.h" + +/* TODO (at least...): + * 1. Indexed consts + ARL + * 2. Arb. swz/negation + * 3. NV_vp11, NV_vp2, NV_vp3 features + * - extra arith opcodes + * - branching + * - texture sampling + * - indexed attribs + * - indexed results + * 4. bugs + */ + +#define SWZ_X 0 +#define SWZ_Y 1 +#define SWZ_Z 2 +#define SWZ_W 3 +#define MASK_X 8 +#define MASK_Y 4 +#define MASK_Z 2 +#define MASK_W 1 +#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) +#define DEF_SCALE 0 +#define DEF_CTEST 0 +#include "nv30_shader.h" + +#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv30_sr_neg((s)) +#define abs(s) nv30_sr_abs((s)) + +struct nv30_vpc { + struct nv30_vertex_program *vp; + + struct nv30_vertex_program_exec *vpi; + + unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; + + int high_temp; + int temp_temp_count; + + struct nv30_sreg *imm; + unsigned nr_imm; +}; + +static struct nv30_sreg +temp(struct nv30_vpc *vpc) +{ + int idx; + + idx = vpc->temp_temp_count++; + idx += vpc->high_temp + 1; + return nv30_sr(NV30SR_TEMP, idx); +} + +static struct nv30_sreg +constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) +{ + struct nv30_vertex_program *vp = vpc->vp; + struct nv30_vertex_program_data *vpd; + int idx; + + if (pipe >= 0) { + for (idx = 0; idx < vp->nr_consts; idx++) { + if (vp->consts[idx].index == pipe) + return nv30_sr(NV30SR_CONST, idx); + } + } + + idx = vp->nr_consts++; + vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); + vpd = &vp->consts[idx]; + + vpd->index = pipe; + vpd->value[0] = x; + vpd->value[1] = y; + vpd->value[2] = z; + vpd->value[3] = w; + return nv30_sr(NV30SR_CONST, idx); +} + +#define arith(cc,s,o,d,m,s0,s1,s2) \ + nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) + +static void +emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t sr = 0; + + switch (src.type) { + case NV30SR_TEMP: + sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT); + sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT); + break; + case NV30SR_INPUT: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + vp->ir |= (1 << src.index); + hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT); + break; + case NV30SR_CONST: + sr |= (NV30_VP_SRC_REG_TYPE_CONST << + NV30_VP_SRC_REG_TYPE_SHIFT); + assert(vpc->vpi->const_index == -1 || + vpc->vpi->const_index == src.index); + vpc->vpi->const_index = src.index; + break; + case NV30SR_NONE: + sr |= (NV30_VP_SRC_REG_TYPE_INPUT << + NV30_VP_SRC_REG_TYPE_SHIFT); + break; + default: + assert(0); + } + + if (src.negate) + sr |= NV30_VP_SRC_NEGATE; + + if (src.abs) + hw[0] |= (1 << (21 + pos)); + + sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) | + (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) | + (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) | + (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT)); + +/* + * |VVV| + * d�.�b + * \u/ + * + */ + + switch (pos) { + case 0: + hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >> + NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT; + hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) << + NV30_VP_INST_SRC0L_SHIFT; + break; + case 1: + hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT; + break; + case 2: + hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >> + NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT; + hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) << + NV30_VP_INST_SRC2L_SHIFT; + break; + default: + assert(0); + } +} + +static void +emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) +{ + struct nv30_vertex_program *vp = vpc->vp; + + switch (dst.type) { + case NV30SR_TEMP: + hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT); + break; + case NV30SR_OUTPUT: + switch (dst.index) { + case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; + case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; + case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; + case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; + case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; + case NV30_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; + case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; + case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; + case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; + case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; + case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; + case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; + case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; + case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; + default: + break; + } + + hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); + hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); + + /*XXX: no way this is entirely correct, someone needs to + * figure out what exactly it is. + */ + hw[3] |= 0x800; + break; + default: + assert(0); + } +} + +static void +nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, + struct nv30_sreg dst, int mask, + struct nv30_sreg s0, struct nv30_sreg s1, + struct nv30_sreg s2) +{ + struct nv30_vertex_program *vp = vpc->vp; + uint32_t *hw; + + vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); + vpc->vpi = &vp->insns[vp->nr_insns - 1]; + memset(vpc->vpi, 0, sizeof(*vpc->vpi)); + vpc->vpi->const_index = -1; + + hw = vpc->vpi->data; + + hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT); + hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) | + (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) | + (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) | + (3 << NV30_VP_INST_COND_SWZ_W_SHIFT)); + + hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT); +// hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; +// hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); + + if (dst.type == NV30SR_OUTPUT) { + if (slot) + hw[3] |= (mask << NV30_VP_INST_SDEST_WRITEMASK_SHIFT); + else + hw[3] |= (mask << NV30_VP_INST_VDEST_WRITEMASK_SHIFT); + } else { + if (slot) + hw[3] |= (mask << NV30_VP_INST_STEMP_WRITEMASK_SHIFT); + else + hw[3] |= (mask << NV30_VP_INST_VTEMP_WRITEMASK_SHIFT); + } + + emit_dst(vpc, hw, slot, dst); + emit_src(vpc, hw, 0, s0); + emit_src(vpc, hw, 1, s1); + emit_src(vpc, hw, 2, s2); +} + +static INLINE struct nv30_sreg +tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv30_sreg src; + + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); + break; + case TGSI_FILE_CONSTANT: + src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); + break; + case TGSI_FILE_IMMEDIATE: + src = vpc->imm[fsrc->SrcRegister.Index]; + break; + case TGSI_FILE_TEMPORARY: + if (vpc->high_temp < fsrc->SrcRegister.Index) + vpc->high_temp = fsrc->SrcRegister.Index; + src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); + break; + default: + NOUVEAU_ERR("bad src file\n"); + break; + } + + src.abs = fsrc->SrcRegisterExtMod.Absolute; + src.negate = fsrc->SrcRegister.Negate; + src.swz[0] = fsrc->SrcRegister.SwizzleX; + src.swz[1] = fsrc->SrcRegister.SwizzleY; + src.swz[2] = fsrc->SrcRegister.SwizzleZ; + src.swz[3] = fsrc->SrcRegister.SwizzleW; + return src; +} + +static INLINE struct nv30_sreg +tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + struct nv30_sreg dst; + + switch (fdst->DstRegister.File) { + case TGSI_FILE_OUTPUT: + dst = nv30_sr(NV30SR_OUTPUT, + vpc->output_map[fdst->DstRegister.Index]); + + break; + case TGSI_FILE_TEMPORARY: + dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); + if (vpc->high_temp < dst.index) + vpc->high_temp = dst.index; + break; + default: + NOUVEAU_ERR("bad dst file\n"); + break; + } + + return dst; +} + +static INLINE int +tgsi_mask(uint tgsi) +{ + int mask = 0; + + if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; + if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; + if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; + if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; + return mask; +} + +static boolean +nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, + const struct tgsi_full_instruction *finst) +{ + struct nv30_sreg src[3], dst, tmp; + struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + int mask; + int ai = -1, ci = -1; + int i; + + if (finst->Instruction.Opcode == TGSI_OPCODE_END) + return TRUE; + + vpc->temp_temp_count = 0; + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { + src[i] = tgsi_src(vpc, fsrc); + } + } + + for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { + const struct tgsi_full_src_register *fsrc; + + fsrc = &finst->FullSrcRegisters[i]; + switch (fsrc->SrcRegister.File) { + case TGSI_FILE_INPUT: + if (ai == -1 || ai == fsrc->SrcRegister.Index) { + ai = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + /*XXX: index comparison is broken now that consts come from + * two different register files. + */ + case TGSI_FILE_CONSTANT: + case TGSI_FILE_IMMEDIATE: + if (ci == -1 || ci == fsrc->SrcRegister.Index) { + ci = fsrc->SrcRegister.Index; + src[i] = tgsi_src(vpc, fsrc); + } else { + src[i] = temp(vpc); + arith(vpc, 0, OP_MOV, src[i], MASK_ALL, + tgsi_src(vpc, fsrc), none, none); + } + break; + case TGSI_FILE_TEMPORARY: + /* handled above */ + break; + default: + NOUVEAU_ERR("bad src file\n"); + return FALSE; + } + } + + dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); + mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); + + switch (finst->Instruction.Opcode) { + case TGSI_OPCODE_ABS: + arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); + break; + case TGSI_OPCODE_ADD: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); + break; + case TGSI_OPCODE_ARL: + arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_DP3: + arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DP4: + arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DPH: + arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_DST: + arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_EX2: + arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_EXP: + arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_FLR: + arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_FRC: + arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_LG2: + arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LIT: + arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_LOG: + arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_MAD: + arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); + break; + case TGSI_OPCODE_MAX: + arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MIN: + arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_MOV: + arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); + break; + case TGSI_OPCODE_MUL: + arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_POW: + tmp = temp(vpc); + arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, + swz(src[0], X, X, X, X)); + arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), + swz(src[1], X, X, X, X), none); + arith(vpc, 1, OP_EX2, dst, mask, none, none, + swz(tmp, X, X, X, X)); + break; + case TGSI_OPCODE_RCP: + arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_RET: + break; + case TGSI_OPCODE_RSQ: + arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); + break; + case TGSI_OPCODE_SGE: + arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SGT: + arith(vpc, 0, OP_SGT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SLT: + arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); + break; + case TGSI_OPCODE_SUB: + arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); + break; + case TGSI_OPCODE_XPD: + tmp = temp(vpc); + arith(vpc, 0, OP_MUL, tmp, mask, + swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); + arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), + swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), + neg(tmp)); + break; + default: + NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); + return FALSE; + } + + return TRUE; +} + +static boolean +nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, + const struct tgsi_full_declaration *fdec) +{ + int hw; + + switch (fdec->Semantic.SemanticName) { + case TGSI_SEMANTIC_POSITION: + hw = NV30_VP_INST_DEST_POS; + break; + case TGSI_SEMANTIC_COLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_COL0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_COL1; + } else { + NOUVEAU_ERR("bad colour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_BCOLOR: + if (fdec->Semantic.SemanticIndex == 0) { + hw = NV30_VP_INST_DEST_BFC0; + } else + if (fdec->Semantic.SemanticIndex == 1) { + hw = NV30_VP_INST_DEST_BFC1; + } else { + NOUVEAU_ERR("bad bcolour semantic index\n"); + return FALSE; + } + break; + case TGSI_SEMANTIC_FOG: + hw = NV30_VP_INST_DEST_FOGC; + break; + case TGSI_SEMANTIC_PSIZE: + hw = NV30_VP_INST_DEST_PSZ; + break; + case TGSI_SEMANTIC_GENERIC: + if (fdec->Semantic.SemanticIndex <= 7) { + hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); + } else { + NOUVEAU_ERR("bad generic semantic index\n"); + return FALSE; + } + break; + default: + NOUVEAU_ERR("bad output semantic\n"); + return FALSE; + } + + vpc->output_map[fdec->DeclarationRange.First] = hw; + return TRUE; +} + +static boolean +nv30_vertprog_prepare(struct nv30_vpc *vpc) +{ + struct tgsi_parse_context p; + int nr_imm = 0; + + tgsi_parse_init(&p, vpc->vp->pipe.tokens); + while (!tgsi_parse_end_of_tokens(&p)) { + const union tgsi_full_token *tok = &p.FullToken; + + tgsi_parse_token(&p); + switch(tok->Token.Type) { + case TGSI_TOKEN_TYPE_IMMEDIATE: + nr_imm++; + break; + default: + break; + } + } + tgsi_parse_free(&p); + + if (nr_imm) { + vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); + assert(vpc->imm); + } + + return TRUE; +} + +static void +nv30_vertprog_translate(struct nv30_context *nv30, + struct nv30_vertex_program *vp) +{ + struct tgsi_parse_context parse; + struct nv30_vpc *vpc = NULL; + + tgsi_dump(vp->pipe.tokens,0); + + vpc = CALLOC(1, sizeof(struct nv30_vpc)); + if (!vpc) + return; + vpc->vp = vp; + vpc->high_temp = -1; + + if (!nv30_vertprog_prepare(vpc)) { + FREE(vpc); + return; + } + + tgsi_parse_init(&parse, vp->pipe.tokens); + + while (!tgsi_parse_end_of_tokens(&parse)) { + tgsi_parse_token(&parse); + + switch (parse.FullToken.Token.Type) { + case TGSI_TOKEN_TYPE_DECLARATION: + { + const struct tgsi_full_declaration *fdec; + fdec = &parse.FullToken.FullDeclaration; + switch (fdec->Declaration.File) { + case TGSI_FILE_OUTPUT: + if (!nv30_vertprog_parse_decl_output(vpc, fdec)) + goto out_err; + break; + default: + break; + } + } + break; + case TGSI_TOKEN_TYPE_IMMEDIATE: + { + const struct tgsi_full_immediate *imm; + + imm = &parse.FullToken.FullImmediate; + assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); +// assert(imm->Immediate.Size == 4); + vpc->imm[vpc->nr_imm++] = + constant(vpc, -1, + imm->u.ImmediateFloat32[0].Float, + imm->u.ImmediateFloat32[1].Float, + imm->u.ImmediateFloat32[2].Float, + imm->u.ImmediateFloat32[3].Float); + } + break; + case TGSI_TOKEN_TYPE_INSTRUCTION: + { + const struct tgsi_full_instruction *finst; + finst = &parse.FullToken.FullInstruction; + if (!nv30_vertprog_parse_instruction(vpc, finst)) + goto out_err; + } + break; + default: + break; + } + } + + vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST; + vp->translated = TRUE; +out_err: + tgsi_parse_free(&parse); + FREE(vpc); +} + +static boolean +nv30_vertprog_validate(struct nv30_context *nv30) +{ + struct nouveau_winsys *nvws = nv30->nvws; + struct pipe_winsys *ws = nv30->pipe.winsys; + struct nouveau_grobj *rankine = nv30->screen->rankine; + struct nv30_vertex_program *vp; + struct pipe_buffer *constbuf; + boolean upload_code = FALSE, upload_data = FALSE; + int i; + + vp = nv30->vertprog; + constbuf = nv30->constbuf[PIPE_SHADER_VERTEX]; + + /* Translate TGSI shader into hw bytecode */ + if (!vp->translated) { + nv30_vertprog_translate(nv30, vp); + if (!vp->translated) + return FALSE; + } + + /* Allocate hw vtxprog exec slots */ + if (!vp->exec) { + struct nouveau_resource *heap = nv30->screen->vp_exec_heap; + struct nouveau_stateobj *so; + uint vplen = vp->nr_insns; + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { + while (heap->next && heap->size < vplen) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->exec); + } + + if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) + assert(0); + } + + so = so_new(2, 0); + so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); + so_data (so, vp->exec->start); + so_ref(so, &vp->so); + + upload_code = TRUE; + } + + /* Allocate hw vtxprog const slots */ + if (vp->nr_consts && !vp->data) { + struct nouveau_resource *heap = nv30->screen->vp_data_heap; + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { + while (heap->next && heap->size < vp->nr_consts) { + struct nv30_vertex_program *evict; + + evict = heap->next->priv; + nvws->res_free(&evict->data); + } + + if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) + assert(0); + } + + /*XXX: handle this some day */ + assert(vp->data->start >= vp->data_start_min); + + upload_data = TRUE; + if (vp->data_start != vp->data->start) + upload_code = TRUE; + } + + /* If exec or data segments moved we need to patch the program to + * fixup offsets and register IDs. + */ + if (vp->exec_start != vp->exec->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->has_branch_offset) { + assert(0); + } + } + + vp->exec_start = vp->exec->start; + } + + if (vp->nr_consts && vp->data_start != vp->data->start) { + for (i = 0; i < vp->nr_insns; i++) { + struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + + if (vpi->const_index >= 0) { + vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; + vpi->data[1] |= + (vpi->const_index + vp->data->start) << + NV30_VP_INST_CONST_SRC_SHIFT; + + } + } + + vp->data_start = vp->data->start; + } + + /* Update + Upload constant values */ + if (vp->nr_consts) { + float *map = NULL; + + if (constbuf) { + map = ws->buffer_map(ws, constbuf, + PIPE_BUFFER_USAGE_CPU_READ); + } + + for (i = 0; i < vp->nr_consts; i++) { + struct nv30_vertex_program_data *vpd = &vp->consts[i]; + + if (vpd->index >= 0) { + if (!upload_data && + !memcmp(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float))) + continue; + memcpy(vpd->value, &map[vpd->index * 4], + 4 * sizeof(float)); + } + + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); + OUT_RING (i + vp->data->start); + OUT_RINGp ((uint32_t *)vpd->value, 4); + } + + if (constbuf) { + ws->buffer_unmap(ws, constbuf); + } + } + + /* Upload vtxprog */ + if (upload_code) { +#if 0 + for (i = 0; i < vp->nr_insns; i++) { + NOUVEAU_MSG("VP inst %d: 0x%08x 0x%08x 0x%08x 0x%08x\n", + i, vp->insns[i].data[0], vp->insns[i].data[1], + vp->insns[i].data[2], vp->insns[i].data[3]); + } +#endif + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); + OUT_RING (vp->exec->start); + for (i = 0; i < vp->nr_insns; i++) { + BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); + OUT_RINGp (vp->insns[i].data, 4); + } + } + + if (vp->so != nv30->state.hw[NV30_STATE_VERTPROG]) { + so_ref(vp->so, &nv30->state.hw[NV30_STATE_VERTPROG]); + return TRUE; + } + + return FALSE; +} + +void +nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) +{ + struct nouveau_winsys *nvws = nv30->screen->nvws; + + vp->translated = FALSE; + + if (vp->nr_insns) { + FREE(vp->insns); + vp->insns = NULL; + vp->nr_insns = 0; + } + + if (vp->nr_consts) { + FREE(vp->consts); + vp->consts = NULL; + vp->nr_consts = 0; + } + + nvws->res_free(&vp->exec); + vp->exec_start = 0; + nvws->res_free(&vp->data); + vp->data_start = 0; + vp->data_start_min = 0; + + vp->ir = vp->or = 0; + so_ref(NULL, &vp->so); +} + +struct nv30_state_entry nv30_state_vertprog = { + .validate = nv30_vertprog_validate, + .dirty = { + .pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/, + .hw = NV30_STATE_VERTPROG, + } +}; diff --git a/src/gallium/drivers/nv20/nv30_vertprog.c b/src/gallium/drivers/nv20/nv30_vertprog.c deleted file mode 100644 index 72824559e8b..00000000000 --- a/src/gallium/drivers/nv20/nv30_vertprog.c +++ /dev/null @@ -1,838 +0,0 @@ -#include "pipe/p_context.h" -#include "pipe/p_defines.h" -#include "pipe/p_state.h" - -#include "pipe/p_shader_tokens.h" -#include "tgsi/tgsi_parse.h" -#include "tgsi/tgsi_dump.h" - -#include "nv30_context.h" -#include "nv30_state.h" - -/* TODO (at least...): - * 1. Indexed consts + ARL - * 2. Arb. swz/negation - * 3. NV_vp11, NV_vp2, NV_vp3 features - * - extra arith opcodes - * - branching - * - texture sampling - * - indexed attribs - * - indexed results - * 4. bugs - */ - -#define SWZ_X 0 -#define SWZ_Y 1 -#define SWZ_Z 2 -#define SWZ_W 3 -#define MASK_X 8 -#define MASK_Y 4 -#define MASK_Z 2 -#define MASK_W 1 -#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) -#define DEF_SCALE 0 -#define DEF_CTEST 0 -#include "nv30_shader.h" - -#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) -#define neg(s) nv30_sr_neg((s)) -#define abs(s) nv30_sr_abs((s)) - -struct nv30_vpc { - struct nv30_vertex_program *vp; - - struct nv30_vertex_program_exec *vpi; - - unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; - - int high_temp; - int temp_temp_count; - - struct nv30_sreg *imm; - unsigned nr_imm; -}; - -static struct nv30_sreg -temp(struct nv30_vpc *vpc) -{ - int idx; - - idx = vpc->temp_temp_count++; - idx += vpc->high_temp + 1; - return nv30_sr(NV30SR_TEMP, idx); -} - -static struct nv30_sreg -constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) -{ - struct nv30_vertex_program *vp = vpc->vp; - struct nv30_vertex_program_data *vpd; - int idx; - - if (pipe >= 0) { - for (idx = 0; idx < vp->nr_consts; idx++) { - if (vp->consts[idx].index == pipe) - return nv30_sr(NV30SR_CONST, idx); - } - } - - idx = vp->nr_consts++; - vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts); - vpd = &vp->consts[idx]; - - vpd->index = pipe; - vpd->value[0] = x; - vpd->value[1] = y; - vpd->value[2] = z; - vpd->value[3] = w; - return nv30_sr(NV30SR_CONST, idx); -} - -#define arith(cc,s,o,d,m,s0,s1,s2) \ - nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) - -static void -emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) -{ - struct nv30_vertex_program *vp = vpc->vp; - uint32_t sr = 0; - - switch (src.type) { - case NV30SR_TEMP: - sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT); - sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT); - break; - case NV30SR_INPUT: - sr |= (NV30_VP_SRC_REG_TYPE_INPUT << - NV30_VP_SRC_REG_TYPE_SHIFT); - vp->ir |= (1 << src.index); - hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT); - break; - case NV30SR_CONST: - sr |= (NV30_VP_SRC_REG_TYPE_CONST << - NV30_VP_SRC_REG_TYPE_SHIFT); - assert(vpc->vpi->const_index == -1 || - vpc->vpi->const_index == src.index); - vpc->vpi->const_index = src.index; - break; - case NV30SR_NONE: - sr |= (NV30_VP_SRC_REG_TYPE_INPUT << - NV30_VP_SRC_REG_TYPE_SHIFT); - break; - default: - assert(0); - } - - if (src.negate) - sr |= NV30_VP_SRC_NEGATE; - - if (src.abs) - hw[0] |= (1 << (21 + pos)); - - sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) | - (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) | - (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) | - (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT)); - -/* - * |VVV| - * d�.�b - * \u/ - * - */ - - switch (pos) { - case 0: - hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >> - NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT; - hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) << - NV30_VP_INST_SRC0L_SHIFT; - break; - case 1: - hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT; - break; - case 2: - hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >> - NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT; - hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) << - NV30_VP_INST_SRC2L_SHIFT; - break; - default: - assert(0); - } -} - -static void -emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) -{ - struct nv30_vertex_program *vp = vpc->vp; - - switch (dst.type) { - case NV30SR_TEMP: - hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT); - break; - case NV30SR_OUTPUT: - switch (dst.index) { - case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break; - case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break; - case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break; - case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break; - case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break; - case NV30_VP_INST_DEST_PSZ : vp->or |= (1 << 5); break; - case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break; - case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break; - case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break; - case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break; - case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break; - case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break; - case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break; - case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break; - default: - break; - } - - hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT); - hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20); - - /*XXX: no way this is entirely correct, someone needs to - * figure out what exactly it is. - */ - hw[3] |= 0x800; - break; - default: - assert(0); - } -} - -static void -nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, - struct nv30_sreg dst, int mask, - struct nv30_sreg s0, struct nv30_sreg s1, - struct nv30_sreg s2) -{ - struct nv30_vertex_program *vp = vpc->vp; - uint32_t *hw; - - vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); - vpc->vpi = &vp->insns[vp->nr_insns - 1]; - memset(vpc->vpi, 0, sizeof(*vpc->vpi)); - vpc->vpi->const_index = -1; - - hw = vpc->vpi->data; - - hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT); - hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) | - (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) | - (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) | - (3 << NV30_VP_INST_COND_SWZ_W_SHIFT)); - - hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT); -// hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK; -// hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT); - - if (dst.type == NV30SR_OUTPUT) { - if (slot) - hw[3] |= (mask << NV30_VP_INST_SDEST_WRITEMASK_SHIFT); - else - hw[3] |= (mask << NV30_VP_INST_VDEST_WRITEMASK_SHIFT); - } else { - if (slot) - hw[3] |= (mask << NV30_VP_INST_STEMP_WRITEMASK_SHIFT); - else - hw[3] |= (mask << NV30_VP_INST_VTEMP_WRITEMASK_SHIFT); - } - - emit_dst(vpc, hw, slot, dst); - emit_src(vpc, hw, 0, s0); - emit_src(vpc, hw, 1, s1); - emit_src(vpc, hw, 2, s2); -} - -static INLINE struct nv30_sreg -tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { - struct nv30_sreg src; - - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); - break; - case TGSI_FILE_CONSTANT: - src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); - break; - case TGSI_FILE_IMMEDIATE: - src = vpc->imm[fsrc->SrcRegister.Index]; - break; - case TGSI_FILE_TEMPORARY: - if (vpc->high_temp < fsrc->SrcRegister.Index) - vpc->high_temp = fsrc->SrcRegister.Index; - src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); - break; - default: - NOUVEAU_ERR("bad src file\n"); - break; - } - - src.abs = fsrc->SrcRegisterExtMod.Absolute; - src.negate = fsrc->SrcRegister.Negate; - src.swz[0] = fsrc->SrcRegister.SwizzleX; - src.swz[1] = fsrc->SrcRegister.SwizzleY; - src.swz[2] = fsrc->SrcRegister.SwizzleZ; - src.swz[3] = fsrc->SrcRegister.SwizzleW; - return src; -} - -static INLINE struct nv30_sreg -tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { - struct nv30_sreg dst; - - switch (fdst->DstRegister.File) { - case TGSI_FILE_OUTPUT: - dst = nv30_sr(NV30SR_OUTPUT, - vpc->output_map[fdst->DstRegister.Index]); - - break; - case TGSI_FILE_TEMPORARY: - dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); - if (vpc->high_temp < dst.index) - vpc->high_temp = dst.index; - break; - default: - NOUVEAU_ERR("bad dst file\n"); - break; - } - - return dst; -} - -static INLINE int -tgsi_mask(uint tgsi) -{ - int mask = 0; - - if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X; - if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y; - if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z; - if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W; - return mask; -} - -static boolean -nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, - const struct tgsi_full_instruction *finst) -{ - struct nv30_sreg src[3], dst, tmp; - struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); - int mask; - int ai = -1, ci = -1; - int i; - - if (finst->Instruction.Opcode == TGSI_OPCODE_END) - return TRUE; - - vpc->temp_temp_count = 0; - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) { - src[i] = tgsi_src(vpc, fsrc); - } - } - - for (i = 0; i < finst->Instruction.NumSrcRegs; i++) { - const struct tgsi_full_src_register *fsrc; - - fsrc = &finst->FullSrcRegisters[i]; - switch (fsrc->SrcRegister.File) { - case TGSI_FILE_INPUT: - if (ai == -1 || ai == fsrc->SrcRegister.Index) { - ai = fsrc->SrcRegister.Index; - src[i] = tgsi_src(vpc, fsrc); - } else { - src[i] = temp(vpc); - arith(vpc, 0, OP_MOV, src[i], MASK_ALL, - tgsi_src(vpc, fsrc), none, none); - } - break; - /*XXX: index comparison is broken now that consts come from - * two different register files. - */ - case TGSI_FILE_CONSTANT: - case TGSI_FILE_IMMEDIATE: - if (ci == -1 || ci == fsrc->SrcRegister.Index) { - ci = fsrc->SrcRegister.Index; - src[i] = tgsi_src(vpc, fsrc); - } else { - src[i] = temp(vpc); - arith(vpc, 0, OP_MOV, src[i], MASK_ALL, - tgsi_src(vpc, fsrc), none, none); - } - break; - case TGSI_FILE_TEMPORARY: - /* handled above */ - break; - default: - NOUVEAU_ERR("bad src file\n"); - return FALSE; - } - } - - dst = tgsi_dst(vpc, &finst->FullDstRegisters[0]); - mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask); - - switch (finst->Instruction.Opcode) { - case TGSI_OPCODE_ABS: - arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none); - break; - case TGSI_OPCODE_ADD: - arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]); - break; - case TGSI_OPCODE_ARL: - arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_DP3: - arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DP4: - arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DPH: - arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_DST: - arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_EX2: - arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_EXP: - arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_FLR: - arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_FRC: - arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_LG2: - arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_LIT: - arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_LOG: - arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_MAD: - arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]); - break; - case TGSI_OPCODE_MAX: - arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MIN: - arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_MOV: - arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none); - break; - case TGSI_OPCODE_MUL: - arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_POW: - tmp = temp(vpc); - arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none, - swz(src[0], X, X, X, X)); - arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X), - swz(src[1], X, X, X, X), none); - arith(vpc, 1, OP_EX2, dst, mask, none, none, - swz(tmp, X, X, X, X)); - break; - case TGSI_OPCODE_RCP: - arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_RET: - break; - case TGSI_OPCODE_RSQ: - arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]); - break; - case TGSI_OPCODE_SGE: - arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SGT: - arith(vpc, 0, OP_SGT, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SLT: - arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none); - break; - case TGSI_OPCODE_SUB: - arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1])); - break; - case TGSI_OPCODE_XPD: - tmp = temp(vpc); - arith(vpc, 0, OP_MUL, tmp, mask, - swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none); - arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W), - swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y), - neg(tmp)); - break; - default: - NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode); - return FALSE; - } - - return TRUE; -} - -static boolean -nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, - const struct tgsi_full_declaration *fdec) -{ - int hw; - - switch (fdec->Semantic.SemanticName) { - case TGSI_SEMANTIC_POSITION: - hw = NV30_VP_INST_DEST_POS; - break; - case TGSI_SEMANTIC_COLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV30_VP_INST_DEST_COL0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV30_VP_INST_DEST_COL1; - } else { - NOUVEAU_ERR("bad colour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_BCOLOR: - if (fdec->Semantic.SemanticIndex == 0) { - hw = NV30_VP_INST_DEST_BFC0; - } else - if (fdec->Semantic.SemanticIndex == 1) { - hw = NV30_VP_INST_DEST_BFC1; - } else { - NOUVEAU_ERR("bad bcolour semantic index\n"); - return FALSE; - } - break; - case TGSI_SEMANTIC_FOG: - hw = NV30_VP_INST_DEST_FOGC; - break; - case TGSI_SEMANTIC_PSIZE: - hw = NV30_VP_INST_DEST_PSZ; - break; - case TGSI_SEMANTIC_GENERIC: - if (fdec->Semantic.SemanticIndex <= 7) { - hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex); - } else { - NOUVEAU_ERR("bad generic semantic index\n"); - return FALSE; - } - break; - default: - NOUVEAU_ERR("bad output semantic\n"); - return FALSE; - } - - vpc->output_map[fdec->DeclarationRange.First] = hw; - return TRUE; -} - -static boolean -nv30_vertprog_prepare(struct nv30_vpc *vpc) -{ - struct tgsi_parse_context p; - int nr_imm = 0; - - tgsi_parse_init(&p, vpc->vp->pipe.tokens); - while (!tgsi_parse_end_of_tokens(&p)) { - const union tgsi_full_token *tok = &p.FullToken; - - tgsi_parse_token(&p); - switch(tok->Token.Type) { - case TGSI_TOKEN_TYPE_IMMEDIATE: - nr_imm++; - break; - default: - break; - } - } - tgsi_parse_free(&p); - - if (nr_imm) { - vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); - assert(vpc->imm); - } - - return TRUE; -} - -static void -nv30_vertprog_translate(struct nv30_context *nv30, - struct nv30_vertex_program *vp) -{ - struct tgsi_parse_context parse; - struct nv30_vpc *vpc = NULL; - - tgsi_dump(vp->pipe.tokens,0); - - vpc = CALLOC(1, sizeof(struct nv30_vpc)); - if (!vpc) - return; - vpc->vp = vp; - vpc->high_temp = -1; - - if (!nv30_vertprog_prepare(vpc)) { - FREE(vpc); - return; - } - - tgsi_parse_init(&parse, vp->pipe.tokens); - - while (!tgsi_parse_end_of_tokens(&parse)) { - tgsi_parse_token(&parse); - - switch (parse.FullToken.Token.Type) { - case TGSI_TOKEN_TYPE_DECLARATION: - { - const struct tgsi_full_declaration *fdec; - fdec = &parse.FullToken.FullDeclaration; - switch (fdec->Declaration.File) { - case TGSI_FILE_OUTPUT: - if (!nv30_vertprog_parse_decl_output(vpc, fdec)) - goto out_err; - break; - default: - break; - } - } - break; - case TGSI_TOKEN_TYPE_IMMEDIATE: - { - const struct tgsi_full_immediate *imm; - - imm = &parse.FullToken.FullImmediate; - assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32); -// assert(imm->Immediate.Size == 4); - vpc->imm[vpc->nr_imm++] = - constant(vpc, -1, - imm->u.ImmediateFloat32[0].Float, - imm->u.ImmediateFloat32[1].Float, - imm->u.ImmediateFloat32[2].Float, - imm->u.ImmediateFloat32[3].Float); - } - break; - case TGSI_TOKEN_TYPE_INSTRUCTION: - { - const struct tgsi_full_instruction *finst; - finst = &parse.FullToken.FullInstruction; - if (!nv30_vertprog_parse_instruction(vpc, finst)) - goto out_err; - } - break; - default: - break; - } - } - - vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST; - vp->translated = TRUE; -out_err: - tgsi_parse_free(&parse); - FREE(vpc); -} - -static boolean -nv30_vertprog_validate(struct nv30_context *nv30) -{ - struct nouveau_winsys *nvws = nv30->nvws; - struct pipe_winsys *ws = nv30->pipe.winsys; - struct nouveau_grobj *rankine = nv30->screen->rankine; - struct nv30_vertex_program *vp; - struct pipe_buffer *constbuf; - boolean upload_code = FALSE, upload_data = FALSE; - int i; - - vp = nv30->vertprog; - constbuf = nv30->constbuf[PIPE_SHADER_VERTEX]; - - /* Translate TGSI shader into hw bytecode */ - if (!vp->translated) { - nv30_vertprog_translate(nv30, vp); - if (!vp->translated) - return FALSE; - } - - /* Allocate hw vtxprog exec slots */ - if (!vp->exec) { - struct nouveau_resource *heap = nv30->screen->vp_exec_heap; - struct nouveau_stateobj *so; - uint vplen = vp->nr_insns; - - if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { - while (heap->next && heap->size < vplen) { - struct nv30_vertex_program *evict; - - evict = heap->next->priv; - nvws->res_free(&evict->exec); - } - - if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) - assert(0); - } - - so = so_new(2, 0); - so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1); - so_data (so, vp->exec->start); - so_ref(so, &vp->so); - - upload_code = TRUE; - } - - /* Allocate hw vtxprog const slots */ - if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv30->screen->vp_data_heap; - - if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { - while (heap->next && heap->size < vp->nr_consts) { - struct nv30_vertex_program *evict; - - evict = heap->next->priv; - nvws->res_free(&evict->data); - } - - if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) - assert(0); - } - - /*XXX: handle this some day */ - assert(vp->data->start >= vp->data_start_min); - - upload_data = TRUE; - if (vp->data_start != vp->data->start) - upload_code = TRUE; - } - - /* If exec or data segments moved we need to patch the program to - * fixup offsets and register IDs. - */ - if (vp->exec_start != vp->exec->start) { - for (i = 0; i < vp->nr_insns; i++) { - struct nv30_vertex_program_exec *vpi = &vp->insns[i]; - - if (vpi->has_branch_offset) { - assert(0); - } - } - - vp->exec_start = vp->exec->start; - } - - if (vp->nr_consts && vp->data_start != vp->data->start) { - for (i = 0; i < vp->nr_insns; i++) { - struct nv30_vertex_program_exec *vpi = &vp->insns[i]; - - if (vpi->const_index >= 0) { - vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; - vpi->data[1] |= - (vpi->const_index + vp->data->start) << - NV30_VP_INST_CONST_SRC_SHIFT; - - } - } - - vp->data_start = vp->data->start; - } - - /* Update + Upload constant values */ - if (vp->nr_consts) { - float *map = NULL; - - if (constbuf) { - map = ws->buffer_map(ws, constbuf, - PIPE_BUFFER_USAGE_CPU_READ); - } - - for (i = 0; i < vp->nr_consts; i++) { - struct nv30_vertex_program_data *vpd = &vp->consts[i]; - - if (vpd->index >= 0) { - if (!upload_data && - !memcmp(vpd->value, &map[vpd->index * 4], - 4 * sizeof(float))) - continue; - memcpy(vpd->value, &map[vpd->index * 4], - 4 * sizeof(float)); - } - - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5); - OUT_RING (i + vp->data->start); - OUT_RINGp ((uint32_t *)vpd->value, 4); - } - - if (constbuf) { - ws->buffer_unmap(ws, constbuf); - } - } - - /* Upload vtxprog */ - if (upload_code) { -#if 0 - for (i = 0; i < vp->nr_insns; i++) { - NOUVEAU_MSG("VP inst %d: 0x%08x 0x%08x 0x%08x 0x%08x\n", - i, vp->insns[i].data[0], vp->insns[i].data[1], - vp->insns[i].data[2], vp->insns[i].data[3]); - } -#endif - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1); - OUT_RING (vp->exec->start); - for (i = 0; i < vp->nr_insns; i++) { - BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4); - OUT_RINGp (vp->insns[i].data, 4); - } - } - - if (vp->so != nv30->state.hw[NV30_STATE_VERTPROG]) { - so_ref(vp->so, &nv30->state.hw[NV30_STATE_VERTPROG]); - return TRUE; - } - - return FALSE; -} - -void -nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) -{ - struct nouveau_winsys *nvws = nv30->screen->nvws; - - vp->translated = FALSE; - - if (vp->nr_insns) { - FREE(vp->insns); - vp->insns = NULL; - vp->nr_insns = 0; - } - - if (vp->nr_consts) { - FREE(vp->consts); - vp->consts = NULL; - vp->nr_consts = 0; - } - - nvws->res_free(&vp->exec); - vp->exec_start = 0; - nvws->res_free(&vp->data); - vp->data_start = 0; - vp->data_start_min = 0; - - vp->ir = vp->or = 0; - so_ref(NULL, &vp->so); -} - -struct nv30_state_entry nv30_state_vertprog = { - .validate = nv30_vertprog_validate, - .dirty = { - .pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/, - .hw = NV30_STATE_VERTPROG, - } -}; -- cgit v1.2.3 From b5a3c4272be1561646c8a104e4faae870f16ddee Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sat, 8 Nov 2008 18:04:33 +0200 Subject: Nouveau: name replace for nv20. No functional changes, only changed function, struct, macro etc. names. nv10 -> nv20 nv30 -> nv20 celsius -> kelvin Did not touch fifo command macros. Don't try to build nv20_vertprog.c for now. Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv20/Makefile | 4 +- src/gallium/drivers/nv20/nv20_clear.c | 4 +- src/gallium/drivers/nv20/nv20_context.c | 194 ++++++++++----------- src/gallium/drivers/nv20/nv20_context.h | 104 +++++------ src/gallium/drivers/nv20/nv20_fragprog.c | 8 +- src/gallium/drivers/nv20/nv20_fragtex.c | 46 ++--- src/gallium/drivers/nv20/nv20_miptree.c | 68 ++++---- src/gallium/drivers/nv20/nv20_prim_vbuf.c | 132 +++++++------- src/gallium/drivers/nv20/nv20_screen.c | 54 +++--- src/gallium/drivers/nv20/nv20_screen.h | 14 +- src/gallium/drivers/nv20/nv20_state.c | 270 ++++++++++++++--------------- src/gallium/drivers/nv20/nv20_state.h | 30 ++-- src/gallium/drivers/nv20/nv20_state_emit.c | 200 ++++++++++----------- src/gallium/drivers/nv20/nv20_surface.c | 20 +-- src/gallium/drivers/nv20/nv20_vbo.c | 30 ++-- src/gallium/drivers/nv20/nv20_vertprog.c | 146 ++++++++-------- 16 files changed, 662 insertions(+), 662 deletions(-) diff --git a/src/gallium/drivers/nv20/Makefile b/src/gallium/drivers/nv20/Makefile index 76aafbe8f0b..d777fd3d8b4 100644 --- a/src/gallium/drivers/nv20/Makefile +++ b/src/gallium/drivers/nv20/Makefile @@ -14,8 +14,8 @@ DRIVER_SOURCES = \ nv20_state.c \ nv20_state_emit.c \ nv20_surface.c \ - nv20_vbo.c \ - nv20_vertprog.c + nv20_vbo.c +# nv20_vertprog.c C_SOURCES = \ $(COMMON_SOURCES) \ diff --git a/src/gallium/drivers/nv20/nv20_clear.c b/src/gallium/drivers/nv20/nv20_clear.c index be7e09cf4b0..81b6f3e78ac 100644 --- a/src/gallium/drivers/nv20/nv20_clear.c +++ b/src/gallium/drivers/nv20/nv20_clear.c @@ -2,10 +2,10 @@ #include "pipe/p_defines.h" #include "pipe/p_state.h" -#include "nv10_context.h" +#include "nv20_context.h" void -nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, +nv20_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue) { pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue); diff --git a/src/gallium/drivers/nv20/nv20_context.c b/src/gallium/drivers/nv20/nv20_context.c index e9b61daae7f..2af5b0203ed 100644 --- a/src/gallium/drivers/nv20/nv20_context.c +++ b/src/gallium/drivers/nv20/nv20_context.c @@ -2,101 +2,101 @@ #include "pipe/p_defines.h" #include "pipe/p_winsys.h" -#include "nv10_context.h" -#include "nv10_screen.h" +#include "nv20_context.h" +#include "nv20_screen.h" static void -nv10_flush(struct pipe_context *pipe, unsigned flags, +nv20_flush(struct pipe_context *pipe, unsigned flags, struct pipe_fence_handle **fence) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - draw_flush(nv10->draw); + draw_flush(nv20->draw); FIRE_RING(fence); } static void -nv10_destroy(struct pipe_context *pipe) +nv20_destroy(struct pipe_context *pipe) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - if (nv10->draw) - draw_destroy(nv10->draw); + if (nv20->draw) + draw_destroy(nv20->draw); - FREE(nv10); + FREE(nv20); } -static void nv10_init_hwctx(struct nv10_context *nv10) +static void nv20_init_hwctx(struct nv20_context *nv20) { - struct nv10_screen *screen = nv10->screen; + struct nv20_screen *screen = nv20->screen; struct nouveau_winsys *nvws = screen->nvws; int i; float projectionmatrix[16]; - BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1); + BEGIN_RING(kelvin, NV10TCL_DMA_NOTIFY, 1); OUT_RING (screen->sync->handle); - BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2); + BEGIN_RING(kelvin, NV10TCL_DMA_IN_MEMORY0, 2); OUT_RING (nvws->channel->vram->handle); OUT_RING (nvws->channel->gart->handle); - BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY2, 2); + BEGIN_RING(kelvin, NV10TCL_DMA_IN_MEMORY2, 2); OUT_RING (nvws->channel->vram->handle); OUT_RING (nvws->channel->vram->handle); - BEGIN_RING(celsius, NV10TCL_NOP, 1); + BEGIN_RING(kelvin, NV10TCL_NOP, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 2); + BEGIN_RING(kelvin, NV10TCL_RT_HORIZ, 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1); + BEGIN_RING(kelvin, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1); OUT_RING ((0x7ff<<16)|0x800); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1); + BEGIN_RING(kelvin, NV10TCL_VIEWPORT_CLIP_VERT(0), 1); OUT_RING ((0x7ff<<16)|0x800); for (i=1;i<8;i++) { - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1); + BEGIN_RING(kelvin, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(i), 1); + BEGIN_RING(kelvin, NV10TCL_VIEWPORT_CLIP_VERT(i), 1); OUT_RING (0); } - BEGIN_RING(celsius, 0x290, 1); + BEGIN_RING(kelvin, 0x290, 1); OUT_RING ((0x10<<16)|1); - BEGIN_RING(celsius, 0x3f4, 1); + BEGIN_RING(kelvin, 0x3f4, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_NOP, 1); + BEGIN_RING(kelvin, NV10TCL_NOP, 1); OUT_RING (0); - if (nv10->screen->celsius->grclass != NV10TCL) { + if (nv20->screen->kelvin->grclass != NV10TCL) { /* For nv11, nv17 */ - BEGIN_RING(celsius, 0x120, 3); + BEGIN_RING(kelvin, 0x120, 3); OUT_RING (0); OUT_RING (1); OUT_RING (2); - BEGIN_RING(celsius, NV10TCL_NOP, 1); + BEGIN_RING(kelvin, NV10TCL_NOP, 1); OUT_RING (0); } - BEGIN_RING(celsius, NV10TCL_NOP, 1); + BEGIN_RING(kelvin, NV10TCL_NOP, 1); OUT_RING (0); /* Set state */ - BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_FOG_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_ALPHA_FUNC_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 2); + BEGIN_RING(kelvin, NV10TCL_ALPHA_FUNC_FUNC, 2); OUT_RING (0x207); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2); + BEGIN_RING(kelvin, NV10TCL_TX_ENABLE(0), 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_RC_IN_ALPHA(0), 12); + BEGIN_RING(kelvin, NV10TCL_RC_IN_ALPHA(0), 12); OUT_RING (0x30141010); OUT_RING (0); OUT_RING (0x20040000); @@ -110,22 +110,22 @@ static void nv10_init_hwctx(struct nv10_context *nv10) OUT_RING (0x300e0300); OUT_RING (0x0c091c80); - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_BLEND_FUNC_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2); + BEGIN_RING(kelvin, NV10TCL_DITHER_ENABLE, 2); OUT_RING (1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_LINE_SMOOTH_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VERTEX_WEIGHT_ENABLE, 2); + BEGIN_RING(kelvin, NV10TCL_VERTEX_WEIGHT_ENABLE, 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_SRC, 4); + BEGIN_RING(kelvin, NV10TCL_BLEND_FUNC_SRC, 4); OUT_RING (1); OUT_RING (0); OUT_RING (0); OUT_RING (0x8006); - BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 8); + BEGIN_RING(kelvin, NV10TCL_STENCIL_MASK, 8); OUT_RING (0xff); OUT_RING (0x207); OUT_RING (0); @@ -134,103 +134,103 @@ static void nv10_init_hwctx(struct nv10_context *nv10) OUT_RING (0x1e00); OUT_RING (0x1e00); OUT_RING (0x1d01); - BEGIN_RING(celsius, NV10TCL_NORMALIZE_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_NORMALIZE_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 2); + BEGIN_RING(kelvin, NV10TCL_FOG_ENABLE, 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_LIGHT_MODEL, 1); + BEGIN_RING(kelvin, NV10TCL_LIGHT_MODEL, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_COLOR_CONTROL, 1); + BEGIN_RING(kelvin, NV10TCL_COLOR_CONTROL, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_ENABLED_LIGHTS, 1); + BEGIN_RING(kelvin, NV10TCL_ENABLED_LIGHTS, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3); + BEGIN_RING(kelvin, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3); OUT_RING (0); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + BEGIN_RING(kelvin, NV10TCL_DEPTH_FUNC, 1); OUT_RING (0x201); - BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_DEPTH_WRITE_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_DEPTH_TEST_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_FACTOR, 2); + BEGIN_RING(kelvin, NV10TCL_POLYGON_OFFSET_FACTOR, 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + BEGIN_RING(kelvin, NV10TCL_POINT_SIZE, 1); OUT_RING (8); - BEGIN_RING(celsius, NV10TCL_POINT_PARAMETERS_ENABLE, 2); + BEGIN_RING(kelvin, NV10TCL_POINT_PARAMETERS_ENABLE, 2); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_LINE_WIDTH, 1); + BEGIN_RING(kelvin, NV10TCL_LINE_WIDTH, 1); OUT_RING (8); - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_LINE_SMOOTH_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + BEGIN_RING(kelvin, NV10TCL_POLYGON_MODE_FRONT, 2); OUT_RING (0x1b02); OUT_RING (0x1b02); - BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + BEGIN_RING(kelvin, NV10TCL_CULL_FACE, 2); OUT_RING (0x405); OUT_RING (0x901); - BEGIN_RING(celsius, NV10TCL_POLYGON_SMOOTH_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_POLYGON_SMOOTH_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_CULL_FACE_ENABLE, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_CLIP_PLANE_ENABLE(0), 8); + BEGIN_RING(kelvin, NV10TCL_CLIP_PLANE_ENABLE(0), 8); for (i=0;i<8;i++) { OUT_RING (0); } - BEGIN_RING(celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3); + BEGIN_RING(kelvin, NV10TCL_FOG_EQUATION_CONSTANT, 3); OUT_RING (0x3fc00000); /* -1.50 */ OUT_RING (0xbdb8aa0a); /* -0.09 */ OUT_RING (0); /* 0.00 */ - BEGIN_RING(celsius, NV10TCL_NOP, 1); + BEGIN_RING(kelvin, NV10TCL_NOP, 1); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_FOG_MODE, 2); + BEGIN_RING(kelvin, NV10TCL_FOG_MODE, 2); OUT_RING (0x802); OUT_RING (2); /* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when * using texturing, except when using the texture matrix */ - BEGIN_RING(celsius, NV10TCL_VIEW_MATRIX_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_VIEW_MATRIX_ENABLE, 1); OUT_RING (6); - BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + BEGIN_RING(kelvin, NV10TCL_COLOR_MASK, 1); OUT_RING (0x01010101); /* Set vertex component */ - BEGIN_RING(celsius, NV10TCL_VERTEX_COL_4F_R, 4); + BEGIN_RING(kelvin, NV10TCL_VERTEX_COL_4F_R, 4); OUT_RINGf (1.0); OUT_RINGf (1.0); OUT_RINGf (1.0); OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_COL2_3F_R, 3); + BEGIN_RING(kelvin, NV10TCL_VERTEX_COL2_3F_R, 3); OUT_RING (0); OUT_RING (0); OUT_RING (0); - BEGIN_RING(celsius, NV10TCL_VERTEX_NOR_3F_X, 3); + BEGIN_RING(kelvin, NV10TCL_VERTEX_NOR_3F_X, 3); OUT_RING (0); OUT_RING (0); OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_TX0_4F_S, 4); + BEGIN_RING(kelvin, NV10TCL_VERTEX_TX0_4F_S, 4); OUT_RINGf (0.0); OUT_RINGf (0.0); OUT_RINGf (0.0); OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_TX1_4F_S, 4); + BEGIN_RING(kelvin, NV10TCL_VERTEX_TX1_4F_S, 4); OUT_RINGf (0.0); OUT_RINGf (0.0); OUT_RINGf (0.0); OUT_RINGf (1.0); - BEGIN_RING(celsius, NV10TCL_VERTEX_FOG_1F, 1); + BEGIN_RING(kelvin, NV10TCL_VERTEX_FOG_1F, 1); OUT_RINGf (0.0); - BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_EDGEFLAG_ENABLE, 1); OUT_RING (1); memset(projectionmatrix, 0, sizeof(projectionmatrix)); - BEGIN_RING(celsius, NV10TCL_PROJECTION_MATRIX(0), 16); + BEGIN_RING(kelvin, NV10TCL_PROJECTION_MATRIX(0), 16); projectionmatrix[0*4+0] = 1.0; projectionmatrix[1*4+1] = 1.0; projectionmatrix[2*4+2] = 1.0; @@ -239,11 +239,11 @@ static void nv10_init_hwctx(struct nv10_context *nv10) OUT_RINGf (projectionmatrix[i]); } - BEGIN_RING(celsius, NV10TCL_DEPTH_RANGE_NEAR, 2); + BEGIN_RING(kelvin, NV10TCL_DEPTH_RANGE_NEAR, 2); OUT_RING (0.0); OUT_RINGf (16777216.0); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4); + BEGIN_RING(kelvin, NV10TCL_VIEWPORT_SCALE_X, 4); OUT_RINGf (-2048.0); OUT_RINGf (-2048.0); OUT_RINGf (16777215.0 * 0.5); @@ -253,44 +253,44 @@ static void nv10_init_hwctx(struct nv10_context *nv10) } static void -nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) +nv20_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield) { } struct pipe_context * -nv10_create(struct pipe_screen *pscreen, unsigned pctx_id) +nv20_create(struct pipe_screen *pscreen, unsigned pctx_id) { - struct nv10_screen *screen = nv10_screen(pscreen); + struct nv20_screen *screen = nv20_screen(pscreen); struct pipe_winsys *ws = pscreen->winsys; - struct nv10_context *nv10; + struct nv20_context *nv20; struct nouveau_winsys *nvws = screen->nvws; - nv10 = CALLOC(1, sizeof(struct nv10_context)); - if (!nv10) + nv20 = CALLOC(1, sizeof(struct nv20_context)); + if (!nv20) return NULL; - nv10->screen = screen; - nv10->pctx_id = pctx_id; + nv20->screen = screen; + nv20->pctx_id = pctx_id; - nv10->nvws = nvws; + nv20->nvws = nvws; - nv10->pipe.winsys = ws; - nv10->pipe.screen = pscreen; - nv10->pipe.destroy = nv10_destroy; - nv10->pipe.set_edgeflags = nv10_set_edgeflags; - nv10->pipe.draw_arrays = nv10_draw_arrays; - nv10->pipe.draw_elements = nv10_draw_elements; - nv10->pipe.clear = nv10_clear; - nv10->pipe.flush = nv10_flush; + nv20->pipe.winsys = ws; + nv20->pipe.screen = pscreen; + nv20->pipe.destroy = nv20_destroy; + nv20->pipe.set_edgeflags = nv20_set_edgeflags; + nv20->pipe.draw_arrays = nv20_draw_arrays; + nv20->pipe.draw_elements = nv20_draw_elements; + nv20->pipe.clear = nv20_clear; + nv20->pipe.flush = nv20_flush; - nv10_init_surface_functions(nv10); - nv10_init_state_functions(nv10); + nv20_init_surface_functions(nv20); + nv20_init_state_functions(nv20); - nv10->draw = draw_create(); - assert(nv10->draw); - draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10)); + nv20->draw = draw_create(); + assert(nv20->draw); + draw_set_rasterize_stage(nv20->draw, nv20_draw_vbuf_stage(nv20)); - nv10_init_hwctx(nv10); + nv20_init_hwctx(nv20); - return &nv10->pipe; + return &nv20->pipe; } diff --git a/src/gallium/drivers/nv20/nv20_context.h b/src/gallium/drivers/nv20/nv20_context.h index f3b56de25a7..8ad926db20a 100644 --- a/src/gallium/drivers/nv20/nv20_context.h +++ b/src/gallium/drivers/nv20/nv20_context.h @@ -1,5 +1,5 @@ -#ifndef __NV10_CONTEXT_H__ -#define __NV10_CONTEXT_H__ +#ifndef __NV20_CONTEXT_H__ +#define __NV20_CONTEXT_H__ #include "pipe/p_context.h" #include "pipe/p_defines.h" @@ -15,42 +15,42 @@ #include "nouveau/nouveau_gldefs.h" #define NOUVEAU_PUSH_CONTEXT(ctx) \ - struct nv10_screen *ctx = nv10->screen + struct nv20_screen *ctx = nv20->screen #include "nouveau/nouveau_push.h" -#include "nv10_state.h" +#include "nv20_state.h" #define NOUVEAU_ERR(fmt, args...) \ fprintf(stderr, "%s:%d - "fmt, __func__, __LINE__, ##args); #define NOUVEAU_MSG(fmt, args...) \ fprintf(stderr, "nouveau: "fmt, ##args); -#define NV10_NEW_VERTPROG (1 << 0) -#define NV10_NEW_FRAGPROG (1 << 1) -#define NV10_NEW_VTXARRAYS (1 << 2) -#define NV10_NEW_BLEND (1 << 3) -#define NV10_NEW_BLENDCOL (1 << 4) -#define NV10_NEW_RAST (1 << 5) -#define NV10_NEW_DSA (1 << 6) -#define NV10_NEW_VIEWPORT (1 << 7) -#define NV10_NEW_SCISSOR (1 << 8) -#define NV10_NEW_FRAMEBUFFER (1 << 9) +#define NV20_NEW_VERTPROG (1 << 0) +#define NV20_NEW_FRAGPROG (1 << 1) +#define NV20_NEW_VTXARRAYS (1 << 2) +#define NV20_NEW_BLEND (1 << 3) +#define NV20_NEW_BLENDCOL (1 << 4) +#define NV20_NEW_RAST (1 << 5) +#define NV20_NEW_DSA (1 << 6) +#define NV20_NEW_VIEWPORT (1 << 7) +#define NV20_NEW_SCISSOR (1 << 8) +#define NV20_NEW_FRAMEBUFFER (1 << 9) -#include "nv10_screen.h" +#include "nv20_screen.h" -struct nv10_context { +struct nv20_context { struct pipe_context pipe; struct nouveau_winsys *nvws; - struct nv10_screen *screen; + struct nv20_screen *screen; unsigned pctx_id; struct draw_context *draw; uint32_t dirty; - struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; - struct nv10_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; + struct nv20_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS]; + struct nv20_miptree *tex_miptree[PIPE_MAX_SAMPLERS]; unsigned dirty_samplers; unsigned fp_samplers; unsigned vp_samplers; @@ -60,10 +60,10 @@ struct nv10_context { struct pipe_buffer *zeta; uint32_t lma_offset; - struct nv10_blend_state *blend; + struct nv20_blend_state *blend; struct pipe_blend_color *blend_color; - struct nv10_rasterizer_state *rast; - struct nv10_depth_stencil_alpha_state *dsa; + struct nv20_rasterizer_state *rast; + struct nv20_depth_stencil_alpha_state *dsa; struct pipe_viewport_state *viewport; struct pipe_scissor_state *scissor; struct pipe_framebuffer_state *framebuffer; @@ -90,15 +90,15 @@ struct nv10_context { struct nouveau_resource *exec_heap; struct nouveau_resource *data_heap; - struct nv10_vertex_program *active; + struct nv20_vertex_program *active; - struct nv10_vertex_program *current; + struct nv20_vertex_program *current; } vertprog; */ struct { - struct nv10_fragment_program *active; + struct nv20_fragment_program *active; - struct nv10_fragment_program *current; + struct nv20_fragment_program *current; struct pipe_buffer *constant_buf; } fragprog; @@ -106,45 +106,45 @@ struct nv10_context { struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; }; -static INLINE struct nv10_context * -nv10_context(struct pipe_context *pipe) +static INLINE struct nv20_context * +nv20_context(struct pipe_context *pipe) { - return (struct nv10_context *)pipe; + return (struct nv20_context *)pipe; } -extern void nv10_init_state_functions(struct nv10_context *nv10); -extern void nv10_init_surface_functions(struct nv10_context *nv10); +extern void nv20_init_state_functions(struct nv20_context *nv20); +extern void nv20_init_surface_functions(struct nv20_context *nv20); -extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen); +extern void nv20_screen_init_miptree_functions(struct pipe_screen *pscreen); -/* nv10_clear.c */ -extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps, +/* nv20_clear.c */ +extern void nv20_clear(struct pipe_context *pipe, struct pipe_surface *ps, unsigned clearValue); -/* nv10_draw.c */ -extern struct draw_stage *nv10_draw_render_stage(struct nv10_context *nv10); +/* nv20_draw.c */ +extern struct draw_stage *nv20_draw_render_stage(struct nv20_context *nv20); -/* nv10_fragprog.c */ -extern void nv10_fragprog_bind(struct nv10_context *, - struct nv10_fragment_program *); -extern void nv10_fragprog_destroy(struct nv10_context *, - struct nv10_fragment_program *); +/* nv20_fragprog.c */ +extern void nv20_fragprog_bind(struct nv20_context *, + struct nv20_fragment_program *); +extern void nv20_fragprog_destroy(struct nv20_context *, + struct nv20_fragment_program *); -/* nv10_fragtex.c */ -extern void nv10_fragtex_bind(struct nv10_context *); +/* nv20_fragtex.c */ +extern void nv20_fragtex_bind(struct nv20_context *); -/* nv10_prim_vbuf.c */ -struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ); -extern void nv10_vtxbuf_bind(struct nv10_context* nv10); +/* nv20_prim_vbuf.c */ +struct draw_stage *nv20_draw_vbuf_stage( struct nv20_context *nv20 ); +extern void nv20_vtxbuf_bind(struct nv20_context* nv20); -/* nv10_state.c and friends */ -extern void nv10_emit_hw_state(struct nv10_context *nv10); -extern void nv10_state_tex_update(struct nv10_context *nv10); +/* nv20_state.c and friends */ +extern void nv20_emit_hw_state(struct nv20_context *nv20); +extern void nv20_state_tex_update(struct nv20_context *nv20); -/* nv10_vbo.c */ -extern boolean nv10_draw_arrays(struct pipe_context *, unsigned mode, +/* nv20_vbo.c */ +extern boolean nv20_draw_arrays(struct pipe_context *, unsigned mode, unsigned start, unsigned count); -extern boolean nv10_draw_elements( struct pipe_context *pipe, +extern boolean nv20_draw_elements( struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned prim, unsigned start, unsigned count); diff --git a/src/gallium/drivers/nv20/nv20_fragprog.c b/src/gallium/drivers/nv20/nv20_fragprog.c index 698db5a16a9..4f496369dd3 100644 --- a/src/gallium/drivers/nv20/nv20_fragprog.c +++ b/src/gallium/drivers/nv20/nv20_fragprog.c @@ -6,16 +6,16 @@ #include "tgsi/tgsi_parse.h" #include "tgsi/tgsi_util.h" -#include "nv10_context.h" +#include "nv20_context.h" void -nv10_fragprog_bind(struct nv10_context *nv10, struct nv10_fragment_program *fp) +nv20_fragprog_bind(struct nv20_context *nv20, struct nv20_fragment_program *fp) { } void -nv10_fragprog_destroy(struct nv10_context *nv10, - struct nv10_fragment_program *fp) +nv20_fragprog_destroy(struct nv20_context *nv20, + struct nv20_fragment_program *fp) { } diff --git a/src/gallium/drivers/nv20/nv20_fragtex.c b/src/gallium/drivers/nv20/nv20_fragtex.c index 238634d0bb4..77c34897e2c 100644 --- a/src/gallium/drivers/nv20/nv20_fragtex.c +++ b/src/gallium/drivers/nv20/nv20_fragtex.c @@ -1,4 +1,4 @@ -#include "nv10_context.h" +#include "nv20_context.h" static INLINE int log2i(int i) { @@ -33,14 +33,14 @@ static INLINE int log2i(int i) NV10TCL_TX_FORMAT_FORMAT_##tf, \ } -struct nv10_texture_format { +struct nv20_texture_format { boolean defined; uint pipe; int format; }; -static struct nv10_texture_format -nv10_texture_formats[] = { +static struct nv20_texture_format +nv20_texture_formats[] = { _(A8R8G8B8_UNORM, A8R8G8B8), _(A1R5G5B5_UNORM, A1R5G5B5), _(A4R4G4B4_UNORM, A4R4G4B4), @@ -54,10 +54,10 @@ nv10_texture_formats[] = { {}, }; -static struct nv10_texture_format * -nv10_fragtex_format(uint pipe_format) +static struct nv20_texture_format * +nv20_fragtex_format(uint pipe_format) { - struct nv10_texture_format *tf = nv10_texture_formats; + struct nv20_texture_format *tf = nv20_texture_formats; while (tf->defined) { if (tf->pipe == pipe_format) @@ -70,16 +70,16 @@ nv10_fragtex_format(uint pipe_format) static void -nv10_fragtex_build(struct nv10_context *nv10, int unit) +nv20_fragtex_build(struct nv20_context *nv20, int unit) { #if 0 - struct nv10_sampler_state *ps = nv10->tex_sampler[unit]; - struct nv10_miptree *nv10mt = nv10->tex_miptree[unit]; - struct pipe_texture *pt = &nv10mt->base; - struct nv10_texture_format *tf; + struct nv20_sampler_state *ps = nv20->tex_sampler[unit]; + struct nv20_miptree *nv20mt = nv20->tex_miptree[unit]; + struct pipe_texture *pt = &nv20mt->base; + struct nv20_texture_format *tf; uint32_t txf, txs, txp; - tf = nv10_fragtex_format(pt->format); + tf = nv20_fragtex_format(pt->format); if (!tf || !tf->defined) { NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format); return; @@ -107,9 +107,9 @@ nv10_fragtex_build(struct nv10_context *nv10, int unit) return; } - BEGIN_RING(celsius, NV10TCL_TX_OFFSET(unit), 8); - OUT_RELOCl(nv10mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - OUT_RELOCd(nv10mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); + BEGIN_RING(kelvin, NV10TCL_TX_OFFSET(unit), 8); + OUT_RELOCl(nv20mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + OUT_RELOCd(nv20mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/); OUT_RING (ps->wrap); OUT_RING (0x40000000); /* enable */ OUT_RING (txs); @@ -120,30 +120,30 @@ nv10_fragtex_build(struct nv10_context *nv10, int unit) } void -nv10_fragtex_bind(struct nv10_context *nv10) +nv20_fragtex_bind(struct nv20_context *nv20) { #if 0 - struct nv10_fragment_program *fp = nv10->fragprog.active; + struct nv20_fragment_program *fp = nv20->fragprog.active; unsigned samplers, unit; - samplers = nv10->fp_samplers & ~fp->samplers; + samplers = nv20->fp_samplers & ~fp->samplers; while (samplers) { unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - BEGIN_RING(celsius, NV10TCL_TX_ENABLE(unit), 1); + BEGIN_RING(kelvin, NV10TCL_TX_ENABLE(unit), 1); OUT_RING (0); } - samplers = nv10->dirty_samplers & fp->samplers; + samplers = nv20->dirty_samplers & fp->samplers; while (samplers) { unit = ffs(samplers) - 1; samplers &= ~(1 << unit); - nv10_fragtex_build(nv10, unit); + nv20_fragtex_build(nv20, unit); } - nv10->fp_samplers = fp->samplers; + nv20->fp_samplers = fp->samplers; #endif } diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c index ad084e72b81..dbfd779de9d 100644 --- a/src/gallium/drivers/nv20/nv20_miptree.c +++ b/src/gallium/drivers/nv20/nv20_miptree.c @@ -2,13 +2,13 @@ #include "pipe/p_defines.h" #include "pipe/p_inlines.h" -#include "nv10_context.h" -#include "nv10_screen.h" +#include "nv20_context.h" +#include "nv20_screen.h" static void -nv10_miptree_layout(struct nv10_miptree *nv10mt) +nv20_miptree_layout(struct nv20_miptree *nv20mt) { - struct pipe_texture *pt = &nv10mt->base; + struct pipe_texture *pt = &nv20mt->base; boolean swizzled = FALSE; uint width = pt->width[0], height = pt->height[0]; uint offset = 0; @@ -27,12 +27,12 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); if (swizzled) - nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; + nv20mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; else - nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; - nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; + nv20mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; + nv20mt->level[l].pitch = (nv20mt->level[l].pitch + 63) & ~63; - nv10mt->level[l].image_offset = + nv20mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = MAX2(1, width >> 1); @@ -42,28 +42,28 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) for (f = 0; f < nr_faces; f++) { for (l = 0; l <= pt->last_level; l++) { - nv10mt->level[l].image_offset[f] = offset; - offset += nv10mt->level[l].pitch * pt->height[l]; + nv20mt->level[l].image_offset[f] = offset; + offset += nv20mt->level[l].pitch * pt->height[l]; } } - nv10mt->total_size = offset; + nv20mt->total_size = offset; } static struct pipe_texture * -nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) +nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) { struct pipe_winsys *ws = screen->winsys; - struct nv10_miptree *mt; + struct nv20_miptree *mt; - mt = MALLOC(sizeof(struct nv10_miptree)); + mt = MALLOC(sizeof(struct nv20_miptree)); if (!mt) return NULL; mt->base = *pt; mt->base.refcount = 1; mt->base.screen = screen; - nv10_miptree_layout(mt); + nv20_miptree_layout(mt); mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL, mt->total_size); @@ -76,72 +76,72 @@ nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) } static void -nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) +nv20_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) { struct pipe_texture *mt = *pt; *pt = NULL; if (--mt->refcount <= 0) { - struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt; + struct nv20_miptree *nv20mt = (struct nv20_miptree *)mt; int l; - pipe_buffer_reference(screen, &nv10mt->buffer, NULL); + pipe_buffer_reference(screen, &nv20mt->buffer, NULL); for (l = 0; l <= mt->last_level; l++) { - if (nv10mt->level[l].image_offset) - FREE(nv10mt->level[l].image_offset); + if (nv20mt->level[l].image_offset) + FREE(nv20mt->level[l].image_offset); } - FREE(nv10mt); + FREE(nv20mt); } } static void -nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, +nv20_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, uint face, uint levels) { } static struct pipe_surface * -nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, +nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { struct pipe_winsys *ws = screen->winsys; - struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt; + struct nv20_miptree *nv20mt = (struct nv20_miptree *)pt; struct pipe_surface *ps; ps = ws->surface_alloc(ws); if (!ps) return NULL; - pipe_buffer_reference(screen, &ps->buffer, nv10mt->buffer); + pipe_buffer_reference(screen, &ps->buffer, nv20mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; ps->height = pt->height[level]; ps->block = pt->block; ps->nblocksx = pt->nblocksx[level]; ps->nblocksy = pt->nblocksy[level]; - ps->stride = nv10mt->level[level].pitch; + ps->stride = nv20mt->level[level].pitch; if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset = nv10mt->level[level].image_offset[face]; + ps->offset = nv20mt->level[level].image_offset[face]; } else { - ps->offset = nv10mt->level[level].image_offset[0]; + ps->offset = nv20mt->level[level].image_offset[0]; } return ps; } static void -nv10_miptree_surface_release(struct pipe_screen *screen, +nv20_miptree_surface_release(struct pipe_screen *screen, struct pipe_surface **surface) { } -void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen) +void nv20_screen_init_miptree_functions(struct pipe_screen *pscreen) { - pscreen->texture_create = nv10_miptree_create; - pscreen->texture_release = nv10_miptree_release; - pscreen->get_tex_surface = nv10_miptree_surface_get; - pscreen->tex_surface_release = nv10_miptree_surface_release; + pscreen->texture_create = nv20_miptree_create; + pscreen->texture_release = nv20_miptree_release; + pscreen->get_tex_surface = nv20_miptree_surface_get; + pscreen->tex_surface_release = nv20_miptree_surface_release; } diff --git a/src/gallium/drivers/nv20/nv20_prim_vbuf.c b/src/gallium/drivers/nv20/nv20_prim_vbuf.c index 62a8f6d89da..a51d657d274 100644 --- a/src/gallium/drivers/nv20/nv20_prim_vbuf.c +++ b/src/gallium/drivers/nv20/nv20_prim_vbuf.c @@ -42,18 +42,18 @@ #include "pipe/p_inlines.h" #include "pipe/p_winsys.h" -#include "nv10_context.h" -#include "nv10_state.h" +#include "nv20_context.h" +#include "nv20_state.h" #include "draw/draw_vbuf.h" /** - * Primitive renderer for nv10. + * Primitive renderer for nv20. */ -struct nv10_vbuf_render { +struct nv20_vbuf_render { struct vbuf_render base; - struct nv10_context *nv10; + struct nv20_context *nv20; /** Vertex buffer */ struct pipe_buffer* buffer; @@ -66,13 +66,13 @@ struct nv10_vbuf_render { }; -void nv10_vtxbuf_bind( struct nv10_context* nv10 ) +void nv20_vtxbuf_bind( struct nv20_context* nv20 ) { int i; for(i = 0; i < 8; i++) { - BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1); - OUT_RING(0/*nv10->vtxbuf*/); - BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1); + BEGIN_RING(kelvin, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1); + OUT_RING(0/*nv20->vtxbuf*/); + BEGIN_RING(kelvin, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1); OUT_RING(0/*XXX*/); } } @@ -80,75 +80,75 @@ void nv10_vtxbuf_bind( struct nv10_context* nv10 ) /** * Basically a cast wrapper. */ -static INLINE struct nv10_vbuf_render * -nv10_vbuf_render( struct vbuf_render *render ) +static INLINE struct nv20_vbuf_render * +nv20_vbuf_render( struct vbuf_render *render ) { assert(render); - return (struct nv10_vbuf_render *)render; + return (struct nv20_vbuf_render *)render; } static const struct vertex_info * -nv10_vbuf_render_get_vertex_info( struct vbuf_render *render ) +nv20_vbuf_render_get_vertex_info( struct vbuf_render *render ) { - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; + struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render); + struct nv20_context *nv20 = nv20_render->nv20; - nv10_emit_hw_state(nv10); + nv20_emit_hw_state(nv20); - return &nv10->vertex_info; + return &nv20->vertex_info; } static void * -nv10_vbuf_render_allocate_vertices( struct vbuf_render *render, +nv20_vbuf_render_allocate_vertices( struct vbuf_render *render, ushort vertex_size, ushort nr_vertices ) { - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; - struct pipe_winsys *winsys = nv10->pipe.winsys; + struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render); + struct nv20_context *nv20 = nv20_render->nv20; + struct pipe_winsys *winsys = nv20->pipe.winsys; size_t size = (size_t)vertex_size * (size_t)nr_vertices; - assert(!nv10_render->buffer); - nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); + assert(!nv20_render->buffer); + nv20_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size); - nv10->dirty |= NV10_NEW_VTXARRAYS; + nv20->dirty |= NV20_NEW_VTXARRAYS; return winsys->buffer_map(winsys, - nv10_render->buffer, + nv20_render->buffer, PIPE_BUFFER_USAGE_CPU_WRITE); } static void -nv10_vbuf_render_set_primitive( struct vbuf_render *render, +nv20_vbuf_render_set_primitive( struct vbuf_render *render, unsigned prim ) { - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - nv10_render->hwprim = prim + 1; + struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render); + nv20_render->hwprim = prim + 1; } static void -nv10_vbuf_render_draw( struct vbuf_render *render, +nv20_vbuf_render_draw( struct vbuf_render *render, const ushort *indices, uint nr_indices) { - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; + struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render); + struct nv20_context *nv20 = nv20_render->nv20; int push, i; - nv10_emit_hw_state(nv10); + nv20_emit_hw_state(nv20); - BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); - OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); + BEGIN_RING(kelvin, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1); + OUT_RELOCl(nv20_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); - OUT_RING(nv10_render->hwprim); + BEGIN_RING(kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + OUT_RING(nv20_render->hwprim); if (nr_indices & 1) { - BEGIN_RING(celsius, NV10TCL_VB_ELEMENT_U32, 1); + BEGIN_RING(kelvin, NV10TCL_VB_ELEMENT_U32, 1); OUT_RING (indices[0]); indices++; nr_indices--; } @@ -157,7 +157,7 @@ nv10_vbuf_render_draw( struct vbuf_render *render, // XXX too big/small ? check the size push = MIN2(nr_indices, 1200 * 2); - BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1); + BEGIN_RING_NI(kelvin, NV10TCL_VB_ELEMENT_U16, push >> 1); for (i = 0; i < push; i+=2) OUT_RING((indices[i+1] << 16) | indices[i]); @@ -165,33 +165,33 @@ nv10_vbuf_render_draw( struct vbuf_render *render, indices += push; } - BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); + BEGIN_RING(kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1); OUT_RING (0); } static void -nv10_vbuf_render_release_vertices( struct vbuf_render *render, +nv20_vbuf_render_release_vertices( struct vbuf_render *render, void *vertices, unsigned vertex_size, unsigned vertices_used ) { - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - struct nv10_context *nv10 = nv10_render->nv10; - struct pipe_winsys *winsys = nv10->pipe.winsys; - struct pipe_screen *pscreen = &nv10->screen->pipe; - - assert(nv10_render->buffer); - winsys->buffer_unmap(winsys, nv10_render->buffer); - pipe_buffer_reference(pscreen, &nv10_render->buffer, NULL); + struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render); + struct nv20_context *nv20 = nv20_render->nv20; + struct pipe_winsys *winsys = nv20->pipe.winsys; + struct pipe_screen *pscreen = &nv20->screen->pipe; + + assert(nv20_render->buffer); + winsys->buffer_unmap(winsys, nv20_render->buffer); + pipe_buffer_reference(pscreen, &nv20_render->buffer, NULL); } static void -nv10_vbuf_render_destroy( struct vbuf_render *render ) +nv20_vbuf_render_destroy( struct vbuf_render *render ) { - struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render); - FREE(nv10_render); + struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render); + FREE(nv20_render); } @@ -199,38 +199,38 @@ nv10_vbuf_render_destroy( struct vbuf_render *render ) * Create a new primitive render. */ static struct vbuf_render * -nv10_vbuf_render_create( struct nv10_context *nv10 ) +nv20_vbuf_render_create( struct nv20_context *nv20 ) { - struct nv10_vbuf_render *nv10_render = CALLOC_STRUCT(nv10_vbuf_render); + struct nv20_vbuf_render *nv20_render = CALLOC_STRUCT(nv20_vbuf_render); - nv10_render->nv10 = nv10; + nv20_render->nv20 = nv20; - nv10_render->base.max_vertex_buffer_bytes = 16*1024; - nv10_render->base.max_indices = 1024; - nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info; - nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices; - nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive; - nv10_render->base.draw = nv10_vbuf_render_draw; - nv10_render->base.release_vertices = nv10_vbuf_render_release_vertices; - nv10_render->base.destroy = nv10_vbuf_render_destroy; + nv20_render->base.max_vertex_buffer_bytes = 16*1024; + nv20_render->base.max_indices = 1024; + nv20_render->base.get_vertex_info = nv20_vbuf_render_get_vertex_info; + nv20_render->base.allocate_vertices = nv20_vbuf_render_allocate_vertices; + nv20_render->base.set_primitive = nv20_vbuf_render_set_primitive; + nv20_render->base.draw = nv20_vbuf_render_draw; + nv20_render->base.release_vertices = nv20_vbuf_render_release_vertices; + nv20_render->base.destroy = nv20_vbuf_render_destroy; - return &nv10_render->base; + return &nv20_render->base; } /** * Create a new primitive vbuf/render stage. */ -struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 ) +struct draw_stage *nv20_draw_vbuf_stage( struct nv20_context *nv20 ) { struct vbuf_render *render; struct draw_stage *stage; - render = nv10_vbuf_render_create(nv10); + render = nv20_vbuf_render_create(nv20); if(!render) return NULL; - stage = draw_vbuf_stage( nv10->draw, render ); + stage = draw_vbuf_stage( nv20->draw, render ); if(!stage) { render->destroy(render); return NULL; diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c index 27a9edf9bba..b7f5ea85128 100644 --- a/src/gallium/drivers/nv20/nv20_screen.c +++ b/src/gallium/drivers/nv20/nv20_screen.c @@ -1,13 +1,13 @@ #include "pipe/p_screen.h" -#include "nv10_context.h" -#include "nv10_screen.h" +#include "nv20_context.h" +#include "nv20_screen.h" static const char * -nv10_screen_get_name(struct pipe_screen *screen) +nv20_screen_get_name(struct pipe_screen *screen) { - struct nv10_screen *nv10screen = nv10_screen(screen); - struct nouveau_device *dev = nv10screen->nvws->channel->device; + struct nv20_screen *nv20screen = nv20_screen(screen); + struct nouveau_device *dev = nv20screen->nvws->channel->device; static char buffer[128]; snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset); @@ -15,13 +15,13 @@ nv10_screen_get_name(struct pipe_screen *screen) } static const char * -nv10_screen_get_vendor(struct pipe_screen *screen) +nv20_screen_get_vendor(struct pipe_screen *screen) { return "nouveau"; } static int -nv10_screen_get_param(struct pipe_screen *screen, int param) +nv20_screen_get_param(struct pipe_screen *screen, int param) { switch (param) { case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: @@ -60,7 +60,7 @@ nv10_screen_get_param(struct pipe_screen *screen, int param) } static float -nv10_screen_get_paramf(struct pipe_screen *screen, int param) +nv20_screen_get_paramf(struct pipe_screen *screen, int param) { switch (param) { case PIPE_CAP_MAX_LINE_WIDTH: @@ -80,7 +80,7 @@ nv10_screen_get_paramf(struct pipe_screen *screen, int param) } static boolean -nv10_screen_is_format_supported(struct pipe_screen *screen, +nv20_screen_is_format_supported(struct pipe_screen *screen, enum pipe_format format, enum pipe_texture_target target, unsigned tex_usage, unsigned geom_flags) @@ -114,7 +114,7 @@ nv10_screen_is_format_supported(struct pipe_screen *screen, } static void * -nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, +nv20_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, unsigned flags ) { struct pipe_winsys *ws = screen->winsys; @@ -128,7 +128,7 @@ nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface, } static void -nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) +nv20_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) { struct pipe_winsys *ws = screen->winsys; @@ -136,21 +136,21 @@ nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface) } static void -nv10_screen_destroy(struct pipe_screen *pscreen) +nv20_screen_destroy(struct pipe_screen *pscreen) { - struct nv10_screen *screen = nv10_screen(pscreen); + struct nv20_screen *screen = nv20_screen(pscreen); struct nouveau_winsys *nvws = screen->nvws; nvws->notifier_free(&screen->sync); - nvws->grobj_free(&screen->celsius); + nvws->grobj_free(&screen->kelvin); FREE(pscreen); } struct pipe_screen * -nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) +nv20_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) { - struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen); + struct nv20_screen *screen = CALLOC_STRUCT(nv20_screen); unsigned celsius_class; unsigned chipset = nvws->channel->device->chipset; int ret; @@ -174,7 +174,7 @@ nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) return NULL; } - ret = nvws->grobj_alloc(nvws, celsius_class, &screen->celsius); + ret = nvws->grobj_alloc(nvws, celsius_class, &screen->kelvin); if (ret) { NOUVEAU_ERR("Error creating 3D object: %d\n", ret); return FALSE; @@ -184,24 +184,24 @@ nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws) ret = nvws->notifier_alloc(nvws, 1, &screen->sync); if (ret) { NOUVEAU_ERR("Error creating notifier object: %d\n", ret); - nv10_screen_destroy(&screen->pipe); + nv20_screen_destroy(&screen->pipe); return NULL; } screen->pipe.winsys = ws; - screen->pipe.destroy = nv10_screen_destroy; + screen->pipe.destroy = nv20_screen_destroy; - screen->pipe.get_name = nv10_screen_get_name; - screen->pipe.get_vendor = nv10_screen_get_vendor; - screen->pipe.get_param = nv10_screen_get_param; - screen->pipe.get_paramf = nv10_screen_get_paramf; + screen->pipe.get_name = nv20_screen_get_name; + screen->pipe.get_vendor = nv20_screen_get_vendor; + screen->pipe.get_param = nv20_screen_get_param; + screen->pipe.get_paramf = nv20_screen_get_paramf; - screen->pipe.is_format_supported = nv10_screen_is_format_supported; + screen->pipe.is_format_supported = nv20_screen_is_format_supported; - screen->pipe.surface_map = nv10_surface_map; - screen->pipe.surface_unmap = nv10_surface_unmap; + screen->pipe.surface_map = nv20_surface_map; + screen->pipe.surface_unmap = nv20_surface_unmap; - nv10_screen_init_miptree_functions(&screen->pipe); + nv20_screen_init_miptree_functions(&screen->pipe); return &screen->pipe; } diff --git a/src/gallium/drivers/nv20/nv20_screen.h b/src/gallium/drivers/nv20/nv20_screen.h index 3f8750a13f7..8f2f2e341db 100644 --- a/src/gallium/drivers/nv20/nv20_screen.h +++ b/src/gallium/drivers/nv20/nv20_screen.h @@ -1,22 +1,22 @@ -#ifndef __NV10_SCREEN_H__ -#define __NV10_SCREEN_H__ +#ifndef __NV20_SCREEN_H__ +#define __NV20_SCREEN_H__ #include "pipe/p_screen.h" -struct nv10_screen { +struct nv20_screen { struct pipe_screen pipe; struct nouveau_winsys *nvws; /* HW graphics objects */ - struct nouveau_grobj *celsius; + struct nouveau_grobj *kelvin; struct nouveau_notifier *sync; }; -static INLINE struct nv10_screen * -nv10_screen(struct pipe_screen *screen) +static INLINE struct nv20_screen * +nv20_screen(struct pipe_screen *screen) { - return (struct nv10_screen *)screen; + return (struct nv20_screen *)screen; } #endif diff --git a/src/gallium/drivers/nv20/nv20_state.c b/src/gallium/drivers/nv20/nv20_state.c index d2375aa2f64..c3b87230b71 100644 --- a/src/gallium/drivers/nv20/nv20_state.c +++ b/src/gallium/drivers/nv20/nv20_state.c @@ -5,16 +5,16 @@ #include "tgsi/tgsi_parse.h" -#include "nv10_context.h" -#include "nv10_state.h" +#include "nv20_context.h" +#include "nv20_state.h" static void * -nv10_blend_state_create(struct pipe_context *pipe, +nv20_blend_state_create(struct pipe_context *pipe, const struct pipe_blend_state *cso) { - struct nv10_blend_state *cb; + struct nv20_blend_state *cb; - cb = MALLOC(sizeof(struct nv10_blend_state)); + cb = MALLOC(sizeof(struct nv20_blend_state)); cb->b_enable = cso->blend_enable ? 1 : 0; cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) | @@ -33,17 +33,17 @@ nv10_blend_state_create(struct pipe_context *pipe, } static void -nv10_blend_state_bind(struct pipe_context *pipe, void *blend) +nv20_blend_state_bind(struct pipe_context *pipe, void *blend) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->blend = (struct nv10_blend_state*)blend; + nv20->blend = (struct nv20_blend_state*)blend; - nv10->dirty |= NV10_NEW_BLEND; + nv20->dirty |= NV20_NEW_BLEND; } static void -nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso) +nv20_blend_state_delete(struct pipe_context *pipe, void *hwcso) { FREE(hwcso); } @@ -82,13 +82,13 @@ wrap_mode(unsigned wrap) { } static void * -nv10_sampler_state_create(struct pipe_context *pipe, +nv20_sampler_state_create(struct pipe_context *pipe, const struct pipe_sampler_state *cso) { - struct nv10_sampler_state *ps; + struct nv20_sampler_state *ps; uint32_t filter = 0; - ps = MALLOC(sizeof(struct nv10_sampler_state)); + ps = MALLOC(sizeof(struct nv20_sampler_state)); ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) | (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT)); @@ -204,41 +204,41 @@ nv10_sampler_state_create(struct pipe_context *pipe, } static void -nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) +nv20_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); unsigned unit; for (unit = 0; unit < nr; unit++) { - nv10->tex_sampler[unit] = sampler[unit]; - nv10->dirty_samplers |= (1 << unit); + nv20->tex_sampler[unit] = sampler[unit]; + nv20->dirty_samplers |= (1 << unit); } } static void -nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso) +nv20_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { FREE(hwcso); } static void -nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr, +nv20_set_sampler_texture(struct pipe_context *pipe, unsigned nr, struct pipe_texture **miptree) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); unsigned unit; for (unit = 0; unit < nr; unit++) { - nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit]; - nv10->dirty_samplers |= (1 << unit); + nv20->tex_miptree[unit] = (struct nv20_miptree *)miptree[unit]; + nv20->dirty_samplers |= (1 << unit); } } static void * -nv10_rasterizer_state_create(struct pipe_context *pipe, +nv20_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) { - struct nv10_rasterizer_state *rs; + struct nv20_rasterizer_state *rs; int i; /*XXX: ignored: @@ -249,7 +249,7 @@ nv10_rasterizer_state_create(struct pipe_context *pipe, * multisample * offset_units / offset_scale */ - rs = MALLOC(sizeof(struct nv10_rasterizer_state)); + rs = MALLOC(sizeof(struct nv20_rasterizer_state)); rs->templ = cso; @@ -312,30 +312,30 @@ nv10_rasterizer_state_create(struct pipe_context *pipe, } static void -nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast) +nv20_rasterizer_state_bind(struct pipe_context *pipe, void *rast) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->rast = (struct nv10_rasterizer_state*)rast; + nv20->rast = (struct nv20_rasterizer_state*)rast; - draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL)); + draw_set_rasterizer_state(nv20->draw, (nv20->rast ? nv20->rast->templ : NULL)); - nv10->dirty |= NV10_NEW_RAST; + nv20->dirty |= NV20_NEW_RAST; } static void -nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) +nv20_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso) { FREE(hwcso); } static void * -nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, +nv20_depth_stencil_alpha_state_create(struct pipe_context *pipe, const struct pipe_depth_stencil_alpha_state *cso) { - struct nv10_depth_stencil_alpha_state *hw; + struct nv20_depth_stencil_alpha_state *hw; - hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state)); + hw = MALLOC(sizeof(struct nv20_depth_stencil_alpha_state)); hw->depth.func = nvgl_comparison_op(cso->depth.func); hw->depth.write_enable = cso->depth.writemask ? 1 : 0; @@ -358,55 +358,55 @@ nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe, } static void -nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa) +nv20_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa; + nv20->dsa = (struct nv20_depth_stencil_alpha_state*)dsa; - nv10->dirty |= NV10_NEW_DSA; + nv20->dirty |= NV20_NEW_DSA; } static void -nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) +nv20_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso) { FREE(hwcso); } static void * -nv10_vp_state_create(struct pipe_context *pipe, +nv20_vp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *templ) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - return draw_create_vertex_shader(nv10->draw, templ); + return draw_create_vertex_shader(nv20->draw, templ); } static void -nv10_vp_state_bind(struct pipe_context *pipe, void *shader) +nv20_vp_state_bind(struct pipe_context *pipe, void *shader) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); + draw_bind_vertex_shader(nv20->draw, (struct draw_vertex_shader *) shader); - nv10->dirty |= NV10_NEW_VERTPROG; + nv20->dirty |= NV20_NEW_VERTPROG; } static void -nv10_vp_state_delete(struct pipe_context *pipe, void *shader) +nv20_vp_state_delete(struct pipe_context *pipe, void *shader) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader); + draw_delete_vertex_shader(nv20->draw, (struct draw_vertex_shader *) shader); } static void * -nv10_fp_state_create(struct pipe_context *pipe, +nv20_fp_state_create(struct pipe_context *pipe, const struct pipe_shader_state *cso) { - struct nv10_fragment_program *fp; + struct nv20_fragment_program *fp; - fp = CALLOC(1, sizeof(struct nv10_fragment_program)); + fp = CALLOC(1, sizeof(struct nv20_fragment_program)); fp->pipe.tokens = tgsi_dup_tokens(cso->tokens); tgsi_scan_shader(cso->tokens, &fp->info); @@ -415,51 +415,51 @@ nv10_fp_state_create(struct pipe_context *pipe, } static void -nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso) +nv20_fp_state_bind(struct pipe_context *pipe, void *hwcso) { - struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_fragment_program *fp = hwcso; + struct nv20_context *nv20 = nv20_context(pipe); + struct nv20_fragment_program *fp = hwcso; - nv10->fragprog.current = fp; - nv10->dirty |= NV10_NEW_FRAGPROG; + nv20->fragprog.current = fp; + nv20->dirty |= NV20_NEW_FRAGPROG; } static void -nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso) +nv20_fp_state_delete(struct pipe_context *pipe, void *hwcso) { - struct nv10_context *nv10 = nv10_context(pipe); - struct nv10_fragment_program *fp = hwcso; + struct nv20_context *nv20 = nv20_context(pipe); + struct nv20_fragment_program *fp = hwcso; - nv10_fragprog_destroy(nv10, fp); + nv20_fragprog_destroy(nv20, fp); FREE((void*)fp->pipe.tokens); FREE(fp); } static void -nv10_set_blend_color(struct pipe_context *pipe, +nv20_set_blend_color(struct pipe_context *pipe, const struct pipe_blend_color *bcol) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->blend_color = (struct pipe_blend_color*)bcol; + nv20->blend_color = (struct pipe_blend_color*)bcol; - nv10->dirty |= NV10_NEW_BLENDCOL; + nv20->dirty |= NV20_NEW_BLENDCOL; } static void -nv10_set_clip_state(struct pipe_context *pipe, +nv20_set_clip_state(struct pipe_context *pipe, const struct pipe_clip_state *clip) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - draw_set_clip_state(nv10->draw, clip); + draw_set_clip_state(nv20->draw, clip); } static void -nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, +nv20_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, const struct pipe_constant_buffer *buf ) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); struct pipe_winsys *ws = pipe->winsys; assert(shader < PIPE_SHADER_TYPES); @@ -469,8 +469,8 @@ nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, void *mapped; if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ))) { - memcpy(nv10->constbuf[shader], mapped, buf->size); - nv10->constbuf_nr[shader] = + memcpy(nv20->constbuf[shader], mapped, buf->size); + nv20->constbuf_nr[shader] = buf->size / (4 * sizeof(float)); ws->buffer_unmap(ws, buf->buffer); } @@ -478,111 +478,111 @@ nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index, } static void -nv10_set_framebuffer_state(struct pipe_context *pipe, +nv20_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->framebuffer = (struct pipe_framebuffer_state*)fb; + nv20->framebuffer = (struct pipe_framebuffer_state*)fb; - nv10->dirty |= NV10_NEW_FRAMEBUFFER; + nv20->dirty |= NV20_NEW_FRAMEBUFFER; } static void -nv10_set_polygon_stipple(struct pipe_context *pipe, +nv20_set_polygon_stipple(struct pipe_context *pipe, const struct pipe_poly_stipple *stipple) { NOUVEAU_ERR("line stipple hahaha\n"); } static void -nv10_set_scissor_state(struct pipe_context *pipe, +nv20_set_scissor_state(struct pipe_context *pipe, const struct pipe_scissor_state *s) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->scissor = (struct pipe_scissor_state*)s; + nv20->scissor = (struct pipe_scissor_state*)s; - nv10->dirty |= NV10_NEW_SCISSOR; + nv20->dirty |= NV20_NEW_SCISSOR; } static void -nv10_set_viewport_state(struct pipe_context *pipe, +nv20_set_viewport_state(struct pipe_context *pipe, const struct pipe_viewport_state *vpt) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - nv10->viewport = (struct pipe_viewport_state*)vpt; + nv20->viewport = (struct pipe_viewport_state*)vpt; - draw_set_viewport_state(nv10->draw, nv10->viewport); + draw_set_viewport_state(nv20->draw, nv20->viewport); - nv10->dirty |= NV10_NEW_VIEWPORT; + nv20->dirty |= NV20_NEW_VIEWPORT; } static void -nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count, +nv20_set_vertex_buffers(struct pipe_context *pipe, unsigned count, const struct pipe_vertex_buffer *vb) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count); - nv10->dirty |= NV10_NEW_VTXARRAYS; + memcpy(nv20->vtxbuf, vb, sizeof(*vb) * count); + nv20->dirty |= NV20_NEW_VTXARRAYS; - draw_set_vertex_buffers(nv10->draw, count, vb); + draw_set_vertex_buffers(nv20->draw, count, vb); } static void -nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count, +nv20_set_vertex_elements(struct pipe_context *pipe, unsigned count, const struct pipe_vertex_element *ve) { - struct nv10_context *nv10 = nv10_context(pipe); + struct nv20_context *nv20 = nv20_context(pipe); - memcpy(nv10->vtxelt, ve, sizeof(*ve) * count); - nv10->dirty |= NV10_NEW_VTXARRAYS; + memcpy(nv20->vtxelt, ve, sizeof(*ve) * count); + nv20->dirty |= NV20_NEW_VTXARRAYS; - draw_set_vertex_elements(nv10->draw, count, ve); + draw_set_vertex_elements(nv20->draw, count, ve); } void -nv10_init_state_functions(struct nv10_context *nv10) +nv20_init_state_functions(struct nv20_context *nv20) { - nv10->pipe.create_blend_state = nv10_blend_state_create; - nv10->pipe.bind_blend_state = nv10_blend_state_bind; - nv10->pipe.delete_blend_state = nv10_blend_state_delete; - - nv10->pipe.create_sampler_state = nv10_sampler_state_create; - nv10->pipe.bind_sampler_states = nv10_sampler_state_bind; - nv10->pipe.delete_sampler_state = nv10_sampler_state_delete; - nv10->pipe.set_sampler_textures = nv10_set_sampler_texture; - - nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create; - nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind; - nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete; - - nv10->pipe.create_depth_stencil_alpha_state = - nv10_depth_stencil_alpha_state_create; - nv10->pipe.bind_depth_stencil_alpha_state = - nv10_depth_stencil_alpha_state_bind; - nv10->pipe.delete_depth_stencil_alpha_state = - nv10_depth_stencil_alpha_state_delete; - - nv10->pipe.create_vs_state = nv10_vp_state_create; - nv10->pipe.bind_vs_state = nv10_vp_state_bind; - nv10->pipe.delete_vs_state = nv10_vp_state_delete; - - nv10->pipe.create_fs_state = nv10_fp_state_create; - nv10->pipe.bind_fs_state = nv10_fp_state_bind; - nv10->pipe.delete_fs_state = nv10_fp_state_delete; - - nv10->pipe.set_blend_color = nv10_set_blend_color; - nv10->pipe.set_clip_state = nv10_set_clip_state; - nv10->pipe.set_constant_buffer = nv10_set_constant_buffer; - nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state; - nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple; - nv10->pipe.set_scissor_state = nv10_set_scissor_state; - nv10->pipe.set_viewport_state = nv10_set_viewport_state; - - nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers; - nv10->pipe.set_vertex_elements = nv10_set_vertex_elements; + nv20->pipe.create_blend_state = nv20_blend_state_create; + nv20->pipe.bind_blend_state = nv20_blend_state_bind; + nv20->pipe.delete_blend_state = nv20_blend_state_delete; + + nv20->pipe.create_sampler_state = nv20_sampler_state_create; + nv20->pipe.bind_sampler_states = nv20_sampler_state_bind; + nv20->pipe.delete_sampler_state = nv20_sampler_state_delete; + nv20->pipe.set_sampler_textures = nv20_set_sampler_texture; + + nv20->pipe.create_rasterizer_state = nv20_rasterizer_state_create; + nv20->pipe.bind_rasterizer_state = nv20_rasterizer_state_bind; + nv20->pipe.delete_rasterizer_state = nv20_rasterizer_state_delete; + + nv20->pipe.create_depth_stencil_alpha_state = + nv20_depth_stencil_alpha_state_create; + nv20->pipe.bind_depth_stencil_alpha_state = + nv20_depth_stencil_alpha_state_bind; + nv20->pipe.delete_depth_stencil_alpha_state = + nv20_depth_stencil_alpha_state_delete; + + nv20->pipe.create_vs_state = nv20_vp_state_create; + nv20->pipe.bind_vs_state = nv20_vp_state_bind; + nv20->pipe.delete_vs_state = nv20_vp_state_delete; + + nv20->pipe.create_fs_state = nv20_fp_state_create; + nv20->pipe.bind_fs_state = nv20_fp_state_bind; + nv20->pipe.delete_fs_state = nv20_fp_state_delete; + + nv20->pipe.set_blend_color = nv20_set_blend_color; + nv20->pipe.set_clip_state = nv20_set_clip_state; + nv20->pipe.set_constant_buffer = nv20_set_constant_buffer; + nv20->pipe.set_framebuffer_state = nv20_set_framebuffer_state; + nv20->pipe.set_polygon_stipple = nv20_set_polygon_stipple; + nv20->pipe.set_scissor_state = nv20_set_scissor_state; + nv20->pipe.set_viewport_state = nv20_set_viewport_state; + + nv20->pipe.set_vertex_buffers = nv20_set_vertex_buffers; + nv20->pipe.set_vertex_elements = nv20_set_vertex_elements; } diff --git a/src/gallium/drivers/nv20/nv20_state.h b/src/gallium/drivers/nv20/nv20_state.h index 3a3fd0d4f4f..34f402fdcbf 100644 --- a/src/gallium/drivers/nv20/nv20_state.h +++ b/src/gallium/drivers/nv20/nv20_state.h @@ -1,10 +1,10 @@ -#ifndef __NV10_STATE_H__ -#define __NV10_STATE_H__ +#ifndef __NV20_STATE_H__ +#define __NV20_STATE_H__ #include "pipe/p_state.h" #include "tgsi/tgsi_scan.h" -struct nv10_blend_state { +struct nv20_blend_state { uint32_t b_enable; uint32_t b_srcfunc; uint32_t b_dstfunc; @@ -14,14 +14,14 @@ struct nv10_blend_state { uint32_t d_enable; }; -struct nv10_sampler_state { +struct nv20_sampler_state { uint32_t wrap; uint32_t en; uint32_t filt; uint32_t bcol; }; -struct nv10_rasterizer_state { +struct nv20_rasterizer_state { uint32_t shade_model; uint32_t line_width; @@ -43,24 +43,24 @@ struct nv10_rasterizer_state { const struct pipe_rasterizer_state *templ; }; -struct nv10_vertex_program_exec { +struct nv20_vertex_program_exec { uint32_t data[4]; boolean has_branch_offset; int const_index; }; -struct nv10_vertex_program_data { +struct nv20_vertex_program_data { int index; /* immediates == -1 */ float value[4]; }; -struct nv10_vertex_program { +struct nv20_vertex_program { const struct pipe_shader_state *pipe; boolean translated; - struct nv10_vertex_program_exec *insns; + struct nv20_vertex_program_exec *insns; unsigned nr_insns; - struct nv10_vertex_program_data *consts; + struct nv20_vertex_program_data *consts; unsigned nr_consts; struct nouveau_resource *exec; @@ -73,12 +73,12 @@ struct nv10_vertex_program { uint32_t or; }; -struct nv10_fragment_program_data { +struct nv20_fragment_program_data { unsigned offset; unsigned index; }; -struct nv10_fragment_program { +struct nv20_fragment_program { struct pipe_shader_state pipe; struct tgsi_shader_info info; @@ -89,7 +89,7 @@ struct nv10_fragment_program { uint32_t *insn; int insn_len; - struct nv10_fragment_program_data *consts; + struct nv20_fragment_program_data *consts; unsigned nr_consts; struct pipe_buffer *buffer; @@ -99,7 +99,7 @@ struct nv10_fragment_program { }; -struct nv10_depth_stencil_alpha_state { +struct nv20_depth_stencil_alpha_state { struct { uint32_t func; uint32_t write_enable; @@ -124,7 +124,7 @@ struct nv10_depth_stencil_alpha_state { } alpha; }; -struct nv10_miptree { +struct nv20_miptree { struct pipe_texture base; struct pipe_buffer *buffer; diff --git a/src/gallium/drivers/nv20/nv20_state_emit.c b/src/gallium/drivers/nv20/nv20_state_emit.c index 46c7e1d7536..23029433f3f 100644 --- a/src/gallium/drivers/nv20/nv20_state_emit.c +++ b/src/gallium/drivers/nv20/nv20_state_emit.c @@ -1,108 +1,108 @@ -#include "nv10_context.h" -#include "nv10_state.h" +#include "nv20_context.h" +#include "nv20_state.h" -static void nv10_state_emit_blend(struct nv10_context* nv10) +static void nv20_state_emit_blend(struct nv20_context* nv20) { - struct nv10_blend_state *b = nv10->blend; + struct nv20_blend_state *b = nv20->blend; - BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_DITHER_ENABLE, 1); OUT_RING (b->d_enable); - BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3); + BEGIN_RING(kelvin, NV10TCL_BLEND_FUNC_ENABLE, 3); OUT_RING (b->b_enable); OUT_RING (b->b_srcfunc); OUT_RING (b->b_dstfunc); - BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1); + BEGIN_RING(kelvin, NV10TCL_COLOR_MASK, 1); OUT_RING (b->c_mask); } -static void nv10_state_emit_blend_color(struct nv10_context* nv10) +static void nv20_state_emit_blend_color(struct nv20_context* nv20) { - struct pipe_blend_color *c = nv10->blend_color; + struct pipe_blend_color *c = nv20->blend_color; - BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1); + BEGIN_RING(kelvin, NV10TCL_BLEND_COLOR, 1); OUT_RING ((float_to_ubyte(c->color[3]) << 24)| (float_to_ubyte(c->color[0]) << 16)| (float_to_ubyte(c->color[1]) << 8) | (float_to_ubyte(c->color[2]) << 0)); } -static void nv10_state_emit_rast(struct nv10_context* nv10) +static void nv20_state_emit_rast(struct nv20_context* nv20) { - struct nv10_rasterizer_state *r = nv10->rast; + struct nv20_rasterizer_state *r = nv20->rast; - BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2); + BEGIN_RING(kelvin, NV10TCL_SHADE_MODEL, 2); OUT_RING (r->shade_model); OUT_RING (r->line_width); - BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1); + BEGIN_RING(kelvin, NV10TCL_POINT_SIZE, 1); OUT_RING (r->point_size); - BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2); + BEGIN_RING(kelvin, NV10TCL_POLYGON_MODE_FRONT, 2); OUT_RING (r->poly_mode_front); OUT_RING (r->poly_mode_back); - BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2); + BEGIN_RING(kelvin, NV10TCL_CULL_FACE, 2); OUT_RING (r->cull_face); OUT_RING (r->front_face); - BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2); + BEGIN_RING(kelvin, NV10TCL_LINE_SMOOTH_ENABLE, 2); OUT_RING (r->line_smooth_en); OUT_RING (r->poly_smooth_en); - BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_CULL_FACE_ENABLE, 1); OUT_RING (r->cull_face_en); } -static void nv10_state_emit_dsa(struct nv10_context* nv10) +static void nv20_state_emit_dsa(struct nv20_context* nv20) { - struct nv10_depth_stencil_alpha_state *d = nv10->dsa; + struct nv20_depth_stencil_alpha_state *d = nv20->dsa; - BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1); + BEGIN_RING(kelvin, NV10TCL_DEPTH_FUNC, 1); OUT_RING (d->depth.func); - BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_DEPTH_WRITE_ENABLE, 1); OUT_RING (d->depth.write_enable); - BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_DEPTH_TEST_ENABLE, 1); OUT_RING (d->depth.test_enable); #if 0 - BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_STENCIL_ENABLE, 1); OUT_RING (d->stencil.enable); - BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7); + BEGIN_RING(kelvin, NV10TCL_STENCIL_MASK, 7); OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7); #endif - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1); + BEGIN_RING(kelvin, NV10TCL_ALPHA_FUNC_ENABLE, 1); OUT_RING (d->alpha.enabled); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 1); + BEGIN_RING(kelvin, NV10TCL_ALPHA_FUNC_FUNC, 1); OUT_RING (d->alpha.func); - BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_REF, 1); + BEGIN_RING(kelvin, NV10TCL_ALPHA_FUNC_REF, 1); OUT_RING (d->alpha.ref); } -static void nv10_state_emit_viewport(struct nv10_context* nv10) +static void nv20_state_emit_viewport(struct nv20_context* nv20) { } -static void nv10_state_emit_scissor(struct nv10_context* nv10) +static void nv20_state_emit_scissor(struct nv20_context* nv20) { // XXX this is so not working -/* struct pipe_scissor_state *s = nv10->scissor; - BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2); +/* struct pipe_scissor_state *s = nv20->scissor; + BEGIN_RING(kelvin, NV10TCL_SCISSOR_HORIZ, 2); OUT_RING (((s->maxx - s->minx) << 16) | s->minx); OUT_RING (((s->maxy - s->miny) << 16) | s->miny);*/ } -static void nv10_state_emit_framebuffer(struct nv10_context* nv10) +static void nv20_state_emit_framebuffer(struct nv20_context* nv20) { - struct pipe_framebuffer_state* fb = nv10->framebuffer; + struct pipe_framebuffer_state* fb = nv20->framebuffer; struct pipe_surface *rt, *zeta = NULL; uint32_t rt_format, w, h; int colour_format = 0, zeta_format = 0; @@ -140,32 +140,32 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10) } if (zeta) { - BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + BEGIN_RING(kelvin, NV10TCL_RT_PITCH, 1); OUT_RING (rt->stride | (zeta->stride << 16)); } else { - BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1); + BEGIN_RING(kelvin, NV10TCL_RT_PITCH, 1); OUT_RING (rt->stride | (rt->stride << 16)); } - nv10->rt[0] = rt->buffer; + nv20->rt[0] = rt->buffer; if (zeta_format) { - nv10->zeta = zeta->buffer; + nv20->zeta = zeta->buffer; } - BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3); + BEGIN_RING(kelvin, NV10TCL_RT_HORIZ, 3); OUT_RING ((w << 16) | 0); OUT_RING ((h << 16) | 0); OUT_RING (rt_format); - BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); + BEGIN_RING(kelvin, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2); OUT_RING (((w - 1) << 16) | 0 | 0x08000800); OUT_RING (((h - 1) << 16) | 0 | 0x08000800); } -static void nv10_vertex_layout(struct nv10_context *nv10) +static void nv20_vertex_layout(struct nv20_context *nv20) { - struct nv10_fragment_program *fp = nv10->fragprog.current; + struct nv20_fragment_program *fp = nv20->fragprog.current; uint32_t src = 0; int i; struct vertex_info vinfo; @@ -193,108 +193,108 @@ static void nv10_vertex_layout(struct nv10_context *nv10) } void -nv10_emit_hw_state(struct nv10_context *nv10) +nv20_emit_hw_state(struct nv20_context *nv20) { int i; - if (nv10->dirty & NV10_NEW_VERTPROG) { - //nv10_vertprog_bind(nv10, nv10->vertprog.current); - nv10->dirty &= ~NV10_NEW_VERTPROG; + if (nv20->dirty & NV20_NEW_VERTPROG) { + //nv20_vertprog_bind(nv20, nv20->vertprog.current); + nv20->dirty &= ~NV20_NEW_VERTPROG; } - if (nv10->dirty & NV10_NEW_FRAGPROG) { - nv10_fragprog_bind(nv10, nv10->fragprog.current); - /*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */ - nv10->dirty_samplers |= (1<<10); - nv10->dirty_samplers = 0; + if (nv20->dirty & NV20_NEW_FRAGPROG) { + nv20_fragprog_bind(nv20, nv20->fragprog.current); + /*XXX: clear NV20_NEW_FRAGPROG if no new program uploaded */ + nv20->dirty_samplers |= (1<<10); + nv20->dirty_samplers = 0; } - if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) { - nv10_fragtex_bind(nv10); - nv10->dirty &= ~NV10_NEW_FRAGPROG; + if (nv20->dirty_samplers || (nv20->dirty & NV20_NEW_FRAGPROG)) { + nv20_fragtex_bind(nv20); + nv20->dirty &= ~NV20_NEW_FRAGPROG; } - if (nv10->dirty & NV10_NEW_VTXARRAYS) { - nv10->dirty &= ~NV10_NEW_VTXARRAYS; - nv10_vertex_layout(nv10); - nv10_vtxbuf_bind(nv10); + if (nv20->dirty & NV20_NEW_VTXARRAYS) { + nv20->dirty &= ~NV20_NEW_VTXARRAYS; + nv20_vertex_layout(nv20); + nv20_vtxbuf_bind(nv20); } - if (nv10->dirty & NV10_NEW_BLEND) { - nv10->dirty &= ~NV10_NEW_BLEND; - nv10_state_emit_blend(nv10); + if (nv20->dirty & NV20_NEW_BLEND) { + nv20->dirty &= ~NV20_NEW_BLEND; + nv20_state_emit_blend(nv20); } - if (nv10->dirty & NV10_NEW_BLENDCOL) { - nv10->dirty &= ~NV10_NEW_BLENDCOL; - nv10_state_emit_blend_color(nv10); + if (nv20->dirty & NV20_NEW_BLENDCOL) { + nv20->dirty &= ~NV20_NEW_BLENDCOL; + nv20_state_emit_blend_color(nv20); } - if (nv10->dirty & NV10_NEW_RAST) { - nv10->dirty &= ~NV10_NEW_RAST; - nv10_state_emit_rast(nv10); + if (nv20->dirty & NV20_NEW_RAST) { + nv20->dirty &= ~NV20_NEW_RAST; + nv20_state_emit_rast(nv20); } - if (nv10->dirty & NV10_NEW_DSA) { - nv10->dirty &= ~NV10_NEW_DSA; - nv10_state_emit_dsa(nv10); + if (nv20->dirty & NV20_NEW_DSA) { + nv20->dirty &= ~NV20_NEW_DSA; + nv20_state_emit_dsa(nv20); } - if (nv10->dirty & NV10_NEW_VIEWPORT) { - nv10->dirty &= ~NV10_NEW_VIEWPORT; - nv10_state_emit_viewport(nv10); + if (nv20->dirty & NV20_NEW_VIEWPORT) { + nv20->dirty &= ~NV20_NEW_VIEWPORT; + nv20_state_emit_viewport(nv20); } - if (nv10->dirty & NV10_NEW_SCISSOR) { - nv10->dirty &= ~NV10_NEW_SCISSOR; - nv10_state_emit_scissor(nv10); + if (nv20->dirty & NV20_NEW_SCISSOR) { + nv20->dirty &= ~NV20_NEW_SCISSOR; + nv20_state_emit_scissor(nv20); } - if (nv10->dirty & NV10_NEW_FRAMEBUFFER) { - nv10->dirty &= ~NV10_NEW_FRAMEBUFFER; - nv10_state_emit_framebuffer(nv10); + if (nv20->dirty & NV20_NEW_FRAMEBUFFER) { + nv20->dirty &= ~NV20_NEW_FRAMEBUFFER; + nv20_state_emit_framebuffer(nv20); } /* Emit relocs for every referenced buffer. * This is to ensure the bufmgr has an accurate idea of how * the buffer is used. This isn't very efficient, but we don't * seem to take a significant performance hit. Will be improved - * at some point. Vertex arrays are emitted by nv10_vbo.c + * at some point. Vertex arrays are emitted by nv20_vbo.c */ /* Render target */ // XXX figre out who's who for NV10TCL_DMA_* and fill accordingly -// BEGIN_RING(celsius, NV10TCL_DMA_COLOR0, 1); -// OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); - OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); +// BEGIN_RING(kelvin, NV10TCL_DMA_COLOR0, 1); +// OUT_RELOCo(nv20->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(kelvin, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv20->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - if (nv10->zeta) { + if (nv20->zeta) { // XXX -// BEGIN_RING(celsius, NV10TCL_DMA_ZETA, 1); -// OUT_RELOCo(nv10->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1); - OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); +// BEGIN_RING(kelvin, NV10TCL_DMA_ZETA, 1); +// OUT_RELOCo(nv20->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(kelvin, NV10TCL_ZETA_OFFSET, 1); + OUT_RELOCl(nv20->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); /* XXX for when we allocate LMA on nv17 */ -/* BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1); - OUT_RELOCl(nv10->zeta + lma_offset);*/ +/* BEGIN_RING(kelvin, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1); + OUT_RELOCl(nv20->zeta + lma_offset);*/ } /* Vertex buffer */ - BEGIN_RING(celsius, NV10TCL_DMA_VTXBUF0, 1); - OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); - BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1); - OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(kelvin, NV10TCL_DMA_VTXBUF0, 1); + OUT_RELOCo(nv20->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); + BEGIN_RING(kelvin, NV10TCL_COLOR_OFFSET, 1); + OUT_RELOCl(nv20->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); /* Texture images */ for (i = 0; i < 2; i++) { - if (!(nv10->fp_samplers & (1 << i))) + if (!(nv20->fp_samplers & (1 << i))) continue; - BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 1); - OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM | + BEGIN_RING(kelvin, NV10TCL_TX_OFFSET(i), 1); + OUT_RELOCl(nv20->tex[i].buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD); - BEGIN_RING(celsius, NV10TCL_TX_FORMAT(i), 1); - OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format, + BEGIN_RING(kelvin, NV10TCL_TX_FORMAT(i), 1); + OUT_RELOCd(nv20->tex[i].buffer, nv20->tex[i].format, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD | NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0, NV10TCL_TX_FORMAT_DMA1); diff --git a/src/gallium/drivers/nv20/nv20_surface.c b/src/gallium/drivers/nv20/nv20_surface.c index 875e4c58589..41b6d6ad35c 100644 --- a/src/gallium/drivers/nv20/nv20_surface.c +++ b/src/gallium/drivers/nv20/nv20_surface.c @@ -26,39 +26,39 @@ * **************************************************************************/ -#include "nv10_context.h" +#include "nv20_context.h" #include "pipe/p_defines.h" #include "pipe/p_winsys.h" #include "pipe/p_inlines.h" #include "util/u_tile.h" static void -nv10_surface_copy(struct pipe_context *pipe, unsigned do_flip, +nv20_surface_copy(struct pipe_context *pipe, unsigned do_flip, struct pipe_surface *dest, unsigned destx, unsigned desty, struct pipe_surface *src, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - struct nv10_context *nv10 = nv10_context(pipe); - struct nouveau_winsys *nvws = nv10->nvws; + struct nv20_context *nv20 = nv20_context(pipe); + struct nouveau_winsys *nvws = nv20->nvws; nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy, width, height); } static void -nv10_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, +nv20_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest, unsigned destx, unsigned desty, unsigned width, unsigned height, unsigned value) { - struct nv10_context *nv10 = nv10_context(pipe); - struct nouveau_winsys *nvws = nv10->nvws; + struct nv20_context *nv20 = nv20_context(pipe); + struct nouveau_winsys *nvws = nv20->nvws; nvws->surface_fill(nvws, dest, destx, desty, width, height, value); } void -nv10_init_surface_functions(struct nv10_context *nv10) +nv20_init_surface_functions(struct nv20_context *nv20) { - nv10->pipe.surface_copy = nv10_surface_copy; - nv10->pipe.surface_fill = nv10_surface_fill; + nv20->pipe.surface_copy = nv20_surface_copy; + nv20->pipe.surface_fill = nv20_surface_fill; } diff --git a/src/gallium/drivers/nv20/nv20_vbo.c b/src/gallium/drivers/nv20/nv20_vbo.c index d0e788ac036..4edc4efebd8 100644 --- a/src/gallium/drivers/nv20/nv20_vbo.c +++ b/src/gallium/drivers/nv20/nv20_vbo.c @@ -2,31 +2,31 @@ #include "pipe/p_context.h" #include "pipe/p_state.h" -#include "nv10_context.h" -#include "nv10_state.h" +#include "nv20_context.h" +#include "nv20_state.h" #include "nouveau/nouveau_channel.h" #include "nouveau/nouveau_pushbuf.h" -boolean nv10_draw_elements( struct pipe_context *pipe, +boolean nv20_draw_elements( struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned prim, unsigned start, unsigned count) { - struct nv10_context *nv10 = nv10_context( pipe ); - struct draw_context *draw = nv10->draw; + struct nv20_context *nv20 = nv20_context( pipe ); + struct draw_context *draw = nv20->draw; unsigned i; - nv10_emit_hw_state(nv10); + nv20_emit_hw_state(nv20); /* * Map vertex buffers */ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (nv10->vtxbuf[i].buffer) { + if (nv20->vtxbuf[i].buffer) { void *buf = pipe->winsys->buffer_map(pipe->winsys, - nv10->vtxbuf[i].buffer, + nv20->vtxbuf[i].buffer, PIPE_BUFFER_USAGE_CPU_READ); draw_set_mapped_vertex_buffer(draw, i, buf); } @@ -44,18 +44,18 @@ boolean nv10_draw_elements( struct pipe_context *pipe, } draw_set_mapped_constant_buffer(draw, - nv10->constbuf[PIPE_SHADER_VERTEX], - nv10->constbuf_nr[PIPE_SHADER_VERTEX]); + nv20->constbuf[PIPE_SHADER_VERTEX], + nv20->constbuf_nr[PIPE_SHADER_VERTEX]); /* draw! */ - draw_arrays(nv10->draw, prim, start, count); + draw_arrays(nv20->draw, prim, start, count); /* * unmap vertex/index buffers */ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (nv10->vtxbuf[i].buffer) { - pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer); + if (nv20->vtxbuf[i].buffer) { + pipe->winsys->buffer_unmap(pipe->winsys, nv20->vtxbuf[i].buffer); draw_set_mapped_vertex_buffer(draw, i, NULL); } } @@ -67,10 +67,10 @@ boolean nv10_draw_elements( struct pipe_context *pipe, return TRUE; } -boolean nv10_draw_arrays( struct pipe_context *pipe, +boolean nv20_draw_arrays( struct pipe_context *pipe, unsigned prim, unsigned start, unsigned count) { - return nv10_draw_elements(pipe, NULL, 0, prim, start, count); + return nv20_draw_elements(pipe, NULL, 0, prim, start, count); } diff --git a/src/gallium/drivers/nv20/nv20_vertprog.c b/src/gallium/drivers/nv20/nv20_vertprog.c index 72824559e8b..a885fcd7a56 100644 --- a/src/gallium/drivers/nv20/nv20_vertprog.c +++ b/src/gallium/drivers/nv20/nv20_vertprog.c @@ -6,8 +6,8 @@ #include "tgsi/tgsi_parse.h" #include "tgsi/tgsi_dump.h" -#include "nv30_context.h" -#include "nv30_state.h" +#include "nv20_context.h" +#include "nv20_state.h" /* TODO (at least...): * 1. Indexed consts + ARL @@ -32,47 +32,47 @@ #define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W) #define DEF_SCALE 0 #define DEF_CTEST 0 -#include "nv30_shader.h" +#include "nv20_shader.h" -#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) -#define neg(s) nv30_sr_neg((s)) -#define abs(s) nv30_sr_abs((s)) +#define swz(s,x,y,z,w) nv20_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w) +#define neg(s) nv20_sr_neg((s)) +#define abs(s) nv20_sr_abs((s)) -struct nv30_vpc { - struct nv30_vertex_program *vp; +struct nv20_vpc { + struct nv20_vertex_program *vp; - struct nv30_vertex_program_exec *vpi; + struct nv20_vertex_program_exec *vpi; unsigned output_map[PIPE_MAX_SHADER_OUTPUTS]; int high_temp; int temp_temp_count; - struct nv30_sreg *imm; + struct nv20_sreg *imm; unsigned nr_imm; }; -static struct nv30_sreg -temp(struct nv30_vpc *vpc) +static struct nv20_sreg +temp(struct nv20_vpc *vpc) { int idx; idx = vpc->temp_temp_count++; idx += vpc->high_temp + 1; - return nv30_sr(NV30SR_TEMP, idx); + return nv20_sr(NV30SR_TEMP, idx); } -static struct nv30_sreg -constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) +static struct nv20_sreg +constant(struct nv20_vpc *vpc, int pipe, float x, float y, float z, float w) { - struct nv30_vertex_program *vp = vpc->vp; - struct nv30_vertex_program_data *vpd; + struct nv20_vertex_program *vp = vpc->vp; + struct nv20_vertex_program_data *vpd; int idx; if (pipe >= 0) { for (idx = 0; idx < vp->nr_consts; idx++) { if (vp->consts[idx].index == pipe) - return nv30_sr(NV30SR_CONST, idx); + return nv20_sr(NV30SR_CONST, idx); } } @@ -85,16 +85,16 @@ constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w) vpd->value[1] = y; vpd->value[2] = z; vpd->value[3] = w; - return nv30_sr(NV30SR_CONST, idx); + return nv20_sr(NV30SR_CONST, idx); } #define arith(cc,s,o,d,m,s0,s1,s2) \ - nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) + nv20_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2)) static void -emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) +emit_src(struct nv20_vpc *vpc, uint32_t *hw, int pos, struct nv20_sreg src) { - struct nv30_vertex_program *vp = vpc->vp; + struct nv20_vertex_program *vp = vpc->vp; uint32_t sr = 0; switch (src.type) { @@ -163,9 +163,9 @@ emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src) } static void -emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) +emit_dst(struct nv20_vpc *vpc, uint32_t *hw, int slot, struct nv20_sreg dst) { - struct nv30_vertex_program *vp = vpc->vp; + struct nv20_vertex_program *vp = vpc->vp; switch (dst.type) { case NV30SR_TEMP: @@ -205,12 +205,12 @@ emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst) } static void -nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, - struct nv30_sreg dst, int mask, - struct nv30_sreg s0, struct nv30_sreg s1, - struct nv30_sreg s2) +nv20_vp_arith(struct nv20_vpc *vpc, int slot, int op, + struct nv20_sreg dst, int mask, + struct nv20_sreg s0, struct nv20_sreg s1, + struct nv20_sreg s2) { - struct nv30_vertex_program *vp = vpc->vp; + struct nv20_vertex_program *vp = vpc->vp; uint32_t *hw; vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi)); @@ -248,13 +248,13 @@ nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op, emit_src(vpc, hw, 2, s2); } -static INLINE struct nv30_sreg -tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { - struct nv30_sreg src; +static INLINE struct nv20_sreg +tgsi_src(struct nv20_vpc *vpc, const struct tgsi_full_src_register *fsrc) { + struct nv20_sreg src; switch (fsrc->SrcRegister.File) { case TGSI_FILE_INPUT: - src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); + src = nv20_sr(NV30SR_INPUT, fsrc->SrcRegister.Index); break; case TGSI_FILE_CONSTANT: src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0); @@ -265,7 +265,7 @@ tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { case TGSI_FILE_TEMPORARY: if (vpc->high_temp < fsrc->SrcRegister.Index) vpc->high_temp = fsrc->SrcRegister.Index; - src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); + src = nv20_sr(NV30SR_TEMP, fsrc->SrcRegister.Index); break; default: NOUVEAU_ERR("bad src file\n"); @@ -281,18 +281,18 @@ tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) { return src; } -static INLINE struct nv30_sreg -tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) { - struct nv30_sreg dst; +static INLINE struct nv20_sreg +tgsi_dst(struct nv20_vpc *vpc, const struct tgsi_full_dst_register *fdst) { + struct nv20_sreg dst; switch (fdst->DstRegister.File) { case TGSI_FILE_OUTPUT: - dst = nv30_sr(NV30SR_OUTPUT, + dst = nv20_sr(NV30SR_OUTPUT, vpc->output_map[fdst->DstRegister.Index]); break; case TGSI_FILE_TEMPORARY: - dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index); + dst = nv20_sr(NV30SR_TEMP, fdst->DstRegister.Index); if (vpc->high_temp < dst.index) vpc->high_temp = dst.index; break; @@ -317,11 +317,11 @@ tgsi_mask(uint tgsi) } static boolean -nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, +nv20_vertprog_parse_instruction(struct nv20_vpc *vpc, const struct tgsi_full_instruction *finst) { - struct nv30_sreg src[3], dst, tmp; - struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0); + struct nv20_sreg src[3], dst, tmp; + struct nv20_sreg none = nv20_sr(NV30SR_NONE, 0); int mask; int ai = -1, ci = -1; int i; @@ -484,7 +484,7 @@ nv30_vertprog_parse_instruction(struct nv30_vpc *vpc, } static boolean -nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, +nv20_vertprog_parse_decl_output(struct nv20_vpc *vpc, const struct tgsi_full_declaration *fdec) { int hw; @@ -539,7 +539,7 @@ nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc, } static boolean -nv30_vertprog_prepare(struct nv30_vpc *vpc) +nv20_vertprog_prepare(struct nv20_vpc *vpc) { struct tgsi_parse_context p; int nr_imm = 0; @@ -560,7 +560,7 @@ nv30_vertprog_prepare(struct nv30_vpc *vpc) tgsi_parse_free(&p); if (nr_imm) { - vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg)); + vpc->imm = CALLOC(nr_imm, sizeof(struct nv20_sreg)); assert(vpc->imm); } @@ -568,21 +568,21 @@ nv30_vertprog_prepare(struct nv30_vpc *vpc) } static void -nv30_vertprog_translate(struct nv30_context *nv30, - struct nv30_vertex_program *vp) +nv20_vertprog_translate(struct nv20_context *nv20, + struct nv20_vertex_program *vp) { struct tgsi_parse_context parse; - struct nv30_vpc *vpc = NULL; + struct nv20_vpc *vpc = NULL; tgsi_dump(vp->pipe.tokens,0); - vpc = CALLOC(1, sizeof(struct nv30_vpc)); + vpc = CALLOC(1, sizeof(struct nv20_vpc)); if (!vpc) return; vpc->vp = vp; vpc->high_temp = -1; - if (!nv30_vertprog_prepare(vpc)) { + if (!nv20_vertprog_prepare(vpc)) { FREE(vpc); return; } @@ -599,7 +599,7 @@ nv30_vertprog_translate(struct nv30_context *nv30, fdec = &parse.FullToken.FullDeclaration; switch (fdec->Declaration.File) { case TGSI_FILE_OUTPUT: - if (!nv30_vertprog_parse_decl_output(vpc, fdec)) + if (!nv20_vertprog_parse_decl_output(vpc, fdec)) goto out_err; break; default: @@ -626,7 +626,7 @@ nv30_vertprog_translate(struct nv30_context *nv30, { const struct tgsi_full_instruction *finst; finst = &parse.FullToken.FullInstruction; - if (!nv30_vertprog_parse_instruction(vpc, finst)) + if (!nv20_vertprog_parse_instruction(vpc, finst)) goto out_err; } break; @@ -643,35 +643,35 @@ out_err: } static boolean -nv30_vertprog_validate(struct nv30_context *nv30) +nv20_vertprog_validate(struct nv20_context *nv20) { - struct nouveau_winsys *nvws = nv30->nvws; - struct pipe_winsys *ws = nv30->pipe.winsys; - struct nouveau_grobj *rankine = nv30->screen->rankine; - struct nv30_vertex_program *vp; + struct nouveau_winsys *nvws = nv20->nvws; + struct pipe_winsys *ws = nv20->pipe.winsys; + struct nouveau_grobj *rankine = nv20->screen->rankine; + struct nv20_vertex_program *vp; struct pipe_buffer *constbuf; boolean upload_code = FALSE, upload_data = FALSE; int i; - vp = nv30->vertprog; - constbuf = nv30->constbuf[PIPE_SHADER_VERTEX]; + vp = nv20->vertprog; + constbuf = nv20->constbuf[PIPE_SHADER_VERTEX]; /* Translate TGSI shader into hw bytecode */ if (!vp->translated) { - nv30_vertprog_translate(nv30, vp); + nv20_vertprog_translate(nv20, vp); if (!vp->translated) return FALSE; } /* Allocate hw vtxprog exec slots */ if (!vp->exec) { - struct nouveau_resource *heap = nv30->screen->vp_exec_heap; + struct nouveau_resource *heap = nv20->screen->vp_exec_heap; struct nouveau_stateobj *so; uint vplen = vp->nr_insns; if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) { while (heap->next && heap->size < vplen) { - struct nv30_vertex_program *evict; + struct nv20_vertex_program *evict; evict = heap->next->priv; nvws->res_free(&evict->exec); @@ -691,11 +691,11 @@ nv30_vertprog_validate(struct nv30_context *nv30) /* Allocate hw vtxprog const slots */ if (vp->nr_consts && !vp->data) { - struct nouveau_resource *heap = nv30->screen->vp_data_heap; + struct nouveau_resource *heap = nv20->screen->vp_data_heap; if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) { while (heap->next && heap->size < vp->nr_consts) { - struct nv30_vertex_program *evict; + struct nv20_vertex_program *evict; evict = heap->next->priv; nvws->res_free(&evict->data); @@ -718,7 +718,7 @@ nv30_vertprog_validate(struct nv30_context *nv30) */ if (vp->exec_start != vp->exec->start) { for (i = 0; i < vp->nr_insns; i++) { - struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + struct nv20_vertex_program_exec *vpi = &vp->insns[i]; if (vpi->has_branch_offset) { assert(0); @@ -730,7 +730,7 @@ nv30_vertprog_validate(struct nv30_context *nv30) if (vp->nr_consts && vp->data_start != vp->data->start) { for (i = 0; i < vp->nr_insns; i++) { - struct nv30_vertex_program_exec *vpi = &vp->insns[i]; + struct nv20_vertex_program_exec *vpi = &vp->insns[i]; if (vpi->const_index >= 0) { vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK; @@ -754,7 +754,7 @@ nv30_vertprog_validate(struct nv30_context *nv30) } for (i = 0; i < vp->nr_consts; i++) { - struct nv30_vertex_program_data *vpd = &vp->consts[i]; + struct nv20_vertex_program_data *vpd = &vp->consts[i]; if (vpd->index >= 0) { if (!upload_data && @@ -792,8 +792,8 @@ nv30_vertprog_validate(struct nv30_context *nv30) } } - if (vp->so != nv30->state.hw[NV30_STATE_VERTPROG]) { - so_ref(vp->so, &nv30->state.hw[NV30_STATE_VERTPROG]); + if (vp->so != nv20->state.hw[NV30_STATE_VERTPROG]) { + so_ref(vp->so, &nv20->state.hw[NV30_STATE_VERTPROG]); return TRUE; } @@ -801,9 +801,9 @@ nv30_vertprog_validate(struct nv30_context *nv30) } void -nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) +nv20_vertprog_destroy(struct nv20_context *nv20, struct nv20_vertex_program *vp) { - struct nouveau_winsys *nvws = nv30->screen->nvws; + struct nouveau_winsys *nvws = nv20->screen->nvws; vp->translated = FALSE; @@ -829,8 +829,8 @@ nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp) so_ref(NULL, &vp->so); } -struct nv30_state_entry nv30_state_vertprog = { - .validate = nv30_vertprog_validate, +struct nv20_state_entry nv20_state_vertprog = { + .validate = nv20_vertprog_validate, .dirty = { .pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/, .hw = NV30_STATE_VERTPROG, -- cgit v1.2.3 From 0da43322bbc6ead4eeb1b9fe079a33e0d57bece5 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sat, 8 Nov 2008 18:32:33 +0200 Subject: Nouveau: build, link and use nv20. Signed-off-by: Pekka Paalanen --- configs/default | 2 +- src/gallium/drivers/nouveau/nouveau_winsys.h | 6 ++++++ src/gallium/winsys/drm/nouveau/Makefile | 1 + src/gallium/winsys/drm/nouveau/nouveau_winsys.c | 5 ++++- src/gallium/winsys/g3dvl/nouveau/Makefile | 3 ++- 5 files changed, 14 insertions(+), 3 deletions(-) diff --git a/configs/default b/configs/default index 6c123cc3c55..7ea27b48ce4 100644 --- a/configs/default +++ b/configs/default @@ -76,7 +76,7 @@ EGL_DRIVERS_DIRS = demo # Gallium directories and GALLIUM_AUXILIARY_DIRS = draw translate cso_cache pipebuffer tgsi sct rtasm util GALLIUM_AUXILIARIES = $(foreach DIR,$(GALLIUM_AUXILIARY_DIRS),$(TOP)/src/gallium/auxiliary/$(DIR)/lib$(DIR).a) -GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv04 nv10 nv30 nv40 nv50 failover +GALLIUM_DRIVER_DIRS = softpipe i915simple i965simple nv04 nv10 nv20 nv30 nv40 nv50 failover GALLIUM_DRIVERS = $(foreach DIR,$(GALLIUM_DRIVER_DIRS),$(TOP)/src/gallium/drivers/$(DIR)/lib$(DIR).a) GALLIUM_WINSYS_DIRS = xlib egl_xlib diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index 48feeba309d..a89b056244d 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -69,6 +69,12 @@ nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); extern struct pipe_context * nv10_create(struct pipe_screen *, unsigned pctx_id); +extern struct pipe_screen * +nv20_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); + +extern struct pipe_context * +nv20_create(struct pipe_screen *, unsigned pctx_id); + extern struct pipe_screen * nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *); diff --git a/src/gallium/winsys/drm/nouveau/Makefile b/src/gallium/winsys/drm/nouveau/Makefile index be630ff6d16..81562ca78d3 100644 --- a/src/gallium/winsys/drm/nouveau/Makefile +++ b/src/gallium/winsys/drm/nouveau/Makefile @@ -10,6 +10,7 @@ PIPE_DRIVERS = \ $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ $(TOP)/src/gallium/drivers/nv04/libnv04.a \ $(TOP)/src/gallium/drivers/nv10/libnv10.a \ + $(TOP)/src/gallium/drivers/nv20/libnv20.a \ $(TOP)/src/gallium/drivers/nv30/libnv30.a \ $(TOP)/src/gallium/drivers/nv40/libnv40.a \ $(TOP)/src/gallium/drivers/nv50/libnv50.a diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys.c index 0878840dcc0..364340e1d3d 100644 --- a/src/gallium/winsys/drm/nouveau/nouveau_winsys.c +++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys.c @@ -101,10 +101,13 @@ nouveau_pipe_create(struct nouveau_context *nv) switch (chipset & 0xf0) { case 0x10: - case 0x20: hws_create = nv10_screen_create; hw_create = nv10_create; break; + case 0x20: + hws_create = nv20_screen_create; + hw_create = nv20_create; + break; case 0x30: hws_create = nv30_screen_create; hw_create = nv30_create; diff --git a/src/gallium/winsys/g3dvl/nouveau/Makefile b/src/gallium/winsys/g3dvl/nouveau/Makefile index 5d11bde322b..ff43327778a 100644 --- a/src/gallium/winsys/g3dvl/nouveau/Makefile +++ b/src/gallium/winsys/g3dvl/nouveau/Makefile @@ -26,11 +26,12 @@ LDFLAGS += -L${DRMDIR}/lib \ -L${GALLIUMDIR}/auxiliary/rtasm \ -L${GALLIUMDIR}/auxiliary/cso_cache \ -L${GALLIUMDIR}/drivers/nv10 \ + -L${GALLIUMDIR}/drivers/nv20 \ -L${GALLIUMDIR}/drivers/nv30 \ -L${GALLIUMDIR}/drivers/nv40 \ -L${GALLIUMDIR}/drivers/nv50 -LIBS += -ldriclient -ldrm -lnv10 -lnv30 -lnv40 -lnv50 -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lm +LIBS += -ldriclient -ldrm -lnv10 -lnv20 -lnv30 -lnv40 -lnv50 -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lm ############################################# -- cgit v1.2.3 From 903ae9d04eb7d0066c4ba1e30acb44630ca129f4 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sat, 8 Nov 2008 19:07:04 +0200 Subject: Nouveau: update nv20 miptree according to nv40. glxinfo doesn't degfault anymore. Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv20/nv20_miptree.c | 29 ++++++++++++++++++----------- 1 file changed, 18 insertions(+), 11 deletions(-) diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c index dbfd779de9d..b014f4c22e3 100644 --- a/src/gallium/drivers/nv20/nv20_miptree.c +++ b/src/gallium/drivers/nv20/nv20_miptree.c @@ -94,25 +94,18 @@ nv20_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt) } } -static void -nv20_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt, - uint face, uint levels) -{ -} - - static struct pipe_surface * nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags) { - struct pipe_winsys *ws = screen->winsys; struct nv20_miptree *nv20mt = (struct nv20_miptree *)pt; struct pipe_surface *ps; - ps = ws->surface_alloc(ws); + ps = CALLOC_STRUCT(pipe_surface); if (!ps) return NULL; + pipe_texture_reference(&ps->texture, pt); pipe_buffer_reference(screen, &ps->buffer, nv20mt->buffer); ps->format = pt->format; ps->width = pt->width[level]; @@ -121,9 +114,14 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, ps->nblocksx = pt->nblocksx[level]; ps->nblocksy = pt->nblocksy[level]; ps->stride = nv20mt->level[level].pitch; + ps->usage = flags; + ps->status = PIPE_SURFACE_STATUS_DEFINED; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv20mt->level[level].image_offset[face]; + } else + if (pt->target == PIPE_TEXTURE_3D) { + ps->offset = nv20mt->level[level].image_offset[zslice]; } else { ps->offset = nv20mt->level[level].image_offset[0]; } @@ -132,9 +130,18 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, } static void -nv20_miptree_surface_release(struct pipe_screen *screen, - struct pipe_surface **surface) +nv20_miptree_surface_release(struct pipe_screen *pscreen, + struct pipe_surface **psurface) { + struct pipe_surface *ps = *psurface; + + *psurface = NULL; + if (--ps->refcount > 0) + return; + + pipe_texture_reference(&ps->texture, NULL); + pipe_buffer_reference(pscreen, &ps->buffer, NULL); + FREE(ps); } void nv20_screen_init_miptree_functions(struct pipe_screen *pscreen) -- cgit v1.2.3 From 72ae2bd85624ae994709c0c4b5e0933002c61854 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sun, 9 Nov 2008 00:58:19 +0200 Subject: Nouveau: fix nv20_vertex_layout() The function should update nv20->vertex_info, and not just a local struct that's thrown away immediately. Fixes a SIGFPE due to vbuf->vertex_size = 0 in vbuf_alloc_vertices(). Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv20/nv20_state_emit.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/src/gallium/drivers/nv20/nv20_state_emit.c b/src/gallium/drivers/nv20/nv20_state_emit.c index 23029433f3f..e8dd22925c4 100644 --- a/src/gallium/drivers/nv20/nv20_state_emit.c +++ b/src/gallium/drivers/nv20/nv20_state_emit.c @@ -168,28 +168,28 @@ static void nv20_vertex_layout(struct nv20_context *nv20) struct nv20_fragment_program *fp = nv20->fragprog.current; uint32_t src = 0; int i; - struct vertex_info vinfo; + struct vertex_info *vinfo = &nv20->vertex_info; - memset(&vinfo, 0, sizeof(vinfo)); + memset(vinfo, 0, sizeof(*vinfo)); for (i = 0; i < fp->info.num_inputs; i++) { switch (fp->info.input_semantic_name[i]) { case TGSI_SEMANTIC_POSITION: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_LINEAR, src++); break; case TGSI_SEMANTIC_COLOR: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++); + draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_LINEAR, src++); break; default: case TGSI_SEMANTIC_GENERIC: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); break; case TGSI_SEMANTIC_FOG: - draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); + draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++); break; } } - draw_compute_vertex_size(&vinfo); + draw_compute_vertex_size(vinfo); } void -- cgit v1.2.3 From 94631e63a8d56562b782ec30a6ec5b9d77ca70a0 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Sun, 9 Nov 2008 01:02:10 +0200 Subject: Mesa: fix number of buffers in st_draw_vbo(). The clean-up call to pipe->set_vertex_buffers() should use the same number of buffers as the first call. Signed-off-by: Pekka Paalanen --- src/mesa/state_tracker/st_draw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index bdf8648ef7c..fbf13aa8377 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -640,7 +640,7 @@ st_draw_vbo(GLcontext *ctx, pipe_buffer_reference(pipe->screen, &vbuffer[attr].buffer, NULL); assert(!vbuffer[attr].buffer); } - pipe->set_vertex_buffers(pipe, vp->num_inputs, vbuffer); + pipe->set_vertex_buffers(pipe, num_vbuffers, vbuffer); } -- cgit v1.2.3 From 66eacb3fed6bc1926e1925b164b7168a6e0a82d5 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Nov 2008 15:58:03 +1100 Subject: nouveau: pass object handle not pointer to GPU... --- src/gallium/winsys/drm/nouveau/nv04_surface.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/winsys/drm/nouveau/nv04_surface.c b/src/gallium/winsys/drm/nouveau/nv04_surface.c index 5bf89e1952e..d08955de718 100644 --- a/src/gallium/winsys/drm/nouveau/nv04_surface.c +++ b/src/gallium/winsys/drm/nouveau/nv04_surface.c @@ -386,7 +386,7 @@ nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc) BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE, 1); OUT_RING (chan, nvc->channel->vram->handle); BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE, 1); - OUT_RING (chan, nvc->NvSwzSurf); + OUT_RING (chan, nvc->NvSwzSurf->handle); BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN, 1); OUT_RING (chan, 0); BEGIN_RING(chan, nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_ROP, 1); -- cgit v1.2.3 From e658950d4e6c0e5f8b09fa89718d358030885e08 Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Nov 2008 16:12:28 +1100 Subject: nv40: init pipe_surface correctly --- src/gallium/drivers/nv40/nv40_miptree.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index f8d4497bf75..f321b721492 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -123,6 +123,8 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps->stride = mt->level[level].pitch; ps->usage = flags; ps->status = PIPE_SURFACE_STATUS_DEFINED; + ps->refcount = 1; + ps->winsys = pscreen->winsys; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = mt->level[level].image_offset[face]; -- cgit v1.2.3 From 1d6cba6572cc008760e9d4ff803b0b7e1420e41b Mon Sep 17 00:00:00 2001 From: Ben Skeggs Date: Mon, 10 Nov 2008 16:33:31 +1100 Subject: nouveau: remove previous hack around x86_64 breakage This hack now causes breakage *doh*. :) --- src/gallium/auxiliary/draw/draw_context.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c index b439bc4059f..41a4cba1ddd 100644 --- a/src/gallium/auxiliary/draw/draw_context.c +++ b/src/gallium/auxiliary/draw/draw_context.c @@ -64,10 +64,8 @@ struct draw_context *draw_create( void ) if (!draw_pt_init( draw )) goto fail; -#ifdef PIPE_ARCH_X86 if (!draw_vs_init( draw )) goto fail; -#endif return draw; -- cgit v1.2.3 From 3086b1ecbe718d05bdf016e01cff9f5928c42e63 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Wed, 12 Nov 2008 22:37:01 +0200 Subject: nv04..nv30: fix pipe_surface init Signed-off-by: Pekka Paalanen --- src/gallium/drivers/nv04/nv04_miptree.c | 2 ++ src/gallium/drivers/nv10/nv10_miptree.c | 2 ++ src/gallium/drivers/nv20/nv20_miptree.c | 2 ++ src/gallium/drivers/nv30/nv30_miptree.c | 2 ++ 4 files changed, 8 insertions(+) diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c index 02f7d210e31..0cbb91e187b 100644 --- a/src/gallium/drivers/nv04/nv04_miptree.c +++ b/src/gallium/drivers/nv04/nv04_miptree.c @@ -109,6 +109,8 @@ nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps->nblocksx = pt->nblocksx[level]; ps->nblocksy = pt->nblocksy[level]; ps->stride = nv04mt->level[level].pitch; + ps->refcount = 1; + ps->winsys = pscreen->winsys; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv04mt->level[level].image_offset[face]; diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index ad084e72b81..943f9e21e98 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -121,6 +121,8 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, ps->nblocksx = pt->nblocksx[level]; ps->nblocksy = pt->nblocksy[level]; ps->stride = nv10mt->level[level].pitch; + ps->refcount = 1; + ps->winsys = screen->winsys; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv10mt->level[level].image_offset[face]; diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c index b014f4c22e3..c6106d58c43 100644 --- a/src/gallium/drivers/nv20/nv20_miptree.c +++ b/src/gallium/drivers/nv20/nv20_miptree.c @@ -116,6 +116,8 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt, ps->stride = nv20mt->level[level].pitch; ps->usage = flags; ps->status = PIPE_SURFACE_STATUS_DEFINED; + ps->refcount = 1; + ps->winsys = screen->winsys; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv20mt->level[level].image_offset[face]; diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index 5c4f4da9481..9124db03e83 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -121,6 +121,8 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps->stride = nv30mt->level[level].pitch; ps->usage = flags; ps->status = PIPE_SURFACE_STATUS_DEFINED; + ps->refcount = 1; + ps->winsys = pscreen->winsys; if (pt->target == PIPE_TEXTURE_CUBE) { ps->offset = nv30mt->level[level].image_offset[face]; -- cgit v1.2.3